• 专利附录
  • 专利说明
  • 权利要求
  • 类似技术
  • 同族专利
  • 引用文献
  • 法律状态
  • 优先权
    • 专利摘要:
    A method for estimating the state of health of a battery of an electric or hybrid vehicle under conditions of use, comprising the following steps: a) during the operation of said battery, acquiring a time series of speed measurements ( v) or acceleration of said vehicle and, simultaneously, at least one time series of measurements (I, U, P) of a quantity chosen from: a current or a power delivered by said battery, and a voltage at its terminals; b) extracting segments of said time series corresponding to patterns of speed or acceleration satisfying at least one predefined condition; and c) determining estimates of health status of said battery by applying at least one continuous estimation or classification model to said segments of said time series. Device and system for implementing such a method. A method of constructing such a continuous estimation or classification model
    公开号:FR3016218A1
    申请号:FR1450025
    申请日:2014-01-03
    公开日:2015-07-10
    发明作者:Anthony Barre；Mathias Gerard；Frederic Suard
    申请人:Commissariat a lEnergie Atomique CEA；Commissariat a lEnergie Atomique et aux Energies Alternatives CEA；
    IPC主号:
    专利说明:

    [0001] METHOD, DEVICE AND SYSTEM FOR ESTIMATING THE HEALTH CONDITION OF A BATTERY OF AN ELECTRIC OR HYBRID VEHICLE WHEN USED, AND METHOD FOR CONSTRUCTING A MODEL FOR SUCH ESTIMATION The invention relates to a method , on a device and on a system for estimating the state of health of a battery of an electric or hybrid vehicle. The invention also relates to a method of constructing a model for estimating the state of health of such a battery. The state of health, or level of aging, of a battery 10 can be quantified by different variables. The most used are changes in capacity, resistance or the impedance of the battery studied. An indicator normally used is the SOH ("State of Health", nominal capacity at time tx, that is to say, state of health in English), defined by: initial capacity 100% Alternatively, the SOH is sometimes defined as from the resistance of the battery. Another frequently used indicator is Remaining Useful Life (RUL), which represents the proportion of time (or the number of cycles, for example) remaining to an EOL end-of-life criterion. "End of Life" in English), usually defined by a remaining capacity threshold, in percent RUL can also be called SOL ("State of Life"). Whatever the parameter used to define it, the state of health of a battery must be known in real time by the user in order to avoid the risk of an untimely failure, or an unexpected degradation of the performances. of the device or machine powered by the battery This is particularly important in the case of vehicle batteries - especially electric or hybrid cars - A direct calculation of the SOH by measuring the capacity or the resistance of the battery is possible in principle, but requires lengthy and complex measurements that can not be implemented in real time. As for the RUL / SOL, it can only be estimated.
    [0002] Due to the technical and economic importance of the problem, numerous methods for estimating the state of health of a battery have been proposed. For a review of these methods we can refer to the article by A. Barré et al. "A review on lithium-ion battery aging mechanisms and estimates for automotive applications", Journal of Power Sources 241 (2013) pages 680-689. Most of the existing solutions dealing with the problem of estimating the health level of a battery in real use, and in particular on an electric vehicle, use an equivalent circuit modeling the battery.
    [0003] The equivalent circuits are different from one proposition to another, according to the dynamics and validity ranges of these empirical models. Indeed, the modeling of a battery by equivalent circuit is very difficult due to the complexity of the many physico-chemical phenomena involved in its aging. On the other hand, this methodology is not sufficiently flexible because the parameters of an equivalent circuit must be adapted for each battery technology. Another major disadvantage of this methodology is that the model for estimating the level of aging receives input variables requiring prior estimates. For example, the state of charge and the resistance of the battery must be either measured (which requires many specific tests) or estimated (which is a complex problem in itself). Thus, these variables induce biases as soon as the model enters, which then causes a divergence of the results over time. Finally, such models also prove to be unrepresentative of the uses in real conditions because they are mainly based on tests in controlled conditions (test benches), which are not all significant of the real uses. By way of example, mention may be made of the article by B. Saha, K. Goebel, S. Poll and J. Christophersen, "Prognostics methods for battery health monitoring using a bayesian framework", IEEE Transactions on instrumentation and measurement 58 ( 2) (2009) 291-297. The method described in this paper uses an evolutionary equivalent circuit, characterized by parameters whose values are estimated by electrochemical impedance spectroscopy measurements. Aging curves, representing the evolution of these parameters, are determined "offline" by RVM (Relevance Vector Machine) relevant vector regression; then, the evolutionary model thus developed is used in a process of dynamic state estimation, of the PF ("Particle Filter") type. Other methods of estimating the state of health of a battery known from the prior art use aging maps defined in prior tests; see for example document FR2975188. These maps combine, for example, a measured resistance with a prediction of the capacity of the battery, or use a maximum measured voltage and a temperature to estimate a state of health. This methodology is not adaptable to real conditions. Indeed, in order to be representative of the real conditions, a cartography should take into account all the parameters that can intervene in the phenomena of aging. However, these are too numerous and interdependent to be effectively taken into account, which leads to a lack of reliability of the estimates thus obtained. The physical modeling approach is also widespread in the issues related to the aging of batteries; see for example US20130030739. It consists in determining equations modeling the evolution of the state of health of a battery. These equations are determined to be in agreement with data obtained on a test bench, but are not very suitable for modeling in real conditions, since the phenomena of degradation are very complex and come from many interdependent parameters, which leads to very difficult modeling. In addition these methods are not applicable online because the calculations requested are too complex for the power of embedded computers. Still other methods use learning methods such as neural networks and / or fuzzy logic from estimated signals and parameters. See, for example, US 2010/0324848. These methods can be used online; their main disadvantages are related to the use of data obtained on test bench. The invention aims to overcome, in whole or in part, at least some of the aforementioned drawbacks of the prior art. More particularly, the invention aims to allow a reliable estimate and "online" (that is to say, in use) of the state of health of a battery of an electric or hybrid vehicle. An object of the invention making it possible to achieve this goal is a method of estimating the state of health of a battery of an electric or hybrid vehicle under conditions of use, comprising the following steps: a) during the operation of said battery, acquiring a time series of measurements of speed or acceleration of said vehicle and, simultaneously, at least one time series of measurements of a quantity (chosen from: a current or a power delivered by said battery, and a voltage across its terminals, b) extracting segments of said time series corresponding to patterns of speed or acceleration satisfying at least one predefined condition; and c) determining estimates of the state of health of said battery by applying at least one continuous estimation or classification model to said segments of said time series. According to various embodiments of the invention: Said step a) may also include the simultaneous acquisition of a time series of temperature measurements of said battery; said step b) may also include extracting segments of said time series of temperature measurements corresponding to said speed or acceleration patterns; and said step c) may comprise the application of said or each said continuous estimation or classification model also to said segments of said time series of temperature measurements, or to an average temperature value associated with each said segment. In said step b), a segment of said time series of velocity or acceleration measurements can be considered as satisfying said predefined condition when a variation of velocity or acceleration, respectively, included in a first predefined range is produced in a time interval within a second predefined range. Said step c) may comprise an operation for resetting said segments of said time series of measurements, prior to the application of said or each said continuous estimation or classification model, said resetting operation comprising, for each said speed pattern : identifying a transformation converting said velocity pattern to a reference velocity pattern; and applying said transformation, or a transformation associated therewith, to each said segment of said time series corresponding to said speed pattern. Said or at least one said continuous estimation or classification model may be based on a metric or pseudo-metric chosen from: a pseudo-metric of dynamic temporal deformation; and a global alignment metric. Said or at least one said continuous estimation or classification model may be a kernel model. The method may also comprise a step d) of updating said one or more continuous estimation or classification models, or of a posteriori correction of said estimates, on the basis of estimates of the state of health of said battery obtained by characterization 25 off-line. Another object of the invention is a device for estimating the state of health of a battery of an electric or hybrid vehicle under conditions of use, comprising: at least a first input port for a signal an indication of a speed or acceleration of said vehicle; at least one second input port for a signal indicative of a current or a power delivered by said battery, or a voltage at its terminals; and a data processing module configured or programmed to implement a method as mentioned above using said signals. Yet another object of the invention is a system for estimating the state of health of a battery of an electric or hybrid vehicle under conditions of use, comprising: - such a device; at least one velocity or acceleration sensor of a vehicle, connected to said first port of said device; and at least one current or voltage sensor connected to said second port of said device. Yet another object of the invention is a method of constructing a model for estimating the state of health of a battery of an electric or hybrid vehicle under conditions of use, comprising the following steps: ) during a plurality of operating periods of said battery, acquiring a time series of velocity or acceleration measurements of said vehicle and, simultaneously, at least one time series of measurements of a magnitude selected from: a current or a power delivered by said battery, and a voltage at its terminals; B) extracting segments of said time series corresponding to patterns of speed or acceleration satisfying at least one predefined condition; C) determining reference health states of said battery during said operating periods by interpolating estimates of said health status obtained by offline characterization performed between said operating periods; and D) constructing at least one continuous estimation or classification model from said segments of said time series and corresponding reference health states. According to particular embodiments of such a method: Said step A) may also include the simultaneous acquisition of a time series of temperature measurements of said battery; said step B) may also include extracting segments of said time series of temperature measurements corresponding to said patterns of speed or acceleration; and said step D) may comprise constructing said continuous estimation or classification model also from said segments of said time series of temperature measurements, or an average temperature value associated with each said segment. In said step B), a segment of said time series of velocity or acceleration measurements can be considered as satisfying said predefined condition when a variation of velocity or acceleration, respectively, included in a first predefined range becomes produced in a time interval within a second predefined range. Said step D) may comprise a resetting operation of said segments of said time series of measurements, prior to the construction of said or each said continuous estimation or classification model, said resetting operation comprising, for each said speed pattern: identifying a transform converting said velocity pattern to a reference velocity pattern; and applying said transformation, or a transformation associated therewith, to each said segment of said time series corresponding to said speed pattern. Said or at least one said continuous estimation or classification model may be based on a metric or pseudo-metric chosen from: a pseudo-metric of dynamic temporal strain; and a global alignment metric. Said or at least one said continuous estimation or classification model may be a kernel model. Other characteristics, details and advantages of the invention will emerge on reading the description given with reference to the appended drawings given by way of example and which represent, respectively: FIG. 1, a block diagram of a system estimation of the state of health of a battery of an electric or hybrid vehicle according to one embodiment of the invention; FIG. 2 is a flow chart of a method for estimating the state of health of a battery of an electric or hybrid vehicle and of a method of constructing a model for such a two-mode estimation. embodiment of the invention; FIGS. 3A and 3B, a step of extracting time series segments of measurements corresponding to vehicle speed patterns satisfying a predefined condition, according to an embodiment of the invention; FIG. 4, the pseudo-metric of dynamic time deformation (DTW), used in an advantageous embodiment of the invention; FIGS. 5A and 5B, time series segments of speed and current measurements obtained during the implementation of a method according to one embodiment of the invention; FIGS. 6A, 6B, 6C and 7A, 7B, 7C, graphs illustrating an operation (optional) of resetting; and FIG. 8, the results of a continuous estimation of the state of health of a battery obtained by implementing a method according to one embodiment of the invention. FIG. 1 represents a BATT electric battery embedded in an electric or hybrid terrestrial vehicle VEL, supplying an electric motor ME and connected to a system for estimating its state of health according to one embodiment of the invention. This system, also on board, comprises a data processing module MTD and a plurality of sensors, and in particular: a voltage sensor CU for measuring the voltage U (t) at the terminals of the battery; a current sensor C1 for measuring a current 1 (t) supplied (or absorbed) by the battery, a temperature sensor CT for measuring an internal temperature T (t) of the battery and a speed sensor CV measuring the instantaneous speed v Other sensors may also be present, including other temperature sensors for measuring temperatures at different locations in the battery or its environment. Conversely, the temperature sensor CT and / or one of the two sensors CU, Cl (but not both at once) can be omitted. The speed sensor may be replaced or accompanied by a vehicle acceleration sensor, or any other sensor measuring a characteristic parameter of a state of motion of the latter. The data processing module MTD receives the signals generated by these sensors as input and outputs an estimate of the state of health of the battery (indicated by "SOH" in the figure, but it can be any parameter indicative of such a state of health, as for example the RUL). This module can notably comprise, by a processor programmed in a timely manner, accompanied by a memory storing one or more programs for the implementation of a method according to the invention, parameters of one or more estimation models of the state of health of the battery and possibly time series of measurements from said sensors (which is useful for the construction and / or offline update models). It may also include one or more other circuits, analog or digital, signal processing.
    [0004] The MTD data processing module, as well as all or some of the Cl, CU, CT and Cv sensors, may be part of a Battery Management System (BMS). As will be explained in detail later, the construction of the model or models for estimating the state of health of the battery 30 is done using both the signals from the sensors C1, CU, CT, Cv and the results of "offline" characterizations of the battery. This construction can be performed by the data processing module MTD (which must then receive the aforementioned results) or by an external computer, interfaced with the MTD module. According to the invention, the state of health of the battery BATT is estimated directly from signals from the battery and the vehicle, obtained via the sensors C1, CU, CT, Cv. Figure 2 illustrates: - in its left part, a method of constructing a model for estimating the state of health of the battery BATT; and in its right part, a method for estimating the health status of the BATT battery from this model. Both of these methods constitute two aspects of the present invention. They both use the signals generated by the sensors Cl, CU, CT, Cv during the actual operation of the battery and the vehicle. The construction method of the estimation model also uses health status baseline values obtained by offline characterization. The health status estimation method, on the other hand, is done entirely "online" or "in real time". The different steps of these two methods will now be described with reference, where necessary, to FIGS. 31, 3B and 4. I. Construction of the model or models for estimating the state of health of the battery (part of left of Figure 2). The construction of the model or models for estimating the state of health of the battery comprises the following steps: obtaining, in real time, data relating to the battery (current and / or voltage and / or power, possibly temperature ...) and to the vehicle (speed and / or acceleration), the extraction of reference speed patterns, as well as patterns of current and / or voltage and / or power and temperature values corresponding to these patterns. ; the mapping of these patterns with reference values of the state of health of the battery, obtained by interpolation of measurements made offline; the comparison of the extracted motifs, preceded 3016 2 1 8 11 or accompanied by a possible registration; and finally the actual construction of models of continuous or discrete (classification) estimation of the state of health of the battery. i. Obtaining the battery and vehicle data - blocks 100 200 and 300 of the flowchart in Figure 2. This first step is to collect data directly from the batteries and vehicles studied. These batteries (or only one) must have been used long enough to obtain complete and diverse data. The reference criterion is the end of life (EOL) of the battery, usually defined, in the case of electric vehicles, as the moment when a battery reaches 80% of its nominal initial capacity. During this phase, the batteries must be instrumented allowing the permanent acquisition of data during use (block 200), which can then be used in the invention. These values can be the temperature T of the batteries, the voltage U at their terminals, the delivered current I and the power delivered P (the latter can be obtained from measurements of voltage and current: P = U.I). The other variable extracted during the uses is the speed (and / or the acceleration) of the vehicles. All these acquisitions are made as and when tests (block 100). It is sufficient to have at least one of the I, U, P signals to establish a predictive model; however, several of these signals (I, U or I, P or U, P or I, U, P) can also be taken into account. The temperature information of each of the batteries can also be added as additional information, but is not necessary to set up the process. The construction of the models being in a decentralized setting, the data is stored, for example in a memory of the MTD processing module, to be processed at the end of the actual test data acquisition process. This then makes it possible to perform the calculations by means of a computer other than the BMS (battery management system) which acquires the data.
    [0005] On the other hand, it is necessary for the methodology to have health status references in order to build the models. These references must be obtained periodically during the experiment (block 300). This can be done through complete characterizations of the studied batteries or the vehicle (bench tests) or by other methods: empty voltage, etc. These tests make it possible to obtain aging parameters of the batteries, for example the maximum capacity or the value of the resistance of the battery at the times of the characterizations. These values serve as an aging reference for model construction. Interpolation (linear, cubic, etc.) provides a continuous evolution of these health status values of the battery. The "temporal" axis can be the experimentation time or the energy delivered, or even the distance according to the variables obtained during the tests. We thus obtain continuous health status changes for each of the batteries that have been studied in this process. Thus, there exist in this context three different types of data dependent on each other: - Battery data during their use: instantaneous current I, and / or voltage U and / or power P, and possibly temperature T; - Vehicle data during taxiing: f9; - Evolution of the reference of the state of health of each of the batteries. Subsequently, note S all the signals from the battery. Thus, S contains at least one of (I, U, and P) and may also contain temperature information T. By convention, the plural will be used for the set S, although the latter may comprise only one. signal (I or U or P). ii. Extraction of motifs v references - block 110. An idea underlying the present invention is to compare the differences between signals from the battery over time in order to predict the aging suffered by the battery. To do this, it is necessary to take a comparison criterion, in order to quantify the modification of the signals over time for identical or similar uses. Another idea underlying the invention is to extract time series of measurements from the sensors, repetitive patterns used as a reference to comparisons made thereafter. These comparisons will be established in order to identify the differences in signal behavior according to the level of aging corresponding to that moment. The signal that serves as a reference is the speed of the vehicle (in other embodiments, it could be acceleration). To extract repeating patterns, certain criteria must be set in order to perform this detection automatically. In the case of a speed signal, the extraction criteria can be the length of the pattern, as well as low and high speed thresholds. In this case, for a speed pattern to be selected, it is required that a speed variation within a first predefined range occur in a time interval within a second predefined range. In the example of FIG. 3A, it is required that a variation of speed of at least 20 km / h between a low speed of 20 km / h and a high speed of 40 km / h (first range) occurs in one time not greater than 2.5 seconds and not less than 3.7 seconds (second range); the lower limit of this range could be set to zero (consider all "fast" accelerations), but a non-zero lower bound is useful to avoid taking into account signals from measurement errors. In the example of FIG. 3A, a vehicle is traveling at a cruising speed of 50 km / h and undergoes five deceleration episodes followed by an acceleration that brings it back to cruising speed; then it accelerates to a new cruising speed of 100 km / h. Only acceleration phases are considered (which is not an essential limitation). The first acceleration phase is discarded because the variation from 20 km / h to 40 km / h occurs in a time greater than the upper limit of the second range; The second and the fifth acceleration phase, as well as the last acceleration that brings the vehicle to a speed of 100 km / h are discarded because the speed does not cross the low threshold of 20 km / h. On the other hand, the third and fourth acceleration phases satisfy the criterion indicated above.
    [0006] The level set for the thresholds delimiting the range of speed variation has a significant influence on the sensitivity and accuracy of the process. Thus, a higher threshold of high speed will induce a small number of saved patterns, which may lead to a model that is less efficient due to lack of data. On the contrary, the choice of a too narrow range of variation will lead to the extraction of a large number of patterns, but these will be too short to contain significant information on the state of health of the battery under study. In addition, the amplitude of the speed variation range may be chosen in accordance with the data acquisition frequency. As a result, a low acquisition frequency induces a wide speed variation range in order to be able to identify dynamics in the signals. One possible criterion consists in considering a limit length of 20 values per extracted segment, which represents, for example, a segment of 2 seconds for an acquisition frequency of 10 Hz. The purpose of this extraction is to characterize phenomena related to the aging of a battery, the pattern is preferably adapted to a strong braking, or a strong acceleration. iii. Extraction of time series of measurements corresponding to the speed patterns - block 210. Following the process of extracting the patterns of the reference variable (speed, or even acceleration), it is necessary to extract segments corresponding to said patterns from the time series of signals S from the battery. In other words, the information on the temporal location of the extracted reference motifs in the complete signals i is used in order to obtain segments or patterns of the set S (according to the variables taken into account) associated with the reference patterns (speeds). . This step is illustrated in FIG. 3B, which shows the extraction of current (I), voltage (U) and power (P) time series segments corresponding to the two speed patterns identified in the previous step.
    [0007] If the temperature T of the battery is used, it is sufficient to retain its average value in correspondence of the speed patterns extracted. Indeed, the temperature has a slow dynamic compared to the other variables.
    [0008] At the end of this step, there are n velocity profiles, associated with n segments of time series of measurements of each variable considered (I and / or U and / or P), and optionally n average values of temperature T. It is important to keep the location information in the experiment time, and the corresponding battery. iv. Taking into account the aging level - block 400 Then, each of the segments of S thus extracted is associated with one or more health status references (capacity, impedance, etc.), previously determined and stored (step Ii and block 300 in Figure 2). This leads to n sets of data each containing: - a reference pattern; At least one segment of a time series of measurements S chosen from I, U and P, and optionally a value T; and at least one reference of the state of health of the battery corresponding to the segment (s) of Set in the reference pattern. v. Comparison of the extracted segments, construction of distance matrices and, if applicable, block kernels 500. The objective of this step is to study the modification of the extracted profiles, according to the level of aging of the battery, in order to build health status estimation models. However, it must be considered that these variable patterns are sensitive to alterations in the reference pattern. Indeed, to be able to compare the extracted segments ideally, it would be necessary to have exactly the same reference patterns 30 (speeds), from the same conditions (temperature, load level, wind, driver ...). In such a case, the perceived changes in the S segments would be due solely to aging phenomena. However, obtaining exactly the same conditions is impracticable in the context of data in actual use. It is therefore useful to use a methodology that takes into account the modifications of the reference motifs. To do this, it may be envisaged to carry out a registration 5 by applying appropriate transformations to the reference motifs, so as to make them identical to each other, then to apply these same transformations - or corresponding transformations - on the segments of S associates. Such a resetting process can then be performed by wavelet methods, or by resetting from the Dynamic Time Warping (DTW) or by simply interpolating the signals. The principle of dynamic time deformation will now be illustrated with the help of FIG. 4. Let P = (pi, PN) and Q = (qi, qm) be two time series of lengths N and M. If NEM, these two series can not be compared by a simple Euclidean distance. Dynamic Time Warp (DTW) allows contraction and expansion of the time axis, alleviating alignment problems. The principle of the DTW metric (in fact, it is a pseudo-metric) consists of constructing a matrix of cost D (N x M), with a measure (1) (p; q), often defined as the Euclidean distance (ph q) = 11p; - q112, then find, among the set of alignments A (N, M) possible, the alignment Tc which minimizes the cumulative costs between P and Q. An alignment TC is of length In1 = L, and is composed of L -tuples (n1,7c2), such that: 1 = 7 [1 (1) 7 [1 (L) = N 1 = 7 [2 (1) <--- 7 [2 (L) = M 25 Thus, the distance DTW between two signals P and Q is defined by: DT1 / 1 / P. with: Dp, Q q, 2 (0) i = i As mentioned above, DTW (P, Q) is not rigorously a metric (one speaks of "pseudo-metric") because it does not satisfy the triangular identity: DTW (P, R) DTW (P, Q) + DTW (Q, R), VP, Q, R If the difference of the length between the time series is not too large (less than a factor of 2), it is possible to use a global alignment metric (GA, for the English "Global Alignment"). A distance GA takes into account all the costs more precisely the global alignment distance kGA is given by 10. Another possibility consists in not modifying the extracted signals and using a suitable comparison system taking into account this problem. It is then a question of considering the signals as they have been extracted, and of comparing them from a (pseudo) metric taking into account the temporal differences (for example, DTW). The DTW registration will now be illustrated using an example. Consider a set 51 of signals (V1,11, U1) (in solid lines in FIGS. 6A, 6B and 6C) and another set S2 of signals (V2, I2, U2) (in dashed line), each of these two sets from the same 20 speed extraction criteria (10-60 km / h between 7 and 10 seconds). DTW registration is then performed on the signals V1 and V2, in other words the transformation Tc described above is sought. As a reminder, this transformation (or alignment) matches the vectors V1 and V2 by time deformation. The process then consists in preserving this temporal deformation by directly applying it to I and U. The result of this method is visible in FIGS. 7A, 7B and 7C.
    [0009] Whatever the option chosen, it is necessary to apply one or more (pseudo) metrics, in order to quantify the differences between extracted segments. The choice of a metric must allow the initial changes due to the data recording conditions to be taken into account. The Euclidean distance can be used in the case of segments of identical length (which is unlikely in the real-life use). Other metrics such as the Manhattan distance [Mattausch02], or the "Complexity Invariant Distance" [Batista11] can be used. Other possibilities, not requiring that the segments be of identical length, are the distance from the DTW [Keogh05] or the cross-correlation between signals [Hirata08], for example. The particular advantage of the distance from the DTW is that it can be applied regardless of the length or shape of the segments. In addition, this method makes it possible to calculate the difference between two segments by taking into account the temporal distortions. These can, in the case studied here, be due to fluctuations in recording conditions (temperature, rain, wind, driving ...). Therefore the use of DTW seems particularly suitable to solve the problems related to changing conditions. Each of these metrics provides a value representing the difference between two extracted segments. It is then possible to apply one or more of these metrics to obtain a matrix (or matrices) of dissimilarity between each extracted segments. These matrices quantizing, in different ways, different segments, will be employed in the subsequent construction of the models. Finally, in the framework of model constructions by statistical methods, it can be chosen to calculate one (of) kernel (x) from the previously calculated metrics, or directly from the segments. The nuclei used may be directly derived from a dot product. For example, there may be mentioned, in a nonlimiting manner: Triangular core: 19 = (1 - K (zi - Gaussian kernel: - Epanechnikov kernel: (1 - L u corresponding to a distance between signals.
    [0010] Kernels calculated from the DTW are also conceivable. For example, there may be mentioned, in a nonlimiting manner: - Gaussian nucleus DTW (GDTW) [Lei08]; - DTW Negative (NDTW) [Gudmundsson08]; - DTW core "Softmax" [De Vries12]; - Gaussian elastic metric core (GEMK) [Zhang10]. Finally, kernels inspired by the DTW can also be calculated, such as those derived from the Global Alignment (GA) approach. For example, mention may be made, in a nonlimiting manner: - Global alignment nucleus (GAK) [Cuturi06]; - "logGA" core [Cuturi11]; - Triangular GAK Core (TGAK) [Joder08]. The kernels directly derived from the DTW are not, in all rigor, defined positive, although they are most often in application cases, because the DTW is not a metric, but a pseudo-metric.
    [0011] This characteristic is the reason for the GA approach that tries to solve this disadvantage. A quick review of the different kernels is done in an article [Joder08]. Each of these nuclei requires, for its construction, parameters to be fixed beforehand, for example by cross validation, which is a well-known method in statistics. Of course, it is also possible to use another type of kernel afterwards. In addition, T values, if taken into account, can be compared by a declining nucleus of Euclidean distance. At the end of this step, n sets of data (S, i, aging) created during the previous step are thus available, as well as distance matrices between the segments of S extracted and possibly of one or more nuclei. calculated for each variable (U and / or I and / or P and / or T). All this information can be used to build models for estimating the state of health of a battery simply from at least one pattern extracted from S and an associated reference pattern i. vi. Model construction - block 600A and / or 600B All the necessary information extracted by the previous steps forms a reference base. The last step of this part then consists in constructing models taking input of segments of S, as well as optionally a value T, associated with a reference pattern i, and which have the output of a health status estimate of the battery. corresponding. The output can be discrete or continuous depending on the type of method used in the model construction. 600A: Discrete Classification Model The purpose of the classification models is to predict a health status class for a new signal. Thus, the result of this type of model is not continuous, but discrete. Many algorithms are suitable for this type of problem. Health status classes can be intervals, regular or not, of values. The important thing is that all values are contained in one and only one class. Here is a non-exhaustive list of the methodologies closest to the problematic: - k-Nearest Neighbors (k-PPV): For a new S segment provided as input, this method determines the k segments closest to this one. here, from the distance taken into account in the previous step. (from a distance matrix or a kernel), and assigns the new segment to the majority class among the k neighbors.
    [0012] The only parameter to choose is the number of k neighbors chosen [Hastie01, p 463-468]. - Average k (kMeans): This method forms k clusters of segments taken as references, each associated with the majority health status class. The methodology of this algorithm is described in the literature [Hastie01, p. 460-461]. Thus, for a new segment, its assigned class will be that of the nearest cluster. This notion can be defined by the minimum average distance between the segment to be classified and all the segments of a cluster, or by the distance between the segment to be classified and the centroid of the segments of a cluster. - Hierarchical classification: In the same way as the kMeans method, this method divides a sample of reference segments into k clusters. The methodology here is to build a hierarchical tree from the distances calculated in the previous step. From this hierarchical tree, the algorithm consists in pruning it in order to form k clusters of segments. Once these clusters are formed, the diagnostic process is identical to that explained for the kMeans method [Hastie01, p. 520-525]. - Support Vector Machine (SVM): This supervised learning method, unlike the previous ones, requires a prior kernel (x) build. Moreover, it is the only one able to create a model built on several segments resulting from S, and also on one (of) nucleus (x) calculated from the values of T. To do this, the nuclei of the reference motifs are then associated (Multiple Kernel Learning, or MKL), as detailed in [Gonen11]. On the other hand, the full functioning process of the SVM method is also described in many articles such as [Hastie01, p. 423-431]. This method allows the more accurate detection of changes due to changing health states of the battery than is provided by the other methods explained.30 600A: Continuous Estimation Models The methodology used in the continuous estimation of The state of health of a battery is mainly based on the core (s) built in the previous step. As a result, if no kernel was calculated in this step, the continuous estimation will be done mainly by regression methods. The methods envisaged to carry out a continuous estimation of the state of health of a battery are, for example: - Regression methods: Regression algorithms, based on "shrinkage", such as the LASSO method [Hastie01, p. 68-69], the Ridge method [Hastie01, p. 61-64], or the LARS method [Efron03], - Vector Support Regression (SVR): This method is an extension of the SVM allowing continuous output. The process is detailed in the literature [Smola03]. As in the case of SVM, it is possible to take into account several kernels, resulting from the comparison of extracted S segments and T values. - Relevance Vector Machine (RVM): This is the most common kernel method in continuous output problems The principle is explained extensively in [Tipping01] In the same way as all kernel methods, the implementation of this method requires a choice in the values of the parameters. - Ridge Regression (Kernel Ridge Regression): An alternative to Ridge Regression is to use kernels in this method, as detailed in [Welling] 11. Diagnosing a Battery in Real Use (right side of the figure) 2) 3 This part deals with the application of the models built in part I., in a context of real use of a battery on an electric or hybrid vehicle.The aim is to reach a diagnosis of the ed state of health of the battery, without need of particular use. The process of estimating the state of health of the battery uses many steps described in part I. Thus, the application also requires an instrumented battery and vehicle, allowing the real-time acquisition of the same data as in Part II (I and / or U, and / or P, optionally T, fy) - see blocks 1200 and 1100 on the right-hand part of Figure 2. In addition, the estimation model (s) constructed during step I.vi. are here introduced into the methodology in the form of decision functions 10 of the input-output type. The processing of the data obtained from the battery consists initially in an extraction, as and when acquisitions, speed patterns corresponding to the criteria set in section I.ii. (blocks 1110 and 1220).
    [0013] Subsequently, as soon as a speed pattern has been extracted (block 1400), it can be used with the corresponding time series segments of S for estimating the state of health of the battery by application of a or several models built during step I.vi (block 1500). For this, the same process as that described in I.ii. at I.iv. is applied, thereby obtaining a signal I, U and / or P corresponding to the extracted speed pattern (as this is done in real time during acquisitions). If necessary, a registration is also applied, as during the construction of the models. In the case of using a single model, the response obtained on the state of health of the battery directly provides the estimate. On the contrary, if several models are used, as many estimates are obtained. Thus, the user can choose to consider all the estimates (display of all the results, which then forms a confidence interval), or to take into account all the values in order to calculate a diagnosis of health status from the estimators. This may consist, among other things, in calculating an average, a median, a selection of nearby values, or in prioritizing a method. It is important to note that this method is implemented during the use of an electric vehicle and in real conditions. Indeed, the diagnostics are provided when a reference speed pattern has been detected, which is why the definition of the latter is very important (thresholds and length of the pattern). Estimates are therefore provided immediately after the extraction of a suitable speed pattern. Indeed, once the stages I. of model construction completed, the calculation times are compatible with embedded use. Models for estimating the state of health of the battery can be updated. For this, it is necessary to obtain one (or more) new exact values of aging, resulting from specific tests. This can then be done during tests during a passage of the vehicle in a specialized garage. Two possibilities then make it possible to update: either the construction of new models with these data, or a correction of the bias of the estimates made. The first option is to reconstruct new models by the process explained in steps I; this requires an offline step of calculations. In the second case, it is only a question of applying a correction to the estimates made in order to correct the measured bias. In other words, if the last estimate predicts a resistance of 0.8 and the exact measured value specifically is 0.81, then a "+ 1.25%" fix will be applied to the new estimates. The technical result of the invention will now be illustrated by considering a specific example of implementation, using FIGS. 5A and 5B. The criteria chosen for the extraction of the speed patterns are then an acceleration of 20 to 40 km / h between 2.5 and 3.7 seconds, which corresponds to FIGS. 3A and 3B discussed above. The patterns obtained (FIG. 5A) illustrate the problem related to offsets due to external conditions. Here, only the current signals will be considered in order to make an estimate of the capacity of the battery. These signals I corresponding to the speed patterns are shown in FIG. 5B. From these data - and the reference values of the state of health of the battery - two models are developed to predict a level of capacity related to the initial capacity (SOE indicator): a discrete model resulting from the method kNN by DTW distance, another from the SVM method using logGAK cores. In the case of the kNN method, cross-validation allows the number of nearest neighbors (in terms of distance DTW) to be set to 24. The following four classes are considered: C1 = [100% - 96.75%], C2 = [96.75% - 93.50%], C3 = [93.50% - 90.25%], C4 = [ 90.25% - 87%]. The results of the method are illustrated by a matrix of confusion, presenting the percentages of good classifications, for an overall accuracy (percentage of signals whose classification is exact) of 59%: Class predicted Cl C2 C3 C4 Class real Cl 55.1% 28.6 % 12 .2% 4.1% C2 36.7% 40.9% 16.3% 6.1% C3 19% 4.8% 66.7% 9.5% C4 6% 8% 12% 74% An SVM classification was also performed on these same data, considering two classes C1 '= [100% - 93.50%] and C2' = [93.50% - 87%]. The SVM method also requires parameters to be set automatically by cross-validation, and more specifically a soft margin parameter C and a parameter conditioning the associated quadratic programming method (QP). The results then allow a global classification accuracy of 80%. The confusion matrix is: Predicted class C1 'C2' Real class Cl '74.4% 25.6% C2' 15.1% 84.9% Thus, these two methods provide two classes as a result. The choice made in terms of decision is then to take the averages of the bounds of these classes in the case where the results are different. Thus, when the results of the estimations at a given moment are respectively C2 = [96.75% - 93.50 %], and C2 '= [93.50% - 87%], one will retain the interval [95.125%, 90.25%]. A continuous RVM estimation model with a DTWK core was also tested using I and U patterns from accelerations between 10 and 60 km / h in 7 to 10 seconds. The results are illustrated in Figure 8. References: [Batista11] G.E.A.P.A. Batista, X. Wang, E. Keogh, A 15 Complex Invariant Distance Measure for Time Series, Proceedings of the Eleventh International Conference on Data Mining, SDM 2011, April 2830, 2011, Mesa, Arizona, USA. [Cuturi06] M. Cuturi, J.P. Green, O. Birkenes, T. Matsui, A kernel for time series based on global alignments, Acoustics, Speech and Signal Processing, 2007. ICASSP 2007, IEEE International Conference on. Flight. 2. IEEE, 2007. [Cuturil 1] M. Cuturi, Fast Global Alignment Kernels, Proceedings of the International Machine Learning Conference (ICML-11) (2011). [De Vriesl2] G. K. D. De Vries, Kernel methods for vessel trajectories (2012) p.45-66. [Efron03] B. Efron, T. Hastie, I. Johnstone, R. Tibshirani, Least angle regression, Ann.Statist. Volume 32, Number 2 (2004), 407-499. [Gonen11] M Giinen, E Alpaydin, Multiple kernel learning algorithms, Journal of Machine Learning Research, 999999 (2011): 22112268 [Gudmundsson08] S. Gudmundsson, TP Runarsson, S. Sigurdsson, Support Vector Machines and Dynamic Time Warping for Time Series , Neural Networks, 2008. IJCNN 2008. (IEEE World Congress on Computational Intelligence). IEEE International Joint Conference on. IEEE, 2008. [Hastie01] T. Hastie, R. Tibshirani, J. Friedman, The Elements of Statistical Learning, Vol. 2. New York: Springer (2001), p.59-64, 377-384, 412-417,472-479. [Hirata08] S. Hirata, M.K. Kurosawa, T. Katagiri, Cross-Correlation by Single-Bit Signal Processing for Ultrasonic Distance Measurement, IEICE Transactions on Fundamentals of Electronics, Communications and Computer Science 91.4 (2008): 1031-1037. [Joder08] C. Joder, S. Essid, G. Richard, Alignment kernels 15 for audio classification with application to music instrument recognition, European Signal Processing Conference (EUSIPCO 2008), Lausanne, Switzerland. 2008. [Keogh05] E. Keogh, CA Ratanamahatana, Exact Indexing of Dynamic Time Warping, Knowledge and Information Systems (7) (2005) 2 0 358 386. [Lei08] H. Lei, V. Govindaraju, A Study on the Dynamic Time Warping in Kernel Machines, Signal-Image Technologies and Internet-Based System, 2007. SITIS '07. Third International IEEE Conference on (pp. 839845). [Mattausch02] HJ Mattausch, N. Omori, S. Fukae, T. Koide, T. Gyoten, Fully-parallel pattern-matching engine with dynamic adaptability to Hamming or Manhattan Distance, VLSI Circuits Digest of Technical Papers 2002. Symposium on, IEEE (2002). [Smola03] A.J. Smola, B. Schiilkopf, A tutorial on support 30 vector regression, Statistics and Computing, - 14 (3): 199-222. [Tipping01] E. Tipping, Sparse Bayesian Learning and the Relevant Vector Machine, J. Mach. Learn. Res. 1 (2001) 211_244. [Welling] Mr Welling, Kernel Ridge Department of Computer Science, University of Toronto, Toronto (Kanada). [Zhang10] D. Zhang, W. Zuo, Zhang D., H. Zhang, Time series classification using Gaussian elastic metric 5 kernel, Proceedings of International Conference on Pattern Recognition (2010) IEEE (pp. 29-32) ).
    权利要求:
    Claims (15)
    [0001]
    REVENDICATIONS1. A method for estimating the state of health of a battery (BATT) of an electric or hybrid vehicle (VEL) under use conditions, comprising the following steps: a) during operation of said battery, acquiring a series temporal measurement of speed or acceleration (v) of said vehicle and, simultaneously, at least one time series of measurements of a quantity (I, U, P) chosen from: a current or a power delivered by said battery, and a voltage at its terminals; b) extracting segments of said time series corresponding to patterns of speed or acceleration satisfying at least one predefined condition; and c) determining estimates of health status of said battery by applying at least one continuous estimation or classification model to said segments of said time series.
    [0002]
    2. The method of claim 1 wherein: said step a) also comprises the simultaneous acquisition of a time series of temperature measurements (T) of said battery; said step b) also comprises extracting segments of said time series of temperature measurements corresponding to said patterns of speed or acceleration; and said step c) comprises the application of said or each said continuous estimation or classification model also to said segments of said time series of temperature measurements, or to an average temperature value associated with each said segment.
    [0003]
    The method according to one of the preceding claims, wherein in said step b), a segment of said time series of velocity or acceleration measurements satisfies said predefined condition when velocity or acceleration variation occurs. respectively in a first predefined range occurs in a time interval within a second predefined range.
    [0004]
    4. Method according to one of the preceding claims wherein said step c) comprises a resetting operation of said segments of said time series of measurements, prior to the application of said or each said model of continuous estimation or classification, said operation. resetting device comprising, for each said speed pattern: - identifying a transformation converting said speed pattern into a reference speed pattern; and - applying said transformation, or a transformation associated therewith, to each said segment of said time series corresponding to said speed pattern. 15
    [0005]
    5. Method according to one of the preceding claims wherein said or at least one said continuous estimation or classification model is based on a metric or pseudo-metric selected from: a pseudo-metric dynamic time deformation; and 20 - a global alignment metric.
    [0006]
    6. Method according to one of the preceding claims wherein said or at least one said model of continuous estimation or classification is a kernel model. 25
    [0007]
    7. Method according to one of the preceding claims also comprising a step d) of updating said one or more continuous estimation or classification models, or of a posteriori correction of said estimates, based on estimates of the state of health. of said battery 30 obtained by offline characterization.
    [0008]
    8. Device for estimating the state of health of a battery (BATT) of an electric or hybrid vehicle in use conditions, comprising: - at least a first input port for a signal (v) 5 indicative of a speed or acceleration of said vehicle; at least one second input port for a signal (I, U, P) indicative of a current or a power delivered by said battery, or a voltage at its terminals; and a data processing module (MTD) configured or programmed to implement a method according to one of the preceding claims using said signals.
    [0009]
    9. System for estimating the state of health of a battery of an electric or hybrid vehicle under conditions of use, comprising: - a device according to claim 8; at least one speed or acceleration sensor (Cv) of a vehicle, connected to said first port of said device; and at least one current or voltage sensor (C1, CU) connected to said second port of said device. 20
    [0010]
    10. A method for constructing a model for estimating the state of health of a battery of an electric or hybrid vehicle under conditions of use, comprising the following steps: A) during a plurality of periods operating said battery, acquiring a time series of measurements of velocity (v) or acceleration of said vehicle and, simultaneously, at least one time series of measurements (I, U, P) of a quantity selected from: a current or a power delivered by said battery, and a voltage at its terminals; B) extracting segments of said time series 30 corresponding to patterns of speed or acceleration satisfying at least one predefined condition; C) determining reference health states of said battery during said periods of operation by interpolation of estimates. said health status obtained by means of offline characterization performed between said operating periods; and D) constructing at least one continuous estimation or classification model from said segments of said time series and corresponding reference health states.
    [0011]
    11. The method of claim 10 wherein: said step A) also comprises the simultaneous acquisition of a time series of temperature measurements (T) of said battery; said step B) also comprises extracting segments of said time series of temperature measurements corresponding to said speed or acceleration patterns; and said step D) comprises constructing said continuous estimation or classification model also from said segments of said time series of temperature measurements, or from an average temperature value associated with each said segment. 20
    [0012]
    The method according to one of claims 10 or 11 wherein, in said step B), a segment of said time series of speed or acceleration measurements satisfies said predefined condition when a variation of speed or acceleration , respectively, within a first predefined range occurs at a time interval within a second predefined range.
    [0013]
    13. Method according to one of claims 10 to 12 wherein said step D) comprises a resetting operation of said segments 30 of said time series of measurements, prior to the construction of said or each said continuous estimation model or classification said resetting operation comprising, for each said speed pattern: - identifying a transformation converting said velocity pattern to a reference velocity pattern; and - applying said transformation, or a transformation associated therewith, to each said segment of said time series corresponding to said speed pattern.
    [0014]
    14. Method according to one of claims 10 to 13 wherein said or at least one said continuous estimation or classification model is based on a metric or pseudo-metric selected from: - a pseudo-metric of dynamic temporal deformation ; and - a global alignment metric.
    [0015]
    15. The method according to one of claims 10 to 14 wherein said or at least one said continuous estimation or classification model is a kernel model.
    类似技术:
    公开号 | 公开日 | 专利标题
    EP3089888B1|2020-10-14|Method, device and system for estimating the state of health of an electric or hybrid vehicle in running condition, and method for building a model for such estimation
    EP3044601B1|2019-12-11|Method, device and system for estimating the state of charge of a battery
    Li et al.2021|Online capacity estimation of lithium-ion batteries with deep long short-term memory networks
    EP2434304B1|2013-07-24|System and method of in situ battery diagnosis by electrochemical impedance spectroscopy
    EP2368161B1|2016-11-09|Detection of anomalies in an aircraft engine
    EP2888599B1|2020-05-27|Method for estimating the ageing of a battery
    EP2627982A1|2013-08-21|System for monitoring an engine test bed
    FR2990516A1|2013-11-15|ESTIMATING THE CHARGE OF A BATTERY
    EP3047290B1|2019-12-18|Method for estimating the ageing of a cell of a storage battery
    FR3008506A1|2015-01-16|METHOD AND DEVICE FOR PROVIDING POSITIVE POSITIVE DATA FOR AN OPERATING MODEL BASED ON DATA
    FR2991076A1|2013-11-29|METHOD FOR PROCESSING A SET OF QUADRUPLETS OF VALUES RELATING TO OPERATING POINTS OF AN ELECTROCHEMICAL ACCUMULATOR, METHOD FOR DETERMINING A STATE OF ENERGY FROM DATA DERIVED FROM THE PROCESSING PROCESS, RECORDING MEDIUM, COMPUTER PROGRAM AND DEVICE
    FR3067124B1|2019-07-05|METHOD AND SYSTEM FOR REAL-TIME DIAGNOSIS OF THE OPERATING STATE OF AN ELECTROCHEMICAL SYSTEM, AND ELECTROCHEMICAL SYSTEM INTEGRATING THE DIAGNOSTIC SYSTEM
    EP3588301A1|2020-01-01|Automatic and auto-optimised determination of parameters for executing a software application on an information processing platform
    Sun et al.2021|Battery state of health estimation method based on sparse auto‐encoder and backward propagation fading diversity among battery cells
    EP2775313A1|2014-09-10|Method of determining the residual capacity of a battery
    FR3084170A1|2020-01-24|METHOD FOR DETERMINING A COVARIANCE STATE NOISE MATRIX FOR SETTING A BATTERY CHARGE OBSERVER AND DEVICE THEREOF
    FR2944358A1|2010-10-15|Battery&#39;s e.g. lithium-ion battery, health state estimating device for electrical traction of e.g. hybrid vehicle, has samplers to calculate difference between symptomatic and effective parameters, and comparator to indicate health state
    FR3010200A1|2015-03-06|METHOD AND DEVICE FOR NORMALIZING VALUES OF OPERATING PARAMETERS OF AN AIRCRAFT ENGINE
    FR3062733A1|2018-08-10|METHOD FOR MONITORING EQUIPMENT OF ELECTROMECHANICAL ACTUATOR TYPE
    EP3754353A1|2020-12-23|Method for determining a robust estimator of the capacity of an electric battery
    EP3883029A1|2021-09-22|Method and system for determining a degradation indicator of a fuel cell
    FR2981754A1|2013-04-26|METHOD AND SYSTEM FOR DETERMINING THE CHARGE STATE OF AN ELEMENTARY CELL AND BATTERY
    FR3107597A1|2021-08-27|Method for estimating the energy state of a battery
    FR3040156A1|2017-02-24|AUTOMATED BRAKE ACTUATOR CONTROL METHOD AND VEHICLE OR VEHICLE BRAKE USING THE SAME
    FR3106415A1|2021-07-23|Rapid and offline diagnostic procedures for accumulators and associated devices
    同族专利:
    公开号 | 公开日
    EP3089888A1|2016-11-09|
    EP3089888B1|2020-10-14|
    WO2015101570A1|2015-07-09|
    JP2017509103A|2017-03-30|
    FR3016218B1|2016-01-01|
    US20170003352A1|2017-01-05|
    引用文献:
    公开号 | 申请日 | 公开日 | 申请人 | 专利标题
    US20130300377A1|2012-05-08|2013-11-14|GM Global Technology Operations LLC|Battery state-of-charge estimator using robust h observer|CN111090047A|2019-12-09|2020-05-01|泉州装备制造研究所|Lithium battery health state estimation method based on multi-model fusion|
    CN113504473A|2021-09-10|2021-10-15|四川大学|RUL prediction method suitable for power lithium battery nonlinear regression process|JP4353154B2|2005-08-04|2009-10-28|トヨタ自動車株式会社|Fuel cell vehicle|
    JP4238875B2|2006-02-10|2009-03-18|トヨタ自動車株式会社|Battery life evaluation system for hybrid vehicles|
    JP4179330B2|2006-03-31|2008-11-12|トヨタ自動車株式会社|Battery information display device for hybrid vehicle|
    KR100823507B1|2006-08-29|2008-04-21|삼성에스디아이 주식회사|Battery manegement system and the operating method thereof|
    US9274157B2|2007-07-17|2016-03-01|Midtronics, Inc.|Battery tester for electric vehicle|
    US8935195B2|2010-05-11|2015-01-13|The Royal Institution For The Advancement Of Learning/Mcgill University|Method of identification and devices thereof|
    FR2980274B1|2011-09-15|2014-06-13|Peugeot Citroen Automobiles Sa|METHOD FOR ESTIMATING AGING INDICATORS OF A TRACTION BATTERY OF AN ELECTRIC OR HYBRID MOTOR VEHICLE|
    US9157966B2|2011-11-25|2015-10-13|Honeywell International Inc.|Method and apparatus for online determination of battery state of charge and state of health|US9419992B2|2014-08-13|2016-08-16|Palantir Technologies Inc.|Unwanted tunneling alert system|
    CN104459560B|2014-12-29|2017-04-05|合肥工业大学|Lithium battery method for predicting residual useful life based on wavelet de-noising and Method Using Relevance Vector Machine|
    US9407652B1|2015-06-26|2016-08-02|Palantir Technologies Inc.|Network anomaly detection|
    KR20170011010A|2015-07-21|2017-02-02|삼성전자주식회사|Method and apparatus for estimating state of battery|
    US9537880B1|2015-08-19|2017-01-03|Palantir Technologies Inc.|Anomalous network monitoring, user behavior detection and database system|
    US10044745B1|2015-10-12|2018-08-07|Palantir Technologies, Inc.|Systems for computer network security risk assessment including user compromise analysis associated with a network of devices|
    US9857778B1|2016-10-07|2018-01-02|International Business Machines Corporation|Forecasting solar power generation using real-time power data, weather data, and complexity-based similarity factors|
    US10661805B2|2016-11-22|2020-05-26|Samsung Electronics Co., Ltd.|Vehicle control unitand operating method thereof|
    GB201621434D0|2016-12-16|2017-02-01|Palantir Technologies Inc|Processing sensor logs|
    CN110036304B|2016-12-21|2021-08-20|沃尔沃卡车集团|Battery management system and method for controlling battery management system|
    US10819116B2|2017-02-28|2020-10-27|International Business Machines Corporation|Forecasting solar power generation using real-time power data|
    FR3067124B1|2017-06-02|2019-07-05|Universite De Franche-Comte|METHOD AND SYSTEM FOR REAL-TIME DIAGNOSIS OF THE OPERATING STATE OF AN ELECTROCHEMICAL SYSTEM, AND ELECTROCHEMICAL SYSTEM INTEGRATING THE DIAGNOSTIC SYSTEM|
    CN107884718B|2017-10-11|2021-02-19|深圳天珑无线科技有限公司|Battery remaining capacity calculation method, mobile terminal and storage device|
    CN109613440B|2019-01-17|2021-06-08|清华-伯克利深圳学院筹备办公室|Battery grading method, device, equipment and storage medium|
    CN110223193A|2019-03-27|2019-09-10|东北电力大学|The method of discrimination and system of operation of power networks state are used for based on fuzzy clustering and RS-KNN model|
    DE102019205843A1|2019-04-24|2020-10-29|Robert Bosch Gmbh|Procedure for assessing the health of a high-voltage battery and battery tester|
    WO2021226505A1|2020-05-07|2021-11-11|Zitara Technologies, Inc.|Battery analysis system and method|
    法律状态:
    2015-02-02| PLFP| Fee payment|Year of fee payment: 2 |
    2016-02-01| PLFP| Fee payment|Year of fee payment: 3 |
    2017-01-31| PLFP| Fee payment|Year of fee payment: 4 |
    2018-01-31| PLFP| Fee payment|Year of fee payment: 5 |
    2020-01-30| PLFP| Fee payment|Year of fee payment: 7 |
    2021-01-28| PLFP| Fee payment|Year of fee payment: 8 |
    优先权:
    申请号 | 申请日 | 专利标题
    FR1450025A|FR3016218B1|2014-01-03|2014-01-03|METHOD, DEVICE AND SYSTEM FOR ESTIMATING THE HEALTH CONDITION OF A BATTERY OF AN ELECTRIC OR HYBRID VEHICLE WHEN USED, AND METHOD FOR CONSTRUCTING A MODEL FOR SUCH ESTIMATION|FR1450025A| FR3016218B1|2014-01-03|2014-01-03|METHOD, DEVICE AND SYSTEM FOR ESTIMATING THE HEALTH CONDITION OF A BATTERY OF AN ELECTRIC OR HYBRID VEHICLE WHEN USED, AND METHOD FOR CONSTRUCTING A MODEL FOR SUCH ESTIMATION|
    PCT/EP2014/079230| WO2015101570A1|2014-01-03|2014-12-23|Method, device and system for estimating the state of health of a battery in an electric or hybrid vehicle during operation thereof, and method for creating a model for an estimation of said type|
    US15/106,776| US20170003352A1|2014-01-03|2014-12-23|Method, device and system for estimating the state of health of a battery in an electric or hybrid vehicle during operation thereof, and method for creating model for estimation of said type|
    EP14821659.1A| EP3089888B1|2014-01-03|2014-12-23|Method, device and system for estimating the state of health of an electric or hybrid vehicle in running condition, and method for building a model for such estimation|
    JP2016544405A| JP2017509103A|2014-01-03|2014-12-23|Method, apparatus, and system for estimating battery degradation state during operation of electric or hybrid vehicle, and method for constructing model for estimation|
    [返回顶部]