比利时专利BE1025140B1 OPTIMIZING THE FUNCTIONING OF HOT WATER SYSTEMS

专利PDF首页>>比利时专利

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 optimizing the operation of at least one of a plurality of hot water systems in a plurality of buildings, the method comprising: collecting sensor data from the plurality of hot water systems in the plurality of buildings via a communication network, the collected sensor data comprising a plurality of sets of at least one hot water system data record and a corresponding hot water system type; estimating a behavior for each of the hot water system types of the multiple hot water systems based on the collected sensor data; optimizing the operation of at least one hot water system based on the estimated behavior of the corresponding type of hot water system.
公开号:BE1025140B1
申请号:E2017/5596
申请日:2017-08-29
公开日:2018-11-20
发明作者:Hussain Kazmi；Stefan Lodeweyckx
申请人:Enervalis Nv；
IPC主号:

专利说明:

(30) Priority data:
01/09/2016 EP 16186776.7 (73) Holder (s):
ENERVALIS NV
3530, WOODEN COLLECTION
Belgium (72) Inventor (s):
KAZMI Hussain
3500 HASSELT
Belgium
LODEWEYCKX Stefan
2990 WUUSTWEZEL-GOOREIND
Belgium (54) OPTIMIZING THE OPERATION OF HOT WATER SYSTEMS (57) Method for optimizing the operation of at least one of several hot water systems in multiple buildings, the method comprising: collecting sensor data from the multiple hot water systems in the multiple buildings via a communication network wherein the collected sensor data includes multiple sets of at least one hot water system data record and a corresponding hot water system type; estimate a behavior for each of the hot water system types of the multiple hot water systems based on the sensor data collected; optimize the operation of at least one hot water system based on the estimated behavior of the corresponding type of hot water system.

BELGIAN INVENTION PATENT
FPS Economy, K.M.O., Self-employed & Energy
Publication number: 1025140
Filing number: BE2017 / 5596
Intellectual Property Office
International classification: G05B 13/04
Date of grant: 20/11/2018
The Minister of Economy,
Having regard to the Paris Convention of 20 March 1883 for the Protection of Industrial Property;
Having regard to the Law of March 28, 1984 on inventive patents, Article 22, for patent applications filed before September 22, 2014;
Having regard to Title 1 Invention Patents of Book XI of the Economic Law Code, Article XI.24, for patent applications filed from September 22, 2014;
Having regard to the Royal Decree of 2 December 1986 on the filing, granting and maintenance of inventive patents, Article 28;
Having regard to the application for an invention patent received by the Intellectual Property Office on 29/08/2017.
Whereas for patent applications that fall within the scope of Title 1, Book XI, of the Code of Economic Law (hereinafter WER), in accordance with Article XI.19, § 4, second paragraph, of the WER, the granted patent will be limited. to the patent claims for which the novelty search report was prepared, when the patent application is the subject of a novelty search report indicating a lack of unity of invention as referred to in paragraph 1, and when the applicant does not limit his filing and does not file a divisional application in accordance with the search report.
Decision:
Article 1
ENERVALIS NV, Greenville Campus Center-South 1111, 3530 HOUTHALEN-HELCHTEREN Belgium;
represented by
PHILIPPAERTS Yannick, Meir 24 box 17, 2000, ANTWERP;
a Belgian invention patent with a term of 20 years, subject to payment of the annual fees as referred to in Article XI.48, § 1 of the Code of Economic Law, for:
OPTIMIZING THE OPERATION OF HOT WATER SYSTEMS.
INVENTOR (S):
KAZMI Hussain, Bosstraat 195/1, 3500, HASSELT;
LODEWEYCKX Stefan, Bredabaan 1032, 2990, WUUSTWEZEL-GOOREIND;
PRIORITY :
01/09/2016 EP 16186776.7;
BREAKDOWN:
Split from basic application:
Filing date of the basic application:
Article 2. - This patent is granted without prior investigation into the patentability of the invention, without warranty of the merit of the invention, nor of the accuracy of its description and at the risk of the applicant (s).
Brussels, 20/11/2018,
With special authorization:
BE2017 / 5596
Optimizing the operation of hot water systems
The present invention relates to a method for optimizing the operation of hot water systems and / or their interaction with the energy grid.
Anthropogenic climate change is one of the greatest challenges for humanity. Roughly 40% of global energy consumption and resulting greenhouse gases come from built environments, whether residential or commercial. In many cases, heating, ventilation and air conditioning (HVAC) and hot water (HW) facilities are the two most substantial sources. Two approaches have been vigorously pursued in recent years to reduce this situation: improving energy efficiency and encouraging renewable energy sources (such as sun, wind, etc.).
These improvements have paved the way for energy-neutral buildings (NZEBs), which produce as much energy (for example, using solar PV, etc.) as they consume on average over time. However, renewable energy sources are intermittent and stochastic by nature. When large-scale storage is absent, they are only available when the source (wind or sun, etc.) is of course available.
Furthermore, while the energy efficiency standards dramatically reduce energy consumption for HVAC in a new building, they do not significantly reduce the energy required for hot water supplies, as this is highly resident dependent and independent of facade improvements etc. to the building. These innovations also fail to affect the vast majority of construction stock that is over 20 years old and highly inefficient.
US 2016/0010878 describes a machine learning-based smart water heater controller using wireless sensor networks. This document describes to use multiple sensors in multiple places in the building to create a statistical model for the probability of a water decrease over time. Building characteristics are also determined, including heat loss properties and heating behavior. A disadvantage of this system is that many sensors are required in the building to provide enough data to determine the use and construction characteristics. These sensors must be implemented separately in each building. Implementing these sensors can be cumbersome, labor intensive, and costly. Furthermore, these sensors are optimized to teach user behavior, which means that unless otherwise specified, the operation of the hot water system remains unknown.
A standard approach to solve this problem of modeling the operation of the hot water system is to build a detailed thermodynamic model of the storage vessel using tests performed in a test laboratory in a controlled, offline setting. However, this step of building the model is impractical in the
BE2017 / 5596 sense that it must be repeated for each hot water system to calibrate the relevant parameters.
It is an object of the present invention to provide a method for automatic optimization of a hot water system operation that is cheaper, more efficient and less cumbersome to install.
To this end, the invention provides a method for optimizing the operation of at least one of several hot water systems in several buildings, the method comprising:
collecting sensor data from the multiple hot water systems in the multiple buildings via a communication network, the sensor data collected containing multiple sets of at least one hot water system data record and a corresponding hot water system type;
estimating a behavior for each of the hot water system types of the multiple hot water systems based on the sensor data collected; optimize the operation of at least one of the hot water systems based on the estimated behavior of the corresponding type of hot water systems.
The invention is based on the insight that hot water systems of the same type exhibit highly similar behavior, even when they are placed in different environments and when they are used differently. In fact, observing the same system in different environments with different disturbances allows to learn a more general model, i.e. one that is capable of accurately representing the hot water system under a wide range of conditions. Therefore, data coming from multiple buildings, and thus collected from multiple different hot water systems, can be combined and used together, especially when related to a hot water system of the same type, to estimate the behavior of the system. The practical consequence is that the minimum number of sensors required for a satisfactory optimization of operation drops significantly. A further consequence is that learning can be done more quickly by making different aspects of the hot water system visible, i.e. a substantial reduction in time required for learning the model to estimate the behavior of this system. As a result, fewer sensors are needed and additional or additional or specific sensors can be added to any building based on opportunity.
Opportunity based use and / or placement of sensors is significantly less cumbersome and expensive. Opportunity based use and / or placement of sensors allows to perform a cost-benefit analysis before sensors are provided. because of this
BE2017 / 5596 sensors can be provided in places that are easily accessible and / or in places where communication lines are already provided and / or in situations where a certain additional sensor is expected to provide relevant additional information. Furthermore, many buildings are already equipped with sensors such as temperature sensors, energy flow sensors such as gas flow and / or electricity flow meters. These sensors can be used in the method according to the invention and are regarded as opportunity-based sensors.
The combination of the use of the predetermined limited number of required sensors and a further number of additional sensors, added based on expediency, allows to optimize the operation of a hot water system in a building at a reduced cost. According to the invention, different types of data can be collected for different buildings and used together when the data relates to the same hot water system type, to estimate the behavior of the hot water system. In addition, the invention uses a data-driven modeling approach to estimate the behavior of the hot water system. The behavioral estimator may include a function approach technique such as a non-linear regression method, for example a neural network or a decision tree method.
The function approaches link the observed sensor information with a computer-based model of the hot water system. These function approaches take sensor data as input, such as observed vessel temperature, ambient temperature, user consumption as well as internal hot water system parameters. The output of this functional approach is a characterization of the entire hot water system. More specifically, the entire temperature distribution current in the storage vessel (in ° C or ° F), as well as the amount of energy (in watt-hours) required to reheat the storage vessel to any set point and the power (in Watts) that such an operation would consume , supplied. Preferably, the function approach technique used to teach the hot water system is appropriate to address the uncertainty surrounding its estimates. Many nonlinear function approximation techniques exist, which in addition to the estimated status also provide estimates of uncertainty including, but not limited to, a Boom based method, Gaussian method, and neural networks. Coulston et al., Approximating Prediction Uncertainty for Random Forest Regression Models, Photogrammatic engineering and remote sensing, Vol. 82, no. 3, March 2016, pp. 189-197, is hereby incorporated by reference for explaining tree-based methods. The document reference C.E. Rasmussen & C.K.I. Williams, Gaussian Processes for Machine Learning, the MIT Press, 2006, ISBN 026218253X is hereby incorporated by reference for explaining the Gaussian method. The document with reference G. Papadopoulos et al., Confidence Estimation Methods for Neural
BE2017 / 5596
Networks: A Practical Comparison, ESANN'2000 Proceedings, ISBN 2-930307-00-5, pp. 75-80 is hereby incorporated by reference for explaining neural networks.
This estimate of uncertainty is intrinsically higher for statuses that the job counselor has never seen before and is low for statuses that occur frequently. In particular, the estimation of uncertainty can be used to guide future system estimates to explore statuses that are currently unknown, but which offer the potential to improve overall operation. Combining multiple data streams for the same hot water system as input for the function approximator mainly reduces this uncertainty much faster and leads to a better estimate of the status.
The sensor data that are collected from the multiple buildings and that relate to that hot water system types are fed in a model so that the behavior is estimated based on the data collected. This is a so-called black box approach and improves the generalization potential of the estimate while minimizing manual modeling effort. The generalization potential refers to the ability to accurately estimate behavior under a wide range of operating conditions. Furthermore, the results of the behavior will be applicable to any hot water system of that type, including those not used in the model learning procedure. After the behavior is estimated for a hot water system type, the operation can be optimized based on this estimated behavior. This forms an iterative, closed-loop process, using the estimated behavior of the hot water system for optimization, and new data resulting from this optimization can be used to improve behavior estimation.
This also applies to newly installed hot water systems, the vessel type of which corresponds to a type whose behavior has previously been estimated based on previous installations from which data has been collected. The operation of such new installations can be immediately optimized based on the previously estimated support. Furthermore, previously learned behavioral assessment models for similar, but not the same, hot water systems can be used before sufficient data is collected in new installations to accelerate learning. This entire process can be further supplemented using thermodynamic models of hot water systems, when available. However, this thermodynamic model is only intended to additionally interact with the data-driven behavior estimate and does not reduce the general applicability of the invention.
A precise, reliable knowledge of the hot water system opens up the possibilities to perform active control to improve energy efficiency and to provide demand-driven services to the grid, etc. Because the optimization method builds on
BE2017 / 5596 historical data of users, guarantees can be built in to ensure resident comfort at all times. This approach is noticeably different from the current industrial trend in which research focuses on optimizing the 'how' demand, in particular energy efficiency is only marginally increased by controlling the heat cycles of the heating mechanism. By making such low-level thermodynamics the core of the optimization, the door is opened for near-local optimal control. When a more systematic approach of "when" is chosen to provide hot water depending on the user need and grid constraints, a more desirable result is achieved in terms of both energy efficiency and system boundaries. This invention relates in particular to this problem of local-global energy optimization: an exact estimate of the behavior of the hot water system and how the residents interact with it will lead to minimal energy consumption with minimal impact on the grid, with user comfort being virtually unaffected.
Preferably, the hot water system includes at least one hot water tank and a heating mechanism, and the hot water system type includes a corresponding at least one hot water tank type and a heating mechanism type. Different buildings can be equipped with different combinations of heating mechanisms and hot water tanks. Behavioral modeling is therefore preferably a combination of at least one model of the hot water vessel and a model of the heating mechanism. Preferably, the steps of estimating the behavior for each of the hot water system type include estimating a behavior for each of the hot water tank type and each of the heating mechanism type. When behavior is estimated for each of the hot water tank type and heating mechanism type, the behavior can be estimated more accurately, and different combinations can be made in buildings of heating systems and / or hot water tanks, the combination of which can be optimized by combining corresponding behavioral estimates.
Preferably, the method includes a training phase and an operational phase, wherein at least in the training phase, the at least hot water system data record includes at least one of the following:
- a water flow data record;
- an energy flow data record; and
- a water temperature data record.
Further preferably, in the operational phase, the at least one hot water system data record contains at least one energy flow data record. More specifically, the energy flow data record; and either the water flow data record or the water temperature data record together the minimum sensor data required. Furthermore, preferably can
BE2017 / 5596 additional data records are added. The additional data records are generated by sensors that are added to the hot water system and / or the building, based on opportunity. The water temperature data record can be recorded via a single sensor or a combination of sensors, and there is no limitation on sensor placement, at least when it is known approximately a priori.
Preferably, each of the plural sets of at least one hot water system data record and a corresponding hot water system type further includes a timestamp that provides a time-related indication of relevance of the hot water system data record. Hot water usage or time can be monitored via the time indication to extract usage patterns. These usage patterns can differ from building to building and from hot water system to hot water system. The hot water system behavior model is independent or at least mainly independent of the usage pattern. The hot water system status at a given time, on the other hand, depends on the usage pattern. Therefore, when on the one hand the usage pattern is known, and on the other hand the behavior of the system is known, the operation can be optimized. Furthermore, adding a timestamp allows to provide further data to the estimator, such as weather related data and / or dynamic energy price information. In this context it will be clear that the weather can also influence the behavior of the hot water system, so that such data in the estimator can improve the behavioral estimate.
Preferably, the estimation step includes:
- setting up a predetermined adaptive hot water system model for each of the hot water system types;
- supplying the collected sensor data to the corresponding hot water system models; and
- adapting the adaptive hot water system model to learn the behavior of the corresponding hot water system using machine learning algorithms.
The predetermined adaptive hot water system model is preferably formed by a general non-linear function approximation method. Sensor data is collected from all hot water systems belonging to the same family and fed to the function approach method. Optionally, real sensor data can be supplemented with data sampled from a thermodynamic model, if available. This aspect of the invention is suitable for explicitly combining data-based machine learning algorithms with the adaptive hot water system model, when available. Such an adaptive hot water system model can use thermodynamic behavioral principles or previously collected data from similar, but not the same, hot water system types,
BE2017 / 5596 and where the model is parameterized based on the data supplied to the model.
Preferably, the operation optimization step includes balancing the energy input to the hot water system over time to improve energy consumption and meet the hot water requirements. The general goal of optimizing the operation of a hot water system is that the requirements for hot water are met while a minimal amount of energy is consumed. Energy efficiency can then be maximized by providing a just-in-time heating, being by matching the supply of heat with the user's demand for hot water. A simple example can demonstrate why this behavior is desired. As long as the demand is met, any additional amount of hot water in the system will be immaterial for user comfort and thereby lead to unnecessary thermodynamic losses to the environment. Furthermore, systems such as can work more (or less) efficiently depending on the ambient weather conditions.
Additionally, the energy consumption profile over time can also affect energy costs. This will be made clear with a simple example. Drawing a quantity from the energy grid by means of an energy peak, in a period where no significant amount of green energy is produced, will result in a high electricity cost. When the same amount of energy is drawn from the electricity grid over a period of time and at a rate at which green energy is produced, the electricity will be noticeably less expensive. By mapping the hot water requirements and taking into account the behavior of the hot water system, energy consumption can be balanced to positively influence the environmental impact. Energy costs may also depend on variable rate schedules such as a time-of-use and time-of-day pricing schedule.
Preferably improving energy consumption includes at least one of:
- reducing general energy consumption;
- reducing energy consumption peaks;
- reducing the general energy cost;
- responding to energy grid imbalances; and
- matching energy availability with energy consumption.
The advantages illustrated above using the simple example apply when peaks are reduced and when available energy is matched with energy consumption.
Preferably, the hot water supplies are determined for each of the multiple hot water systems from sensor data collected, to be at least an amount over time and / or
BE2017 / 5596 temperature over the time of hot water used to indicate for each of the hot water systems. By mapping the hot water supplies, user comfort can be kept maximal while the hot water system is used over its maximum capacity over time. The amount of hot water available in the hot water system is monitored based on these requirements. Furthermore, the behavior of the hot water system can be taken into account in such a way that energy consumption can be balanced in an optimal way. This includes the rate at which the hot water system can be reheated to an acceptable temperature.
The invention further relates to a data storage device and encoder in machine-readable and machine-executable form for carrying out the steps of the method according to the invention. The invention further relates to a computer program in machine-readable and machine-executable form for carrying out the steps of the method according to the invention.
The invention will now be described in more detail with reference to drawings illustrating some preferred embodiments of the invention. In the drawings:
figure 1 schematically shows an embodiment of the invention;
Figure 2 is the problem that is at least partially solved by the invention;
Figure 3 shows a diagram for processing data from several buildings; dn figure 4 see a diagram illustrating the iterative character of the model building and the optimization process.
In the drawings, the same reference numeral is assigned to the same or analogous element.
In the description, the term energy efficiency refers to the possible increase or decrease in amount of energy required to provide a predetermined service, e.g. heating a water vessel. In this context, the predetermined service can be defined as having the same level of user comfort. The terms response-to-demand refer to the response to the status of the energy grid by switching the hot water system operation from times when energy supply is low to when supply is higher. In addition, automatic response-to-question will relate to doing this automatically, ie without direct feedback from the resident. The term building refers to a combination of a residential, commercial or industrial building that is equipped with a hot water system. The term hot water vessel is the storage vessel that stores the water and is an important aspect in both energy efficiency and response-to-demand concerns in this invention.
BE2017 / 5596
In this context it is admitted that the invention is independent of the physical characteristics of a particular hot water tank. The term heating mechanism or heat source is defined as the element used to convert an energy medium into hot water where the hot water is for use or storage in the hot water tank. In this context, it is noted that the invention is independent of the heating mechanism as long as the energy is measured either directly or in an aggregated form which can then be disaggregated. The term smart meter is defined as an electricity or gas consumption measuring device at the user's premises characterized by the measurement made with a specified granularity (e.g. 5 minutes). Preferably, the output of the smart meter can be sent over a communication network, for example a secure internet connection. The term flow meter refers to a sensor that, analogous to the smart meter, measures the flow of hot water as an amount over time. The term data aggregation means combining multiple flows of information into one flow. The term disaggregation refers to the inverse step of data aggregation. The term hot water tank status refers to the temperature spread in the tank that can be used to estimate the amount of residual hot water, i.e. above a certain predetermined temperature threshold, which would be available to the resident if he emptied the hot water tank completely. The term transition in the hot water tank status is the change in status depending on a disturbance caused by the behavior of a resident.
There are an estimated 50 million hot water storage tanks in the EU in 2016 at residential level only. Converting to thermal to electrical potential, this represents energy storage in the order of hundreds of TWhs. Also, the peak power potential is on the order of hundreds of GWs. Each of these figures alone is far above the requirements or effectively installed capacity of smaller European countries such as Belgium and the Netherlands. Applying the invention to only a fraction of these vessels will allow tremendous energy savings and the ability to respond to demand of the energy grid.
Empirical results show that applying the principles of the invention can allow energy savings of, on average, more than 10%. These energy savings can, when needed, be used to deliver response-on-demand and subordinate services to the energy grid without risking compromising resident comfort. These guarantees of resident comfort are in sharp contrast to the forced shutdown of certain services at the time of supply-demand imbalance and the alleged risk of complete extinction or blackout in many European countries at the time.
Finally, the invention also guarantees that it provides these services at the lowest possible system cost in an automated manner and in a manner that is robust to the presence or absence of certain sensors. It is important to note that
BE2017 / 5596 would be the closest alternative to providing all year round flexibility to install large scale electric batteries. However, it would take untold amounts of investment to get close to the amount of flexibility that can be achieved from these hot water tanks at a fraction of the cost.
This is further substantiated by the fact that:
- hot water vessels represent the year around energy flexibility for the energy grid, in the form of thermal storage to dampen intermittent, stochastic renewable energy generation; this contrasts with the thermal mass of the buildings, which is only available at certain times of the year, depending on the local climate.
- The energy consumed for hot water supply increases relatively because the HVAC consumption is drastically reduced due to specific energy efficiency gains.
- The power used for hot water supply represents a temporarily concentrated peak load while space heating (HVAC) is a more distributed load over time.
Figure 1 schematically illustrates an embodiment of the invention. Figure 1 shows several buildings 1. Figure 1 illustrates a first number of buildings 1a at a first geographical location. A second number of buildings 1b is provided in a second geographic location different from the first location, for example, in a different state or country. Figure 1 further shows a building 1c in more detail, 1c is shown only as an example to illustrate the invention.
Each of several buildings 1 is provided with at least one hot water system 2. In addition, each building unit typically has its own hot water system 2. Some buildings, such as an apartment building, have several building units and can therefore contain several hot water systems. Alternatively, a single hot water system can be used in a building to supply multiple building units with hot water.
A hot water system 2 includes at least a heating mechanism 4 for heating hot water. Preferably, the hot water system further includes a hot water vessel 3 adapted to store a predetermined amount of hot water. When a combination of a hot water vessel 3 and heating means 4 is provided as a hot water system, the water in the vessel 3 is typically kept at a predetermined minimum temperature by the heating mechanism. This is sub-optimal, because hot water vessels 3 are rarely completely emptied by a user. While such a configuration of a hot water system 2 containing a hot water vessel 3 and a heating mechanism 4 is highly comfortable for a user, it is highly inefficient. The user comfort is the result of a large amount of hot water that is constantly available to the user at any given time. The inefficiency can best be explained by means of a
BE2017 / 5596 simple example. If no hot water is used by a user, the hot water in the hot water tank will still be reheated periodically. A threshold temperature is set for this and when the temperature of the water in the water vessel falls below the threshold value, the heating mechanism 4 is activated to reheat the water in the vessel 3. This reheating mechanism is ignorant of the fact that there may be no demand for hot water from the occupant of the building as well as the non-linear dynamics of the hot water tank, in particular the hot water rising to the top and the temperature distribution in the storage vessel is not uniform. Furthermore, ambient temperature dependent heating systems such as heat pumps may be more or less efficient depending on when the reheating cycle occurs. Also, the external availability of energy at the time of reheating or the current energy cost is not taken into account. Furthermore, the rate of reheating, which is related to the power used to reheat the water, will not be actively controlled. From this example, it is clear that reheating can lead to a situation where the heating mechanism starts to reheat the storage vessel at 10:30 pm, when no green energy is available from solar panels and when, according to the specific models built on historical data, the chances are very small that the user will consume hot water in the coming hours. Drawing energy from the energy grid at such a time is most likely sub-optimal. However, optimization options can only be discovered when the behavior of the hot water system 2 is known and preferably when the hot water needs in the building have been mapped.
In the invention, sensors are provided in multiple buildings 1, which collects data regarding hot water use in the buildings and / or energy consumption / availability in the buildings. In addition, different sensors and a different number of sensors can be provided in different buildings with different purposes, as will be clear from the explanation and examples below.
The main effect of the invention is that the sensor requirements to optimize the hot water supply are minimized. A smart meter that measures the energy flow to the hot water system is considered the absolute minimum condition for the system to work. The temperature and flow meter are only needed for the initial phase in which the model is built for a new type of storage vessel. Each of these can be installed non-invasively at the exit of the storage vessel, if the vessel is not already equipped with it. More instances of hot water systems, available during the initial phase of building the model, can lead to a faster assessment of the behavior of the hot water system.
BE2017 / 5596
When a temperature sensor is available in the center of the storage vessel, or at another location in the vessel, it is also sufficient to build the model. Furthermore, the type of smart meter and the heating mechanism will not matter, practically this means that the system can work just as well with a gas heating system as with electric heaters, heat pumps and combined heat and power (CHP) systems etc. This means that the potential for generalizing the invention exceeds that which can be obtained through a naive black box alone or with a more specialized gray or white box model.
The smart meter is illustrated in figure 1 with reference numeral 5. The smart meter 5 connects the electricity or gas power system in the house 1c to the electricity or gas grid 6. The power meter 5 provides an indication of energy availability and / or energy consumption. Based on the data from the smart meter 5, in combination with the estimated behavior of the hot water system 2, which will be further described below, the hot water supply can be optimized. As described above, in order to build a model to estimate the behavior of the hot water system 2, additionally to the smart meter 5, a flow and temperature sensor 10 is also necessary. The flow and temperature sensor 10 measures the flow of the hot water and the temperature of the hot water consumed from the hot water system 2. In figure 1 a hot water vessel 3 is provided, and all the hot water consumed is extracted from the hot water vessel so that the temperature sensor and the flow sensor 10, illustrated as a single sensor, are provided at an output of the hot water vessel 3. It will be clear to the skilled person that the flow sensor and temperature sensor can also be designed in different elements and can be carried out in different locations. being placed. Based on the combination of the data from the smart meter 5 and the temperature and flow meter 10, the behavior of the hot water system 2 can be estimated by aggregating information from all hot water systems of the same type.
Additional sensors can be added based on opportunity. By using the additional sensors, the behavior of the hot water system 2 can be estimated in more detail and the hot water use can be mapped in more detail. In the example of figure 1, the building lc is provided with solar panels. Solar panels are typically connected to the building's internal electricity system via a converter 8. This converter 8 collects data specifically related to the availability of green energy in the building lc. It will become clear to those skilled in the art that this information about green energy availability can be used to optimize the energy consumption of the hot water system. In this invention, this information is considered data from sensors that is added based on expediency. The sensors are present in it
BE2017 / 5596 solar panel system and converter, and can provide data for the further optimization of the hot water system.
Similarly, hot water consuming appliances 9 can be provided with sensors to provide further information about the hot water usage. For example, the hot water consuming appliance 9 can provide information about the moments in time when hot water has been used, and can also provide information about the temperature at which the water has been used. When hot water is used by all hot water consuming appliances 9 at a maximum of 50 ° C, it would be sub-optimal to store the hot water in the hot water tank 3 at, for example, 80 ° C. In a reverse situation, the hot water needs could not be met. Therefore, such data represents knowledge that can be collected from the building and used in the invention. Figure 1 further shows a weather station 11 in building 1c, which generates weather-related data, which can also be used in the invention as additional sensor data. It will be clear to the skilled person that the weather can influence both the behavior of the heating system 2 and the need for hot water. The additional sensors described above are only mentioned as examples of data that directly and / or indirectly relate to the hot water system behavior and / or the hot water requirements in building 1.
Sensor data from the multiple buildings 1 are sent as illustrated in Figure 1 with arrows 12a, 12b and 12c to a data aggregator 15. The data aggregator 15 is preferably positioned on a server side 14. The server side 14 forms a virtual environment by a server and / or computer and / or network of servers and / or computers. In Figure 1, the server side is shown on the left side of the figure, separated from the field side 13, which is shown on the right side of the figure. On the field side 13, the data is collected at a geographical location by physically present sensors.
The transmitted data 12 is transmitted in sets of at least one data record and a hot water system type. Such sets can be formed directly by combining the data records of the sensor and the hot water system type installed in the building in which the sensor data is collected. Alternatively, the sets may be indirectly formed by identifying the building where the sensor data was collected when transmitting the sensor data, and separately providing a hot water system type information related to the identified building. At the end, the data aggregator 15 can group the incoming data from the multiple buildings 1 according to hot water system type. This is illustrated in Figure 1 with the arrows between the data aggregator 15 and the hot water vessels 16 and the heating mechanisms 17. When the hot water system 2 is constituted by a hot water vessel 3 and a heating mechanism 4, the data can be grouped accordingly.
BE2017 / 5596
At the server side 14 or locally in the field 13, a generalizable, nonlinear function approximation model using an array of regression methods including neural network and random forest based estimators is preferably used to estimate the behavior of the hot water system. Such a model is provided to learn the behavior of each hot water tank type, indicated in Figure 1a with 16a, 16b, ..., 16x, and for each type of heating mechanism, indicated in Figure 1a with 17a, 17b, ..., 17x. The function approximation can be any regression or classification method with sufficient representational power to correctly approximate the nonlinear dynamics of the hot water tank status and the energy consumption of the heating system. This assembly method provides estimates for both the current vessel status and how uncertain the position approximator is about the estimate. This uncertainty estimate is a function of whether the same data has previously been observed. This hot water system model can be used to simulate the status of the hot water tank based on a start status and the expected behavior of residents. Furthermore, it can predict the expected amount of energy that the heating mechanism will consume to reheat the vessel, along with an estimate of the uncertainty about this prediction, as well as the distribution of this energy over time, in particular power consumption.
The expected current status and future transitions of the hot water tank are used together with the predictions of the heating mechanism and the opportunity-based sensors to optimize the operation of the hot water system. This may refer to maximizing energy efficiency, minimizing energy costs, and matching energy supply to energy demand, etc. However, because data is used to teach the model in online settings, it may be uncertain or even incorrect for system states never seen before. This magnitude of uncertainty about the condition of the vessel and the heating mechanism is incorporated in the process of making a decision and by guiding the exploration strategy of the hot water system, being to identify regions of the space's status leading to better optimization can lead.
By supplying the aggregated data to the corresponding behavior models 16 and 17, the behavior of each hot water tank type and of each heating mechanism type can be modeled. Preferably, self-learning computer algorithms are provided for parameterizing and fine-tuning the behavioral model based on the data. The aim is to form the behavioral model in such a way that the response of the model to an input corresponds to the response of the hot water system 2 to a similar input, which response is measured by the sensors.
BE2017 / 5596
By collecting a large amount of data from multiple buildings 1, in which different types of sensor data are collected based on opportunity, a highly detailed behavioral model can be generated at a surprisingly low cost. After these models have been generated, they can be used to optimize the operation of the corresponding hot water system. Therefore, the left side of Figure 1 is separated into an upper part 18 showing the model building phase and a lower part 19 showing the optimization phase. The modeling phase is preferably performed on the server side 18, while the optimization phase 19 can be performed remotely as well as on the server side 14 or locally in the field in distributed settings 13. Figure 1 shows an example where the optimization phase is also performed on the server side 13 .
Based on the sensor data 12 coming from each building, hot water supplies are mapped and energy availability is monitored as illustrated with block 20 in Figure 1. The dotted line 12c has only been added to clarify that the mapping of the hot water supplies and / whether the energy availability has been determined for each of the multiple buildings 1 separately. In figure 1 the operation of the hot water system of building lc is optimized. For this purpose, the sensor data sent from this specific building lc 12c is used to map hot water supplies and energy availability 20. Furthermore, it is known for this building lc that the building lc contains a hot water tank 3 corresponding to model 16b and contains a heating mechanism 4 corresponding to model 17a. Therefore, models 16b and 17a are used together with the hot water supplies and energy availability to calculate optimization parameters 21 for the hot water system. These calculated parameters 21 can be provided, as illustrated with arrow 22, to the building. The parameters 21 form the basis for the operating settings of the hot water system 2, so that a superior operation is obtained. The calculation of optimization parameters 21 can focus on optimizing energy efficiency, and / or user comfort, and / or optimizing the use of green energy and / or maximally reducing energy peaks consumed by the hot water system 2, or can focus on other optimization goals . Multiple goals can also be set simultaneously, whereby the goals are balanced with respect to each other to obtain optimal functioning of the hot water system. In this context, those skilled in the art will understand that optimization goals can be prioritized based on user preferences and / or properties of the energy grid used by the hot water system.
Figure 2 shows in a simple diagram the solar energy available over time with line 23, and shows the hot water usage over time with line 24. Using the hot water system model, the behavior of the hot water system of which can be determined, the operation of the hot water system are optimized in such a way that the energy consumption
BE2017 / 5596 can be shifted maximum to line 23. Preferably, the knowledge that a peak of hot water usage will arrive in the late afternoon can be used to prepare the hot water needed during the peak around noon, using solar energy 23. The skilled person will immediately recognize how this information can be used to optimize the operation of the hot water system 2.
Figure 3 shows a self-explanatory diagram illustrating how aggregated data from multiple buildings can be used to optimize the operation of the hot water system and to build the behavior model. It shows one possible configuration for the expected system behavior in the presence or absence of certain sensor configurations. Figure 3 also illustrates the case of incomplete sensor data; once a model has been trained for the hot water system, it is possible to optimize the hot water system operation. As an example, when a model is taught for a particular type of hot water system, the operation of that type of hot water system, when placed in new buildings, can be operated without the need to install the flow meter for measuring hot water consumption. In this case, the previously learned hot water vessel model and the temperature sensor data in the new building can be used to estimate the flow of hot water, allowing future optimization of the operation of the hot water system. A number of other options also exist to handle cases where other, or additional, sensors are not present.
Figure 4 shows the possible iterative character of the model building and optimization process; where a model for the hot water system is built up with collected data, it is then used to optimize the operation of the hot water system. Afterwards, additional data records can be obtained, which can be used to update the model and its uncertain estimates. The frequency of how often this model needs to be updated can be time independent (e.g. once a week), time dependent (e.g. once a day at the start of operation and once a month after substantial data has been collected) or data dependent (e.g. in function of gathered information, once enough new data has been collected to substantially improve the existing model). This figure shows only one example, and other schemes can also be used to manage the aggregated data.
Based on the description above and the corresponding figures, those skilled in the art will be able to understand the advantages and effects of the invention. Furthermore, the skilled person will immediately understand that the invention is not limited to the examples given, and that much more sensor data, behavioral estimates and optimization options are available to the skilled person. In addition, the invention is not limited to the examples described above, and the scope of protection is defined only in the claims.
BE2017 / 5596

权利要求:
Claims (14)
[1]
Conclusions
Method for optimizing the operation of at least one of several hot water systems in several buildings, the method comprising:
- collecting sensor data from the multiple hot water systems in the multiple buildings via a communication network, the sensor data collected comprising multiple sets of at least one hot water system data record and a corresponding hot water system type;
- estimate a behavior for each of the hot water system types of the multiple hot water systems based on the sensor data collected;
- optimizing the operation of at least one hot water system based on the estimated behavior of the corresponding type of hot water system.
[2]
The method of claim 1, wherein the hot water system includes at least one hot water vessel and a heating mechanism, and wherein the hot water system type includes a corresponding at least one hot water vessel and heating mechanism type.
[3]
The method of claim 2, wherein the step of estimating a behavior for each of the hot water system types includes estimating a behavior for each of the hot water vessel types and each of the heating mechanism types.
[4]
A method according to any preceding claim, wherein the method includes a training phase and an operational phase, wherein at least in the training phase, said at least one hot water system data record includes at least one of the following:
- a water flow data record;
- an energy flow data record; and
- a water temperature data record.
[5]
The method of claim 4, wherein in the operational phase said at least one hot water system data record includes at least one energy flow data record.
[6]
The method of claim 4 or 5, wherein the plural sets of at least one hot water system data record and a corresponding hot water system type further include a time indication that provides a time-related indication of relevance of the hot water system data record.
[7]
The method of any preceding claim, wherein the estimating step includes:
- setting up a generalizable function approach to learn the hot water system behavior with which a hot water system model is formed for each of the hot water system types;
BE2017 / 5596
- aggregating the data collected from the multiple buildings into a form suitable for the job approximator;
- supplying the collected sensor data to the corresponding hot water system model; and
- adapting the adaptive hot water system model to learn the behavior of the corresponding hot water system using machine learning algorithms.
[8]
The method of claim 7, wherein improving energy consumption includes at least one of the following:
- reducing general energy consumption;
- reducing energy consumption peaks;
- reducing general energy costs;
- responding to energy grid imbalances; and
- matching energy availability with energy consumption.
[9]
A method according to claim 7 or 8, wherein the hot water supplies are determined for each of the multiple hot water systems from the sensor data collected, at least one of an amount as a function of time and / or a temperature as a function of time of the hot designate water consumed for each of the hot water systems.
[10]
The method of claim 7, 8 or 9, the method further comprising updating the hot water system model based on a combination of sensor data collected and estimation of the model's uncertainty.
[11]
The method according to any of the preceding claims, wherein the step of optimizing the operation comprises balancing the energy input to the hot water system over time to improve energy consumption and to supply the hot water supplies.
[12]
The method of any preceding claim, wherein the operation optimization step is based on a combination of corresponding sensor data collected and the estimated behavior.
[13]
A data storage device containing a program in machine-readable and machine-executable form to perform the method steps of any of the preceding claims 1-12.
[14]
A computer program in a machine-readable and machine-executable form for performing the method steps of any of the preceding claims 1-12.
BE2017 / 5596
BE2017 / 5596
BE2017 / 5596
BE2017 / 5596
BE2017 / 5596
Optimizing the operation of hot water systems

类似技术:

公开号 | 公开日 | 专利标题

Vázquez-Canteli et al.2019|Reinforcement learning for demand response: A review of algorithms and modeling techniques

Reynolds et al.2018|A zone-level, building energy optimisation combining an artificial neural network, a genetic algorithm, and model predictive control

Reynders et al.2018|Energy flexible buildings: An evaluation of definitions and quantification methodologies applied to thermal storage

Corradi et al.2012|Controlling electricity consumption by forecasting its response to varying prices

Lindberg et al.2016|Methodology for optimal energy system design of Zero Energy Buildings using mixed-integer linear programming

Pallonetto et al.2020|On the assessment and control optimisation of demand response programs in residential buildings

Scott et al.2013|Residential demand response under uncertainty

Molina et al.2013|Model predictive and genetic algorithm-based optimization of residential temperature control in the presence of time-varying electricity prices

JP6671344B2|2020-03-25|Method and apparatus for operating a smart system for optimizing power consumption

Finck et al.2019|Economic model predictive control for demand flexibility of a residential building

Gholamzadehmir et al.2020|Adaptive-predictive control strategy for HVAC systems in smart buildings–A review

Dorini et al.2013|Chance-constrained optimization of demand response to price signals

Schachter et al.2014|A short-term load forecasting model for demand response applications

Balint et al.2019|Determinants of energy flexibility in residential hot water systems

Oprea et al.2018|Flattening the electricity consumption peak and reducing the electricity payment for residential consumers in the context of smart grid by means of shifting optimization algorithm

Finck et al.2018|Review of applied and tested control possibilities for energy flexibility in buildings

Lissa et al.2021|Deep reinforcement learning for home energy management system control

De Lorenzi et al.2020|Setup and testing of smart controllers for small-scale district heating networks: An integrated framework

Becker et al.2015|Building energy management in the FZI house of living labs

Golmohamadi et al.2021|Hierarchical flexibility potentials of residential buildings with responsive heat pumps: A case study of Denmark

BE1025140B1|2018-11-20|OPTIMIZING THE FUNCTIONING OF HOT WATER SYSTEMS

Parisio et al.2018|Energy management systems for intelligent buildings in smart grids

Vivekananthan et al.2014|Energy efficient home with price sensitive stochastically programmable TCAs

Darwazeh et al.2022|Review of peak load management strategies in commercial buildings

Marszal-Pomianowska et al.2020|Principles of Energy Flexible Buildings: Energy in Buildings and Communities Programme

同族专利:

公开号 | 公开日

BE1025140A1|2018-11-12|

EP3291030A1|2018-03-07|

引用文献:

公开号 | 申请日 | 公开日 | 申请人 | 专利标题

US20160216007A1|2008-07-01|2016-07-28|Carina Technology, Inc.|Water Heater Demand Side Management System|

US20120232701A1|2011-03-07|2012-09-13|Raphael Carty|Systems and methods for optimizing energy and resource management for building systems|

US10451294B2|2014-07-14|2019-10-22|Santa Clara University|Machine learning based smart water heater controller using wireless sensor networks|

法律状态:
2018-12-13| FG| Patent granted|Effective date: 20181120 |

2020-05-29| MM| Lapsed because of non-payment of the annual fee|Effective date: 20190831 |

优先权:

申请号 | 申请日 | 专利标题

EP16186776.7|2016-09-01|

EP16186776.7A|EP3291030A1|2016-09-01|2016-09-01|Optimizing operation of hot water systems|

[返回顶部]