• 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 processing calibration data of different origin comprises reading the calibration data consisting of a description file and a data part and carrying out a basic statistical analysis. Subsequently, a conversion and compression of the data portion of the calibration data into a uniform format and the analysis of the converted and compressed calibration data according to certain additional criteria.
    公开号:AT511286A2
    申请号:T50225/2012
    申请日:2012-06-08
    公开日:2012-10-15
    发明作者:
    申请人:Avl List Gmbh；
    IPC主号:
    专利说明:

    Printed: 11-06 * 2012 E014.1
    102012/50225 AV-3485 AT
    Method for processing calibration data
    The invention relates to a method for processing calibration data of different origin, comprising reading out the calibration data consisting of a description file and a data part and performing a statistical analysis.
    The test and test runs of vehicles or their subsystems produce an enormous amount of calibration data, but at the moment they are stored completely independently of each other and are not related to each other. Therefore, knowledge stored indirectly in these historical data sources can not be used.
    Examples of so-called "Data Mining" for the evaluation of historical data are for example the extraction / ranking in search by search engines, the classification in Spamfiltem, the shopping cart analysis in online department stores, but also the correction of damage to auto parts.
    The problem of "Data Mining" in the area of the calibration data the following example becomes clear:
    A data base contains between 15,000 and 30,000 different labels (parameters). There are about 100 data states per project for each vehicle / engine / transmission variant (specification). If the time dependency is also taken into account, this results in a number of approx. 300 million labels (30,000 labels * 100 variants * 50 weeks * 2 years) for a single project that runs for two years.
    Conventional comparisons or evaluations fail because of the large amount of data, the lack of visualization and the powerful comparison. The currently available algorithms (> 200) for " Data mining " Failure on characteristic curves (2D) or characteristic diagrams (3D), which form an essential part of calibration data.
    The object of the present invention was therefore to extract from calibration data of any origin and in any format technically relevant knowledge and provide for further use.
    To solve this problem, the method described above is improved in that following the first analysis, a step of converting and compressing the data portion of the calibration data into a uniform format and the analysis of the converted and compressed calibration data according to certain additional criteria. In this way, any possible correlation between labels of the calibration data can advantageously be indicated, it is possible to perform a preliminary assessment of new projects on the basis of historical information, a clustering or the grouping of 08-06-2012 1
    AV-3485 AT
    Labels can be made (label A always has the same values in variants A, B, C), there is the possibility for time series analyzes (statements about date, effort and quality of a data status), for plausibility check and / or error detection (is the value of a label plausible in connection with historical data statuses) as well as a calibration process optimization by uncovering weak points in the process can be carried out.
    In order to make the label data comparable, according to an advantageous variant of the method, the data belonging to the same parameters are provided with uniform axes and the same number of supporting points. Thus - and in further improved embodiments as explained below - the comparison can be carried out regardless of the type / unit / axis of the labels of the calibration data less expensive and faster.
    Preferably, the data is normalized with respect to the axes.
    A further advantageous embodiment variant provides that the minimum of the respective, preferably normalized, data is shifted to the zero line.
    In this case, a further simplification and thus faster processing can be achieved by scaling the data to a range of preferably zero to 2 high 32-2.
    The further optional feature that the data is converted into a base 64 number system results in a saving of required storage space.
    A further advantageous embodiment of the invention is characterized in that the normalized data are stored with a checksum, whereby a faster comparison is possible.
    Another optional option is to take into account the maturity of the data during extraction. Thus, both a data reduction and the increase in the quality of the data can be achieved.
    In the following description, the invention will be explained in more detail with reference to the accompanying drawings.
    1 a to 1 c show exemplary one-, two- and three-dimensional calibration data, and FIGS. 2 a and 2 b show two diagrams for clarifying statements about the quality of calibration data due to the changes over time.
    From a large amount of calibration data, which as shown by way of example in FIGS. 1 a to 1 c can be present in different dimensionality, information should be obtained in the simplest possible way and used for current data states. A conventional comparison is often not possible because it would be very time consuming 2
    Printed: 11-06-2012 E014.1
    10 2012/50225 AV-3485 AT and the calibration data may also differ in terms of dimension, axis values, number of nodes and units. Therefore, the calibration data consisting of a description file and a data part are only read out of the database in a first step, read into the system according to the invention and a basic statistical analysis is performed. This includes a minimum / maximum analysis, the determination of a "maturity level". using certain markers, etc.
    Subsequently, in a next step, a conversion and compression of the data part of the calibration data is performed in a uniform format. Thereafter, the same labels are provided with uniform axes and the same number of nodes.
    The start or end value for the axes results from the minimum or maximum axis values of all the same labels. The number of points of use can advantageously be fixed statically to a specific value, for example 25. In the next step, the label values are then normalized to match the new axes. This processing now makes it possible for the first time to perform a perfect comparison that is independent of the original axes, types and / or units.
    Since one usually has to deal with an enormous amount of data, according to an advantageous continuation of the invention, these data are further processed prior to insertion into the database in order to save storage space. First, the values are shifted so that the minimum is exactly at the zero point and you only have to work with positive values. Finally, there is a scaling, so that only more integers from 0 to 2 are high 32-2. Preferably, a checksum (e.g., CRC32) is also created over the normalized and scaled values of each label. This unique checksum allows two maps (3D) to be checked for equality with a single command, and also allows values to be clumped.
    It can be provided as a further step, to convert these integers into a separate number system, will be described in more detail below. It should be mentioned earlier that there are some characteristic curves (2D) or characteristic diagrams (3D) in the label data, which always have the same value. This would be a straight line or a flat surface.
    In this case, only a single z-value is stored, as it always repeats, which avoids unnecessary memory consumption.
    The conversion of the data into a number system with a larger number of characters than the decimal system, preferably of 64 characters, also contributes to less storage space requirement. Such number systems consume in comparison to the decimal system 08-06-2012 3
    AV-3485 AT less memory because large numbers can be displayed with fewer characters. Preferably, the base 64 is chosen because the conversion between the decimal system and a base-2 Zahl number system can be dunched with fast bit-shifting operations (see Table 1).
    The converted and compressed data is preferably stored in a "data warehouse". stored centrally and enable the rapid generation of complex analyzes. An example of this is explained in FIGS. 2a and 2b. In FIG. 2a the frequency and / or extent of changes in the calibration data is plotted over time. It can be seen that after a start-up phase of the respective project there are major changes in the early stages, until at the end of the project only minor adjustments, the "fine tuning", take place. If, therefore, an analysis of calibration data results in a diagram as in FIG. 2b which shows an exactly opposite course of change, this can be used to draw conclusions about poor project execution and / or incorrect data.
    Table 1:
    Value . ....... character value | Characters Werl Zöchen Value ... -: Characters j 0 0 16: G 32 W 48 .... Ί m Ι 1 1 17 i H 33 ........... 1 ...... j X 49 2 2 18 'I 34 1 ............. ... 50 i Ο i 3 3 19: 3 35 .......... .......... j Λ ....... 51 P 4 j 4 20; κ 36 a 52 ί q 1 5 5 21 i L ... 37 Tb 53 Γ r I 6 6 22 i M 38 ..... ic 54 s 1 7 7 23 39 d 55 t ί 8 8 24; o 40 e 56 ru______________i 9 9 25 p 41 ........... J J .................... 57 viv ................. j 10 A 26: Q 42 .. ....... i9 ...... -.... 58 -i --.....-: ΪΪ IB 27: R 43 h 59 txi 12! C 28! s 44 i 60 jy ........... I 13 r ............................... D 29! T 45 ........! I .................. 61, ................ 1 14 ... 1E .......... 30; u 46 k 62 · < ! 15 f .._. 31 V 47. ≫ .......................... 63 LI ......................!
    To reduce the amount of data and to improve data quality, the maturity of each label is used. The maturity level is specified for each label and is between 0% (initial) and 100% (finished). When creating the above-mentioned "Data Warehouse" a kind of filtering can be done in which only those labels 4
    Printed: 11-06-2012 E014.1
    10 2012/50225 AV-3485 AT, which has a degree of maturity > own n%. Also, a weighting can be done by means of the degree of ripeness. For example, when creating a report or an evaluation, those labels with a higher degree of maturity are preferred. For the recognition of outliers as well as for the representation of the distribution function (see figure: normal distribution curve) those labels with a score of 0% are not considered. 08-06-2012 5
    权利要求:
    Claims (8)
    [1]
    AV-3485 AT Claims: 1. A method for processing calibration data of different origin, comprising reading the calibration data consisting of a description file and a data part and performing a basic statistical analysis, characterized by the subsequent step of converting and compressing the data part of the calibration data in a uniform format and the analysis of the converted and compressed calibration data according to certain additional criteria.
    [2]
    2. The method according to claim 1, characterized in that the data associated with the same parameters data are provided with uniform axes and gle'cher number of support points.
    [3]
    3. The method according to claim 2, characterized in that the data are normalized with respect to the axes.
    [4]
    4. The method according to claim 2 or 3, characterized in that the minimum of the respective, preferably normalized data is shifted to the zero line.
    [5]
    A method according to claim 4, characterized in that the data is scaled to a range of preferably zero to 2 up to 32-2.
    [6]
    6. The method according to claim 5, characterized in that the data is converted into a number system to the base 64.
    [7]
    7. The method according to any one of claims 3 to 6, characterized in that the normalized data are stored with a checksum.
    [8]
    8. The method according to any one of claims 1 to 7, characterized in that the degree of maturity of the data is taken into account in the extraction. 6
    类似技术:
    公开号 | 公开日 | 专利标题
    DE60204005T2|2006-06-01|METHOD AND DEVICE FOR RECOGNIZING A HANDWRITTEN PATTERN
    DE102011003155A1|2011-08-04|Map data, map data generation method, map data storage medium and navigation device
    DE10357780A1|2004-07-01|Electronic device
    WO2004102291A1|2004-11-25|Configuration of tools and processes for metal forming
    DE102007040070B4|2017-03-09|Image density conversion method, image enhancement processing device and program therefor
    Broda2005|Marketing-Praxis
    DE69433620T2|2004-08-12|Data compression method
    AT511286A2|2012-10-15|Method for processing calibration data
    DE60037497T2|2008-12-04|METHOD AND DEVICE FOR DISPLAYING OR SELECTING AN OBJECT IN AN IMAGE OR A COMPUTER READABLE RECORDING MEDIUM
    DE112019003861T5|2021-05-27|MOLD PACKING TECHNOLOGY
    EP2602591B1|2016-04-20|Method for simplified storage of data representing shapes
    DE102020001541A1|2020-10-01|Method for transforming acquired sensor data from a first data domain into a second data domain
    AT15663U1|2018-04-15|Method for processing calibration data
    Hildebrandt et al.2016|CSR und Energiewirtschaft
    WO2020187394A1|2020-09-24|Method for training an autoencoder and for classifying data, and autoencoder and computer program for this purpose
    DE102014207870A1|2015-10-29|Method for the automated creation of a data record characterizing two technical drawings
    DE10240940A1|2004-03-18|Computer system and method for comparing data sets of three-dimensional bodies
    DE102013207066A1|2014-10-23|Method for storing altitude profile data, as well as profile data and method for generating a height profile from these profile data
    Šedivý2019|Digitale Erschließung der | Quellen in der Slowakei
    DE102019219924B4|2022-02-17|Method and device for creating and providing a database with sensor data patches stored therein for use in quilting
    DE102018129138A1|2020-05-20|Method and system for determining a pair of table columns for linking
    DE102016216218A1|2018-03-01|Method for determining surface areas of water from OSM data
    EP3764273A1|2021-01-13|System and method for identifying a pallet
    DE102014220546A1|2016-04-14|Apparatus and method for encoding time series
    DE102015014199A1|2016-06-02|Method and system for improving the object recognition performance of a 3D recognition unit
    同族专利:
    公开号 | 公开日
    EP2859415A1|2015-04-15|
    KR101710286B1|2017-02-24|
    AT511286B1|2018-07-15|
    JP2015525401A|2015-09-03|
    CN104380214A|2015-02-25|
    US20150154817A1|2015-06-04|
    IN2014KN02797A|2015-05-08|
    AT511286A3|2018-05-15|
    CN104380214B|2017-12-08|
    EP2859415B1|2017-03-22|
    WO2013182681A1|2013-12-12|
    KR20150018631A|2015-02-23|
    US9489780B2|2016-11-08|
    引用文献:
    公开号 | 申请日 | 公开日 | 申请人 | 专利标题
    DE3115692A1|1981-04-18|1982-11-11|Klöckner-Humboldt-Deutz AG, 5000 Köln|ADAPTIVE RULING PROCEDURE|
    JP2715431B2|1988-03-14|1998-02-18|日本電気株式会社|Command judgment method|
    JP2005267025A|2004-03-17|2005-09-29|Tsutomu Ukyo|Area extraction system for analysis model, method, program and program medium|
    US7348988B2|2005-05-06|2008-03-25|Via Technologies, Inc.|Texture cache control using an adaptive missing data table in a multiple cache computer graphics environment|
    US7526405B2|2005-10-14|2009-04-28|Fisher-Rosemount Systems, Inc.|Statistical signatures used with multivariate statistical analysis for fault detection and isolation and abnormal condition prevention in a process|
    US7941260B2|2006-05-09|2011-05-10|GM Global Technology Operations LLC|Rapid engine mapping and modeling|
    US20070271229A1|2006-05-17|2007-11-22|Jauder Ho|System and method for data searching among multiple enterprise applications|
    US7720553B2|2006-08-31|2010-05-18|Dell Products L.P.|Service metric portfolio management|
    US7984362B2|2006-12-22|2011-07-19|Detroit Diesel Corporation|Method for static fault code synchronization in an internal combustion engine|
    CN101639896A|2009-05-19|2010-02-03|上海闻泰电子科技有限公司|Data filtering and smoothing method applied to touch screen|DE102015014478A1|2015-11-10|2017-05-11|Avl List Gmbh|System and method for calibrating a vehicle component|
    CN111145380A|2019-12-31|2020-05-12|上海能塔智能科技有限公司|Reported data processing method and device adaptive to vehicle-mounted equipment and electronic equipment|
    法律状态:
    优先权:
    申请号 | 申请日 | 专利标题
    ATA50225/2012A|AT511286B1|2012-06-08|2012-06-08|Method for processing calibration data|ATA50225/2012A| AT511286B1|2012-06-08|2012-06-08|Method for processing calibration data|
    CN201380030010.3A| CN104380214B|2012-06-08|2013-06-07|method for processing data|
    US14/405,637| US9489780B2|2012-06-08|2013-06-07|Method for processing data|
    KR1020157000422A| KR101710286B1|2012-06-08|2013-06-07|Method for processing data|
    JP2015515537A| JP2015525401A|2012-06-08|2013-06-07|Data processing method|
    PCT/EP2013/061796| WO2013182681A1|2012-06-08|2013-06-07|Method for processing data|
    IN2797KON2014| IN2014KN02797A|2012-06-08|2013-06-07|
    EP13728163.0A| EP2859415B1|2012-06-08|2013-06-07|Method for processing data|
    [返回顶部]