Real-time battery health monitoring system based on electrochemical sensor

Abstract

The present application provides a real-time battery health monitoring system based on an electrochemical sensor, which is applied in the field of data processing. The system includes: an electrochemical sensor for obtaining multiple dimensional parameters of the battery within a preset time length and sending them to a monitoring and analysis module; a monitoring and analysis module for determining the correlation between the multiple dimensional parameters; and based on the correlation, determining the degree of abnormality of the multiple dimensional parameters at different times within the preset time length; a state determination module for determining the working state of the battery based on the degree of abnormality of the multiple dimensional parameters at different times within the preset time length. The system processes the multiple dimensional parameters by extracting the correlation information between the data dimensions to improve the effect of principal component analysis; and can use the principal components during reconstruction to obtain the degree of abnormality based on time, reducing the error of single-dimensional parameters in evaluating the health status of the battery.

Description

Technical Field

[0001] The present application relates to the field of data processing, and more specifically, to a real-time battery health monitoring system based on an electrochemical sensor in the field of data processing. Background Art

[0002] An electrochemical sensor is a sensor that uses electrochemical principles to detect and measure the concentration of chemical substances. It is usually composed of electrodes and electrolytes. Electrochemical sensors are sensitive to target molecules or ions and can monitor the concentration changes of target substances in real time. They are suitable for real-time monitoring and control of equipment. For example, by collecting battery voltage, current, temperature and other parameters to analyze the performance of the battery, the stability, life and other health conditions of the battery can be evaluated, and problems can be found in time and corresponding measures can be taken.

[0003] When using electrochemical sensors to monitor the health of batteries, related technologies usually collect and analyze various parameters of the battery separately, and use anomaly detection algorithms to detect whether there are abnormal parameters to determine the operating status of the battery. However, the operating status of the battery is affected by many factors, and its various parameters will change significantly during operation. Therefore, the abnormal detection of single-dimensional battery parameters is not only prone to errors, but it is also difficult to directly determine the abnormal state of the battery through the abnormality of a certain dimension. Summary of the invention

[0004] The present application provides a real-time battery health monitoring system based on electrochemical sensors. The system can better extract the correlation information between data dimensions and improve the effect of principal component analysis; and can use the principal components during reconstruction to obtain the degree of abnormality based on time, thereby reducing the error of single-dimensional parameters in evaluating the battery health status.

[0005] The present application embodiment provides a real-time battery health monitoring system based on an electrochemical sensor, the system comprising:

[0006] Electrochemical sensor, used to obtain multiple dimensional parameters of the battery within a preset time and send them to the monitoring and analysis module;

[0007] A monitoring and analysis module, used to determine the correlation between the multiple dimensional parameters; and based on the correlation, determine the abnormality degree of the multiple dimensional parameters at different times within a preset time period;

[0008] The state determination module is used to determine the working state of the battery based on the abnormality degree of the multiple dimensional parameters at different times within a preset time period.

[0009] In the above scheme, the multi-dimensional parameters of the battery within a preset time are collected by an electrochemical sensor and sent to the monitoring and analysis module, so that the monitoring and analysis module determines the correlation between the multi-dimensional parameters; and based on the correlation, the abnormality degree of the multi-dimensional parameters at different times is determined; in this way, through the correlation between each dimensional parameter, the correlation information between the data dimensions can be better extracted, the effect of the principal component analysis can be improved, and the determined abnormality degree can be made more accurate. Finally, the working state of the battery is determined based on the abnormality degree of the multi-dimensional parameters at different times through the state determination module; in this way, since the correlation between each dimensional parameter is introduced in the process of determining the abnormality degree at different times, the accuracy of analyzing the abnormality degree is improved, thereby improving the accuracy of monitoring the working state of the battery.

[0010] In the above system, the monitoring and analysis module includes:

[0011] A first determination submodule, configured to determine a basic abnormality degree of each dimensional parameter based on time series variation information corresponding to the plurality of dimensional parameters;

[0012] A first clustering submodule is used to perform similarity clustering on the multiple dimensional parameters based on the basic abnormality degree of each dimensional parameter to obtain the correlation between the multiple dimensional parameters;

[0013] The reconstruction submodule is used to reconstruct the multiple dimensional parameters based on the correlation between the multiple dimensional parameters to determine the abnormality degree of the multiple dimensional parameters at different times.

[0014] In the above system, the first determining submodule includes:

[0015] A construction unit, configured to construct a neighborhood of the reference data with the reference data in the multiple dimensional parameters as the center and a preset number of data as the radius; wherein the parameters in the neighborhood are continuous with the reference data in time sequence; and the reference data is any dimensional parameter in the multiple dimensional parameters;

[0016] A first determining unit, configured to determine a severity of a short-term change of the reference data based on the neighborhood and the plurality of dimensional parameters;

[0017] A second determining unit, configured to determine a degree of deviation of the reference data based on the reference data and an overall mean of the plurality of dimensional parameters;

[0018] The first fusion unit is used to fuse the short-term change severity and the deviation degree to obtain the basic abnormality degree of the reference data.

[0019] In the above system, the first determining unit includes:

[0020] A first determining subunit, configured to determine a first-order difference of each parameter in the neighborhood and a first-order difference of each dimensional parameter in the multiple dimensional parameters;

[0021] A second determination subunit is used to perform averaging processing on the first-order differences of multiple parameters in the neighborhood to obtain an average change trend of the neighborhood;

[0022] A third determining subunit, configured to determine an overall average change trend of the multiple dimensional parameters based on the first-order differences of the multiple dimensional parameters;

[0023] The fourth determining subunit is used to determine the difference between the average change trend of the neighborhood and the overall average change trend as the severity of the short-term change.

[0024] In the above system, the second determining unit includes:

[0025] a fifth determining subunit, configured to determine a reference difference between the reference data and the overall mean;

[0026] a sixth determining subunit, configured to determine differences between the plurality of dimensional parameters and the overall mean, to obtain a plurality of differences;

[0027] The seventh determining subunit is used to determine the ratio of the reference difference to the maximum difference among the multiple differences as the deviation degree of the reference data.

[0028] In the above system, the first clustering submodule includes:

[0029] A third determining unit, configured to determine a temporal distance between each parameter in a neighborhood of the reference data and the reference data;

[0030] A second fusion unit is used to fuse the difference between each parameter in the neighborhood and the reference data, and the time series distance to obtain change information of the reference data in the neighborhood;

[0031] a fourth determining unit, configured to determine a degree of stability of a change of the reference data within the neighborhood based on a second-order difference of the reference data within the neighborhood;

[0032] A third fusion unit is used to fuse the change information of the reference data in the neighborhood and the change stability to obtain the fluctuation degree of the reference data;

[0033] a fifth determining unit, configured to determine, for different dimensional parameters among the plurality of dimensional parameters, similarities between the different dimensional parameters based on time differences, fluctuation degree differences, and basic abnormality degree differences between the different dimensional parameters;

[0034] The clustering unit is used to determine the correlation between the different dimensional parameters based on the similarity between the different dimensional parameters.

[0035] In the above system, the clustering unit includes:

[0036] A clustering subunit, used for clustering the similarities between the parameters of different dimensions to obtain a plurality of clusters;

[0037] An eighth determination subunit is used to respectively determine the data proportion of the first dimension parameter and the second dimension parameter in the same cluster; wherein the first dimension parameter and the second dimension parameter are any two different dimension parameters in the multiple dimensions;

[0038] A ninth determining subunit, configured to determine, based on the data proportion, a degree of proximity between the first dimensional parameter and the second dimensional parameter in the plurality of clusters;

[0039] The first fusion subunit is used to fuse the data proportion and the degree of proximity of the data proportion to obtain the correlation between the first dimensional parameter and the second dimensional parameter.

[0040] In the above system, the reconstruction submodule includes:

[0041] An extraction unit, used for performing principal component extraction on the multiple dimensional parameters to obtain multiple principal components;

[0042] a sixth determining unit, configured to determine, for an mth principal component among the plurality of principal components, a contribution rate of the mth principal component in the dimensional parameter to which it belongs, and a factor loading amount of the mth principal component in the plurality of dimensional parameters; wherein m is an integer greater than or equal to 1 and less than or equal to the number of the plurality of principal components;

[0043] a seventh determining unit, configured to determine the degree of preference of the mth principal component based on the contribution rate, the factor loading amount and the correlation between the multiple dimension parameters;

[0044] A selection unit, configured to select, according to the preference degree of each principal component, at least one target principal component whose cumulative contribution rate reaches a preset threshold value from among the plurality of principal components;

[0045] A reconstruction unit, configured to reconstruct the at least one target principal component to obtain a reconstruction error of the at least one target principal component at the different moments;

[0046] The fourth fusion unit is used to fuse the reconstruction error and basic abnormality degree of the at least one target principal component at the different moments to obtain the abnormality degree at the different moments.

[0047] In the above system, the seventh determining unit includes:

[0048] a tenth determination subunit, used to determine the load difference between the factor loads corresponding to the mth principal component in each of the two dimensional parameters;

[0049] An eleventh determination subunit is used to determine the ratio between the correlation between each two dimensional parameters and the load difference;

[0050] The second fusion subunit is used to fuse the contribution rate and the proportion to obtain the preferred degree of the mth principal component.

[0051] In the above system, the selection unit includes:

[0052] A sorting subunit, used for sorting the plurality of principal components according to the preference degree of each principal component to obtain sorted principal components;

[0053] The selection subunit is used to continuously select at least one target principal component whose cumulative contribution rate reaches the preset threshold value from the sorted principal components, starting with the principal component with the greatest degree of preference. BRIEF DESCRIPTION OF THE DRAWINGS

[0054] Figure 1 It is a schematic diagram of an implementation environment provided by an embodiment of the present application;

[0055] Figure 2 It is a schematic diagram of the composition structure of a real-time battery health monitoring system based on an electrochemical sensor provided in an embodiment of the present application;

[0056] Figure 3 is another structural diagram of a real-time battery health monitoring system based on an electrochemical sensor provided in an embodiment of the present application;

[0057] Figure 4 is a flow chart of implementing real-time battery health monitoring based on an electrochemical sensor provided in an embodiment of the present application;

[0058] Figure 5 is another flow chart of implementing real-time battery health monitoring based on electrochemical sensors provided in an embodiment of the present application;

[0059] Figure 6 It is a structural diagram of a computer device provided in an embodiment of the present application. DETAILED DESCRIPTION

[0060] The technical solution in the present application will be described clearly and in detail below in conjunction with the accompanying drawings. In the description of the embodiments of the present application, unless otherwise specified, " / " means or, for example, A / B can mean A or B: "and / or" in the text is only a description of the association relationship of associated objects, indicating that there can be three relationships, for example, A and / or B can mean: A exists alone, A and B exist at the same time, and B exists alone. In addition, in the description of the embodiments of the present application, "multiple" means two or more than two.

[0061] In the following, the terms "first" and "second" are used for descriptive purposes only and are not to be understood as suggesting or implying relative importance or implicitly indicating the number of technical features indicated. Thus, a feature defined as "first" or "second" may explicitly or implicitly include one or more of the features.

[0062] The implementation environment of the embodiment of the present application is introduced as follows. Figure 1 As shown, the implementation environment of the real-time battery health monitoring system based on electrochemical sensors provided in the embodiment of the present application includes a data acquisition terminal 101, a monitoring and analysis system 102, and a battery status feedback terminal 103.

[0063] The data acquisition terminal 101 and the battery status feedback terminal 103 are connected via a wireless network, and the battery status feedback terminal 103 is connected to the monitoring and analysis system 102 via a wireless or wired network. The data acquisition terminal 101 can be any type of terminal, such as a data acquisition sensor. The data acquisition terminal 101 collects multiple dimensional parameters of the battery and sends them to the monitoring and analysis system 102; the monitoring and analysis system 102 can better extract the correlation between the data dimensions through the correlation between each dimension, thereby improving the effect of the principal component analysis, and then consider the correlation between each dimension in the process of determining the abnormality degree at different times, so that the determined abnormality degree is more accurate; and then the battery status feedback terminal 103 is used to more accurately feedback the working status of the battery according to the abnormality degree of multiple dimensional parameters at different times.

[0064] The monitoring and analysis system 102 is an independent physical server, or a server cluster or distributed system composed of multiple physical servers, or a cloud server that provides basic cloud computing services such as cloud services, cloud databases, cloud computing, cloud functions, cloud storage, network services, cloud communications, middleware services, domain name services, security services, distribution networks (Content Delivery Network, CDN), as well as big data and artificial intelligence platforms.

[0065] The following is an introduction to the technical solution provided in the embodiment of the present application. The embodiment of the present application provides a real-time battery health monitoring system based on an electrochemical sensor. Figure 2 , Figure 2 200 is a schematic diagram of the composition structure of a real-time battery health monitoring system based on an electrochemical sensor provided in an embodiment of the present application. The system 200 includes:

[0066] The electrochemical sensor 201 is used to obtain multiple dimensional parameters of the battery within a preset time period and send them to the monitoring and analysis module.

[0067] Here, the electrochemical sensor may include: oxygen sensor, pH sensor, ion selective electrode, gas sensor, etc. The electrochemical sensor is correctly and firmly connected to the battery, so that the electrochemical sensor can collect multiple dimensional parameters of the battery in real time and send them to the monitoring and analysis module. Among them, the multiple dimensional parameters include: voltage, current, temperature, internal resistance, capacity, etc. The preset time length can be a custom time length, which indicates that the obtained multi-dimensional parameters are the time length of the same sample. For example, the preset time length is 3 hours, so that the same sample includes 3 hours of multi-dimensional parameters.

[0068] In some possible implementations, the electrochemical sensor can collect multiple dimensional parameters of the battery at a specific frequency, for example, the specific frequency is once per minute. In this way, the electrochemical sensor can collect multiple dimensional parameters of the battery once per minute, and send the three hours of continuous collection of multiple dimensional parameters as the same sample data to the monitoring and analysis module.

[0069] The monitoring and analysis module 202 is used to determine the correlation between the multiple dimensional parameters; and based on the correlation, determine the abnormality degree of the multiple dimensional parameters at different times within a preset time period.

[0070] Here, the monitoring and analysis module 202 can be implemented by a monitoring and analysis system. After the monitoring and analysis system receives the multiple dimensional parameters sent by the electrochemical sensor, the basic abnormality degree of the multiple dimensional parameters is determined by the time series changes of the multiple dimensional parameters within a preset time length, and similarity clustering is performed based on the basic abnormality degree to analyze the correlation between the multiple dimensional parameters. In this way, the correlation between the multiple dimensional parameters can reflect the strength of the association between the different dimensional parameters, so as to facilitate the selection of the principal component with a larger proportion of parameters with strong correlation. Afterwards, based on the correlation between the parameters, the correlation between the parameters is used for reconstruction to obtain the abnormality degree of the multiple dimensional parameters at each moment. In this way, the abnormality degree can characterize the health state of the battery. That is, the greater the abnormality degree, the more likely it is that there is an abnormality in the current operating state of the battery; obviously, the smaller the abnormality degree, the more normal the current operating state of the battery.

[0071] The state determination module 203 is used to determine the working state of the battery based on the abnormality degree of the multiple dimensional parameters at different times within a preset time period.

[0072] Here, after obtaining the abnormality degree of multiple dimensional parameters at each moment, the overall abnormality degree is obtained by summing the abnormality degrees at multiple moments. The overall abnormality degree is compared with the preset abnormality degree threshold to determine the working state of the battery. The preset abnormality degree threshold can be a custom setting. For example, the preset abnormality degree threshold is set to 0.8. When the overall abnormality degree is greater than or equal to 0.8, the working state of the battery is determined to be an abnormal state, and a prompt message is returned to prompt the battery to be troubleshooted. When the overall abnormality degree is less than 0.8, the working state of the battery is determined to be normal.

[0073] In the embodiment of the present application, the multi-dimensional parameters of the battery within a preset time are collected by an electrochemical sensor and sent to the monitoring and analysis module; the monitoring and analysis module can better extract the correlation information between the data dimensions through the correlation between each dimensional parameter, improve the effect of principal component analysis, and thus make the determined abnormality more accurate. Finally, the state determination module determines the working state of the battery based on the abnormality of multiple dimensional parameters at different times; in this way, since the correlation between each dimensional parameter is introduced in the process of determining the abnormality at different times, the accuracy of analyzing the abnormality is improved, thereby improving the accuracy of monitoring the working state of the battery.

[0074] In some embodiments, the monitoring and analysis module may be composed of a first determination submodule, a first clustering submodule and a reconstruction submodule, such as Figure 3 As shown, the monitoring and analysis module 202 includes:

[0075] The first determination submodule 301 is used to determine the basic abnormality degree of each dimensional parameter based on the time series change information corresponding to the multiple dimensional parameters.

[0076] Here, the time series change information corresponding to the multiple dimensional parameters is used to represent the time series relationship between the multiple dimensional parameters. The neighborhood of any dimensional parameter is constructed according to the time series change information, and the basic abnormality degree of the dimensional parameter is determined by the data in the neighborhood and the overall data of the multiple dimensional parameters.

[0077] In some possible implementations, the first determining submodule 301 may be implemented by a construction unit, a first determining unit, a second determining unit, and a first fusion unit (not shown in the figure):

[0078] A construction unit is used to construct a neighborhood of the reference data with the reference data in the multiple dimensional parameters as the center and a preset number of data as the radius.

[0079] The parameters in the neighborhood are continuous with the reference data in time sequence; the reference data is any dimension parameter among the multiple dimension parameters. For example, the reference data is The preset number of data is 5, that is, with the reference data as the center, the 5 parameter radii before and after the reference data in time sequence construct the neighborhood of the reference data.

[0080] The first determination unit is used to determine the severity of short-term changes in the reference data based on the neighborhood and the multiple dimensional parameters.

[0081] Here, the dimensional parameters in the neighborhood and the overall multiple dimensional parameters are combined for comprehensive analysis to determine the severity of short-term changes in the reference data. The severity of short-term changes indicates whether the reference data has suddenly changed dramatically, showing obvious mutations or jumps, or abnormal peaks and valleys, etc. It also indicates whether the reference data has a dramatic change in the time series trend, or has a large deviation relative to the overall data.

[0082] In some possible implementations, the first determining unit is implemented by a first determining subunit, a second determining subunit, a third determining subunit, and a fourth determining subunit to determine a severity of a short-term change in the reference data, wherein:

[0083] The first determination subunit is used to determine the first-order difference of each parameter in the neighborhood and the first-order difference of each dimensional parameter in the multiple dimensional parameters.

[0084] Here, the neighborhood is based on the reference data The neighborhood includes 5 parameters before and after the reference data in time series. The first-order differences of the 10 parameters before and after the reference data in the neighborhood and the first-order differences of each dimension are determined.

[0085] The second determining subunit is used to determine the average change trend of the neighborhood based on the first-order differences of multiple parameters in the neighborhood.

[0086] Here, by averaging the first-order differences of multiple parameters in the neighborhood, the average change trend of the neighborhood can be obtained.

[0087] The third determining subunit is used to determine the overall average change trend of the multiple dimensional parameters based on the first-order differences of the multiple dimensional parameters.

[0088] Here, by averaging the first-order differences of multiple dimensional parameters, the overall average change trend can be obtained.

[0089] The fourth determining subunit is used to determine the difference between the average change trend of the neighborhood and the overall average change trend as the severity of the short-term change.

[0090] Here, by first determining the difference between the average change trend of the neighborhood and the overall average change trend, and then taking the norm of the difference, the degree of short-term change intensity can be obtained. In this way, by characterizing the degree of short-term change intensity of the reference data by the part where the average change trend in the neighborhood of the reference data is higher than the overall average change trend, the change of the reference data can be more accurately reflected.

[0091] The second determination unit is used to determine the degree of deviation of the reference data based on the reference data and the overall mean of the multiple dimensional parameters.

[0092] Here, the deviation degree of the reference data is obtained by calculating the difference between the reference data and the overall mean, and then further calculating the ratio between the difference and multiple dimension parameters. The deviation degree of the reference data can characterize the degree to which the difference of the reference data approaches the maximum difference. The greater the deviation degree, the higher the abnormality of the reference data.

[0093] In some possible implementations, the second determining unit is constituted by the fifth determining subunit, the sixth determining subunit, and the seventh determining subunit, wherein:

[0094] The fifth determining subunit is used to determine a reference difference between the reference data and the overall mean.

[0095] Here, the absolute value of the difference between the reference data and the overall mean is taken as the reference difference between the reference data and the overall mean.

[0096] The sixth determination subunit is used to determine the differences between the multiple dimensional parameters and the overall mean to obtain multiple differences.

[0097] Here, multiple differences are obtained by calculating the difference between each dimension parameter and the overall mean.

[0098] The seventh determining subunit is used to determine the ratio of the reference difference to the maximum value of the multiple differences as the deviation degree of the reference data.

[0099] Here, the maximum difference is first selected from multiple differences, and then the ratio between the reference difference and the maximum difference is calculated, and the ratio is used as the deviation degree of the reference data. In this way, the deviation degree of the reference data is analyzed by the difference between the reference data and the overall mean, and the ratio between the target dimension parameters with the largest difference from the overall mean, so that the abnormality of the reference data can be reflected through the deviation degree.

[0100] The first fusion unit is used to fuse the short-term change severity and the deviation degree to obtain the basic abnormality degree of the reference data.

[0101] Here, the short-term change intensity and the deviation degree are multiplied to obtain the multiplication result, which is used as the basic abnormality degree of the reference data. In this way, by constructing the neighborhood of the reference data, the short-term change intensity of the reference data is further analyzed, and the basic abnormality degree of the reference data is determined in combination with the deviation degree of the reference data, which can improve the accuracy of the basic abnormality degree.

[0102] The first clustering submodule 302 is used to perform similarity clustering on the multiple dimensional parameters based on the basic abnormality degree of each dimensional parameter to obtain the correlation between the multiple dimensional parameters.

[0103] Here, after obtaining the basic abnormality degree of each dimensional parameter, multiple dimensional parameters are clustered by adopting the similarity clustering method, and the basic abnormality degree is introduced in the clustering process, so that the dimensional parameters with similar basic abnormality degrees are classified as one category, and then the correlation between multiple dimensional parameters is obtained.

[0104] In some possible implementations, the first clustering submodule 302 may be implemented by a third determining unit, a second fusion unit, a fourth determining unit, a third fusion unit, a fifth determining unit, and a clustering unit, wherein:

[0105] The third determining unit is used to determine the temporal distance between each parameter in the neighborhood of the reference data and the reference data.

[0106] Here, the number of parameters between each parameter in the neighborhood and the reference data is taken as the temporal distance between each parameter and the reference data.

[0107] The second fusion unit is used to fuse the difference between each parameter in the neighborhood and the reference data, as well as the time series distance, to obtain change information of the reference data in the neighborhood.

[0108] Here, the difference between each parameter in the neighborhood and the reference data is first calculated, and the absolute value of the difference is taken. Then, the difference is multiplied by the inverse of the time series distance, and the multiplication results are summed to obtain the change information. In this way, the change information can characterize the change of the reference data in the neighborhood. The larger the value, the greater the fluctuation of the reference data.

[0109] The fourth determining unit is used to determine the stability of the change of the reference data in the neighborhood based on the second-order difference of the reference data in the neighborhood.

[0110] Here, the second-order difference of the parameter data in the neighborhood is first determined, so as to obtain multiple second-order difference values, and the multiple second-order difference values ​​are summed, and the summation result is used as the change stability of the reference data in the neighborhood. In this way, the change stability can represent the change stability of the reference data in the neighborhood. The larger the value, the more unstable the data change and the greater the fluctuation.

[0111] The third fusion unit is used to fuse the change information of the reference data in the neighborhood and the change stability to obtain the fluctuation degree of the reference data.

[0112] Here, the fluctuation degree of the reference data can be obtained by multiplying the change information and the change stability and performing a norm operation on the product.

[0113] The fifth determination unit is used to determine the similarity between different dimensional parameters among the multiple dimensional parameters based on the time difference, fluctuation degree difference and basic abnormality degree difference between the different dimensional parameters.

[0114] Here, for each two different dimensional parameters, the time difference, fluctuation degree difference, and basic abnormality degree difference between the two dimensional parameters are multiplied, and the obtained product is used as an exponent with a base of 2, so that the obtained result is used as the similarity between the two dimensional parameters. In this way, the closer the fluctuation degree, time, and basic abnormality degree between the two dimensional parameters are, the higher their similarity is.

[0115] The clustering unit is used to determine the correlation between the different dimensional parameters based on the similarity between the different dimensional parameters.

[0116] Here, multiple dimensional parameters are clustered according to the phase velocity between different dimensional parameters, thereby obtaining multiple clusters, and each cluster includes at least one dimensional parameter; then the correlation between different dimensional parameters is calculated according to the dimensional parameters included in each cluster. In this way, for different dimensional parameters, by analyzing the differences in the time, fluctuation degree and basic abnormality degree of the different dimensional parameters, the similarity between the different dimensional parameters is obtained, and by performing similarity clustering, the correlation between the different dimensional parameters can be analyzed more accurately.

[0117] In some possible implementations, the clustering unit may be implemented by a clustering subunit, an eighth determining subunit, a ninth determining subunit, and a first fusion subunit, wherein:

[0118] The clustering subunit is used to cluster the similarities between the parameters of different dimensions to obtain multiple clusters.

[0119] Here, after obtaining the similarity between every two dimensional parameters, the multiple dimensional parameters are clustered by similarity clustering, so that the multiple dimensional parameters are divided into multiple classes, that is, multiple clusters are obtained. The parameters in the same cluster are parameters with similar time series and similar change patterns, and the same cluster can contain multiple dimensional parameters. If the proportion of several dimensional parameters in any cluster is large and very close, it means that these dimensional parameters are highly correlated in this part, and the more clusters with similar proportions, the stronger the correlation of these dimensional parameters.

[0120] The eighth determination subunit is used to respectively determine the data proportions of the first dimension parameter and the second dimension parameter in the same cluster.

[0121] Here, the first dimension parameter and the second dimension parameter are any two different dimension parameters in the multiple dimensions. Calculate the proportion of any two dimension parameters in the same cluster. For example, is the total number of data contained in the k-th cluster, , are the data proportions of dimensions A and B in cluster k respectively.

[0122] The ninth determining subunit is used to determine, based on the data proportion, the degree of proximity between the first dimensional parameter and the second dimensional parameter in the multiple clusters.

[0123] Here, we first determine the data proportion of the first dimension parameter and the second dimension parameter in each cluster, then calculate the difference between the two data proportions, multiply the absoluteness of the difference by the minimum variance of the difference, and obtain multiple products; then, we average the multiple products obtained, and use the average as the degree of proximity of the data proportions. The smaller the value, the greater the degree of proximity, and the stronger the correlation between the first dimension parameter and the second dimension parameter.

[0124] The first fusion subunit is used to fuse the data proportion and the degree of proximity of the data proportion to obtain the correlation between the first dimensional parameter and the second dimensional parameter.

[0125] Here, after obtaining the proportion of the two dimensional parameters in all clusters, the smaller proportions of the two dimensional parameters in all clusters are summed, the sum result is multiplied by the degree of proximity of the data proportions, and the product is used as an exponential with base e to obtain the correlation between the two dimensional parameters. In this way, for any two different dimensional parameters, by combining the data proportions of the two dimensional parameters in the same cluster and the degree of proximity of the data proportions in multiple clusters, the correlation between the two dimensional parameters can be analyzed more accurately.

[0126] The reconstruction submodule 303 is used to reconstruct the multiple dimensional parameters based on the correlation between the multiple dimensional parameters to determine the abnormality degree of the multiple dimensional parameters at different times.

[0127] Here, after obtaining the correlation between each two dimensional parameters, the multi-dimensional parameters are reconstructed by principal component analysis (PCA), so as to obtain the abnormal degree of multiple dimensional parameters at each moment. In this way, by combining the correlation between multiple dimensional parameters and reconstructing multiple dimensional parameters, it is possible to better capture the important structures and patterns in the data, thereby improving the accuracy of determining the abnormal degree at different moments.

[0128] In some possible implementations, the reconstruction submodule may be implemented by an extraction unit, a sixth determination unit, a seventh determination unit, a selection unit, a reconstruction unit, and a fourth fusion unit, wherein:

[0129] The extraction unit is used to perform principal component extraction on the multiple dimensional parameters to obtain multiple principal components.

[0130] Here, principal component extraction is performed on each of the multiple dimensional parameters, so that at least one principal component can be extracted from each dimensional parameter. For example, PCA is used to extract principal components from all dimensional parameters of the battery to obtain the principal component , , …, .

[0131] The sixth determination unit is used to determine, for the mth principal component among the multiple principal components, the contribution rate of the mth principal component in the dimensional parameter to which it belongs, and the factor loading of the mth principal component in the multiple dimensional parameters.

[0132] Wherein, m is an integer greater than or equal to 1 and less than or equal to the number of principal components; the contribution rate of the mth principal component in the dimension parameter to which it belongs can be realized by formula (5), that is, by calculating the proportion of the principal component in all components, the contribution rate of the principal component can be obtained.

[0133] The seventh determination unit is used to determine the preference degree of the mth principal component based on the contribution rate, the factor loading amount and the correlation between the multiple dimensional parameters.

[0134] Here, the preference of the mth principal component is obtained by determining the ratio of the correlation between the two dimensions and the factor loading in the mth principal component, as well as the product of the contribution rate and the factor loading. In this way, the higher the preference, the more important the mth principal component is in the dimension parameters.

[0135] In some possible implementations, the seventh determining unit may be implemented by a tenth determining subunit, an eleventh determining subunit, and a twelfth determining subunit, wherein:

[0136] The tenth determination subunit is used to determine the load difference between the factor loads corresponding to the mth principal component in every two dimensional parameters.

[0137] Here, since the factor loading of the mth principal component in each dimension parameter has been obtained, the difference of the factor loadings corresponding to any two dimensional parameters of the mth principal component can be obtained.

[0138] The eleventh determination subunit is used to determine the ratio between the correlation between each two dimensional parameters and the load difference.

[0139] Here, the correlation between each two dimensional parameters is divided by the load difference, and the division results are summed to obtain the ratio.

[0140] The second fusion subunit is used to fuse the contribution rate and the proportion to obtain the preferred degree of the mth principal component.

[0141] Here, the contribution rate is multiplied by the ratio, and the result of the multiplication is used as the priority of the mth principal component. In this way, by combining the correlation between the parameters of different dimensions and the factor loading, it is possible to extract more accurate principal components and obtain the priority of the principal components more accurately.

[0142] The selection unit is used to select at least one target principal component whose cumulative contribution rate reaches a preset threshold value from the plurality of principal components according to the preference degree of each principal component.

[0143] Here, after obtaining the preferred degree of each principal component, at least one target principal component whose cumulative contribution rate reaches the preset threshold the fastest is selected.

[0144] In some possible implementations, the selection unit may be implemented by a sorting subunit and a selection subunit, wherein:

[0145] The sorting subunit is used to sort the multiple principal components according to the preference degree of each principal component to obtain sorted principal components.

[0146] Here, the sorted principal component can be obtained by sorting multiple principal components from large to small in terms of their preference, or from small to large in terms of their preference.

[0147] The selection subunit is used to continuously select at least one target principal component whose cumulative contribution rate reaches the preset threshold value from the sorted principal components, starting with the principal component with the greatest degree of preference.

[0148] Here, if the plurality of principal components are sorted from large to small according to the degree of preference, then at least one of the first few target principal components whose cumulative contribution rate is greater than or equal to the preset threshold is selected. In this way, by sorting the plurality of principal components first, the target principal component whose cumulative contribution rate is greater than the preset threshold can be quickly selected.

[0149] The reconstruction unit is used to reconstruct the at least one target principal component to obtain the reconstruction error of the at least one target principal component at the different moments.

[0150] Here, at least one target principal component can better retain the main information of the original dimension parameter. These target principal components are reconstructed to obtain the reconstruction error of at least one target principal component at the different moments. For example, It represents the reconstruction error of the c-th dimension data at time t. The larger the error, the greater the abnormality at time t.

[0151] The fourth fusion unit is used to fuse the reconstruction error and basic abnormality degree of the at least one target principal component at the different moments to obtain the abnormality degree at the different moments.

[0152] Here, the abnormality degree of multiple dimensional parameters at different times can be obtained by formula (8); that is, the reconstruction error of each target principal component at different times is multiplied by the basic abnormality degree of the dimensional parameter at the corresponding time, and then the average is calculated to obtain the abnormality degree at different times. In this way, through the correlation between different dimensional parameters, the multiple dimensional parameters are reconstructed so that the principal component retains more valuable data information. In this way, the abnormality degree based on time is obtained by using the principal component feature vector at the time of reconstruction, which can improve the accuracy in evaluating the battery health status.

[0153] The embodiment of the present application obtains the corresponding basic abnormality degree according to the time series characteristics of the parameters of each dimension when the battery is working, and then calculates the correlation between the data of each dimension by clustering, and uses the PCA algorithm to screen the principal components according to the correlation and obtain the reconstruction error of the data, and finally obtains the abnormality degree of the battery at each moment to complete the monitoring of the battery health; combined with Figure 4 The steps shown are explained below:

[0154] 401, collect raw data samples of the battery.

[0155] Here, the electrochemical sensor is installed in the battery system so that the connection between the sensor and the battery is correct and stable. The sensor collects various reference data of the battery in real time, including voltage, current, temperature, internal resistance, capacity, etc. The collection frequency is once a minute, and the same sample contains three hours of data. The collected data is then transmitted to the monitoring and analysis system for processing. Among them, the collection frequency and sample size are set empirically, and can be adjusted according to actual needs in specific applications.

[0156] 402, analyzing the degree of abnormal performance of the battery at different times when the battery is working.

[0157] Here, in the process of real-time health monitoring of the battery, it is necessary to monitor the abnormal changes in various battery parameters. Since the parameter variation range of the battery is large during operation and the various parameters will affect each other, the abnormal degree of the parameters in a single dimension is prone to detection errors and cannot directly indicate that there is a problem with the health status of the battery. Therefore, it is necessary to analyze the initial abnormal degree of the parameters in each dimension and combine the correlation of the parameters in multiple dimensions, and use the reconstruction error of PCA to obtain the battery abnormality accurate to the moment, so as to better judge the health status and performance of the battery.

[0158] In some possible implementations, the process of processing the collected parameters of the battery during operation and obtaining the abnormal performance of the battery at each moment can be performed by Figure 5 The steps shown achieve:

[0159] 501, the basic abnormality degree of each dimension parameter is obtained according to the temporal variation characteristics of the data.

[0160] Here, during the normal operation of the battery, various parameters will change within a certain range, and the trend of change is relatively stable. When certain parameters suddenly change drastically, showing obvious mutations or jumps, or abnormal peak and valley values ​​appear, it means that the health of the battery may be abnormal. Therefore, for any parameter of the battery, if the trend of a certain data in the time series changes drastically, or it has a large deviation relative to the overall data, the degree of abnormality of the data is high.

[0161] According to the above logic, any parameter of a certain dimension of the battery is regarded as reference data , set a fixed-size neighborhood for the data, which contains 5 data before and after it in time series, then the basic abnormality of the data As shown in formula (1):

[0162] (1);

[0163] In formula (1), represents the normalization function, that is, Perform normalization. Reference data, Indicates the first order in the neighborhood of the reference data. data, Indicates the first data, is the total number of data in the sample, that is, the total number of data for multiple dimension parameters, , are the first-order differences of the data in the neighborhood and the data in the sample, respectively. The average change trend in the neighborhood of the reference data is higher than the average change trend of the whole sample, which represents the degree of short-term change of the reference data. The larger the value, the higher the abnormality of the reference data. is the mean of all data in the sample, Indicates the difference between the reference data and the mean of the overall data. represents the maximum difference among all the data in the sample, It means that the difference of the reference data approaches the maximum difference, which represents the degree of deviation of the reference data. The larger the ratio is, the greater the degree of deviation is, and the higher the degree of abnormality of the reference data is.

[0164] The above formula can be used to obtain the basic abnormality level of all dimensional parameters in the battery working process, and make a preliminary judgment on the health status of the battery. Then, on this basis, the abnormality level accurate to the moment can be further analyzed.

[0165] 502, the correlation between parameters of different dimensions is obtained through similarity clustering.

[0166] Here, after obtaining the basic abnormality of the data, in order to further obtain the abnormality of the battery at each moment, it is necessary to use the PCA algorithm to extract and reconstruct the principal components of the multidimensional data. In this process, it is also necessary to screen out a part of the principal components that retain more original data information and discard other principal components. However, there is a strong correlation between the various parameters of the battery, and the correlation strength of parameters in different dimensions is different. Therefore, when screening the principal components, we cannot only consider the original data information contained, but also need to try to select principal components with a larger proportion of parameters with strong correlation. For example, the voltage and current of the battery are strongly correlated, but the correlation between voltage and capacity is weak. Therefore, when selecting the principal components, we should try to select the principal components with higher loads and smaller differences in voltage data and current data, so as to better capture the important structures and patterns in the data.

[0167] Since the relationship between various parameters of the battery is relatively complex when it is working, and there are the influence of outliers, the data correlation calculation method in the related technology cannot obtain more accurate results. For example, the Pearson correlation coefficient is difficult to capture the parameters of this nonlinear relationship, and it is easy to cause deviation or distortion of the results due to a small number of outliers. Therefore, the embodiment of the present application uses the similarity clustering method to obtain the correlation between the parameters of each dimension of the battery. By standardizing the data of all dimensions and constructing a unique feature vector, the data with similar features are clustered into the same cluster. Finally, the correlation of parameters of different dimensions is judged according to the distribution characteristics of the parameters of each dimension in the cluster, so as to better capture the complex relationship between the various parameters of the battery.

[0168] Based on this, we first standardize the data of all dimensions by using the standard score (Z-score), which is achieved by subtracting the mean from the data and then dividing it by the standard deviation. Then, we treat all the standardized data as a sample and perform K-means clustering on the data in the sample, with the clustering condition being the similarity between the data. and Similarity between As shown in formula (2):

[0169] (2);

[0170] In formula (2), exp{.} is an exponential function with the natural constant (e) as the base; Indicates reference data the degree of volatility; Data and The degree of fluctuation can be obtained by changing the formula Reference data in Replace with get. and Indicates the time of data, Data and The basic abnormality level, It means data and The closer the fluctuation degree, time and basic abnormality between the two data points, the closer the data points are. and The higher the similarity between them, the lower it is; For reference data, Indicates the first order in the neighborhood of the reference data. data, For data and The temporal distance, It indicates the variation of the reference data within the neighborhood. The larger the value, the greater the fluctuation of the reference data. is the second-order difference of the reference data in the neighborhood, It indicates the stability of the reference data within the neighborhood. The larger the value, the more unstable the data change and the greater the fluctuation.

[0171] The similarity obtained by the above formula replaces the distance similarity in K-means clustering, sets the number of clusters to 20 (empirical value, which can be adjusted according to the actual scenario), and completes the clustering of all data. In the clustering results obtained, the data in the same cluster are data with similar time series and similar change patterns, and the same cluster is likely to contain multiple dimensional parameters. If the data of several dimensions in a cluster account for a large proportion and are very close, it means that the data of these dimensions are highly correlated in this part, and the more clusters with similar proportions, the stronger the correlation of these dimensions.

[0172] According to the above logic, we analyze any two data dimensions A and B and record the number of dimensions A and B in the kth cluster as , , then the correlation between A and B As shown in formula (3):

[0173] (3);

[0174] In formula (3), For the The total number of data contained in the clusters, , The dimensions are 、 In the cluster The proportion of data in . Representation Dimension , The closer the proportion of data in all clusters is, the smaller the value is, the closer the proportion is. , The stronger the correlation; Representation Dimension , In the cluster The smaller proportion of It represents the dimension , The sum of the smaller proportions in all clusters. The larger the value, the greater the dimension. 、 The more data in the same cluster, the stronger the correlation; is the variance of all smaller ratios, indicating the dimension 、 The smaller the value is, the more evenly distributed the data is in the same cluster, and the stronger the correlation is. In this way, the correlation between each dimension and all other dimensions can be obtained, and then the principal components after PCA can be selected using the correlation.

[0175] 503, PCA reconstruction is performed on the data using correlation and the abnormality degree at each moment is obtained according to the reconstruction error.

[0176] Here, PCA reconstruction refers to the use of the PCA algorithm to extract the principal components of multidimensional data, and then use matrix transformation to restore it to the original data after filtering out some of the principal components. The loss of data in this process is the reconstruction error. The larger the error, the greater the degree of abnormality of the original data. In related technologies, the selection of principal components will try to select principal components that retain more original data information, but the principal components selected in this way are not necessarily meaningful. Therefore, when selecting the principal components of battery parameters, in addition to the information contained, the correlation between data dimensions must also be considered.

[0177] According to the above logic, PCA is used to extract the principal components of all dimensional parameters of the battery to obtain the principal components. , , …, , and then calculate the preference of these principal components As shown in formula (4):

[0178] (4);

[0179] In formula (4), Indicates principal components, For the The contribution rate of each principal component. The greater the contribution rate, the higher the preference of the principal component. Indicates Data dimensions, Representation Dimension Outside Data dimensions, is the factor loading of the data dimension, is the total dimension of battery parameters, For Dimension , of relevance. This means that in the principal component The ratio of the correlation between two dimensions to the factor loadings. The stronger the correlation and the closer the factor loadings, the closer the two dimensions are in the principal component. The better the performance is, the higher the preference of the principal component is.

[0180] Among them, the contribution rate of the kth principal component can be obtained through the following process:

[0181] There are P variables in the original sample. After principal component analysis, the number of variables will be reduced a lot. The number of principal components should be selected based on the principal component contribution rate and the cumulative contribution rate. The contribution rate of the kth principal component is As shown in formula (5):

[0182] = (5);

[0183] In formula (5), Indicates the contribution rate of the kth principal component The total contribution rate of the principal components of P variables The contribution rate of the kth principal component The larger it is, the more information of the original data is preserved, and the greater the cumulative contribution rate of the first m principal components of the sample.

[0184] Among them, the factor loading of the data dimension can be obtained through the following process:

[0185] In practical applications, when extracting m principal components from p variables, it is necessary to give a reasonable explanation to these new m variables. The explanation often needs to be judged according to the specific problem and combined with the proportion of the original variables preserved in the principal component, that is, to solve the factor loading of the principal component and the original variables. . Factor loading As shown in formula (6):

[0186] (6);

[0187] In formula (6), Indicates the factor loading of the kth principal component on the original i-th variable (indicates the variable X i For the principal component Y k importance of). represents the element in the i-th row and i-th column of the covariance matrix, represents the i-th row of the k-th eigenvector. If the principal component analysis is performed using the correlation matrix, the factor loading As shown in formula (7):

[0188] (7);

[0189] In formula (7), represents the i-th row and i-th column of the relevant matrix, that is, 1.

[0190] Get All The degree of preference of the principal components is Sort by size, select the first If the cumulative contribution rate of the principal component is greater than 75%, then The principal components can better preserve the main information of the original data. The principal components are reconstructed to obtain the reconstructed data. Using the reconstruction error and basic abnormality of the data, we can obtain The final abnormality level at the moment As shown in formula (8):

[0191] (8);

[0192] In formula (8), is the total number of dimensions of battery parameters, express Time next The reconstruction error of the dimension parameter, the larger the error, The greater the abnormality of the moment; express Time next The basic abnormality of the dimension parameter. The larger the value, the The greater the abnormality of the moment.

[0193] 403, the battery health status monitoring is completed.

[0194] Here, through the above steps 401 to 403, the abnormal degree of battery parameters at all times is calculated. The abnormality level represents the current health status and performance of the battery. The greater the abnormality level, the more likely the battery's current operating state is to be abnormal. Set the threshold to 0.8, which specifies the abnormality level of the battery parameter at a certain moment. When ≥0.8, it is determined that the current status of the battery is abnormal and troubleshooting is required to complete real-time health monitoring of the battery.

[0195] In an embodiment of the present application, the parameter changes of multiple dimensions of the battery during operation are first analyzed, and the characteristic performance of the data is analyzed through the time series characteristics of the parameters of each dimension to determine the initial abnormality of each battery parameter. Then, the correlation between each dimension is calculated by similarity clustering, which can better extract the correlation information between the data dimensions and improve the effect of principal component analysis. Finally, the data is reconstructed by PCA based on the correlation between the dimensions, so that the principal component retains more valuable data information, and the principal component feature vector at the time of reconstruction can be used to obtain the abnormality degree based on the moment, reducing the error of the single-dimensional parameter in evaluating the battery health status.

[0196] Optionally, the transmission medium can be a wired link (such as but not limited to coaxial cable, optical fiber and digital subscriber line (DSL)) or a wireless link (such as but not limited to wireless Fidelity (WIFI), Bluetooth and mobile device network).

[0197] It should be noted that: the system provided in the above embodiment is only illustrated by the division of the above functional modules. In actual applications, the above functions can be assigned to different functional modules as needed, that is, the internal structure of the computer device is divided into different functional modules to complete all or part of the functions described above. In addition, the shopping guide device and the image extension method embodiment provided in the above embodiment belong to the same concept. The specific implementation process is detailed in the method embodiment and will not be repeated here.

[0198] Figure 6 is a schematic diagram of the structure of a computer device provided in an embodiment of the present application. For example, Figure 6As shown, the computer device 600 includes: a memory 601, a processor 602, and a computer program 603 stored in the memory 601 and running on the processor 602, wherein when the processor 602 executes the computer program 603, the computer device can execute any one of the image extension methods based on the generative image model introduced above.

[0199] In addition, an embodiment of the present application also protects a device, which may include a memory and a processor, wherein the memory stores executable program code, and the processor is used to call and execute the executable program code to execute an image extension method based on a generative image model provided in an embodiment of the present application.

[0200] In this embodiment, the functional modules of the device can be divided according to the above method example. For example, each functional module can be corresponded, or two or more functions can be integrated into one processing module. The above integrated module can be implemented in the form of hardware. It should be noted that the division of modules in this embodiment is schematic and is only a logical function division. There may be other division methods in actual implementation.

[0201] In the case of dividing each module according to each function, the device may also include a signal uploading module, a determining module, an adjusting module, etc. It should be noted that all relevant contents of each step involved in the above method embodiment can be referred to the functional description of the corresponding functional module, which will not be repeated here.

[0202] It should be understood that the device provided in this embodiment is used to execute the above-mentioned image extension method based on the generative image model, and thus can achieve the same effect as the above-mentioned implementation method.

[0203] In the case of an integrated unit, the device may include a processing module and a storage module. When the device is applied to a device, the processing module may be used to control and manage the actions of the device. The storage module may be used to support the device to execute mutual program codes, etc.

[0204] The processing module may be a processor or a controller, which may implement or execute various exemplary logic blocks, modules and circuits disclosed in the present application. The processor may also be a combination that implements computing functions, such as a combination of one or more microprocessors, a combination of digital signal processing (DSP) and a microprocessor, etc. The storage module may be a memory.

[0205] In addition, the device provided in the embodiments of the present application may specifically be a chip, a component or a module, and the chip may include a connected processor and a memory; wherein the memory is used to store instructions, and when the processor calls and executes the instructions, the chip can execute an image extension method based on a generative image model provided in the above embodiment.

[0206] This embodiment also provides a computer-readable storage medium, which stores computer program code. When the computer program code runs on a computer, the computer executes the above-mentioned related method steps to implement an image extension method based on a generative image model provided in the above embodiment.

[0207] This embodiment also provides a computer program product. When the computer program product runs on a computer, it enables the computer to execute the above-mentioned related steps to implement an image extension method based on a generative image model provided by the above embodiment.

[0208] Among them, the device, computer-readable storage medium, computer program product or chip provided in this embodiment is used to execute the corresponding method provided above. Therefore, the beneficial effects that can be achieved can refer to the beneficial effects in the corresponding method provided above, and will not be repeated here.

[0209] Through the description of the above implementation methods, technical personnel in the relevant field can understand that for the convenience and simplicity of description, only the division of the above-mentioned functional modules is used as an example. In actual applications, the above-mentioned functions can be assigned to different functional modules as needed, that is, the internal structure of the device can be divided into different functional modules to complete all or part of the functions described above.

[0210] In the embodiments provided in the present application, it should be understood that the disclosed devices and methods can be implemented in other ways. For example, the device embodiments described above are only schematic, for example, the division of modules or units is only a logical function division, and there may be other division methods in actual implementation, such as multiple units or components can be combined or integrated into another device, or some features can be ignored or not executed. Another point is that the mutual coupling or direct coupling or communication connection shown or discussed can be through some interfaces, indirect coupling or communication connection of devices or units, which can be electrical, mechanical or other forms.

[0211] The above contents are only specific implementation methods of the present application, but the protection scope of the present application is not limited thereto. Any technician familiar with the technical field can easily think of changes or substitutions within the technical scope disclosed in the present application, which should be included in the protection scope of the present application. Therefore, the protection scope of the present application should be based on the protection scope of the claims.

Claims

1. Real-time battery health monitoring system based on electrochemical sensor, characterized in that: The system comprises: Electrochemical sensor, used to obtain multiple dimensional parameters of the battery within a preset time and send them to the monitoring and analysis module; A monitoring and analysis module, used to determine the correlation between the multiple dimensional parameters; and based on the correlation, determine the abnormality degree of the multiple dimensional parameters at different times within a preset time period; Among them, the monitoring and analysis module includes: A first determination submodule, configured to determine a basic abnormality degree of each dimensional parameter based on time series variation information corresponding to the plurality of dimensional parameters; A first clustering submodule is used to perform similarity clustering on the multiple dimensional parameters based on the basic abnormality degree of each dimensional parameter to obtain the correlation between the multiple dimensional parameters; A reconstruction submodule, configured to reconstruct the multiple dimensional parameters based on the correlation between the multiple dimensional parameters, so as to determine the abnormality degree of the multiple dimensional parameters at different times; Wherein, the first determination submodule includes: A construction unit, configured to construct a neighborhood of the reference data with the reference data in the multiple dimensional parameters as the center and a preset number of data as the radius; wherein the parameters in the neighborhood are continuous with the reference data in time sequence; and the reference data is any dimensional parameter in the multiple dimensional parameters; A first determining unit, configured to determine a severity of a short-term change of the reference data based on the neighborhood and the plurality of dimensional parameters; A second determining unit, configured to determine a degree of deviation of the reference data based on the reference data and an overall mean of the plurality of dimensional parameters; A first fusion unit is used to fuse the short-term change severity and the deviation degree to obtain a basic abnormality degree of the reference data; Wherein, the first determination unit includes: A first determining subunit, configured to determine a first-order difference of each parameter in the neighborhood and a first-order difference of each dimensional parameter in the multiple dimensional parameters; A second determination subunit is used to perform averaging processing on the first-order differences of multiple parameters in the neighborhood to obtain an average change trend of the neighborhood; A third determining subunit, configured to determine an overall average change trend of the multiple dimensional parameters based on the first-order differences of the multiple dimensional parameters; A fourth determining subunit is used to determine the difference between the average change trend of the neighborhood and the overall average change trend as the severity of the short-term change; Wherein, the second determining unit includes: a fifth determining subunit, configured to determine a reference difference between the reference data and the overall mean; a sixth determining subunit, configured to determine differences between the plurality of dimensional parameters and the overall mean, to obtain a plurality of differences; a seventh determining subunit, configured to determine a ratio of the reference difference to a maximum difference among the plurality of differences as a deviation degree of the reference data; Wherein, the first clustering submodule includes: a third determining unit, configured to determine a temporal distance between each parameter in a neighborhood of the reference data and the reference data; A second fusion unit is used to fuse the difference between each parameter in the neighborhood and the reference data, and the time series distance to obtain change information of the reference data in the neighborhood; a fourth determining unit, configured to determine a degree of stability of a change of the reference data within the neighborhood based on a second-order difference of the reference data within the neighborhood; A third fusion unit is used to fuse the change information of the reference data in the neighborhood and the change stability to obtain the fluctuation degree of the reference data; a fifth determining unit, configured to determine, for different dimensional parameters among the plurality of dimensional parameters, similarities between the different dimensional parameters based on time differences, fluctuation degree differences, and basic abnormality degree differences between the different dimensional parameters; A clustering unit, configured to determine the correlation between the parameters of different dimensions based on the similarities between the parameters of different dimensions; Wherein, the clustering unit comprises: A clustering subunit, used for clustering the similarities between the parameters of different dimensions to obtain a plurality of clusters; An eighth determination subunit is used to respectively determine the data proportion of the first dimension parameter and the second dimension parameter in the same cluster; wherein the first dimension parameter and the second dimension parameter are any two different dimension parameters in the multiple dimensions; A ninth determining subunit, configured to determine, based on the data proportion, a degree of proximity between the first dimensional parameter and the second dimensional parameter in the plurality of clusters; A first fusion subunit is used to fuse the data proportion and the degree of proximity of the data proportion to obtain the correlation between the first dimensional parameter and the second dimensional parameter; The state determination module is used to determine the working state of the battery based on the abnormality degree of the multiple dimensional parameters at different times within a preset time period.

2. The system according to claim 1, characterized in that The reconstruction submodule comprises: An extraction unit, used for performing principal component extraction on the multiple dimensional parameters to obtain multiple principal components; a sixth determining unit, configured to determine, for an mth principal component among the plurality of principal components, a contribution rate of the mth principal component in the dimensional parameter to which it belongs, and a factor loading amount of the mth principal component in the plurality of dimensional parameters; wherein m is an integer greater than or equal to 1 and less than or equal to the number of the plurality of principal components; a seventh determining unit, configured to determine the degree of preference of the mth principal component based on the contribution rate, the factor loading amount and the correlation between the multiple dimension parameters; A selection unit, configured to select, according to the preference degree of each principal component, at least one target principal component whose cumulative contribution rate reaches a preset threshold value from among the plurality of principal components; A reconstruction unit, configured to reconstruct the at least one target principal component to obtain a reconstruction error of the at least one target principal component at the different moments; The fourth fusion unit is used to fuse the reconstruction error and basic abnormality degree of the at least one target principal component at the different moments to obtain the abnormality degree at the different moments.

3. The system according to claim 2, characterized in that The seventh determining unit includes: a tenth determination subunit, used to determine the load difference between the factor loads corresponding to the mth principal component in each of the two dimensional parameters; An eleventh determination subunit is used to determine the ratio between the correlation between each two dimensional parameters and the load difference; The second fusion subunit is used to fuse the contribution rate and the proportion to obtain the preferred degree of the mth principal component.

4. The system according to claim 2, characterized in that The selection unit comprises: A sorting subunit, used for sorting the plurality of principal components according to the preference degree of each principal component to obtain sorted principal components; The selection subunit is used to continuously select at least one target principal component whose cumulative contribution rate reaches the preset threshold value from the sorted principal components, starting with the principal component with the greatest degree of preference.

Images