CN117761543A - Multi-sparse observer fusion power battery abnormal voltage identification method and system - Google Patents

Abstract

A multi-sparse observer fused power battery abnormal voltage identification method and system. The method includes extracting battery cell voltage data from historical data uploaded by electric vehicles and preprocessing; based on the preprocessed battery cell voltage data, determine The length of the time window is used to construct the voltage fluctuation characteristic quantity; the voltage fluctuation characteristic quantity is divided into modes to construct battery fluctuation characteristic quantity data sets under different modes; the voltage fluctuation characteristic quantity data sets under different modes are sparsely processed to construct sparse observations Collect the detectors and perform conditional judgment. If the conditions are met, sparse observer fusion is performed to determine the abnormal voltage threshold; based on the abnormal voltage threshold, abnormal voltage identification is performed on the vehicle's real-time data. Based on the voltage fluctuation status of the vehicle in different modes, the present invention uses a multi-sparse observer fusion method to determine the threshold for the abnormal voltage of the power battery. It can provide early warning for vehicles with thermal runaway and potential failure risks, and improve driving safety.

Description

A multi-sparse observer fusion power battery abnormal voltage identification method and system

Technical field

The invention belongs to the technical field of power battery safety, and specifically relates to a multi-sparse observer fusion power battery abnormal voltage identification method and system.

Background technique

As global energy shortages continue to intensify, the popularity of electric vehicles continues to increase. Electric vehicles are currently recognized as a key direction for the future development of the automotive industry. Therefore, research on the safety of vehicle battery systems has received increasing attention from industry and academia. Voltage, as an external representation parameter of power batteries, is an important indicator for evaluating the safety of battery systems. A large number of studies have shown that abnormal voltage fluctuations may indicate that the safety level of the battery system is declining, and if not controlled, it may eventually lead to voltage failure or even thermal runaway. Therefore, accurate identification of abnormal voltage fluctuations is extremely important for early detection of battery system failures and ensuring the safe operation of operating vehicles.

The threshold-based abnormal voltage identification method is widely used in real-time safety monitoring of large-scale vehicles due to its short vehicle identification time. The determination of anomaly thresholds in the current study is often based on experimental results obtained on one or a small number of vehicles and remains constant throughout. However, due to huge differences in battery specifications, driver driving habits and operating environments, different vehicles have different "abnormal voltage ranges". The constant abnormal voltage threshold used for anomaly detection is difficult to generalize to large-scale vehicle monitoring. Therefore, setting adaptive abnormality determination thresholds according to different vehicle operating conditions is of great significance for quickly and accurately identifying abnormal voltages.

Contents of the invention

The purpose of the present invention is to provide a multi-sparse observer fused power battery abnormal voltage identification method and system to solve the above-mentioned problems in the prior art. Through vehicle historical data and based on the voltage fluctuation status of the vehicle in different modes, the multi-sparse observer is used to identify the abnormal voltage of the power battery. The observer fusion method determines the threshold for the abnormal voltage of the power battery, and can adaptively set the abnormal voltage threshold for the bicycle.

In order to achieve the above objects, the present invention has the following technical solutions:

In the first aspect, a multi-sparse observer fusion power battery abnormal voltage identification method is provided, including:

From the historical data uploaded by electric vehicles, the battery cell voltage data is extracted and preprocessed;

Based on the preprocessed battery cell voltage data, determine the length of the time window and construct the voltage fluctuation characteristic quantity;

Divide the voltage fluctuation characteristic quantities into modes and construct battery fluctuation characteristic quantity data sets under different modes;

Perform sparse processing on the voltage fluctuation feature data sets under different modes, construct a sparse observer set, and perform conditional judgments. If the conditions are met, sparse observer fusion will be performed to determine the abnormal voltage threshold;

Based on the abnormal voltage threshold, abnormal voltage identification is performed on vehicle real-time data.

As an optimal solution, upload the historical data of electric vehicles according to the national standard GB/T 32960 "Technical Specifications for Remote Service and Management Systems of Electric Vehicles", extract the battery cell voltage data and perform preprocessing. The preprocessing steps include:

For time disorder and data duplication in the sample data, sort the data by time and perform redundant operations;

If there is a null value in the single cell voltage value list, it will be considered as invalid frame data;

If the highest cell voltage value is greater than 6V, it is considered invalid frame data;

If the lowest cell voltage value is less than 0V, it is considered invalid frame data.

As a preferred solution, the voltage fluctuation characteristic quantity is the voltage fluctuation of a single battery cell in a specific time period, and the calculation expression is as follows:

In the formula, x _i represents the voltage value at time i in the observation window, μ represents the average of the voltage values at all times in the observation window, n represents the number of samples in the observation window, and Std _i represents the voltage of the i-th battery cell in the fixed time window. standard deviation;

Calculate the standard deviation after eliminating cells that exceed the median of the battery cell voltage standard deviation by more than 10 times:

In the formula, Std _median represents the median of the standard deviation of all cell voltages in a fixed time window, Std _std represents the standard deviation of the standard deviation of all cell voltages in a fixed time window after being processed by the Pre-standard method, and Std_standard _i represents the i-th The abnormal voltage fluctuation score of a battery cell after being processed by the Pre-standard method.

As a preferred solution, the step of dividing the voltage fluctuation characteristic quantity into modes includes: taking the median of the fluctuation characteristic quantity of all battery cells at the same time to represent the general fluctuation trend of the power battery pack at the corresponding time, which is called the safety mode; taking the same value The maximum value of the fluctuation characteristic quantity of all battery cells at a time represents the limit fluctuation trend of the power battery pack at the corresponding time, which is called the limit mode.

As a preferred solution, the step of sparsely processing the voltage fluctuation feature data sets in different modes and constructing a sparse observer set includes:

For the battery fluctuation feature set M under different modes, each object in the set is an observer; multiple objects are randomly selected from the battery fluctuation feature set M to form an observer set. According to the statistical sampling theory, the specified confidence is obtained The number of observers required for the interval and sampling error is calculated as follows:

In the formula, k is the number of observers required, m is the number of objects in the battery fluctuation feature set M, σ is the standard deviation of the battery fluctuation feature set M, ε is the limit error without repeated sampling, and Z is the confidence coefficient;

Calculate the Euclidean distance D _i,j of each observer and each object in the battery fluctuation feature set M, and construct a distance matrix D, that is, D={D _i,j ,i∈(1,2,..., m), j∈(1,2,...,k)}; Construct an observation matrix P to store the indices of the x observers closest to each object in the data set;

Calculate all index values in the observation matrix P and create a matrix U to represent the number of occurrences of each observer, that is, P = {P _j ,j∈(1,2,...,k)}; set according to the matrix U The inertia threshold q, the value of q is calculated as follows:

q＝ _Qρ (U)

Among them, Q _ρ (.) is the quantile function;

When the number of occurrences of an observer is less than q, the corresponding observer is considered a lazy observer, and the lazy observer is deleted from the observer set; after the lazy observer is removed, k _act observers remain and become active observers , a low-density model used to characterize the distribution characteristics of the training data set.

As a preferred solution, the step of performing conditional judgment, performing sparse observer fusion if the condition is met, and determining the abnormal voltage threshold includes:

Obtain the safe mode observer set Saf_sample and the extreme mode observer set Ext_sample, and use the fluctuation coefficient fluc_coefficient to characterize the fluctuation characteristics of the power battery itself. The calculation expression is as follows:

fluc_coefficient=min(Ext_sample)-max(Saf_sample)

Taking fluc_threshold as the threshold to determine vehicle voltage abnormality, the calculation expression is as follows:

When the fluctuation characteristic quantity is greater than fluc_threshold, it is identified as abnormal voltage;

When sampling, ensure that fluc_threshold is greater than the minimum value of Safe_sample. If this condition is met, the two observer sets will be fused to detect voltage fluctuation anomalies. If not, resampling will be required.

As a preferred solution, the step of identifying abnormal voltage on vehicle real-time data based on abnormal voltage thresholds includes: when judging the abnormality degree of new object o, calculate the difference between new object o and each observer in the low-density model. Euclidean distance, forming a distance matrix N with a length of k _act , recording the index values of the nearest x observers, and constructing an observation array P; at the same time, calculating the average distance between the new object o and the nearest x observers, expressed Abnormal degree y _o :

In the formula, M(.) is the median function, and d(,) is the Euclidean distance.

In the second aspect, a multi-sparse observer fusion power battery abnormal voltage identification system is provided, including:

The battery cell voltage data extraction module is used to extract the battery cell voltage data from the historical data uploaded by electric vehicles and perform preprocessing;

The voltage fluctuation characteristic quantity building module is used to determine the length of the time window and construct the voltage fluctuation characteristic quantity based on the preprocessed battery cell voltage data;

The battery fluctuation characteristic quantity data set division module is used to divide the voltage fluctuation characteristic quantity into modes and construct battery fluctuation characteristic quantity data sets in different modes;

The abnormal voltage threshold determination module is used to sparsely process voltage fluctuation feature data sets in different modes, construct a sparse observer set, and perform conditional judgments. If the conditions are met, sparse observer fusion is performed to determine the abnormal voltage threshold;

The abnormal voltage identification module is used to identify abnormal voltages on vehicle real-time data based on abnormal voltage thresholds.

In a third aspect, an electronic device is provided, including:

A memory that stores at least one instruction; and a processor that executes the instructions stored in the memory to implement the multi-sparse observer fusion power battery abnormal voltage identification method.

In a fourth aspect, a computer-readable storage medium is provided. At least one instruction is stored in the computer-readable storage medium. The at least one instruction is executed by a processor in an electronic device to implement the multi-sparse observer fusion. Method for identifying abnormal voltage of power battery.

Compared with the prior art, the present invention at least has the following beneficial effects:

For electric vehicles in operation, different vehicles have different definition ranges of abnormal voltages due to differences in battery types, driver driving habits, operating environments, etc. The present invention focuses on the identification of abnormal voltages of electric vehicle power batteries and uploads data to Based on the big data of actual vehicle operation, in view of the differences in the voltage fluctuation range of different vehicles due to differences in battery types, driver driving habits, operating environments, etc., a power battery abnormal voltage adaptation based on the fusion of multiple sparse observers is proposed The identification method divides the voltage fluctuation characteristic quantity into patterns, constructs battery fluctuation characteristic quantity data sets in different modes, performs sparse processing on the voltage fluctuation characteristic quantity data sets in different modes, constructs a sparse observer set, and performs Conditional judgment, if the conditions are met, sparse observer fusion is performed to determine the abnormal voltage threshold. Finally, based on the determined abnormal voltage threshold, abnormal voltage identification is performed on the vehicle's real-time data. By using a sparse representation method to characterize the normal fluctuation range of the battery, and using multi-sparse observer fusion to identify abnormal voltages, the present invention can provide early warning for vehicles with thermal runaway and potential failure risks, thereby improving driving safety.

Description of the drawings

In order to more clearly illustrate the technical solutions in the embodiments of the present invention, the drawings needed to be used in the description of the embodiments will be briefly introduced below. Obviously, the drawings in the following description are some embodiments of the present invention. Those of ordinary skill in the art can also obtain other drawings based on these drawings without exerting creative efforts.

Figure 1 is an overall flow chart of the abnormal voltage identification method of a multi-sparse observer fused power battery according to an embodiment of the present invention;

Figure 2 is a voltage curve diagram of each battery cell in the first group of vehicle battery cell data according to the embodiment of the present invention;

Figure 3 is a Std_standard curve diagram of each battery cell in the first group of vehicle battery cell data according to the embodiment of the present invention;

Figure 4 is a graph of the abnormality coefficient of each battery cell in the first group of vehicle battery cell data according to the embodiment of the present invention;

Figure 5 is a voltage curve diagram of each battery cell in the second set of vehicle battery cell data according to the embodiment of the present invention;

Figure 6 is a graph of anomaly coefficients of each battery cell in the second set of vehicle battery cell data according to the embodiment of the present invention.

Detailed ways

The present invention will be described in further detail below in conjunction with the accompanying drawings.

Please refer to Figure 1. A multi-sparse observer fusion power battery abnormal voltage identification method proposed by an embodiment of the present invention includes the following steps:

(1) According to the national standard GB/T 32960 "Technical Specification for Electric Vehicle Remote Service and Management System", electric vehicles upload historical data, extract battery cell voltage data and perform preprocessing;

(2) Based on the battery cell voltage data, determine the length of the time window and construct the voltage fluctuation characteristic quantity;

(3) Divide the voltage fluctuation characteristic quantities into modes and construct battery fluctuation characteristic quantity data sets under different modes;

(4) Perform sparse processing on the voltage fluctuation feature data sets in different modes, construct a sparse observer set, and perform conditional judgments. If the conditions are met, sparse observer fusion is performed and the abnormal voltage threshold is determined;

(5) Based on the abnormal voltage threshold, abnormal voltage identification is performed on vehicle real-time data.

In a possible implementation, in step (1), the data uploaded by the electric vehicle according to the national standard GB/T 32960 "Technical Specification for Electric Vehicle Remote Service and Management System" needs to be preprocessed and analyzed. The preprocessing principles are as follows:

1) In view of the problems of time disorder and data duplication in the sample data, sort the data by time and perform redundant operations;

2) If there is a null value in the single cell voltage value list, it means that not all voltage values of this frame data are valid, so it is regarded as invalid frame data;

3) If the highest cell voltage value is greater than 6V, it is considered invalid frame data;

4) If the lowest cell voltage value is less than 0V, it is considered invalid frame data.

In a possible implementation, step (2) selects the standard deviation as a characteristic quantity to characterize the voltage fluctuation of a single battery cell. In order to characterize the voltage fluctuation of a single battery cell in a specific time period, observation window technology is used. Calculated as follows:

In the formula, x _i represents the voltage value at time i in the observation window, μ represents the average of the voltage values at all times in the observation window, n represents the number of samples in the observation window, and Std _i represents the voltage of the i-th battery cell in the fixed time window. standard deviation.

In order to characterize the horizontal comparison between different battery cells within the same time period, the Pre-standard method was proposed. In view of the voltage fluctuation characteristics in the early stage of power battery failure, the Pre-standard method is improved based on the Z-score standardization method. Regardless of the small abnormal fluctuations in the early stages of the battery pack or the instantaneous changes in the battery cells in the early stage of failure, abnormal battery cells often occur in a few of the many cells in the battery pack. Therefore, in the Pre-standard method, battery cells are used The median of the body voltage standard deviation replaces the average value to characterize the normal state of the battery pack. Similarly, when calculating the standard deviation, it is concluded from the analysis of multiple vehicles with thermal runaway failures that the standard deviation of abnormal cells in the early stage of failure often exceeds the normal state by more than 10 times. Therefore, the standard deviation of the battery cell voltage exceeding the standard deviation is The standard deviation is calculated after the monomers that are more than 10 times the median are eliminated to ensure that they are not affected by voltage mutations in the early stage of failure. The calculation formula is as follows:

Among them, Std _median represents the median of the standard deviation of all single cell voltages in a fixed time window, and Std _std represents the standard deviation of the standard deviation of all single cells in a fixed time window after being processed by the Pre-standard method. Std_standard _i represents the abnormal voltage fluctuation score of the i-th battery cell after being processed by the Pre-standard method.

In one possible implementation, based on the operating characteristics of the power battery, step (3) proposes two modes that can be used by the sparse observer to characterize the fluctuation range of the power battery voltage in the normal state. For power battery packs, small abnormal fluctuations in the early stages of failure are often reflected in a single or a few cells. Therefore, for the vast majority of cells, in the presence of abnormal cell fluctuations, they can still guarantee normal operation. status running. Therefore, the median of the fluctuation characteristics of all battery cells at the same time is taken to represent the general fluctuation trend of the power battery pack at that time, which is called the safety mode; the maximum value of the fluctuation characteristics of all battery cells at the same time is taken to represent the power battery pack at that time. The extreme fluctuation trend is called the extreme mode. Save the battery fluctuation feature sets in different modes independently to construct battery fluctuation feature data sets in different modes.

In a possible implementation, in step (4), an abnormal voltage threshold determination method based on multi-sparse observer fusion is proposed. First, initialize the observer. For the battery fluctuation feature set M under different modes, each object in the set can be called an observer. Multiple objects are randomly selected from M to form an observer set. According to the statistical sampling theory, the number of observers required to specify the confidence interval and sampling error can be obtained. The calculation formula is as follows:

Among them, k is the number of observers required, m is the number of objects in M, σ is the standard deviation of M, ε is the limit error of non-repeated sampling, take ε = 0.1σ, Z is the confidence coefficient, take 95% confidence interval, so Z=1.96.

Secondly, the observation matrix is constructed, the Euclidean distance D _i,j of each observer and each object in M is calculated, and the distance matrix D is constructed, that is, D={D _i,j ,i∈(1,2,..., m),j∈(1,2,...,k)}. The observation matrix P is constructed to store the indices of the x nearest observers to each object in the data set. The general value of x is 3-10, and it will not cause a large deviation in the identification of abnormal points, but it is related to the value of k and the scene characteristics. Therefore, comprehensive considerations should be taken when selecting to ensure that the problems of oversimplification and overfitting of the model will not occur.

Then the lazy observer is removed, all index values in the observation matrix are counted, and a matrix U is created to represent the number of occurrences of each observer, that is, P={P _j ,j∈(1,2,...,k) }. Set the inertia threshold q according to the matrix U, and the calculation formula is as follows:

q＝ _Qρ (U)

Among them, Q _ρ (.) is the quantile function. According to empirical tests, when ρ is 0.3, the effect is better.

When the number of occurrences of an observer is less than q, it is considered a lazy observer and is deleted from the observer set. This filtering can ensure that the remaining observers can represent the medium and high density areas of the training data and avoid the interference of outliers on the recognition effect. After removing the lazy observers, the remaining k _act observers become active observers, which are low-density models that can be used to characterize the distribution characteristics of the training data set.

Finally, the fluctuation characteristic quantities in the two modes are subjected to the above operations respectively to obtain the safe mode observer set Saf_sample and the extreme mode observer set Ext_sample. The fluctuation coefficient fluc_coefficient is used to characterize the fluctuation characteristics of the power battery itself. The calculation formula is as follows:

fluc_coefficient=min(Ext_sample)-max(Saf_sample)

It can be seen from the above formula that different vehicles and different training samples will have different effects on the fluctuation coefficient, which exactly reflects the uniqueness of voltage fluctuations of different vehicles under the influence of multiple factors. And for the same vehicle, as the battery usage time is different, As the degree of aging gradually intensifies, the fluctuation coefficient should also show corresponding changes. The larger the fluctuation coefficient, it means that the fluctuation of the battery in the environment is more severe than other batteries, and the voltage fluctuation range under normal operation is larger, so the standard for determining abnormal voltage should also be increased.

Taking fluc_threshold as the threshold to determine vehicle voltage abnormality, the calculation formula is as follows:

When the fluctuation characteristic quantity is greater than fluc_threshold, it is determined to be an abnormal voltage. Taking into account the distribution characteristics of the two modes and the systematic errors in sparse representation, to avoid the problem of unbalanced sample distribution, it should be ensured that fluc_threshold is greater than the minimum value of Safe_sample when sampling. If satisfied, the two observer sets will be fused for voltage Detection of fluctuation abnormal points, if not satisfied, re-sampling is required.

In a possible implementation, in step (5), when a new object o enters and its abnormality degree needs to be judged, the Euclidean distance between the new object o and each observer in the low-density model should be calculated, forming A distance matrix N of length k _act records the index values of the x nearest observers to construct an observation array P. At the same time, the average distance between the new object o and the x nearest observers is calculated, which can be expressed as the anomaly degree y _o . The calculation formula is as follows:

Among them, M(.) is the median function and d(,) is the Euclidean distance.

The multi-sparse observer fused power battery abnormal voltage identification method of the present invention focuses on the identification of abnormal voltage of the electric vehicle power battery. Based on the real vehicle operation big data uploaded by the electric vehicle, it is targeted at different vehicles due to battery types, driver driving habits, Differences in the operating environment and other aspects lead to differences in the voltage fluctuation range. An adaptive identification method for power battery abnormal voltage based on the fusion of multiple sparse observers is proposed. Feature quantities based on variance and Pre-standard methods are used to characterize power battery voltage fluctuations. Sparse The representation method represents the normal fluctuation range of the battery, and uses multi-sparse observer fusion to identify abnormal voltages, which can provide early warning for vehicles with thermal runaway and potential failure risks, thereby improving driving safety.

In order to verify the early warning effect of the method of the present invention on vehicles with thermal runaway and potential failure risks, the operating data of two actually operating vehicles were used for verification. The first set of vehicle battery cell data uses six days of data from August 11, 2019 to August 16, 2019 for training, and the data of September 29, 2019 is used for testing. Figure 2 shows the voltage changes of each single cell during the test time segment. From Figure 2, it can be found that the No. 28 battery cell has experienced several maximum and minimum voltage changes in a short period of time, and the voltage fluctuations are relatively large. The remaining cells are more violent, which is undoubtedly an abnormal fluctuation of voltage. Figure 3 shows the change of its fluctuation characteristic quantity Std_standard. It can be seen from Figure 3 that the Std_standard at this moment is significantly higher than other moments, so Std_standard accurately characterizes this voltage abnormality. The fluctuation characteristic quantity of each battery cell is put into the model for testing, and the cell voltage abnormal coefficient fluctuation chart shown in Figure 4 is obtained. The figure shows whether the voltage fluctuation at the current moment is abnormal compared with the vehicle's historical data. From It can be seen from the figure that the abnormal coefficient of voltage fluctuation at this moment is higher than the abnormal threshold, so it is identified as an abnormal point.

The second set of vehicle battery cell data uses seven days of data from January 13, 2020 to January 19, 2020 for training, and the data of March 2, 2020 is used for testing. The vehicle experienced a thermal event on March 2. Out of control failure. Figure 5 shows the voltage changes of each single battery during the test time segment. It can be seen from the figure that before 325 frames of data, the vehicle was in a charging state and was nearly full. The voltage of each single battery continued to increase and the voltage fluctuated. It has always shown good consistency. At the 325th frame of data, the voltage of monomer No. 23-27 suddenly dropped, reaching a minimum of 0.25V. The voltage of monomer No. 28 suddenly rose, reaching 4.998V, and This change continues until a thermal runaway failure occurs. The fluctuation characteristics of each battery cell were put into the model for testing, and the cell voltage abnormal coefficient fluctuation chart shown in Figure 6 was obtained. At the moment when the faulty cell suddenly mutated, cells No. 23-28 were all higher than those of the vehicle according to the vehicle. The abnormal threshold determined by historical data, therefore, the method proposed by the present invention can identify abnormal voltage fluctuations in a timely manner and play a predictive role in faults to a certain extent.

For electric vehicles in operation, different vehicles have different definition ranges of abnormal voltage due to differences in battery types, driver driving habits, operating environments, etc. Therefore, the present invention uses a sparse representation method to characterize the voltage based on the multi-sparse observer fusion method. The normal fluctuation range is generated and an adaptive threshold is generated to identify and locate the voltage moment exceeding the threshold. In addition, the present invention greatly reduces the amount of calculation in terms of feature construction, sparse representation, pattern division, etc., ensuring that the method proposed by the present invention is more suitable for monitoring real vehicle online data and ensuring the accuracy of the abnormal voltage identification method. Real-time performance provides theoretical support for real-time safety risk monitoring of operating vehicles.

Another embodiment of the present invention also proposes a multi-sparse observer fusion power battery abnormal voltage identification system, including:

The battery cell voltage data extraction module is used to extract the battery cell voltage data from the historical data uploaded by electric vehicles and perform preprocessing;

The voltage fluctuation characteristic quantity building module is used to determine the length of the time window and construct the voltage fluctuation characteristic quantity based on the preprocessed battery cell voltage data;

The battery fluctuation characteristic quantity data set division module is used to divide the voltage fluctuation characteristic quantity into modes and construct battery fluctuation characteristic quantity data sets in different modes;

The abnormal voltage threshold determination module is used to sparsely process voltage fluctuation feature data sets in different modes, construct a sparse observer set, and perform conditional judgments. If the conditions are met, sparse observer fusion is performed to determine the abnormal voltage threshold;

The abnormal voltage identification module is used to identify abnormal voltages on vehicle real-time data based on abnormal voltage thresholds.

Another embodiment of the present invention also provides an electronic device, including:

A memory that stores at least one instruction; and a processor that executes the instructions stored in the memory to implement the multi-sparse observer fusion power battery abnormal voltage identification method.

Another embodiment of the present invention also provides a computer-readable storage medium. The computer-readable storage medium stores at least one instruction, and the at least one instruction is executed by a processor in an electronic device to implement the multiple tasks. Sparse observer fusion power battery abnormal voltage identification method.

Exemplarily, the instructions stored in the memory may be divided into one or more modules/units, and the one or more modules/units are stored in a computer-readable storage medium and executed by the processor, To complete the multi-sparse observer fusion power battery abnormal voltage identification method of the present invention. The one or more modules/units may be a series of computer-readable instruction segments capable of completing specific functions. The instruction segments are used to describe the execution process of the computer program in the server.

The electronic device may be a computing device such as a smartphone, a notebook, a PDA, a cloud server, etc. The electronic device may include, but is not limited to, a processor and a memory. Those skilled in the art can understand that the electronic device may also include more or less components, or a combination of certain components, or different components. For example, the electronic device may also include input and output devices, network access devices, bus etc.

The processor can be a central processing unit (CentraL Processing Unit, CPU), or other general-purpose processor, digital signal processor (Digital Signal Processor, DSP), application specific integrated circuit (AppLication Specific Integrated Circuit, ASIC), off-the-shelf Programmable gate array (Field-Programmable Gate Array, FPGA) or other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components, etc. A general-purpose processor may be a microprocessor or the processor may be any conventional processor, etc.

The memory may be an internal storage unit of the server, such as a hard disk or memory of the server. The memory may also be an external storage device of the server, such as a plug-in hard disk, a smart media card (SMC), a secure digital (SD) card, a flash memory card ( FLash Card) etc. Further, the memory may also include both an internal storage unit of the server and an external storage device. The memory is used to store the computer readable instructions and other programs and data required by the server. The memory may also be used to temporarily store data that has been output or is to be output.

It should be noted that the information interaction, execution process, etc. between the above-mentioned module units are based on the same concept as the method embodiments, and their specific functions and technical effects can be found in the method embodiments section, which will not be discussed here. Repeat.

Those skilled in the art can clearly understand that for the convenience and simplicity of description, only the division of the above functional units and modules is used as an example. In actual applications, the above functions can be allocated to different functional units and modules according to needs. Module completion means dividing the internal structure of the device into different functional units or modules to complete all or part of the functions described above. Each functional unit and module in the embodiment can be integrated into one processing unit, or each unit can exist physically alone, or two or more units can be integrated into one unit. The above-mentioned integrated unit can be hardware-based. It can also be implemented in the form of software functional units. In addition, the specific names of each functional unit and module are only for the convenience of distinguishing each other and are not used to limit the scope of protection of the present application. For the specific working processes of the units and modules in the above system, please refer to the corresponding processes in the foregoing method embodiments, and will not be described again here.

If the integrated unit is implemented in the form of a software functional unit and sold or used as an independent product, it may be stored in a computer-readable storage medium. Based on this understanding, this application can implement all or part of the processes in the methods of the above embodiments by instructing relevant hardware through a computer program. The computer program can be stored in a computer-readable storage medium. The computer program When executed by a processor, the steps of each of the above method embodiments may be implemented. Wherein, the computer program includes computer program code, which may be in the form of source code, object code, executable file or some intermediate form. The computer-readable medium may at least include: any entity or device capable of carrying computer program code to the camera device/terminal device, recording media, computer memory, read-only memory (ROM, Read-Only Memory), random access memory (RAM, RandomAccess Memory), electrical carrier signals, telecommunications signals, and software distribution media. For example, U disk, mobile hard disk, magnetic disk or CD, etc.

In the above embodiments, each embodiment is described with its own emphasis. For parts that are not detailed or documented in a certain embodiment, please refer to the relevant descriptions of other embodiments.

The above-described embodiments are only used to illustrate the technical solutions of the present application, but not to limit them; although the present application has been described in detail with reference to the foregoing embodiments, those of ordinary skill in the art should understand that they can still implement the above-mentioned implementations. The technical solutions described in the examples are modified, or some of the technical features are equivalently replaced; and these modifications or substitutions do not cause the essence of the corresponding technical solutions to deviate from the spirit and scope of the technical solutions in the embodiments of this application, and should be included in within the protection scope of this application.

Claims (10)

1. A multi-sparse observer fusion power battery abnormal voltage identification method, which is characterized by including: From the historical data uploaded by electric vehicles, the battery cell voltage data is extracted and preprocessed; Based on the preprocessed battery cell voltage data, determine the length of the time window and construct the voltage fluctuation characteristic quantity; Divide the voltage fluctuation characteristic quantities into modes and construct battery fluctuation characteristic quantity data sets under different modes; Perform sparse processing on the voltage fluctuation feature data sets under different modes, construct a sparse observer set, and perform conditional judgments. If the conditions are met, sparse observer fusion will be performed to determine the abnormal voltage threshold; Based on the abnormal voltage threshold, abnormal voltage identification is performed on vehicle real-time data. 2. The multi-sparse observer fusion power battery abnormal voltage identification method according to claim 1, characterized in that the historical data of electric vehicles are uploaded according to the national standard GB/T 32960 "Technical Specification for Remote Service and Management Systems of Electric Vehicles" and extracted. Battery cell voltage data and preprocessing, the preprocessing steps include: For time disorder and data duplication in the sample data, sort the data by time and perform redundant operations; If there is a null value in the single cell voltage value list, it will be considered as invalid frame data; If the highest cell voltage value is greater than 6V, it is considered invalid frame data; If the lowest cell voltage value is less than 0V, it is considered invalid frame data. 3. The multi-sparse observer fusion power battery abnormal voltage identification method according to claim 1, characterized in that the voltage fluctuation characteristic quantity is the voltage fluctuation of a single battery cell in a specific time period, and the calculation expression is as follows: In the formula, x _i represents the voltage value at time i in the observation window, μ represents the average of the voltage values at all times in the observation window, n represents the number of samples in the observation window, and Std _i represents the voltage of the i-th battery cell in the fixed time window. standard deviation; Calculate the standard deviation after eliminating cells that exceed the median of the battery cell voltage standard deviation by more than 10 times: In the formula, Std _median represents the median of the standard deviation of all cell voltages in a fixed time window, Std _std represents the standard deviation of the standard deviation of all cell voltages in a fixed time window after being processed by the Pre-standard method, and Std_standard _i represents the i-th The abnormal voltage fluctuation score of a battery cell after being processed by the Pre-standard method. 4. The multi-sparse observer fusion power battery abnormal voltage identification method according to claim 1, characterized in that the step of pattern dividing the voltage fluctuation characteristic quantity includes: taking the median of all battery cell fluctuation characteristic quantities at the same time. The number represents the general fluctuation trend of the power battery pack at the corresponding time, which is called the safety mode; the maximum value of the fluctuation characteristic quantity of all battery cells at the same time represents the extreme fluctuation trend of the power battery pack at the corresponding time, which is called the limit mode. 5. The multi-sparse observer fusion power battery abnormal voltage identification method according to claim 4, characterized in that the step of performing sparse processing on the voltage fluctuation feature data sets in different modes and constructing a sparse observer set includes : For the battery fluctuation feature set M under different modes, each object in the set is an observer; multiple objects are randomly selected from the battery fluctuation feature set M to form an observer set. According to the statistical sampling theory, the specified confidence is obtained The number of observers required for the interval and sampling error is calculated as follows: In the formula, k is the number of observers required, m is the number of objects in the battery fluctuation feature set M, σ is the standard deviation of the battery fluctuation feature set M, ε is the limit error without repeated sampling, and Z is the confidence coefficient; Calculate the Euclidean distance D _i,j of each observer and each object in the battery fluctuation feature set M, and construct a distance matrix D, that is, D={D _i,j ,i∈(1,2,..., m), j∈(1,2,...,k)}; Construct an observation matrix P to store the indices of the x observers closest to each object in the data set; Calculate all index values in the observation matrix P and create a matrix U to represent the number of occurrences of each observer, that is, P = {P _j ,j∈(1,2,...,k)}; set according to the matrix U The inertia threshold q, the value of q is calculated as follows: q＝ _Qρ (U) Among them, Q _ρ (.) is the quantile function; When the number of occurrences of an observer is less than q, the corresponding observer is considered a lazy observer, and the lazy observer is deleted from the observer set; after the lazy observer is removed, k _act observers remain and become active observers , a low-density model used to characterize the distribution characteristics of the training data set. 6. The multi-sparse observer fusion power battery abnormal voltage identification method according to claim 5, characterized in that the step of performing conditional judgment, performing sparse observer fusion if the conditions are met, and determining the abnormal voltage threshold includes: Obtain the safe mode observer set Saf_sample and the extreme mode observer set Ext_sample, and use the fluctuation coefficient fluc_coefficient to characterize the fluctuation characteristics of the power battery itself. The calculation expression is as follows: fluc_coefficient=min(Ext_sample)-max(Saf_sample) Taking fluc_threshold as the threshold to determine vehicle voltage abnormality, the calculation expression is as follows: When the fluctuation characteristic quantity is greater than fluc_threshold, it is identified as abnormal voltage; When sampling, ensure that fluc_threshold is greater than the minimum value of Safe_sample. If this condition is met, the two observer sets will be fused to detect voltage fluctuation anomalies. If not, resampling will be required. 7. The multi-sparse observer fusion power battery abnormal voltage identification method according to claim 6, characterized in that the step of identifying abnormal voltage on vehicle real-time data based on abnormal voltage thresholds includes: When judging the abnormality of a new object o, calculate the Euclidean distance between the new object o and each observer in the low-density model, form a distance matrix N with a length of k _act , and record the index values of the nearest x observers , construct the observation array P; at the same time, calculate the average distance between the new object o and the nearest x observers, indicating the abnormality degree y _o : In the formula, M(.) is the median function, and d(,) is the Euclidean distance. 8. A multi-sparse observer fusion power battery abnormal voltage identification system, which is characterized by including: The battery cell voltage data extraction module is used to extract the battery cell voltage data from the historical data uploaded by electric vehicles and perform preprocessing; The voltage fluctuation characteristic quantity building module is used to determine the length of the time window and construct the voltage fluctuation characteristic quantity based on the preprocessed battery cell voltage data; The battery fluctuation characteristic quantity data set division module is used to divide the voltage fluctuation characteristic quantity into modes and construct battery fluctuation characteristic quantity data sets in different modes; The abnormal voltage threshold determination module is used to sparsely process voltage fluctuation feature data sets in different modes, construct a sparse observer set, and perform conditional judgments. If the conditions are met, sparse observer fusion is performed to determine the abnormal voltage threshold; The abnormal voltage identification module is used to identify abnormal voltages on vehicle real-time data based on abnormal voltage thresholds. 9. An electronic device, characterized in that it includes: a memory to store at least one instruction; and A processor that executes instructions stored in the memory to implement the multi-sparse observer fusion power battery abnormal voltage identification method according to any one of claims 1 to 7. 10. A computer-readable storage medium, characterized in that: the computer-readable storage medium stores at least one instruction, and the at least one instruction is executed by a processor in an electronic device to implement claims 1 to 7 The multi-sparse observer fusion power battery abnormal voltage identification method described in any one of the above.