Method, apparatus, and work vehicle for determining monobloc cell failure type

Abstract

Embodiments of the present application provide a method, device, engineering vehicle, processor and storage medium for determining a monomer battery fault type. The method is applied to an engineering vehicle including at least one battery pack, and includes: collecting running state of the engineering vehicle and state data of each monomer battery in the battery pack within a preset time period, wherein the battery pack includes a plurality of monomer batteries; determining a target monomer battery that has a fault in the battery pack according to the running state and the state data; sequentially inputting the state data and the running state of each target monomer battery into a trained long short-term memory artificial neural network to determine a fault type of each target monomer battery through the trained long short-term memory artificial neural network. The method considers the running state of the new energy engineering vehicle in the fault diagnosis process, and can avoid misdiagnosis caused by state transformation.

Description

Technical Field

[0001] This application relates to the field of battery fault diagnosis technology, specifically to a method, apparatus, engineering vehicle, storage medium, and processor for determining the fault type of a single battery cell. Background Technology

[0002] New energy vehicles are one of the most important ways to reduce carbon emissions. In recent years, my country's new energy vehicle industry has entered a stage of rapid development. As the most crucial component of new energy pump trucks, the operating status of the power battery directly affects the overall performance of the vehicle. Power battery failures can range from affecting the construction efficiency and operational experience of the pump truck to causing short circuits or even explosions, leading to fires and posing a significant threat to the safety of the pump truck and its personnel. Fault diagnosis of the power battery in new energy pump trucks can prevent problems before they occur, avoid major accidents caused by power battery failures, extend the battery's lifespan, reduce damage to other components of the pump truck caused by battery failures, and reduce vehicle operating costs related to the power battery.

[0003] However, in existing technologies, fault diagnosis is usually performed on the entire vehicle's battery pack, which results in low accuracy in fault diagnosis and increases the difficulty of maintenance. Summary of the Invention

[0004] The purpose of this application is to provide a method, apparatus, engineering vehicle, storage medium, and processor for determining the fault type of a single battery cell.

[0005] To achieve the above objectives, a first aspect of this application provides a method for determining the fault type of a single battery cell, applied to an engineering vehicle, the engineering vehicle including at least one battery pack, the method comprising:

[0006] Collect the operating status of engineering vehicles and the status data of each individual battery cell in the battery pack within a preset time period. The battery pack includes multiple individual batteries.

[0007] Identify the target cell in the battery pack that has failed based on its operating status and status data;

[0008] The state data and operating status of each target cell are sequentially input into a trained long short-term memory artificial neural network to determine the fault type of each target cell.

[0009] In the embodiments of this application, the operating state includes at least a start-up state, a fabrication state, a normal transport state, and an emergency stop state. The state data includes at least the voltage, current, temperature, temperature change rate, state of charge, and internal resistance of a single battery cell. Determining the target single battery cell that has failed in the battery pack based on the operating state and state data includes: forming a battery set from any two single batteries in the battery pack; determining the characteristic distance between two single batteries in the battery set based on the state data of each single battery cell in the battery set; determining a target battery set whose characteristic distance is greater than a preset distance threshold; and determining the single battery cell included in at least two target battery sets as the target single battery cell.

[0010] In embodiments of this application, the types of characteristic distances include at least Manhattan distance, Euclidean distance, and Chebyshev distance. Determining the characteristic distance between two individual cells in the battery set based on the state data of each individual cell in the battery set includes: determining the Manhattan distance, Euclidean distance, and Chebyshev distance between two individual cells in the battery set based on the state data of each individual cell in the battery set; determining the weighting coefficient corresponding to each type of characteristic distance; and performing a weighted summation of each type of characteristic distance based on the weighting coefficient, and finally determining the value obtained as the characteristic distance between two individual cells in the battery set.

[0011] In the embodiments of this application, determining the characteristic distance between two individual cells in the battery set based on the state data of each individual cell in the battery set includes calculating the characteristic distance d(i,j) according to formula (1):

[0012]

[0013] Where p is the number of battery features used to measure distance, and x ip Let x be the eigenvalue of the p-th characteristic of the first individual cell in the battery set. jp Let h represent the eigenvalue of the p-th feature of the second individual cell in the battery set, and h represent the coefficient of the feature distance. The value of h will change depending on the type of feature distance. The values ​​of h are 1, 2, and ∞, which correspond to Manhattan distance, Euclidean distance, and Chebyshev distance, respectively.

[0014] In the embodiments of this application, the operating state includes at least a start-up state, a fabrication state, a normal transport state, and an emergency stop state. The state data and operating state of each target cell are sequentially input into a trained Long Short-Term Memory (LSTM) artificial neural network to determine the fault type of each target cell. This includes: for each target cell, extracting the original features of its state data and operating state; performing feature interaction on any two original features to obtain new features; merging the original features with the new features and performing normalization processing after feature selection; and inputting the processed features into the trained LTM artificial neural network to determine the fault type of each target cell.

[0015] In the embodiments of this application, the long short-term memory artificial neural network is an LSTM neural network. The hidden layer of the LSTM neural network includes a first hidden layer and a second hidden layer. The number of LSTM neurons in the first hidden layer is the number of delay variables. The second hidden layer contains 8 neurons and uses the ReLU activation function. Finally, Softmax is used to classify the fault types. The number of output nodes is the same as the number of fault types, corresponding to each fault type.

[0016] In embodiments of this application, the method further includes a training step for a long short-term memory artificial neural network. The training step includes: acquiring the historical operating status of battery packs of multiple engineering vehicles and the historical status data of individual cells in each battery pack; extracting historical features from the historical operating status and historical status data; determining the historical fault types of individual cells with faults, and adding fault type labels to each historical feature according to the historical fault types; classifying the historical features into delayed variables and non-delayed variables; performing feature interaction between delayed variables and non-delayed variables to obtain new delayed variables and non-delayed variables; inputting the delayed variables into a first hidden layer to train the long short-term memory artificial neural network; and inputting the non-delayed variables and the output of the first hidden layer into a second hidden layer to train the long short-term memory artificial neural network.

[0017] In embodiments of this application, the engineering vehicle includes a display device, and the method further includes: sending fault type, operating status and / or status data to the display device for display.

[0018] In the embodiments of this application, the engineering vehicle is a pump truck.

[0019] A second aspect of this application provides a processor configured to perform the above-described method for determining the type of fault in a single battery cell.

[0020] A third aspect of this application provides an apparatus for determining the type of fault in a single battery cell, comprising:

[0021] Multiple sensors are installed in the battery pack to collect status data of each individual cell in the battery pack;

[0022] A communication module, installed in the battery pack, is used to transmit status data and the operating status of the engineering vehicle; and

[0023] The processor mentioned above.

[0024] A fourth aspect of this application provides an engineering vehicle, characterized in that it includes the aforementioned device for determining the type of fault in a single battery cell.

[0025] A fifth aspect of this application provides a machine-readable storage medium storing instructions that, when executed by a processor, configure the processor to perform the aforementioned method for determining the type of fault in a single battery cell.

[0026] By considering the operating status of new energy engineering vehicles during fault diagnosis, the above technical solution avoids misdiagnosis caused by state changes. After locating a single battery cell fault, an improved neural network is used to analyze the fault type, further determining the fault type and facilitating subsequent battery repair. This method determines the fault type based on identifying the single cell fault, extending battery life and improving the efficiency of subsequent repairs. Incorporating historical power battery data and vehicle operating status into the neural network training data also improves the accuracy of fault diagnosis.

[0027] Other features and advantages of the embodiments of this application will be described in detail in the following detailed description section. Attached Figure Description

[0028] The accompanying drawings are provided to further illustrate the embodiments of this application and form part of the specification. They are used together with the following detailed description to explain the embodiments of this application, but do not constitute a limitation on the embodiments of this application. In the drawings:

[0029] Figure 1 The schematic diagram illustrates a process flow diagram of a method for determining the fault type of a single cell according to an embodiment of this application;

[0030] Figure 2 A flowchart illustrating the recursive feature elimination process according to an embodiment of this application is shown schematically.

[0031] Figure 3 This illustration schematically shows an LSTM fault diagnosis network structure diagram according to an embodiment of the present application;

[0032] Figure 4The schematic diagram illustrates a process flow diagram of a method for determining the fault type of a single cell according to another embodiment of this application;

[0033] Figure 5 This schematic diagram illustrates a framework of a power battery fault diagnosis system according to an embodiment of this application.

[0034] Figure 6 The diagram illustrates the internal structure of a computer device according to an embodiment of this application. Detailed Implementation

[0035] To make the objectives, technical solutions, and advantages of the embodiments of this application clearer, the technical solutions of the embodiments of this application will be clearly and completely described below with reference to the accompanying drawings. It should be understood that the specific embodiments described herein are only for illustration and explanation of the embodiments of this application and are not intended to limit the embodiments of this application. All other embodiments obtained by those skilled in the art based on the embodiments of this application without creative effort are within the scope of protection of this application.

[0036] Figure 1 The illustration schematically shows a flowchart of a method for determining the fault type of a single battery cell according to an embodiment of this application. Figure 1 As shown in one embodiment of this application, a method for determining the fault type of a single battery cell is provided, comprising the following steps:

[0037] Step 101: Collect the operating status of the engineering vehicle and the status data of each individual cell in the battery pack within a preset time period. The battery pack includes multiple individual cells.

[0038] Step 102: Determine the target cell in the battery pack that has failed based on the operating status and status data.

[0039] Step 103: The state data and operating status of each target cell are sequentially input into the trained Long Short-Term Memory (LSTM) artificial neural network to determine the fault type of each target cell.

[0040] The above solution is applied to new energy engineering vehicles, where each vehicle includes at least one battery pack, specifically a power battery pack. First, the operating status of the engineering vehicle and the status data of each individual cell within each battery pack can be collected over a preset time period. Taking a new energy pump truck as an example, the operating status can include four states: start-up, material placement, normal transportation, and emergency stop. Different operating states cause changes in the power battery parameters of the engineering vehicle. This solution incorporates the operating status of the new energy pump truck into the fault diagnosis process, reducing the false diagnosis rate. For a single battery pack, each pack can include multiple individual cells. The status data of each individual cell can include at least the cell's voltage, current, temperature, temperature change rate, state of charge, and internal resistance. Furthermore, communication devices can be installed on the relevant power battery packs to transmit and store the collected status data in real time.

[0041] In this solution, voltage sensors, current sensors, and temperature sensors can be installed in the power battery pack to collect the state data of each individual cell in the new energy pump truck's power battery. After acquiring the operating status of the engineering vehicle and the state data of each individual cell, the target cell in the battery pack that has failed can be identified based on these two sets of data. Then, the state data of each target cell and the operating status of the engineering vehicle can be sequentially input into a trained Long Short-Term Memory (LSTM) artificial neural network, which can then determine the fault type of each target cell. The LSTM artificial neural network can also be referred to as an LSTM neural network.

[0042] In one embodiment, determining the target cell that has failed in the battery pack based on the operating status and status data includes: forming a battery set by any two cells in the battery pack; determining the characteristic distance between two cells in the battery set based on the status data of each cell in the battery set; determining the target battery set whose characteristic distance is greater than a preset distance threshold; and determining the cell included in at least two target battery sets as the target cell.

[0043] In this embodiment, a faulty individual battery cell in the power battery pack can be located based on distance calculation. First, any two individual batteries in the battery pack can be grouped into a battery set to determine the characteristic distance between any two individual batteries. Then, for each battery set, the characteristic distance between two individual batteries in that set can be determined based on the state data of each individual battery cell in that set. After obtaining the characteristic distance, target battery sets with characteristic distances greater than a preset distance threshold can be selected, thus identifying individual batteries included in at least two target battery sets as target individual batteries. For example, suppose there are target battery sets A, B, and C with characteristic distances greater than the preset distance threshold. If individual battery X1 is included in sets A, B, and C, then it can be determined that individual battery X1 is faulty, and thus individual battery X1 can be identified as the faulty target individual battery. It can be seen that in this embodiment, only individual batteries that appear multiple times in the target battery sets are counted, or it can be considered that only the individual battery that appears most frequently in the target battery sets is counted and identified as the faulty target individual battery. The characteristic distance between two individual batteries is calculated by quantifying the state data of individual batteries to determine the characteristic distance between any two individual batteries.

[0044] In one embodiment, determining the characteristic distance between two individual cells in the battery set based on the state data of each individual cell in the battery set includes: determining the Manhattan distance, Euclidean distance, and Chebyshev distance between two individual cells in the battery set based on the state data of each individual cell in the battery set; determining the weighting coefficient corresponding to each type of characteristic distance; and performing a weighted summation of each type of characteristic distance based on the weighting coefficient, and finally determining the value obtained as the characteristic distance between two individual cells in the battery set.

[0045] The feature distance includes at least three types of distances: Manhattan distance, Euclidean distance, and Chebyshev distance. For any battery set, the Manhattan distance, Euclidean distance, and Chebyshev distance between two individual cells in the battery set can be determined based on the state data of each individual cell in the battery set. Specifically, in one embodiment, the feature distance d(i,j) between two individual cells in each battery set under each type can be calculated according to the following formula (1):

[0046]

[0047] Where p is the number of battery features used to measure distance, and x ip Let x be the eigenvalue of the p-th characteristic of the first individual cell in the battery set. jThis represents the second individual cell in the battery set. h represents the coefficient of the feature distance. The value of h will change depending on the type of feature distance. The values ​​of h are 1, 2, and ∞, which correspond to Manhattan distance, Euclidean distance, and Chebyshev distance, respectively.

[0048] For example, suppose we need to calculate the Manhattan distance between two batteries, then the value of h is 1, and we can consider x to be... i1 The voltage of the first battery, x i2 The current of the first battery, x i3 The temperature of the first battery, etc., can be used to calculate the Manhattan distance between any two individual batteries in a battery set. Furthermore, an iterative calculation method can be used to calculate the distance relationships between individual batteries. For example, with 8 individual batteries, battery 1 is first grouped into its corresponding battery set with batteries 2-8, and the distance between battery 1 and batteries 2-8 is calculated. Then, the distance between battery 2 and batteries 3-8 is calculated, and this process is repeated until the calculation is complete.

[0049] For any set of batteries, after calculating the Manhattan distance d1, Euclidean distance d2, and Chebyshev distance d3 between any two individual batteries in the set, the three types of characteristic distances can be weighted and summed based on the weighting coefficients of each type of characteristic distance. The final characteristic distance d is then calculated using the following formula (2):

[0050] d=λ1d1+λ2d2+λ3d3 (2)

[0051] Where the weighting coefficients for d1, d2, and d3 are λ1, λ2, and λ3, respectively. Specifically, the values ​​of λ1, λ2, and λ3 depend on the dispersion of the distance data set after h takes different values. This scheme uses variance to measure the dispersion of the distance, then:

[0052]

[0053]

[0054] Where n is the total distance between the corresponding individual cells, and x is the average distance of the distance group. i For the i-th distance value, σ 2 Let σ1, σ2, and σ3 be the variances. σ1, σ2, and σ3 correspond to the standard deviations of d1, d2, and d3, respectively.

[0055] In one embodiment, the state data and operating status of each target cell are sequentially input into a trained Long Short-Term Memory (LSTM) artificial neural network to determine the fault type of each target cell. This includes: extracting the original features of the state data and operating status of each target cell; performing feature interaction on any two original features to obtain new features; merging the original features with the new features and performing normalization processing after feature selection; and inputting the processed features into the trained LSM artificial neural network to determine the fault type of each target cell.

[0056] After identifying the target cell that failed in each battery pack, the state data and operational features of each target cell can be extracted. For ease of description, these features can be referred to as the original features. To increase feature richness, improve the model's expressive power, and uncover the hidden relationship between feature quantities and failure modes, this scheme performs feature interaction based on the features. Let the original feature set be X = [x1, x2, ..., x...]. n Based on this, feature interaction is performed. Any two original features can be multiplied to obtain a new feature set X′. That is, X′=[x1x2, x1x3, ..., x...]. n-1 x n Then, the original features can be merged with the new features, and normalization processing can be performed after feature selection. The processed features are then input into a trained Long Short-Term Memory (LSTM) artificial neural network to determine the fault type of each target cell. Further, as... Figure 2 As shown, after merging features, all features can be sorted according to their weights, and feature selection is performed using RFE (Recursive Feature Elimination). Specifically, the evaluator is first trained on the initial feature set. Then, weights are assigned to the features and sorted based on the training results. Finally, features with the lowest weights are pruned. If the number of features does not meet the set expectation, this process is recursively executed; otherwise, the features after feature selection are output. The normalization process is as follows:

[0057]

[0058] Where, x n The data is normalized, and x is the original data. max x min These represent the maximum and minimum values ​​in the dataset, respectively. After normalization, feature selection is performed to reduce overfitting and improve the model's generality.

[0059] In one embodiment, the Long Short-Term Memory (LSTM) artificial neural network is an LSTM neural network. The LSTM neural network has a first hidden layer and a second hidden layer. The number of LSTM neurons in the first hidden layer is equal to the number of delay variables. The second hidden layer contains 8 neurons and uses the ReLU activation function. Finally, Softmax is used to classify the fault types. The number of output nodes is the same as the number of fault types, corresponding to each fault type. For example, if there are 5 output nodes, it means there are 5 fault types.

[0060] like Figure 3 As shown, n is the number of delayed variables and m is the number of non-delayed variables. Delayed variables are used as the input of the LSTM unit in the first hidden layer, and non-delayed variables are used directly as the input of the second hidden layer. Finally, the probability of each fault type is calculated by Softmax to determine the fault type.

[0061] In one embodiment, the method described above for determining the fault type of a single battery cell further includes a training step for a long short-term memory artificial neural network, such as... Figure 4 As shown, the training steps include:

[0062] Step 401: Obtain the historical operating status of the battery packs of multiple engineering vehicles and the historical status data of individual cells in each battery pack;

[0063] Step 402: Extract historical features of historical operating status and historical status data;

[0064] Step 403: Determine the historical fault types of the individual cells with faults, and add a fault type label to each historical feature according to the historical fault types;

[0065] Step 404: Classify historical features into delayed variables and non-delayed variables;

[0066] Step 405: Perform feature interaction on the delayed variables and non-delayed variables to obtain new delayed variables and non-delayed variables;

[0067] Step 406: Input the delay variable into the first hidden layer to train the long short-term memory artificial neural network;

[0068] Step 407: Input the non-delayed variables and the output of the first hidden layer into the second hidden layer to train the long short-term memory artificial neural network.

[0069] Before officially deploying a Long Short-Term Memory (LSTM) artificial neural network, it needs to be trained. Once trained, the fault type of each target battery cell can be predicted. Specifically, the historical operating states of battery packs from multiple engineering vehicles and the historical state data of individual cells in each battery pack under each historical operating state can be obtained. Then, historical features of the historical operating states and the historical state data under each historical operating state are extracted. For example, for engineering vehicle N1 in the laying state, data such as voltage, current, temperature, temperature change rate, state of charge, and internal resistance of each individual cell in the battery pack are collected. The historical state data of each individual cell in engineering vehicle N1 under the laying state is then used as a set of sample data to determine if any individual cells have failed and the historical fault type of those individual cells. Since these sample data are historical data, fault type labels can be added to each historical feature of the individual cell based on its known fault types. For example, the first set of sample data for engineering vehicle N1 is the historical state data of each individual cell in the laying state, and cell number 2 in this set of sample data has failed, with the fault type being thermal runaway. Then, based on each feature included in the first set of data for engineering vehicle N1, a fault type label corresponding to the fault type can be added.

[0070] Furthermore, historical features can be categorized into delayed variables and non-delayed variables. Delayed variables are features that are not strongly correlated with time, while non-delayed variables are features that are strongly correlated with time, such as temperature and the rate of temperature change. Then, delayed and non-delayed variables can be interacted to obtain new delayed and non-delayed variables. Feature interaction can refer to multiplying any two delayed variables with the non-delayed variables to obtain new delayed and non-delayed variables. Specifically, delayed variable × non-delayed variable = delayed variable. If it is a delayed variable, it can be directly input into the first hidden layer of the Long Short-Term Memory (LSTM) artificial neural network to train the LSM; if it is a delayed variable, it can be directly input into the second hidden layer of the LSM to train the LSM, thus extending the network's training time. Furthermore, when training the Long Short-Term Memory (LSTM) artificial neural network by inputting each historical feature, the operating state can be one-hot encoded. The codes corresponding to the four states (start-up state, fabric laying state, normal transport state, and emergency stop state) are [1,0,0,0], [0,1,0,0], [0,0,1,0], and [0,0,0,1], respectively. The output nodes of the LTM artificial neural network can also be one-hot encoded. Taking the five fault types as an example, the specific details are shown in Table 1 below:

[0071] Table 1 Fault Types and Causes

[0072]

[0073] Therefore, when the Long Short-Term Memory (LSTM) artificial neural network outputs predictions of fault types for individual battery cells, it can also output corresponding codes, which can then be used to determine the corresponding fault type. The rate of temperature change reflects the stability of the power battery operation, and its calculation method is as follows:

[0074]

[0075] Where Δt is the time difference between time 1 and time 2, and T1 and T2 are the temperatures at time 1 and time 2, respectively.

[0076] In one embodiment, the engineering vehicle described above includes a display device, and the method further includes: sending fault type, operating status and / or status data to the display device for display.

[0077] Furthermore, the display device can utilize Spring Cloud microservice technology to achieve agile development and deployment, supporting the display of information such as vehicle operating status, power battery parameters, fault types, and fault repair suggestions, while also supporting high concurrency. Alternatively, a web-based visual interface can be designed using the MVC (Model-View-Control) framework. This interface primarily displays the operating status of the engineering vehicle and relevant power battery parameters; if a fault is detected, it can also display the fault type and fault repair suggestions.

[0078] By considering the operating status of new energy engineering vehicles during fault diagnosis, the above technical solution avoids misdiagnosis caused by state changes. A distance-based calculation method is used for preliminary diagnosis of the new energy power battery and fault location of individual battery cells, fully utilizing various battery characteristics to improve the accuracy of individual cell fault detection. After locating the individual battery cell fault, an improved neural network is used to analyze the fault type, further determining the fault type and facilitating subsequent battery repair. Finally, a visual display is provided via Web and microservices, allowing users to more intuitively observe the vehicle's operating status and the power battery's status. This method determines the fault type based on identifying individual cell faults, extending battery life and improving the efficiency of subsequent repairs. Incorporating historical power battery data and vehicle operating status into the neural network training data also improves the accuracy of fault diagnosis.

[0079] like Figure 5 As shown, in one embodiment, a data acquisition module is installed on the battery pack of the engineering vehicle to collect status data of each individual cell in the battery pack. The data acquisition module may include devices such as voltage sensors, current sensors, and temperature sensors. The collected status data is transmitted in real time via a communication device installed on the power battery pack to a processor. The processor monitors the status data for preliminary battery diagnosis. If the status data of the individual cell indicates that the battery is normal, it is determined to be a normal battery; otherwise, it is determined to be a faulty individual cell.

[0080] Next, the state data and operating status of each target battery cell can be input into a trained Long Short-Term Memory (LSTM) artificial neural network to determine the fault type of each target battery cell. Then, the battery fault type and corresponding repair suggestions can be displayed using a microservice module and a web presentation module.

[0081] Figure 1 This is a flowchart illustrating a method for determining the fault type of a single battery cell in one embodiment. It should be understood that, although... Figure 1 The steps in the flowchart are shown sequentially as indicated by the arrows, but these steps are not necessarily executed in the order indicated by the arrows. Unless otherwise specified herein, there is no strict order in which these steps are executed, and they can be performed in other orders. Figure 1 At least some of the steps in the process may include multiple sub-steps or multiple stages. These sub-steps or stages are not necessarily completed at the same time, but can be executed at different times. The execution order of these sub-steps or stages is not necessarily sequential, but can be executed in turn or alternately with other steps or at least some of the sub-steps or stages of other steps.

[0082] This application provides a storage medium storing a program that, when executed by a processor, implements the method described above for determining the fault type of a single battery cell.

[0083] This application provides a processor for running a program, wherein the program executes the above-described method for determining the fault type of a single battery cell.

[0084] In one embodiment, this application also provides an apparatus for determining the fault type of a single battery cell, comprising:

[0085] Multiple sensors are installed in the battery pack to collect status data of each individual cell in the battery pack;

[0086] A communication module, installed in the battery pack, is used to transmit status data and the operating status of the engineering vehicle; and

[0087] Based on the processor described above.

[0088] The system includes multiple sensors, such as voltage sensors, current sensors, and temperature sensors, to collect status data of each individual cell in the power battery of the new energy pump truck. The communication module can include 4G or 5G communication devices for real-time transmission and storage of the collected status data. The processor, based on the status data transmitted by the communication module and the operating status of the engineering vehicle, identifies the target cell in the battery pack that has malfunctioned. It then sequentially inputs the status data and operating status of each target cell into a trained Long Short-Term Memory (LSTM) artificial neural network to determine the fault type of each target cell.

[0089] The processor contains a kernel, which retrieves the corresponding program unit from memory. One or more kernels can be configured, and adjusting kernel parameters allows for different methods used to determine the type of fault in a single battery cell.

[0090] The memory may include non-permanent memory in computer-readable media, such as random access memory (RAM) and / or non-volatile memory, such as read-only memory (ROM) or flash RAM, and the memory includes at least one memory chip.

[0091] In one embodiment, a computer device is provided, which may be a server, and its internal structure diagram may be as follows: Figure 6 As shown. The computer device includes a processor A01, a network interface A02, a memory (not shown), and a database (not shown) connected via a system bus. The processor A01 provides computing and control capabilities. The memory includes internal memory A03 and a non-volatile storage medium A04. The non-volatile storage medium A04 stores an operating system B01, a computer program B02, and a database (not shown). The internal memory A03 provides an environment for the operation of the operating system B01 and the computer program B02 in the non-volatile storage medium A04. The database stores the vehicle's operating status and the status data of each individual battery cell in the battery pack. The network interface A02 communicates with external terminals via a network connection. When executed by the processor A01, the computer program B02 implements a method for determining the type of fault in an individual battery cell.

[0092] Those skilled in the art will understand that Figure 6 The structure shown is merely a block diagram of a portion of the structure related to the present application and does not constitute a limitation on the computer device to which the present application is applied. Specific computer devices may include more or fewer components than those shown in the figure, or combine certain components, or have different component arrangements.

[0093] This application provides an apparatus including a processor, a memory, and a program stored in the memory and executable on the processor. When the processor executes the program, it implements the steps of a method for determining the fault type of a single cell.

[0094] This application also provides a computer program product that, when executed on a data processing device, is adapted to perform a program that initializes a method with the following steps for determining the fault type of a single cell battery.

[0095] Those skilled in the art will understand that embodiments of this application can be provided as methods, systems, or computer program products. Therefore, this application can take the form of a completely hardware embodiment, a completely software embodiment, or an embodiment combining software and hardware aspects. Furthermore, this application can take the form of a computer program product embodied on one or more computer-usable storage media (including but not limited to disk storage, CD-ROM, optical storage, etc.) containing computer-usable program code.

[0096] This application is described with reference to flowchart illustrations and / or block diagrams of methods, apparatus (systems), and computer program products according to embodiments of this application. It will be understood that each block of the flowchart illustrations and / or block diagrams, and combinations of blocks in the flowchart illustrations and / or block diagrams, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, special-purpose computer, embedded processor, or other programmable data processing apparatus to produce a machine, such that the instructions, which execute via the processor of the computer or other programmable data processing apparatus, generate instructions for implementing the flowchart... Figure 1 One or more processes and / or boxes Figure 1 A device that provides the functions specified in one or more boxes.

[0097] These computer program instructions may also be stored in a computer-readable storage medium that can direct a computer or other programmable data processing device to function in a particular manner, such that the instructions stored in the computer-readable storage medium produce an article of manufacture including instruction means, which are implemented in a process Figure 1 One or more processes and / or boxes Figure 1 The function specified in one or more boxes.

[0098] These computer program instructions may also be loaded onto a computer or other programmable data processing equipment to cause a series of operational steps to be performed on the computer or other programmable equipment to produce a computer-implemented process, thereby providing instructions that execute on the computer or other programmable equipment for implementing the process. Figure 1 One or more processes and / or boxes Figure 1 The steps of the function specified in one or more boxes.

[0099] In a typical configuration, a computing device includes one or more processors (CPU), input / output interfaces, network interfaces, and memory.

[0100] Memory may include non-persistent memory in computer-readable media, such as random access memory (RAM) and / or non-volatile memory, such as read-only memory (ROM) or flash RAM. Memory is an example of computer-readable media.

[0101] Computer-readable media includes both permanent and non-permanent, removable and non-removable media that can store information using any method or technology. Information can be computer-readable instructions, data structures, modules of programs, or other data. Examples of computer storage media include, but are not limited to, phase-change memory (PRAM), static random access memory (SRAM), dynamic random access memory (DRAM), other types of random access memory (RAM), read-only memory (ROM), electrically erasable programmable read-only memory (EEPROM), flash memory or other memory technologies, CD-ROM, digital versatile optical disc (DVD) or other optical storage, magnetic tape, magnetic magnetic disk storage or other magnetic storage devices, or any other non-transferable medium that can be used to store information accessible by a computing device. As defined herein, computer-readable media does not include transient computer-readable media, such as modulated data signals and carrier waves.

[0102] It should also be noted that the terms "comprising," "including," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such process, method, article, or apparatus. Unless otherwise specified, an element defined by the phrase "comprising one..." does not exclude the presence of other identical elements in the process, method, article, or apparatus that includes that element.

[0103] The above are merely embodiments of this application and are not intended to limit the scope of this application. Various modifications and variations can be made to this application by those skilled in the art. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of this application should be included within the scope of the claims of this application.

Claims

1. A method for determining the fault type of a single battery cell, characterized in that, Applied to engineering vehicles, the engineering vehicles including at least one battery pack, the method includes: The system collects the operating status of the engineering vehicle within a preset time period and the status data of each individual battery cell in the battery pack. The battery pack includes multiple individual batteries. The operating status includes at least the start-up status, material laying status, normal transportation status, and emergency stop status. The status data includes at least the voltage, current, temperature, temperature change rate, state of charge, and internal resistance of each individual battery cell. Based on the operating status and the status data, the target single battery cell in the battery pack that has failed is determined; The state data and operating state of each target cell are sequentially input into a trained long short-term memory artificial neural network to determine the fault type of each target cell through the trained long short-term memory artificial neural network. The step of determining the target cell in the battery pack that has failed based on the operating status and the status data includes: Any two individual cells in the battery pack can be combined into a battery assembly. Based on the state data of each individual cell in the battery set, determine the Manhattan distance, Euclidean distance, and Chebyshev distance between two individual cells in the battery set. Determine the weighting coefficients corresponding to each type of feature distance; The feature distances of each type are weighted and summed according to the weighting coefficients, and the resulting value is finally determined as the feature distance between two individual cells in the battery set. Identify the target battery set whose feature distance is greater than a preset distance threshold; The target single cell is defined as a cell included in at least two target cell sets.

2. The method for determining the fault type of a single battery cell according to claim 1, characterized in that, Determining the characteristic distance between two individual cells in the battery set based on the state data of each individual cell in the battery set includes calculating the characteristic distance according to formula (1). : （1） Where p is the number of battery features used to measure distance. Let p be the eigenvalue of the p-th characteristic of the first individual cell in the battery set. Let h represent the eigenvalue of the p-th feature of the second individual cell in the battery set, and h represent the coefficient of the feature distance. The value of h will change depending on the type of feature distance. The values ​​of h are 1, 2, and ∞, which correspond to Manhattan distance, Euclidean distance, and Chebyshev distance, respectively.

3. The method for determining the fault type of a single battery cell according to claim 1, characterized in that, The operating states include at least the start-up state, the fabric-laying state, the normal transportation state, and the emergency stop state. The step of sequentially inputting the state data and operating state of each target cell into a trained Long Short-Term Memory (LSTM) artificial neural network to determine the fault type of each target cell through the trained LSM artificial neural network includes: For each target cell, extract the state data and the original features of the operating state of the target cell; Interact with any two original features to obtain new features; The original features are merged with the new features, and normalization is performed after feature selection. The processed features are input into a trained long short-term memory artificial neural network to determine the fault type of each target cell.

4. The method for determining the fault type of a single battery cell according to claim 1, characterized in that, The Long Short-Term Memory (LSTM) artificial neural network is a type of LSTM neural network. The hidden layers of the LSTM neural network include a first hidden layer and a second hidden layer. The number of LSTM neurons in the first hidden layer is equal to the number of delay variables. The second hidden layer contains 8 neurons and uses the ReLU activation function. Finally, Softmax is used to classify the fault types. The number of output nodes is the same as the number of fault types, corresponding to each fault type.

5. The method for determining the fault type of a single battery cell according to claim 4, characterized in that, The method further includes a training step for a long short-term memory artificial neural network, the training step comprising: Acquire the historical operating status of battery packs of multiple engineering vehicles and the historical status data of individual cells in each battery pack; Extract the historical operating states and the historical features of the historical state data; Identify the historical fault types of individual cells that have faults, and add a fault type tag to each historical feature based on the historical fault types; The historical features are classified into delayed variables and non-delayed variables. By performing feature interaction between the delayed variable and the non-delayed variable, new delayed variables and non-delayed variables are obtained; The delay variable is input into the first hidden layer to train the long short-term memory artificial neural network; The non-delayed variables and the output of the first hidden layer are input into the second hidden layer to train the long short-term memory artificial neural network.

6. The method for determining the fault type of a single battery cell according to claim 1, characterized in that, The engineering vehicle includes a display device, and the method further includes: The fault type, the operating status, and / or the status data are sent to the display device for display.

7. The method for determining the fault type of a single battery cell according to claim 1, characterized in that, The engineering vehicle in question is a pump truck.

8. A processor, characterized in that, It is configured to perform the method for determining the fault type of a single cell as described in any one of claims 1 to 7.

9. An apparatus for determining the type of fault in a single battery cell, characterized in that, include: Multiple sensors are installed in the battery pack to collect state data of each individual cell in the battery pack. A communication module, installed in the battery pack, is used to transmit the status data and the operating status of the engineering vehicle; as well as The processor according to claim 8.

10. An engineering vehicle, characterized in that, Includes the apparatus for determining the type of fault in a single battery cell as described in claim 9.

11. The engineering vehicle according to claim 10, characterized in that, The engineering vehicle in question is a pump truck.

12. A machine-readable storage medium storing instructions thereon, characterized in that, When executed by a processor, this instruction causes the processor to be configured to perform a method for determining the type of fault in a single cell according to any one of claims 1 to 7.

Images