Lithium ion battery internal short circuit diagnosis method and system based on incremental capacity curve

Abstract

The invention discloses a lithium ion battery internal short circuit diagnosis method and system based on an incremental capacity curve, and the method comprises the steps: obtaining internal short circuit data of a lithium ion battery, carrying out the data preprocessing, and obtaining a preprocessed lithium ion battery internal short circuit data set; a multi-head self-attention mechanism is introduced, and a double-branch deep learning model is constructed; and inputting the pre-processed lithium ion battery internal short circuit data set into a double-branch deep learning model for data diagnosis processing to obtain lithium ion battery internal short circuit fault degree grade data. By using the method and the system, the ISC diagnosis of the lithium ion battery can be carried out by carrying out feature enhancement in combination with a multi-head self-attention mechanism, and the judgment precision of the short circuit in the lithium ion battery is improved. The lithium ion battery internal short circuit diagnosis method and system based on the incremental capacity curve can be widely applied to the technical field of lithium ion battery internal short circuit diagnosis.

Description

Technical Field

[0001] The present invention relates to the technical field of lithium ion battery internal short circuit diagnosis, and in particular to a lithium ion battery internal short circuit diagnosis method and system based on an incremental capacity curve. Background Art

[0002] Lithium-ion batteries have become the power source of the electric vehicle industry due to their excellent performance. However, they also have certain potential challenges, one of which is thermal runaway (TR). Internal short circuit (ISC) is considered to be one of the main causes of battery thermal runaway accidents. The process of ISC failure in lithium-ion batteries is mostly a long process, which can be roughly divided into early, middle and late stages. Among them, ISC failure in the middle and late stages has already caused damage to battery performance, especially in the late stage, when the battery has a large area short circuit, generating a lot of heat instantly, which can easily lead to TR. This stage lasts for a very short time. Therefore, in order to improve the safety of lithium-ion batteries, it is necessary to detect ISC failures in batteries and conduct graded warnings in the early stages of failure.

[0003] At present, the ISC diagnosis technology of lithium-ion batteries can be mainly summarized as: model-based and data-driven diagnosis methods. The core of the model-based method is to build an accurate model to predict and diagnose faults. However, given the inherent complex nonlinear characteristics of lithium-ion batteries, it is extremely challenging to build an accurate model. The high accuracy of the model is often accompanied by a more complex model structure, which inevitably increases the complexity of the calculation and resource consumption. The advantage of the data-driven method is that it does not rely on an in-depth understanding of the battery physical process, nor does it require an accurate physical model. When there is sufficient data, the data-driven method can provide performance that is superior to traditional models. The current data-driven method mainly performs ISC diagnosis by constructing characteristics such as voltage, current and electrochemical impedance spectroscopy between normal batteries and ISC batteries. However, the resistance of early ISC faults is usually in the order of hundreds of ohms, which leads to a high similarity between the voltage characteristics of early ISC fault batteries and normal batteries. Using voltage data for early ISC diagnosis may affect the accuracy of the diagnostic results. Therefore, the existing technology is not accurate in judging the internal short circuit of lithium-ion batteries. Summary of the invention

[0004] In order to solve the above technical problems, the purpose of the present invention is to provide a lithium-ion battery internal short circuit diagnosis method and system based on an incremental capacity curve, which can perform ISC diagnosis of lithium-ion batteries by combining a multi-head self-attention mechanism to enhance features and improve the judgment accuracy of lithium-ion batteries internal short circuits.

[0005] The first technical solution adopted by the present invention is: a method for diagnosing internal short circuit of a lithium-ion battery based on an incremental capacity curve, comprising the following steps:

[0006] Acquire internal short circuit data of lithium-ion batteries and perform data preprocessing to obtain a preprocessed lithium-ion battery internal short circuit data set;

[0007] Introduce a multi-head self-attention mechanism to build a dual-branch deep learning model;

[0008] The preprocessed lithium-ion battery internal short-circuit data set is input into the dual-branch deep learning model for data diagnosis and processing to obtain the lithium-ion battery internal short-circuit fault degree data.

[0009] Further, the step of obtaining the internal short circuit data of the lithium ion battery and performing data preprocessing to obtain a preprocessed lithium ion battery internal short circuit data set specifically includes:

[0010] Obtaining charging capacity data and charging voltage data of the lithium-ion battery;

[0011] The relationship between the charging power data of the lithium-ion battery and the charging voltage data of the lithium-ion battery is calculated to obtain the change in the capacity of the lithium-ion battery with the voltage;

[0012] According to the change of lithium-ion battery capacity with voltage, a relationship curve between lithium-ion battery capacity increment and voltage is constructed and a local weighted scatter plot smoothing process is performed to obtain a smooth IC curve;

[0013] The smooth IC curve is converted by Gram angle and field to obtain a two-dimensional image of the relationship between the capacity increment and voltage of the lithium-ion battery;

[0014] The preprocessed lithium-ion battery internal short-circuit data set is constructed by combining the two-dimensional image of the relationship between the lithium-ion battery capacity increment and voltage and the smoothed IC curve.

[0015] Furthermore, the calculation expression of the lithium-ion battery capacity variation with voltage is specifically as follows:

[0016]

[0017] In the above formula, Q c Indicates the charge capacity of the lithium-ion battery, V c Indicates the charging voltage of lithium-ion batteries, Q c,1 Indicates the power of the lithium-ion battery at the first moment, Q c,2 Indicates the power of the lithium-ion battery at the second moment, ΔQ c Indicates the change in the charge capacity of a lithium-ion battery, ΔV c Indicates the change in charging voltage of a lithium-ion battery.

[0018] Furthermore, the dual-branch deep learning model includes a convolution module, a gated recurrent unit module, a feature fusion layer, a multi-head attention layer, a fully connected layer and an output layer, wherein the output end of the convolution module is connected to the first input end of the feature fusion layer, the output end of the gated recurrent unit module is connected to the second input end of the feature fusion layer, the output end of the feature fusion layer is connected to the input end of the multi-head attention layer, the output end of the multi-head attention layer is connected to the input end of the fully connected layer, and the output end of the fully connected layer is connected to the input end of the output layer.

[0019] Furthermore, the convolution module includes an image input layer, a convolution layer, a pooling layer and a flattening layer, wherein the image input layer, the convolution layer, the pooling layer and the flattening layer are connected in sequence.

[0020] Furthermore, the step of inputting the preprocessed lithium-ion battery internal short-circuit data set into the dual-branch deep learning model for data diagnosis processing to obtain lithium-ion battery internal short-circuit fault degree level data specifically includes:

[0021] The preprocessed lithium-ion battery internal short-circuit dataset is input into the dual-branch deep learning model;

[0022] Based on the convolution module of the dual-branch deep learning model, the smooth IC curve is subjected to feature extraction processing to obtain the IC curve feature vector;

[0023] Based on the gated recurrent unit module of the dual-branch deep learning model, feature extraction is performed on the two-dimensional image of the relationship between the capacity increment and voltage of the lithium-ion battery to obtain a two-dimensional feature image;

[0024] Based on the feature fusion layer of the dual-branch deep learning model, the IC curve feature vector and the two-dimensional feature image are fused to obtain the fused feature vector of the relationship between the capacity increment and voltage of the lithium-ion battery;

[0025] Based on the multi-head attention layer of the dual-branch deep learning model, the fused feature vector of the relationship between the capacity increment and the voltage of the lithium-ion battery is subjected to multi-head attention feature fusion processing to obtain the multi-head feature vector of the relationship between the capacity increment and the voltage of the lithium-ion battery;

[0026] Based on the fully connected layer of the dual-branch deep learning model, the multi-head feature vector of the relationship between the capacity increment and the voltage of the lithium-ion battery is classified and processed to obtain the level data of the short-circuit fault degree of the lithium-ion battery to be output;

[0027] Based on the output layer of the dual-branch deep learning model, the lithium-ion battery internal short-circuit fault degree level data to be output is output to obtain the lithium-ion battery internal short-circuit fault degree level data.

[0028] Furthermore, the convolution module based on the dual-branch deep learning model performs feature extraction processing on the smooth IC curve to obtain the IC curve feature vector, which specifically includes:

[0029] The smoothed IC curve is input into the convolution module of the two-branch deep learning model;

[0030] Based on the image input layer of the convolution module, a smooth IC curve is obtained;

[0031] Based on the convolution layer of the convolution module, the smooth IC curve is convolved to obtain the convolved IC curve;

[0032] Based on the pooling layer of the convolution module, the convolution IC curve is pooled to obtain the pooled IC curve;

[0033] Based on the flat layer of the convolution module, the IC curve after pooling is transformed to obtain the IC curve feature vector.

[0034] The second technical solution adopted by the present invention is: a lithium-ion battery internal short circuit diagnosis system based on an incremental capacity curve, comprising:

[0035] A preprocessing module is used to obtain the internal short circuit data of the lithium-ion battery and perform data preprocessing to obtain a preprocessed lithium-ion battery internal short circuit data set;

[0036] A building block for introducing a multi-head self-attention mechanism and building a dual-branch deep learning model;

[0037] The diagnostic module is used to input the preprocessed lithium-ion battery internal short-circuit data set into the dual-branch deep learning model for data diagnosis and processing to obtain the lithium-ion battery internal short-circuit fault degree level data.

[0038] The beneficial effects of the method and system of the present invention are as follows: the present invention obtains the internal short-circuit data of the lithium-ion battery and performs data preprocessing, further introduces a multi-head self-attention mechanism, constructs a dual-branch deep learning model, adopts CNN and GRU dual branches to extract features respectively, combines the multi-head self-attention mechanism to perform feature enhancement to perform ISC diagnosis of the lithium-ion battery, extracts the image features of the incremental capacity curve through CNN, extracts the time series features of the incremental capacity curve through GRU, and fuses the two modal features, distinguishes the features between tiny ISC batteries and normal batteries, and improves the judgment accuracy of internal short circuits in lithium-ion batteries. BRIEF DESCRIPTION OF THE DRAWINGS

[0039] Figure 1 It is a flow chart of the steps of the lithium ion battery internal short circuit diagnosis method based on the incremental capacity curve of the present invention;

[0040] Figure 2 It is a structural block diagram of a lithium-ion battery internal short circuit diagnosis system based on an incremental capacity curve of the present invention;

[0041] Figure 3 It is a schematic diagram of a dual-branch deep learning model diagnostic process provided by a specific embodiment of the present invention;

[0042] Figure 4 It is a schematic diagram of the structure of a dual-branch deep learning model provided by a specific embodiment of the present invention;

[0043] Figure 5 It is a schematic diagram of a partial incremental capacity curve provided by a specific embodiment of the present invention;

[0044] Figure 6 It is a schematic diagram of the diagnosis result of short circuit in lithium battery provided by a specific embodiment of the present invention. DETAILED DESCRIPTION

[0045] The present invention is further described in detail below in conjunction with the accompanying drawings and specific embodiments. The step numbers in the following embodiments are only provided for the convenience of explanation and description, and the order between the steps is not limited in any way. The execution order of each step in the embodiment can be adaptively adjusted according to the understanding of those skilled in the art.

[0046] Reference Figure 1 and Figure 3 The present invention provides a method for diagnosing internal short circuit of a lithium ion battery based on an incremental capacity curve, the method comprising the following steps:

[0047] S100, acquiring internal short-circuit data of a lithium-ion battery and performing data preprocessing to obtain a preprocessed lithium-ion battery internal short-circuit data set;

[0048] Specifically, the charging power data and the charging voltage data of the lithium-ion battery are obtained; the relationship between the charging power data and the charging voltage data of the lithium-ion battery is calculated to obtain the change of the capacity of the lithium-ion battery with the voltage; the relationship curve between the capacity increment and the voltage of the lithium-ion battery is constructed according to the change of the capacity of the lithium-ion battery with the voltage and a local weighted scatter plot smoothing process is performed to obtain a smooth IC curve; the smoothed IC curve is converted by Gram angle and field to obtain a two-dimensional image of the relationship between the capacity increment and the voltage of the lithium-ion battery; and the two-dimensional image of the relationship between the capacity increment and the voltage of the lithium-ion battery and the smoothed IC curve are combined to construct a pre-processed lithium-ion battery internal short circuit data set.

[0049] In the embodiment of the present invention, the original data set is to extract the charging power and charging voltage from the collected lithium-ion battery data set, calculate the capacity change with voltage, obtain the relationship curve between capacity increment and voltage, and then use the local weighted scatter plot smoothing method to obtain a smooth IC curve. Some IC curves are as follows: Figure 5 As shown, the smoothed IC curve is converted into a two-dimensional image using the Gram angle and field, and the original data set consists of the smoothed IC curve and the two-dimensional image.

[0050] Among them, the calculation expression of the change of lithium-ion battery capacity with voltage is as follows:

[0051]

[0052] In the above formula, Q c Indicates the charge capacity of the lithium-ion battery, V c Indicates the charging voltage of lithium-ion batteries, Q c,1 Indicates the power of the lithium-ion battery at the first moment, Q c,2 Indicates the power of the lithium-ion battery at the second moment, ΔQ c Indicates the change in the charge capacity of a lithium-ion battery, ΔV c Indicates the change in charging voltage of a lithium-ion battery.

[0053] S200, introduce multi-head self-attention mechanism to build a dual-branch deep learning model;

[0054] Specifically, the dual-branch deep learning model includes a convolution module, a gated recurrent unit module, a feature fusion layer, a multi-head attention layer, a fully connected layer and an output layer, wherein the output end of the convolution module is connected to the first input end of the feature fusion layer, the output end of the gated recurrent unit module is connected to the second input end of the feature fusion layer, the output end of the feature fusion layer is connected to the input end of the multi-head attention layer, the output end of the multi-head attention layer is connected to the input end of the fully connected layer, and the output end of the fully connected layer is connected to the input end of the output layer.

[0055] The convolution module includes an image input layer, a convolution layer, a pooling layer and a flattening layer, wherein the image input layer, the convolution layer, the pooling layer and the flattening layer are connected in sequence.

[0056] In a specific embodiment of the present invention, the CNN-GRU-MHSA (dual-branch deep learning model) fusion method is mainly composed of an input layer, three CNN layers, a GRU layer, a feature fusion layer, an MHSA layer, a fully connected layer and an output layer; wherein the number of convolution kernels of the three convolution layers is 32, 64, and 128 respectively, the kernel size of each convolution layer is the same, which is 3×3, the stride is 2×2, and the activation function is ReLU. In order to adjust the size of the convolution, a maximum pooling layer with a kernel size of 3×3 and a stride of 2×2 is used. After the convolution layer and the pooling layer, the feature map is flattened into a one-dimensional vector. The GRU layer has 128 hidden units and the activation function is tanh. The GRU layer is used to capture the dynamic changes and trends in the one-dimensional data of the IC curve. The MHSA layer uses 8 self-attention heads to enable the method to learn different weight distributions and key information, thereby improving the expression and prediction performance of the method. The fully connected layer uses categorical cross entropy as a loss function to output the final level of short-circuit fault in the lithium-ion battery.

[0057] S300, inputting the preprocessed lithium-ion battery internal short-circuit data set into the dual-branch deep learning model for data diagnosis and processing to obtain lithium-ion battery internal short-circuit fault degree level data.

[0058] Specifically, the preprocessed lithium-ion battery internal short-circuit data set is input into the dual-branch deep learning model; based on the convolution module of the dual-branch deep learning model, the smoothed IC curve is subjected to feature extraction processing to obtain the IC curve feature vector; based on the gated cyclic unit module of the dual-branch deep learning model, the feature extraction processing is performed on the two-dimensional image of the relationship between the capacity increment and the voltage of the lithium-ion battery to obtain the two-dimensional feature image; based on the feature fusion layer of the dual-branch deep learning model, the feature fusion processing is performed on the IC curve feature vector and the two-dimensional feature image to obtain the fused feature vector of the relationship between the capacity increment and the voltage of the lithium-ion battery; The multi-head attention layer of the dual-branch deep learning model performs multi-head attention feature fusion processing on the fused feature vector of the relationship between the lithium-ion battery capacity increment and the voltage, and obtains the multi-head feature vector of the relationship between the lithium-ion battery capacity increment and the voltage; based on the fully connected layer of the dual-branch deep learning model, the multi-head feature vector of the relationship between the lithium-ion battery capacity increment and the voltage is classified and processed to obtain the lithium-ion battery internal short-circuit fault degree level data to be output; based on the output layer of the dual-branch deep learning model, the lithium-ion battery internal short-circuit fault degree level data to be output is output to obtain the lithium-ion battery internal short-circuit fault degree level data.

[0059] Among them, the smooth IC curve is input into the convolution module of the dual-branch deep learning model; based on the image input layer of the convolution module, a smooth IC curve is obtained; based on the convolution layer of the convolution module, the smooth IC curve is convolved to obtain the convolved IC curve; based on the pooling layer of the convolution module, the convolved IC curve is pooled to obtain the pooled IC curve; based on the flattening layer of the convolution module; the pooled IC curve is converted to obtain the IC curve feature vector.

[0060] In a specific embodiment of the present invention, the CNN-GRU-MHSA fusion method adopts a CNN and GRU dual-branch parallel approach, such as Figure 4 As shown, the GRU branch inputs the data of the smoothed lithium-ion battery IC curve into the GRU layer to extract its feature vector; the CNN branch inputs the two-dimensional image of the lithium-ion battery IC curve converted by the GASF into the CNN layer to extract its features. Then, the feature vectors extracted by the two branches are connected and fused, and the fused feature vectors are fed to the MHSA layer and processed by multiple self-attention heads. Finally, the degree of short-circuit fault in the lithium-ion battery is output through the fully connected layer.

[0061] Finally, the data annotation method is shown in Table 1, and the ISC lithium battery failure degree and normal lithium batteries are divided into 6 categories.

[0062] Table 1 Data numbering method

[0063] serial number resistance Number Description 1 10-50Ω Ⅰ 2 51-250Ω Ⅱ 3 251-500Ω Ⅲ 4 501-750Ω Ⅳ 5 751-1000Ω Ⅴ 6 normal nor

[0064] In summary, the confusion matrix of the method of the present invention in the test data set under the initial SOC of 0% charging condition is as follows: Figure 6 As shown. The vertical axis in the figure corresponds to the actual ISC fault degree of the test sample and the normal battery. For example, the normal label corresponds to "nor" and the ISC degree of levels Ⅰ-Ⅴ, corresponding to the "10-50Ω", "51-250Ω", "251-500Ω", "501-750Ω" and "751-1000Ω" ISC equivalent resistance intervals. The horizontal axis represents the result of the test sample recognition. From the results in the figure, it can be seen that CNN-GRU-MHSA has only 9 misjudged samples and completely and accurately identifies the two ISC severity levels Ⅰ and Ⅱ, and there are only a small number of misjudged samples in adjacent ISC levels. It can be seen that the ISC diagnosis accuracy of the method in the present invention is 92.5%.

[0065] Reference Figure 2 , a lithium-ion battery internal short circuit diagnosis system based on incremental capacity curve, including:

[0066] A preprocessing module 201 is used to obtain the internal short circuit data of the lithium ion battery and perform data preprocessing to obtain a preprocessed lithium ion battery internal short circuit data set;

[0067] A construction module 202 is used to introduce a multi-head self-attention mechanism and construct a dual-branch deep learning model;

[0068] The diagnosis module 203 is used to input the preprocessed lithium-ion battery internal short-circuit data set into the dual-branch deep learning model for data diagnosis processing to obtain the lithium-ion battery internal short-circuit fault degree level data.

[0069] The contents of the above method embodiments are all applicable to the present system embodiments. The functions specifically implemented by the present system embodiments are the same as those of the above method embodiments, and the beneficial effects achieved are also the same as those achieved by the above method embodiments.

[0070] The above is a specific description of the preferred implementation of the present invention, but the invention is not limited to the embodiments. Those skilled in the art may make various equivalent modifications or substitutions without violating the spirit of the present invention. These equivalent modifications or substitutions are all included in the scope defined by the claims of this application.

Claims

1. A method for diagnosing internal short circuit in a lithium-ion battery based on an incremental capacity curve, characterized in that: The following steps are involved: Acquire internal short circuit data of lithium-ion batteries and perform data preprocessing to obtain a preprocessed lithium-ion battery internal short circuit data set; Introduce a multi-head self-attention mechanism to build a dual-branch deep learning model; The preprocessed lithium-ion battery internal short-circuit data set is input into the dual-branch deep learning model for data diagnosis and processing to obtain the lithium-ion battery internal short-circuit fault degree data.

2. The method for diagnosing internal short circuit in a lithium-ion battery based on an incremental capacity curve according to claim 1, characterized in that: The step of obtaining the internal short-circuit data of the lithium-ion battery and performing data preprocessing to obtain a preprocessed lithium-ion battery internal short-circuit data set specifically includes: Obtaining charging capacity data and charging voltage data of the lithium-ion battery; The relationship between the charging power data of the lithium-ion battery and the charging voltage data of the lithium-ion battery is calculated to obtain the change in the capacity of the lithium-ion battery with the voltage; According to the change of lithium-ion battery capacity with voltage, a relationship curve between lithium-ion battery capacity increment and voltage is constructed and a local weighted scatter plot smoothing process is performed to obtain a smooth IC curve; The smooth IC curve is converted by Gram angle and field to obtain a two-dimensional image of the relationship between the capacity increment and voltage of the lithium-ion battery; The preprocessed lithium-ion battery internal short-circuit data set is constructed by combining the two-dimensional image of the relationship between the lithium-ion battery capacity increment and voltage and the smoothed IC curve.

3. The method for diagnosing internal short circuit in a lithium-ion battery based on an incremental capacity curve according to claim 2, characterized in that: The calculation expression of the lithium-ion battery capacity variation with voltage is specifically as follows: In the above formula, Q c Indicates the charge capacity of the lithium-ion battery, V c Indicates the charging voltage of lithium-ion batteries, Q c,1 Indicates the power of the lithium-ion battery at the first moment, Q c,2 Indicates the power of the lithium-ion battery at the second moment, ΔQ c Indicates the change in the charge capacity of a lithium-ion battery, ΔV c Indicates the change in charging voltage of a lithium-ion battery.

4. The method for diagnosing internal short circuit in a lithium-ion battery based on an incremental capacity curve according to claim 3, characterized in that: The dual-branch deep learning model includes a convolution module, a gated recurrent unit module, a feature fusion layer, a multi-head attention layer, a fully connected layer and an output layer, wherein the output end of the convolution module is connected to the first input end of the feature fusion layer, the output end of the gated recurrent unit module is connected to the second input end of the feature fusion layer, the output end of the feature fusion layer is connected to the input end of the multi-head attention layer, the output end of the multi-head attention layer is connected to the input end of the fully connected layer, and the output end of the fully connected layer is connected to the input end of the output layer.

5. The method for diagnosing internal short circuit in a lithium-ion battery based on an incremental capacity curve according to claim 4, characterized in that: The convolution module includes an image input layer, a convolution layer, a pooling layer and a flattening layer, wherein the image input layer, the convolution layer, the pooling layer and the flattening layer are connected in sequence.

6. The method for diagnosing internal short circuit in a lithium-ion battery based on an incremental capacity curve according to claim 5, characterized in that: The step of inputting the preprocessed lithium-ion battery internal short-circuit data set into the dual-branch deep learning model for data diagnosis processing to obtain lithium-ion battery internal short-circuit fault degree level data specifically includes: The preprocessed lithium-ion battery internal short-circuit dataset is input into the dual-branch deep learning model; Based on the convolution module of the dual-branch deep learning model, the smooth IC curve is subjected to feature extraction processing to obtain the IC curve feature vector; Based on the gated recurrent unit module of the dual-branch deep learning model, feature extraction is performed on the two-dimensional image of the relationship between the capacity increment and voltage of the lithium-ion battery to obtain a two-dimensional feature image; Based on the feature fusion layer of the dual-branch deep learning model, the IC curve feature vector and the two-dimensional feature image are fused to obtain the fused feature vector of the relationship between the capacity increment and voltage of the lithium-ion battery; Based on the multi-head attention layer of the dual-branch deep learning model, the fused feature vector of the relationship between the capacity increment and the voltage of the lithium-ion battery is subjected to multi-head attention feature fusion processing to obtain the multi-head feature vector of the relationship between the capacity increment and the voltage of the lithium-ion battery; Based on the fully connected layer of the dual-branch deep learning model, the multi-head feature vector of the relationship between the capacity increment and the voltage of the lithium-ion battery is classified and processed to obtain the level data of the short-circuit fault degree of the lithium-ion battery to be output; Based on the output layer of the dual-branch deep learning model, the lithium-ion battery internal short-circuit fault degree level data to be output is output to obtain the lithium-ion battery internal short-circuit fault degree level data.

7. The method for diagnosing internal short circuit of a lithium-ion battery based on an incremental capacity curve according to claim 6, characterized in that: The convolution module based on the dual-branch deep learning model performs feature extraction processing on the smooth IC curve to obtain the IC curve feature vector, which specifically includes: The smoothed IC curve is input into the convolution module of the two-branch deep learning model; Based on the image input layer of the convolution module, a smooth IC curve is obtained; Based on the convolution layer of the convolution module, the smooth IC curve is convolved to obtain the convolved IC curve; Based on the pooling layer of the convolution module, the convolution IC curve is pooled to obtain the pooled IC curve; based on the flattening layer of the convolution module, the pooled IC curve is converted to obtain the IC curve feature vector.

8. A lithium-ion battery internal short circuit diagnosis system based on an incremental capacity curve, characterized in that: Includes the following modules: A preprocessing module is used to obtain the internal short circuit data of the lithium-ion battery and perform data preprocessing to obtain a preprocessed lithium-ion battery internal short circuit data set; A building block for introducing a multi-head self-attention mechanism and building a dual-branch deep learning model; The diagnostic module is used to input the preprocessed lithium-ion battery internal short-circuit data set into the dual-branch deep learning model for data diagnosis and processing to obtain the lithium-ion battery internal short-circuit fault degree level data.

Images