New energy vehicle battery performance detection system and method
Through deep learning technology, the cycle life feature vector of lithium iron phosphate batteries is extracted and decoded and regressed using a decoder, which solves the problems of detection time and data randomness in the existing technology, and realizes the accurate prediction of the cycle life of lithium iron phosphate batteries , optimized battery management and performance.
Patent Information
- Application Number
- CN202410498597.0
- Authority / Receiving Office
- CN · China
- Patent Type
- Patents(China)
- Current Assignee / Owner
- Filing Date
- 2024-04-24
- Publication Date
- 2024-12-06
- Estimated Expiration
- 2044-04-24
AI Technical Summary
The prior art is difficult to accurately detect the cycle life of lithium iron phosphate batteries. The detection method takes a long time and the data is very random, so it cannot truly reflect the changes in the battery structure and physical and chemical properties.
Deep learning technology is adopted to obtain the tomographic images and electrochemical characteristics information of lithium iron phosphate batteries, and the cycle life characteristic vector is extracted, and decoded regression is used to accurately predict the cycle life of the battery.
It realizes accurate prediction of the cycle life of lithium iron phosphate batteries, provides an important reference, and optimizes battery usage and performance for the battery management system.
Smart Images

Figure CN118362898B_ABST
Abstract
Description
Technical Field
[0001] The present application relates to the field of intelligent detection of new energy vehicle performance, and more specifically, to a new energy vehicle battery performance detection system and method. Background Art
[0002] Power batteries play a vital role in new energy electric vehicles, directly affecting the performance of the vehicle and the development prospects of the electric vehicle industry. According to electric vehicle theory, the charge and discharge current of the power battery will fluctuate with changes in vehicle driving conditions and driver behavior. In addition, the power battery will also be affected by external factors such as temperature changes and vibrations, which will directly affect the power output and service life of the power battery.
[0003] Therefore, real-time monitoring of the status of power batteries is crucial, which is one of the key technologies to ensure the efficient and safe operation of power batteries and electric vehicles. However, the detection and prediction of the cycle life of power batteries (such as lithium iron phosphate batteries) is still one of the challenges facing the power battery field worldwide.
[0004] To detect the cycle life of lithium iron phosphate power batteries, the relationship between the structural morphology of the battery components and their performance should be determined by comprehensive measurement and analysis of the battery's structural morphology and physical and chemical properties. However, the general method for determining the battery cycle life is to perform charge and discharge cycle tests on the battery, which is not only difficult to truly reflect the changes in the structural morphology and physical and chemical properties of the battery components, but also takes a long time to detect, is time-consuming, and has high randomness in the data.
[0005] Therefore, we look forward to a new energy vehicle battery performance testing system and method that uses deep learning technology to analyze and predict the lithium iron phosphate battery to be tested, so as to achieve accurate prediction of the cycle life of new energy vehicle batteries. Summary of the invention
[0006] In order to solve the above technical problems, the present application is proposed. The embodiment of the present application provides a new energy vehicle battery performance detection system and method, which first obtains the tomographic images of multiple predetermined time points of the lithium iron phosphate battery to be detected under different simulated working conditions and the electrochemical characteristic information of multiple predetermined time points of the lithium iron phosphate battery to be detected under different simulated working conditions as input data, and then uses deep learning technology to extract and analyze these input data to obtain the cycle life feature vector of the lithium iron phosphate battery, and then uses a decoder to decode and regress the lithium iron phosphate battery cycle life feature vector to obtain a decoded value for representing the cycle life of the lithium iron phosphate battery to be detected. In this way, according to the decoded value, the cycle life of the lithium iron phosphate battery can be accurately predicted. In this way, it can provide an important reference for the battery management system, which helps to optimize the use and performance of the battery.
[0007] According to the first aspect of the present application, a new energy vehicle battery performance detection system is provided, which includes:
[0008] A lithium iron phosphate battery data acquisition module, used to obtain tomographic images of a lithium iron phosphate battery to be tested at multiple predetermined time points under different simulated working conditions and electrochemical characteristic information of a lithium iron phosphate battery to be tested at multiple predetermined time points under different simulated working conditions;
[0009] A lithium iron phosphate battery feature acquisition module, used to extract and analyze the tomographic images of the lithium iron phosphate battery to be detected at multiple predetermined time points under the different simulated working conditions and the electrochemical characteristic information of the lithium iron phosphate battery to be detected at multiple predetermined time points under the different simulated working conditions to obtain a tomographic global feature vector and an electrochemical information global feature matrix;
[0010] A lithium iron phosphate battery feature fusion module, used for multiplying the tomography global feature vector and the electrochemical information global feature matrix to obtain a lithium iron phosphate battery cycle life feature vector;
[0011] The lithium iron phosphate battery feature decoding module is used to obtain a decoding value based on the lithium iron phosphate battery cycle life feature vector, and the decoding value is used to represent the cycle life of the lithium iron phosphate battery to be detected.
[0012] In combination with the first aspect of the present application, in a new energy vehicle battery performance detection system of the first aspect of the present application, the lithium iron phosphate battery feature acquisition module includes: a tomographic feature extraction unit, which is used to feature encode the tomographic images of multiple predetermined time points of the lithium iron phosphate battery to be detected under the different simulated working conditions to obtain the tomographic global feature vector; an electrochemical information feature extraction unit, which is used to feature encode the electrochemical characteristic information of the lithium iron phosphate battery to be detected under the different simulated working conditions at multiple predetermined time points to obtain the electrochemical information global feature matrix.
[0013] Among them, the tomographic feature extraction unit includes: a depth feature extraction subunit, which is used to pass the tomographic images of multiple predetermined time points of the lithium iron phosphate battery to be tested under the different simulated working conditions through a tomographic feature extractor based on a depth feature fusion module to obtain multiple tomographic feature matrices under each simulated working condition; a tomographic timing feature extraction subunit, which is used to arrange the multiple tomographic feature matrices under each simulated working condition into tomographic timing input tensors under each simulated working condition according to the time dimension, and then pass them through a tomographic timing feature extraction module based on a three-dimensional convolutional neural network model to obtain a tomographic timing feature vector under each simulated working condition; a tomographic global feature extraction subunit, which is used to arrange the tomographic timing feature vectors under each simulated working condition into a tomographic global input vector, and then pass them through a tomographic global feature extraction module based on a feature encoder to obtain the tomographic global feature vector.
[0014] In particular, the shallow-depth feature extraction subunit is used to: extract a shallow feature matrix from the i-th layer of the tomographic feature extractor, the i-th layer being the first to sixth layers of the tomographic feature extractor; extract a deep feature matrix from the j-th layer of the tomographic feature extractor, the ratio between the j-th layer and the i-th layer being greater than or equal to 5; and use the shallow-depth feature fusion module to fuse the shallow feature matrix and the deep feature matrix to obtain each tomographic feature matrix in the multiple tomographic feature matrices under the various simulation conditions.
[0015] In addition, the electrochemical information feature extraction unit includes: an electrochemical information arrangement subunit, which is used to arrange the electrochemical characteristic information of multiple predetermined time points of the lithium iron phosphate battery to be tested under the different simulated working conditions according to the time dimension into multiple electrochemical information input vectors under each simulated working condition; an electrochemical feature extraction subunit, which is used to pass the multiple electrochemical information input vectors under each simulated working condition through an electrochemical information feature extraction module based on a one-dimensional convolutional neural network model to obtain multiple electrochemical information feature vectors under each simulated working condition; an electrochemical timing feature extraction subunit, which is used to pass the multiple electrochemical information feature vectors under each simulated working condition through an electrochemical information change feature extraction module based on a bidirectional long short-term memory neural network model to obtain electrochemical information change feature vectors under each simulated working condition; an electrochemical global feature extraction subunit, which is used to arrange the electrochemical information change feature vectors under each simulated working condition into an electrochemical information global input matrix and then pass it through an electrochemical information global feature extraction module containing a two-dimensional convolution structure to obtain the electrochemical information global feature matrix.
[0016] In particular, the electrochemical time series feature extraction subunit is used to: arrange the multiple electrochemical information feature vectors under the various simulated working conditions in one dimension to obtain a sequence of multiple long-distance correlation feature vectors; and use the electrochemical information change feature extraction module based on the bidirectional long short-term memory neural network model to perform long-distance correlation feature encoding on the sequence of the multiple long-distance correlation feature vectors to obtain the electrochemical information change feature vectors under the various simulated working conditions.
[0017] In combination with the first aspect of the present application, in a new energy vehicle battery performance detection system of the first aspect of the present application, the lithium iron phosphate battery feature decoding module includes: a lithium iron phosphate battery cycle life feature acquisition unit, which is used to pass the lithium iron phosphate battery cycle life feature vector through a lithium iron phosphate battery cycle life decoding feature extraction module based on a convolutional neural network model to obtain a lithium iron phosphate battery cycle life decoding feature vector; a lithium iron phosphate battery cycle life feature optimization unit, which is used to perform implicit group optimization with spatial sparsity constraints on the lithium iron phosphate battery cycle life decoding feature vector to obtain an optimized lithium iron phosphate battery cycle life decoding feature vector; a lithium iron phosphate battery cycle life decoding unit, which is used to pass the optimized lithium iron phosphate battery cycle life decoding feature vector through a decoder to obtain the decoding value.
[0018] The lithium iron phosphate battery cycle life feature optimization unit includes: calculating the implicit group optimization factor of the spatial sparsity restriction of the lithium iron phosphate battery cycle life decoding feature vector by the following formula, wherein the formula is:
[0019]
[0020] Among them, v c represents the cycle life decoding feature vector of lithium iron phosphate battery, σ(v c ) represents the variance of the eigenvalue set of the cycle life decoding eigenvector of the lithium iron phosphate battery, v i is the characteristic value of each position in the cycle life decoding feature vector of the lithium iron phosphate battery, and L is the length of the cycle life decoding feature vector of the lithium iron phosphate battery, and ||v c ||1 represents the norm of the lithium iron phosphate battery cycle life decoding feature vector, w is the implicit group optimization factor of the spatial sparsity restriction of the lithium iron phosphate battery cycle life decoding feature vector; and the lithium iron phosphate battery cycle life decoding feature vector is weighted with the implicit group optimization factor of the spatial sparsity restriction as the weight to obtain the optimized lithium iron phosphate battery cycle life decoding feature vector.
[0021] According to the second aspect of the present application, a new energy vehicle battery performance detection method is provided, which includes:
[0022] Obtaining tomographic images of a lithium iron phosphate battery to be tested at multiple predetermined time points under different simulated working conditions and electrochemical characteristic information of a lithium iron phosphate battery to be tested at multiple predetermined time points under different simulated working conditions;
[0023] Respectively extracting and analyzing the tomographic images at multiple predetermined time points of the lithium iron phosphate battery to be tested under the different simulated working conditions and the electrochemical characteristic information at multiple predetermined time points of the lithium iron phosphate battery to be tested under the different simulated working conditions to obtain a tomographic global feature vector and an electrochemical information global feature matrix;
[0024] Multiplying the tomography global feature vector and the electrochemical information global feature matrix to obtain a cycle life feature vector of the lithium iron phosphate battery;
[0025] A decoding value is obtained based on the cycle life feature vector of the lithium iron phosphate battery, and the decoding value is used to represent the cycle life of the lithium iron phosphate battery to be detected.
[0026] In combination with the second aspect of the present application, in a new energy vehicle battery performance detection method of the second aspect of the present application, feature extraction and analysis are performed on the tomographic images of multiple predetermined time points of the lithium iron phosphate battery to be detected under the different simulated working conditions and the electrochemical characteristic information of the lithium iron phosphate battery to be detected under the different simulated working conditions to obtain a tomographic global feature vector and an electrochemical information global feature matrix, including: a tomographic feature extraction unit, used to feature encode the tomographic images of multiple predetermined time points of the lithium iron phosphate battery to be detected under the different simulated working conditions to obtain the tomographic global feature vector; an electrochemical information feature extraction unit, used to feature encode the electrochemical characteristic information of the lithium iron phosphate battery to be detected under the different simulated working conditions to obtain the electrochemical information global feature matrix.
[0027] Compared with the prior art, the present application provides a new energy vehicle battery performance detection system and method, which first obtains the tomographic images of multiple predetermined time points of the lithium iron phosphate battery to be detected under different simulated working conditions and the electrochemical characteristic information of multiple predetermined time points of the lithium iron phosphate battery to be detected under different simulated working conditions as input data, and then uses deep learning technology to extract and analyze these input data to obtain the lithium iron phosphate battery cycle life feature vector, and then uses a decoder to decode and regress the lithium iron phosphate battery cycle life feature vector to obtain a decoded value for representing the cycle life of the lithium iron phosphate battery to be detected. In this way, according to the decoded value, the cycle life of the lithium iron phosphate battery can be accurately predicted. In this way, it can provide an important reference for the battery management system and help optimize the use and performance of the battery. BRIEF DESCRIPTION OF THE DRAWINGS
[0028] By describing the embodiments of the present application in more detail in conjunction with the accompanying drawings, the above and other purposes, features and advantages of the present application will become more apparent. The accompanying drawings are used to provide a further understanding of the embodiments of the present application and constitute a part of the specification. Together with the embodiments of the present application, they are used to explain the present application and do not constitute a limitation of the present application. In the accompanying drawings, the same reference numerals generally represent the same components or steps.
[0029] Figure 1 The figure illustrates a schematic block diagram of a new energy vehicle battery performance detection system according to an embodiment of the present application.
[0030] Figure 2 The figure illustrates a schematic block diagram of a lithium iron phosphate battery feature acquisition module in a new energy vehicle battery performance detection system according to an embodiment of the present application.
[0031] Figure 3 The figure illustrates a schematic block diagram of a tomographic feature extraction unit in a lithium iron phosphate battery feature acquisition module in a new energy vehicle battery performance detection system according to an embodiment of the present application.
[0032] Figure 4 The figure illustrates a schematic block diagram of an electrochemical information feature extraction unit in a lithium iron phosphate battery feature acquisition module in a new energy vehicle battery performance detection system according to an embodiment of the present application.
[0033] Figure 5 The figure illustrates a schematic block diagram of a lithium iron phosphate battery feature decoding module in a new energy vehicle battery performance detection system according to an embodiment of the present application.
[0034] Figure 6 The figure illustrates a flow chart of a new energy vehicle battery performance detection method according to an embodiment of the present application. DETAILED DESCRIPTION
[0035] Below, the exemplary embodiments according to the present application will be described in detail with reference to the accompanying drawings. Obviously, the described embodiments are only part of the embodiments of the present application, rather than all the embodiments of the present application, and it should be understood that the present application is not limited to the exemplary embodiments described here.
[0036] Exemplary Systems
[0037] Figure 1 FIG. 1 is a schematic block diagram of a new energy vehicle battery performance detection system according to an embodiment of the present application. Figure 1As shown, according to the embodiment of the present application, the new energy vehicle battery performance detection system 100 includes: a lithium iron phosphate battery data acquisition module 110, which is used to obtain the tomographic images of the lithium iron phosphate battery to be detected at multiple predetermined time points under different simulated working conditions and the electrochemical characteristic information of the lithium iron phosphate battery to be detected at multiple predetermined time points under different simulated working conditions; a lithium iron phosphate battery feature acquisition module 120, which is used to extract and analyze the tomographic images of the lithium iron phosphate battery to be detected under the different simulated working conditions and the electrochemical characteristic information of the lithium iron phosphate battery to be detected under the different simulated working conditions respectively to obtain a tomographic global feature vector and an electrochemical information global feature matrix; a lithium iron phosphate battery feature fusion module 130, which is used to multiply the tomographic global feature vector and the electrochemical information global feature matrix to obtain a lithium iron phosphate battery cycle life feature vector; a lithium iron phosphate battery feature decoding module 140, which is used to obtain a decoding value based on the lithium iron phosphate battery cycle life feature vector, and the decoding value is used to represent the cycle life of the lithium iron phosphate battery to be detected.
[0038] Lithium iron phosphate battery is a lithium-ion battery with the characteristics of high safety, stability and long cycle life. Compared with other types of lithium batteries, lithium iron phosphate battery performs better in safety and stability.
[0039] Lithium iron phosphate batteries are widely used in new energy vehicles, mainly as power batteries. Due to their high safety and stability, lithium iron phosphate batteries can provide reliable power support for electric vehicles, and at the same time have a long cycle life, which can meet the needs of electric vehicles for high-performance batteries.
[0040] However, the charge and discharge current of power batteries will fluctuate with changes in vehicle driving conditions and driver behavior. In addition, power batteries are also subject to external influences such as temperature changes and vibrations, which will directly affect the power output and service life of power batteries. Therefore, real-time monitoring of the status of power batteries is crucial, which is one of the key technologies to ensure the efficient and safe operation of power batteries and electric vehicles. At present, the detection and prediction of the cycle life of power batteries (such as lithium iron phosphate batteries) is still one of the challenges facing the power battery field worldwide.
[0041] To detect the cycle life of lithium iron phosphate power batteries, the relationship between the structural morphology of the battery components and their performance should be determined by comprehensive measurement and analysis of the battery's structural morphology and physical and chemical properties. However, the general method for judging the battery cycle life is to perform charge and discharge cycle tests on the battery, which is not only difficult to truly reflect the changes in the structural morphology and physical and chemical properties of the battery components, but also has a long detection time, is time-consuming, and has high data randomness. Therefore, a new energy vehicle battery performance detection system and method is expected to use deep learning technology to analyze and predict the lithium iron phosphate battery to be detected, so as to achieve accurate prediction of the cycle life of new energy vehicle batteries.
[0042] It should be understood that deep learning is a branch of artificial intelligence. It builds artificial neural networks by simulating the connection between neurons in the human brain to achieve complex pattern recognition and learning tasks. Deep learning technology has achieved great success in recent years, mainly due to its powerful processing capabilities and efficient use of large-scale data. Deep learning technology is widely used in computer vision, data processing, and natural language processing.
[0043] In the new energy vehicle battery performance detection system of the embodiment of the present application, deep learning technology plays an important role. For example, the deep learning model can automatically learn features from battery data, identify key influencing factors, and extract useful information. Therefore, in the specific embodiment of the present application, the input data is feature extracted and analyzed by deep learning technology.
[0044] In an embodiment of the present application, the lithium iron phosphate battery data acquisition module 110 is used to obtain tomographic images of multiple predetermined time points of the lithium iron phosphate battery to be detected under different simulated working conditions and electrochemical characteristic information of multiple predetermined time points of the lithium iron phosphate battery to be detected under different simulated working conditions. It should be understood that the tomographic image provides intuitive information about the internal structure and component distribution of the battery, which can help understand the structure, material distribution and possible defects or damage of the battery. By analyzing the tomographic image, structural changes or abnormal conditions can be discovered in time, providing important clues for cycle life prediction. Battery performance parameters, such as voltage, current, charge / discharge capacity, internal resistance and other electrochemical characteristic information reflect the battery performance, which is directly related to the working state and health of the battery. Comprehensively considering the internal structure, component distribution and electrochemical characteristics of the battery, a rich data source can be provided, which helps the model to better understand the working mechanism of the battery, thereby achieving accurate prediction of the cycle life of the lithium iron phosphate battery to be detected. Furthermore, considering that the performance of batteries under different working conditions will be different, by collecting data under multiple working conditions, the generalization ability of the deep learning model can be improved, so that it can better adapt to the battery performance prediction task under different working conditions, rather than being limited to the prediction of specific working conditions. Therefore, in order to more accurately predict the cycle life of lithium iron phosphate batteries, tomographic images (image information about the internal structure and component distribution of the battery) and electrochemical characteristic information (voltage, current, charge / discharge capacity, internal resistance, etc.) of the lithium iron phosphate battery to be tested under different simulated working conditions are collected as input data.
[0045] Specifically, when obtaining tomographic images of the lithium iron phosphate battery to be tested at multiple predetermined time points under different simulated working conditions, technologies such as X-ray CT scanning, magnetic resonance imaging, and infrared thermal imaging can be used. When collecting electrochemical characteristic information of the lithium iron phosphate battery to be tested at multiple predetermined time points under different simulated working conditions, a special electrochemical testing instrument, such as an electrochemical workstation or a charge and discharge test system, can be used to perform charge and discharge cycle tests on the battery. These testing instruments can record the electrochemical characteristic information of the battery at different time points, such as voltage, current, charge / discharge capacity, internal resistance, etc.
[0046] In the embodiment of the present application, the lithium iron phosphate battery feature acquisition module 120 is used to extract and analyze the tomographic images at multiple predetermined time points of the lithium iron phosphate battery to be detected under the different simulated working conditions and the electrochemical characteristic information at multiple predetermined time points of the lithium iron phosphate battery to be detected under the different simulated working conditions to obtain the tomographic global feature vector and the electrochemical information global feature matrix. It should be understood that after collecting these input data, the input data are further subjected to feature extraction and analysis tasks.
[0047] Specifically, Figure 2The figure shows a schematic block diagram of a lithium iron phosphate battery feature acquisition module in a new energy vehicle battery performance detection system according to an embodiment of the present application. Figure 2 As shown, the lithium iron phosphate battery feature acquisition module 120 includes: a tomographic feature extraction unit 121, which is used to feature encode the tomographic images of the lithium iron phosphate battery to be detected at multiple predetermined time points under the different simulated working conditions to obtain the tomographic global feature vector; an electrochemical information feature extraction unit 122, which is used to feature encode the electrochemical characteristic information of the lithium iron phosphate battery to be detected at multiple predetermined time points under the different simulated working conditions to obtain the electrochemical information global feature matrix.
[0048] First, feature extraction and analysis are performed on the tomographic images of the lithium iron phosphate battery to be tested at multiple predetermined time points under the different simulated working conditions. Specifically, Figure 3 The figure shows a schematic block diagram of a tomographic feature extraction unit in a lithium iron phosphate battery feature acquisition module in a new energy vehicle battery performance detection system according to an embodiment of the present application. Figure 3 As shown, the tomographic feature extraction unit 121 includes: a depth feature extraction subunit 121-1, which is used to pass the tomographic images of multiple predetermined time points of the lithium iron phosphate battery to be tested under the different simulated working conditions through a tomographic feature extractor based on a depth feature fusion module to obtain multiple tomographic feature matrices under each simulated working condition; a tomographic time series feature extraction subunit 121-2, which is used to arrange the multiple tomographic feature matrices under each simulated working condition into tomographic time series input tensors under each simulated working condition according to the time dimension, and then pass them through a tomographic time series feature extraction module based on a three-dimensional convolutional neural network model to obtain a tomographic time series feature vector under each simulated working condition; a tomographic global feature extraction subunit 121-3, which is used to arrange the tomographic time series feature vectors under each simulated working condition into a tomographic global input vector, and then pass them through a tomographic global feature extraction module based on a feature encoder to obtain the tomographic global feature vector.
[0049] It should be understood that the tomographic images of lithium iron phosphate batteries under different simulated working conditions contain not only shallow features such as texture, color, and edge, but also abstract features such as category and shape. Therefore, in order to more comprehensively capture the internal structure and component distribution characteristics of lithium iron phosphate batteries, a tomographic feature extractor based on a deep and shallow feature fusion module is used to perform feature extraction on the tomographic images of the lithium iron phosphate batteries to be tested at multiple predetermined time points under the different simulated working conditions to obtain multiple tomographic feature matrices under each simulated working condition.
[0050] In a specific embodiment of the present application, the shallow-depth feature extraction subunit 121-1 is used to: extract a shallow feature matrix from the i-th layer of the tomographic feature extractor, the i-th layer being the first to sixth layers of the tomographic feature extractor; extract a deep feature matrix from the j-th layer of the tomographic feature extractor, the ratio between the j-th layer and the i-th layer being greater than or equal to 5; and use the shallow-depth feature fusion module to fuse the shallow feature matrix and the deep feature matrix to obtain each tomographic feature matrix in the multiple tomographic feature matrices under the various simulation conditions.
[0051] Next, considering that the internal structure and component distribution characteristics of the lithium iron phosphate battery under each simulated working condition change over time, based on this, in order to better retain these time series change characteristics, the multiple tomographic feature matrices under each simulated working condition are first arranged according to the time dimension as tomographic time series input tensors under each simulated working condition. Then, a tomographic time series feature extraction module based on a three-dimensional convolutional neural network model is used to extract the time series change characteristics of the tomographic time series input tensors under each simulated working condition to obtain the tomographic time series feature vectors under each simulated working condition. Among them, the three-dimensional convolutional neural network model can simultaneously learn the features in the time series, space and channel dimensions, more comprehensively capture the feature information in the data, and help improve the performance and generalization ability of the model.
[0052] In a specific embodiment of the present application, the tomographic timing feature extraction subunit 121-2 is used to: use each layer of the three-dimensional convolutional neural network model to perform the following on the input data in the forward transfer of the layer: convolution processing on the input data to obtain a convolution feature map; pooling processing based on a feature matrix on the convolution feature map to obtain a pooled feature map; nonlinear activation on the pooled feature map to obtain an activation feature map; wherein the input of the first layer of the three-dimensional convolutional neural network model is the tomographic timing input tensor under the various simulation conditions, and the output of the last layer of the three-dimensional convolutional neural network model is the tomographic timing feature vector under the various simulation conditions.
[0053] Furthermore, considering that the internal structure and component distribution characteristics of lithium iron phosphate batteries may be different under different working conditions, a single analysis of the features under each working condition may affect the accuracy of feature extraction. Therefore, in order to comprehensively analyze the features under different working conditions, the tomography time series feature vectors under the various simulated working conditions are first arranged as a tomography global input vector. Then, a tomography global feature extraction module based on a feature encoder is used to perform global feature extraction on the tomography global input vector to obtain a tomography global feature vector. In this way, it is helpful to capture the internal structure and component distribution of lithium iron phosphate batteries more comprehensively and accurately. In this way, the battery performance under different working conditions can be fully and comprehensively considered, the generalization ability and prediction accuracy of the model can be improved, and the accurate prediction of the cycle life of new energy vehicle batteries can be achieved.
[0054] In a specific embodiment of the present application, the tomographic global feature extraction subunit 121-3 is used to: use the layers of the feature encoder to perform the following operations on the input data in the forward pass of the layer: use the convolution units of the layers of the feature encoder to perform convolution processing based on a one-dimensional convolution kernel on the input data to obtain a convolution feature map; use the pooling units of the layers of the feature encoder to perform pooling processing on the convolution feature map along the channel dimension to obtain a pooling feature map; and use the activation units of the layers of the feature encoder to perform nonlinear activation on the feature values of each position in the pooling feature map to obtain an activation feature map; wherein the output of the last layer of the feature encoder is the tomographic global feature vector.
[0055] Then, feature extraction and analysis are performed on the electrochemical characteristic information of the lithium iron phosphate battery to be tested at multiple predetermined time points under the different simulated working conditions. Specifically, Figure 4 The figure shows a schematic block diagram of an electrochemical information feature extraction unit in a lithium iron phosphate battery feature acquisition module in a new energy vehicle battery performance detection system according to an embodiment of the present application. Figure 4As shown, the electrochemical information feature extraction unit 122 includes: an electrochemical information arrangement subunit 122-1, which is used to arrange the electrochemical characteristic information of the lithium iron phosphate battery to be tested under the different simulated working conditions at multiple predetermined time points according to the time dimension into multiple electrochemical information input vectors under each simulated working condition; an electrochemical feature extraction subunit 122-2, which is used to pass the multiple electrochemical information input vectors under each simulated working condition through an electrochemical information feature extraction module based on a one-dimensional convolutional neural network model to obtain multiple electrochemical information feature vectors under each simulated working condition; an electrochemical time series feature extraction subunit 122-3, which is used to pass the multiple electrochemical information feature vectors under each simulated working condition through an electrochemical information change feature extraction module based on a bidirectional long short-term memory neural network model to obtain electrochemical information change feature vectors under each simulated working condition; an electrochemical global feature extraction subunit 122-4, which is used to arrange the electrochemical information change feature vectors under each simulated working condition into an electrochemical information global input matrix and then pass it through an electrochemical information global feature extraction module containing a two-dimensional convolution structure to obtain the electrochemical information global feature matrix.
[0056] It should be understood that electrochemical characteristics change over time, including the performance of the battery at different time points. By arranging the time series, the time series information in the data can be retained, so that the model can learn the law of battery performance changes over time. Therefore, in order to better retain this time series information, the electrochemical characteristic information of the lithium iron phosphate battery to be tested at multiple predetermined time points under the different simulation conditions is arranged according to the time dimension as multiple electrochemical information input vectors under each simulation condition. In this way, the electrochemical information input vector corresponding to each predetermined time point under each simulation condition can be obtained.
[0057] Next, considering that the one-dimensional convolutional neural network model can effectively capture local features when processing time series data, sliding the convolution kernel on the input sequence and extracting local features helps to identify important patterns and structures in the data. Based on this, an electrochemical information feature extraction module based on a one-dimensional convolutional neural network model is used to extract features of the multiple electrochemical information input vectors under the various simulated working conditions to obtain multiple electrochemical information feature vectors under the various simulated working conditions. In this way, support can be provided for the accurate prediction of the cycle life of new energy vehicle batteries.
[0058] In a specific embodiment of the present application, the electrochemical feature extraction subunit 122-2 is used to: use each layer of the one-dimensional convolutional neural network model to perform the following on the input data in the forward transfer of the layer: perform convolution processing based on a one-dimensional convolution kernel on the input data to obtain a convolution feature map; perform pooling processing based on a feature matrix on the convolution feature map to obtain a pooled feature map; perform nonlinear activation on the pooled feature map to obtain an activation feature map; wherein the input of the first layer of the one-dimensional convolutional neural network model is a plurality of electrochemical information input vectors under the respective simulation conditions, and the output of the last layer of the one-dimensional convolutional neural network model is a plurality of electrochemical information feature vectors under the respective simulation conditions.
[0059] Furthermore, it should be understood that the electrochemical information of lithium iron phosphate batteries changes dynamically in the time dimension, that is, the electrochemical information is time series data. The bidirectional long short-term memory neural network model, as a neural network structure specially designed to process time series data, can better capture the time series information and change patterns in the data. Therefore, in order to better capture the changing trends and patterns of electrochemical information in the time dimension, an electrochemical information change feature extraction module based on a bidirectional long short-term memory neural network model is used to extract and analyze the time series change features of multiple electrochemical information feature vectors under the various simulated working conditions to obtain the electrochemical information change feature vectors under each simulated working condition. In this way, the time series dependency in the input sequence can be captured, which helps to more comprehensively understand the changing trend of electrochemical information.
[0060] In a specific embodiment of the present application, the electrochemical timing feature extraction subunit 122-3 is used to: arrange the multiple electrochemical information feature vectors under the respective simulation conditions in one dimension to obtain a sequence of multiple long-distance correlation feature vectors; and use the electrochemical information change feature extraction module based on the bidirectional long short-term memory neural network model to perform long-distance correlation feature encoding on the sequence of the multiple long-distance correlation feature vectors to obtain the electrochemical information change feature vectors under the respective simulation conditions.
[0061] Furthermore, it should be understood that the dynamic changes of electrochemical information in the time dimension under different simulated working conditions may be different, and analyzing the electrochemical information under each simulated working condition separately may not fully reflect the performance and life changes of the battery under different working conditions. Therefore, in order to comprehensively analyze the telephone writing information under each simulated working condition, the characteristic vectors of the electrochemical information changes under the various simulated working conditions are first arranged into an electrochemical information global input matrix. In this way, the electrochemical information features at different time points can be arranged spatially, retaining the global pattern of the electrochemical information. Then, an electrochemical information global feature extraction module containing a two-dimensional convolution structure is used to perform a two-dimensional convolution operation on the electrochemical information global input matrix to obtain an electrochemical information global feature matrix. In this way, it is helpful to better understand the global patterns and structures in the data, thereby improving the ability to accurately predict the cycle life of the battery.
[0062] In a specific embodiment of the present application, the electrochemical global feature extraction subunit 122-4 is used to: use the layers of the electrochemical information global feature extraction module containing a two-dimensional convolution structure to perform the following operations on the input data in the forward transmission of the layer: use the convolution units of the layers of the electrochemical information global feature extraction module containing a two-dimensional convolution structure to perform convolution processing based on a two-dimensional convolution kernel on the input data to obtain a convolution feature map; use the pooling units of the layers of the electrochemical information global feature extraction module containing a two-dimensional convolution structure to perform pooling processing on the convolution feature map along the channel dimension to obtain a pooling feature map; and use the activation units of the layers of the electrochemical information global feature extraction module containing a two-dimensional convolution structure to perform nonlinear activation on the eigenvalues of each position in the pooling feature map to obtain an activation feature map; wherein the output of the last layer of the electrochemical information global feature extraction module containing a two-dimensional convolution structure is the electrochemical information global feature matrix.
[0063] In an embodiment of the present application, the lithium iron phosphate battery feature fusion module 130 is used to multiply the tomography global feature vector and the electrochemical information global feature matrix to obtain the lithium iron phosphate battery cycle life feature vector. It should be understood that there may be a certain complementarity between the tomography image information and the electrochemical characteristic information. Electrochemical characteristic information may more directly reflect the real-time performance of the battery, while tomography image information can provide more detailed information about the internal structure and component distribution of the battery. By fusing these two types of information, the complementarity between them can be comprehensively utilized to improve the ability to predict the battery cycle life. Therefore, in order to better analyze and predict the cycle life of the lithium iron phosphate battery, the tomography global feature vector and the electrochemical information global feature matrix are further feature fused to obtain the lithium iron phosphate battery cycle life feature vector.
[0064] In the embodiment of the present application, the lithium iron phosphate battery feature decoding module 140 is used to obtain a decoding value based on the lithium iron phosphate battery cycle life feature vector, and the decoding value is used to represent the cycle life of the lithium iron phosphate battery to be detected. Specifically, Figure 5 The figure shows a schematic block diagram of a lithium iron phosphate battery feature decoding module in a new energy vehicle battery performance detection system according to an embodiment of the present application. Figure 5 As shown, the lithium iron phosphate battery feature decoding module 140 includes: a lithium iron phosphate battery cycle life feature acquisition unit 141, which is used to pass the lithium iron phosphate battery cycle life decoding feature extraction module based on a convolutional neural network model to obtain a lithium iron phosphate battery cycle life decoding feature vector; a lithium iron phosphate battery cycle life feature optimization unit 142, which is used to perform implicit group optimization with spatial sparsity constraints on the lithium iron phosphate battery cycle life decoding feature vector to obtain an optimized lithium iron phosphate battery cycle life decoding feature vector; a lithium iron phosphate battery cycle life decoding unit 143, which is used to pass the optimized lithium iron phosphate battery cycle life decoding feature vector through a decoder to obtain the decoded value.
[0065] It should be understood that the feature vector obtained after feature fusion may contain a large amount of original information, but this information may not always be the most effective way to represent the key features of the battery cycle life. Through the feature extraction process of models such as convolutional neural networks, the input features can be further converted into higher-level, more abstract feature representations, thereby better capturing the complex relationships between features. Therefore, in order to capture higher-level and more representative abstract features, a lithium iron phosphate battery cycle life decoding feature extraction module based on a convolutional neural network model is used to perform deep convolution encoding on the lithium iron phosphate battery cycle life feature vector to obtain a lithium iron phosphate battery cycle life decoding feature vector.
[0066] In particular, in the technical solution of the present application, it is considered that there may be information loss in the feature extraction process of the tomographic image and the electrochemical characteristic information. If the key features, such as the relevant features of the cycle life, are not accurately captured and retained, the information of the decoded feature vector of the cycle life of the lithium iron phosphate battery will be incomplete and the degree of aggregation will be reduced. At the same time, considering that there are fewer or unevenly distributed features about the cycle life of the lithium iron phosphate battery in the data set, it will also lead to a low degree of information aggregation of the decoded feature vector. The low degree of information aggregation of the decoded feature vector of the cycle life of the lithium iron phosphate battery will result in the feature vector being unable to comprehensively and accurately characterize the cycle life characteristics of the lithium iron phosphate battery, which may affect the accuracy of the final cycle life prediction and distort the prediction results or make them unstable. The lack of feature vectors with high information aggregation may lead to the inability to accurately identify the key factors affecting the cycle life of the lithium iron phosphate battery, making it impossible to optimize and improve them in a targeted manner, limiting the possibility of improving battery performance. The low degree of information aggregation will affect the quality of the decoded feature vector of the cycle life of the lithium iron phosphate battery, causing the performance of the final cycle life prediction model to decline, affecting the use effect and reliability of the entire lithium iron phosphate battery. Therefore, in order to solve the problem of low information aggregation of the cycle life decoding feature vector of the lithium iron phosphate battery, in the technical solution of the present application, implicit group optimization with spatial sparsity restriction is performed on the cycle life decoding feature vector of the lithium iron phosphate battery.
[0067] In a specific embodiment of the present application, the lithium iron phosphate battery cycle life feature optimization unit 142 includes: calculating the implicit group optimization factor of the spatial sparsity restriction of the lithium iron phosphate battery cycle life decoding feature vector by the following formula, wherein the formula is:
[0068]
[0069] Among them, v c represents the cycle life decoding feature vector of lithium iron phosphate battery, σ(v c ) represents the variance of the eigenvalue set of the cycle life decoding eigenvector of the lithium iron phosphate battery, v i is the characteristic value of each position in the cycle life decoding feature vector of the lithium iron phosphate battery, and L is the length of the cycle life decoding feature vector of the lithium iron phosphate battery, and ||v c ||1 represents the norm of the lithium iron phosphate battery cycle life decoding feature vector, w is the implicit group optimization factor of the spatial sparsity restriction of the lithium iron phosphate battery cycle life decoding feature vector; and the lithium iron phosphate battery cycle life decoding feature vector is weighted with the implicit group optimization factor of the spatial sparsity restriction as the weight to obtain the optimized lithium iron phosphate battery cycle life decoding feature vector.
[0070] Here, the lithium iron phosphate battery cycle life decoding feature vector is subjected to implicit group optimization with spatial sparsity constraints, which evaluates the fitness of each particle in the lithium iron phosphate battery cycle life decoding feature vector in a high-dimensional feature space, and updates the lithium iron phosphate battery cycle life decoding feature vector according to the optimal solution of each particle in the lithium iron phosphate battery cycle life decoding feature vector and the group optimal solution to satisfy the updated lithium iron phosphate battery cycle life decoding feature vector to meet the spatial sparsity constraints, thereby enhancing the aggregation degree of the characteristic manifold of the lithium iron phosphate battery cycle life decoding feature vector by restoring the basic information in the full-precision information representation space, thereby improving the expression effect of the lithium iron phosphate battery cycle life decoding feature vector.
[0071] Next, it should be understood that the decoder plays the role of converting abstract features into specific prediction values in the neural network. It enables the model to predict specific cycle life values based on the input feature representation by learning the mapping relationship from feature space to output space. Therefore, in order to improve the prediction ability and accuracy of the model, the decoder is used to decode and regress the cycle life decoding feature vector of the lithium iron phosphate battery to obtain a decoded value for representing the cycle life of the lithium iron phosphate battery to be detected. Among them, the decoded value provides a prediction result for the cycle life of the lithium iron phosphate battery to be detected, which can be used as a basis for further decision-making and action. Through the analysis and application of the decoded value, the model can be continuously optimized and the prediction results can be improved, so as to be better applied to actual battery management and maintenance.
[0072] In a specific embodiment of the present application, the lithium iron phosphate battery cycle life decoding unit 143 is used to: use the decoder to decode and regress the optimized lithium iron phosphate battery cycle life decoding feature vector using the following formula to obtain the decoded value used to represent the cycle life of the optimized lithium iron phosphate battery to be detected, wherein the formula is Wherein, X is the optimized cycle life decoding feature vector of the lithium iron phosphate battery, Y is the decoding value used to represent the cycle life of the lithium iron phosphate battery to be detected, W is the weight matrix, Represents matrix multiplication.
[0073] It is worth mentioning that in addition to using the decoder to decode and regress the decoded feature vector of the optimized lithium iron phosphate battery cycle life, the support vector machine (SVM) can also be used for regression prediction. SVM is a supervised learning algorithm that can be used for both classification and regression problems. In this case, we can use the decoded feature vector of the optimized lithium iron phosphate battery cycle life as the input feature, the actual cycle life value as the target label, and perform regression prediction through the SVM algorithm.
[0074] The specific implementation process is as follows: 1. Data preparation: prepare training data sets and test data sets with optimized lithium iron phosphate battery cycle life decoding feature vectors and corresponding actual cycle life values. 2. Feature standardization: standardize the feature vectors to ensure that the numerical ranges between different features are consistent and avoid problems caused by different feature scales. 3. Select SVM regression model: In the regression problem, select an appropriate SVM regression model. You can choose linear kernel function, polynomial kernel function or Gaussian kernel function according to the specific situation. 4. Model training: Use the training data set to train the selected SVM regression model. The model will learn how to map the input features to the target cycle life value. 5. Model prediction: Use the trained SVM regression model to predict the feature vectors in the test data set to obtain the cycle life prediction value of the lithium iron phosphate battery to be tested. 6. Evaluate model performance: By comparing the difference between the model prediction value and the actual cycle life value, the performance of the SVM regression model can be evaluated and tuned and improved as needed.
[0075] In summary, the new energy vehicle battery performance detection system according to the embodiment of the present application is explained, which first obtains the tomographic images of multiple predetermined time points of the lithium iron phosphate battery to be detected under different simulated working conditions and the electrochemical characteristic information of multiple predetermined time points of the lithium iron phosphate battery to be detected under different simulated working conditions as input data, and then uses deep learning technology to extract and analyze these input data to obtain the cycle life feature vector of the lithium iron phosphate battery, and then uses a decoder to decode and regress the lithium iron phosphate battery cycle life feature vector to obtain a decoded value for representing the cycle life of the lithium iron phosphate battery to be detected. In this way, according to the decoded value, the cycle life of the lithium iron phosphate battery can be accurately predicted. In this way, it can provide an important reference for the battery management system and help optimize the use and performance of the battery.
[0076] As described above, the new energy vehicle battery performance detection system 100 according to the embodiment of the present application can be implemented in various wireless terminals, such as a server for new energy vehicle battery performance detection. In one example, the new energy vehicle battery performance detection system 100 according to the embodiment of the present application can be integrated into the wireless terminal as a software module and / or a hardware module. For example, the new energy vehicle battery performance detection system 100 can be a software module in the operating system of the wireless terminal, or it can be an application developed for the wireless terminal; of course, the new energy vehicle battery performance detection system 100 can also be one of the many hardware modules of the wireless terminal.
[0077] Alternatively, in another example, the new energy vehicle battery performance detection system 100 and the wireless terminal may also be separate devices, and the new energy vehicle battery performance detection system 100 may be connected to the wireless terminal via a wired and / or wireless network and transmit interactive information in accordance with an agreed data format.
[0078] Exemplary Methods
[0079] Figure 6 The figure shows a flow chart of a method for detecting the performance of a new energy vehicle battery according to an embodiment of the present application. Figure 6 As shown, according to the new energy vehicle battery performance detection method of the embodiment of the present application, the method includes: S1, obtaining tomographic images of a lithium iron phosphate battery to be detected at multiple predetermined time points under different simulated working conditions and electrochemical characteristic information of the lithium iron phosphate battery to be detected at multiple predetermined time points under different simulated working conditions; S2, respectively extracting and analyzing the tomographic images of the lithium iron phosphate battery to be detected at multiple predetermined time points under the different simulated working conditions and the electrochemical characteristic information of the lithium iron phosphate battery to be detected under the different simulated working conditions to obtain a tomographic global feature vector and an electrochemical information global feature matrix; S3, multiplying the tomographic global feature vector and the electrochemical information global feature matrix to obtain a lithium iron phosphate battery cycle life feature vector; S4, obtaining a decoding value based on the lithium iron phosphate battery cycle life feature vector, and the decoding value is used to represent the cycle life of the lithium iron phosphate battery to be detected.
[0080] Here, those skilled in the art can understand that the specific functions and operations of each step in the above-mentioned new energy vehicle battery performance detection method have been referenced above. Figure 1 The new energy vehicle battery performance detection system has been introduced in detail, and therefore, its repeated description will be omitted.
[0081] In the several embodiments provided by the present invention, it should be understood that the disclosed systems, devices, apparatuses and methods can be implemented in other ways. For example, the system embodiments described above are only illustrative, for example, the division of the modules is only a logical function division, and there may be other division methods in actual implementation.
[0082] In addition, each functional module in each embodiment of the present invention may be integrated into one processing unit, or each unit may exist physically separately, or two or more units may be integrated into one unit.
[0083] It is obvious to those skilled in the art that the present invention is not limited to the details of the above exemplary embodiments, and that the present invention can be implemented in other specific forms without departing from the spirit or essential characteristics of the present invention.
[0084] Therefore, no matter from which point of view, the embodiments should be regarded as illustrative and non-restrictive, and the scope of the present invention is limited by the appended claims rather than the above description, so it is intended that all changes falling within the meaning and scope of the equivalent elements of the claims are included in the present invention. Any attached figure mark in the claims should not be regarded as limiting the claims involved.
[0085] In addition, it is clear that the word "comprising" does not exclude other units or steps, and the singular does not exclude the plural. Multiple units or systems stated in a system claim can also be implemented by one unit or system through software or hardware. Secondary words are used to indicate names, but not to indicate any particular order.
[0086] Finally, it should be noted that the above embodiments are only used to illustrate the technical solution of the present invention rather than to limit it. Although the present invention has been described in detail with reference to the preferred embodiments, those skilled in the art should understand that the technical solution of the present invention can be modified or replaced by equivalents without departing from the spirit of the technical solution of the present invention.
Claims
1. A new energy vehicle battery performance detection system, characterized in that: include: A lithium iron phosphate battery data acquisition module, used to obtain tomographic images of a lithium iron phosphate battery to be tested at multiple predetermined time points under different simulated working conditions and electrochemical characteristic information of a lithium iron phosphate battery to be tested at multiple predetermined time points under different simulated working conditions; A lithium iron phosphate battery feature acquisition module, used to extract and analyze the tomographic images of the lithium iron phosphate battery to be detected at multiple predetermined time points under the different simulated working conditions and the electrochemical characteristic information of the lithium iron phosphate battery to be detected at multiple predetermined time points under the different simulated working conditions to obtain a tomographic global feature vector and an electrochemical information global feature matrix; A lithium iron phosphate battery feature fusion module, used for multiplying the tomography global feature vector and the electrochemical information global feature matrix to obtain a lithium iron phosphate battery cycle life feature vector; A lithium iron phosphate battery feature decoding module, used to obtain a decoding value based on the lithium iron phosphate battery cycle life feature vector, wherein the decoding value is used to represent the cycle life of the lithium iron phosphate battery to be detected; Wherein, the lithium iron phosphate battery feature acquisition module includes: A tomographic feature extraction unit, used for performing feature encoding on the tomographic images of the lithium iron phosphate battery to be tested at multiple predetermined time points under the different simulated working conditions to obtain the tomographic global feature vector; An electrochemical information feature extraction unit, used for feature encoding the electrochemical characteristic information of the lithium iron phosphate battery to be tested at multiple predetermined time points under the different simulated working conditions to obtain the electrochemical information global feature matrix; Wherein, the tomographic feature extraction unit comprises: A depth feature extraction subunit, used for respectively passing the tomographic images of the lithium iron phosphate battery to be tested at multiple predetermined time points under the different simulated working conditions through a tomographic feature extractor based on a depth feature fusion module to obtain multiple tomographic feature matrices under each simulated working condition; A tomographic time series feature extraction subunit, used for arranging the multiple tomographic feature matrices under the various simulation conditions according to the time dimension into tomographic time series input tensors under the various simulation conditions, and then passing them through a tomographic time series feature extraction module based on a three-dimensional convolutional neural network model to obtain a tomographic time series feature vector under the various simulation conditions; A tomographic global feature extraction subunit, used for arranging the tomographic time series feature vectors under the respective simulation conditions into a tomographic global input vector and then passing the tomographic global feature extraction module based on a feature encoder to obtain the tomographic global feature vector; Wherein, the electrochemical information feature extraction unit comprises: An electrochemical information arrangement subunit, used to arrange the electrochemical characteristic information of the lithium iron phosphate battery to be tested at multiple predetermined time points under the different simulated working conditions according to the time dimension into multiple electrochemical information input vectors under each simulated working condition; An electrochemical feature extraction subunit, used for respectively passing the multiple electrochemical information input vectors under the respective simulated working conditions through an electrochemical information feature extraction module based on a one-dimensional convolutional neural network model to obtain multiple electrochemical information feature vectors under the respective simulated working conditions; An electrochemical time series feature extraction subunit, used for obtaining the electrochemical information change feature vectors under each simulation condition by passing the multiple electrochemical information feature vectors under each simulation condition through an electrochemical information change feature extraction module based on a bidirectional long short-term memory neural network model; An electrochemical global feature extraction subunit, used for arranging the electrochemical information change feature vectors under the respective simulated working conditions into an electrochemical information global input matrix and then passing the matrix through an electrochemical information global feature extraction module including a two-dimensional convolution structure to obtain the electrochemical information global feature matrix; Wherein, the lithium iron phosphate battery feature decoding module includes: A lithium iron phosphate battery cycle life feature acquisition unit, used for passing the lithium iron phosphate battery cycle life feature vector through a lithium iron phosphate battery cycle life decoding feature extraction module based on a convolutional neural network model to obtain a lithium iron phosphate battery cycle life decoding feature vector; A lithium iron phosphate battery cycle life feature optimization unit, used for performing implicit group optimization with spatial sparsity restriction on the lithium iron phosphate battery cycle life decoding feature vector to obtain an optimized lithium iron phosphate battery cycle life decoding feature vector; The lithium iron phosphate battery cycle life decoding unit is used to pass the optimized lithium iron phosphate battery cycle life decoding feature vector through a decoder to obtain the decoding value.
2. The new energy vehicle battery performance detection system according to claim 1 is characterized in that: The depth feature extraction subunit is used to: Extracting a shallow feature matrix from the i-th layer of the tomographic feature extractor, wherein the i-th layer is the first to sixth layers of the tomographic feature extractor; Extracting a deep feature matrix from the jth layer of the tomographic feature extractor, wherein a ratio between the jth layer and the ith layer is greater than or equal to 5; as well as The shallow-deep feature fusion module is used to fuse the shallow feature matrix and the deep feature matrix to obtain each tomographic feature matrix in the plurality of tomographic feature matrices under each simulation condition.
3. The new energy vehicle battery performance detection system according to claim 2 is characterized in that: The electrochemical time series feature extraction subunit is used to: Arranging the multiple electrochemical information feature vectors under the respective simulation conditions in one dimension to obtain a sequence of multiple long-range correlation feature vectors; and The electrochemical information change feature extraction module based on the bidirectional long short-term memory neural network model is used to perform long-distance correlation feature encoding on the sequences of the multiple long-distance correlation feature vectors respectively to obtain the electrochemical information change feature vectors under the various simulation conditions.
4. The new energy vehicle battery performance detection system according to claim 3 is characterized in that: The lithium iron phosphate battery cycle life characteristic optimization unit comprises: The implicit group optimization factor of the spatial sparsity restriction of the cycle life decoding feature vector of the lithium iron phosphate battery is calculated by the following formula, wherein the formula is: Among them, v c represents the cycle life decoding feature vector of lithium iron phosphate battery, σ(v c ) represents the variance of the eigenvalue set of the cycle life decoding eigenvector of the lithium iron phosphate battery, v i is the characteristic value of each position in the cycle life decoding feature vector of the lithium iron phosphate battery, and L is the length of the cycle life decoding feature vector of the lithium iron phosphate battery, and ||v c || 1 represents a norm of the cycle life decoding feature vector of the lithium iron phosphate battery, w is an implicit group optimization factor constrained by the spatial sparsity of the cycle life decoding feature vector of the lithium iron phosphate battery; and The implicit group optimization factor of the spatial sparsity restriction is used as a weight to weight the cycle life decoding feature vector of the lithium iron phosphate battery to obtain the optimized cycle life decoding feature vector of the lithium iron phosphate battery.
5. A new energy vehicle battery performance detection method, characterized in that: include: Obtaining tomographic images of a lithium iron phosphate battery to be tested at multiple predetermined time points under different simulated working conditions and electrochemical characteristic information of a lithium iron phosphate battery to be tested at multiple predetermined time points under different simulated working conditions; Respectively extracting and analyzing the tomographic images at multiple predetermined time points of the lithium iron phosphate battery to be tested under the different simulated working conditions and the electrochemical characteristic information at multiple predetermined time points of the lithium iron phosphate battery to be tested under the different simulated working conditions to obtain a tomographic global feature vector and an electrochemical information global feature matrix; Multiplying the tomography global feature vector and the electrochemical information global feature matrix to obtain a cycle life feature vector of the lithium iron phosphate battery; Obtaining a decoding value based on the lithium iron phosphate battery cycle life feature vector, wherein the decoding value is used to represent the cycle life of the lithium iron phosphate battery to be detected; Among them, the tomographic images of the lithium iron phosphate battery to be tested at multiple predetermined time points under the different simulated working conditions and the electrochemical characteristic information of the lithium iron phosphate battery to be tested at multiple predetermined time points under the different simulated working conditions are respectively extracted and analyzed to obtain the tomographic global feature vector and the electrochemical information global feature matrix, including: A tomographic feature extraction unit, used for performing feature encoding on the tomographic images of the lithium iron phosphate battery to be tested at multiple predetermined time points under the different simulated working conditions to obtain the tomographic global feature vector; An electrochemical information feature extraction unit, used for feature encoding the electrochemical characteristic information of the lithium iron phosphate battery to be tested at multiple predetermined time points under the different simulated working conditions to obtain the electrochemical information global feature matrix; Among them, feature encoding is performed on the tomographic images of the lithium iron phosphate battery to be tested at multiple predetermined time points under the different simulated working conditions to obtain the tomographic global feature vector, including: The tomographic images of the lithium iron phosphate battery to be tested at multiple predetermined time points under the different simulated working conditions are respectively passed through a tomographic feature extractor based on a deep and shallow feature fusion module to obtain multiple tomographic feature matrices under each simulated working condition; Arranging the multiple tomographic feature matrices under the various simulation conditions according to the time dimension into tomographic timing input tensors under the various simulation conditions, and then passing them through a tomographic timing feature extraction module based on a three-dimensional convolutional neural network model to obtain a tomographic timing feature vector under the various simulation conditions; Arranging the tomographic time series feature vectors under the various simulation conditions as a tomographic global input vector and then passing it through a tomographic global feature extraction module based on a feature encoder to obtain the tomographic global feature vector; Among them, feature encoding is performed on the electrochemical characteristic information of the lithium iron phosphate battery to be tested at multiple predetermined time points under the different simulated working conditions to obtain the electrochemical information global feature matrix, including: Arranging the electrochemical characteristic information of the lithium iron phosphate battery to be tested at multiple predetermined time points under the different simulated working conditions according to the time dimension into multiple electrochemical information input vectors under each simulated working condition; The multiple electrochemical information input vectors under the respective simulated working conditions are respectively passed through an electrochemical information feature extraction module based on a one-dimensional convolutional neural network model to obtain multiple electrochemical information feature vectors under the respective simulated working conditions; The multiple electrochemical information feature vectors under the various simulated working conditions are passed through an electrochemical information change feature extraction module based on a bidirectional long short-term memory neural network model to obtain an electrochemical information change feature vector under the various simulated working conditions; Arranging the electrochemical information change feature vectors under the various simulation conditions into an electrochemical information global input matrix and then passing it through an electrochemical information global feature extraction module including a two-dimensional convolution structure to obtain the electrochemical information global feature matrix; Wherein, a decoding value is obtained based on the cycle life characteristic vector of the lithium iron phosphate battery, and the decoding value is used to represent the cycle life of the lithium iron phosphate battery to be detected, including: The lithium iron phosphate battery cycle life feature vector is passed through a lithium iron phosphate battery cycle life decoding feature extraction module based on a convolutional neural network model to obtain a lithium iron phosphate battery cycle life decoding feature vector; Performing spatial sparsity-restricted implicit group optimization on the cycle life decoding feature vector of the lithium iron phosphate battery to obtain an optimized cycle life decoding feature vector of the lithium iron phosphate battery; The optimized lithium iron phosphate battery cycle life decoding feature vector is passed through a decoder to obtain the decoded value.
