Performance evaluation method and system for lithium battery pole piece detection

Abstract

Disclosed is a performance evaluation method and system for lithium battery pole piece detection. First, the performance indicators of the lithium battery to be detected are arranged as a performance indicator input vector, then the performance indicator input vector is passed through a first one-dimensional convolutional neural network model to obtain a first-scale performance indicator associated feature vector, then the performance indicator input vector is passed through a second one-dimensional convolutional neural network model to obtain a second-scale performance indicator associated feature vector, then the first-scale performance indicator associated feature vector and the second-scale performance indicator associated feature vector are fused to obtain a classification feature vector, and finally, the classification feature vector is passed through a classifier to obtain a classification result for indicating whether the performance of the lithium battery to be detected meets a predetermined standard. In this way, the performance of the lithium battery pole piece can be comprehensively and accurately evaluated, thereby better ensuring the quality and stability of the lithium battery.

Description

Technical Field

[0001] The present application relates to the field of intelligent evaluation, and more specifically, to a performance evaluation method and system for lithium battery pole piece detection. Background Art

[0002] As the power source of new energy vehicles, power batteries are the most important system in the whole vehicle and the iconic component that distinguishes them from other traditional fuel vehicles. The heart of traditional fuel vehicles is the engine, while the heart of new energy vehicles is the power battery. Ternary polymer lithium battery, referred to as ternary lithium battery, refers to a lithium battery whose positive electrode material uses ternary positive electrode materials of lithium nickel cobalt manganese oxide (Li(NiCoMn)O2) or lithium nickel cobalt aluminum oxide. The ternary composite positive electrode material is made of nickel salt, cobalt salt and manganese salt. The ratio of nickel, cobalt and manganese can be adjusted according to actual needs. The battery with ternary material as the positive electrode is safer than that of lithium cobalt oxide battery. Ternary lithium battery is a kind of energy storage device that integrates high energy density and high voltage. It has been widely used in mobile and wireless electronic devices, power tools, hybrid power and electric vehicles.

[0003] As a battery itself, the appearance directly provided by a lithium battery is the battery pole piece, so the most commonly used appearance inspection for lithium batteries is pole piece inspection, including its quantity and position. However, the traditional method of detecting the number and position of lithium battery pole pieces has shortcomings. In many factories, only the leakage of batch batteries or the false detection rate are roughly adopted, resulting in incomplete detailed judgments, which cannot fully reflect the quality of the current batch of batteries and the effectiveness of the detection system.

[0004] Therefore, an optimized performance evaluation scheme for lithium battery pole piece detection is desired. Summary of the invention

[0005] In order to solve the above technical problems, the present application is proposed. The embodiment of the present application provides a performance evaluation method and system for lithium battery pole piece detection. It first arranges the performance indicators of the lithium battery to be detected as a performance indicator input vector, then passes the performance indicator input vector through a first one-dimensional convolutional neural network model to obtain a first-scale performance indicator associated feature vector, then passes the performance indicator input vector through a second one-dimensional convolutional neural network model to obtain a second-scale performance indicator associated feature vector, then, fuses the first-scale performance indicator associated feature vector and the second-scale performance indicator associated feature vector to obtain a classification feature vector, and finally, passes the classification feature vector through a classifier to obtain a classification result for indicating whether the performance of the lithium battery to be detected meets a predetermined standard. In this way, the performance of the lithium battery pole piece can be comprehensively and accurately evaluated, thereby better ensuring the quality and stability of the lithium battery.

[0006] According to one aspect of the present application, a performance evaluation method for lithium battery pole piece detection is provided, which includes:

[0007] Obtaining performance indicators of the lithium battery to be tested, the performance indicators including negative electrode sheet-quantity error, positive electrode sheet-quantity error, negative electrode sheet-quantity accuracy, positive electrode sheet-quantity accuracy, positive and negative electrode sheet pair-overall accuracy, negative electrode sheet position error, positive electrode sheet position error and positive and negative electrode sheet pair-overall position error;

[0008] Arranging the performance indicators of the lithium battery to be tested into a performance indicator input vector according to sample dimensions;

[0009] Passing the performance indicator input vector through a first one-dimensional convolutional neural network model to obtain a first scale performance indicator associated feature vector;

[0010] Passing the performance indicator input vector through a second one-dimensional convolutional neural network model to obtain a second-scale performance indicator associated feature vector, wherein the second one-dimensional convolutional neural network model and the first one-dimensional convolutional neural network model respectively use one-dimensional convolution kernels with different scales;

[0011] fusing the first scale performance indicator associated feature vector and the second scale performance indicator associated feature vector to obtain a classification feature vector; and

[0012] The classification feature vector is passed through a classifier to obtain a classification result, and the classification result is used to indicate whether the performance of the lithium battery to be tested meets a predetermined standard.

[0013] In the above-mentioned performance evaluation method for lithium battery electrode detection, the calculation formula of the negative electrode quantity error is expressed as:

[0014]

[0015] The calculation formula of the positive electrode sheet-quantity error is expressed as:

[0016]

[0017] The calculation formula of the negative electrode sheet-quantity accuracy is expressed as:

[0018]

[0019] The calculation formula of the positive electrode sheet-quantity accuracy is expressed as:

[0020]

[0021] The calculation formula of the positive and negative electrode pair-overall accuracy is expressed as:

[0022]

[0023] The calculation formula of the negative electrode position error is expressed as:

[0024]

[0025] The calculation formula of the positive electrode position error is expressed as:

[0026]

[0027] The calculation formula of the positive and negative electrode pair-overall position error is expressed as:

[0028]

[0029] in, Indicates the negative electrode sheet-quantity error, Indicates the positive electrode sheet-quantity error, Indicates the accuracy of the number of negative electrode sheets, Indicates the positive electrode sheet-quantity accuracy, Indicates the overall accuracy of the positive and negative electrode pairs, represents the position error of the negative electrode sheet, represents the positive electrode position error, Indicates the overall position error of the positive and negative electrode pairs, Indicates the number of negative plates to be tested. Indicates the number of negative plates corresponding to the true value, Indicates the number of positive plates to be tested. Indicates the number of positive plates corresponding to the true value, Indicates the total number of positive and negative electrodes to be tested. Indicates the total number of positive and negative electrodes of the corresponding true value, represents the spatial coordinates of the jth negative electrode sheet of the i-th sample to be tested, represents the spatial coordinates of the jth negative electrode of the ith sample of the corresponding true value, represents the spatial coordinates of the jth positive electrode of the i-th sample to be tested, represents the spatial coordinates of the jth negative electrode sheet of the i-th sample of the corresponding true value, and N represents the total number of samples in the test set.

[0030] In the above-mentioned performance evaluation method for lithium battery pole piece detection, the performance index input vector is passed through a first one-dimensional convolutional neural network model to obtain a first scale performance index associated feature vector, including:

[0031] Use each layer of the first one-dimensional convolutional neural network model to perform one-dimensional convolution processing, mean pooling processing and nonlinear activation processing on the input data in the forward pass of the layer so that the last layer of the first one-dimensional convolutional neural network model outputs the first-scale performance indicator associated feature vector, wherein the input of the first layer of the first one-dimensional convolutional neural network model is the performance indicator input vector.

[0032] In the above-mentioned performance evaluation method for lithium battery pole piece detection, the performance index input vector is passed through a second one-dimensional convolutional neural network model to obtain a second-scale performance index associated feature vector, including:

[0033] Use each layer of the second one-dimensional convolutional neural network model to perform one-dimensional convolution processing, mean pooling processing and nonlinear activation processing on the input data in the forward pass of the layer so that the last layer of the second one-dimensional convolutional neural network model outputs the second-scale performance indicator associated feature vector, wherein the input of the first layer of the second one-dimensional convolutional neural network model is the performance indicator input vector.

[0034] In the above-mentioned performance evaluation method for lithium battery pole piece detection, the first scale performance index associated feature vector and the second scale performance index associated feature vector are fused to obtain a classification feature vector, including:

[0035] The first scale performance indicator associated feature vector and the second scale performance indicator associated feature vector are fused using the following cascade formula to obtain the classification feature vector;

[0036] Wherein, the cascade formula is:

[0037] V c =Concat[V a , V b ]

[0038] Among them, V a represents the associated feature vector of the first scale performance index, V b represents the associated feature vector of the second scale performance index, Concat[·] represents the cascade function, V c represents the classification feature vector.

[0039] The above-mentioned performance evaluation method for lithium battery pole piece detection also includes a training step: for training the first one-dimensional convolutional neural network model, the second one-dimensional convolutional neural network model and the classifier.

[0040] In the above-mentioned performance evaluation method for lithium battery pole piece detection, the training step includes:

[0041] Acquire training data, wherein the training data includes a training performance index of the lithium battery to be tested and a true value of whether the performance of the lithium battery to be tested meets a predetermined standard;

[0042] Arranging the training performance indicators of the lithium battery to be tested into a training performance indicator input vector according to the sample dimension;

[0043] Passing the training performance indicator input vector through the first one-dimensional convolutional neural network model to obtain a training first-scale performance indicator associated feature vector;

[0044] Passing the training performance indicator input vector through the second one-dimensional convolutional neural network model to obtain a training second-scale performance indicator associated feature vector, wherein the second one-dimensional convolutional neural network model and the first one-dimensional convolutional neural network model respectively use one-dimensional convolution kernels with different scales;

[0045] fusing the training first scale performance indicator associated feature vector and the training second scale performance indicator associated feature vector to obtain a training classification feature vector;

[0046] Passing the training classification feature vector through the classifier to obtain a classification loss function value;

[0047] Calculating the streaming refinement loss function value of the training first-scale performance indicator associated feature vector and the training second-scale performance indicator associated feature vector; and

[0048] The weighted sum of the classification loss function value and the streaming refinement loss function value is used as the loss function value, and the first one-dimensional convolutional neural network model, the second one-dimensional convolutional neural network model and the classifier are trained through back propagation of gradient descent.

[0049] In the above-mentioned performance evaluation method for lithium battery pole piece detection, the training classification feature vector is passed through the classifier to obtain the classification loss function value, including:

[0050] The classifier processes the training classification feature vector using the following classification loss formula to obtain a training classification result. The classification loss formula is: softmax{(W n ,B n ):…:(W1,B1)|X}, where W1 to W n is the weight matrix, B1 to B n is a bias vector, X is the training classification feature vector; and

[0051] A cross entropy value between the training classification result and the true value is calculated as the classification loss function value.

[0052] In the above-mentioned performance evaluation method for lithium battery pole piece detection, the streaming refinement loss function value of the training first scale performance index associated feature vector and the training second scale performance index associated feature vector is calculated, including:

[0053] The streaming refinement loss function value of the training first scale performance indicator associated feature vector and the training second scale performance indicator associated feature vector is calculated using the following optimization formula;

[0054] Wherein, the optimization formula is:

[0055]

[0056] Wherein, V1 represents the associated characteristic vector of the training first scale performance indicator, V2 represents the associated characteristic vector of the training second scale performance indicator, represents the square of the vector's two-norm, and and ⊙ represent position-by-position subtraction and multiplication of a vector, respectively. exp(·) represents the exponential operation of a vector. The exponential operation of a vector represents the calculation of the value of a natural exponential function raised to the power of the eigenvalues ​​at each position in the vector. Represents the streaming refinement loss function value.

[0057] According to another aspect of the present application, a performance evaluation method for lithium battery pole piece detection is provided, which includes:

[0058] A performance index acquisition module is used to acquire performance indicators of the lithium battery to be tested, wherein the performance indicators include negative electrode sheet-quantity error, positive electrode sheet-quantity error, negative electrode sheet-quantity accuracy, positive electrode sheet-quantity accuracy, positive and negative electrode sheet pair-overall accuracy, negative electrode sheet position error, positive electrode sheet position error and positive and negative electrode sheet pair-overall position error;

[0059] An input vector arrangement module, used to arrange the performance indicators of the lithium battery to be tested into a performance indicator input vector according to the sample dimension;

[0060] A first convolutional encoding module, used for passing the performance indicator input vector through a first one-dimensional convolutional neural network model to obtain a first scale performance indicator associated feature vector;

[0061] A second convolutional encoding module, used for passing the performance indicator input vector through a second one-dimensional convolutional neural network model to obtain a second-scale performance indicator associated feature vector, wherein the second one-dimensional convolutional neural network model and the first one-dimensional convolutional neural network model respectively use one-dimensional convolution kernels with different scales;

[0062] a vector fusion module, configured to fuse the first scale performance indicator associated feature vector and the second scale performance indicator associated feature vector to obtain a classification feature vector; and

[0063] The classification module is used to pass the classification feature vector through a classifier to obtain a classification result, and the classification result is used to indicate whether the performance of the lithium battery to be tested meets a predetermined standard.

[0064] Compared with the prior art, the performance evaluation method and system for lithium battery pole piece detection provided by the present application first arrange the performance indicators of the lithium battery to be detected into a performance indicator input vector, then pass the performance indicator input vector through a first one-dimensional convolutional neural network model to obtain a first-scale performance indicator associated feature vector, then pass the performance indicator input vector through a second one-dimensional convolutional neural network model to obtain a second-scale performance indicator associated feature vector, then fuse the first-scale performance indicator associated feature vector and the second-scale performance indicator associated feature vector to obtain a classification feature vector, and finally, pass the classification feature vector through a classifier to obtain a classification result for indicating whether the performance of the lithium battery to be detected meets a predetermined standard. In this way, the performance of the lithium battery pole piece can be comprehensively and accurately evaluated, thereby better ensuring the quality and stability of the lithium battery. BRIEF DESCRIPTION OF THE DRAWINGS

[0065] In order to more clearly illustrate the technical solutions of the embodiments of the present application, the following is a brief introduction to the drawings required for the description of the embodiments. Obviously, the drawings described below are only some embodiments of the present application. For ordinary technicians in this field, other drawings can be obtained based on these drawings without creative work. The following drawings are not deliberately scaled according to the actual size, and the focus is on illustrating the main purpose of the present application.

[0066] Figure 1 This is a diagram of an application scenario of a performance evaluation method for lithium battery pole piece detection according to an embodiment of the present application.

[0067] Figure 2 Flow chart of a performance evaluation method for lithium battery pole piece detection according to an embodiment of the present application.

[0068] Figure 3 Schematic diagram of the architecture of a performance evaluation method for lithium battery pole piece detection according to an embodiment of the present application.

[0069] Figure 4 It is a flowchart of the training steps further included in the performance evaluation method of lithium battery pole piece detection according to the embodiment of the present application.

[0070] Figure 5 4 is a block diagram of a performance evaluation system for lithium battery pole piece detection according to an embodiment of the present application. DETAILED DESCRIPTION

[0071] The technical solutions in the embodiments of the present application will be described clearly and completely below in conjunction with the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present application, not all embodiments. Based on the embodiments of the present application, all other embodiments obtained by ordinary technicians in this field without creative work also fall within the scope of protection of the present application.

[0072] As shown in this application and claims, unless the context clearly indicates an exception, the words "a", "an", "an" and / or "the" do not refer to the singular and may also include the plural. Generally speaking, the terms "include" and "comprise" only indicate the inclusion of the steps and elements that have been clearly identified, and these steps and elements do not constitute an exclusive list. The method or device may also include other steps or elements.

[0073] Although the present application makes various references to certain modules in the system according to the embodiments of the present application, any number of different modules can be used and run on the user terminal and / or server. The modules are only illustrative, and different aspects of the system and method can use different modules.

[0074] Flowcharts are used in the present application to illustrate the operations performed by the system according to the embodiments of the present application. It should be understood that the preceding or following operations are not necessarily performed accurately in order. On the contrary, various steps may be processed in reverse order or simultaneously as required. Meanwhile, other operations may also be added to these processes, or a certain step or several steps of operations may be removed from these processes.

[0075] 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.

[0076] As mentioned above, in the production process of lithium batteries, the appearance directly provided by lithium batteries is the battery pole piece. Therefore, the most commonly used appearance inspection for lithium batteries is pole piece inspection, and parameters such as the number and position of pole pieces have a great influence on the performance of the battery, so it is very necessary to detect these parameters. However, the traditional method of detecting the number and position of lithium battery pole pieces has shortcomings. In many factories, only the rough use of batch battery leakage or false detection rate, etc., leads to incomplete detailed judgments, which cannot fully reflect the quality of the current batch of batteries and the effectiveness of the detection system. Therefore, an optimized performance evaluation scheme for lithium battery pole piece detection is desired.

[0077] Specifically, in the technical solution of the present application, in terms of lithium battery detection indicators, a comprehensive analysis is conducted on the quantity, position, single pole, group level, accuracy, and error rate, so as to utilize 8 comprehensive complementary indicators of lithium battery performance to evaluate the performance of lithium batteries, wherein the performance indicators of lithium batteries include the number of errors of negative electrode sheets, the number of errors of positive electrode sheets, the number of accuracy of negative electrode sheets, the number of accuracy of positive electrode sheets, the number of accuracy of positive electrode sheets, the overall accuracy of positive and negative electrode sheet pairs, the position error of negative electrode sheets, the position error of positive electrode sheets, and the overall position error of positive and negative electrode sheet pairs. Accordingly, considering that these 8 indicators have different degrees of correlation under different indicator category spans, how to explore the correlation relationship between these indicators, so as to comprehensively evaluate the number errors and position errors of positive and negative electrodes, as well as the overall accuracy and inter-electrode spacing of electrode sheets through the mutual constraints between the 8 indicators, is a difficult point.

[0078] In recent years, deep learning and neural networks have been widely used in computer vision, natural language processing, text signal processing and other fields. The development of deep learning and neural networks provides new solutions and solutions for mining the implicit feature correlation information between the various indicators.

[0079] More specifically, in the technical solution of the present application, first, the performance indicators of the lithium battery to be tested are obtained, and the performance indicators include the number of negative electrode sheets-errors, the number of positive electrode sheets-errors, the number of negative electrode sheets-accuracy, the number of positive electrode sheets-accuracy, the overall accuracy of the positive and negative electrode sheets, the position error of the negative electrode sheets, the position error of the positive electrode sheets, and the overall position error of the positive and negative electrode sheets. It should be understood that in the production process of lithium batteries, parameters such as the number and position of the electrodes have a great influence on the performance of the battery. Therefore, when testing lithium battery electrodes, it is necessary to comprehensively and accurately evaluate these parameters. In particular, these performance indicators include the number errors and accuracy of negative and positive electrode sheets, as well as the position errors, as well as the overall accuracy and position errors, so that through the analysis of the comprehensive correlation between these indicators, the performance of the lithium battery electrodes can be comprehensively and accurately evaluated, thereby better ensuring the quality and stability of the lithium battery.

[0080] Next, considering that the performance indicators of the various lithium batteries are mutually correlated, in order to fully express the correlation characteristics between the performance indicators of the various lithium batteries, in the technical solution of the present application, the performance indicators of the lithium batteries to be tested are further arranged as performance indicator input vectors according to the sample dimensions, so as to facilitate the subsequent capture and extraction of indicator correlation characteristics.

[0081] Then, considering that the performance indicators of each lithium battery present different correlation characteristics under different indicator category spans, in order to comprehensively and accurately evaluate the performance of lithium battery pole pieces to ensure the quality and stability of lithium batteries, in the technical solution of the present application, a convolutional neural network model with one-dimensional convolution kernels of different scales is further used to perform feature mining of the performance indicator input vector to extract multi-scale correlation feature information between the performance indicators of each lithium battery under different indicator category spans. Specifically, the performance indicator input vector is processed through a first one-dimensional convolutional neural network model, so as to extract the first scale correlation features of the performance indicators of each lithium battery by using a one-dimensional convolution kernel with a first scale, thereby obtaining a first scale performance indicator correlation feature vector; and the performance indicator input vector is processed through a second one-dimensional convolutional neural network model, so as to extract the second scale correlation features of the performance indicators of each lithium battery by using a one-dimensional convolution kernel with a second scale, thereby obtaining a second scale performance indicator correlation feature vector. In particular, here, the second one-dimensional convolutional neural network model and the first one-dimensional convolutional neural network model respectively use one-dimensional convolution kernels with different scales to extract multi-scale correlation feature information under different receptive fields between the various lithium battery performance indicators.

[0082] Furthermore, the first-scale performance indicator associated feature vector and the second-scale performance indicator associated feature vector are fused to fuse multi-scale lithium battery performance indicator associated feature information under different performance indicator category spans, and this is used as a classification feature vector for classification processing through a classifier to obtain a classification result indicating whether the performance of the lithium battery to be tested meets the predetermined standard.

[0083] That is, in the technical solution of the present application, the label of the classifier includes the performance of the lithium battery to be detected meets the predetermined standard (first label), and the performance of the lithium battery to be detected does not meet the predetermined standard (second label), wherein the classifier determines which classification label the classification feature vector belongs to through a soft maximum function. It is worth noting that the first label p1 and the second label p2 here do not contain artificially set concepts. In fact, during the training process, the computer model does not have the concept of "whether the performance of the lithium battery to be detected meets the predetermined standard". It only has two classification labels and the probability of the output feature under these two classification labels, that is, the sum of p1 and p2 is one. Therefore, the classification result of whether the performance of the lithium battery to be detected meets the predetermined standard is actually converted from the classification label to a binary classification class probability distribution that conforms to the laws of nature. In essence, the physical meaning of the natural probability distribution of the label is used, rather than the linguistic text meaning of "whether the performance of the lithium battery to be detected meets the predetermined standard". It should be understood that in the technical solution of the present application, the classification label of the classifier is a detection and evaluation label for whether the performance of the lithium battery to be tested meets the predetermined standards. Therefore, after obtaining the classification result, the quantity error, position error of the positive and negative poles, and the overall accuracy of the pole pair and the pole spacing can be comprehensively evaluated based on the classification result, thereby improving the accuracy of lithium battery quality detection.

[0084] In particular, in the technical solution of the present application, here, when the first-scale performance indicator associated feature vector and the second-scale performance indicator associated feature vector are fused to obtain the classification feature vector, what is essentially done is to map the parametric associated feature representations of the performance indicators at different scales expressed by the first-scale performance indicator associated feature vector and the second-scale performance indicator associated feature vector to a high-dimensional fusion feature space. Therefore, if the correlation between the first-scale performance indicator associated feature vector and the second-scale performance indicator associated feature vector obtained by the first one-dimensional convolutional neural network model and the second one-dimensional convolutional neural network model in the serialized parametric associated feature representation and the high-dimensional feature fusion space dimension can be improved, the expression effect of the classification feature vector can be improved.

[0085] Based on this, the applicant of the present application further introduces a streaming refinement loss function for the first scale performance index associated feature vector and the second scale performance index associated feature vector in addition to the classification loss function based on the classification feature vector, which is expressed as:

[0086]

[0087] Where V1 and V2 are the first scale performance indicator associated feature vector and the second scale performance indicator associated feature vector, respectively, and Represents the square of the two-norm of a vector.

[0088] Here, the streaming refinement loss function is based on the transformation of the first scale performance index associated feature vector V1 and the second scale performance index associated feature vector V2 in the parameter serialization associated streaming distribution to the spatial distribution in the high-dimensional fusion feature space, and the super-resolution improvement of the spatial distribution in the high-dimensional fusion feature space is achieved by synchronously performing the interpolation under the sequence distribution of the vector, so as to provide a more refined alignment for the distribution difference in the high-dimensional fusion feature space through the mutual probability relationship under the balanced sequence, so as to jointly present the cross-dimensional (inter-dimensional) context association on the parameter-associated feature dimension of the performance index at different scales and the fusion feature space dimension, so as to improve the expression effect of the classification feature vector, and then improve the accuracy of the classification result obtained by the classification feature vector through the classifier. In this way, the number error, position error, overall accuracy and inter-pole spacing of the positive and negative poles can be comprehensively evaluated, thereby improving the accuracy of lithium battery quality detection.

[0089] Figure 1 FIG. 1 is an application scenario diagram of the performance evaluation method for lithium battery pole piece detection according to an embodiment of the present application. Figure 1 As shown, in this application scenario, first, the performance index of the lithium battery to be tested is obtained (for example, Figure 1 D), the performance indicators include negative electrode sheet-quantity error, positive electrode sheet-quantity error, negative electrode sheet-quantity accuracy, positive electrode sheet-quantity accuracy, positive and negative electrode sheet pair-overall accuracy, negative electrode sheet position error, positive electrode sheet position error and positive and negative electrode sheet pair-overall position error, and then the performance indicators of the lithium battery to be tested are input into a server deployed with a performance evaluation algorithm for lithium battery electrode detection (for example, Figure 1 S) as shown in the figure, wherein the server is capable of using the performance evaluation algorithm of the lithium battery pole piece detection to process the performance indicators of the lithium battery to be detected to obtain a classification result indicating whether the performance of the lithium battery to be detected meets the predetermined standards.

[0090] After introducing the basic principles of the present application, various non-limiting embodiments of the present application will be described in detail with reference to the accompanying drawings.

[0091] Figure 2 Flow chart of the performance evaluation method of lithium battery pole piece detection according to the embodiment of the present application. Figure 2As shown, according to the performance evaluation method of lithium battery pole piece detection in the embodiment of the present application, the steps include: S110, obtaining performance indicators of the lithium battery to be detected, wherein the performance indicators include negative pole piece-quantity error, positive pole piece-quantity error, negative pole piece-quantity accuracy, positive pole piece-quantity accuracy, positive and negative pole piece pair-overall accuracy, negative pole piece position error, positive pole piece position error and positive and negative pole piece pair-overall position error; S120, arranging the performance indicators of the lithium battery to be detected into a performance indicator input vector according to the sample dimension; S130, passing the performance indicator input vector through a first one-dimensional convolutional neural network model to obtain a first scale performance indicator. S140, passing the performance indicator input vector through a second one-dimensional convolutional neural network model to obtain a second-scale performance indicator associated feature vector, wherein the second one-dimensional convolutional neural network model and the first one-dimensional convolutional neural network model respectively use one-dimensional convolution kernels with different scales; S150, fusing the first-scale performance indicator associated feature vector and the second-scale performance indicator associated feature vector to obtain a classification feature vector; and, S160, passing the classification feature vector through a classifier to obtain a classification result, wherein the classification result is used to indicate whether the performance of the lithium battery to be tested meets a predetermined standard.

[0092] Figure 3 Schematic diagram of the performance evaluation method for lithium battery pole piece detection according to an embodiment of the present application. Figure 3 As shown, in the network architecture, first, the performance indicators of the lithium battery to be detected are obtained, and the performance indicators include negative electrode sheet-quantity error, positive electrode sheet-quantity error, negative electrode sheet-quantity accuracy, positive electrode sheet-quantity accuracy, positive and negative electrode sheet pair-overall accuracy, negative electrode sheet position error, positive electrode sheet position error and positive and negative electrode sheet pair-overall position error; then, the performance indicators of the lithium battery to be detected are arranged as performance indicator input vectors according to sample dimensions; then, the performance indicator input vector is passed through a first one-dimensional convolutional neural network model to obtain a first-scale performance indicator associated feature vector; then, the performance indicator input vector is passed through a second one-dimensional convolutional neural network model to obtain a second-scale performance indicator associated feature vector, wherein the second one-dimensional convolutional neural network model and the first one-dimensional convolutional neural network model respectively use one-dimensional convolution kernels with different scales; then, the first-scale performance indicator associated feature vector and the second-scale performance indicator associated feature vector are fused to obtain a classification feature vector; finally, the classification feature vector is passed through a classifier to obtain a classification result, and the classification result is used to indicate whether the performance of the lithium battery to be detected meets the predetermined standard.

[0093] More specifically, in step S110, the performance indicators of the lithium battery to be tested are obtained, and the performance indicators include the negative electrode sheet-quantity error, the positive electrode sheet-quantity error, the negative electrode sheet-quantity accuracy, the positive electrode sheet-quantity accuracy, the positive and negative electrode sheet pair-overall accuracy, the negative electrode sheet position error, the positive electrode sheet position error, and the positive and negative electrode sheet pair-overall position error. It should be understood that in the production process of lithium batteries, parameters such as the number and position of the electrode sheets have a great influence on the performance of the battery. Therefore, when testing the lithium battery electrode sheets, it is necessary to comprehensively and accurately evaluate these parameters. In particular, these performance indicators include the number and accuracy of the negative electrode sheets, the positive electrode sheets, and the position error, as well as the overall accuracy and position error, so that through the analysis of the comprehensive correlation between these indicators, the performance of the lithium battery electrode sheets can be comprehensively and accurately evaluated, thereby better ensuring the quality and stability of the lithium battery.

[0094] Accordingly, in a specific example, the calculation formula of the negative electrode sheet-number error (Anode-Num-MAE) is expressed as:

[0095]

[0096] The calculation formula of the cathode plate-number error (Cachode-Num-MAE) is expressed as:

[0097]

[0098] The calculation formula of the negative electrode sheet-number accuracy (Anode-Num-Acc) is expressed as:

[0099]

[0100] The calculation formula of the cathode plate-number accuracy (Cachode-Num-Acc) is expressed as:

[0101]

[0102] The calculation formula of the positive and negative electrode pair-overall accuracy (Pair-Num-Acc) is expressed as:

[0103]

[0104] The calculation formula of the negative electrode position error (Anode-Location-MAE) is expressed as:

[0105]

[0106] The calculation formula of the cathode plate position error (Cachode-Location-MAE) is expressed as:

[0107]

[0108] The calculation formula of the positive and negative electrode pair-overall position error is expressed as:

[0109]

[0110] in, Indicates the negative electrode sheet-quantity error, Indicates the positive electrode sheet-quantity error, Indicates the accuracy of the number of negative electrode sheets, Indicates the positive electrode sheet-quantity accuracy, Indicates the overall accuracy of the positive and negative electrode pairs, represents the position error of the negative electrode sheet, represents the positive electrode position error, Indicates the overall position error of the positive and negative electrode pairs, Indicates the number of negative plates to be tested. Indicates the number of negative plates corresponding to the true value, Indicates the number of positive plates to be tested. Indicates the number of positive plates corresponding to the true value, Indicates the total number of positive and negative electrodes to be tested. Indicates the total number of positive and negative electrodes of the corresponding true value, represents the spatial coordinates of the jth negative electrode sheet of the i-th sample to be tested, represents the spatial coordinates of the jth negative electrode of the ith sample of the corresponding true value, represents the spatial coordinates of the jth positive electrode of the i-th sample to be tested, The spatial coordinates of the jth negative electrode sheet of the i-th sample of the corresponding true value, N represents the total number of samples in the test set. The unit of the negative electrode sheet position error, the positive electrode sheet position error and the positive-negative electrode sheet pair-overall position error is pixel.

[0111] Through the mutual constraints among the above 8 indicators, the quantity error, position error of the positive and negative poles, the overall accuracy of the pole pair, and the pole spacing can be comprehensively evaluated.

[0112] More specifically, in step S120, the performance indicators of the lithium battery to be detected are arranged as a performance indicator input vector according to the sample dimension. Since the performance indicators of the lithium batteries are mutually correlated, in order to fully express the correlation characteristics between the performance indicators of the lithium batteries, in the technical solution of the present application, the performance indicators of the lithium battery to be detected are further arranged as a performance indicator input vector according to the sample dimension, so as to facilitate the subsequent capture and extraction of indicator correlation characteristics.

[0113] Taking into account that the performance indicators of the various lithium batteries present different correlation characteristics under different indicator category spans, in order to be able to comprehensively and accurately evaluate the performance of lithium battery pole pieces to ensure the quality and stability of lithium batteries, in the technical solution of the present application, a convolutional neural network model with one-dimensional convolution kernels of different scales is further used to perform feature mining of the performance indicator input vector to extract multi-scale correlation feature information between the various lithium battery performance indicators under different indicator category spans.

[0114] More specifically, in step S130, the performance indicator input vector is passed through a first one-dimensional convolutional neural network model to obtain a first-scale performance indicator associated feature vector, thereby extracting the first-scale associated features of each lithium battery performance indicator by using a one-dimensional convolution kernel with a first scale.

[0115] It should be understood that Convolutional Neural Network (CNN) is an artificial neural network that is widely used in image recognition and other fields. Convolutional neural network may include input layer, hidden layer and output layer, wherein the hidden layer may include convolution layer, pooling layer, activation layer and fully connected layer, etc. The previous layer performs corresponding operations according to the input data and outputs the operation results to the next layer. The input initial data is processed through multiple layers of operations to obtain a final result.

[0116] Accordingly, in a specific example, the performance indicator input vector is passed through a first one-dimensional convolutional neural network model to obtain a first-scale performance indicator associated feature vector, including: using each layer of the first one-dimensional convolutional neural network model to perform one-dimensional convolution processing, mean pooling processing, and nonlinear activation processing on the input data in the forward pass of the layer so that the last layer of the first one-dimensional convolutional neural network model outputs the first-scale performance indicator associated feature vector, wherein the input of the first layer of the first one-dimensional convolutional neural network model is the performance indicator input vector.

[0117] More specifically, in step S140, the performance indicator input vector is passed through a second one-dimensional convolutional neural network model to obtain a second-scale performance indicator associated feature vector, wherein the second one-dimensional convolutional neural network model and the first one-dimensional convolutional neural network model respectively use one-dimensional convolution kernels with different scales. In this way, the second-scale associated features of each lithium battery performance indicator are extracted by using a one-dimensional convolution kernel with a second scale, thereby obtaining a second-scale performance indicator associated feature vector.

[0118] Accordingly, in a specific example, the performance indicator input vector is passed through a second one-dimensional convolutional neural network model to obtain a second-scale performance indicator associated feature vector, including: using each layer of the second one-dimensional convolutional neural network model to perform one-dimensional convolution processing, mean pooling processing, and nonlinear activation processing on the input data in the forward pass of the layer to output the second-scale performance indicator associated feature vector from the last layer of the second one-dimensional convolutional neural network model, wherein the input of the first layer of the second one-dimensional convolutional neural network model is the performance indicator input vector.

[0119] In particular, here, the second one-dimensional convolutional neural network model and the first one-dimensional convolutional neural network model respectively use one-dimensional convolution kernels with different scales to extract multi-scale correlation feature information under different receptive fields between the various lithium battery performance indicators.

[0120] More specifically, in step S150, the first scale performance indicator associated feature vector and the second scale performance indicator associated feature vector are fused to obtain a classification feature vector, so as to fuse multi-scale lithium battery performance indicator associated feature information under different performance indicator category spans.

[0121] Accordingly, in a specific example, fusing the first scale performance indicator associated feature vector and the second scale performance indicator associated feature vector to obtain a classification feature vector includes: fusing the first scale performance indicator associated feature vector and the second scale performance indicator associated feature vector using the following cascade formula to obtain the classification feature vector; wherein the cascade formula is:

[0122] V c =Concat[V a , V b ]

[0123] Among them, V a represents the associated feature vector of the first scale performance index, V b represents the associated feature vector of the second scale performance index, Concat[·] represents the cascade function, V c represents the classification feature vector.

[0124] More specifically, in step S160, the classification feature vector is passed through a classifier to obtain a classification result, and the classification result is used to indicate whether the performance of the lithium battery to be detected meets the predetermined standard. That is, in the technical solution of the present application, the label of the classifier includes the performance of the lithium battery to be detected meets the predetermined standard (first label), and the performance of the lithium battery to be detected does not meet the predetermined standard (second label), wherein the classifier determines which classification label the classification feature vector belongs to through a soft maximum function.

[0125] It should be understood that the role of the classifier is to use the given categories and known training data to learn classification rules and classifiers, and then classify (or predict) unknown data. Logistic regression and SVM are often used to solve binary classification problems. For multi-class classification problems, logistic regression or SVM can also be used, but multiple binary classifications are required to form a multi-classification. However, this is prone to errors and is not efficient. Commonly used multi-classification methods include the Softmax classification function.

[0126] Accordingly, in a specific example, the performance evaluation method for lithium battery pole piece detection further includes a training step: for training the first one-dimensional convolutional neural network model, the second one-dimensional convolutional neural network model and the classifier. Figure 4 As shown, the training step includes: S210, obtaining training data, the training data including the training performance indicators of the lithium battery to be detected, and whether the performance of the lithium battery to be detected meets the real value of the predetermined standard; S220, arranging the training performance indicators of the lithium battery to be detected into a training performance indicator input vector according to the sample dimension; S230, passing the training performance indicator input vector through the first one-dimensional convolutional neural network model to obtain a training first-scale performance indicator associated feature vector; S240, passing the training performance indicator input vector through the second one-dimensional convolutional neural network model to obtain a training second-scale performance indicator associated feature vector, wherein the second one-dimensional convolutional neural network model and the first one-dimensional convolutional neural network model are respectively Use one-dimensional convolution kernels with different scales; S250, fuse the training first-scale performance indicator associated feature vector and the training second-scale performance indicator associated feature vector to obtain a training classification feature vector; S260, pass the training classification feature vector through the classifier to obtain a classification loss function value; S270, calculate the streaming refinement loss function value of the training first-scale performance indicator associated feature vector and the training second-scale performance indicator associated feature vector; and, S280, use the weighted sum of the classification loss function value and the streaming refinement loss function value as the loss function value, and train the first one-dimensional convolutional neural network model, the second one-dimensional convolutional neural network model and the classifier through back propagation of gradient descent.

[0127] Accordingly, in a specific example, the training classification feature vector is passed through the classifier to obtain a classification loss function value, including: the classifier processes the training classification feature vector using the following classification loss formula to obtain a training classification result, and the classification loss formula is: softmax{(W n ,Bn ):…:(W1,B1)|X}, where W1 to W n is the weight matrix, B1 to B n is the bias vector, X is the training classification feature vector; and the cross entropy value between the training classification result and the true value is calculated as the classification loss function value.

[0128] In particular, in the technical solution of the present application, here, when the training first-scale performance indicator associated feature vector and the training second-scale performance indicator associated feature vector are fused to obtain the training classification feature vector, what is essentially done is to map the parametric associated feature representation of the performance indicators at different scales expressed by the training first-scale performance indicator associated feature vector and the training second-scale performance indicator associated feature vector to a high-dimensional fusion feature space. Therefore, if the correlation between the training first-scale performance indicator associated feature vector and the training second-scale performance indicator associated feature vector obtained by the first one-dimensional convolutional neural network model and the second one-dimensional convolutional neural network model in the serialized parametric associated feature representation and the high-dimensional feature fusion space dimension can be improved, the expression effect of the training classification feature vector can be improved. Based on this, the applicant of the present application further introduces a streaming refinement loss function for the training first-scale performance indicator associated feature vector and the training second-scale performance indicator associated feature vector in addition to the classification loss function based on the training classification feature vector.

[0129] Accordingly, in a specific example, calculating the streaming refinement loss function value of the training first-scale performance indicator associated feature vector and the training second-scale performance indicator associated feature vector includes: calculating the streaming refinement loss function value of the training first-scale performance indicator associated feature vector and the training second-scale performance indicator associated feature vector using the following optimization formula; wherein the optimization formula is:

[0130]

[0131] Wherein, V1 represents the associated characteristic vector of the training first scale performance indicator, V2 represents the associated characteristic vector of the training second scale performance indicator, represents the square of the vector's two-norm, and and ⊙ represent position-by-position subtraction and multiplication of a vector, respectively. exp(·) represents the exponential operation of a vector. The exponential operation of a vector represents the calculation of the value of a natural exponential function raised to the power of the eigenvalues ​​at each position in the vector. Represents the streaming refinement loss function value.

[0132] Here, the streaming refinement loss function is based on the transformation of the parameter-serialized associated streaming distribution of the training first-scale performance index associated feature vector and the training second-scale performance index associated feature vector to the spatial distribution in the high-dimensional fusion feature space, and realizes the super-resolution improvement of the spatial distribution in the high-dimensional fusion feature space by synchronously performing the interpolation under the sequence distribution of the vector, thereby providing a more refined alignment of the distribution difference in the high-dimensional fusion feature space through the mutual probability relationship under the balanced sequence, so as to jointly present the cross-dimensional contextual association in the parameter-associated feature dimension of the performance index at different scales and the fusion feature space dimension, thereby improving the expression effect of the training classification feature vector, and then improving the accuracy of the training classification result obtained by the training classification feature vector through the classifier. In this way, the number error, position error, overall accuracy and inter-pole spacing of the positive and negative poles can be comprehensively evaluated, thereby improving the accuracy of lithium battery quality detection.

[0133] In summary, based on the performance evaluation method of lithium battery pole piece detection in the embodiment of the present application, the performance indicators of the lithium battery to be detected are first arranged as a performance indicator input vector, then the performance indicator input vector is passed through a first one-dimensional convolutional neural network model to obtain a first-scale performance indicator associated feature vector, then the performance indicator input vector is passed through a second one-dimensional convolutional neural network model to obtain a second-scale performance indicator associated feature vector, then the first-scale performance indicator associated feature vector and the second-scale performance indicator associated feature vector are fused to obtain a classification feature vector, and finally, the classification feature vector is passed through a classifier to obtain a classification result for indicating whether the performance of the lithium battery to be detected meets a predetermined standard. In this way, the performance of the lithium battery pole piece can be comprehensively and accurately evaluated, thereby better ensuring the quality and stability of the lithium battery.

[0134] Figure 5 FIG. 1 is a block diagram of a performance evaluation system 100 for lithium battery electrode detection according to an embodiment of the present application. Figure 5As shown, according to the performance evaluation system 100 for lithium battery pole piece detection of the embodiment of the present application, it includes: a performance index acquisition module 110, which is used to obtain the performance index of the lithium battery to be detected, and the performance index includes a negative pole piece-quantity error, a positive pole piece-quantity error, a negative pole piece-quantity accuracy, a positive pole piece-quantity accuracy, a positive pole piece pair-overall accuracy, a negative pole piece position error, a positive pole piece position error, and a positive and negative pole piece pair-overall position error; an input vector arrangement module 120, which is used to arrange the performance index of the lithium battery to be detected into a performance index input vector according to the sample dimension; a first convolution encoding module 130, which is used to pass the performance index input vector through a first one-dimensional convolutional neural network model to obtain a first A scale performance indicator associated feature vector; a second convolutional coding module 140, used to pass the performance indicator input vector through a second one-dimensional convolutional neural network model to obtain a second scale performance indicator associated feature vector, wherein the second one-dimensional convolutional neural network model and the first one-dimensional convolutional neural network model respectively use one-dimensional convolution kernels with different scales; a vector fusion module 150, used to fuse the first scale performance indicator associated feature vector and the second scale performance indicator associated feature vector to obtain a classification feature vector; and a classification module 160, used to pass the classification feature vector through a classifier to obtain a classification result, wherein the classification result is used to indicate whether the performance of the lithium battery to be tested meets a predetermined standard.

[0135] In one example, in the above-mentioned lithium battery electrode plate detection performance evaluation system 100, the calculation formula of the negative electrode plate-quantity error is expressed as:

[0136]

[0137] The calculation formula of the positive electrode sheet-quantity error is expressed as:

[0138]

[0139] The calculation formula of the negative electrode sheet-quantity accuracy is expressed as:

[0140]

[0141] The calculation formula of the positive electrode sheet-quantity accuracy is expressed as:

[0142]

[0143] The calculation formula of the positive and negative electrode pair-overall accuracy is expressed as:

[0144]

[0145] The calculation formula of the negative electrode position error is expressed as:

[0146]

[0147] The calculation formula of the positive electrode position error is expressed as:

[0148]

[0149] The calculation formula of the positive and negative electrode pair-overall position error is expressed as:

[0150]

[0151] in, Indicates the negative electrode sheet-quantity error, Indicates the positive electrode sheet-quantity error, Indicates the accuracy of the number of negative electrode sheets, Indicates the positive electrode sheet-quantity accuracy, Indicates the overall accuracy of the positive and negative electrode pairs, represents the position error of the negative electrode sheet, represents the positive electrode position error, Indicates the overall position error of the positive and negative electrode pairs, Indicates the number of negative plates to be tested. Indicates the number of negative electrodes corresponding to the true value, Indicates the number of positive plates to be tested. Indicates the number of positive plates corresponding to the true value, Indicates the total number of positive and negative electrodes to be tested. Indicates the total number of positive and negative electrodes of the corresponding true value, represents the spatial coordinates of the jth negative electrode sheet of the i-th sample to be tested, represents the spatial coordinates of the jth negative electrode of the ith sample of the corresponding true value, represents the spatial coordinates of the jth positive electrode of the i-th sample to be tested, represents the spatial coordinates of the jth negative electrode sheet of the i-th sample of the corresponding true value, and N represents the total number of samples in the test set.

[0152] In one example, in the above-mentioned performance evaluation system 100 for lithium battery pole piece detection, the first convolution encoding module 130 is used to: use each layer of the first one-dimensional convolutional neural network model to perform one-dimensional convolution processing, mean pooling processing and nonlinear activation processing on the input data in the forward pass of the layer so as to output the first-scale performance indicator associated feature vector from the last layer of the first one-dimensional convolutional neural network model, wherein the input of the first layer of the first one-dimensional convolutional neural network model is the performance indicator input vector.

[0153] In one example, in the above-mentioned performance evaluation system 100 for lithium battery pole piece detection, the second convolution encoding module 140 is used to: use each layer of the second one-dimensional convolutional neural network model to perform one-dimensional convolution processing, mean pooling processing and nonlinear activation processing on the input data in the forward pass of the layer to output the second-scale performance indicator associated feature vector from the last layer of the second one-dimensional convolutional neural network model, wherein the input of the first layer of the second one-dimensional convolutional neural network model is the performance indicator input vector.

[0154] In one example, in the above-mentioned lithium battery pole piece detection performance evaluation system 100, the vector fusion module 150 is used to: fuse the first scale performance index associated feature vector and the second scale performance index associated feature vector using the following cascade formula to obtain the classification feature vector; wherein the cascade formula is:

[0155] V c =Concat[V a , V b ]

[0156] Among them, V a represents the associated feature vector of the first scale performance index, V b represents the associated feature vector of the second scale performance index, Concat[·] represents the cascade function, V c represents the classification feature vector.

[0157] In one example, in the above-mentioned performance evaluation system 100 for lithium battery pole piece detection, it also includes a training module for training the first one-dimensional convolutional neural network model, the second one-dimensional convolutional neural network model and the classifier. The training module is used to: obtain training data, the training data includes the training performance indicators of the lithium battery to be detected, and the true value of whether the performance of the lithium battery to be detected meets the predetermined standard; arrange the training performance indicators of the lithium battery to be detected into a training performance indicator input vector according to the sample dimension; pass the training performance indicator input vector through the first one-dimensional convolutional neural network model to obtain a training first-scale performance indicator associated feature vector; pass the training performance indicator input vector through the second one-dimensional convolutional neural network model to obtain a training second-scale performance indicator associated feature vector, wherein the second one-dimensional convolutional neural network model and the first one-dimensional convolutional neural network model are respectively used One-dimensional convolution kernels with different scales; fusing the training first-scale performance indicator associated feature vector and the training second-scale performance indicator associated feature vector to obtain a training classification feature vector; passing the training classification feature vector through the classifier to obtain a classification loss function value; calculating the streaming refinement loss function value of the training first-scale performance indicator associated feature vector and the training second-scale performance indicator associated feature vector; and, using the weighted sum of the classification loss function value and the streaming refinement loss function value as the loss function value, and training the first one-dimensional convolutional neural network model, the second one-dimensional convolutional neural network model and the classifier through back propagation of gradient descent.

[0158] In one example, in the above-mentioned lithium battery pole piece detection performance evaluation system 100, the training classification feature vector is passed through the classifier to obtain the classification loss function value, including: the classifier processes the training classification feature vector with the following classification loss formula to obtain the training classification result, and the classification loss formula is: softmax{(W n ,B n ):…:(W1,B1)|X}, where W1 to W n is the weight matrix, B1 to B n is the bias vector, X is the training classification feature vector; and the cross entropy value between the training classification result and the true value is calculated as the classification loss function value.

[0159] In one example, in the above-mentioned performance evaluation system 100 for lithium battery pole piece detection, calculating the streaming refinement loss function value of the training first scale performance index associated feature vector and the training second scale performance index associated feature vector includes: calculating the streaming refinement loss function value of the training first scale performance index associated feature vector and the training second scale performance index associated feature vector using the following optimization formula; wherein the optimization formula is:

[0160]

[0161] Wherein, V1 represents the associated characteristic vector of the training first scale performance indicator, V2 represents the associated characteristic vector of the training second scale performance indicator, represents the square of the vector's two-norm, and and ⊙ represent position-by-position subtraction and multiplication of a vector, respectively. exp(·) represents the exponential operation of a vector. The exponential operation of a vector represents the calculation of the value of a natural exponential function raised to the power of the eigenvalues ​​at each position in the vector. Represents the streaming refinement loss function value.

[0162] Here, those skilled in the art will appreciate that the specific functions and operations of each module in the performance evaluation system 100 for lithium battery pole piece detection have been described in the above reference. Figures 1 to 4 The performance evaluation method of lithium battery pole piece detection has been introduced in detail in the description, and therefore, its repeated description will be omitted.

[0163] As described above, the performance evaluation system 100 for lithium battery pole piece detection according to the embodiment of the present application can be implemented in various wireless terminals, such as a server having a performance evaluation algorithm for lithium battery pole piece detection. In one example, the performance evaluation system 100 for lithium battery pole piece detection according to the embodiment of the present application can be integrated into a wireless terminal as a software module and / or a hardware module. For example, the performance evaluation system 100 for lithium battery pole piece detection can be a software module in the operating system of the wireless terminal, or can be an application developed for the wireless terminal; of course, the performance evaluation system 100 for lithium battery pole piece detection can also be one of the many hardware modules of the wireless terminal.

[0164] Alternatively, in another example, the performance evaluation system 100 for lithium battery pole piece detection and the wireless terminal may also be separate devices, and the performance evaluation system 100 for lithium battery pole piece detection 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.

[0165] According to another aspect of the present application, a non-volatile computer-readable storage medium is provided, on which computer-readable instructions are stored. When the instructions are executed by a computer, the above-mentioned method can be executed.

[0166] The program part of the technology can be considered as a "product" or "manufactured product" in the form of executable code and / or related data, which is participated in or realized by computer-readable media. Tangible and permanent storage media can include any memory or storage used by any computer, processor, or similar device or related module. For example, various semiconductor memories, tape drives, disk drives or any similar devices that can provide storage functions for software.

[0167] All or part of the software may sometimes communicate over a network, such as the Internet or other communications network. Such communications can load software from one computer device or processor to another. For example: from a server or host computer of a video target detection device to a hardware platform of a computer environment, or other computer environment that implements the system, or a system with similar functions related to providing information required for target detection. Therefore, another medium capable of transmitting software elements can also be used as a physical connection between local devices, such as light waves, radio waves, electromagnetic waves, etc., transmitted through cables, optical cables or air. Physical media used to carry carriers, such as cables, wireless connections or optical cables and the like, can also be considered as media that carry software. As used herein, unless limited to tangible "storage" media, other terms referring to computer or machine "readable media" refer to media involved in the process of the processor executing any instructions.

[0168] The present application uses specific words to describe the embodiments of the present application. For example, "first / second embodiment", "one embodiment", and / or "some embodiments" refer to a certain feature, structure or characteristic related to at least one embodiment of the present application. Therefore, it should be emphasized and noted that "one embodiment" or "an embodiment" or "an alternative embodiment" mentioned twice or multiple times in different positions in this specification does not necessarily refer to the same embodiment. In addition, some features, structures or characteristics in one or more embodiments of the present application can be appropriately combined.

[0169] In addition, it will be appreciated by those skilled in the art that various aspects of the present application may be illustrated and described by a number of patentable categories or situations, including any new and useful process, machine, product or combination of substances, or any new and useful improvements thereto. Accordingly, various aspects of the present application may be performed entirely by hardware, entirely by software (including firmware, resident software, microcode, etc.), or by a combination of hardware and software. The above hardware or software may all be referred to as "data blocks", "modules", "engines", "units", "components" or "systems". In addition, various aspects of the present application may be represented as a computer product located in one or more computer-readable media, which includes computer-readable program code.

[0170] Unless otherwise defined, all terms (including technical and scientific terms) used herein have the same meaning as commonly understood by one of ordinary skill in the art to which the invention belongs. It should also be understood that terms such as those defined in common dictionaries should be interpreted as having a meaning consistent with their meaning in the context of the relevant technology and should not be interpreted in an idealized or extremely formal sense, unless explicitly defined as such herein.

[0171] The above is an explanation of the present invention and should not be considered as a limitation thereof. Although several exemplary embodiments of the present invention have been described, it will be readily appreciated by those skilled in the art that many modifications may be made to the exemplary embodiments without departing from the novel teachings and advantages of the present invention. Therefore, all such modifications are intended to be included within the scope of the present invention as defined in the claims. It should be understood that the above is an explanation of the present invention and should not be considered as being limited to the specific embodiments disclosed, and modifications to the disclosed embodiments and other embodiments are intended to be included within the scope of the appended claims. The present invention is defined by the claims and their equivalents.

Claims

1. A performance evaluation method for lithium battery pole piece detection, characterized in that: include: Obtaining performance indicators of the lithium battery to be tested, the performance indicators including negative electrode sheet-quantity error, positive electrode sheet-quantity error, negative electrode sheet-quantity accuracy, positive electrode sheet-quantity accuracy, positive and negative electrode sheet pair-overall accuracy, negative electrode sheet position error, positive electrode sheet position error and positive and negative electrode sheet pair-overall position error; Arranging the performance indicators of the lithium battery to be tested into a performance indicator input vector according to sample dimensions; Passing the performance indicator input vector through a first one-dimensional convolutional neural network model to obtain a first scale performance indicator associated feature vector; Passing the performance indicator input vector through a second one-dimensional convolutional neural network model to obtain a second-scale performance indicator associated feature vector, wherein the second one-dimensional convolutional neural network model and the first one-dimensional convolutional neural network model respectively use one-dimensional convolution kernels with different scales; fusing the first scale performance indicator associated feature vector and the second scale performance indicator associated feature vector to obtain a classification feature vector; and The classification feature vector is passed through a classifier to obtain a classification result, and the classification result is used to indicate whether the performance of the lithium battery to be tested meets a predetermined standard.

2. The performance evaluation method for lithium battery pole piece detection according to claim 1, characterized in that: The calculation formula of the negative electrode sheet-quantity error is expressed as: The calculation formula of the positive electrode sheet-quantity error is expressed as: The calculation formula of the negative electrode sheet-quantity accuracy is expressed as: The calculation formula of the positive electrode sheet-quantity accuracy is expressed as: The calculation formula of the positive and negative electrode pair-overall accuracy is expressed as: The calculation formula of the negative electrode position error is expressed as: The calculation formula of the positive electrode position error is expressed as: The calculation formula of the positive and negative electrode pair-overall position error is expressed as: in, Indicates the negative electrode sheet-quantity error, Indicates the positive electrode sheet-quantity error, Indicates the accuracy of the number of negative electrode sheets, Indicates the positive electrode sheet-quantity accuracy, Indicates the overall accuracy of the positive and negative electrode pairs, represents the position error of the negative electrode sheet, represents the positive electrode position error, Indicates the overall position error of the positive and negative electrode pairs, Indicates the number of negative plates to be tested. Indicates the number of negative electrodes corresponding to the true value, Indicates the number of positive plates to be tested. Indicates the number of positive plates corresponding to the true value, Indicates the total number of positive and negative electrodes to be tested. Indicates the total number of positive and negative electrodes of the corresponding true value, represents the spatial coordinates of the jth negative electrode sheet of the i-th sample to be tested, represents the spatial coordinates of the jth negative electrode of the ith sample of the corresponding true value, represents the spatial coordinates of the jth positive electrode of the i-th sample to be tested, represents the spatial coordinates of the jth negative electrode sheet of the i-th sample of the corresponding true value, and N represents the total number of samples in the test set.

3. The performance evaluation method for lithium battery pole piece detection according to claim 2, characterized in that: Passing the performance indicator input vector through a first one-dimensional convolutional neural network model to obtain a first scale performance indicator associated feature vector, including: Use each layer of the first one-dimensional convolutional neural network model to perform one-dimensional convolution processing, mean pooling processing and nonlinear activation processing on the input data in the forward pass of the layer so that the last layer of the first one-dimensional convolutional neural network model outputs the first-scale performance indicator associated feature vector, wherein the input of the first layer of the first one-dimensional convolutional neural network model is the performance indicator input vector.

4. The performance evaluation method for lithium battery pole piece detection according to claim 3, characterized in that: Passing the performance indicator input vector through a second one-dimensional convolutional neural network model to obtain a second scale performance indicator associated feature vector, including: Use each layer of the second one-dimensional convolutional neural network model to perform one-dimensional convolution processing, mean pooling processing and nonlinear activation processing on the input data in the forward pass of the layer so that the last layer of the second one-dimensional convolutional neural network model outputs the second-scale performance indicator associated feature vector, wherein the input of the first layer of the second one-dimensional convolutional neural network model is the performance indicator input vector.

5. The performance evaluation method for lithium battery pole piece detection according to claim 4, characterized in that: Fusion of the first scale performance indicator associated feature vector and the second scale performance indicator associated feature vector to obtain a classification feature vector includes: The first scale performance indicator associated feature vector and the second scale performance indicator associated feature vector are fused using the following cascade formula to obtain the classification feature vector; Wherein, the cascade formula is: In c =Concat[V a ,In b ] Among them, V a represents the associated feature vector of the first scale performance index, V b represents the associated feature vector of the second scale performance index, Concat[·] represents the cascade function, V c represents the classification feature vector.

6. The performance evaluation method for lithium battery pole piece detection according to claim 5, characterized in that: It also includes a training step: used to train the first one-dimensional convolutional neural network model, the second one-dimensional convolutional neural network model and the classifier.

7. The performance evaluation method for lithium battery pole piece detection according to claim 6, characterized in that: The training step comprises: Acquire training data, wherein the training data includes a training performance index of the lithium battery to be tested and a true value of whether the performance of the lithium battery to be tested meets a predetermined standard; Arranging the training performance indicators of the lithium battery to be tested into a training performance indicator input vector according to the sample dimension; Passing the training performance indicator input vector through the first one-dimensional convolutional neural network model to obtain a training first-scale performance indicator associated feature vector; Passing the training performance indicator input vector through the second one-dimensional convolutional neural network model to obtain a training second-scale performance indicator associated feature vector, wherein the second one-dimensional convolutional neural network model and the first one-dimensional convolutional neural network model respectively use one-dimensional convolution kernels with different scales; fusing the training first scale performance indicator associated feature vector and the training second scale performance indicator associated feature vector to obtain a training classification feature vector; Passing the training classification feature vector through the classifier to obtain a classification loss function value; Calculating the streaming refinement loss function value of the training first-scale performance indicator associated feature vector and the training second-scale performance indicator associated feature vector; and The weighted sum of the classification loss function value and the streaming refinement loss function value is used as the loss function value, and the first one-dimensional convolutional neural network model, the second one-dimensional convolutional neural network model and the classifier are trained through back propagation of gradient descent.

8. The performance evaluation method for lithium battery pole piece detection according to claim 7, characterized in that: Passing the training classification feature vector through the classifier to obtain a classification loss function value includes: The classifier processes the training classification feature vector using the following classification loss formula to obtain a training classification result. The classification loss formula is: softmax{(W n , B n ):...:(W1,B1)|X}, where W1 to W n is the weight matrix, B1 to B n is a bias vector, X is the training classification feature vector; and A cross entropy value between the training classification result and the true value is calculated as the classification loss function value.

9. The performance evaluation method for lithium battery pole piece detection according to claim 8, characterized in that: Calculating the streaming refinement loss function value of the training first scale performance indicator associated feature vector and the training second scale performance indicator associated feature vector, including: The streaming refinement loss function value of the training first scale performance indicator associated feature vector and the training second scale performance indicator associated feature vector is calculated using the following optimization formula; Wherein, the optimization formula is: Wherein, V1 represents the associated characteristic vector of the training first scale performance indicator, V2 represents the associated characteristic vector of the training second scale performance indicator, represents the square of the two-norm of the vector, and θ and ⊙ represent the position-by-position subtraction and multiplication of the vector respectively, exp(·) represents the exponential operation of the vector, and the exponential operation of the vector represents the calculation of the natural exponential function value raised to the power of the eigenvalue of each position in the vector, Represents the streaming refinement loss function value.

10. A performance evaluation method for lithium battery pole piece detection, characterized in that: include: A performance index acquisition module is used to acquire performance indicators of the lithium battery to be tested, wherein the performance indicators include negative electrode sheet-quantity error, positive electrode sheet-quantity error, negative electrode sheet-quantity accuracy, positive electrode sheet-quantity accuracy, positive and negative electrode sheet pair-overall accuracy, negative electrode sheet position error, positive electrode sheet position error and positive and negative electrode sheet pair-overall position error; An input vector arrangement module, used to arrange the performance indicators of the lithium battery to be tested into a performance indicator input vector according to the sample dimension; A first convolutional encoding module, used for passing the performance indicator input vector through a first one-dimensional convolutional neural network model to obtain a first scale performance indicator associated feature vector; A second convolutional encoding module, used for passing the performance indicator input vector through a second one-dimensional convolutional neural network model to obtain a second-scale performance indicator associated feature vector, wherein the second one-dimensional convolutional neural network model and the first one-dimensional convolutional neural network model respectively use one-dimensional convolution kernels with different scales; a vector fusion module, configured to fuse the first scale performance indicator associated feature vector and the second scale performance indicator associated feature vector to obtain a classification feature vector; and The classification module is used to pass the classification feature vector through a classifier to obtain a classification result, and the classification result is used to indicate whether the performance of the lithium battery to be tested meets a predetermined standard.

Images