Method and equipment for identifying and classifying state of electric pile, and storage medium

Abstract

The invention discloses an electric pile state identification and classification method, equipment and a storage medium, and relates to the field of battery data calibration. The method comprises the following steps: according to the similarity between each type of galvanic pile parameters and galvanic pile voltage, determining the correlation between each type of galvanic pile parameters and galvanic pile voltage; selecting galvanic pile parameters with large correlation as training data; training the training data through a self-organizing mapping neural network to obtain an optimal clustering model for clustering the pile parameters; clustering the training data through the optimal clustering model to obtain a plurality of clusters of clustering data; generating a pseudo pile state label for each type of clustering data and then forming a sample set; and training the sample set through a one-dimensional convolutional neural network to obtain a pile state classification model for pile state classification. According to the method, the workload of manual labeling is reduced, and the state of the galvanic pile can be effectively identified and classified.

Description

Technical Field

[0001] This application relates to the field of battery data calibration, specifically to a method, device, and storage medium for identifying and classifying the state of a battery stack. Background Technology

[0002] With the increasing demands for environmental protection and energy, fuel cells, as a clean and efficient energy conversion technology, have received widespread attention. However, fuel cells may encounter various faults during operation, which not only affect their performance and lifespan but also pose potential threats to the environment and safety. Therefore, accurate fault diagnosis is necessary to ensure the reliable operation of fuel cells and extend their service life.

[0003] Data-driven fuel cell fault diagnosis utilizes a large amount of real-world operational data, employing advanced algorithms and models to monitor and analyze various parameters of the fuel cell in real time, thereby accurately diagnosing faults and performance status. Data-driven fault diagnosis models require a large amount of labeled data; however, manually labeling vast amounts of fuel cell data is extremely resource-intensive, significantly extending the development cycle of fuel cell fault diagnosis models.

[0004] Chinese invention patent CN115598536A, entitled "A Fault Diagnosis Method for PEMFC Based on Fuzzy C-means Clustering and Probabilistic Neural Network," is based on a modified Fouquet equivalent circuit model of a fuel cell stack. Combined with EIS impedance spectrum experimental data of the stack, it obtains the correspondence between four health states of the stack—membrane dryness, flooding, oxygen starvation, and normal—and the circuit model parameters. Appropriate fault features are then extracted as feature inputs for the clustering algorithm. The invention then uses a fuzzy C-means clustering algorithm to cluster fault sample data, forming standard cluster centers. A probabilistic neural network algorithm is employed to classify multiple faults in the fault samples, effectively removing outliers and improving the model's classification accuracy. An example analysis of 200 sets of experimental data is presented, and the proposed method is compared with support vector machines and the K-nearest neighbor method, demonstrating that it can quickly identify the four stack operating states.

[0005] but: The fuzzy c-means clustering used requires setting a certain number of categories, and it cannot effectively identify new faults. Furthermore, this clustering algorithm is sensitive to the initial value setting and has high computational complexity.

[0006] At the same time, using probabilistic neural networks as classifiers is more suitable for small-scale data and not for large-scale data processing. Summary of the Invention

[0007] To address the shortcomings of existing technologies, the technical problem solved by this invention is: how to effectively identify and classify fuel cell stack conditions while reducing the workload of manual labeling.

[0008] To achieve the above objectives, in a first aspect, embodiments of this application provide a method for identifying and classifying the state of an electric stack, the method comprising the following steps: Based on the similarity between the parameters and voltage of each type of fuel cell stack, the correlation between the parameters and voltage of each type of fuel cell stack is determined; voltage-sensitive fuel cell stacks and fuel cell stack parameters with high correlation with them are selected as training data. After training the training data through a self-organizing map neural network, the optimal clustering model for clustering the fuel cell stack parameters is obtained. The training data is clustered using the optimal clustering model to obtain several clustered data classes; a pseudo-pile state label is generated for each clustered data class to form a sample set; The sample set is trained using a one-dimensional convolutional neural network to obtain a fuel cell state classification model. After classifying the fuel cell state using the fuel cell state classification model, the fuel cell state label is obtained.

[0009] In conjunction with the first aspect, in one embodiment, the process of determining the similarity between each type of fuel cell stack parameters and the fuel cell stack voltage includes: determining the similarity between each type of fuel cell stack parameters and the fuel cell stack voltage based on the covariance between each type of fuel cell stack parameters and the fuel cell stack voltage.

[0010] In conjunction with the first aspect, in one embodiment, the formula for calculating the similarity r between each type of stack parameter and stack voltage is as follows:

[0011] Where x and y represent parameters for different types of fuel cell stacks, such as x being the stack voltage and y being the stack current. Represents the mean of x. This represents the mean of y; The absolute value of r represents the correlation score between each type of stack parameter and the stack voltage.

[0012] In conjunction with the first aspect, in one implementation, the training process of the optimal clustering model includes: Initialize the neuron weights of the competing layer nodes in the self-organizing map neural network, and set the initial learning rate and neighborhood radius; All training data are divided into several groups of node data according to the time when the test data is acquired, and each group of node data serves as an input layer node. Based on the similarity between each input layer node and each competing layer node, determine the winning node in the competing layer corresponding to each input layer node; Update the neighborhood radius of the winning node in the competition layer, update the learning rate based on the updated neighborhood radius, and determine the neighborhood range of the winning node in the competition layer; based on the updated learning rate, update the neuron weights of other nodes in the competition layer within the neighborhood range of the winning node in the competition layer.

[0013] In conjunction with the first aspect, in one embodiment, the process of training the sample set to obtain the electric pile state classification model through a one-dimensional convolutional neural network includes: dividing the sample set into K subsets with the same number of data points and different data points; performing K training cycles through the one-dimensional convolutional neural network, selecting K-1 subsets as the training set and the remaining subsets as the test set during each training cycle.

[0014] In conjunction with the first aspect, in one implementation, the fuel cell stack state classification model includes: One-dimensional convolution is used to extract features from input data. Max pooling layers are used to: extract key features; The fully connected layer is used to provide the probability of each stack state category, with the category with the highest probability being the final prediction.

[0015] In conjunction with the first aspect, in one embodiment, the process of obtaining the battery stack parameters includes: testing the battery stack to obtain test data, and after denoising and normalizing the test data, classifying it into several types of battery stack parameters according to type.

[0016] In conjunction with the first aspect, in one implementation, the highly correlated stack parameters are: the stack parameters corresponding to the first P correlation scores of each category, where P is a preset value.

[0017] Secondly, embodiments of this application provide a device for identifying and classifying the state of an electric stack. The device includes a processor, a memory, and a program for identifying and classifying the state of an electric stack stored in the memory and executable by the processor. When the program for identifying and classifying the state of an electric stack is executed by the processor, it implements the method provided in the first aspect.

[0018] Thirdly, embodiments of this application provide a computer-readable storage medium storing a program for identifying and classifying the state of an electric stack, wherein the computer program, when executed, implements the method provided in the first aspect.

[0019] Compared with the prior art, the advantages of the present invention are as follows: This application obtains the optimal clustering model through training a self-organizing map neural network. The number of nodes in the competition layer of the self-organizing map neural network is flexible, as it is set according to the maximum number of categories required. Unlike the fuzzy C-means in existing technologies, this setting does not require prior knowledge of the specific number of categories in the samples. Therefore, this application effectively avoids the difficulties caused by the unknown number of categories in the samples, and has greater flexibility and practicality. Moreover, compared with the manual annotation of large amounts of data in existing technologies, this application only requires clustering data obtained from the optimal clustering model, significantly reducing the workload of manual annotation.

[0020] Furthermore, this application uses pseudo-stack state labels obtained through training a self-organizing map neural network to obtain a stack state classification model after training a one-dimensional convolutional neural network. This model can be used for fuel cell stack state classification to classify stack states and obtain stack state labels. The method is simple and effective and can provide a reliable decision basis for classification tasks.

[0021] In summary, this application not only reduces the workload of manual annotation, but also enables effective identification and classification of fuel cell stack states. Furthermore, the method of training a combination of self-organizing mapping neural networks and one-dimensional convolutional neural networks is applicable to large amounts of data. Attached Figure Description

[0022] To more clearly illustrate the technical solutions in the embodiments of the present invention, the accompanying drawings used in the description of the embodiments will be briefly introduced below. Obviously, the accompanying drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0023] Figure 1 This is a flowchart illustrating the identification and classification of the fuel cell stack state in an embodiment of this application; Figure 2 This is a schematic diagram showing the correlation coefficients between the fuel cell stack parameters and voltage in the embodiments of this application; Figure 3 This is a flowchart illustrating the training process of the self-organizing map neural network in this application embodiment; Figure 4 This is a comparison chart of different clustering algorithms and manually labeled classifications in the embodiments of this application; Figure 5 This is a schematic diagram of the structure of the fuel cell stack classification model in the embodiments of this application; Figure 6 This is a schematic diagram of the hardware structure of the fuel cell stack state identification and classification device involved in the embodiments of this application. Detailed Implementation

[0024] To make the objectives, technical solutions, and advantages of the embodiments of this application clearer, the technical solutions of the embodiments of this application will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of this application, not all embodiments. Based on the embodiments of this application, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of this application.

[0025] The flowchart shown in the attached diagram is for illustrative purposes only and does not necessarily include all content and operations / steps, nor does it necessarily have to be performed in the order described. For example, some operations / steps can be broken down, combined, or partially merged, so the actual execution order may change depending on the actual situation.

[0026] To make the objectives, technical solutions, and advantages of this application clearer, the embodiments of this application will be described in further detail below with reference to the accompanying drawings.

[0027] In a first aspect, embodiments of this application provide a method for identifying and classifying the state of an electric stack, the method comprising the following steps: S1: Determine the required stack parameters based on the battery stack test data. The types of stack parameters include stack voltage.

[0028] S2: Determine the correlation between the parameters and voltage of each type of fuel cell stack based on the similarity between the parameters and voltage of each type of fuel cell stack; select voltage-sensitive fuel cell stacks and fuel cell stack parameters with high correlation with them as training data.

[0029] S3: After training the training data through a self-organizing map neural network, the optimal clustering model for clustering the stack parameters is obtained.

[0030] S4: Cluster the training data using the optimal clustering model to obtain several cluster data classes; generate pseudo-pile state labels for each cluster data class to form a sample set.

[0031] S5: Train the sample set through a one-dimensional convolutional neural network to obtain a fuel cell state classification model. After classifying the fuel cell state through the fuel cell state classification model, obtain the fuel cell state label.

[0032] Therefore, this application obtains the optimal clustering model through training a self-organizing map neural network. The number of nodes in the competition layer of the self-organizing map neural network is flexible, as it is set according to the maximum number of categories required. Unlike the fuzzy C-means in existing technologies, this setting does not require prior knowledge of the specific number of categories in the samples. Thus, this application effectively avoids the difficulties caused by the unknown number of categories in the samples, exhibiting greater flexibility and practicality. Moreover, compared to the manual annotation of large amounts of data in existing technologies, this application only requires clustering data obtained from the optimal clustering model, significantly reducing the workload of manual annotation.

[0033] Furthermore, this application uses pseudo-pile state labels obtained through training a self-organizing map neural network to train a one-dimensional convolutional neural network, which can then be used to obtain a pile state classifier. This classifier can be used for pile state classification, and the method is simple and effective, providing a reliable decision basis for classification tasks.

[0034] In summary, this application not only reduces the workload of manual annotation, but also enables effective identification and classification of fuel cell stack states. Furthermore, the method of training a combination of self-organizing mapping neural networks and one-dimensional convolutional neural networks is applicable to large amounts of data.

[0035] In one embodiment, the process of determining the battery stack parameters to be used based on the test data of the battery stack in S1 above includes: testing the battery stack to obtain test data (this step is a conventional process), and after denoising and normalizing the test data, classifying it into several types of battery stack parameters according to type.

[0036] Specifically, the process of denoising test data includes: filtering duplicate test data (to ensure data uniqueness), outlier data with values ​​exceeding the theoretical range, non-numerical data, and data with missing values.

[0037] In one embodiment, the process of determining the similarity between each type of stack parameter and stack voltage in S2 above includes: determining the similarity between each type of stack parameter and stack voltage based on the covariance between each type of stack parameter and stack voltage.

[0038] Specifically, the formula for calculating the similarity r between the parameters and voltage of each type of fuel cell stack is as follows:

[0039] Where x and y represent parameters for different types of fuel cell stacks, such as x being the stack voltage and y being the stack current. Represents the mean of x. This represents the mean of y.

[0040] Based on this, the absolute value of r represents the correlation score between each type of stack parameter and the stack voltage. The design principle is as follows: the numerator of the above formula calculates the covariance of two different variables, and the denominator represents the standardization of the covariance; in the correlation score, a positive number indicates a positive correlation, and a negative number indicates a negative correlation. The closer the score is to 0, the weaker the correlation; while the closer the score is to 1 or -1, the stronger the correlation; therefore, the larger the absolute value of r, the stronger the correlation.

[0041] Meanwhile, the above-mentioned highly correlated fuel cell stack parameters are: the fuel cell stack parameters corresponding to the first P correlation scores of each category. The value of P is a preset value, and the specific value of P can be set according to the specific classification and diagnostic requirements.

[0042] For the stack characteristics other than voltage selected in this embodiment, and their correlation scores (i.e., correlation coefficients) with voltage, please refer to [link to relevant documentation]. Figure 2 As shown.

[0043] See Figure 3 As shown, the training process of the optimal clustering model in S3 above includes: S301: Initialize the neuron weights of the competing layer nodes in the self-organizing map neural network, and set the initial learning rate and neighborhood radius.

[0044] S302: Divide all training data into several groups of node data according to the test data acquisition time, and each group of node data serves as an input layer node (i.e., all training data obtained at the same time serve as an input layer node).

[0045] S303: Determine the winning node of the competitive layer corresponding to each input layer node based on the similarity between each input layer node and each competing layer node.

[0046] S304: Update the neighborhood radius based on the winning node of the competition layer, update the learning rate based on the updated neighborhood radius, and determine the neighborhood range of the winning node of the competition layer; based on the updated learning rate, update the neuron weights of other nodes in the neighborhood range of the winning node of the competition layer.

[0047] S305: Determine if the maximum number of iterations has been reached. If yes, select the optimal clustering model and end the process; otherwise, proceed to S303.

[0048] The above describes the training process of a self-organizing map neural network (SOM) with specific parameters. To obtain the optimal clustering results, different parameter combinations need to be tried when training the SOM. Therefore, a parametric grid algorithm is used to search for the parameter combination that yields the optimal clustering performance. Specifically, it is necessary to set the key parameter limits for the SOM, covering hyperparameters such as the learning rate and the number of training iterations. The parametric grid search strategy is then used to optimize the clustering performance. The training epochs range for the SOM is set to [10000, 60000], and the upper and lower limits of the learning rate are set to [0.001, 0.01]. The parametric grid method is used to combine these parameters. During the combination iteration process, the optimal clustering performance of each parameter group is recorded and compared with the current globally optimal clustering performance. If it is better, the optimal parameter combination is updated, and the search for the optimal parameter combination continues until the iteration condition is met. The optimal parameter combination at the end of the iteration is the globally optimal combination.

[0049] See Figure 4 The image shows the results of self-organizing map neural network (SOM), K-Means clustering algorithm, and manually labeled algorithm under the same input feature parameters. Figure 4 The results show that the color alignment of the manually labeled data with the output labels of the self-organizing map neural network has not yet been performed, so there are instances where the manually labeled data and the clustering labels, despite being classified in the same way, have inconsistent colors. Figure 4 It can be seen that, under the same input feature parameters, the clustering results obtained by the K-Means algorithm differ significantly from the manual labels, and the number of clusters obtained does not conform to the actual situation. In contrast, the clustering results obtained by the self-organizing map neural network are basically close to the manual labels, and can basically correctly classify the five different stack states (normal state, high activity, dry membrane, low activity, and wet membrane).

[0050] In one embodiment, the process of training the sample set to obtain the stack state classification model through a one-dimensional convolutional neural network in S5 above includes: dividing the sample set into K subsets with the same number of data points and different data points (the number of K is set according to different scenarios, generally 3 to 5); performing K training cycles through a one-dimensional convolutional neural network, selecting K-1 subsets as the training set each time, and using the remaining subsets as the test set, with the stack state categories of the test set being normal, dry membrane, wet membrane, high activity, and low activity.

[0051] For details, see Figure 5 As shown, the fuel cell stack state classification model includes: One-dimensional convolution is used to: efficiently extract multiple features from input data by utilizing the characteristics of diverse convolution kernels; thus, it is not only good at capturing local details of data, but also can automatically learn valuable feature representations during the training process, providing a solid foundation for data analysis and model building.

[0052] Max pooling layers are used to: extract key features, thereby reducing data redundancy and preventing overfitting, while expanding the receptive field of convolutional layers and retaining key features, thus improving the performance and generalization ability of the model.

[0053] The fully connected layer is used to provide the probability of each stack state category. After processing by the fully connected layer, the model can output the predicted probability corresponding to each category. Among these probabilities, the category corresponding to the largest one is the final prediction result.

[0054] To verify the effectiveness of the above-mentioned fuel cell state classification model, the same training set was trained three times using the above-mentioned fuel cell state classification model and the SVM-based fuel cell state classifier, respectively, to obtain fuel cell state classification labels. The labels for the test set were manually labeled.

[0055] The fuel cell state classification labels obtained from each training of the fuel cell state classification model and the SVM-based fuel cell state classifier were compared with manually labeled labels. The average accuracy of the SVM-based fuel cell state classifier (accuracies of 89.26%, 86.52%, and 88.60% in three tests) was 88.13%, while the average accuracy of the aforementioned fuel cell state classification model (accuracies of 97.19%, 96.39%, and 96.39% in three tests) was 96.66%.

[0056] Secondly, embodiments of this application provide a device for identifying and classifying the state of an electric stack. This device can be a personal computer (PC), a laptop computer, a server, or other device with data processing capabilities.

[0057] Reference Figure 6 , Figure 6 This is a schematic diagram of the hardware structure of the fuel cell state identification and classification device involved in the embodiments of this application. In the embodiments of this application, the fuel cell state identification and classification device may include a processor, a memory, a communication interface, and a communication bus.

[0058] The communication bus can be of any type and is used to interconnect the processor, memory, and communication interface.

[0059] The communication interface includes input / output (I / O) interfaces, physical interfaces, and logical interfaces used for interconnecting devices within the fuel cell state identification and classification device, as well as interfaces used for interconnecting the fuel cell state identification and classification device with other devices (such as other computing devices or user equipment). Physical interfaces can be Ethernet interfaces, fiber optic interfaces, ATM interfaces, etc.; user equipment can be displays, keyboards, etc.

[0060] Memory can be various types of storage media, such as random access memory (RAM), read-only memory (ROM), non-volatile RAM (NVRAM), flash memory, optical storage, hard disk, programmable ROM (PROM), erasable PROM (EPROM), electrically erasable PROM (EEPROM), etc.

[0061] The processor can be a general-purpose processor, which can call the fuel cell state identification and classification program stored in the memory and execute the fuel cell state identification and classification method provided in the embodiments of this application. For example, the general-purpose processor can be a central processing unit (CPU). The method executed when the fuel cell state identification and classification program is called can refer to the various embodiments of the fuel cell state identification and classification method of this application, and will not be repeated here.

[0062] Those skilled in the art will understand that Figure 6 The hardware structure shown does not constitute a limitation of this application and may include more or fewer components than shown, or combine certain components, or have different component arrangements.

[0063] Thirdly, embodiments of this application also provide a computer-readable storage medium.

[0064] The computer-readable storage medium of this application stores a program for identifying and classifying the state of a fuel cell, wherein when the program for identifying and classifying the state of a fuel cell is executed by a processor, it implements the steps of the method for identifying and classifying the state of a fuel cell as described above.

[0065] The method implemented when the fuel cell state identification and classification procedure is executed can be referred to in various embodiments of the fuel cell state identification and classification method of this application, and will not be repeated here.

[0066] It should be noted that the sequence numbers of the embodiments in this application are for descriptive purposes only and do not represent the superiority or inferiority of the embodiments.

[0067] The terms "comprising" and "having," and any variations thereof, in the specification, claims, and accompanying drawings of this application are intended to cover non-exclusive inclusion. For example, a process, method, system, product, or apparatus that includes a series of steps or units is not limited to the listed steps or units, but may optionally include steps or units not listed, or may optionally include other steps or units inherent to such process, method, product, or apparatus. The terms "first," "second," and "third," etc., are used to distinguish different objects, etc., and do not indicate a sequence, nor do they limit "first," "second," and "third" to different types.

[0068] In the description of the embodiments of this application, terms such as "exemplary," "for example," or "for instance" are used to indicate examples, illustrations, or explanations. Any embodiment or design described as "exemplary," "for example," or "for instance" in the embodiments of this application should not be construed as being more preferred or advantageous than other embodiments or designs. Specifically, the use of terms such as "exemplary," "for example," or "for instance" is intended to present the relevant concepts in a concrete manner.

[0069] In the description of the embodiments of this application, unless otherwise stated, " / " means "or". For example, A / B can mean A or B. The "and / or" in the text is merely a description of the relationship between related objects, indicating that there can be three relationships. For example, A and / or B can mean: A exists alone, A and B exist simultaneously, and B exists alone. In addition, in the description of the embodiments of this application, "multiple" means two or more.

[0070] In some processes described in the embodiments of this application, multiple operations or steps are included in a specific order. However, it should be understood that these operations or steps may not be executed in the order they appear in the embodiments of this application, or they may be executed in parallel. The sequence number of the operation is only used to distinguish different operations, and the sequence number itself does not represent any execution order. In addition, these processes may include more or fewer operations, and these operations or steps may be executed sequentially or in parallel, and these operations or steps may be combined.

[0071] Through the above description of the embodiments, those skilled in the art can clearly understand that the methods of the above embodiments can be implemented by means of software plus necessary general-purpose hardware platforms. Of course, they can also be implemented by hardware, but in many cases the former is a better implementation method. Based on this understanding, the technical solution of this application, in essence, or the part that contributes to the prior art, can be embodied in the form of a software product. This computer software product is stored in a storage medium (such as ROM / RAM, magnetic disk, optical disk) as described above, and includes several instructions to cause a terminal device to execute the methods described in the various embodiments of this application.

[0072] The above are merely specific embodiments of the present invention, but the protection scope of the present invention is not limited thereto. Any person skilled in the art can easily conceive of various equivalent modifications or substitutions within the technical scope disclosed in the present invention, and these modifications or substitutions should all be covered within the protection scope of the present invention. Therefore, the protection scope of the present invention should be determined by the scope of the claims.

Claims

1. A method for identifying and classifying the state of an electric fuel cell stack, characterized in that, The method includes the following steps: Based on the similarity between the parameters and voltage of each type of fuel cell stack, the correlation between the parameters and voltage of each type of fuel cell stack is determined; voltage-sensitive fuel cell stacks and fuel cell stack parameters with high correlation with them are selected as training data. After training the training data through a self-organizing map neural network, the optimal clustering model for clustering the fuel cell stack parameters is obtained. The training data is clustered using the optimal clustering model to obtain several clustered data classes; a pseudo-pile state label is generated for each clustered data class to form a sample set; The sample set is trained using a one-dimensional convolutional neural network to obtain a fuel cell state classification model. After classifying the fuel cell state using the fuel cell state classification model, the fuel cell state label is obtained.

2. The method for identifying and classifying the state of a fuel cell stack as described in claim 1, characterized in that, The process of determining the similarity between each type of fuel cell stack parameters and fuel cell stack voltage includes: determining the similarity between each type of fuel cell stack parameters and fuel cell stack voltage based on the covariance between each type of fuel cell stack parameters and fuel cell stack voltage.

3. The method for identifying and classifying the state of a fuel cell stack as described in claim 2, characterized in that, The formula for calculating the similarity r between the parameters of each type of fuel cell stack and the stack voltage is as follows: Where x and y represent parameters for different types of fuel cell stacks, such as x being the stack voltage and y being the stack current. Represents the mean of x. This represents the mean of y; The absolute value of r represents the correlation score between each type of stack parameter and the stack voltage.

4. The method for identifying and classifying the state of a fuel cell stack as described in claim 1, characterized in that, The training process of the optimal clustering model includes: Initialize the neuron weights of the competing layer nodes in the self-organizing map neural network, and set the initial learning rate and neighborhood radius; All training data are divided into several groups of node data according to the time when the test data is acquired, and each group of node data serves as an input layer node. Based on the similarity between each input layer node and each competing layer node, determine the winning node in the competing layer corresponding to each input layer node; Update the neighborhood radius of the winning node in the competition layer, update the learning rate based on the updated neighborhood radius, and determine the neighborhood range of the winning node in the competition layer; based on the updated learning rate, update the neuron weights of other nodes in the competition layer within the neighborhood range of the winning node in the competition layer.

5. The method for identifying and classifying the state of a fuel cell stack as described in claim 1, characterized in that, The process of training a fuel cell state classification model using a one-dimensional convolutional neural network to obtain a sample set includes: dividing the sample set into K subsets with the same number of data points and different data points; performing K training iterations using a one-dimensional convolutional neural network, selecting K-1 subsets as the training set and the remaining subsets as the test set during each training iteration.

6. The method for identifying and classifying the state of a fuel cell stack as described in claim 1, characterized in that, The fuel cell stack state classification model includes: One-dimensional convolution is used to extract features from input data. Max pooling layers are used to: extract key features; The fully connected layer is used to provide the probability of each stack state category, with the category with the highest probability being the final prediction.

7. The method for identifying and classifying the state of a fuel cell stack as described in any one of claims 1 to 6, characterized in that, The process for obtaining the battery stack parameters includes: testing the battery stack to obtain test data, denoising and normalizing the test data, and then classifying the test data into several types of battery stack parameters according to their type.

8. The method for identifying and classifying the state of a fuel cell stack as described in any one of claims 1 to 6, characterized in that, The highly correlated stack parameters are: the stack parameters corresponding to the first P correlation scores of each category, where P is a preset value.

9. A device for identifying and classifying the state of an electric stack, characterized in that, The fuel cell state identification and classification device includes a processor, a memory, and a fuel cell state identification and classification program stored in the memory and executable by the processor, wherein when the fuel cell state identification and classification program is executed by the processor, it implements the steps of the fuel cell state identification and classification method as described in any one of claims 1 to 8.

10. A computer-readable storage medium, characterized in that, The computer-readable storage medium stores a program for identifying and classifying the state of a fuel cell stack, wherein when the program for identifying and classifying the state of a fuel cell stack is executed, it implements the steps of the method for identifying and classifying the state of a fuel cell stack as described in any one of claims 1 to 8.

Images