Battery RUL prediction system based on MH-MoE model, expansion stress and other characteristics and prediction method thereof

Through the battery RUL prediction system based on the MH-MoE model and expansion stress characteristics, multiple expert sub-models and gated networks are used to process the battery data, which solves the problems of insufficient generalization ability and low efficiency of traditional models in complex data processing. , achieving more accurate and efficient battery RUL prediction, improving the effectiveness of battery health management.

CN118938026AInactive Publication Date: 2024-11-12NANTONG LE CHUANGXIN ENERGY CO LTD

Patent Information

Application Number
CN202411404709.8
Authority / Receiving Office
CN · China
Patent Type
Applications(China)
Current Assignee / Owner
Filing Date
2024-10-10
Publication Date
2024-11-12
Estimated Expiration
Not applicable · inactive patent

AI Technical Summary

Technical Problem

Traditional machine learning models have limited generalization capabilities when processing complex and diverse battery data, which are prone to overfitting problems, and are not efficient when processing large-scale data, making it difficult to accurately predict the remaining service life of the battery (RUL).

Method used

Using a battery RUL prediction system based on features such as MH-MoE model and expansion stress, the input layer, expert layer and gated network layer are used to process data using multiple expert sub-models and gated networks, and weight allocation is performed in combination with the Gumbel softmax algorithm. , dynamically select and activate expert sub-models, integrating outputs to improve prediction accuracy and efficiency.

Benefits of technology

It improves the accuracy and model expressiveness of battery RUL prediction, enhances the ability to adapt to changes in battery state, can process complex data more effectively, and provides efficient RUL prediction results, which is of great significance to battery health management.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN118938026A_ABST
    Figure CN118938026A_ABST
Patent Text Reader

Abstract

The invention discloses a battery RUL prediction system and method based on an MH-MoE model, expansion stress and other characteristics, and relates to the technical field of battery life prediction methods.The battery RUL prediction system comprises an input layer, an expert layer and a gating network layer, data input by the input layer comprises current, voltage, expansion stress and temperature of a battery, the expert layer comprises M expert sub-models, and M is a positive integer; the input data is processed from different angles, the gating network layer comprises N gating networks, and the weight of each expert sub-model is independently calculated; the input data passes through a plurality of gating networks, each gating network generates a group of weights according to the input data, and expert sub-models are selected and activated; the activated expert sub-models are used for processing the input data; and finally, summarizing the outputs of all the activated expert sub-models to form a final output battery remaining service life prediction result. And the effects of effectively processing the battery operation data and providing accurate and efficient RUL prediction are achieved.
Need to check novelty before this filing date? Find Prior Art

Description

Technical Field

[0001] The present invention relates to the technical field of battery life prediction methods, and in particular to a battery RUL prediction system and a prediction method thereof based on features such as an MH-MoE model and expansion stress. Background Art

[0002] The remaining useful life (RUL) of a battery is one of the key indicators for measuring the health of a battery. Usually, the battery RUL can be expressed by the number of remaining cycles. The importance of battery RUL is mainly reflected in the following aspects: it is used to evaluate the health of the battery, guide the use and maintenance of the battery, provide early warning of potential battery failures, evaluate the capacity attenuation of the battery, optimize the battery management strategy, and evaluate the value of battery reuse.

[0003] Predicting the remaining useful life (RUL) of a battery is of great importance in practical applications. During the operation of the battery, the following data can be measured: expansion stress (internal pressure generated by the chemical reaction during the charge and discharge process of the battery), current (the current during the charge and discharge process, which can be charging or discharging), voltage, temperature; Using current, voltage, temperature and expansion stress to predict the RUL of a battery is a comprehensive approach that takes advantage of the multiple characteristics of the battery during the charge and discharge process. This approach can help identify early signs of battery aging, thereby enabling more effective battery management, extending the battery life, and ensuring safe operation of the device.

[0004] Modern data analysis and machine learning techniques, such as deep learning, random forests, and support vector machines, can process multidimensional data and extract useful information from it to predict RUL. However, traditional machine learning models usually use a single model to process the entire data set, but this approach often seems inadequate when faced with complex and diverse data. These models have limited generalization capabilities, are prone to overfitting problems, and are inefficient when processing large-scale data. Summary of the invention

[0005] The purpose of the present invention is to provide a battery RUL prediction system and a prediction method based on the MH-MoE model and expansion stress and other characteristics, so as to improve the expressiveness and computational efficiency of the model, more effectively process battery operation data, and provide accurate and efficient RUL prediction, which is of great significance for battery health management.

[0006] The above technical objectives of the present invention are achieved through the following technical solutions: A battery RUL prediction system based on the MH-MoE model and features such as expansion stress, including an input layer, an expert layer and a gated network layer. The input data of the input layer includes multiple battery characteristic parameter data. The expert layer includes M expert sub-models, extracts corresponding features from different angles, and processes the input data. The gated network layer includes N gated networks, and independently calculates the weight of each expert sub-model. Each gating network generates a set of weights based on the input data to select and activate the expert sub-model; the activated expert sub-model processes the input data; finally, the outputs of all activated expert sub-models are aggregated to form the final output of the remaining battery life prediction result.

[0007] Furthermore, each expert sub-model has a 4-layer structure, where the bottom input layer receives input data, there are 2 middle layers with dimensions of 64 and 32 respectively, and the output layer outputs a 16-dimensional vector.

[0008] Furthermore, the gated network adopts a multi-layer perceptron model, which is a 3-layer network structure, with an input layer receiving input data and a middle layer being a 32-dimensional vector.

[0009] Furthermore, the output layer of each gating network is an M-dimensional vector, corresponding to the weights of the M expert sub-models, and the sum of the M weights is 1.

[0010] Furthermore, the battery expansion stress is measured by a micro pressure sensor mounted inside or outside the battery.

[0011] Furthermore, the output of the gating network is further weighted by the gumbel softmax algorithm.

[0012] The present application also discloses a battery RUL prediction method based on the MH-MoE model and expansion stress and other characteristics, comprising the following steps: Step S10, testing and collecting battery operation data and RUL data; Step S20, constructing an MH-MoE model; the MH-MoE model includes M expert sub-models and N gated network models, and the output of the gated network is further weighted using the gumbel softmax algorithm; Step S30, training and verifying the MH-MoE model, dividing the training data into a training set and a test set in proportion, training the model on the training set, and verifying the model on the test set; Step S40: Use the model to predict the RUL of the battery.

[0013] Furthermore, in step S20, the MH-MoE model performs weighted summation on the weight vectors of each expert sub-model and the outputs of the corresponding M expert sub-models to obtain the corresponding comprehensive feature representation, and then performs averaging calculation to obtain the average value of the comprehensive feature representation; then, the average value of the comprehensive feature representation is connected to a fully connected layer, and finally, the battery remaining service life prediction result is output through the linear regression layer.

[0014] Furthermore, the Gumbel-Softmax algorithm in step S20 includes the following steps: Initial logits: Assume there is a logits vector z containing X elements, which represent the pre-activation scores for each expert; Add Gumbel noise; add a noise g sampled from a Gumbel(0,1) distribution to each logit: z′=z+g Where g=−log(−log(u)), u is a vector sampled from the uniform distribution U(0,1); Softmax mapping; Apply the Softmax function to the logits with Gumbel noise added to obtain a probability distribution π: ; in, is the temperature parameter that controls the sharpness of the softmax distribution.

[0015] In summary, the present invention has the following beneficial effects: 1. Multi-angle feature learning: Each "head" specializes in learning different aspects of battery operation data, so that the information of battery status can be captured more comprehensively, thereby improving the accuracy of prediction; 2. Flexible gating mechanism: MH-MoE’s gating network can dynamically select the appropriate expert sub-model according to the characteristics of the input data, which enhances the model’s ability to adapt to changes in battery status; 3. Enhanced model expressiveness: By integrating the outputs of multiple expert sub-models, MH-MoE can learn more complex battery degradation patterns, thereby improving the expressiveness of RUL prediction; 4. Easy to expand: If you need to process more data features or more complex tasks, you can expand the MH-MoE model by increasing the number of "heads" without making large-scale changes to the model structure; In general, the RUL prediction method based on MH-MoE takes advantage of its structural advantages and can process battery operation data more effectively to provide accurate and efficient RUL prediction, which is of great significance for battery health management. BRIEF DESCRIPTION OF THE DRAWINGS

[0016] Figure 1 It is a schematic diagram of the model structure of a battery RUL prediction system based on the MH-MoE model and expansion stress and other characteristics of the present invention; Figure 2 It is a flow chart of a battery RUL prediction method based on the MH-MoE model and expansion stress and other characteristics of the present invention; Figure 3 It is a structure diagram of an expert sub-model in a battery RUL prediction method based on the MH-MoE model and expansion stress and other characteristics of the present invention; Figure 4 It is a structural diagram of a gated network model in a battery RUL prediction method based on the MH-MoE model and expansion stress and other characteristics of the present invention; Figure 5 This is a sample diagram of prediction results of a battery RUL prediction method based on the MH-MoE model and expansion stress and other characteristics of the present invention. DETAILED DESCRIPTION

[0017] The specific implementation of the present invention is further described below in conjunction with the accompanying drawings, and this embodiment does not constitute a limitation of the present invention. Based on the embodiments in this application, all other embodiments obtained by ordinary technicians in this field without making creative work are within the scope of protection of this application.

[0018] A battery RUL prediction system based on the MH-MoE model and features such as expansion stress, such as Figure 1 As shown, it includes input layer, expert layer and gated network layer. The data inputted by the input layer includes a plurality of battery characteristic parameter data, including Current, Voltage, Expansion and Temperature in this embodiment, which respectively represent the current, voltage, expansion stress and temperature of the battery; The expert layer includes M expert sub-models, so that the expert sub-model is composed of multiple independent sub-networks. Each expert sub-model extracts corresponding features from different perspectives and processes input data. The gated network layer includes N gated networks, which independently calculate the weight of each expert sub-model; each gated network acts as a "head" to select and activate the corresponding expert sub-model according to the input data. Multiple gated networks can process input data in parallel, and their output is an M-dimensional vector, corresponding to the weights of M expert sub-models, and the sum of M weights is 1; the multi-head mechanism allows multiple gated networks to work simultaneously, and each gated network generates its own weight distribution for selecting and activating a part of the expert sub-model.

[0019] The workflow of the above MH-MoE is as follows: the input data passes through multiple gating networks (heads), each of which generates a set of weights based on the input data to select and activate the expert sub-model; then, each gating network selects and activates several corresponding expert sub-models, and these activated expert sub-models process the input data; finally, the outputs of all activated expert sub-models are aggregated to form the final output of the battery remaining service life prediction result; This design enables MH-MoE to process complex data more effectively, improve the generalization ability of the model, and maintain high efficiency when processing large-scale data.

[0020] This embodiment takes three multi-layer perceptron (MLP) expert sub-models as examples, takes two gated networks based on multi-layer perceptrons (MLP) as examples, and also uses the gumbel softmax algorithm to further weight the output of the gated network, and it is also possible not to use this algorithm; The input data is input into three expert sub-models and two gated network models respectively; the three expert sub-models extract corresponding features from different angles, and the two gated network models independently calculate the weight of each expert sub-model; The outputs of the three expert sub-models are output_1, output_2 and output_3. These three outputs are all high-dimensional vectors (32 dimensions are used as an example in this embodiment), which respectively represent the feature representation of each sub-expert model for the input data; the two gated network models independently calculate the weight of each expert sub-model, and the output of each gated network is a 3-dimensional vector, which corresponds to the weights of the three expert sub-models, and the sum of the three weights is 1.

[0021] Gumbel-Softmax implements differentiable sampling in neural networks. Gradient descent can be used to optimize discrete selection parameters during training. It combines the Gumbel distribution and the Softmax function to make the originally non-differentiable discrete sampling process differentiable. The Gumbel distribution is a continuous distribution used for sampling indexes or sorting, while the Softmax function is used to convert real vectors into probability distributions. Therefore, in some embodiments, the 3-dimensional vectors output by the two gating network models are respectively used as inputs of the Gumbel-Softmax algorithm, and the Gumbel-Softmax algorithm re-outputs a 3-dimensional vector as the weight of the expert sub-model by adding Gumbel noise and Softmax mapping; Then, the weight vector of the first expert sub-model is weighted and summed with the outputs of the corresponding three expert sub-models to obtain weighted_expert_outputs_1, which is used as the first comprehensive feature representation; The weight vector of the second expert sub-model is weighted and summed with the outputs of the corresponding three expert sub-models to obtain weighted_expert_outputs_2, which is used as the second comprehensive feature representation; Then, the first comprehensive feature representation and the second comprehensive feature representation are averaged to obtain weighted_expert_output_average as the average value of the comprehensive feature representation; Afterwards, the comprehensive feature representation average is connected to a fully connected layer; Finally, after passing through the linear regression layer, RUL is output.

[0022] This embodiment also discloses a battery RUL prediction method based on the MH-MoE model and expansion stress and other characteristics, such as Figure 2 As shown, the following steps are included: Step S10, testing and collecting battery operation data and RUL data; collecting battery expansion stress, current, voltage, temperature and RUL (remaining useful life) data requires the use of specific sensors and measuring equipment, as well as corresponding data acquisition and processing systems; The collection steps are as follows: (1) Design sensor layout: Ensure that sensors are placed in locations that accurately reflect the battery status; (2) Install the sensor: Install the sensor on the battery or battery management system; (3) Calibrate the sensor: Calibrate the sensor to ensure that its readings are accurate and reliable; (4) Connecting the data acquisition system: Connect all sensors to the data acquisition system; (5) Data synchronization: ensuring that all sensor data are synchronized during collection to facilitate accurate correlation analysis; (6) Data recording: real-time data recording during the battery charging and discharging process; (7) Data processing: pre-processing the collected data, such as filtering and normalization, for subsequent analysis; In order to effectively collect data of the battery under normal working conditions, the specific corresponding data collection method is as follows: Battery expansion stress collection; Pressure sensor: Install miniature pressure sensors inside or outside the battery to measure expansion stress. These sensors can directly measure the internal pressure changes caused by chemical reactions during the battery's charge and discharge process; Data logger: Connect the pressure sensor's signal to a data logger to record real-time data; Battery current collection; Current sensor: Use a current sensor such as a Hall effect sensor or a shunt resistor to measure the current flowing through the battery; Data acquisition system: Connect the current sensor to a data acquisition system (such as a data acquisition card or battery management system) to record the current data; Battery voltage collection; Voltmeter: Use a high-precision voltmeter or a voltage measurement circuit built into the battery management system to measure the terminal voltage of the battery; Data acquisition system: Similar to the current sensor, voltage data can also be recorded through the data acquisition system; Battery temperature collection; Temperature sensor: You can use thermocouples, thermistors (NTC or PTC), or infrared sensors to measure the temperature of the battery; Data acquisition system: Connect the temperature sensor to the data acquisition system to monitor and record temperature changes in real time; Battery RUL data collection; Battery Management System (BMS): RUL is usually calculated by the battery management system, which combines current, voltage and other sensor data to estimate RUL.

[0023] Step S20, constructing an MH-MoE model; the MH-MoE model includes M (3) expert sub-models and N (2) gated network models. In this embodiment, each sub-model is an MLP model; Among them, the MH-MoE model performs weighted summation on the weight vectors of each expert sub-model and the outputs of the corresponding M expert sub-models to obtain the corresponding comprehensive feature representation, and then performs averaging calculation to obtain the average value of the comprehensive feature representation; then, the average value of the comprehensive feature representation is connected to a fully connected layer, and finally, the battery remaining service life prediction result is output through the linear regression layer.

[0024] like Figure 3 As shown, the three expert sub-models used in this embodiment are each an MLP model; each expert sub-model has a four-layer structure, wherein the input bottom layer receives input data, the middle layer has two layers, the dimensions are 64 and 32 respectively, and the output layer outputs a 16-dimensional vector; like Figure 4 As shown in the figure, the gated network model adopts a multi-layer perceptron model, which is a 3-layer network structure. The input layer receives input data, the middle layer is a 32-dimensional vector, the activation function is the relu function, and the output layer is a 3-dimensional vector. The activation function is softmax, which corresponds to the weights of the three expert sub-models respectively, and the sum of the three weights is 1.

[0025] In some embodiments, the output of the gating network is further weighted using a gumbel softmax algorithm; the steps of the Gumbel-Softmax algorithm are as follows: (1) Initial logits Assume that we have a logits vector z containing X elements, which represent the pre-activation scores for each expert, and in this embodiment, the weight vector of the expert sub-model output by the corresponding gating network model.

[0026] (2) Add Gumbel noise Add a noise g sampled from a Gumbel(0,1) distribution to each logit: z′=z+g Here, g=−log(−log(u)), and u is a vector sampled from the uniform distribution U(0,1).

[0027] (3) Softmax Mapping Apply the Softmax function to the logits z′ with Gumbel noise added to obtain a probability distribution π: ; in, is the temperature parameter, which controls the sharpness of the softmax distribution. When it approaches 0, the distribution becomes very sharp, close to a one-hot vector. As it approaches infinity, the distribution approaches a uniform distribution.

[0028] The 3D vectors output by the two gating network models are used as the input of the Gumbel-Softmax algorithm. The Gumbel-Softmax algorithm re-outputs two 3D vectors as the weights of the expert sub-model by adding Gumbel noise and Softmax mapping. Then, the weight vector of the first expert sub-model is weighted and summed with the outputs of the corresponding three expert sub-models to obtain the first comprehensive feature representation; The weight vector of the second expert sub-model is weighted and summed with the outputs of the corresponding three expert sub-models to obtain the second comprehensive feature representation; Then, the first comprehensive feature representation and the second comprehensive feature representation are averaged to obtain the average value of the comprehensive feature representation; Afterwards, the comprehensive feature representation average is connected to a fully connected layer; Finally, after passing through the linear regression layer, RUL is output.

[0029] Step S30, training and verifying the MH-MoE model. In this embodiment, the MH-MoE model code is implemented and trained on the Keras platform. During training, the number of epochs is 1000, and the number of batch samples is batch_size=1024. The optimizer for model training is the adam algorithm, and the loss function is mean square error (mse); The training data is divided into a training set and a test set in a ratio of 8:2. The model is trained on the training set and verified on the test set. Step S40, using the model to predict the RUL of the battery; after the MH-MoE model is trained, if the model effect is good, the model can be deployed and the model can be used to predict the RUL of the battery; The loss function of the model is Mean Square Error (MSE); the optimizer is Adam; the evaluation metric is Mean Absolute Error (MAE), which is the sum of the absolute values ​​of the differences between the target value and the predicted value.

[0030] Forecast data display: This embodiment tests the effect on a public data set (https: / / deepblue.lib.umich.edu / data / concern / file_sets / gq67jr501), and uses the current, voltage, expansion stress and temperature in the cycling_wExpansion.csv file in the folder numbered 01 to predict the battery RUL. The following table shows the battery expansion force stress and other characteristics and RUL data set samples (part); Current [mA] Voltage [V] Expansion [mu m] Temperature [C] Cycle number 0 3.3584943 NaN 22.832777 0 0 3.3584549 32.4 22.896147 0 0 3.3584156 32.4 22.991203 0 0 3.3584943 32.46998001 22.864462 0 1000.2375 3.3844109 32.56998001 22.785248 0 1000.2966 3.3900037 32.6 23.078339 0 1000.3163 3.394927 32.66998002 23.323898 0 1000.1784 3.3995352 32.76998003 23.276371 0 1000.139 3.4038284 32.86998003 23.094181 0 …… …… …… …… …… -999.77649 3.068372 231.565911 25.502266 329 -999.91443 3.052381 231.465911 25.280468 329 -999.7962 3.0350509 231.365911 25.383444 329 -999.83557 3.015594 231.3 25.327997 329 -999.7962 2.9999969 231.295111 25.494345 329 -999.83557 3.0001545 231.295091 25.494345 329 999.68585 3.0220928 231.295051 25.494345 330 1000.2572 3.0801487 231.2049491 25.478502 330 Among them, Current represents current, Voltage represents voltage, Expansion represents expansion stress, Temperature represents temperature, and Cycle number represents the remaining number of battery cycles, i.e. RUL.

[0031] RUL prediction results display: The mean of the RUL test set is 181.2381, and the standard deviation is 109.5206. The mean of the prediction set is 181.0707, and the standard deviation is 109.636, which shows that the data distribution of the prediction results is basically consistent with the data distribution of the real data. The root mean square error between the prediction set and the test set of the battery RUL is only 3.3125, indicating that the prediction effect of the model is good.

[0032] The prediction results are shown in the following table (Sample of battery RUL prediction results based on MH-MoE model): Battery RUL real value Battery RUL prediction value Battery RUL error 58 54.27 3.73 11 11.76 -0.76 236 233.55 2.45 102 101.56 0.44 309 309.78 -0.78 like Figure 5 As shown in the figure, the solid line above represents the true value, the dotted line represents the predicted value, and the solid square represents the error value. The numbers represent the true values. It can be seen from the figure that the prediction effect is good.

[0033] The above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. Those skilled in the art may make various modifications or equivalent substitutions to the present invention within the essence and protection scope of the present invention, and such modifications or equivalent substitutions should also be deemed to fall within the protection scope of the technical solution of the present invention.

Claims

1. A battery RUL prediction system based on MH-MoE model and expansion stress and other characteristics, characterized by: It includes an input layer, an expert layer and a gated network layer. The input layer inputs multiple battery characteristic parameter data. The expert layer includes M expert sub-models, extracts corresponding features from different angles, and processes input data. The gated network layer includes N gated networks, which independently calculate the weight of each expert sub-model. Each gating network generates a set of weights based on the input data to select and activate the expert sub-model; the activated expert sub-model processes the input data; finally, the outputs of all activated expert sub-models are aggregated to form the final output of the remaining battery life prediction result.

2. A battery RUL prediction system based on the MH-MoE model and expansion stress and other characteristics according to claim 1, characterized in that: Each expert sub-model has a 4-layer structure. Its input bottom layer receives input data, there are 2 middle layers with dimensions of 64 and 32 respectively, and the output layer outputs a 16-dimensional vector.

3. A battery RUL prediction system based on the MH-MoE model and expansion stress and other characteristics according to claim 1 or 2, characterized in that: The gated network uses a multi-layer perceptron model, which is a 3-layer network structure. The input layer receives input data and the middle layer is a 32-dimensional vector.

4. A battery RUL prediction system based on the MH-MoE model and expansion stress and other characteristics according to claim 3, characterized in that: The output layer of each gating network is an M-dimensional vector, corresponding to the weights of the M expert sub-models, and the sum of the M weights is 1.

5. A battery RUL prediction system based on the MH-MoE model and expansion stress and other characteristics according to claim 1, characterized in that: Battery expansion stress is measured by a miniature pressure sensor mounted inside or outside the battery.

6. A battery RUL prediction system based on the MH-MoE model and expansion stress and other characteristics according to claim 1, characterized in that: The output of the gating network is further weighted by the gumbel softmax algorithm.

7. A prediction method for a battery RUL prediction system based on the MH-MoE model and expansion stress and other characteristics as claimed in claim 1, characterized in that: The following steps are included: Step S10, testing and collecting battery operation data and RUL data; Step S20, constructing an MH-MoE model; the MH-MoE model includes M expert sub-models and N gated network models, and the output of the gated network is further weighted using the gumbel softmax algorithm; Step S30, training and verifying the MH-MoE model, dividing the training data into a training set and a test set in proportion, training the model on the training set, and verifying the model on the test set; Step S40: Use the model to predict the RUL of the battery.

8. A battery RUL prediction method based on the MH-MoE model and expansion stress and other characteristics according to claim 7, characterized in that: In step S20, the MH-MoE model performs weighted summation on the weight vectors of each expert sub-model and the outputs of the corresponding M expert sub-models to obtain the corresponding comprehensive feature representation, and then performs averaging calculation to obtain the average value of the comprehensive feature representation; then, the average value of the comprehensive feature representation is connected to a fully connected layer, and finally, the battery remaining service life prediction result is output through the linear regression layer.

9. The method for predicting battery RUL based on the MH-MoE model and expansion stress and other characteristics according to claim 7, characterized in that: The Gumbel-Softmax algorithm in step S20 includes the following steps: Initial logits: Assume there is a logits vector z containing X elements, which represent the pre-activation scores for each expert; Add Gumbel noise; add a noise g sampled from a Gumbel(0,1) distribution to each logit: z′=z+g Where g=−log(−log(u)), u is a vector sampled from the uniform distribution U(0,1); Softmax mapping; Apply the Softmax function to the logits with Gumbel noise added to obtain a probability distribution π: ; in, is the temperature parameter that controls the sharpness of the softmax distribution.

Citation Information

Patent Citations

  • Method for predicting residual capacity of battery based on ELM (Extreme Learning Model) and battery impedance spectrum data

    CN115902664A

  • Precise model and lightweight model combined depression identification method and system

    CN118098289A

  • Method for estimating SOC (State of Charge) of lithium battery, battery management system, medium and electric energy equipment

    CN118191637A

  • SOH prediction method based on MOE model and current, voltage and temperature of battery

    CN118584357A

  • Method for predicting RUL based on MOE model and current resistance impedance of battery

    CN118625152A

Cited By

  • Method and system for indirectly predicting state of health of battery based on expansion stress

    CN119936686A

  • MOE-based battery disassembly flexible scheduling method and system

    CN121119654A