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

Abstract

The embodiment of the invention relates to the technical field of battery performance prediction, and discloses a method for indirectly predicting the state of health of a battery based on expansion stress, and the method comprises the steps: collecting the time sequence values of the expansion stress, current, voltage, working temperature, SOC and RUL in a plurality of charging and discharging processes of the battery, and indirectly calculating the SOH data of each time sequence through the SOC and the RUL; establishing a data set containing the expansion stress, the current, the voltage, the working temperature and the SOH under different time sequences; constructing a prediction model based on a sparse hybrid expert model; training and verifying the prediction model by using the data set; and predicting the SOH of the battery by using the trained prediction model, and outputting a prediction result. According to the scheme, the technical problems that a traditional prediction method is single in dimension, lack of dynamics and real-time performance, limited in prediction precision, low in efficiency and the like are at least solved.

Description

Technical Field

[0001] The present application relates to the technical field of battery performance prediction, and in particular to a method and system for indirectly predicting the health status of a battery based on expansion stress. Background Art

[0002] The state of health (SOH) of a battery is a measure of the battery's performance relative to its new state. SOH describes the percentage of the battery's maximum capacity relative to its original capacity after a certain period of use and aging. The battery's SOH is a key parameter for the battery management system (BMS) because it is directly related to the reliability and safety of the battery. As battery technology develops, accurately monitoring and predicting SOH becomes increasingly important.

[0003] However, traditional battery health status prediction methods mainly rely on the monitoring of battery current, voltage and temperature (U / I / T for short). Although these methods can reflect the basic status of the battery to a certain extent, they have obvious limitations and cannot comprehensively and accurately evaluate the health status of the battery. Traditional methods have the following shortcomings: Single dimension: It mainly relies on the data of three dimensions: U / I / T. However, these data do not vary significantly at different stages of the battery life cycle, and only fluctuate greatly in emergencies. This monitoring method cannot capture subtle changes in the internal structure and electrochemical performance of the battery, and it is difficult to achieve a comprehensive and real-time assessment of the battery's health status.

[0004] Lack of dynamics and real-time performance: Battery aging is a dynamic and complex process involving multiple physical and chemical changes. Traditional methods can usually only provide static or phased evaluation results, making it difficult to track changes in battery health status in real time.

[0005] Limited prediction accuracy: Due to the complexity of the internal structure and chemical reactions of the battery, traditional methods often find it difficult to establish an accurate degradation mechanism model to predict the remaining life and health status of the battery. Therefore, the prediction results often have large uncertainties.

[0006] Low efficiency: Traditional machine learning models often use a single model to fit the entire data set, which is inadequate when faced with complex and highly heterogeneous data. Such models have limited generalization capabilities, are prone to overfitting problems, and are inefficient when processing large-scale data.

[0007] In view of the above shortcomings, a more comprehensive and accurate battery health status prediction method is urgently needed. Summary of the invention

[0008] One purpose of the present application is to provide a method and system for indirectly predicting the health status of a battery based on expansion stress, at least to solve the technical problems of traditional prediction methods such as single dimension, lack of dynamics and real-time performance, limited prediction accuracy, and low efficiency.

[0009] To achieve the above objectives, some embodiments of the present application provide the following aspects: In a first aspect, an embodiment of the present application provides a method for indirectly predicting a battery health state based on expansion stress, comprising: Collect the time series values ​​of expansion stress, current, voltage, operating temperature, state of charge SOC and remaining service life RUL during several charge and discharge processes of the battery, and indirectly calculate the battery health status SOH data of each time series through the state of charge SOC and remaining service life RUL data; Establishing a data set including the expansion stress, current, voltage, operating temperature and battery health state SOH in different time series; Constructing a prediction model based on a sparse hybrid expert model, wherein the prediction model includes a plurality of expert sub-models and a gating network model; Each of the expert sub-models has a 5-layer structure, the input layer receives input data, the middle layer has 3 layers, the dimensions are 128, 64 and 32 respectively, and the output layer outputs a 16-dimensional vector; The gated network model is a 3-layer network structure, the input layer receives input data, the middle layer is a 32-dimensional vector, and the output layer 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; wherein each expert sub-model is a multi-layer perceptron, the gated network model is a multi-layer perceptron, and sparse activation is implemented using the Gumbel-Softmax algorithm; Using the data set to train and verify the prediction model; The trained prediction model is used to predict the battery's state of health (SOH) and output the prediction result.

[0010] Furthermore, the indirect calculation of the battery health state SOH data of each time series through the state of charge SOC and the remaining service life RUL data includes: The batteries are grouped by remaining useful life RUL, and the maximum value of the state of charge SOC is calculated in each group. This maximum value is used as the SOH value of each data record in the corresponding remaining useful life RUL cycle period. The entire battery health state SOH sequence is then removed and filled with abnormal values ​​to obtain normal battery health state SOH sequence data.

[0011] Furthermore, the number of the expert sub-models is N, and the outputs of the N expert sub-models are output_1, output_2, ..., output_N, and the N outputs are all high-dimensional vectors; The output of the gating network model is an N-dimensional vector, which corresponds to the weights of the N expert sub-models, and the sum of the N weights is 1.

[0012] Furthermore, the N-dimensional vector output by the gated network model is used as the input of the Gumbel-Softmax algorithm, and the Gumbel-Softmax algorithm re-outputs an N-dimensional vector as the weight of the expert sub-model by adding Gumbel noise and Softmax mapping, thereby realizing sparse activation of the expert sub-model; Performing a weighted summation of the expert sub-model weight and the output of the corresponding expert sub-model to obtain a comprehensive feature representation; Connecting the comprehensive features to a fully connected layer; Through the linear regression layer, the battery's state of health SOH is output.

[0013] Furthermore, wherein N is 3.

[0014] Further, the steps of the Gumbel-Softmax algorithm are as follows: (1) Initial logits Assume a logits vector z containing M elements, which represents the pre-activation score for each expert, corresponding to the weight vector of the expert sub-model output by the gating network model; (2) Add Gumbel noise To each logit, add a noise g sampled from a Gumbel(0, 1) distribution: z′=z+g Among them, z′ is the logits vector with Gumbel noise added, , u is a vector sampled from the uniform distribution U(0,1); (3) Softmax Mapping Apply the Softmax function to the logits z′ with Gumbel noise added to obtain a probability distribution π:

[0015] in, represents the i-th element in the probability distribution π, that is, the probability of the i-th category; Represents the i-th element of the logits vector after adding Gumbel noise; is the temperature parameter, which controls the sharpness of the softmax distribution; when 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.

[0016] In a second aspect, an embodiment of the present application further provides a system for indirectly predicting a battery health state based on expansion stress, comprising: The acquisition unit is used to collect the time series values ​​of expansion stress, current, voltage, operating temperature, state of charge SOC and remaining service life RUL during several charge and discharge processes of the battery, and indirectly calculate the battery health status SOH data of each time series through the state of charge SOC and remaining service life RUL data; An establishing unit, used for establishing a data set including the expansion stress, current, voltage, operating temperature and battery health state SOH in different time series; A construction unit, used to construct a prediction model based on a sparse hybrid expert model, wherein the prediction model includes a plurality of expert sub-models and a gating network model; A training unit, used to train and verify the prediction model using the data set; The prediction unit is used to predict the health state SOH of the battery by using the trained prediction model and output the prediction result.

[0017] Compared with the prior art, this application achieves the following beneficial effects: The present invention supplements the traditional current, voltage and temperature characteristics based on the expansion stress characteristics.

[0018] The present invention proposes a battery health status prediction method based on multi-dimensional monitoring. By introducing a new health factor, expansion stress, a four-dimensional health assessment system is constructed together with voltage, current, and temperature. By real-time monitoring of the expansion stress changes during battery operation and combining the battery management system to realize signal processing and analysis, a comprehensive and real-time assessment of the battery health status is achieved, thereby improving the accuracy of SOH prediction. This innovative method not only makes up for the shortcomings of the traditional three-dimensional evaluation system, but also improves the accuracy and long-term stability of battery health status prediction, providing a more reliable basis for battery maintenance and replacement. When indirectly predicting battery SOH through a sparse mixed expert model, the model's expressive power and prediction accuracy can be improved, the robustness and generalization ability of the overall model are improved, the risk of overfitting is reduced, the computational cost is greatly reduced, and the model's operational efficiency is improved. BRIEF DESCRIPTION OF THE DRAWINGS

[0019] One or more embodiments are exemplarily described by pictures in the corresponding drawings, and these exemplified descriptions do not constitute limitations on the embodiments. Elements with the same reference numerals in the drawings represent similar elements, and unless otherwise stated, the figures in the drawings do not constitute proportional limitations.

[0020] Figure 1 A flow chart of a method for indirectly predicting battery health status based on expansion stress provided in the first embodiment of the present application; Figure 2 A structural diagram of a prediction model provided for the first embodiment of the present application; Figure 3 This is a structural diagram of an expert sub-model in the prediction model provided in the first embodiment of the present application; Figure 4 A structural diagram of a gated network model in a prediction model provided in the first embodiment of the present application; Figure 5 A trend chart of the preliminary calculation results of SOH in the first stage of the method provided in the first embodiment of the present application; Figure 6 A trend diagram of the SOH correction calculation results in the first stage of the method provided in the first embodiment of the present application; Figure 7 This is a sample diagram of the indirect prediction results of battery SOH in the second stage of the method provided in the first embodiment of the present application; Figure 8 A system unit module diagram for indirectly predicting the battery health status based on expansion stress provided in the second embodiment of the present application; Fig. 9 A schematic diagram of an exemplary structure of an electronic device provided for the third embodiment of the present application. DETAILED DESCRIPTION

[0021] In order to make the purpose, technical solution and advantages of the embodiments of the present application clearer, the technical solution in the embodiments of the present application will be clearly and completely described below in conjunction with the drawings in the embodiments of the present application. Obviously, the described embodiments are part of the embodiments of the present application, not all of the embodiments. Based on the embodiments in the present application, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of this application.

[0022] In some scenarios, BMS only collects data such as battery state of charge (SOC) and remaining useful life (RUL), but no direct SOH data. In this case, it is necessary to indirectly calculate SOH through certain methods and build a prediction model to predict SOH. The present invention calls it the SOH indirect prediction method.

[0023] During battery operation, the following data can be measured: 1. Expansion stress: The internal pressure generated by the chemical reaction during the charge and discharge process of the battery.

[0024] 2. Current: The current during the battery charging and discharging process, which can be charging or discharging.

[0025] 3. Voltage: The voltage across the battery.

[0026] 4. Temperature: The operating temperature of the battery.

[0027] Expansion stress, current, voltage and temperature are key factors affecting battery performance, which are closely related to the internal chemical reactions and physical state of the battery.

[0028] Expansion stress, current, voltage and temperature can be used to predict battery SOH, which is mainly reflected in the following points: 1. Expansion stress: The volume of the battery will expand and shrink during the charge and discharge process. The accumulation of this mechanical stress will cause damage to the battery structure, such as electrode peeling, diaphragm rupture, etc. Excessive expansion stress may cause micro short circuits or capacity loss inside the battery, thereby reducing SOH. By monitoring the degree of battery expansion, the accumulation of internal stress in the battery can be indirectly evaluated and the health status of the battery can be predicted.

[0029] 2. Current: The magnitude and duration of the current will affect the cycle life of the battery. High current discharge will cause more heat and stress inside the battery. Long-term high current discharge will accelerate battery aging, reduce its effective capacity, and thus reduce SOH. By analyzing the waveform and statistical characteristics of the current, the charge and discharge behavior of the battery can be evaluated and the health of the battery can be predicted.

[0030] 3. Voltage: The voltage change of the battery reflects the state of the chemical reaction inside the battery. As the battery ages, its full charge and full discharge voltage will decrease. The voltage drop may mean an increase in the internal impedance of the battery, which is a sign of battery aging. By monitoring the voltage curve and voltage stability of the battery, the battery's SOH can be inferred.

[0031] 4. Temperature: The temperature of a battery has a significant impact on its performance and life. Temperatures that are too high or too low will accelerate battery aging. Temperature changes affect the rate of chemical reactions inside the battery, which in turn affects the battery's capacity and power output. By monitoring the temperature changes of the battery, the battery's thermal management status can be evaluated and the battery's health status can be predicted.

[0032] These parameters can be used to predict SOH because they are closely related to the chemical and physical processes inside the battery, which determine the performance and life of the battery. By comprehensively analyzing these parameters, predictive models of battery health status can be established. These models are usually based on machine learning algorithms and can be trained with historical data to predict the future performance of the battery. Predicting SOH helps to identify potential battery degradation problems in advance, so that appropriate maintenance measures can be taken to extend the service life of the battery.

[0033] Sparse Mixture of Experts (SMoE) is a machine learning architecture designed for complex prediction and classification problems. The core of the model is to use multiple "expert" sub-models, each of which specializes in processing a specific part of the data set or a specific type of input. The prediction results of these sub-models are weighted and aggregated through a gating network to obtain the final prediction. The "sparse" nature of SMoE is reflected in that not all expert sub-models are activated at each prediction, but only a part of them are selected to participate in the calculation. This selective activation mechanism greatly reduces the computational cost and improves the operational efficiency of the model. The implementation of the sparse strategy can be achieved through some algorithms such as the Gumbel algorithm, that is, a few experts are selected to be activated among multiple experts, while most of the others remain inactive.

[0034] The present invention proposes a method and system for indirectly predicting the battery state of health SOH based on expansion stress and sparse hybrid expert model. By introducing the new health factor of expansion stress, a four-dimensional health assessment system is constructed together with voltage, current and temperature. This method realizes a comprehensive and real-time assessment of the battery health state by real-time monitoring of the changes in expansion stress during battery operation and combining the battery management system to realize signal processing and analysis. This innovative method not only makes up for the shortcomings of the traditional three-dimensional assessment system, but also improves the accuracy and long-term stability of battery health state prediction, providing a more reliable basis for battery maintenance and replacement.

[0035] First embodiment Figure 1 A flow chart of a method for indirectly predicting battery health status based on expansion stress provided in the first embodiment of the present application, such as Figure 1 As shown, a method 100 for indirectly predicting a battery health state based on expansion stress comprises: S110: collecting time series values ​​of expansion stress, current, voltage, operating temperature, state of charge SOC and remaining service life RUL during several charge and discharge processes of the battery, and indirectly calculating battery health status SOH data of each time series through the state of charge SOC and remaining service life RUL data; Among them, expansion stress is a novel and unique indicator of the internal state of the battery, which reflects the internal pressure changes caused by chemical reactions during the charge and discharge process of the battery. Utilizing this feature can provide a new perspective on the battery's state of health, which helps to predict SOH more accurately. The expansion stress feature can complement the traditional current, voltage and temperature features to provide more comprehensive battery health status information, thereby improving the accuracy of SOH prediction.

[0036] In step S110, specific sensors and measuring equipment are required to collect the expansion stress, current, voltage, temperature, SOC and RUL data of the battery. The following are some basic steps and methods: 1. Expansion stress collection Pressure sensor: During the battery design process, micro pressure sensors can be installed 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.

[0037] Data logger: Connect the signal from the pressure sensor to a data logger to record real-time data.

[0038] 2. 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.

[0039] 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.

[0040] 3. Voltage collection Voltmeter: Use a high-precision voltmeter or the voltage measurement circuit built into the battery management system to measure the terminal voltage of the battery.

[0041] Data acquisition system: Similar to the current sensor, the voltage data can also be recorded by the data acquisition system.

[0042] 4. Temperature collection Temperature sensor: You can use a thermocouple, thermistor (NTC or PTC), or infrared sensor to measure the temperature of the battery.

[0043] Data Acquisition System: Connect the temperature sensor to the data acquisition system to monitor and record temperature changes in real time.

[0044] 5. SOC and RUL data collection Battery Management System (BMS): SOC and RUL are usually calculated by the battery management system, which combines current, voltage and other sensor data to estimate SOC and RUL.

[0045] The specific collection steps are as follows: (1) Design sensor layout: Ensure that the sensors are placed in a location that accurately reflects the battery status.

[0046] (2) Install the sensor: Install the sensor on the battery or battery management system.

[0047] (3) Calibrate the sensor: Calibrate the sensor to ensure its readings are accurate and reliable.

[0048] (4) Connect the data acquisition system: Connect all sensors to the data acquisition system.

[0049] (5) Data synchronization: Ensure that all sensor data are synchronized during collection to facilitate accurate correlation analysis.

[0050] (6) Data recording: Record data in real time during the battery charging and discharging process.

[0051] (7) Data processing: Preprocess the collected data, such as filtering and normalization, for subsequent analysis.

[0052] Through the above steps, relevant data of the battery under normal working conditions can be effectively collected.

[0053] The model of the present invention is divided into two stages, wherein step S110 relates to the first stage (stage 1) of the model, namely indirectly calculating the battery state of health SOH through the state of charge SOC and the remaining service life data RUL.

[0054] Specifically, the indirect calculation method of SOH is: in a complete charge and discharge cycle, the data of SOC and RUL are also complete. In a cycle, the maximum value of SOC is the SOH value corresponding to the cycle. Therefore, the calculation process is grouped by RUL, and the maximum value of SOC is calculated in each group. The maximum value is used as the SOH value of each data record in the corresponding RUL cycle, and the entire SOH sequence is subjected to outlier removal and data filling to obtain normal SOH sequence data.

[0055] The steps of SOH outlier removal and data filling in the present invention are as follows: (1) Identify outliers: Use a sliding window to identify points that are significantly different from surrounding values. The window size of the sliding window is 10,000. The sliding mean and standard deviation are calculated within each window. Outliers are those values ​​that differ from the sliding mean by more than 2 times the sliding standard deviation. (2) Eliminate outliers: Set the identified outliers to NaN; (3) Interpolation supplement: Use linear interpolation to fill in the data of outliers.

[0056] S120: establishing a data set including the expansion stress, current, voltage, operating temperature and battery health state SOH in different time series; In step S120, according to the data obtained in step S110, a data set containing the expansion stress, current, voltage, operating temperature and battery health state SOH in different time series is established. The data set is used as training data and is divided into a training set and a test set in a ratio of 8:2.

[0057] S130: Construct a prediction model based on a sparse mixture of experts (SMoE) model, wherein the prediction model includes multiple expert sub-models and a gating network model; Figure 2 This is a diagram of the prediction model structure provided in the embodiment of the present application. Figure 2 As shown in the figure, the sparse mixture of experts model is a machine learning architecture designed for complex prediction and classification problems. SMoE allows different experts to learn specific aspects of the data. For example, some experts may focus on current changes, while others may focus on expansion stress patterns. This multi-expert model can improve the accuracy of predictions. Through the sparse selection mechanism, only part of the expert network is activated, which significantly reduces the amount of computation and improves the training and prediction efficiency of the model.

[0058] The SMoE model includes a multi-expert model and a gating network model.

[0059] Each expert sub-model is a multi-layer perceptron (MLP) model, and the expert sub-models are all multi-layer structures. The number of expert sub-models is N, and the outputs of the N expert sub-models are output 1 (output_1), output 2 (output_2), ..., output N (output_N), and these N outputs are all high-dimensional vectors.

[0060] The gated network model adopts a multi-layer perceptron (MLP) model. The output of the gated network model is an N-dimensional vector, which corresponds to the weights of N expert sub-models, and the sum of the N weights is 1.

[0061] The output of the gated network is further weighted using the gumbel softmax algorithm to achieve sparse activation of the expert sub-model. That is, sparse activation is achieved using the gumbel-softmax algorithm.

[0062] Among them, the Gumbel-Softmax algorithm is a technique for implementing differentiable sampling in neural networks, which allows us to use gradient descent to optimize discretely selected parameters during training. The Gumbel-Softmax algorithm 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.

[0063] Specifically, the steps of the Gumbel-Softmax algorithm are as follows: (1) Initial logits Assume a logits vector z containing M elements, which represents the pre-activation score for each expert, corresponding to the weight vector of the expert sub-model output by the gating network model; (2) Add Gumbel noise To each logit, add a noise g sampled from a Gumbel(0, 1) distribution: z′=z+g Among them, z′ is the logits vector with Gumbel noise added, , u is a vector sampled from the uniform distribution U(0,1); (3) Softmax Mapping Apply the Softmax function to the logits z′ with Gumbel noise added to obtain a probability distribution π:

[0064] in, represents the i-th element in the probability distribution π, that is, the probability of the i-th category; Represents the i-th element of the logits vector after adding Gumbel noise; is the temperature parameter, which controls the sharpness of the softmax distribution; when 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.

[0065] Furthermore, the weighted summation process includes: The N-dimensional vector output by the gated network model is used as the input of the Gumbel-Softmax algorithm. The Gumbel-Softmax algorithm re-outputs an N-dimensional vector as the weight of the expert sub-model by adding Gumbel noise and Softmax mapping, thereby realizing sparse activation of the expert sub-model. Performing a weighted summation of the expert sub-model weight and the output of the corresponding expert sub-model to obtain a comprehensive feature representation; Connecting the comprehensive features to a fully connected layer; Through the linear regression layer, the battery's state of health SOH is output.

[0066] Preferably, in some embodiments, N is 3, such as Figure 3 As shown in , the expert sub-models are all 5-layer structures. The input layer receives input data, the middle layer has 3 layers, and the dimensions are 128, 64, and 32 respectively. The output layer outputs a 16-dimensional vector. Figure 4 As shown in the figure, the gated network model 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, and the sum of the three weights is 1.

[0067] The input data are fed into three expert sub-models respectively to extract corresponding features from different angles.

[0068] The outputs of the three expert sub-models are output 1 (output1), output 2 (output2) and output 3 (output3), respectively. These three outputs are all high-dimensional vectors. The present invention takes 32 dimensions as an example, which respectively represent the feature representation of each sub-expert model for the input data; the output of the 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.

[0069] The output 3D vector of the gated network model is used as the input of the Gumbel-Softmax algorithm. The Gumbel-Softmax algorithm re-outputs a 3D vector as the weight of the expert sub-model by adding Gumbel noise and Softmax mapping. Then, the expert sub-model weight is weighted and summed with the output of the corresponding expert sub-model to obtain a comprehensive feature representation. After that, the comprehensive feature is connected to a fully connected layer. Finally, the SOH is output through a linear regression layer.

[0070] S140: Using the data set to train and verify the prediction model; In step S140, the input data of the model are Current, Voltage, Expansion and Temperature, which respectively represent the current, voltage, expansion stress and temperature of the battery; the output data of the model is SOH, which represents the health state of the battery.

[0071] 1. Prediction data display The present invention tests the effect on a public data set, which can be obtained through the website https: / / deepblue.lib.umich.edu / data / concern / file_sets / gq67jr501. The current, voltage, expansion stress and temperature in the cycling_wExpansion.csv file in the folder numbered 01 are used to indirectly predict the battery SOH. Table 1 is a sample (part) of the battery expansion stress and other characteristics and SOC and RUL data sets. Among them, Current represents current, Voltage represents voltage, Expansion represents expansion stress, Temperature represents temperature, Capacity represents battery capacity, i.e. SOC, and Cycle number represents the remaining number of battery cycles, i.e. RUL.

[0072] Table 1 Battery expansion stress and other characteristics and SOC and RUL data set examples (partial)

[0073] 2.SOH calculation results display The data is grouped by cycle number, and the maximum value of the SOC sequence in each group is calculated as the SOH value corresponding to each data in the group. The SOH sequence calculation results are as follows: Figure 5 As shown. Figure 5 From the SOH trend graph in , we can see that the overall trend of SOH is steadily decreasing, which is in line with the law of SOH changes. However, there are a few outliers, namely, two places where the figure suddenly drops, which require outlier processing and data filling. After SOH outlier removal and data filling, the calculation result of SOH is as follows: Figure 6 shown.

[0074] A specific sample (part) of SOH calculation results is shown in Table 2. Table 2 contains the current, voltage, expansion stress and temperature data in Table 1 and the SOH calculation results.

[0075] Table 2 Battery expansion stress and other characteristics and SOH data set examples (partial)

[0076] The present invention implements the prediction model code and trains it on the Keras platform, with the number of training epochs=1000 and the number of batch samples batch_size=1024. The optimizer for model training is the adam algorithm, and the loss function is mean square error (mse).

[0077] The training set and test set in Table 2 are used to train and verify the model respectively.

[0078] S150: Use the trained prediction model to predict the battery's state of health SOH and output the prediction result.

[0079] After the prediction model training is completed in step S140, the model can be deployed and used to predict the SOH of the battery if the model effect is good.

[0080] In step S150, the prediction model is fed with the following data: Current, Voltage, Expansion, and Temperature, which represent the battery current, voltage, expansion stress, and temperature, respectively. The prediction target of the prediction model is SOH, i.e., the battery health status. The loss function of the model is the mean square error (MSE); the optimizer is adam; the evaluation metric is the mean absolute error (MAE), i.e., the sum of the absolute values ​​of the difference between the target value and the predicted value.

[0081] 3.SOH prediction results display The mean of the SOH test set is 4.2739, the standard deviation is 0.38516, and the mean of the prediction set is 4.2759, the standard deviation is 0.38521, which shows that the data distribution of the prediction results is basically consistent with the data distribution of the real data.

[0082] The root mean square error between the predicted set and the test set of battery SOH is only 0.01236, indicating that the prediction effect of the model is good. The prediction results are shown in Table 3 below.

[0083] Table 3. Example of indirect prediction results of battery SOH

[0084] The prediction results are as follows: Figure 7 shown. Figure 7 The solid line in the upper middle 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. Figure 7 It can be seen that the prediction effect is good.

[0085] According to the above embodiment of the present invention, the method introduces a new health factor, expansion stress, to construct a four-dimensional health assessment system together with voltage, current and temperature. By real-time monitoring of the changes in expansion stress during battery operation and combining the battery management system to realize signal processing and analysis, a comprehensive and real-time assessment of the battery health status can be achieved, which not only makes up for the shortcomings of the traditional three-dimensional assessment system, but also improves the accuracy and long-term stability of battery health status prediction, and provides a more reliable basis for battery maintenance and replacement. The SOH is indirectly calculated through data such as SOC and RUL, and the SOH is predicted, making the monitoring indicators of the BMS system more complete. When the SMoE model indirectly predicts the battery SOH, it can improve the expression ability and prediction accuracy of the model, improve the robustness and generalization ability of the overall model, and reduce the risk of overfitting.

[0086] The steps of the above method are divided only for clear description. When implemented, they can be combined into one step or some steps can be split and decomposed into multiple steps. As long as they include the same logical relationship, they are all within the scope of protection of this application; adding insignificant modifications to the algorithm or process or introducing insignificant designs without changing the core design of the algorithm and process are all within the scope of protection of this application.

[0087] Second embodiment Figure 8 A system unit module diagram for indirectly predicting the battery health status based on expansion stress provided in the second embodiment of the present application, such as Figure 8 As shown, a system 200 for indirectly predicting battery health status based on expansion stress includes: The acquisition unit 210 is used to acquire the time series values ​​of expansion stress, current, voltage, operating temperature, state of charge SOC and remaining service life RUL during several charge and discharge processes of the battery, and indirectly calculate the battery health state SOH data of each time series through the state of charge SOC and remaining service life data RUL; An establishing unit 220, used to establish a data set including the expansion stress, current, voltage, operating temperature and battery health state SOH in different time series; A construction unit 230, configured to construct a prediction model based on a sparse hybrid expert model, wherein the prediction model includes a plurality of expert sub-models and a gating network model; A training unit 240, configured to train and verify the prediction model using the data set; The prediction unit 250 is used to predict the battery's state of health SOH using the trained prediction model and output a prediction result.

[0088] It is not difficult to find that this embodiment is a system embodiment corresponding to the first embodiment, and this embodiment can be implemented in conjunction with the first embodiment. The relevant technical details mentioned in the first embodiment are still valid in this embodiment, and in order to reduce repetition, they are not repeated here. Accordingly, the relevant technical details mentioned in this embodiment can also be applied in the first embodiment.

[0089] It is worth mentioning that all modules involved in this embodiment are logic modules. In practical applications, a logic unit can be a physical unit, a part of a physical unit, or a combination of multiple physical units. In addition, in order to highlight the innovative part of this application, this embodiment does not introduce units that are not closely related to solving the technical problems proposed by this application, but this does not mean that there are no other units in this embodiment.

[0090] Third embodiment In addition, some embodiments of the present application also provide an electronic device. The electronic device may be a digital computer in various forms, such as a laptop computer, a desktop computer, a workbench, a personal digital assistant, a server, a blade server, a mainframe computer, etc. The electronic device may also be a mobile device in various forms, such as a personal digital processing, a cellular phone, a smart phone, a wearable device, and other similar computing devices.

[0091] The electronic device includes: one or more processors; and a memory storing computer program instructions, wherein when the computer program instructions are executed, the processor executes the steps of the method provided in any one or more of the above embodiments. Fig. 9 An exemplary structural diagram of the electronic device is disclosed. Fig. 9 As shown, the electronic device includes: one or more processors 1101, a memory 1102, and interfaces for connecting various components, including high-speed interfaces and low-speed interfaces. The various components are connected to each other using different buses and can be installed on a common motherboard or installed in other ways as needed. The processor can process instructions executed in the electronic device, including instructions stored in or on the memory to display graphical information of the GUI on an external input / output device (such as a display device coupled to the interface). In some other embodiments, if necessary, multiple processors and / or multiple buses can be used with multiple memories and multiple memories. Similarly, multiple electronic devices can be connected, and each device provides some necessary operations (for example, as a server array, a group of blade servers, or a multi-processor system). Among them, the components shown in this article, their connections and relationships, and their functions are only examples, and are not intended to limit the implementation of the present application described and / or required herein.

[0092] The electronic device may further include: an input device 1103 and an output device 1104. The processor 1101, the memory 1102, the input device 1103 and the output device 1104 may be connected via a bus or other means. Fig. 9 The example of connecting through bus is taken in the following.

[0093] The input device 1103 can receive input digital or character information, and generate key signal input related to the user settings and function control of the electronic device, such as a touch screen, a keypad, a mouse, a track pad, a touch pad, an indicator rod, one or more mouse buttons, a trackball, a joystick and other input devices. The output device 1104 may include a display device, an auxiliary lighting device (e.g., an LED) and a tactile feedback device (e.g., a vibration motor), etc. The display device may include, but is not limited to, a liquid crystal display (LCD), a light emitting diode (LED) display and a plasma display. In some embodiments, the display device may be a touch screen.

[0094] To provide interaction with a user, the electronic device may be a computer. The computer has: a display device (e.g., a CRT (cathode ray tube) or LCD (liquid crystal display) monitor) for displaying information to the user; and a keyboard and a pointing device (e.g., a mouse or a trackball), through which the user can provide input to the computer. Other types of devices may also be used to provide interaction with a user; for example, the feedback provided to the user may be any form of sensory feedback (e.g., visual feedback, auditory feedback, or tactile feedback); and input from the user may be received in any form (including acoustic input, voice input, or tactile input).

[0095] In the embodiments of the present application, a computer program / instruction is stored on a computer-readable medium, and when the computer program / instruction is executed by a processor, the steps of the method provided by any one or more of the above embodiments are implemented. The computer-readable medium may be included in the electronic device described in the above embodiments; or it may exist independently without being assembled into the device. The above computer-readable medium carries one or more computer-readable instructions.

[0096] The memory 1102 can be used as a non-transient computer-readable storage medium, which can be used to store non-transient software programs, non-transient computer executable programs and modules. The processor 1101 executes various functional applications and data processing of the server by running the non-transient software programs, instructions and modules stored in the memory 1102, so as to implement the program instructions / modules corresponding to the method provided by any one or more embodiments in the embodiments of the present application.

[0097] The memory 1102 may include a program storage area and a data storage area, wherein the program storage area may store an operating system, an application required for at least one function; the data storage area may store data created according to the use of the electronic device, etc. In addition, the memory 1102 may include a high-speed random access memory, and may also include a non-transient memory, such as at least one disk storage device, a flash memory device, or other non-transient solid-state storage device. In some embodiments, the memory 1102 may optionally include a memory remotely arranged relative to the processor 1101, and these remote memories may be connected to the electronic device via a network. Examples of the above-mentioned network include, but are not limited to, the Internet, an intranet, a local area network, a mobile communication network, and combinations thereof.

[0098] It should be noted that the computer-readable medium described in this application may be a computer-readable signal medium or a computer-readable storage medium or any combination of the above two. The computer-readable medium may be, for example, but not limited to, an electrical, magnetic, optical, electromagnetic, infrared or semiconductor system, device or device, or any combination of the above. More specific examples of computer-readable storage media may include, but are not limited to: an electrical connection with one or more wires, a portable computer disk, a hard disk, a random access memory (RAM), a read-only memory (ROM), an erasable programmable read-only memory (EPROM or flash memory), an optical fiber, a portable compact disk read-only memory (CD-ROM), an optical storage device, a magnetic storage device, or any suitable combination of the above. In this application, a computer-readable medium may be any tangible medium containing or storing a program that can be used by or in combination with an instruction execution system, device or device.

[0099] Computer readable media include permanent and non-permanent, removable and non-removable media, and can be implemented by any method or technology to store information. Information can be computer readable instructions, data structures, modules of programs or other data. Examples of computer storage media include, but are not limited to, phase change memory (PRAM), static random access memory (SRAM), dynamic random access memory (DRAM), other types of random access memory (RAM), read-only memory (ROM), electrically erasable programmable read-only memory (EEPROM), flash memory or other memory technology, read-only compact disk (CD-ROM), digital versatile disk (DVD) or other optical storage, magnetic cassettes, magnetic tape disk storage or other magnetic storage devices or any other non-transmission media that can be used to store information that can be accessed by a computing device.

[0100] Computer program code for performing the operations of the present application may be written in one or more programming languages ​​or a combination thereof, including object-oriented programming languages, such as Java, Smalltalk, C++, and conventional procedural programming languages, such as "C" or similar programming languages. The program code may be executed entirely on the user's computer, partially on the user's computer, as a separate software package, partially on the user's computer and partially on a remote computer, or entirely on a remote computer or server. In the case of a remote computer, the remote computer may be connected to the user's computer via any type of network, including a local area network (LAN) or a wide area network (WAN), or may be connected to an external computer (e.g., via the Internet using an Internet service provider).

[0101] In the above embodiments, all or part of the embodiments may be implemented by software, hardware, firmware or any combination thereof. For example, an application specific integrated circuit (ASIC), a general-purpose computer or any other similar hardware device may be used to implement the embodiments. In some embodiments, the software program of the present application may be executed by a processor to implement the above steps or functions. Similarly, the software program of the present application (including related data structures) may be stored in a computer-readable recording medium, such as a RAM memory, a magnetic or optical drive or a floppy disk and the like. In addition, some steps or functions of the present application may be implemented by hardware, for example, as a circuit that cooperates with a processor to perform various steps or functions.

[0102] The computer program product provided in the embodiment of the present application includes one or more computer programs / instructions, and when the computer program / instructions are executed by the processor, the process or function described in the embodiment of the present application is generated in whole or in part. The computer may be a general-purpose computer, a special-purpose computer, a computer network, or other programmable devices. The computer instructions may be stored in a computer-readable storage medium, or transmitted from one computer-readable storage medium to another computer-readable storage medium. For example, the computer instructions may be transmitted from a website site, a computer, a server, or a data center to another website site, a computer, a server, or a data center by wired (e.g., coaxial cable, optical fiber, digital subscriber line (DSL)) or wireless (e.g., infrared, wireless, microwave, etc.) mode. The computer-readable storage medium may be any available medium that a computer can access or a data storage device such as a server or a data center that includes one or more available media integrations. The available medium may be a magnetic medium, (e.g., a floppy disk, a hard disk, a tape), an optical medium (e.g., a DVD), or a semiconductor medium (e.g., a solid state disk (SSD)), etc.

[0103] The flow chart or block diagram in the accompanying drawings shows the possible architecture, function and operation of the equipment, method and computer program product according to various embodiments of the present application. In this regard, each box in the flow chart or block diagram can represent a module, a program segment or a part of a code, and the module, the program segment or a part of the code contains one or more executable instructions for realizing the specified logical function. It should also be noted that in some alternative implementations, the functions marked in the box can also occur in a different order from the order marked in the accompanying drawings. For example, two boxes represented in succession can actually be executed substantially in parallel, and they can sometimes be executed in the opposite order, depending on the functions involved. It should also be noted that each box in the block diagram and / or flow chart, and the combination of the boxes in the block diagram and / or flow chart can be implemented with a dedicated system for hardware that performs a specified function or operation, or can be implemented with a combination of dedicated hardware and computer instructions.

[0104] The scope of the present application is defined by the appended claims rather than the above description, and it is intended that all changes that fall within the meaning and scope of the equivalent elements of the claims are included in the present application. Any figure mark in the claims should not be regarded as limiting the claims involved. In addition, it is clear that the word "comprising" does not exclude other units or steps, and the singular does not exclude the plural. Multiple units or devices stated in the device claim may also be implemented by one unit or device through software or hardware. The words "first", "second", etc. are only used to distinguish the description, and do not indicate any particular order, nor can they be understood as indicating or implying relative importance.

[0105] The above is only a specific implementation of the present application, but the protection scope of the present application is not limited thereto. Any person skilled in the art can easily mention changes or substitutions within the technical scope disclosed in the present application, which should be included in the protection scope of the present application. Therefore, the protection scope of the present application should be based on the protection scope of the claims, and the above embodiments should be regarded as exemplary and non-restrictive.

Claims

1. A method for indirectly predicting battery health status based on expansion stress, characterized in that: include: Collect the time series values ​​of expansion stress, current, voltage, operating temperature, state of charge SOC and remaining service life RUL during several charge and discharge processes of the battery, and indirectly calculate the battery health status SOH data of each time series through the state of charge SOC and remaining service life RUL data; Establishing a data set including the expansion stress, current, voltage, operating temperature and battery health state SOH in different time series; Constructing a prediction model based on a sparse hybrid expert model, wherein the prediction model includes a plurality of expert sub-models and a gating network model; Each of the expert sub-models has a 5-layer structure, the input layer receives input data, the middle layer has 3 layers, the dimensions are 128, 64 and 32 respectively, and the output layer outputs a 16-dimensional vector; The gated network model is a 3-layer network structure, the input layer receives input data, the middle layer is a 32-dimensional vector, and the output layer 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; the gated network outputs a 3-dimensional weight vector and is sparsely activated through Gumbel-Softmax; Each expert sub-model is a multi-layer perceptron, the gated network model is a multi-layer perceptron, and sparse activation is implemented using the Gumbel-Softmax algorithm; Using the data set to train and verify the prediction model; The trained prediction model is used to predict the battery's state of health (SOH) and output the prediction result.

2. The method according to claim 1, characterized in that in, The indirect calculation of the battery health state SOH data of each time series through the state of charge SOC and the remaining service life RUL data includes: The batteries are grouped by remaining useful life RUL, and the maximum value of the state of charge SOC is calculated in each group. This maximum value is used as the SOH value of each data record in the corresponding remaining useful life RUL cycle period. The entire battery health state SOH sequence is then removed and filled with abnormal values ​​to obtain normal battery health state SOH sequence data.

3. The method according to claim 1, characterized in that in, The number of the expert sub-models is N, and the outputs of the N expert sub-models are output_1, output_2, ..., output_N, and the N outputs are all high-dimensional vectors; The output of the gating network model is an N-dimensional vector, which corresponds to the weights of the N expert sub-models, and the sum of the N weights is 1.

4. The method according to claim 3, characterized in that in, The N-dimensional vector output by the gated network model is used as the input of the Gumbel-Softmax algorithm. The Gumbel-Softmax algorithm re-outputs an N-dimensional vector as the weight of the expert sub-model by adding Gumbel noise and Softmax mapping, thereby realizing sparse activation of the expert sub-model. Performing a weighted summation of the expert sub-model weight and the output of the corresponding expert sub-model to obtain a comprehensive feature representation; Connecting the comprehensive features to a fully connected layer; Through the linear regression layer, the battery's state of health SOH is output.

5. The method according to claim 4, characterized in that in, N is 3.

6. The method according to claim 4, characterized in that in, The steps of the Gumbel-Softmax algorithm are as follows: (1) Initial logits Assume a logits vector z containing M elements, which represents the pre-activation score for each expert, corresponding to the weight vector of the expert sub-model output by the gating network model; (2) Add Gumbel noise Add a noise g sampled from a Gumbel(0, 1) distribution to each logit: z′=z+g Among them, z′ is the logits vector with Gumbel noise added, , u is a vector sampled from the uniform distribution U(0,1); (3) Softmax Mapping Apply the Softmax function to the logits z′ with Gumbel noise added to obtain a probability distribution π: ； in, represents the i-th element in the probability distribution π, that is, the probability of the i-th category; Represents the i-th element of the logits vector after adding Gumbel noise; is the temperature parameter, which controls the sharpness of the softmax distribution; when 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.

7. A system for indirectly predicting the health status of a battery based on expansion stress, used to implement the method for indirectly predicting the health status of a battery based on expansion stress as claimed in any one of claims 1 to 6, characterized in that: include: The acquisition unit is used to collect the time series values ​​of expansion stress, current, voltage, operating temperature, state of charge SOC and remaining service life RUL during several charge and discharge processes of the battery, and indirectly calculate the battery health status SOH data of each time series through the state of charge SOC and remaining service life RUL data; An establishing unit, used for establishing a data set including the expansion stress, current, voltage, operating temperature and battery health state SOH in different time series; A construction unit, used to construct a prediction model based on a sparse hybrid expert model, wherein the prediction model includes a plurality of expert sub-models and a gating network model; A training unit, used to train and verify the prediction model using the data set; The prediction unit is used to predict the health state SOH of the battery by using the trained prediction model and output the prediction result.

Images