A method and device for intelligently predicting battery health status based on multi-source information

Abstract

The present invention relates to the field of battery technology, and in particular to a method and device for intelligently predicting the health status of a battery based on multi-source information. The method comprises: collecting and processing raw data to obtain processed acoustic emission parameter data and processed cycle aging parameter data; using a gated neural network to weight the processed acoustic emission parameter data and the processed cycle aging parameter data to obtain data weights; constructing and training a model based on a convolutional neural network structure, a long-term and short-term neural network structure, and a STAD model structure of a SOFTS model to obtain a battery capacity prediction model; based on the data weights, inputting the processed acoustic emission parameter data and the processed cycle aging parameter data into the battery capacity prediction model for prediction to obtain a battery health status prediction value. The present invention is an efficient and accurate method for intelligently predicting the health status of a battery based on multi-source information.

Description

Technical Field

[0001] The present invention relates to the field of battery technology, and in particular to a method and device for intelligently predicting battery health status based on multi-source information. Background Art

[0002] Lithium-ion batteries are widely used in various electronic devices, electric vehicles and energy storage systems due to their high energy density, long cycle life and relatively low self-discharge rate. Failure of lithium-ion batteries may cause serious safety issues, such as fire or explosion. By predicting the health status of the battery, potential failures or degradation can be identified in advance, and preventive measures can be taken to avoid safety accidents.

[0003] The remaining life of a lithium-ion battery can be quantified as the battery's charge and discharge capacity. The electrical and acoustic characteristics obtained by acoustic emission during the charge and discharge process of a lithium-ion battery can be used as parameters to detect the battery's health status. Therefore, how to use electrical and acoustic characteristics to predict the battery's health status is an urgent task.

[0004] At present, the technology level mostly uses a temporal hybrid neural network architecture (Multilayer Perceptron-LongShort-Term Memory, MLP-LSTM), and a multilayer perceptron (Multilayer Perceptron, MLP) as the first feature extraction layer, which is more suitable for processing low-dimensional data, and the number of parameters required by its fully connected layer is huge, resulting in low model training efficiency. The model architecture proposed in the present invention can effectively process and reduce the dimension of high-dimensional input data. The number of parameters is reduced through convolution operations, which improves training efficiency and is very helpful in capturing complex patterns and features in battery data.

[0005] Most models, such as Bidirectional Long Short-Term Memory (Bi-LSTM) and Convolutional Neural Network-Long Short-Term Memory (CNN-LSTM), capture the bidirectional dependency of time series through two LSTM layers, forward and backward, but do not have the local feature extraction capability of CNN, and have poor overall generalization capability; the CNN-LSTM model lacks a mechanism for dynamically adjusting weights. Compared with the two, the present invention adopts a star aggregation distribution module (STarAggregate Redistribute, STAD), which can dynamically adjust the attention weight of the model according to the importance of different time steps and features, so that the model pays more attention to key features and important time points, improves prediction accuracy, and more clearly explains which features or time steps contribute the most to the prediction results, increasing the interpretability of the model. The attention mechanism in STAD and the parameter sharing of the convolutional layer can play a certain regularization effect, which is helpful for the generalization ability of the model. The present invention is an efficient and accurate intelligent prediction method for battery health status based on multi-source information. Summary of the invention

[0006] In order to solve the technical problems of poor generalization ability and low processing efficiency in the prior art, an embodiment of the present invention provides a method and device for intelligent prediction of battery health status based on multi-source information. The technical solution is as follows:

[0007] On the one hand, a method for intelligently predicting a battery health state based on multi-source information is provided, the method being implemented by a battery health state intelligent prediction device, the method comprising:

[0008] Collecting original acoustic emission parameter data and original cyclic aging parameter data and performing time series normalization processing to obtain acoustic emission parameter data and cyclic aging parameter data;

[0009] Based on the cyclic aging parameter data, the acoustic emission parameter data is aggregated to obtain processed acoustic emission parameter data;

[0010] Using an autoencoder, performing noise reduction processing on the cyclic aging parameter data to obtain processed cyclic aging parameter data;

[0011] Using a gated neural network, weighting the processed acoustic emission parameter data and the processed cyclic aging parameter data to obtain data weights;

[0012] The model is constructed according to the convolutional neural network structure, the long-term and short-term neural network structure and the STAD model structure of the SOFTS model to obtain the battery capacity prediction model to be trained;

[0013] Using preset training data, training the battery capacity prediction model to be trained to obtain a battery capacity prediction model;

[0014] Based on the data weight, the processed acoustic emission parameter data and the processed cycle aging parameter data are input into the battery capacity prediction model to obtain the battery predicted capacity; and the battery health status prediction value is obtained according to the battery predicted capacity and the preset battery rated capacity.

[0015] On the other hand, a battery health state intelligent prediction device based on multi-source information is provided, and the device is applied to a battery health state intelligent prediction method based on multi-source information, and the device includes:

[0016] A multi-source data acquisition module is used to collect original acoustic emission parameter data and original cyclic aging parameter data and perform time series normalization processing to obtain acoustic emission parameter data and cyclic aging parameter data;

[0017] A parameter aggregation module, used for performing aggregation processing on the acoustic emission parameter data based on the cyclic aging parameter data to obtain processed acoustic emission parameter data;

[0018] A parameter denoising module, used to use an autoencoder to perform denoising processing on the cyclic aging parameter data to obtain processed cyclic aging parameter data;

[0019] A weight distribution module, used for using a gated neural network to perform weight distribution on the processed acoustic emission parameter data and the processed cyclic aging parameter data to obtain data weights;

[0020] A model building module is used to build a model according to the convolutional neural network structure, the long-term and short-term neural network structure and the STAD model structure of the SOFTS model to obtain a battery capacity prediction model to be trained;

[0021] A model training module, used to train the battery capacity prediction model to be trained using preset training data to obtain a battery capacity prediction model;

[0022] The battery health state prediction module is used to input the processed acoustic emission parameter data and the processed cycle aging parameter data into the battery capacity prediction model based on the data weight to obtain the battery predicted capacity; and calculate according to the battery predicted capacity and the preset battery rated capacity to obtain the battery health state prediction value.

[0023] On the other hand, a battery health status intelligent prediction device is provided, which includes: a processor; a memory, wherein the memory stores computer-readable instructions, and when the computer-readable instructions are executed by the processor, any one of the above-mentioned battery health status intelligent prediction methods based on multi-source information is implemented.

[0024] On the other hand, a computer-readable storage medium is provided, in which at least one instruction is stored. The at least one instruction is loaded and executed by a processor to implement any one of the above-mentioned methods for intelligent prediction of battery health status based on multi-source information.

[0025] The beneficial effects brought about by the technical solution provided by the embodiment of the present invention include at least:

[0026] The present invention proposes a method for intelligently predicting the health status of a battery based on multi-source information. By collecting acoustic emission sensors and obtaining acoustic emission signal data, an effective early warning indicator can be provided for battery health status detection. Predicting the health status of a battery based on a battery capacity prediction model can solve the problem of low accuracy in predicting the health status of a battery, thereby preventing accidents caused by the health status of a battery. The battery health status prediction accuracy of the present invention is higher, and the method is simple and easy to operate. The present invention is an efficient and accurate method for intelligently predicting the health status of a battery based on multi-source information. BRIEF DESCRIPTION OF THE DRAWINGS

[0027] In order to more clearly illustrate the technical solutions in the embodiments of the present invention, the following briefly introduces the drawings required for use in the description of the embodiments. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without creative work.

[0028] Figure 1 This is a flow chart of a method for intelligently predicting battery health status based on multi-source information provided by an embodiment of the present invention;

[0029] Figure 2 It is a block diagram of a battery health status intelligent prediction device based on multi-source information provided by an embodiment of the present invention;

[0030] Figure 3 It is a structural schematic diagram of a battery health status intelligent prediction device provided by an embodiment of the present invention. DETAILED DESCRIPTION

[0031] The technical solution of the present invention is described below in conjunction with the accompanying drawings.

[0032] In the embodiments of the present invention, words such as "exemplarily" and "for example" are used to indicate examples, illustrations or explanations. Any embodiment or design described as "example" in the present invention should not be interpreted as being more preferred or more advantageous than other embodiments or designs. Specifically, the use of the word "example" is intended to present the concept in a specific way. In addition, in the embodiments of the present invention, the meaning expressed by "and / or" can be both, or it can be either of the two.

[0033] In the embodiments of the present invention, "image" and "picture" can sometimes be used interchangeably. It should be noted that when the difference between them is not emphasized, the meanings they intend to express are the same. "of", "corresponding, relevant" and "corresponding" can sometimes be used interchangeably. It should be noted that when the difference between them is not emphasized, the meanings they intend to express are the same.

[0034] In the embodiments of the present invention, sometimes a subscript such as W1 may be written as a non-subscript such as W1. When the difference is not emphasized, the meanings to be expressed are the same.

[0035] In order to make the technical problems, technical solutions and advantages to be solved by the present invention more clear, a detailed description will be given below with reference to the accompanying drawings and specific embodiments.

[0036] The embodiment of the present invention provides a method for intelligently predicting the health status of a battery based on multi-source information. The method can be implemented by an intelligent prediction device for the health status of a battery, which can be a terminal or a server. Figure 1 The flowchart of the method for intelligent prediction of battery health status based on multi-source information is shown. The processing flow of the method may include the following steps:

[0037] S1. Collecting original acoustic emission parameter data and original cyclic aging parameter data and performing time series normalization processing to obtain acoustic emission parameter data and cyclic aging parameter data.

[0038] In a feasible implementation manner, the present invention obtains multiple parameters from the collected original acoustic emission signals and the battery cycle aging experiment. It is necessary to normalize the time series corresponding to the parameters of the two sensors. The calculation formula of the process is shown in the following formula (1):

[0039] (1)

[0040] in, Corresponding sensor data The parameter values, Corresponding sensor data The minimum value among the parameters, Corresponding sensor data The maximum value among the parameters, is the sensor data after normalization The parameter value.

[0041] Wherein, the acoustic emission parameter data includes the amplitude and ring count of the acoustic emission signal of the battery;

[0042] The cycle aging parameter data includes the current and voltage of the battery.

[0043] In a feasible implementation, the present invention uses a piezoelectric acoustic emission sensor with high frequency response and high sensitivity to capture the tiny sound wave signals generated by the battery during the aging process. The sensor is tightly attached to the surface of the battery, and a coupling agent (silicone grease) is used to ensure good acoustic contact. A battery charge and discharge tester Arbin Instruments is used to obtain cycle aging parameter data including battery current and voltage.

[0044] The two types of sensors collect data simultaneously, with the battery cycle aging parameter data sampling interval being 0.1S, and the acoustic emission parameter data being collected at the same interval.

[0045] S2. Based on the cyclic aging parameter data, the acoustic emission parameter data is aggregated to obtain processed acoustic emission parameter data.

[0046] Optionally, based on the cyclic aging parameter data, the acoustic emission parameter data is aggregated to obtain processed acoustic emission parameter data, including:

[0047] The acoustic emission parameter data are grouped according to the sampling interval of the cyclic aging parameter data to obtain grouped acoustic emission parameter data;

[0048] According to the grouped acoustic emission parameter data, the acoustic emission parameter data in each group is aggregated to obtain processed acoustic emission parameter data.

[0049] In a feasible implementation, in order to ensure the consistency of the time scale, the acoustic emission parameter data is aggregated and characterized, and the acoustic emission data is segmented according to the sampling window of the battery aging cycle. In each aging sampling, multiple acoustic emission samples within the same time are grouped together, and their standard deviation is calculated as the processed acoustic emission parameter data. The process is shown in the following formula (2):

[0050] (2)

[0051] in, is the standard deviation of the acoustic emission parameter data in the kth time period, is the average value of the acoustic emission parameters in the kth time period, is the total number of data points in the kth time period, Represents the amplitude value or ringing count value of the ith data point in the kth time period.

[0052] S3. Use an autoencoder to perform noise reduction processing on the cyclic aging parameter data to obtain processed cyclic aging parameter data.

[0053] Optionally, an autoencoder is used to perform noise reduction processing on the cyclic aging parameter data to obtain processed cyclic aging parameter data, including:

[0054] Adding Gaussian noise to the cyclic aging parameter data to obtain noisy cyclic aging parameter data;

[0055] The noisy cyclic aging parameter data is input into the autoencoder for noise reduction processing to obtain the processed cyclic aging parameter data.

[0056] In a feasible implementation, in order to remove the influence of noise data in the cycle aging data, the battery aging cycle data is subjected to a denoising autoencoder (DAE), and the autoencoder consists of an input layer, a hidden layer, and an output layer.

[0057] Before the noise reduction process, Gaussian noise is added to the cycle aging data. The cycle aging parameter data is a current and voltage sequence matrix. The mathematical expression of the cycle aging parameter data matrix X is as follows (3):

[0058] (3)

[0059] in, Matrix representing cyclic aging parameter data

[0060] Gaussian noise matrix The mathematical expression of is as follows (4):

[0061] (4)

[0062] in, , which means that each element of the Gaussian noise matrix has a mean of 0 and a standard deviation of is sampled from a normal distribution.

[0063] The data matrix after adding Gaussian noise is as follows (5):

[0064] (5)

[0065] in, is the data after adding Gaussian noise, is the cycle aging parameter data, is the added Gaussian noise, is the input parameter dimension, which is 2 this time.

[0066] The vector that will introduce noise Provided to DAE, encoded and decoded by DAE, the process is as follows (6):

[0067] (6)

[0068] Formula (6) is specifically expanded into the following formulas (7) and (8):

[0069] (7)

[0070] (8)

[0071] Formula (8) is specifically expanded into the following formula (9):

[0072] (9)

[0073] in, is the characteristic expression of the hidden layer in the encoder, and r represents the number of neurons in the hidden layer. Output vector for the decoder. Represents the weight matrix of the autoencoder. and is the bias vector of the autoencoder. and is the nonlinear activation function ReLU function.

[0074] S4. Use a gated neural network to assign weights to the processed acoustic emission parameter data and the processed cyclic aging parameter data to obtain data weights.

[0075] Optionally, a gated neural network is used to assign weights to the processed acoustic emission parameter data and the processed cyclic aging parameter data to obtain data weights, including:

[0076] Assign parameters to the gated neural network according to the preset weight parameters and bias parameters;

[0077] Using the assigned gated neural network, the processed acoustic emission parameter data is calculated to obtain the acoustic emission parameter weights;

[0078] The gated neural network after assignment is used to calculate the processed cyclic aging parameter data to obtain the cyclic aging parameter weights.

[0079] In a feasible implementation, in order to ensure the correlation of the time series, the acoustic emission parameter data and the battery aging cycle data are input into the subsequent deep learning model in the form of a sliding window. In order to control the influence of the acoustic emission data parameters and the battery aging cycle data parameters on the final prediction results, a gating network is set to allocate the weights of the two parameter data.

[0080] The gating network consists of a fully connected neural network of the gating network and an activation function Softmax. The output of the gating neural network is shown in the following equations (10) and (11):

[0081] (10)

[0082] (11)

[0083] in, is the output result of the fully connected neural network, and c is the input acoustic emission feature aggregation vector. and are the weight parameters and bias parameters of the fully connected neural network. G is the weight distribution result after the activation function softmax. , It is the weight distribution of the transmission parameters and battery aging data after the gating network distribution.

[0084] After passing through the gating network, the tensor that is finally input into the neural network architecture for prediction is as follows (12):

[0085] (12)

[0086] in, is the acoustic emission parameter tensor after aggregation of features, is the battery aging cycle parameter tensor after DAE processing, It is the gating network that allocates weights to the two and obtains the tensor that is finally input into the neural network architecture.

[0087] S5. Construct a model based on the convolutional neural network structure, the long-term and short-term neural network structure, and the STAD model structure of the SOFTS model to obtain a battery capacity prediction model to be trained.

[0088] In a feasible implementation manner, the architecture of the battery capacity prediction model proposed in the present invention is CNN-LSTM-STAD.

[0089] Convolutional Neural Network (CNN) is used to initially extract simple feature relationships in data; Long Short-Term Memory (LSTM) is used to capture time dependencies in sequence data; the role of the Star Aggregate Redistribute (STAD) module is to capture the dependencies between different channels in multivariate time series forecasting. The STAD module is a core component in the Series-cOre Fused Time Series forecaster (SOFTS), an efficient multivariate time series forecasting model based on multi-layer perceptron.

[0090] S6. Using preset training data, the battery capacity prediction model to be trained is trained to obtain a battery capacity prediction model.

[0091] In a feasible implementation manner, the present invention uses the first 80% of the training data as a training set, 10% as a validation set, and 10% as a test set, and inputs them into a battery capacity prediction model to be trained for prediction training.

[0092] The obtained capacity estimation vector group of the battery with time step number T is as follows (13) and (14):

[0093] (13)

[0094] (14)

[0095] in, is the capacity estimation vector group predicted within the time step T. The length of the vector group is the sliding window length w. The result is the average value of the entire vector group. .

[0096] S7. Based on the data weight, the processed acoustic emission parameter data and the processed cycle aging parameter data are input into the battery capacity prediction model to obtain the battery predicted capacity; and the battery health status prediction value is obtained according to the battery predicted capacity and the preset battery rated capacity.

[0097] Optionally, based on the data weight, the processed acoustic emission parameter data and the processed cycle aging parameter data are input into a battery capacity prediction model to obtain the battery predicted capacity, including:

[0098] According to the data weight, the processed acoustic emission parameter data and the processed cycle aging parameter data are input into the CNN layer of the battery capacity prediction model for feature extraction to obtain the acoustic emission parameter features and the cycle aging parameter features;

[0099] The acoustic emission parameter characteristics and cycle aging parameter characteristics are input into the LSTM layer of the battery capacity prediction model to perform time series capacity prediction and obtain preliminary predicted capacity characteristics;

[0100] According to the preliminary predicted capacity characteristics, the multi-channel data fusion processing is performed through the STAD module of the battery capacity prediction model to obtain the battery predicted capacity characteristics;

[0101] The battery prediction capacity features are input into the fully connected layer of the battery capacity prediction model to generate prediction results and obtain the battery prediction capacity.

[0102] In a feasible implementation, the CNN layer of the battery capacity prediction model consists of a convolutional layer and a pooling layer, wherein:

[0103] The mathematical expression of the convolutional layer is as follows (15):

[0104] (15)

[0105] The mathematical expression of the pooling layer is as follows (16):

[0106] (16)

[0107] The mathematical expression of the output layer is as follows (17):

[0108] (17)

[0109] in, , are the learned weights and biases of the convolutional layers of the CNN layer,' ' represents the convolution operation, the activation function is ReLU; poolsize is the size of the pooling window, and the output result is the preliminary predicted capacity feature ,u is the sliding window size, and d is the feature dimension.

[0110] After the data passes through the CNN layer, the preliminarily processed features are obtained, and these preliminarily extracted features are input into the LSTM layer for the next step of processing.

[0111] The LSTM layer structure of the battery capacity prediction model is as follows:

[0112] Input: data at the current moment , the hidden state at the previous moment and memory unit .

[0113] The LSTM layer is calculated according to the following equations (18)-(24):

[0114] Forget Gate The mathematical expression of is as follows (18):

[0115] (18)

[0116] Input Gate The mathematical expression of is as follows (19):

[0117] (19)

[0118] Candidate memory cells The mathematical expression of is as follows (20):

[0119] (20)

[0120] The mathematical expression for updating the memory unit is as follows (21):

[0121] (twenty one)

[0122] The mathematical expression of the output gate is as follows (22):

[0123] (twenty two)

[0124] The mathematical expression of the hidden state is as follows (23):

[0125] (twenty three)

[0126] The mathematical expression of the total output of the LSTM layer is as follows (24):

[0127] (twenty four)

[0128] in, is the sigmoid activation function, and tanh is the hyperbolic tangent activation function. and Forget Gate The learnable weights and biases of and Input gate The learnable weights and biases of and They are candidate memory cells The learnable weights and biases of and The output gate The learnable weights and biases of , t represents a specific time point in the time step. , is the LSTM hidden layer output, T is the number of time steps, refers to the number of elements in the input sequence, that is, the length of the input sequence, and d is the feature dimension.

[0129] Predict battery capacity Input the fully connected layer to generate the final prediction result. The mean of the output sequence is used as the model prediction value of the current battery capacity at T time steps.

[0130] The battery state of health (SOH) is calculated. The calculation method of SOH estimation is shown in the following formula (25):

[0131] (25)

[0132] Optionally, according to the preliminary predicted capacity characteristics, a multi-channel data fusion process is performed through the STAD module of the battery capacity prediction model to obtain the battery predicted capacity characteristics, including:

[0133] Input the preliminary predicted capacity feature into the MLP1 layer of the STAD module for multi-channel information integration to obtain the core representation; splice the core representation into each channel of the preliminary predicted capacity feature to obtain the spliced ​​prediction feature;

[0134] The spliced ​​prediction features are input into the MLP2 layer of the STAD module, and the linear regression model is used for data fusion to obtain the battery prediction capacity features.

[0135] In a feasible implementation, the present invention adds a STAD module after the LSTM layer. The STAD module realizes information aggregation and distribution through a centralized structure rather than a distributed structure. This centralized structure aggregates information of all channels through a core, and then distributes the core information to each channel. This step includes the method steps of the MLP1 layer and the MLP2 layer.

[0136] The projection of the MLP1 layer is shown in equation (26):

[0137] (26)

[0138] Among them, MLR1: It is a projection function consisting of two fully connected layers and a GELU activation function. .

[0139] The random pooling process is shown in equation (27):

[0140] (27)

[0141] Among them, StochPool is a random pooling operation, which means that the features extracted by the MLP1 layer integrate multiple channels into a core representation .

[0142] The Repeat_Concat operation is shown in equation (28):

[0143] (28)

[0144] Among them, Repeat_Concat is to splice the core representation o to each series representation On the top, we get the splicing prediction features .

[0145] The MLP2 layer fusion process is as follows:

[0146] (29)

[0147] (30)

[0148] Among them, MLP2 is another projection function MLP2: , and add residual connection to obtain the fused battery prediction capacity characteristics , Linear represents a fully connected layer, and each MLP layer consists of two Linear layers.

[0149] The present invention proposes a method for intelligently predicting the health status of a battery based on multi-source information. By collecting acoustic emission sensors and obtaining acoustic emission signal data, an effective early warning indicator can be provided for battery health status detection. Predicting the health status of a battery based on a battery capacity prediction model can solve the problem of low accuracy in predicting the health status of a battery, thereby preventing accidents caused by the health status of a battery. The battery health status prediction accuracy of the present invention is higher, and the method is simple and easy to operate. The present invention is an efficient and accurate method for intelligently predicting the health status of a battery based on multi-source information.

[0150] Figure 2 The block diagram of a device for intelligently predicting the health status of a battery based on multi-source information according to an exemplary embodiment is shown. The device is used in a method for intelligently predicting the health status of a battery based on multi-source information. Figure 2 The device includes a multi-source data acquisition module 210, a parameter aggregation module 220, a parameter noise reduction module 230, a weight allocation module 240, a model construction module 250, a model training module 260 and a battery health status prediction module 270. Among them:

[0151] The multi-source data acquisition module 210 is used to collect original acoustic emission parameter data and original cyclic aging parameter data and perform time series normalization processing to obtain acoustic emission parameter data and cyclic aging parameter data;

[0152] The parameter aggregation module 220 is used to aggregate the acoustic emission parameter data based on the cyclic aging parameter data to obtain processed acoustic emission parameter data;

[0153] The parameter denoising module 230 is used to perform denoising processing on the cyclic aging parameter data using an autoencoder to obtain processed cyclic aging parameter data;

[0154] The weight distribution module 240 is used to use a gated neural network to distribute weights to the processed acoustic emission parameter data and the processed cyclic aging parameter data to obtain data weights;

[0155] A model building module 250 is used to build a model according to a convolutional neural network structure, a long-term and short-term neural network structure, and a STAD model structure of a SOFTS model to obtain a battery capacity prediction model to be trained;

[0156] A model training module 260 is used to train the battery capacity prediction model to be trained using preset training data to obtain a battery capacity prediction model;

[0157] The battery health status prediction module 270 is used to input the processed acoustic emission parameter data and the processed cycle aging parameter data into the battery capacity prediction model based on the data weight to obtain the battery predicted capacity; and calculate according to the battery predicted capacity and the preset battery rated capacity to obtain the battery health status prediction value.

[0158] The acoustic emission parameter data includes the amplitude and ring count of the acoustic emission signal of the battery;

[0159] Cycle aging parameter data includes battery current and voltage.

[0160] Optionally, the parameter aggregation module 220 is further configured to:

[0161] The acoustic emission parameter data are grouped according to the sampling interval of the cyclic aging parameter data to obtain grouped acoustic emission parameter data;

[0162] According to the grouped acoustic emission parameter data, the acoustic emission parameter data in each group is aggregated to obtain processed acoustic emission parameter data.

[0163] Optionally, the parameter noise reduction module 230 is further used to:

[0164] Adding Gaussian noise to the cyclic aging parameter data to obtain noisy cyclic aging parameter data;

[0165] The noisy cyclic aging parameter data is input into the autoencoder for noise reduction processing to obtain the processed cyclic aging parameter data.

[0166] Optionally, the weight allocation module 240 is further configured to:

[0167] Assign parameters to the gated neural network according to the preset weight parameters and bias parameters;

[0168] Using the assigned gated neural network, the processed acoustic emission parameter data is calculated to obtain the acoustic emission parameter weights;

[0169] The gated neural network after assignment is used to calculate the processed cyclic aging parameter data to obtain the cyclic aging parameter weights.

[0170] Optionally, the battery health status prediction module 270 is further configured to:

[0171] According to the data weight, the processed acoustic emission parameter data and the processed cycle aging parameter data are input into the CNN layer of the battery capacity prediction model for feature extraction to obtain the acoustic emission parameter features and the cycle aging parameter features;

[0172] The acoustic emission parameter characteristics and cycle aging parameter characteristics are input into the LSTM layer of the battery capacity prediction model to perform time series capacity prediction and obtain preliminary predicted capacity characteristics;

[0173] According to the preliminary predicted capacity characteristics, the multi-channel data fusion processing is performed through the STAD module of the battery capacity prediction model to obtain the battery predicted capacity characteristics;

[0174] The battery prediction capacity features are input into the fully connected layer of the battery capacity prediction model to generate prediction results and obtain the battery prediction capacity.

[0175] Optionally, the battery health status prediction module 270 is further configured to:

[0176] Input the preliminary predicted capacity feature into the MLP1 layer of the STAD module for multi-channel information integration to obtain the core representation; splice the core representation into each channel of the preliminary predicted capacity feature to obtain the spliced ​​prediction feature;

[0177] The spliced ​​prediction features are input into the MLP2 layer of the STAD module, and the linear regression model is used for data fusion to obtain the battery prediction capacity features.

[0178] The present invention proposes a method for intelligently predicting the health status of a battery based on multi-source information. By collecting acoustic emission sensors and obtaining acoustic emission signal data, an effective early warning indicator can be provided for battery health status detection. Predicting the health status of a battery based on a battery capacity prediction model can solve the problem of low accuracy in predicting the health status of a battery, thereby preventing accidents caused by the health status of a battery. The battery health status prediction accuracy of the present invention is higher, and the method is simple and easy to operate. The present invention is an efficient and accurate method for intelligently predicting the health status of a battery based on multi-source information.

[0179] Figure 3 is a schematic diagram of the structure of a battery health status intelligent prediction device provided by an embodiment of the present invention, such as Figure 3 As shown, the battery health status intelligent prediction device may include the above Figure 2 Optionally, the battery health state intelligent prediction device 310 may include a first processor 2001 .

[0180] Optionally, the battery health status intelligent prediction device 310 may also include a memory 2002 and a transceiver 2003 .

[0181] The first processor 2001, the memory 2002 and the transceiver 2003 may be connected via a communication bus.

[0182] Combine the following Figure 3 The components of the battery health status intelligent prediction device 310 are specifically introduced:

[0183] The first processor 2001 is the control center of the battery health status intelligent prediction device 310, which can be a processor or a general term for multiple processing elements. For example, the first processor 2001 is one or more central processing units (CPUs), or an application specific integrated circuit (ASIC), or one or more integrated circuits configured to implement the embodiments of the present invention, such as one or more microprocessors (digital signal processors, DSPs), or one or more field programmable gate arrays (field programmable gate arrays, FPGAs).

[0184] Optionally, the first processor 2001 can perform various functions of the battery health status intelligent prediction device 310 by running or executing a software program stored in the memory 2002 and calling data stored in the memory 2002.

[0185] In a specific implementation, as an embodiment, the first processor 2001 may include one or more CPUs, such as Figure 3 CPU0 and CPU1 are shown in FIG.

[0186] In a specific implementation, as an embodiment, the battery health status intelligent prediction device 310 may also include multiple processors, such as Figure 3The first processor 2001 and the second processor 2004 are shown in FIG. Each of these processors may be a single-core processor (single-CPU) or a multi-core processor (multi-CPU). The processor here may refer to one or more devices, circuits, and / or processing cores for processing data (e.g., computer program instructions).

[0187] The memory 2002 is used to store the software program for executing the solution of the present invention, and is controlled to be executed by the first processor 2001. The specific implementation method can refer to the above method embodiment, which will not be repeated here.

[0188] Optionally, the memory 2002 may be a read-only memory (ROM) or other types of static storage devices that can store static information and instructions, a random access memory (RAM) or other types of dynamic storage devices that can store information and instructions, or an electrically erasable programmable read-only memory (EEPROM), a compact disc read-only memory (CD-ROM) or other optical disc storage, optical disc storage (including compressed optical disc, laser disc, optical disc, digital versatile disc, Blu-ray disc, etc.), a magnetic disk storage medium or other magnetic storage device, or any other medium that can be used to carry or store the desired program code in the form of instructions or data structures and can be accessed by a computer, but is not limited thereto. The memory 2002 may be integrated with the first processor 2001, or may exist independently, and may be accessed through the interface circuit ( Figure 3 (not shown) is coupled to the first processor 2001, which is not specifically limited in this embodiment of the present invention.

[0189] The transceiver 2003 is used to communicate with a network device or a terminal device.

[0190] Optionally, the transceiver 2003 may include a receiver and a transmitter ( Figure 3 The receiver is used to implement a receiving function, and the transmitter is used to implement a sending function.

[0191] Optionally, the transceiver 2003 may be integrated with the first processor 2001, or may exist independently, and may be connected to the battery health status intelligent prediction device 310 through an interface circuit ( Figure 3 (not shown) is coupled to the first processor 2001, which is not specifically limited in this embodiment of the present invention.

[0192] It should be noted that Figure 3 The structure of the battery health status intelligent prediction device 310 shown in the figure does not constitute a limitation on the router. The actual knowledge structure identification device may include more or fewer components than shown in the figure, or combine certain components, or arrange the components differently.

[0193] In addition, the technical effects of the battery health status intelligent prediction device 310 can refer to the technical effects of the battery health status intelligent prediction method based on multi-source information described in the above method embodiment, which will not be repeated here.

[0194] It should be understood that the first processor 2001 in the embodiment of the present invention may be a central processing unit (CPU), and the processor may also be other general-purpose processors, digital signal processors (DSP), application specific integrated circuits (ASIC), field programmable gate arrays (FPGA) or other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components, etc. A general-purpose processor may be a microprocessor or the processor may also be any conventional processor, etc.

[0195] It should also be understood that the memory in the embodiments of the present invention may be a volatile memory or a non-volatile memory, or may include both volatile and non-volatile memories. Among them, the non-volatile memory may be a read-only memory (ROM), a programmable read-only memory (PROM), an erasable programmable read-only memory (EPROM), an electrically erasable programmable read-only memory (EEPROM), or a flash memory. The volatile memory may be a random access memory (RAM), which is used as an external cache. By way of example and not limitation, many forms of random access memory (RAM) are available, such as static RAM (SRAM), dynamic random access memory (DRAM), synchronous DRAM (SDRAM), double data rate synchronous dynamic random access memory (DDR SDRAM), enhanced synchronous dynamic random access memory (ESDRAM), synchronous link DRAM (SLDRAM), and direct rambus RAM (DR RAM).

[0196] The above embodiments can be implemented in whole or in part by software, hardware (such as circuits), firmware or any other combination. When implemented by software, the above embodiments can be implemented in whole or in part in the form of a computer program product. The computer program product includes one or more computer instructions or computer programs. When the computer instructions or computer programs are loaded or executed on a computer, the process or function described in the embodiment of the present invention is generated in whole or in part. The computer can be a general-purpose computer, a special-purpose computer, a computer network, or other programmable device. The computer instructions can 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 can be transmitted from one website, computer, server or data center to another website, computer, server or data center by wired (such as infrared, wireless, microwave, etc.). The computer-readable storage medium can be any available medium that can be accessed by a computer or a data storage device such as a server or data center that contains one or more available media sets. The available medium can be a magnetic medium (for example, a floppy disk, a hard disk, a tape), an optical medium (for example, a DVD), or a semiconductor medium. The semiconductor medium can be a solid-state hard disk.

[0197] It should be understood that the term "and / or" in this article is only a description of the association relationship of associated objects, indicating that there can be three relationships. For example, A and / or B can represent: A exists alone, A and B exist at the same time, and B exists alone. A and B can be singular or plural. In addition, the character " / " in this article generally indicates that the associated objects before and after are in an "or" relationship, but it may also indicate an "and / or" relationship. Please refer to the context for specific understanding.

[0198] In the present invention, "at least one" means one or more, and "more than one" means two or more. "At least one of the following" or similar expressions refers to any combination of these items, including any combination of single or plural items. For example, at least one of a, b, or c can mean: a, b, c, ab, ac, bc, or abc, where a, b, and c can be single or multiple.

[0199] It should be understood that in various embodiments of the present invention, the size of the serial numbers of the above-mentioned processes does not mean the order of execution. The execution order of each process should be determined by its function and internal logic, and should not constitute any limitation on the implementation process of the embodiments of the present invention.

[0200] Those skilled in the art will appreciate that the units and algorithm steps of each example described in conjunction with the embodiments disclosed herein can be implemented in electronic hardware, or a combination of computer software and electronic hardware. Whether these functions are performed in hardware or software depends on the specific application and design constraints of the technical solution. Professional and technical personnel can use different methods to implement the described functions for each specific application, but such implementation should not be considered to be beyond the scope of the present invention.

[0201] Those skilled in the art can clearly understand that, for the convenience and brevity of description, the specific working processes of the above-described equipment, devices and units can refer to the corresponding processes in the aforementioned method embodiments and will not be repeated here.

[0202] In the several embodiments provided by the present invention, it should be understood that the disclosed devices, apparatuses and methods can be implemented in other ways. For example, the device embodiments described above are only schematic. For example, the division of the units is only a logical function division. There may be other division methods in actual implementation, such as multiple units or components can be combined or integrated into another device, or some features can be ignored or not executed. Another point is that the mutual coupling or direct coupling or communication connection shown or discussed can be through some interfaces, indirect coupling or communication connection of devices or units, which can be electrical, mechanical or other forms.

[0203] The units described as separate components may or may not be physically separated, and the components shown as units may or may not be physical units, that is, they may be located in one place or distributed on multiple network units. Some or all of the units may be selected according to actual needs to achieve the purpose of the solution of this embodiment.

[0204] In addition, each functional unit in each embodiment of the present invention may be integrated into one processing unit, or each unit may exist physically separately, or two or more units may be integrated into one unit.

[0205] If the functions are implemented in the form of software functional units and sold or used as independent products, they can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of the present invention, or the part that contributes to the prior art, or the part of the technical solution, can be embodied in the form of a software product. The computer software product is stored in a storage medium, including several instructions for enabling a computer device (which can be a personal computer, server, or network device, etc.) to perform all or part of the steps of the methods described in various embodiments of the present invention. The aforementioned storage medium includes: U disk, mobile hard disk, read-only memory (ROM), random access memory (RAM), disk or optical disk, and other media that can store program codes.

[0206] The above is only a specific embodiment of the present invention, but the protection scope of the present invention is not limited thereto. Any person skilled in the art can easily think of changes or substitutions within the technical scope disclosed by the present invention, which should be included in the protection scope of the present invention. Therefore, the protection scope of the present invention should be based on the protection scope of the claims.

Claims

1. A battery health status intelligent prediction method based on multi-source information, characterized in that: The method comprises: Collecting original acoustic emission parameter data and original cyclic aging parameter data and performing time series normalization processing to obtain acoustic emission parameter data and cyclic aging parameter data; Based on the cyclic aging parameter data, the acoustic emission parameter data is aggregated to obtain processed acoustic emission parameter data; Using an autoencoder, performing noise reduction processing on the cyclic aging parameter data to obtain processed cyclic aging parameter data; Using a gated neural network, weighting the processed acoustic emission parameter data and the processed cyclic aging parameter data to obtain data weights; The method of using a gated neural network to weight the processed acoustic emission parameter data and the processed cyclic aging parameter data to obtain data weights includes: Assign parameters to the gated neural network according to the preset weight parameters and bias parameters; Using the gated neural network after the assignment, calculating the processed acoustic emission parameter data to obtain the acoustic emission parameter weight; Using the assigned gated neural network, calculating the processed cyclic aging parameter data to obtain cyclic aging parameter weights; Wherein, in order to ensure the relevance of the time series, the acoustic emission parameter data after treatment and the cyclic aging parameter data after treatment are in the form of data based on a sliding window; The model is constructed according to the convolutional neural network structure, the long-term and short-term neural network structure and the STAD model structure of the SOFTS model to obtain the battery capacity prediction model to be trained; Using preset training data, training the battery capacity prediction model to be trained to obtain a battery capacity prediction model; Based on the data weight, the processed acoustic emission parameter data and the processed cycle aging parameter data are input into the battery capacity prediction model to obtain the battery predicted capacity; and the battery health status prediction value is obtained by calculating according to the battery predicted capacity and the preset battery rated capacity; Wherein, based on the data weight, inputting the processed acoustic emission parameter data and the processed cycle aging parameter data into the battery capacity prediction model to obtain the battery predicted capacity includes: According to the data weight, the processed acoustic emission parameter data and the processed cycle aging parameter data are input into the CNN layer of the battery capacity prediction model for feature extraction to obtain acoustic emission parameter features and cycle aging parameter features; Inputting the acoustic emission parameter characteristics and the cycle aging parameter characteristics into the LSTM layer of the battery capacity prediction model to perform time series capacity prediction to obtain preliminary predicted capacity characteristics; According to the preliminary predicted capacity characteristics, a multi-channel data fusion process is performed through the STAD module of the battery capacity prediction model to obtain a battery predicted capacity characteristic; Inputting the battery predicted capacity feature into the fully connected layer of the battery capacity prediction model to generate a prediction result, thereby obtaining the battery predicted capacity; Wherein, the multi-channel data fusion processing is performed through the STAD module of the battery capacity prediction model according to the preliminary predicted capacity characteristics to obtain the battery predicted capacity characteristics, including: Inputting the preliminary predicted capacity feature into the MLP1 layer of the STAD module for multi-channel information integration to obtain a core representation; splicing the core representation into each channel of the preliminary predicted capacity feature to obtain a spliced ​​prediction feature; The spliced ​​prediction features are input into the MLP2 layer of the STAD module, and a linear regression model is used to perform data fusion to obtain the battery prediction capacity features.

2. The method for intelligent prediction of battery health status based on multi-source information according to claim 1, characterized in that: The acoustic emission parameter data includes the amplitude and ring count of the acoustic emission signal of the battery; The cycle aging parameter data includes the current and voltage of the battery.

3. The method for intelligent prediction of battery health status based on multi-source information according to claim 1, characterized in that: The step of performing aggregation processing on the acoustic emission parameter data based on the cyclic aging parameter data to obtain processed acoustic emission parameter data includes: According to the sampling interval of the cyclic aging parameter data, the acoustic emission parameter data is grouped to obtain grouped acoustic emission parameter data; According to the grouped acoustic emission parameter data, the acoustic emission parameter data in each group is aggregated to obtain processed acoustic emission parameter data.

4. The method for intelligent prediction of battery health status based on multi-source information according to claim 1, characterized in that: The method of using an autoencoder to perform noise reduction processing on the cyclic aging parameter data to obtain processed cyclic aging parameter data includes: Adding Gaussian noise to the cyclic aging parameter data to obtain noisy cyclic aging parameter data; The noisy cyclic aging parameter data is input into an autoencoder for noise reduction processing to obtain processed cyclic aging parameter data.

5. A battery health status intelligent prediction device based on multi-source information, the battery health status intelligent prediction device based on multi-source information is used to implement the battery health status intelligent prediction method based on multi-source information as claimed in any one of claims 1 to 4, characterized in that: The device comprises: A multi-source data acquisition module is used to collect original acoustic emission parameter data and original cyclic aging parameter data and perform time series normalization processing to obtain acoustic emission parameter data and cyclic aging parameter data; A parameter aggregation module, used for performing aggregation processing on the acoustic emission parameter data based on the cyclic aging parameter data to obtain processed acoustic emission parameter data; A parameter denoising module, used to use an autoencoder to perform denoising processing on the cyclic aging parameter data to obtain processed cyclic aging parameter data; A weight distribution module, used for using a gated neural network to perform weight distribution on the processed acoustic emission parameter data and the processed cyclic aging parameter data to obtain data weights; Wherein, the weight allocation module is further used for: Assign parameters to the gated neural network according to the preset weight parameters and bias parameters; Using the gated neural network after the assignment, calculating the processed acoustic emission parameter data to obtain the acoustic emission parameter weight; Using the assigned gated neural network, calculating the processed cyclic aging parameter data to obtain cyclic aging parameter weights; Wherein, in order to ensure the relevance of the time series, the acoustic emission parameter data after treatment and the cyclic aging parameter data after treatment are in the form of data based on a sliding window; A model building module is used to build a model according to the convolutional neural network structure, the long-term and short-term neural network structure and the STAD model structure of the SOFTS model to obtain a battery capacity prediction model to be trained; A model training module, used to train the battery capacity prediction model to be trained using preset training data to obtain a battery capacity prediction model; A battery health state prediction module is used to input the processed acoustic emission parameter data and the processed cycle aging parameter data into the battery capacity prediction model based on the data weight to obtain the battery predicted capacity; and calculate according to the battery predicted capacity and the preset battery rated capacity to obtain the battery health state prediction value; Wherein, the battery health status prediction module is further used for: According to the data weight, the processed acoustic emission parameter data and the processed cycle aging parameter data are input into the CNN layer of the battery capacity prediction model for feature extraction to obtain acoustic emission parameter features and cycle aging parameter features; Inputting the acoustic emission parameter characteristics and the cycle aging parameter characteristics into the LSTM layer of the battery capacity prediction model to perform time series capacity prediction to obtain preliminary predicted capacity characteristics; According to the preliminary predicted capacity characteristics, a multi-channel data fusion process is performed through the STAD module of the battery capacity prediction model to obtain a battery predicted capacity characteristic; Inputting the battery predicted capacity feature into the fully connected layer of the battery capacity prediction model to generate a prediction result, thereby obtaining the battery predicted capacity; Wherein, the battery health status prediction module is further used for: Inputting the preliminary predicted capacity feature into the MLP1 layer of the STAD module for multi-channel information integration to obtain a core representation; splicing the core representation into each channel of the preliminary predicted capacity feature to obtain a spliced ​​prediction feature; The spliced ​​prediction features are input into the MLP2 layer of the STAD module, and a linear regression model is used to perform data fusion to obtain the battery prediction capacity features.

6. A battery health status intelligent prediction device, characterized in that: The battery health status intelligent prediction device comprises: processor; A memory having computer-readable instructions stored thereon, wherein when the computer-readable instructions are executed by the processor, the method according to any one of claims 1 to 4 is implemented.

7. A computer-readable storage medium, characterized in that: The computer-readable storage medium stores program codes, which can be called by a processor to execute the method according to any one of claims 1 to 4.

Images