Battery Remaining Life Prediction Method and Device Based on Sodium-Ion Batteries

Abstract

The present application discloses a method and device for predicting the remaining battery life of a sodium-ion battery, relating to the technical field of battery life prediction, including: obtaining target observation data of a target sodium-ion battery; determining a plurality of time steps according to the target observation data, and generating a target long time series according to the plurality of time steps; inputting the target long time series into a battery life prediction model to obtain the predicted remaining battery life of the target sodium-ion battery, wherein the battery life prediction model includes a dynamic self-attention encoding layer and a temporal attention decoding layer, the dynamic self-attention encoding layer includes a first multi-head self-attention mechanism and a first feed-forward network, the first feed-forward network includes a first fully-connected layer, a second fully-connected layer, and an activation function layer, and the temporal attention decoding layer includes a second multi-head self-attention mechanism and a second feed-forward network. The present application can improve the prediction accuracy.

Description

Technical Field

[0001] The present application relates to the technical field of battery life prediction, and particularly to a method and device for predicting the remaining life of a battery based on a sodium-ion battery. Background Art

[0002] As a new energy storage technology, sodium-ion batteries are widely used in fields such as electric vehicles and energy storage systems. However, the problem of estimating the remaining life of sodium-ion batteries has always been one of the important challenges in battery management and maintenance. Currently, the remaining life of sodium-ion batteries is mainly estimated based on physical models and data-driven methods, such as Kalman filtering, recurrent neural networks, etc. However, traditional physical models often have difficulties in modeling the non-linear characteristics of batteries, thereby limiting the accuracy and reliability of remaining life prediction.

[0003] The above content is only used to assist in understanding the technical solution of the present application, and does not represent an admission that the above content is prior art. Summary of the Invention

[0004] The main purpose of the present application is to provide a method and device for predicting the remaining life of a battery based on a sodium-ion battery, aiming to solve the technical problem of low accuracy and reliability in predicting the life of a sodium-ion battery by using traditional methods.

[0005] To achieve the above purpose, the present application proposes a method for predicting the remaining life of a battery based on a sodium-ion battery, and the method for predicting the remaining life of a battery based on a sodium-ion battery includes:

[0006] Obtain target observation data of a target sodium-ion battery, where the target observation data at least includes current data, voltage data, and temperature data;

[0007] Determine a plurality of time steps according to the target observation data, and generate a target long time series according to the plurality of time steps, where a time step includes an observed value, a time stamp, a time interval, and an identifier;

[0008] Input the target long time series into a battery life prediction model to obtain the predicted remaining life of the target sodium-ion battery, where the battery life prediction model includes a dynamic self-attention encoding layer and a temporal attention decoding layer, the dynamic self-attention encoding layer includes a first multi-head self-attention mechanism and a first feed-forward network, the first feed-forward network includes a first fully connected layer, a second fully connected layer, and an activation function layer, and the temporal attention decoding layer includes a second multi-head self-attention mechanism and a second feed-forward network.

[0009] In one embodiment, the step of inputting the target long time series into a battery life prediction model to obtain the predicted remaining life of the target sodium-ion battery includes:

[0010] Input the target long time series into the dynamic self-attention encoding layer of the battery life prediction model to obtain self-attention features;

[0011] Input the self-attention features into the temporal attention decoding layer of the battery life prediction model to obtain the predicted remaining battery life of the target sodium-ion battery.

[0012] In one embodiment, the first multi-head self-attention mechanism includes multiple attention heads and a fully connected layer; wherein, the step of inputting the target long time series into the dynamic self-attention encoding layer of the battery life prediction model to obtain self-attention features includes:

[0013] Perform parallel calculations on the target long time series through the multiple attention heads of the first multi-head self-attention mechanism to obtain initial self-attention features;

[0014] Perform a linear transformation on the initial self-attention features through the fully connected layer of the first multi-head self-attention mechanism to obtain self-attention features.

[0015] In one embodiment, before the step of inputting the target long time series into the battery life prediction model to obtain the predicted remaining battery life of the target lithium-ion battery, it further includes:

[0016] Collect the observed data of the sodium-ion battery under different working conditions from the sodium-ion battery system, and collect the battery capacity change values of the sodium-ion battery at different time intervals;

[0017] Determine the original data according to the observed data and the battery capacity change values;

[0018] Perform preprocessing operations on the original data to obtain training data, and use the number of battery cycles of the sodium-ion battery as the label value of the training data, wherein the preprocessing operations include cleaning operations, alignment operations, and standardization operations;

[0019] Construct the battery life prediction model based on the training data and the label value.

[0020] In one embodiment, the step of constructing the battery life prediction model based on the training data and the label value includes:

[0021] Based on the training data, train the first battery life prediction model using the logarithmic likelihood strategy of maximizing the observation probability to obtain the second battery life prediction model;

[0022] Optimize the second battery life prediction model based on the self-supervised loss function of KL divergence to obtain the battery life prediction model.

[0023] In one embodiment, after the step of constructing the battery life prediction model based on the training data and the label values, the following steps are further included:

[0024] Determine the self-attention weight distribution of the battery life prediction model;

[0025] Perform statistical analysis on the self-attention weight distribution to determine the mean and standard deviation of the time steps;

[0026] Based on the mean and the standard deviation, determine the importance corresponding to different time steps;

[0027] Normalize the importance of each time step to obtain the normalized importance;

[0028] After mapping the normalized importance to a heat map, determine the model prediction process according to the heat map;

[0029] Optimize the battery life prediction model according to the model prediction process.

[0030] In one embodiment, after the step of constructing the battery life prediction model based on the training data and the label values, the following steps are further included:

[0031] Verify the battery life prediction model using test data to obtain a verification result;

[0032] Based on the verification result, adjust the hyperparameters and network structure of the battery life prediction model.

[0033] In addition, to achieve the above object, the present application further provides a battery remaining life prediction device based on a sodium-ion battery, and the battery remaining life prediction device based on a sodium-ion battery includes:

[0034] An acquisition module, configured to acquire target observation data of a target sodium-ion battery, where the target observation data at least includes current data, voltage data, and temperature data;

[0035] A determination module, configured to determine a plurality of time steps according to the target observation data, and generate a target long time series according to the plurality of time steps, where a time step includes an observation value, a time stamp, a time interval, and an identifier;

[0036] An input module for inputting the target long time series into a battery life prediction model to obtain the predicted remaining battery life of the target sodium-ion battery, wherein the battery life prediction model includes a dynamic self-attention encoding layer and a temporal attention decoding layer, the dynamic self-attention encoding layer includes a first multi-head self-attention mechanism and a first feed-forward network, the first feed-forward network includes a first fully connected layer, a second fully connected layer, and an activation function layer, and the temporal attention decoding layer includes a second multi-head self-attention mechanism and a second feed-forward network.

[0037] In addition, to achieve the above object, the present application also proposes a device for predicting the remaining battery life of a sodium-ion battery, the device includes: a memory, a processor, and a computer program stored on the memory and executable on the processor, the computer program is configured to implement the steps of the method for predicting the remaining battery life of a sodium-ion battery as described above.

[0038] In addition, to achieve the above object, the present application also proposes a storage medium, the storage medium is a computer-readable storage medium, a computer program is stored on the storage medium, and when the computer program is executed by a processor, it implements the steps of the method for predicting the remaining battery life of a sodium-ion battery as described above.

[0039] In addition, to achieve the above object, the present application also provides a computer program product, the computer program product includes a computer program, and when the computer program is executed by a processor, it implements the steps of the method for predicting the remaining battery life of a sodium-ion battery as described above.

[0040] One or more technical solutions proposed by the present application have at least the following technical effects:

[0041] The present application proposes a method and device for predicting the remaining battery life of a sodium-ion battery. By obtaining target observation data of the target sodium-ion battery, where the target observation data at least includes current data, voltage data, and temperature data; determining a plurality of time steps according to the target observation data, and generating a target long time series according to the plurality of time steps, where a time step includes an observation value, a timestamp, a time interval, and an identifier; inputting the target long time series into a battery life prediction model to obtain the predicted remaining battery life of the target sodium-ion battery, where the battery life prediction model includes a dynamic self-attention encoding layer and a temporal attention decoding layer, the dynamic self-attention encoding layer includes a first multi-head self-attention mechanism and a first feed-forward network, the first feed-forward network includes a first fully-connected layer, a second fully-connected layer, and an activation function layer, and the temporal attention decoding layer includes a second multi-head self-attention mechanism and a second feed-forward network. It solves the technical problem that the traditional method for predicting the life of a sodium-ion battery has low accuracy and reliability. Compared with the prior art, the present application can capture the temporal changes in battery performance from the current data, voltage data, and temperature data of the battery through the dynamic self-attention encoding layer, and can also use the features extracted by the encoding layer through the temporal attention decoding layer to predict the remaining battery life. This embodiment fully considers the temporal characteristics of battery performance, effectively improving the accuracy, adaptability, and interpretability of the remaining battery life prediction. BRIEF DESCRIPTION OF THE DRAWINGS

[0042] The accompanying drawings herein are incorporated into the specification and form a part of the specification, showing embodiments consistent with the present application, and are used together with the specification to explain the principles of the present application.

[0043] In order to more clearly illustrate the technical solutions in the embodiments of the present application or the prior art, the following will briefly introduce the drawings required for use in the description of the embodiments or the prior art. Obviously, for those of ordinary skill in the art, without creative efforts, other drawings can also be obtained based on these drawings.

[0044] Figure 1 It is a schematic flowchart provided for Embodiment 1 of the method for predicting the remaining battery life of a sodium-ion battery based on the present application;

[0045] Figure 2 It is a schematic flowchart provided for Embodiment 2 of the method for predicting the remaining battery life of a sodium-ion battery based on the present application;

[0046] Figure 3 It is a schematic module structure diagram of the device for predicting the remaining battery life of a sodium-ion battery based on the embodiments of the present application;

[0047] Figure 4This is a schematic diagram of the device structure of the hardware operating environment involved in the battery remaining life prediction method based on sodium-ion batteries in the embodiments of the present application.

[0048] The implementation, functional features and advantages of the present application will be further described with reference to the embodiments and the accompanying drawings. Specific embodiments

[0049] It should be understood that the specific embodiments described herein are only used to explain the technical solutions of the present application and are not used to limit the present application.

[0050] In order to better understand the technical solutions of the present application, the following will be described in detail with reference to the accompanying drawings of the specification and specific embodiments.

[0051] The main solution of the embodiments of the present application is: obtaining target observation data of a target sodium-ion battery, where the target observation data at least includes current data, voltage data, and temperature data; determining a plurality of time steps according to the target observation data, and generating a target long time series according to the plurality of time steps, where a time step includes an observation value, a timestamp, a time interval, and an identifier; inputting the target long time series into a battery life prediction model to obtain the predicted remaining battery life of the target sodium-ion battery, where the battery life prediction model includes a dynamic self-attention encoding layer and a temporal attention decoding layer, the dynamic self-attention encoding layer includes a first multi-head self-attention mechanism and a first feed-forward network, the first feed-forward network includes a first fully connected layer, a second fully connected layer, and an activation function layer, and the temporal attention decoding layer includes a second multi-head self-attention mechanism and a second feed-forward network.

[0052] In this embodiment, for the convenience of description, the following will be described with an identification-based battery remaining life prediction device for sodium-ion batteries as the execution subject.

[0053] Since traditional physical models often have difficulties in modeling the non-linear characteristics of batteries, the accuracy and reliability of remaining life prediction are limited.

[0054] The present application provides a solution that can capture the temporal changes in battery performance from the current data, voltage data, and temperature data of the battery through the dynamic self-attention encoding layer, and can also use the features extracted by the encoding layer through the temporal attention decoding layer to predict the remaining battery life. This embodiment fully considers the temporal characteristics of battery performance, effectively improving the accuracy, adaptability, and interpretability of battery remaining life prediction.

[0055] It should be noted that the execution subject of this embodiment can be a computing service device with data processing, network communication, and program running functions, such as a tablet computer, a personal computer, a mobile phone, etc., or an electronic device capable of implementing the above functions, a battery remaining life prediction device based on sodium-ion batteries, etc. Hereinafter, a battery remaining life prediction device based on sodium-ion batteries will be taken as an example to illustrate this embodiment and the following embodiments.

[0056] Based on this, an embodiment of the present application provides a method for predicting the remaining life of a battery based on sodium-ion batteries. Refer to Figure 1 , Figure 1 which is a schematic flowchart of the first embodiment of the method for predicting the remaining life of a battery based on sodium-ion batteries in the present application.

[0057] In this embodiment, the method for predicting the remaining life of a battery based on sodium-ion batteries includes steps S10 to S30:

[0058] Step S10, obtaining target observation data of a target sodium-ion battery, where the target observation data at least includes current data, voltage data, and temperature data;

[0059] It should be noted that through the current data, the performance of the battery during charge and discharge can be understood. For example, the usage pattern and aging condition of the battery can be understood through the current change of the battery during charge and discharge cycles; for another example, by combining the current data with the charge and discharge capacity of the battery, the attenuation trend of the battery capacity over time can be revealed; for another example, by analyzing the relationship between the charge and discharge current and voltage, the change of the battery internal resistance can be estimated to reflect battery aging (because the increase in internal resistance is usually a signal of battery aging). For another example, health indicators can be extracted from the current data, such as the peak value of the incremental capacity (IC) during discharge, the voltage value corresponding to the peak value, the peak area, and the peak slope, which can be used as quantitative indicators of the battery health state.

[0060] It should be noted that since the voltage data is closely related to the health state of the battery, by analyzing the voltage change during the charge and discharge process of the battery, indicators related to the health state of the battery can be extracted, such as peak values, the voltage values corresponding to the peak values, the peak area, and the peak slope, etc.; since the attenuation of the battery capacity is often accompanied by the change of the voltage platform, by monitoring these changes, the decline trend of the battery capacity can be evaluated; since the voltage data can reveal the mechanism of battery aging, such as the demarcation of the electrolyte, the structural change of the electrode material, etc.

[0061] It should be noted that since the temperature of the battery will affect the rate of the electrochemical reaction inside it, and thus affect the performance and life of the battery. For example, a high-temperature environment will exacerbate the demarcation of the electrolyte and the degradation of the electrode material, thereby accelerating the aging process of the battery.

[0062] It can be understood that the target observation data may also include discharge capacity, cycle life, peak discharge capacity, capacity attenuation rate, charge and discharge time, health index, internal resistance, and so on.

[0063] Step S20: Determine a plurality of time steps according to the target observation data, and generate a target long time series according to the plurality of time steps, where a time step includes an observation value, a timestamp, a time interval, and an identifier;

[0064] It should be noted that the timestamp represents a specific point in time, the observation value represents the observation data for the target sodium-ion battery at the timestamp (such as current data, voltage data, and temperature data), the time interval represents the time difference between two adjacent observation values, the time interval can be fixed or variable, and the identifier is used to distinguish different batteries.

[0065] Step S30: Input the target long time series into the battery life prediction model to obtain the predicted remaining battery life of the target sodium-ion battery, where the battery life prediction model includes a dynamic self-attention encoding layer and a temporal attention decoding layer, the dynamic self-attention encoding layer includes a first multi-head self-attention mechanism and a first feed-forward network, the first feed-forward network includes a first fully connected layer, a second fully connected layer, and an activation function layer, and the temporal attention decoding layer includes a second multi-head self-attention mechanism and a second feed-forward network.

[0066] It should be noted that the battery life prediction model is a trained model. In the first feed-forward network, an activation function layer is added between the first fully connected layer and the second fully connected layer, that is, the first feed-forward network includes a first fully connected layer, an activation function layer, and a second fully connected layer connected in sequence, where the activation function layer is a ReLU activation function. This feed-forward structure further enhances the expression ability of the model by learning non-linear transformations, so as to better fit the non-linear law of battery life.

[0067] It should be noted that the second multi-head self-attention mechanism and the second feed-forward network can be used to generate the remaining battery life. Through end-to-end training, the advantages of self-attention and feed-forward networks can be fully utilized to achieve accurate prediction of battery life.

[0068] In a feasible implementation manner, the step of inputting the target long time series into the battery life prediction model to obtain the predicted remaining battery life of the target sodium-ion battery includes: inputting the target long time series into the dynamic self-attention encoding layer of the battery life prediction model to obtain self-attention features; inputting the self-attention features into the temporal attention decoding layer of the battery life prediction model to obtain the predicted remaining battery life of the target sodium-ion battery.

[0069] It should be noted that the attention features determined by the first multi-head self-attention mechanism can reflect the long-term dependence relationship between the observed data and the time series in the target long time series, which helps to improve the accuracy of battery life prediction; the temporal attention decoding layer can be used to generate the remaining battery life, and it realizes the accurate prediction of battery life by fully utilizing the advantages of the self-attention mechanism and the feed-forward network through end-to-end training.

[0070] In a feasible implementation manner, the first multi-head self-attention mechanism includes multiple attention heads and a fully connected layer; wherein, the step of inputting the target long time series into the dynamic self-attention encoding layer of the battery life prediction model to obtain self-attention features includes: performing parallel calculations on the target long time series through the multiple attention heads of the first multi-head self-attention mechanism to obtain initial self-attention features; performing a linear transformation on the initial self-attention features through the fully connected layer of the first multi-head self-attention mechanism to obtain self-attention features.

[0071] It should be noted that adding a fully connected layer after the first multi-head self-attention mechanism can further extract and combine features; after inputting the target long time series into the battery life prediction model, multiple attention heads perform parallel calculations, and each attention head learns a different attention weight distribution. For example, after each attention head calculates the query, key, and value matrices, it calculates the attention weights through the scaled dot-product attention mechanism, and outputs the initial self-attention features according to the attention weights. Finally, a fully connected layer is used to perform a linear transformation on the initial self-attention features output by multiple attention heads to obtain the final self-attention features.

[0072] In specific implementation, the specific implementation steps for each attention head to learn a different attention weight distribution are as follows:

[0073] (1) After splitting the target long time series into multiple subsequences, they are respectively input into multiple attention heads, each attention head corresponds to processing one subsequence, and the attention heads process the subsequences in parallel; (2) Each attention head generates query (Q), key (K), and value (V) matrices by performing three different linear transformations (i.e., weight matrix multiplications) on the input subsequence; (3) Calculate the attention scores between them by using the dot product of the query and the key through scaled dot-product attention (used to indicate how much weight should be given in the output). For each query vector, calculate the dot product between it and all key vectors to obtain the dot product result, scale the dot product result by a scaling factor, and normalize the scaled dot product result through the Softmax function to obtain the attention weights for each key.

[0074] In this embodiment, by obtaining the target observation data of the target sodium-ion battery, where the target observation data at least includes current data, voltage data, and temperature data; determining a plurality of time steps according to the target observation data, and generating a target long time series according to the plurality of time steps, where a time step includes an observed value, a timestamp, a time interval, and an identifier; inputting the target long time series into a battery life prediction model to obtain the predicted remaining battery life of the target sodium-ion battery, where the battery life prediction model includes a dynamic self-attention encoding layer and a temporal attention decoding layer, the dynamic self-attention encoding layer includes a first multi-head self-attention mechanism and a first feed-forward network, the first feed-forward network includes a first fully connected layer, a second fully connected layer, and an activation function layer, and the temporal attention decoding layer includes a second multi-head self-attention mechanism and a second feed-forward network. It solves the technical problem that the traditional method for predicting the life of a sodium-ion battery has low accuracy and reliability. Compared with the prior art, the present application can capture the temporal changes in battery performance from the current data, voltage data, and temperature data of the battery through the dynamic self-attention encoding layer, and can also use the features extracted by the encoding layer through the temporal attention decoding layer to predict the remaining battery life. This embodiment fully considers the temporal characteristics of battery performance, effectively improving the accuracy, adaptability, and interpretability of the remaining battery life prediction.

[0075] Based on the first embodiment of the present application, in the second embodiment of the present application, the same or similar content as in the above-mentioned first embodiment can be referred to the above introduction and will not be repeated hereinafter. On this basis, please refer to Figure 2 , before step S30, the step of inputting the target long time series into the battery life prediction model to obtain the predicted remaining battery life of the target lithium-ion battery further includes steps S301 to S304:

[0076] Step S301, collecting the observation data of the sodium-ion battery under different working conditions from the sodium-ion battery system, and collecting the battery capacity change values of the sodium-ion battery at different time intervals;

[0077] Step S302, determining the original data according to the observation data and the battery capacity change values;

[0078] Step S303, performing preprocessing operations on the original data to obtain training data, and using the number of battery cycles of the sodium-ion battery as the label value of the training data, where the preprocessing operations include cleaning operations, alignment operations, and standardization operations;

[0079] It should be noted that the cleaning operation refers to removing outliers, missing values or error records in the original data; the alignment operation refers to adjusting the data to be continuous and aligned in the time series, which is particularly important for time series prediction models; the standardization operation refers to standardizing the data so that it has a zero mean and unit variance, which helps the stability and convergence speed of model training; the number of battery cycles can be used to reflect the number of charge and discharge cycles experienced when the battery capacity drops to a certain percentage of the initial capacity, and can be used to represent the remaining battery life; the original data after the preprocessing operation can be used as the input data for the battery life prediction model.

[0080] Step S304, construct the battery life prediction model based on the training data and the label values.

[0081] It should be noted that during the training process of constructing the battery life prediction model, the model can be trained by maximizing the log-likelihood of the observation probability, and at the same time, the self-supervised loss of KL divergence is used to optimize the model, so as to realize the construction of the battery life prediction model.

[0082] It should be noted that the battery life prediction model is a trained model, which can be directly deployed in actual production to predict the remaining life of sodium-ion batteries in real time. At the same time, it can also collect user feedback information, such as the actual remaining life of sodium-ion batteries, and then compare the predicted battery life with the actual remaining life to continuously optimize the battery life prediction model online and update the model continuously to improve the prediction accuracy.

[0083] In a feasible implementation manner, the step of constructing the battery life prediction model based on the training data and the label values includes: based on the training data, training the first battery life prediction model by using the log-likelihood strategy of maximizing the observation probability to obtain the second battery life prediction model; optimizing the second battery life prediction model based on the self-supervised loss function of KL divergence to obtain the battery life prediction model.

[0084] It should be noted that the log-likelihood strategy of maximizing the observation probability uses maximum likelihood estimation (MLE) as the training objective. During the training process, gradient ascent or other algorithms are used to adjust the model parameters to maximize the log-likelihood function, and the model is trained by minimizing the negative log-likelihood (because the log-likelihood is monotonically increasing, minimizing its negative value is equivalent to maximizing itself).

[0085] It should be noted that the self-supervised loss function of KL divergence includes a KL divergence term, which is mainly used to quantify the difference between the model prediction and the true distribution. The self-supervised loss function can be used to guide the optimization process of the model, and the model parameters are adjusted by minimizing the KL divergence.

[0086] In a specific implementation, when training the model, the Teacher Forcing method is used for sequence generation. For example, during training, a randomly initialized state is used as the starting point of the sequence to generate the first prediction result. Even if the first prediction result may be incorrect, the model still uses the real data (instead of its prediction result) as the input at the next time step.

[0087] In a feasible implementation manner, after the step of constructing the battery life prediction model based on the training data and the label value, the following steps are further included: validating the battery life prediction model using test data to obtain a validation result; based on the validation result, adjusting the hyperparameters and network structure of the battery life prediction model.

[0088] It should be noted that an independent test data set can be used to validate the trained model to obtain a validation result. The validation result can be used to evaluate the prediction accuracy and generalization ability of the model, and you can optimize the model according to the validation result, such as adjusting the hyperparameters of the model, optimizing the network structure, etc., so as to improve the prediction performance of the model.

[0089] In a feasible implementation manner, after the step of constructing the battery life prediction model based on the training data and the label value, the following steps are further included: determining the self-attention weight distribution of the battery life prediction model; performing statistical analysis on the self-attention weight distribution to determine the mean and standard deviation of the time steps; based on the mean and the standard deviation, determining the importance corresponding to different time steps; normalizing the importance of each time step to obtain the normalized importance; after mapping the normalized importance to a heat map, determining the model prediction process according to the heat map; optimizing the battery life prediction model according to the model prediction process.

[0090] It can be understood that the attention weight distribution can represent the attention degree of the model to different time steps when predicting the battery life; for each time step, the mean and standard deviation of this time step are calculated according to all the attention weights corresponding to this time step; the time steps with higher mean and lower standard deviation have higher importance. Specifically, a mean weight and a standard deviation weight can be set, and the importance of the time step can be determined based on the mean of the time step, the standard deviation of the time step, the mean weight, and the standard deviation weight.

[0091] It should be noted that after mapping the normalized importance to the heat map, the importance of the time step can be determined according to the color depth on the heat map. For example, the darker the color, the higher the importance of the time step, and the lighter the color, the lower the importance of the time step. Through the heat map, it is possible to more intuitively identify the time steps that the model focuses on during the prediction process, that is, to understand how the model makes predictions based on time series data, and it is also easier for people to understand the decision-making process of the model, so as to optimize the model according to the model prediction process.

[0092] In this embodiment, by collecting the observation data of the sodium-ion battery under different working conditions from the sodium-ion battery system and collecting the battery capacity change values of the sodium-ion battery at different time intervals; determining the original data according to the observation data and the battery capacity change values; performing preprocessing operations on the original data to obtain training data, and using the number of battery cycles of the sodium-ion battery as the label value of the training data, where the preprocessing operations include cleaning operations, alignment operations, and standardization operations; constructing the battery life prediction model based on the training data and the label value. In this way, it is possible to clean, align, and standardize the original data to obtain training data that is more suitable for training the model, which can not only improve the convergence speed of the model but also enhance the training stability of the model.

[0093] Exemplarily, to help understand the implementation process of the battery remaining life prediction method based on sodium-ion batteries obtained by combining this embodiment with the above Embodiment 1, please refer to Figure 3 , Figure 3 A brief flow schematic diagram of a battery remaining life prediction method based on sodium-ion batteries is provided. Specifically:

[0094] It should be noted that the above examples are only for understanding this application and do not constitute a limitation on the battery remaining life prediction method based on sodium-ion batteries of this application. Any simple transformations in more forms based on this technical concept are within the protection scope of this application.

[0095] This application also provides a battery remaining life prediction device based on sodium-ion batteries. Please refer to Figure 3 , the battery remaining life prediction device based on sodium-ion batteries includes:

[0096] An acquisition module 10, configured to acquire target observation data of a target sodium-ion battery, where the target observation data at least includes current data, voltage data, and temperature data;

[0097] A determination module 20, configured to determine a plurality of time steps according to the target observation data and generate a target long time series according to the plurality of time steps, where the time step includes an observed value, a time stamp, a time interval, and an identifier;

[0098] An input module 30 is configured to input the target long time series into a battery life prediction model to obtain the predicted remaining battery life of the target sodium-ion battery. The battery life prediction model includes a dynamic self-attention encoding layer and a temporal attention decoding layer. The dynamic self-attention encoding layer includes a first multi-head self-attention mechanism and a first feed-forward network. The first feed-forward network includes a first fully-connected layer, a second fully-connected layer, and an activation function layer. The temporal attention decoding layer includes a second multi-head self-attention mechanism and a second feed-forward network.

[0099] The battery remaining life prediction device based on sodium-ion batteries provided in this application adopts the battery remaining life prediction method based on sodium-ion batteries in the above embodiment, and can solve the technical problems of low accuracy and reliability in predicting the life of sodium-ion batteries by traditional methods. Compared with the prior art, the beneficial effects of the battery remaining life prediction device based on sodium-ion batteries provided in this application are the same as those of the battery remaining life prediction method based on sodium-ion batteries provided in the above embodiment, and other technical features in the battery remaining life prediction device based on sodium-ion batteries are the same as the features disclosed in the above embodiment method, and will not be elaborated here.

[0100] This application provides a battery remaining life prediction device based on sodium-ion batteries. The battery remaining life prediction device based on sodium-ion batteries includes: at least one processor; and a memory communicatively connected to the at least one processor. The memory stores instructions executable by the at least one processor. When the instructions are executed by the at least one processor, the at least one processor is enabled to execute the battery remaining life prediction method based on sodium-ion batteries in the first embodiment above.

[0101] Reference is made below to Figure 4 , which shows a schematic structural diagram of a battery remaining life prediction device based on sodium-ion batteries suitable for implementing the embodiments of this application. The battery remaining life prediction device based on sodium-ion batteries in the embodiments of this application may include, but is not limited to, mobile terminals such as mobile phones, laptop computers, digital broadcast receivers, PDAs (Personal Digital Assistants), PADs (Portable Application Descriptions), PMPs (Portable Media Players), vehicle-mounted terminals (such as vehicle-mounted navigation terminals), etc., and fixed terminals such as digital TVs, desktop computers, etc. Figure 4 The battery remaining life prediction device based on sodium-ion batteries shown is only an example and should not impose any limitation on the functions and usage scope of the embodiments of this application.

[0102] As Figure 4 shown, the battery remaining life prediction device based on a sodium-ion battery may include a processing device 1001 (such as a central processing unit, a graphics processing unit, etc.), which may perform various appropriate actions and processes according to a program stored in a read-only memory (ROM: Read Only Memory) 1002 or a program loaded from a storage device 1003 into a random access memory (RAM: Random Access Memory) 1004. In the RAM 1004, various programs and data required for the operation of the battery remaining life prediction device based on a sodium-ion battery are also stored. The processing device 1001, the ROM 1002, and the RAM 1004 are connected to each other via a bus 1005. An input / output (I / O) interface 1006 is also connected to the bus. Generally, the following systems may be connected to the I / O interface 1006: an input device 1007 including, for example, a touch screen, a touch pad, a keyboard, a mouse, an image sensor, a microphone, an accelerometer, a gyroscope, etc.; an output device 1008 including, for example, a liquid crystal display (LCD: Liquid Crystal Display), a speaker, a vibrator, etc.; a storage device 1003 including, for example, a magnetic tape, a hard disk, etc.; and a communication device 1009. The communication device 1009 may allow the battery remaining life prediction device based on a sodium-ion battery to communicate with other devices wirelessly or wiredly to exchange data. Although the figure shows a battery remaining life prediction device based on a sodium-ion battery having various systems, it should be understood that it is not required to implement or have all the shown systems. More or fewer systems may be alternatively implemented or had.

[0103] Specifically, according to the embodiments disclosed in the present application, the processes described above with reference to the flowcharts may be implemented as computer software programs. For example, the embodiments disclosed in the present application include a computer program product, which includes a computer program carried on a computer-readable medium, and the computer program includes program codes for performing the methods shown in the flowcharts. In such an embodiment, the computer program may be downloaded and installed from a network through the communication device, or installed from the storage device 1003, or installed from the ROM 1002. When the computer program is executed by the processing device 1001, the above functions defined in the methods of the embodiments disclosed in the present application are executed.

[0104] The battery remaining life prediction device based on sodium-ion batteries provided by this application adopts the battery remaining life prediction method based on sodium-ion batteries in the above-mentioned embodiments, and can solve the technical problems of low accuracy and reliability in realizing the life prediction of sodium-ion batteries by traditional methods. Compared with the prior art, the beneficial effects of the battery remaining life prediction device based on sodium-ion batteries provided by this application are the same as those of the battery remaining life prediction method based on sodium-ion batteries provided by the above-mentioned embodiments, and other technical features in the battery remaining life prediction device based on sodium-ion batteries are the same as those disclosed in the previous embodiment method, and will not be elaborated here.

[0105] It should be understood that each part disclosed in this application can be implemented by hardware, software, firmware or a combination thereof. In the description of the above embodiments, specific features, structures, materials or characteristics can be combined in a suitable manner in any one or more embodiments or examples.

[0106] As described above, only the specific implementation manners of this application are provided, but the protection scope of this application is not limited thereto. Any person skilled in the art can easily think of changes or substitutions within the technical scope disclosed in this application, and all of them should be covered by the protection scope of this application. Therefore, the protection scope of this application should be subject to the protection scope of the claims.

[0107] This application provides a computer-readable storage medium with computer-readable program instructions (i.e., computer programs) stored thereon, and the computer-readable program instructions are used to execute the battery remaining life prediction method based on sodium-ion batteries in the above-mentioned embodiments.

[0108] The computer-readable storage medium provided by this application can be, for example, a USB flash drive, but is not limited to electrical, magnetic, optical, electromagnetic, infrared, or semiconductor systems, devices, or components, or any combination of the above. More specific examples of computer-readable storage media may include, but are not limited to: electrical connections with one or more wires, portable computer disks, hard disks, random access memory (RAM), read-only memory (ROM), erasable programmable read-only memory (EPROM) or flash memory, optical fibers, portable compact disk read-only memory (CD-ROM), optical storage devices, magnetic storage devices, or any suitable combination of the above. In this embodiment, the computer-readable storage medium can be any tangible medium that contains or stores a program that can be used by or in conjunction with an instruction execution system, device, or component. The program code contained on the computer-readable storage medium can be transmitted using any appropriate medium, including but not limited to: wires, optical cables, RF (radio frequency), etc., or any suitable combination of the above.

[0109] The above computer-readable storage medium can be included in a battery remaining life prediction device based on sodium-ion batteries; it can also exist independently without being assembled into a battery remaining life prediction device based on sodium-ion batteries.

[0110] The above computer-readable storage medium carries one or more programs. When the one or more programs are executed by a battery remaining life prediction device based on sodium-ion batteries, the battery remaining life prediction device based on sodium-ion batteries is caused to: obtain target observation data of a target sodium-ion battery, where the target observation data at least includes current data, voltage data, and temperature data; determine a plurality of time steps according to the target observation data, and generate a target long time series according to the plurality of time steps, where a time step includes an observed value, a timestamp, a time interval, and an identifier; input the target long time series into a battery life prediction model to obtain the predicted remaining battery life of the target sodium-ion battery, where the battery life prediction model includes a dynamic self-attention encoding layer and a temporal attention decoding layer, the dynamic self-attention encoding layer includes a first multi-head self-attention mechanism and a first feed-forward network, the first feed-forward network includes a first fully-connected layer, a second fully-connected layer, and an activation function layer, and the temporal attention decoding layer includes a second multi-head self-attention mechanism and a second feed-forward network.

[0111] Computer program code for performing the operations of this application can be written in one or more programming languages or combinations thereof. The above-mentioned programming languages include object-oriented programming languages such as Java, Smalltalk, C++, and also include conventional procedural programming languages such as the "C" language or similar programming languages. The program code can be executed entirely on the user's computer, partially on the user's computer, executed as an independent 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 can be connected to the user's computer through any type of network, including a local area network (LAN) or a wide area network (WAN), or it can be connected to an external computer (for example, by using an Internet service provider to connect through the Internet).

[0112] The flowcharts and block diagrams in the accompanying drawings illustrate the possible architectures, functions, and operations of systems, methods, and computer program products according to various embodiments of this application. In this regard, each block in the flowchart or block diagram may represent a module, a program segment, or a part of code that contains one or more executable instructions for implementing a specified logical function. It should also be noted that in some alternative implementations, the functions marked in the blocks may occur in a different order than that marked in the accompanying drawings. For example, two consecutively represented blocks may actually be executed substantially in parallel, and they may sometimes be executed in the reverse order, depending on the functions involved. It should also be noted that each block in the block diagram and / or flowchart, as well as combinations of blocks in the block diagram and / or flowchart, can be implemented by a dedicated hardware-based system for performing the specified functions or operations, or can be implemented by a combination of dedicated hardware and computer instructions.

[0113] The modules involved in the embodiments described in this application can be implemented in software or in hardware. Among them, the name of the module does not constitute a limitation to the unit itself in some cases.

[0114] The readable storage medium provided in this application is a computer-readable storage medium. The computer-readable storage medium stores computer-readable program instructions (i.e., computer programs) for performing the above-mentioned battery remaining life prediction method based on sodium-ion batteries, and can solve the technical problems of battery remaining life prediction based on sodium-ion batteries. Compared with the prior art, the beneficial effects of the computer-readable storage medium provided in this application are the same as those of the battery remaining life prediction method based on sodium-ion batteries provided in the above embodiments, and will not be elaborated here.

[0115] The present application also provides a computer program product, including a computer program which, when executed by a processor, implements the steps of the battery remaining life prediction method based on a sodium-ion battery as described above.

[0116] The computer program product provided by the present application can solve the technical problems of low accuracy and reliability in implementing the life prediction of a sodium-ion battery by using a traditional method. Compared with the prior art, the beneficial effects of the computer program product provided by the present application are the same as those of the battery remaining life prediction method based on a sodium-ion battery provided in the above embodiments, and will not be elaborated here.

[0117] The foregoing are only partial embodiments of the present application, and thus do not limit the patent scope of the present application. Any equivalent structural transformation made under the technical concept of the present application by using the content of the specification and drawings of the present application, or direct / indirect application in other related technical fields, is included in the patent protection scope of the present application.

Claims

1. A method for predicting the remaining battery life of a sodium ion battery, characterized in that: The method comprises: Acquire target observation data of a target sodium ion battery, wherein the target observation data includes at least current data, voltage data, and temperature data; Determine a plurality of time steps according to the target observation data, and generate a target long time series according to the plurality of time steps, wherein the time step includes an observation value, a timestamp, a time interval and an identifier; Inputting the target long time series into a battery life prediction model to obtain a predicted battery remaining life of the target sodium-ion battery, wherein the battery life prediction model includes a dynamic self-attention encoding layer and a temporal attention decoding layer, the dynamic self-attention encoding layer includes a first multi-head self-attention mechanism and a first feedforward network, the first feedforward network includes a first fully connected layer, a second fully connected layer and an activation function layer, the temporal attention decoding layer includes a second multi-head self-attention mechanism and a second feedforward network, and the first feedforward network is used to learn nonlinear transformations and fit the nonlinear law of battery life; Before the step of inputting the target long time series into the battery life prediction model to obtain the predicted battery remaining life of the target sodium ion battery, the step further includes: Based on the training data in the sodium-ion battery system, the first battery life prediction model is trained by adopting the log-likelihood strategy of maximizing the observation probability to obtain the second battery life prediction model; Based on the self-supervised loss function of KL divergence, optimizing the second battery life prediction model to obtain the battery life prediction model; Determining a self-attention weight distribution of the battery life prediction model; Performing statistical analysis on the self-attention weight distribution to determine the mean and standard deviation of the time steps; Determining the importance of different time steps based on the mean and the standard deviation; Normalize the importance of each time step to obtain the normalized importance; After mapping the normalized importance to a heat map, determining a model prediction process according to the heat map; The battery life prediction model is optimized according to the model prediction process.

2. The method according to claim 1, characterized in that The step of inputting the target long time series into a battery life prediction model to obtain the predicted remaining battery life of the target sodium ion battery comprises: Inputting the target long time series into the dynamic self-attention encoding layer of the battery life prediction model to obtain self-attention features; The self-attention feature is input into the temporal attention decoding layer of the battery life prediction model to obtain the predicted remaining battery life of the target sodium-ion battery.

3. The method according to claim 1, characterized in that The first multi-head self-attention mechanism includes multiple attention heads and a fully connected layer; wherein the step of inputting the target long time series into the dynamic self-attention encoding layer of the battery life prediction model to obtain the self-attention feature includes: Performing parallel calculations on the target long time series through multiple attention heads of the first multi-head self-attention mechanism to obtain initial self-attention features; The initial self-attention feature is linearly transformed through the fully connected layer of the first multi-head self-attention mechanism to obtain a self-attention feature.

4. The method according to claim 1, characterized in that Before the step of inputting the target long time series into the battery life prediction model to obtain the predicted battery remaining life of the target sodium ion battery, the step further includes: Collect observation data of the sodium-ion battery under different working conditions from the sodium-ion battery system, and collect battery capacity change values ​​of the sodium-ion battery at different time intervals; Determine original data according to the observed data and the battery capacity change value; Performing a preprocessing operation on the raw data to obtain training data, and using the battery cycle number of the sodium ion battery as a label value of the training data, wherein the preprocessing operation includes a cleaning operation, an alignment operation, and a standardization operation; The battery life prediction model is constructed based on the training data and the label value.

5. The method according to claim 4, characterized in that After the step of constructing the battery life prediction model based on the training data and the label value, the following step is further included: Using test data to verify the battery life prediction model to obtain a verification result; Based on the verification result, the hyperparameters and network structure of the battery life prediction model are adjusted.

6. A battery remaining life prediction device based on sodium ion battery, characterized in that: The device comprises: An acquisition module, used to acquire target observation data of a target sodium-ion battery, wherein the target observation data includes at least current data, voltage data and temperature data; A determination module, configured to determine a plurality of time steps according to the target observation data, and generate a target long time series according to the plurality of time steps, wherein a time step includes an observation value, a timestamp, a time interval and an identifier; An input module, used to input the target long time series into a battery life prediction model to obtain the predicted remaining battery life of the target sodium-ion battery, wherein the battery life prediction model includes a dynamic self-attention encoding layer and a temporal attention decoding layer, the dynamic self-attention encoding layer includes a first multi-head self-attention mechanism and a first feedforward network, the first feedforward network includes a first fully connected layer, a second fully connected layer and an activation function layer, the temporal attention decoding layer includes a second multi-head self-attention mechanism and a second feedforward network, and the first feedforward network is used to learn nonlinear transformations and fit the nonlinear law of battery life; The input module is further used for: Based on the training data in the sodium-ion battery system, the first battery life prediction model is trained by adopting the log-likelihood strategy of maximizing the observation probability to obtain the second battery life prediction model; Based on the self-supervised loss function of KL divergence, optimizing the second battery life prediction model to obtain the battery life prediction model; Determining a self-attention weight distribution of the battery life prediction model; Performing statistical analysis on the self-attention weight distribution to determine the mean and standard deviation of the time steps; Determining the importance of different time steps based on the mean and the standard deviation; Normalize the importance of each time step to obtain the normalized importance; After mapping the normalized importance to a heat map, determining a model prediction process according to the heat map; The battery life prediction model is optimized according to the model prediction process.

7. A battery remaining life prediction device based on sodium ion batteries, characterized in that: The device comprises: a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the computer program is configured to implement the steps of the method for predicting the remaining battery life based on a sodium ion battery as described in any one of claims 1 to 5.

8. A storage medium, characterized in that: The storage medium is a computer-readable storage medium, and a computer program is stored on the storage medium. When the computer program is executed by a processor, the steps of the battery remaining life prediction method based on a sodium ion battery are implemented as described in any one of claims 1 to 5.

Images