Sodium-ion battery health state prediction method and system and medium

Abstract

The invention discloses a sodium ion battery health state prediction method and system and a medium. The prediction method comprises the following steps: extracting health features; evaluating the correlation between each health feature and the service life of the battery; optimizing the back-propagation neural network based on a parrot optimization algorithm; and calculating a mean absolute error, a root mean square error, a mean absolute percentage error and a mean square error, and evaluating the prediction performance of the sodium ion battery health state so as to predict the sodium ion battery health state. According to the method, a good initial weight and bias setting can be found for the BP neural network in a short time by using the rapid convergence characteristic of the parrot optimization algorithm, so that the training process of the BP neural network is accelerated. And the parrot optimization algorithm introduces population diversity and a variation strategy, so that the BP neural network is helped to avoid overfitting in the training process, and the generalization ability of the model is improved. The combination of the two can show better performance in the field of sodium ion battery health state prediction.

Description

Technical Field

[0001] The present invention belongs to the field of sodium-ion batteries, and particularly relates to a method, a system and a medium for predicting the state of health of a sodium-ion battery. Background Art

[0002] With the severity of environmental pollution and the shortage of energy, new energy technologies have been widely developed. Currently, lithium-ion batteries are widely used in electric vehicles, renewable energy systems, and consumer electronics. However, with the large-scale development and implementation of energy storage systems, the demand for lithium has increased rapidly, leading to an increasingly prominent lithium shortage. At the same time, safety is also an important issue restricting the application of lithium-ion batteries in large-scale energy storage systems. Compared with Li, Na has the advantages of high abundance, similar electrochemical properties, and mature extraction processes. The latest commercial lithium-ion batteries have an energy efficiency of approximately 95%. In contrast, the efficiency of sodium-ion batteries is slightly lower, approximately 90%, but from the perspective of safety, sodium-ion batteries are significantly superior to lithium-ion batteries. Therefore, sodium-ion batteries are considered another potential energy storage material and show great promise in renewable energy systems and some low-speed electric vehicles. However, after long-term cycling, capacity decay and power degradation can lead to the failure of the battery system's functions and even serious safety problems. In addition, due to the increase in internal resistance, the energy efficiency of the battery will decrease. When the efficiency drops to an unacceptable value, a warning signal should be sent to the system to request battery replacement. By studying the SOH of sodium-ion batteries, it is possible to evaluate whether the battery meets the operating standards, ensure the stable operation of the energy storage system, and reduce energy waste and safety risks. Therefore, accurately and timely evaluating the state of health of the SOH of sodium-ion batteries is crucial for battery management and the protection of energy storage devices.

[0003] The state of health (SOH) of a battery is an indicator that measures the degree of battery aging and is usually defined as the ratio of the current capacity to the initial capacity. Currently, a large amount of research work on SOH prediction has been carried out at home and abroad. Generally speaking, SOH estimation methods can be divided into five categories: direct evaluation methods, adaptive methods, data-driven methods, model-based methods, and fusion methods.

[0004] Direct evaluation methods include Coulomb counting method, open circuit voltage method and impedance spectroscopy method. Among them, the Coulomb counting method is a method based on the integration of electric charge during the charge and discharge cycles of the battery. By measuring the charge and discharge current of the battery over a certain period of time and integrating it over time, the actual capacity of the battery can be obtained, and thus its SOH can be calculated. This method is simple and direct, with low cost, but its main disadvantage is that the measurement accuracy depends on the accuracy and calibration frequency of the equipment. The open circuit voltage method is based on the correlation between the open circuit voltage of the battery and the SOC, and estimates by establishing the corresponding relationship between OCV and SOH. This method performs well under stable temperature and high measurement accuracy, but is greatly affected by the ambient temperature and requires frequent calibration. The impedance spectroscopy method uses electrochemical impedance spectroscopy technology to measure the impedance of the battery at different frequencies and establish the relationship between impedance and SOH. This method is suitable for high-precision measurements in a laboratory environment, but is limited by equipment complexity and environmental factors in actual vehicle applications.

[0005] Adaptive methods include dual sliding mode observer and Kalman filtering method. Among them, the dual sliding mode observer realizes the online estimation of the battery SOH by establishing a battery model and using the sliding mode control theory. This method has good robustness and adaptability, but the algorithm complexity is relatively high and it requires high computing resources. The Kalman filtering method realizes the online estimation of SOH by establishing a battery state space model and using the Kalman filtering algorithm to update the model parameters in real time. This method can effectively process noise data and improve the prediction accuracy, but the algorithm has a large amount of calculation and requires high computing power support.

[0006] Data-driven methods include machine learning algorithms and deep learning algorithms. Among them, machine learning algorithms include support vector machine (SVM), random forest (RF), etc. These methods establish a relationship model between battery aging characteristics and SOH by learning a large amount of historical data. Its advantages are that it can automatically learn laws from data and has high prediction accuracy; the disadvantages are that it requires a large amount of training data and has high requirements for the quality and representativeness of the data. Deep learning algorithms such as convolutional neural network (CNN) and long short-term memory network (LSTM). Deep learning algorithms can handle complex non-linear relationships and extract deep features of battery aging through a multi-level network structure. This method performs excellently on large datasets, but has high computational costs and poor interpretability.

[0007] Model-based methods include electrochemical models and equivalent circuit models. Among them, the electrochemical model details the internal chemical reaction process of the battery and predicts SOH by simulating processes such as lithium ion concentration changes and electrode degradation. This method has high physical accuracy and is suitable for precision analysis, but the model is complex and the amount of calculation is large. The equivalent circuit model simplifies the battery into a circuit model composed of resistors, capacitors and other components, and describes the battery behavior through circuit equations. ECM is easy to implement and calculate, and is suitable for online real-time monitoring, but the prediction accuracy is relatively low.

[0008] The fusion methods include model-model fusion, data-model fusion, and data-data fusion. Among them, model-model fusion combines an electrochemical model with a machine learning model, leveraging the high precision of the electrochemical model and the data-driven capabilities of the machine learning model. This method synthesizes the advantages of different models, improving the accuracy and robustness of predictions, but it is difficult to implement and requires a large amount of experimental data for support. Data-model fusion combines historical data with a physical model and corrects the model parameters through data-driven methods to improve the accuracy of the model. This method not only retains the physical meaning of the model but also has the flexibility of data-driven approaches, but it is necessary to solve the problem of effective fusion between data and the model. Data-data fusion integrates battery data from different sources (such as current, voltage, temperature, etc.) and conducts fusion analysis through machine learning algorithms. This method can comprehensively utilize various information to improve prediction accuracy, but it has high requirements for data preprocessing and fusion algorithms. Summary of the Invention

[0009] The object of the present invention is to overcome the above problems existing in the prior art and provide a method, system, and medium for predicting the state of health of a sodium-ion battery, which is a sodium-ion battery SOH prediction technology based on the PO optimization algorithm combined with a BP neural network model. This method measures parameters in the aging test, including: constant voltage charging time, the area enclosed by the constant voltage charging current-time curve, the peak value of the IC curve, the voltage value corresponding to the peak value of the IC curve, equal voltage charging rise time, equal voltage discharge time during discharge, the maximum temperature during discharge, constant current charging time, and the proportion of constant current charging time in the total charging time, a total of 9 health characteristics. Combine the Pearson correlation coefficient (PCC) and the grey relational grade (GRG) to find the characteristics highly correlated with the battery capacity. The selected characteristics are used as inputs into the PO-BP model for predicting the state of health (SOH) of the sodium-ion battery. The present invention is different from the traditional single-method battery state of health prediction mode. Among them, the PO optimization algorithm can widely explore the solution space by simulating the social behaviors of parrots, such as foraging, staying, communicating, and fear of strangers, which helps to avoid falling into local optimal solutions. Moreover, the PO optimization algorithm has a faster convergence speed compared with traditional optimization algorithms and can exhibit better robustness under different parameter settings. The BP neural network has strong information processing capabilities and can adapt to different tasks and problems. However, when dealing with large-scale data, the training time is long, and it is prone to falling into local optimal solutions, showing poor performance in finding the global optimal solution. Therefore, by combining the characteristics of both, the structure and parameters of the BP neural network are optimized with the parrot optimization algorithm, and a better network configuration is found through global search, thereby improving the prediction performance of the BP neural network and further achieving accurate prediction of the state of health of the sodium-ion battery.

[0010] To achieve the above technical objectives and effects, the present invention is implemented through the following technical solutions:

[0011] A method for predicting the state of health of a sodium-ion battery, comprising:

[0012] Extracting a number of health characteristics in the aging experiment of the sodium-ion battery to complete feature extraction;

[0013] Calculating the Pearson correlation coefficient and grey relational degree corresponding to different health characteristics, and evaluating the correlation between each health characteristic and the battery life, so as to select the health characteristics with high correlation with the battery life and low autocorrelation with other health characteristics as the model input of the backpropagation neural network;

[0014] Optimizing the backpropagation neural network based on the parrot optimization algorithm to improve the convergence speed and prediction accuracy of the backpropagation neural network;

[0015] Calculating the mean absolute error, root mean square error, mean absolute percentage error and mean square error, and evaluating the prediction performance of the state of health of the sodium-ion battery to predict the state of health of the sodium-ion battery.

[0016] Further, the number of health characteristics includes constant voltage charging time, area enclosed by the constant voltage charging current-time curve, peak value of the IC curve, voltage value corresponding to the peak value of the IC curve, equal voltage charging rise time, equal voltage discharge time, maximum temperature during discharge, constant current charging time, and the proportion of constant current charging time in the total charging time.

[0017] Further, the calculation formula of the Pearson correlation coefficient is as follows:

[0018]

[0019] where, F i (k) represents the value of the i-th feature in the k-th cycle, and RUL(k) is the remaining useful life in the k-th cycle;

[0020] The calculation formula of the grey relational degree is as follows:

[0021]

[0022] where, ε i (k) can be expressed as:

[0023]

[0024] where, ρ is the distinguishing coefficient, usually set to 0.5, and n represents the total number of charge and discharge cycles.

[0025] Further, the specific process of the parrot optimization algorithm includes:

[0026] Set the population size, maximum number of iterations, and upper and lower bounds of the search space to complete population initialization;

[0027] Randomly execute any one of the foraging behavior, staying behavior, communication behavior, and fear behavior of strangers at each iteration;

[0028] Judge whether the maximum number of iterations is reached to output the best solution.

[0029] Furthermore, the calculation formula for population initialization is as follows:

[0030]

[0031] where rand(0,1) represents a random number in the range [0,1], represents the position of the i-th parrot in the initial stage;

[0032] The calculation formula for foraging behavior is as follows:

[0033]

[0034] where, represents the updated position, represents the current position. X best represents the best position searched from initialization to the current, and can also represent the current position of the owner; Levy(dim) is the Levy distribution, used to describe the flight situation of the parrot; T represents the current number of iterations, represents the average position within the current population; represents the movement of the parrot relative to the position of the owner, represents the parrot further determining the orientation of the food by observing the positions of the entire population;

[0035] The calculation formula for staying behavior is as follows:

[0036]

[0037] where ones(1,dim) represents a vector of all 1s in dimension dim, X best ·Levy(dim) represents the process of flying towards the owner, and rand(0,1)·ones(1,dim) represents the process of randomly stopping at a certain part of the owner's body;

[0038] The calculation formula for communication behavior is as follows:

[0039]

[0040] where, represents the process of an individual joining the parrot group for communication, represents the process of an individual leaving the group after communication;

[0041] The calculation formula for the fear behavior towards strangers is as follows:

[0042]

[0043] Wherein, represents the process of the parrot reorienting and flying towards the owner, represents the process of the parrot moving away from strangers.

[0044] Furthermore, the backpropagation neural network includes:

[0045] An input layer for receiving raw data, with each neuron corresponding to a feature;

[0046] A hidden layer for extracting non-linear features of the data;

[0047] An output layer for generating the final prediction result.

[0048] Furthermore, optimizing the backpropagation neural network based on the parrot optimization algorithm specifically includes:

[0049] Completing population initialization, where each individual represents a set of weight and bias parameters of the backpropagation neural network;

[0050] Calculating the fitness value of each individual, that is, the prediction error of the backpropagation neural network under the current weights;

[0051] Judging whether the prediction error is lower than the threshold to confirm whether to terminate the iteration.

[0052] Furthermore, the calculation formula for the mean absolute error is as follows:

[0053]

[0054] The calculation formula for the root mean square error is as follows:

[0055]

[0056] The calculation formula for the mean absolute percentage error is as follows:

[0057]

[0058] The calculation formula for the mean square error is as follows:

[0059]

[0060] The present invention also provides a sodium-ion battery state of health prediction system, including:

[0061] A feature extraction module for extracting several health features in the sodium-ion battery aging experiment to complete feature extraction;

[0062] A correlation analysis module is used to calculate the Pearson correlation coefficient and grey relational grade corresponding to different health characteristics, and evaluate the correlation between each health characteristic and the battery life, so as to select health characteristics with high correlation with the battery life and low autocorrelation with other health characteristics as the model input of the backpropagation neural network;

[0063] An algorithm optimization module is used to optimize the backpropagation neural network based on the parrot optimization algorithm to improve the convergence speed and prediction accuracy of the backpropagation neural network;

[0064] A model evaluation module is used to calculate the mean absolute error, root mean square error, mean absolute percentage error and mean square error, and evaluate the prediction performance of the health state of the sodium-ion battery to predict the health state of the sodium-ion battery.

[0065] The present invention also provides a computer storage medium, on which a computer program is stored, and when the computer program is executed, the above prediction method is implemented.

[0066] The beneficial effects of the present invention are:

[0067] By utilizing the fast convergence characteristic of the parrot optimization algorithm, the present invention can find a better initial weight and bias setting for the BP neural network in a short time, thereby accelerating the training process of the BP neural network. And the parrot optimization algorithm helps the BP neural network avoid overfitting and improve the generalization ability of the model during the training process by introducing population diversity and mutation strategies. The combination of the two can show better performance in the field of predicting the health state of sodium-ion batteries. BRIEF DESCRIPTION OF THE DRAWINGS

[0068] The drawings described herein are used to provide a further understanding of the present invention, and constitute a part of this application. The illustrative embodiments of the present invention and their descriptions are used to explain the present invention and do not constitute an improper limitation of the present invention. In the drawings:

[0069] Figure 1 is a flowchart of the prediction method of the present invention;

[0070] Figure 2 is a schematic diagram of the parrot optimization algorithm in the present invention;

[0071] Figure 3 is a structure diagram of the backpropagation neural network in the present invention;

[0072] Figure 4 is a PO optimized BP structure diagram in the present invention;

[0073] Figure 5 is a block diagram of the prediction system structure in the present invention. DETAILED DESCRIPTION OF THE EMBODIMENTS

[0074] Next, in combination with the accompanying drawings in the embodiments of the present invention, the technical solutions in the embodiments of the present invention will be clearly and completely described. Obviously, the described embodiments are only a part of the embodiments of the present invention, rather than all the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those of ordinary skill in the art without creative efforts belong to the scope of protection of the present invention.

[0075] As Figure 1 shown, a method for predicting the state of health of a sodium-ion battery includes:

[0076] Step 1: Extract several health characteristics in the aging experiment of the sodium-ion battery to complete feature extraction.

[0077] The several health characteristics include constant voltage charging time, the area enclosed by the constant voltage charging current-time curve, the peak value of the IC curve, the voltage value corresponding to the peak value of the IC curve, equal voltage charging rise time, equal voltage discharge time during discharge, the maximum temperature during discharge, constant current charging time, and the proportion of constant current charging time in the total charging time, a total of 9 health characteristics.

[0078] Step 2: Calculate the Pearson correlation coefficient and grey relational degree corresponding to different health characteristics, and evaluate the correlation between each health characteristic and the battery life, so as to select health characteristics with high correlation with the battery life and low autocorrelation with other health characteristics as the model input of the backpropagation neural network.

[0079] Two common indicators are used to evaluate the correlation between health characteristics and battery life, namely the Pearson correlation coefficient (PCC) and the grey relational degree (GRG). Among them, PCC provides the direction and strength of linear correlation, while GRG provides a quantitative measurement of system evolution and is applicable to dynamic processes. Generally, the higher the correlation between the input feature and the target, the better the accuracy of the data-driven model.

[0080] Among them, the calculation formula of the Pearson correlation coefficient is as follows:

[0081]

[0082] Among them, F i (k) represents the value of the i-th feature in the k-th cycle, and RUL(k) is the remaining useful life in the k-th cycle;

[0083] The calculation formula of the grey relational degree is as follows:

[0084]

[0085] Among them, ε i (k) can be expressed as:

[0086]

[0087] Among them, ρ is the discrimination coefficient, usually set to 0.5, and n represents the total number of charge and discharge cycles.

[0088] Using the battery aging experiment data, the PCC and GRG with different characteristics can be calculated, the correlation between the characteristics and the battery life can be viewed, as well as the autocorrelation between the characteristics. Selecting the health characteristics with a high correlation with the battery life and a low autocorrelation with other health characteristics as the model input of the neural network can speed up the model operation speed and achieve faster convergence.

[0089] Step 3: Optimize the backpropagation neural network based on the parrot optimization algorithm to improve the convergence speed and prediction accuracy of the backpropagation neural network.

[0090] Such as Figure 2 shown, the specific process of the parrot optimization algorithm includes:

[0091] Set the population size, the maximum number of iterations, and the upper and lower bounds of the search space to complete the population initialization.

[0092] Among them, the population size is set to N, and the maximum number of iterations is Max iter , the search space is limited to lb (lower bound) and ub (upper bound), and the initialization formula of PO is as follows:

[0093]

[0094] Among them, rand(0,1) represents a random number in the range of [0,1], represents the position of the i-th parrot in the initial stage.

[0095] Randomly execute any one of the foraging behavior, staying behavior, communication behavior, and fear behavior towards strangers each time;

[0096] Judge whether the maximum number of iterations is reached to output the optimal solution.

[0097] In the foraging behavior, the parrot mainly estimates the approximate position of the food by observing the position of the food or the owner, and then flies towards the target. Therefore, the position movement follows the following formula:

[0098]

[0099] Among them, represents the updated position, represents the current position. X bestIt represents the best position searched so far since initialization and can also represent the current position of the owner; Levy(dim) is the Levy distribution, which is used to describe the flight of the parrot; T represents the current iteration number. It represents the average position within the current population. It represents that the parrot moves relative to the position of the owner. It represents that the parrot further determines the orientation of the food by observing the positions of the entire population.

[0100] The parrot is a highly social creature. Its staying behavior is to fly to any part of the owner's body and stay there for a period of time. This process can be expressed as:

[0101]

[0102] Among them, ones(1,dim) represents a vector of all 1s in dimension dim, and X best ·Levy(dim) represents the process of flying towards the owner, and rand(0,1)·ones(1,dim) represents the process of randomly staying at a certain part of the owner's body.

[0103] The parrot is a naturally social animal, characterized by close communication within the group. This communication behavior includes communication of flying towards the group and not flying towards the group. In PO, it is assumed that the probabilities of these two behaviors occurring are equal, and the average position of the current population is used to symbolize the center of the group. This process can be expressed as:

[0104]

[0105] Among them, It represents the process of an individual joining the parrot group for communication. It represents the process of an individual leaving the group after communication; both of these behaviors are feasible and can be achieved by P randomly generated within the range [0,1].

[0106] Generally speaking, birds show natural fear of strangers, and parrots are no exception. They keep a distance from unfamiliar people and look for a safe environment with the owner. This behavior can be expressed by the following formula:

[0107]

[0108] Among them, It represents the process of the parrot reorienting and flying towards the owner. It represents the process of the parrot moving away from strangers.

[0109] As Figure 3 shown, the backpropagation neural network includes:

[0110] The input layer is used to receive the original data, and each neuron corresponds to a feature;

[0111] The hidden layer is used to extract the non - linear features of the data;

[0112] The output layer is used to generate the final prediction result.

[0113] As Figure 4 shown, the specific steps of optimizing the back - propagation neural network based on the parrot optimization algorithm include:

[0114] Complete the population initialization, where each individual represents a set of weight and bias parameters of the back - propagation neural network;

[0115] Calculate the fitness value of each individual, that is, the prediction error of the back - propagation neural network under the current weights;

[0116] Judge whether the prediction error is lower than the threshold to confirm whether the iteration terminates.

[0117] Step 4: Calculate the mean absolute error (MAE), root mean square error (RMSE), mean absolute percentage error (MAPE) and mean square error (MSE), and evaluate the prediction performance of the state of health of the sodium - ion battery to predict the state of health of the sodium - ion battery.

[0118] Among them, the formula for calculating the mean absolute error is as follows:

[0119]

[0120] The formula for calculating the root mean square error is as follows:

[0121]

[0122] The formula for calculating the mean absolute percentage error is as follows:

[0123]

[0124] The formula for calculating the mean square error is as follows:

[0125]

[0126] When the MAE, RMSE, MAPE and MSE are closer to 0, the better the model predicts the SOH effect.

[0127] As Figure 5 shown, the second aspect of the present invention also provides a state - of - health prediction system for sodium - ion batteries, including:

[0128] The feature extraction module is used to extract several health features in the aging experiment of the sodium - ion battery to complete the feature extraction;

[0129] A correlation analysis module, which is used to calculate the Pearson correlation coefficient and grey relational degree corresponding to different health characteristics, and evaluate the correlation between each health characteristic and the battery life, so as to select the health characteristics with high correlation with the battery life and low autocorrelation between other health characteristics as the model input of the backpropagation neural network;

[0130] An algorithm optimization module, which is used to optimize the backpropagation neural network based on the parrot optimization algorithm to improve the convergence speed and prediction accuracy of the backpropagation neural network;

[0131] A model evaluation module, which is used to calculate the mean absolute error, root mean square error, mean absolute percentage error and mean square error, and evaluate the prediction performance of the health state of the sodium-ion battery, so as to predict the health state of the sodium-ion battery.

[0132] The third aspect of the present invention also provides a computer storage medium, on which a computer program is stored, and when the computer program is executed, the above method is implemented. The storage medium may include: various media such as USB flash drives, mobile hard disks, read-only memories (ROM), random access memories (RAM), magnetic disks or optical discs that can store program codes.

[0133] In the description of this specification, the descriptions with reference to the terms "one embodiment", "example", "specific example", etc. mean that the specific features, structures, materials or characteristics described in connection with the embodiment or example are included in at least one embodiment or example of the present invention. In this specification, the schematic representations of the above terms do not necessarily refer to the same embodiment or example. Moreover, the specific features, structures, materials or characteristics described can be combined in any one or more embodiments or examples in a suitable manner.

[0134] The above shows and describes the basic principles, main features and advantages of the present invention. Those skilled in the art of this industry should understand that the present invention is not limited by the above embodiments, and what is described in the above embodiments and the specification only illustrates the principle of the present invention. Without departing from the spirit and scope of the present invention, the present invention will have various changes and improvements, and these changes and improvements all fall within the scope of the present invention claimed.

Claims

1. A method for predicting the state of health of a sodium-ion battery, characterized in that, Including: Extract several health characteristics in the aging experiment of sodium-ion batteries to complete feature extraction; Calculate the Pearson correlation coefficient and grey relational degree corresponding to different health characteristics, and evaluate the correlation between each health characteristic and the battery life, so as to select the health characteristics with high correlation with the battery life and low autocorrelation with other health characteristics as the model input of the backpropagation neural network; Optimize the backpropagation neural network based on the parrot optimization algorithm to improve the convergence speed and prediction accuracy of the backpropagation neural network; Calculate the mean absolute error, root mean square error, mean absolute percentage error and mean square error, and evaluate the prediction performance of the health state of sodium-ion batteries to predict the health state of sodium-ion batteries.

2. The method for predicting the state of health of a sodium-ion battery according to claim 1, wherein: Several health characteristics include constant voltage charging time, area enclosed by the constant voltage charging current-time curve, peak value of the IC curve, voltage value corresponding to the peak value of the IC curve, equal voltage charging rise time, equal voltage discharge time during discharge, maximum temperature during discharge, constant current charging time, and the proportion of constant current charging time in the total charging time.

3. A method for predicting the state of health of a sodium-ion battery according to claim 1, characterized in that: The calculation formula of the Pearson correlation coefficient is as follows: Among them, F i (k) represents the value of the i-th feature in the k-th cycle, and RUL(k) is the remaining useful life in the k-th cycle; The calculation formula of the grey relational degree is as follows: where ε i (k) can be expressed as: Among them, ρ is the discrimination coefficient, usually set to 0.5, and n represents the total number of charge-discharge cycles.

4. A method for predicting the state of health of a sodium-ion battery according to any one of claims 1-3, characterized in that, The specific process of the parrot optimization algorithm includes: Set the population size, maximum number of iterations, upper and lower bounds of the search space to complete population initialization; Randomly execute any one of the foraging behavior, staying behavior, communication behavior, and fear behavior of strangers in each iteration; Judge whether the maximum number of iterations is reached to output the best solution.

5. A method for predicting the state of health of a sodium-ion battery according to claim 4, characterized in that, The calculation formula for population initialization is as follows: where rand(0,1) represents a random number within the range of [0,1], represents the position of the i-th parrot in the initial stage; The calculation formula for foraging behavior is as follows: Among them, represents the updated position, represents the current position. X best represents the best position searched so far from initialization, and can also represent the current position of the owner; Levy(dim) is the Levy distribution, used to describe the flight of the parrot; T represents the current iteration number, represents the average position within the current population; represents the movement of the parrot relative to the position of the owner, represents that the parrot further determines the direction of the food by observing the positions of the entire population; The calculation formula for staying behavior is as follows: Among them, ones(1,dim) represents a vector of all 1s in dimension dim, and X best ·Levy(dim) represents the process of flying towards the host, and rand(0,1)·ones(1,dim) represents the process of randomly stopping at a certain part of the host's body; The calculation formula for communication behavior is as follows: Among them, represents the process of an individual joining a parrot group for communication, represents the process of an individual leaving the group after communication; The calculation formula for fear behavior of strangers is as follows: Among them, represents the process of the parrot reorienting and flying towards the owner, represents the process of the parrot moving away from strangers.

6. A method for predicting the state of health of a sodium-ion battery according to claim 5, characterized in that: The backpropagation neural network includes: The input layer is used to receive the original data, and each neuron corresponds to a feature; The hidden layer is used to extract the non-linear features of the data; The output layer is used to generate the final prediction result.

7. A method for predicting the state of health of a sodium-ion battery according to claim 6, wherein, Optimizing the backpropagation neural network based on the parrot optimization algorithm specifically includes: Complete population initialization, and each individual represents a set of weight and bias parameters of the backpropagation neural network; Calculate the fitness value of each individual, that is, the prediction error of the backpropagation neural network under the current weights; Judge whether the prediction error is lower than the threshold to confirm whether the iteration terminates.

8. A method for predicting the state of health of a sodium-ion battery according to claim 7, characterized in that, The calculation formula for the mean absolute error is as follows: The calculation formula for the root mean square error is as follows: The calculation formula for the mean absolute percentage error is as follows: The calculation formula for the mean square error is as follows:

9. A sodium-ion battery state of health prediction system, characterized in that, Including: The feature extraction module is used to extract several health characteristics in the aging experiment of sodium-ion batteries to complete feature extraction; The correlation analysis module is used to calculate the Pearson correlation coefficient and grey relational degree corresponding to different health characteristics, and evaluate the correlation between each health characteristic and the battery life, so as to select the health characteristics with high correlation with the battery life and low autocorrelation with other health characteristics as the model input of the backpropagation neural network; The algorithm optimization module is used to optimize the backpropagation neural network based on the parrot optimization algorithm to improve the convergence speed and prediction accuracy of the backpropagation neural network; The model evaluation module is used to calculate the mean absolute error, root mean square error, mean absolute percentage error, and mean square error, and evaluate the prediction performance of the state of health of the sodium-ion battery to predict the state of health of the sodium-ion battery.

10. A computer storage medium having a computer program stored thereon, characterized in that: When the computer program is executed, it implements the prediction method according to any one of claims 1-8.

Images