Method for predicting residual capacity of vehicle-mounted lithium battery

Abstract

The invention discloses a method for predicting the residual capacity of a vehicle-mounted lithium battery, and the method comprises the steps: achieving the collection of battery data through a battery management system, analyzing the correlation of the multi-dimensional features of the data through a cloud server according to the data of a plurality of electric vehicle battery samples through a Pearson correlation coefficient algorithm, and screening out the strong features which greatly affect the capacity of the battery; performing standardization processing on the data; constructing an LSTM neural network prediction model according to the dimension of the input data; setting parameters of a chaos particle swarm optimization algorithm according to the neural network structure, and applying the optimal parameters obtained by the chaos particle swarm optimization algorithm to the LSTM neural network; and finally completing training and testing of the neural network model to obtain the LSTM neural network model capable of being used for prediction. According to the invention, the Pearson's correlation coefficient algorithm is utilized to carry out battery capacity correlation characteristic analysis, strong characteristics having great influence on the battery capacity are found out, and the accuracy of the model is improved; and the LSTM neural network model optimized by the chaos particle swarm optimization is used for prediction, so that the prediction precision is improved.

Description

Technical Field

[0001] The present invention belongs to the field of vehicle-mounted lithium battery capacity prediction, and specifically relates to a vehicle-mounted lithium battery remaining capacity prediction method based on LSTM neural network. Background Art

[0002] Car lithium batteries have the characteristics of high battery voltage, stable discharge, and long storage life, and can be used in fields such as vehicles that require high battery energy. However, in high temperature environments, the capacity characteristics of the battery will also decay to a certain extent, affecting the continuous use of the battery. Therefore, in order to understand the storage status of the battery, it is necessary to predict the battery capacity to determine the battery performance.

[0003] Currently, there are two methods for predicting battery capacity: model prediction and data-driven prediction. The main model-based prediction method is the equivalent circuit model method, which estimates the battery capacity based on the working history and condition of the power supply by treating the power supply as an equivalent circuit model. The prediction method of the equivalent circuit model is based on the data of the working status of the power supply. It first establishes an equivalent circuit model of the power supply, and then calculates the remaining life of the power supply, so that the battery capacity can be predicted more accurately. However, this method has some limitations, such as the accuracy of the model is highly dependent on the modeling accuracy, and the prediction of new power supplies requires additional model establishment.

[0004] Data-driven methods have become the mainstream, mainly relying on historical data for model training, analyzing and modeling historical battery data to predict the capacity of the power source. In the early days, statistical methods were mainly used for life prediction, such as the particle filter method, which used early cycle data to update the empirical formula for battery capacity decay. However, the use of statistical methods for battery capacity prediction depends on the selection of empirical formulas. If there is a large difference between the trend of the battery capacity decay curve and the empirical formula, the accuracy of life prediction will be greatly reduced. In addition, if the battery jumps during the capacity decay process, the statistical method for battery capacity prediction does not have good adaptability.

[0005] With the gradual development of electric vehicle technology, the functions and uses of battery management systems (BMS) have also been gradually expanded. The earliest battery management system only collected the primary measurement parameters of the battery (voltage, current, temperature, etc.), and then developed to measure and predict secondary parameters (SOC, internal resistance). This technology provides sufficient data premise for data-driven methods.

[0006] At present, the data-driven method is mainly based on deep learning, and the model is trained by using the working status data of the on-board lithium battery, including parameters such as voltage and current, to form a historical data set. The deep learning model can automatically learn and extract the characteristics of the data, improve the generalization ability of the model, and also capture the potential characteristics behind the data, which is of great significance for deepening the working mechanism of the battery. Summary of the invention

[0007] In order to predict the remaining capacity of a vehicle-mounted lithium battery and improve the accuracy of the prediction, the present invention proposes a method for predicting the remaining capacity of a battery based on deep learning optimized by a swarm intelligence algorithm, which effectively improves the accuracy of battery capacity prediction.

[0008] In order to achieve the above object, the method for predicting the remaining battery capacity provided by the present invention comprises the following steps:

[0009] Step 1: Use the battery management system to collect battery data, where the parameters include battery voltage, battery current, battery temperature, remaining capacity, discharge time, maximum discharge voltage, maximum charge voltage, constant current time and charge time for each battery cycle (battery cycle means that the battery undergoes a complete charging and discharging process), and then upload it to the cloud server. The cloud server uses the Pearson correlation coefficient algorithm to analyze the correlation of the multi-dimensional features of the battery data collected from multiple electric vehicle battery sample data, and screen out strong features that have a greater impact on battery capacity; let the total number of batteries be N, and the number of battery cycles of the jth battery sample data be n. j , then the Pearson correlation coefficient of the characteristic data of the j-th battery is as follows:

[0010]

[0011] Then calculate the average Pearson correlation coefficient of this feature

[0012]

[0013] In the formula, x i,j is the capacity data of the jth battery at the ith battery cycle, y i,j is the sampled data value of this feature of the jth battery and the i-th battery cycle number; for each feature, its r is calculated xy ;

[0014] Based on the Pearson correlation coefficient, the strong features that have a greater impact on battery capacity are screened out, including: discharge time, maximum discharge voltage, maximum charge voltage, constant current time and charge time;

[0015] Step 2: Standardize the sample data with the five features selected to obtain normalized data, and divide the data of N batteries into a training set and a test set in proportion;

[0016] Step 3: Set the initial input dimension of the LSTM neural network to 5 and the output dimension of the fully connected layer to 1, and build a neural network model for vehicle lithium battery capacity prediction;

[0017] Step 4: Use the chaotic particle swarm optimization algorithm to optimize the number of neurons, number of layers, learning rate and number of iterations of the LSTM neural network, and assign the obtained optimal hyperparameters to the LSTM neural network model;

[0018] Step 5: Use the training set data of step 2 to train the LSTM neural network model optimized in step 4, then use the test set data to predict the battery capacity of the on-board lithium battery for the trained LSTM neural network model, and finally calculate the evaluation index. The LSTM neural network model with qualified index is used to predict the battery capacity.

[0019] Among them, the method of optimizing the number of neurons, number of layers, learning rate and number of iterations using the chaotic particle swarm optimization algorithm in step 4 is as follows:

[0020] Step 4.1, Initialize the particle swarm: First, initialize a group of particles, each particle represents a set of solutions, that is, the combination of the number of neurons, the number of layers, the learning rate and the number of iterations; introduce the Tent chaos mapping into the particle swarm algorithm, and use the randomness of the chaotic variables to generate a uniformly distributed particle population in the solution space;

[0021] Step 4.2, calculate the fitness value corresponding to each particle: that is, use the number of neurons, number of layers, learning rate and number of iterations corresponding to the particle to train the LSTM model, and calculate the root mean square prediction error of the training set as the fitness value;

[0022] Step 4.3: Update individual optimal position and global optimal position: Update the individual optimal position and global optimal position of each particle according to the fitness value of each particle;

[0023] Step 4.4: Update the position and velocity of the particle: Update the position and velocity of the particle according to the individual optimal position and the global optimal position, as well as the current position and velocity;

[0024] Step 4.5: Determine whether the termination condition is met (the maximum number of iterations is reached or the fitness value has converged to the set value). If the termination condition has not been met, continue with steps 4.2-4.5 until the termination condition is met and output the optimal solution.

[0025] In step 4.1, the Tent chaos mapping formula is:

[0026]

[0027] In the formula, x n represents the nth iteration result of the chaotic mapping, x n+1 represents the n+1th iteration result of the chaotic mapping, a is the scale parameter, β is the nonlinear index, μ n is a random variable that follows a standard normal distribution, ε n is a noise term, which is a uniform random variable between [0,1];

[0028] The above Tent chaos map is used to generate the position of the initial particle population of the particle swarm algorithm. Through the chaotic properties of the Tent chaos map, the initial position of the diffusion particle in the search space is determined by the following formula:

[0029] POS ij =a+(ba)·x n+1

[0030] In the formula, pos ij is the j-th dimension position of the ith particle, and a and b are the lower and upper limits of the search dimension.

[0031] In step 4.4, the chaotic particle swarm optimization algorithm uses an adaptive function to determine the value of the inertia weight, which is as follows:

[0032]

[0033] The particle position update formula and velocity update formula of the algorithm are:

[0034] x i (t+1)=x i (t)+v i (t+1)

[0035] v i (t+1)=ω i (t)·v i (t)+c1·r1·(p i (t)-x i (t))+c2·r2·(p g (t)-x i (t))

[0036] In the formula, v i (t) represents the velocity of particle i at time t, x i (t) represents the position of particle i at time t, p i (t) represents the individual optimal position of particle i at time t, p g (t) represents the global optimal position, ω i(t) represents the inertia weight at time t, c1 and c2 are acceleration factors, r1 and r2 represent random number factors, ω max is the maximum inertia coefficient, ω min is the minimum inertia coefficient.

[0037] The beneficial effects of the present invention are:

[0038] 1. Considering the multi-dimensional characteristics of the battery, the present invention uses the Pearson correlation coefficient algorithm to analyze the relevant characteristics of the battery capacity, find strong features that have a greater impact on the battery capacity, avoid the input of redundant data, and thus improve the accuracy of the model.

[0039] 2. Normalize the original data to eliminate the differences in parameters between different batteries and avoid the impact of specific battery data input, thereby improving the stability and versatility of the prediction model.

[0040] 3. The present invention uses a chaotic particle swarm optimization algorithm to optimize the number of neurons, the number of layers, the learning rate and the number of iterations of the LSTM neural network, introduces the improved Tent chaotic map with normal distribution random variables, diffuses the initial position of particles in the search space, and uses an adaptive function to improve the inertia weight to determine the value of the inertia weight, thereby improving the search capability of the particle swarm algorithm and thus improving the prediction accuracy of the prediction model. BRIEF DESCRIPTION OF THE DRAWINGS

[0041] Figure 1 This is a flow chart of the vehicle battery capacity prediction method proposed by the present invention.

[0042] Figure 2 This is the battery management system BMS data acquisition framework used in the present invention.

[0043] Figure 3 The LSTM neural network structure diagram constructed for the present invention. DETAILED DESCRIPTION

[0044] like Figure 1 As shown, the present invention generally includes the following steps:

[0045] Step 1: First, determine the correlation between the battery's many parameters and battery capacity.

[0046] Use Figure 2 The battery management system collects battery data and then uploads it to the cloud server. Figure 2 The solid arrows in the figure represent data flow, and the dotted arrows represent power flow. The cloud server collects sample data of multiple electric vehicle batteries, and uses the Pearson correlation coefficient algorithm to analyze the correlation of the multi-dimensional features of the battery data, and selects strong features that have a greater impact on the battery capacity.

[0047] Step 2: Standardize the sample data of the features screened out in step 1 to obtain normalized data, and divide the battery sample data into training sets and test sets in proportion.

[0048] Step 3: Set the input dimension and output dimension of the LSTM neural network according to the characteristics of the input data.

[0049] Step 4: Use the chaotic particle swarm optimization algorithm to optimize the number of neurons, number of layers, learning rate, and number of iterations of the LSTM neural network, and assign the optimal hyperparameters to the LSTM model.

[0050] Step 5: Use the training set data to train the LSTM prediction model, and use the test set to verify the model effect. Finally, calculate the evaluation index of the model to obtain the LSTM neural network model that can be used for battery capacity prediction. During the real-time data collection process, in addition to updating the model, the prediction results of the battery capacity are also sent to the electric vehicle.

[0051] The present invention will be described in further detail below with reference to the accompanying drawings and specific embodiments.

[0052] In step 1, the Pearson correlation coefficient algorithm is used to analyze the correlation of the multi-dimensional features of the battery data. The formula is as follows:

[0053]

[0054] In the formula, is the Pearson correlation coefficient of a characteristic of the jth battery, r xy is the average Pearson correlation coefficient of this feature, x i,j is the capacity data of the jth battery at the ith battery cycle, y i,j is the sampled data value of this feature at the i-th battery cycle in the j-th battery. Calculate its r for each feature xy , such as considering relevant characteristics such as battery voltage, battery current, battery temperature, remaining capacity, discharge time, maximum discharge voltage, maximum charge voltage, constant current time and charge time.

[0055] The correlation degree standard corresponding to the Pearson correlation coefficient is shown in Table 1:

[0056] Table 1 Pearson correlation coefficient correlation degree standard

[0057]

[0058] The correlation coefficient between the battery capacity and its related factors is calculated according to the Pearson correlation coefficient algorithm. The correlation coefficient values ​​between each influencing factor (feature) and the battery capacity in the embodiment are shown in the following table:

[0059] Table 2 Correlation coefficient values ​​of influencing factors and battery capacity

[0060]

[0061] Only the correlation coefficient values ​​of the influencing factors with strong correlation and extremely strong correlation and the battery capacity are listed in Table 2. According to the correlation coefficient values ​​of the influencing factors, the influencing factors with higher coefficient values ​​are selected as the model input features. In the present invention, the influencing factors with correlation coefficients greater than 0.6 are selected, including the discharge time (seconds), the maximum discharge voltage (volts), the maximum charge voltage (volts), the constant current time (seconds) and the charging time (seconds) as the five characteristic sequence input models.

[0062] The discharge time refers to the time required for the battery to be fully discharged from a fully charged state;

[0063] The maximum discharge voltage refers to the maximum voltage value allowed by the battery during the discharge process;

[0064] The maximum charging voltage refers to the maximum voltage value allowed by the battery during the charging process;

[0065] Constant current time refers to the time that the battery operates at a constant current during charging or discharging;

[0066] Charging time refers to the time required for the battery to be charged from an empty state to a fully charged state.

[0067] After filtering out the input features, in step 2, an example of the historical data matrix P collected from one lithium battery is as follows.

[0068]

[0069] Among them, each row of the matrix P represents data with the same number of cycles, and each column represents an influencing factor.

[0070] Step 2: Standardize the collected data to eliminate the impact of the dimensional differences between the characteristic data of different battery influencing factors. The normalization formula is as follows:

[0071]

[0072] Where: P i,j represents the element in row i and column j in the historical data matrix, represents the element in row i and column j in the standardized historical data matrix, represents the maximum value of the j-th feature of matrix P, Represents the minimum value of the j-th feature of matrix P.

[0073] The normalized historical data is as follows.

[0074]

[0075] According to the screening results of step 1, the model input is determined to be five feature sequences: discharge time (seconds), maximum discharge voltage (volts), maximum charge voltage (volts), constant current time (seconds) and charge time (seconds). Therefore, in step 3, we set the input dimension of the LSTM neural network to 5 and the dimension of the fully connected layer output layer to 1.

[0076] Step 4: The present invention also optimizes the number of neurons, number of layers, learning rate and number of iterations of the LSTM neural network, specifically by using a chaotic particle swarm optimization algorithm for optimization. The chaotic particle swarm optimization algorithm is an improvement on the PSO (particle swarm) algorithm. In view of the weaknesses of the particle swarm algorithm, such as uneven initial population distribution and unbalanced global search and local search capabilities, chaotic mapping is introduced to improve the particle swarm algorithm. The method of using the chaotic particle swarm optimization algorithm to improve the number of neurons, number of layers, learning rate and number of iterations in the present invention is as follows:

[0077] Step 1: Initialize the particle swarm.

[0078] First, a group of particles are initialized, each particle represents a set of solutions, that is, a combination of the number of hidden layer neurons, the learning rate and the number of iterations. Initially, the position and velocity of each particle can be randomly generated. The present invention introduces Tent chaotic mapping, which uses the randomness of chaotic variables to generate a uniformly distributed particle population in the solution space. The improved Tent chaotic mapping formula with the introduction of normally distributed random variables can be expressed as:

[0079]

[0080] Where: x n represents the nth iteration result of the chaotic mapping, x n+1 represents the n+1th iteration result of the chaotic mapping, a is the scale parameter, β is the nonlinear index, μ n is a random variable that follows a standard normal distribution, ε n is a noise term, which is usually a uniform random variable between [0, 1]. n This term replaces the normal random uniform distribution between 0 and 1. At the same time, a noise term ε is added n Can be used to quickly remove high frequency noise in a map.

[0081] The Tent chaotic map improved by the above-mentioned normal distribution random variable is used to generate the position of the initial particle population of the particle swarm algorithm. Through the chaotic properties of the Tent chaotic map, the initial position of the diffusion particle in the search space is determined. The formula for the initial particle position is as follows:

[0082] POS ij=a+(ba)·x n+1 (5)

[0083] In the formula, pos ij is the j-th dimension position of the ith particle, and a and b are the lower and upper limits of the search dimension. n+1 It is the improved Tent chaos mapping formula with normal distribution random variables.

[0084] Step 2: Calculate the fitness value.

[0085] Calculate the fitness value corresponding to each particle, that is, use the number of neurons, number of layers, learning rate and number of iterations corresponding to the particle to train the LSTM model, and calculate the root mean square prediction error of the training set as the fitness value.

[0086] After setting the relevant parameters of the LSTM neural network, the objective function of the algorithm is determined to be the root mean square prediction error RMSE of the LSTM training set, as shown below

[0087]

[0088] Where T is the number of samples in the training set. t is the actual value of the training set, The predicted value of the training set is obtained by minimizing the fitness function to obtain the optimal number of neurons, number of layers, learning rate, and number of iterations of the LSTM neural network.

[0089] Step 3: Update the individual optimal position and the global optimal position.

[0090] According to the fitness value of each particle, the individual optimal position and global optimal position of each particle are updated. The individual optimal position is the position with the best fitness value in the history of the particle, and the global optimal position is the position with the best fitness value in the history of the entire particle group.

[0091] Step4: Update the position and velocity of the particles.

[0092] According to the individual optimal position and the global optimal position, as well as the current position and velocity, the position and velocity of the particle are updated. This step can be performed using the update formula of the standard PSO algorithm, as shown in Equations (8) and (9), which include the calculation of velocity update and position update.

[0093] For the inertia weight of the particle swarm algorithm, an adaptive function is used to improve the inertia weight to determine the value of the inertia weight. The formula of the inertia weight improved by the adaptive function is as follows:

[0094]

[0095] The particle position update formula and velocity update formula of the improved particle swarm algorithm are as follows:

[0096] x i (t+1)=x i (t)+v i (t+1) (8)

[0097] v i (t+1)=ω i (t)·v i (t)+c1·r1·(p i (t)-x i (t))+c2·r2·(p g (t)-x i (t)) (9)

[0098] In the formula, v i (t) represents the velocity of particle i at time t, v i (t+1) represents the velocity of particle i at time t+1, x i (t) represents the position of particle i at time t, x i (t+1) represents the position of particle i at time t+1, p i (t) represents the individual optimal position of particle i at time t, p g (t) represents the global optimal position, ω i (t) represents the inertia weight at time t, c1 and c2 are acceleration factors, r1 and r2 represent random number factors, ω max is the maximum inertia coefficient, ω min is the minimum inertia coefficient.

[0099] Step 5: Determine the termination condition.

[0100] Determine whether the termination condition is met, such as reaching the maximum number of iterations or the fitness value has converged to a certain extent. If the termination condition has not been met, continue with Step 2-5 until the termination condition is met.

[0101] The parameters of the chaotic particle swarm optimization algorithm include population size, learning factor, etc. The selection of such parameters affects the model prediction results. In the original PSO algorithm, the learning factors c1 and c2 are generally set to 2.0 or 1.49 according to the algorithm search requirements. In this embodiment, the selection of the learning factor and the population size is obtained according to the simulation experiment, and a set of parameters with the smallest error is selected. In this example, the learning factor is selected as 1.49 and the population size is set to 10.

[0102] The optimization results of step 4 are set as the parameters of the LSTM neural network to construct the LSTM prediction model. In step 2, the normalized data of N batteries are divided into training sets and test sets according to the proportion. In step 5, the LSTM prediction model is trained according to the training set. Finally, the trained LSTM prediction model is tested with the test set to obtain the trained LSTM neural network model for vehicle lithium battery capacity prediction.

[0103] In the embodiment, the battery capacity of the vehicle-mounted lithium battery is predicted and analyzed. In order to verify the effectiveness of the present invention, the prediction model proposed by the present invention is compared with the prediction model without the improved algorithm and the original model. The specific prediction models include: unimproved LSTM neural network (first row of Table 3), LSTM neural network optimized by the standard PSO algorithm (second row of Table 3) and LSTM neural network optimized by the chaotic particle swarm optimization algorithm of the present invention (third row of Table 3) prediction model. The evaluation absolute error (MAE) and mean absolute percentage error (MAPE) are used as evaluation indicators, and the calculation formula is as follows:

[0104]

[0105] The battery capacity prediction error is shown in the following table:

[0106] Table 3 Prediction error

[0107]

[0108] It can be seen that the prediction model of the present invention performs random perturbations on particles before the algorithm optimizes the model parameters, thereby improving the distribution of randomness and performing chaotic perturbations on the inertia weights, causing them to change from large to small. The obtained solution has obvious advantages over the standard PSO, and the model prediction effect is better.

[0109] In summary, the present invention uses chaotic reverse learning and chaotic mapping to improve and optimize the particle swarm algorithm, and is used to adjust the number of neurons, the number of layers, the learning rate and the number of iterations of the LSTM neural network, thereby improving the solving ability and accelerating the convergence speed, and can improve the accuracy of prediction.

Claims

1. A method for predicting the remaining capacity of a vehicle-mounted lithium battery, characterized in that: The following steps are involved: Step 1: Use the battery management system to collect battery data, including battery voltage, battery current, battery temperature, remaining capacity, discharge time, maximum discharge voltage, maximum charge voltage, constant current time and charge time of each battery cycle, and upload the data to the cloud server; the cloud server uses the Pearson correlation coefficient algorithm to analyze the correlation of the multi-dimensional features of the battery data for the sample data collected from multiple electric vehicle batteries, and screen out the strong features that have a greater impact on the battery capacity; set The total number of batteries is N, and the number of battery cycles of the j-th battery sample data is recorded as n j , then the Pearson correlation coefficient of the characteristic data of the j-th battery is as follows: Then calculate the average Pearson correlation coefficient of this feature In the formula, x i,j is the capacity data of the jth battery at the ith battery cycle, y i,j is the sampled data value of this feature of the jth battery and the i-th battery cycle number; for each feature, its r is calculated xy ; Based on the Pearson correlation coefficient, the strong features that have a greater impact on battery capacity are screened out, including: discharge time, maximum discharge voltage, maximum charge voltage, constant current time and charge time; Step 2: Standardize the sample data with the five features selected to obtain normalized data, and divide the data of N batteries into a training set and a test set in proportion; Step 3: Set the initial input dimension of the LSTM neural network to 5 and the output dimension of the fully connected layer to 1, and build a neural network model for predicting the remaining capacity of the vehicle lithium battery; Step 4: Use the chaotic particle swarm optimization algorithm to optimize the number of neurons, number of layers, learning rate and number of iterations of the LSTM neural network, and assign the obtained optimal hyperparameters to the LSTM neural network model; Step 5: Use the training set data of step 2 to train the LSTM neural network model optimized in step 4, then use the test set data to predict the battery capacity of the on-board lithium battery for the trained LSTM neural network model, and finally calculate the evaluation index. The LSTM neural network model with qualified index is used to predict the battery capacity.

2. The method for predicting the remaining capacity of a vehicle-mounted lithium battery as claimed in claim 1, wherein: Step 4: The method of using the chaotic particle swarm optimization algorithm to optimize the number of neurons, number of layers, learning rate and number of iterations is as follows: Step 4.1, Initialize the particle swarm: First, initialize a group of particles, each particle represents a set of solutions, that is, the combination of the number of neurons, the number of layers, the learning rate and the number of iterations; introduce the Tent chaos mapping into the particle swarm algorithm, and use the randomness of the chaotic variables to generate a uniformly distributed particle population in the solution space; Step 4.2, calculate the fitness value corresponding to each particle: that is, use the number of neurons, number of layers, learning rate and number of iterations corresponding to the particle to train the LSTM model, and calculate the root mean square prediction error of the training set as the fitness value; Step 4.3: Update individual optimal position and global optimal position: Update the individual optimal position and global optimal position of each particle according to the fitness value of each particle; Step 4.4: Update the position and velocity of the particle: Update the position and velocity of the particle according to the individual optimal position and the global optimal position, as well as the current position and velocity; Step 4.5: Determine whether the termination condition is met. If the termination condition is not met, continue with steps 4.2-4.5 until the termination condition is met and output the optimal solution.

3. The method for predicting the remaining capacity of a vehicle-mounted lithium battery as claimed in claim 2, wherein: In step 4.1, the Tent chaos mapping formula is: In the formula, x n represents the nth iteration result of the chaotic mapping, x n+1 represents the n+1th iteration result of the chaotic mapping, a is the scale parameter, β is the nonlinear index, μ n is a random variable that follows a standard normal distribution, ε n is a noise term, which is a uniform random variable between [0,1]; The above Tent chaos map is used to generate the position of the initial particle population of the particle swarm algorithm. Through the chaotic properties of the Tent chaos map, the initial position of the diffusion particle in the search space is determined by the following formula: pos ij =a+(b-a)·x n+1 In the formula, pos ij is the j-th dimension position of the ith particle, and a and b are the lower and upper limits of the search dimension.

4. The method for predicting the remaining capacity of a vehicle-mounted lithium battery as claimed in claim 2, wherein: In step 4.4, the chaotic particle swarm optimization algorithm uses an adaptive function to determine the value of the inertia weight, which is as follows: The particle position update formula and velocity update formula of the algorithm are: x i (t+1)=x i (t)+v i (t+1) v i (t+1)=ω i (t)·v i (t)+c1·r1·(p i (t)-x i (t))+c2·r2·(p g (t)-x i (t)) In the formula, v i (t) represents the velocity of particle i at time t, x i (t) represents the position of particle i at time t, p i (t) represents the individual optimal position of particle i at time t, p g (t) represents the global optimal position, ω i (t) represents the inertia weight at time t, c1 and c2 are acceleration factors, r1 and r2 represent random number factors, ω max is the maximum inertia coefficient, ω min is the minimum inertia coefficient.

5. The method for predicting the remaining capacity of a vehicle-mounted lithium battery as claimed in claim 2, wherein: The termination condition described in step 4.5 is that the maximum number of iterations is reached or the fitness value has converged to the set value.

Images