Intelligent battery SOH real-time prediction method

Abstract

The invention provides an intelligent battery SOH real-time prediction method, and belongs to the field of battery health state prediction, and the method comprises the steps: obtaining the historical data of the health state of a battery; parameters required by the ISCO-GRU-OL dynamic model are set; based on an improved shark cooperation optimization algorithm and battery health state historical data, optimizing hyper-parameters of a gated circulation unit neural network; based on the optimized hyper-parameters, establishing and training an ISCO-GRU model in combination with historical battery health state data; performing parameter fine tuning on the ISCO-GRU model based on an online learning module according to the acquired latest battery health state data and the online learning frequency so as to update the ISCO-GRU model; future battery health state data are predicted based on the updated ISCO-GRU model, SOH real-time prediction can be accurately carried out, and a basis is provided for battery health management and remaining service life prediction.

Description

Technical Field

[0001] The present invention relates to the field of battery health status prediction, and in particular to an intelligent battery SOH real-time prediction method. Background Art

[0002] The state of health (SOH) of a battery is a key indicator for evaluating battery performance and life, reflecting the ratio of the battery's currently available capacity to its rated capacity. Accurate assessment of SOH is crucial for the effective operation of the battery management system, battery life prediction, and estimation of remaining service life, and will directly affect the reliability and safety of the battery. Existing SOH prediction methods face many challenges. On the one hand, there is a contradiction between model accuracy and real-time performance. Many high-precision SOH prediction models have high computational complexity and are difficult to meet the requirements of real-time prediction. On the other hand, although data-driven methods based on machine learning can fit complex nonlinear relationships, their generalization ability and robustness to noisy data need to be further improved. Overfitting and insufficient model training data are common problems.

[0003] Therefore, how to establish a dynamic real-time prediction model suitable for battery SOH prediction, while improving the model accuracy, real-time and robustness while minimizing the computational cost, is crucial for high-precision and high-efficiency battery SOH prediction. Summary of the invention

[0004] This invention proposes an intelligent battery SOH real-time prediction method, which combines the improved shark swarm optimization algorithm (ISCO), the gated recurrent unit (GRU) neural network and the online learning (OL) strategy. Through the powerful local convergence and global search capabilities of the ISCO algorithm, the efficient processing capability of the GRU network for time series data, and the ability of the online learning strategy to continuously update model parameters, the battery state of health (SOH) is dynamically predicted in real time. This invention aims to improve the accuracy, real-time and robustness of battery SOH prediction, and provide reliable technical support for dynamic real-time prediction of battery SOH.

[0005] The present invention is achieved in that: The technical solution to achieve the purpose of the present invention is: an intelligent battery SOH real-time prediction method, comprising the following steps: Step 1: Obtain battery health status (SOH) historical data; Step 2: Set the parameters required for the ISCO-GRU-OL dynamic model; The ISCO-GRU-OL dynamic model is composed of an ISCO-GRU model and an online learning module; the required parameters include parameters required for the improved shark cooperation optimization algorithm, parameters required for building the ISCO-GRU model and parameters required for the online learning module; Step 3: Based on the improved shark cooperation optimization algorithm and the historical data of battery health status, an optimization strategy based on sliding window flow training is adopted, and the root mean square error of the predicted value of the gated recurrent unit neural network under sliding window flow training is used as the fitness function to optimize the hyperparameters of the gated recurrent unit neural network; Step 4: Based on the optimized hyperparameters, the ISCO-GRU model is established and trained in combination with historical battery health status data; Step 5: Based on the latest battery health status data and online learning frequency F OL , fine-tune the parameters of the ISCO-GRU model based on the online learning module to obtain the ISCO-GRU model after parameter fine-tuning; Step 6: Based on the latest ISCO-GRU model and the latest battery health status data, predict future battery health status data.

[0006] Furthermore, in step 2, the parameters required for the improved shark cooperation optimization algorithm include the total number of individuals N, the proportion of white tip reef shark individuals to the total number of individuals P1 , the proportion of shark leaders to total individuals P2 , maximum number of iterations T ; The parameters required for the ISCO-GRU model include the ISCO-GRU model training window size W 1 , ISCO-GRU model training iterations EPO 1 , Sliding window flow training times WN , upper limit of hyperparameter optimization range ub and lower limit lb ; The parameters required for the online learning module include the online learning frequency F OL , online learning window size W 2 .

[0007] Furthermore, in step 3, the improved shark cooperation optimization algorithm includes two different types of individuals, white tip reef sharks and grey reef sharks, cooperating with each other, and its search step includes four stages: (1) initialization; (2) team hunting; (3) individual tracking; (4) local reinforcement and global exploration; (1) In the initialization phase, randomly generate N 1 White tip reef shark, N 1 The expression is: N 1 =round( N × P1 ) in N Represents the total number of individuals, P1 represents the proportion of white tip reef shark individuals to the total individuals, and the round() function represents rounding the value in the brackets to an integer; The initial position of each white tip reef shark was calculated as follows: M i =lb +( ub - lb )×rand(0,1) in, M i Representative i The location of individual white tip reef sharks, i =1,2,…N1, lb is the lower limit of the search domain, ub is the upper limit of the search domain, M i , lb , ub Both D dimensional vector, where D represents the dimension of the problem to be solved, and rand(0,1) represents a random number between 0 and 1. D dimensional vector; (2) In the team hunting phase, there are three sub-phases: (i) electing a leader; (ii) generating grey reef sharks and forming a team; (iii) team hunting; (i) Leader election sub-phase, calculation N 1 The fitness values ​​of the white tip reef sharks are selected. P White tip reef shark as leader: P =round( N × P2 ) in, P2 Represents the proportion of shark leaders to the total number of individuals; (ii) Generate grey reef sharks and form a team. Each leader generates a number of S i The leader and the gray reef sharks generated around the leader form a leader team. The number of gray reef sharks depends on the fitness value of the leader and the total number of gray reef sharks. i The number of Grey Reef Sharks that spawn when a Whitetip Reef Shark is the leader S i for: S i = round((1 / F(M i ) / F total × N ×(1- P1 )) in F total It is the sum of the reciprocals of the fitness values ​​of all leader whitetip reef sharks, as follows: ,when i When it belongs to a leader; in F () is the fitness function; After all leaders have generated grey reef sharks, determine the actual total number of grey reef sharks generated∑ S i With the expected Grey Reef Shark spawn total 1- P1 Is it consistent? If: ∑ S i > 1- P1 , then the number of gray reef sharks generated by the leader with the best fitness is increased by ∑ S i -(1- P1 ); ∑ S i = 1- P1 , then the number of grey reef sharks remains unchanged; ∑ S i < 1- P1 , then the number of gray reef sharks generated by the leader with the worst fitness is reduced by (1- P1 )-∑ S i , if the number of gray reef sharks generated by the leader with the worst fitness is less than (1- P1 )-∑ S i , then continue to reduce the number of gray reef sharks generated by the second worst fitness leader until ∑ S i = 1- P1 ; The position to spawn a Grey Reef Shark is calculated as follows: M ij = M i +1 / T ×( ub -lb )×(1- t / T ) j =1,2…, S i Among them M ij Indicates leader i The generated j Location of grey reef sharks; T is the maximum number of iterations, t is the current iteration number; if the position of the grey reef shark exceeds the search space boundary in any dimension, the value of that dimension will be adjusted to the nearest boundary value; (iii) Team hunting sub-phase, calculating the fitness value of each grey reef shark and updating the position of the whitetip reef shark to the position of the individual with the best fitness value in the team to which the whitetip reef shark belongs; (3) In the individual tracking phase, there are three sub-phases: (i) selecting the target leader; (ii) calculating the expected update point; (iii) position update; (i) In the target leader selection sub-stage, each whitetip reef shark will choose a specific leader to follow, and the probability of leader i being selected is R ( i ) is expressed as: ,when i When it belongs to a leader; in, e is the base of natural logarithms; (ii) Calculate the expected update point sub-stage. The expected update point of the whitetip reef shark is determined by three factors: individual tracking term, global tracking term and inertia term. If the whitetip reef shark j Choose a leader i As the target, then its expected update point M pj for: M pj =M j +AG j ×( dk1 ) j +(1 - AG j ) ×( dk2 ) j + ( dk3 ) j in, M j White tip reef shark jlocation, (dk1) j is the target tracking item, (dk2) j is a global tracking item, and (dk3) j is the inertia term, AG j is the iteration factor, and its expressions are: ( dk1 ) j = S max × F ij × AG j ×( M i -M j ) ( dk2 ) j = S max × F jbest × AG j )×( M i -M j ) ( dk3 ) j =0.2×( AG j ×( dk1 ) j +( 1- AG j ) ×( dk2 ) j ) × AG j AG j =e (-t / T) in F ij is the target tracking influence term, F jbest is the global tracking influence term, S max is the maximum step length, and the expressions are: F ij = 1-( F(M j ) - F(Mi ) ) / ( F(M) max - F(M) min ) F ibest = 1-( F(M j ) - F(M) min ) / ( F(M) max - F(M) min ) S max = ( ub - lb ) / ( N × T ) (1 / 2) in F(M) max and F(M) min Represent the best and worst fitness values ​​among all whitetip reef sharks, respectively; If the expected update point is beyond the search space boundary, the value of this dimension will be adjusted to the nearest boundary value; (iii) Position update sub-stage: first, evaluate the fitness value of the expected update point of each white-tip reef shark; second, for non-leader white-tip reef sharks, directly update their positions to the expected update point; third, for leader white-tip reef sharks, only when the fitness value of the expected update point position is better than the fitness value of the current position, will its position be updated to the expected update point; (4) In the local reinforcement and global exploration phases, for the white-tip reef shark with the worst fitness, its position is updated to the position of the second-best individual in the team of the leader white-tip reef shark with the best fitness; for the white-tip reef shark with the second-worst fitness, a new position is regenerated within the search domain according to the method of the initialization phase; After each iteration, the optimal position and fitness value are recorded and all grey reef shark individuals are deleted. The termination criteria are evaluated. If the criteria are met, the calculation is terminated and the optimal parameters and optimal fitness value are output. Otherwise, the calculations of stages (2) to (4) are repeated.

[0008] Furthermore, in step three, the gated recurrent unit neural network includes an input layer, a GRU layer including L2 regularization, a dropout layer, a Dense fully connected layer, and a regression layer using an Adam optimizer; the hyperparameters include the number of units of the gated recurrent unit model, the learning rate, the Dropout rate, and the L2 regularization strength.

[0009] Furthermore, in step 3, the optimization strategy of the sliding window flow training means that when optimizing the hyperparameters, each given set of hyperparameters will be optimized in a sliding window flow manner. WN training and prediction, and WN After the sliding window transfer training is completed, WN The root mean square error between the predicted value and the true value is calculated, and the root mean square error is used as the fitness function required to optimize the hyperparameters. WN is the number of sliding window flow training times.

[0010] Further, in step 5, the online learning module refers to each time the online learning frequency is obtained F OL The latest battery status data before the latest data W 2 The data is used as training input, and the latest data is used as training output. The ISCO-GRU model is further trained to achieve parameter fine-tuning. W 2 is the online learning window size.

[0011] The beneficial effects of the present invention are as follows: the present invention provides an intelligent real-time prediction method for battery SOH. Among them, the improved shark cooperative optimization algorithm (ISCO) has fewer parameters, stronger local search capabilities, and faster convergence speed than the original SCO algorithm; the ISCO-GRU model can fully learn the historical state information of the battery SOH to obtain an accurate prediction model; online learning (OL) can fine-tune the ISCO-GRU model and correct the prediction accuracy based on the latest battery SOH data. The present invention can provide a basis for accurate real-time prediction of SOH and battery health management and remaining service life prediction. BRIEF DESCRIPTION OF THE DRAWINGS

[0012] In order to more clearly illustrate the technical solutions of the embodiments of the present invention, the drawings required for use in the embodiments will be briefly introduced below. It should be understood that the following drawings only show certain embodiments of the present invention and therefore should not be regarded as limiting the scope. For ordinary technicians in this field, other related drawings can be obtained based on these drawings without paying creative work.

[0013] Figure 1 It is a flow chart of an intelligent battery SOH real-time prediction method provided by an embodiment of the present invention; Figure 2 is a calculation flow chart of an improved shark cooperation optimization algorithm (ISCO) provided in an embodiment of the present invention; Figure 3 It is a comparison chart of prediction results of the ISCO-GRU-OL method and the comparative method provided in the embodiment of the present invention; DETAILED DESCRIPTION

[0014] In order to make the purpose, technical solutions and advantages of the embodiments of the present invention clearer, the technical solutions in the embodiments of the present invention will be clearly and completely described below in combination with the implementation cases and drawings in the embodiments of the present invention. Obviously, the described embodiments are part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without making creative work belong to the scope of protection of the present invention. Therefore, the following detailed description of the embodiments of the present invention provided in the drawings is not intended to limit the scope of the invention claimed for protection, but merely represents the selected embodiments of the present invention. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without making creative work belong to the scope of protection of the present invention.

[0015] The method of the present invention is described below by taking a specific landslide displacement real-time prediction case as an example.

[0016] like Figure 1 , an intelligent battery SOH real-time prediction method, comprising the following steps: Step 1: Obtain battery health status (SOH) historical data; Step 2: Set the parameters required for the ISCO-GRU-OL dynamic model; The ISCO-GRU-OL dynamic model is composed of an ISCO-GRU model and an online learning module; the required parameters include parameters required for the improved shark cooperation optimization algorithm, parameters required for building the ISCO-GRU model and parameters required for the online learning module; Step 3: Based on the improved shark cooperation optimization algorithm and the historical data of battery health status, an optimization strategy based on sliding window flow training is adopted, and the root mean square error of the predicted value of the gated recurrent unit neural network under sliding window flow training is used as the fitness function to optimize the hyperparameters of the gated recurrent unit neural network; Step 4: Based on the optimized hyperparameters, the ISCO-GRU model is established and trained in combination with historical battery health status data; Step 5: Based on the latest battery health status data and online learning frequency F OL , fine-tune the parameters of the ISCO-GRU model based on the online learning module to obtain the ISCO-GRU model after parameter fine-tuning; Step 6: Based on the latest ISCO-GRU model and the latest battery health status data, predict future battery health status data.

[0017] Furthermore, in step 2, the parameters required for the improved shark cooperation optimization algorithm include the total number of individuals N, the proportion of white tip reef shark individuals to the total number of individuals P1 , the proportion of shark leaders to total individuals P2 , maximum number of iterations T ; The parameters required for the ISCO-GRU model include the ISCO-GRU model training window size W 1 , ISCO-GRU model training iterations EPO 1 , Sliding window flow training times WN , upper limit of hyperparameter optimization range ub and lower limit lb ; The parameters required for the online learning module include the online learning frequency F OL , online learning window size W 2 .

[0018] Furthermore, in step 3, the improved shark cooperation optimization algorithm includes two different types of individuals, white tip reef sharks and grey reef sharks, cooperating with each other, and its search step includes four stages: (1) initialization; (2) team hunting; (3) individual tracking; (4) local reinforcement and global exploration; (1) In the initialization phase, randomly generate N 1 White tip reef shark, N 1 The expression is: N 1 =round( N × P1 ) in N Represents the total number of individuals, P1 represents the proportion of white tip reef shark individuals to the total individuals, and the round() function represents rounding the value in the brackets to an integer; The initial position of each white tip reef shark was calculated as follows: M i =lb +( ub -lb )×rand(0,1) in, M i Representative i The location of individual white tip reef sharks, i =1,2,…N1, lb is the lower limit of the search domain, ub is the upper limit of the search domain, M i , lb , ub Both D dimensional vector, where D represents the dimension of the problem to be solved, and rand(0,1) represents a random number between 0 and 1. D dimensional vector; (2) In the team hunting phase, there are three sub-phases: (i) electing a leader; (ii) generating grey reef sharks and forming a team; (iii) team hunting; (i) Leader election sub-phase, calculation N 1 The fitness values ​​of the white tip reef sharks are selected. P White tip reef shark as leader: P = round( N × P2 ) in, P2 Represents the proportion of shark leaders to the total number of individuals; (ii) Generate grey reef sharks and form a team. Each leader generates a number of S i The leader and the gray reef sharks generated around the leader form a leader team. The number of gray reef sharks depends on the fitness value of the leader and the total number of gray reef sharks. i The number of Grey Reef Sharks that spawn when a Whitetip Reef Shark is the leader S i for: S i = round((1 / F(M i ) / F total × N ×(1- P1 )) in F total It is the sum of the reciprocals of the fitness values ​​of all leader whitetip reef sharks, as follows: ,when i When it belongs to a leader; inF () is the fitness function; After all leaders have generated grey reef sharks, determine the actual total number of grey reef sharks generated∑ S i With the expected Grey Reef Shark spawn total 1- P1 Is it consistent? If: ∑ S i > 1- P1 , then the number of gray reef sharks generated by the leader with the best fitness is increased by ∑ S i -(1- P1 ); ∑ S i = 1- P1 , then the number of grey reef sharks remains unchanged; ∑ S i < 1- P1 , then the number of gray reef sharks generated by the leader with the worst fitness is reduced by (1- P1 )-∑ S i , if the number of gray reef sharks generated by the leader with the worst fitness is less than (1- P1 )-∑ S i , then continue to reduce the number of gray reef sharks generated by the second worst fitness leader until ∑ S i = 1- P1 ; The position to spawn a Grey Reef Shark is calculated as follows: M ij = M i +1 / T ×( ub - lb )×(1- t / T ) j =1,2…, S i Among them M ij Indicates leader i The generated j Location of grey reef sharks; T is the maximum number of iterations, t is the current iteration number; if the position of the grey reef shark exceeds the search space boundary in any dimension, the value of that dimension will be adjusted to the nearest boundary value; (iii) Team hunting sub-phase, calculating the fitness value of each grey reef shark and updating the position of the whitetip reef shark to the position of the individual with the best fitness value in the team to which the whitetip reef shark belongs; (3) In the individual tracking phase, there are three sub-phases: (i) selecting the target leader; (ii) calculating the expected update point; (iii) position update; (i) In the target leader selection sub-stage, each whitetip reef shark will choose a specific leader to follow, and the probability of leader i being selected is R ( i ) is expressed as: ,when i When it belongs to a leader; in, e is the base of natural logarithms; (ii) Calculate the expected update point sub-stage. The expected update point of the whitetip reef shark is determined by three factors: individual tracking term, global tracking term and inertia term. If the whitetip reef shark j Choose a leader i As the target, then its expected update point M pj for: M pj =M j +AG j ×( dk1 ) j +(1 - AG j ) ×( dk2 ) j + ( dk3 ) j in, M j White tip reef shark j location, (dk1) j is the target tracking item, (dk2) j is a global tracking item, and (dk3) j is the inertia term, AG j is the iteration factor, and its expressions are: ( dk1 ) j = S max × F ij × AGj ×( M i -M j ) ( dk2 ) j = S max × F jbest × AG j )×( M i -M j ) ( dk3 ) j =0.2×( AG j ×( dk1 ) j +( 1- AG j ) ×( dk2 ) j ) × AG j AG j =e (-t / T) in F ij is the target tracking influence term, F jbest is the global tracking influence term, S max is the maximum step length, and the expressions are: F ij = 1-( F(M j ) - F(M i ) ) / ( F(M) max - F(M) min ) F ibest = 1-( F(M j ) - F(M) min ) / ( F(M) max - F(M) min ) S max = ( ub - lb ) / ( N × T ) (1 / 2) in F(M) max and F(M) min Represent the best and worst fitness values ​​among all whitetip reef sharks, respectively; If the expected update point is beyond the search space boundary, the value of this dimension will be adjusted to the nearest boundary value; (iii) Position update sub-stage: first, evaluate the fitness value of the expected update point of each white-tip reef shark; second, for non-leader white-tip reef sharks, directly update their positions to the expected update point; third, for leader white-tip reef sharks, only when the fitness value of the expected update point position is better than the fitness value of the current position, will its position be updated to the expected update point; (4) In the local reinforcement and global exploration phases, for the white-tip reef shark with the worst fitness, its position is updated to the position of the second-best individual in the team of the leader white-tip reef shark with the best fitness; for the white-tip reef shark with the second-worst fitness, a new position is regenerated within the search domain according to the method of the initialization phase; After each iteration, the optimal position and fitness value are recorded and all grey reef shark individuals are deleted. The termination criteria are evaluated. If the criteria are met, the calculation is terminated and the optimal parameters and optimal fitness value are output. Otherwise, the calculations of stages (2) to (4) are repeated.

[0019] Furthermore, in step three, the gated recurrent unit neural network includes an input layer, a GRU layer including L2 regularization, a dropout layer, a Dense fully connected layer, and a regression layer using an Adam optimizer; the hyperparameters include the number of units of the gated recurrent unit model, the learning rate, the Dropout rate, and the L2 regularization strength.

[0020] Furthermore, in step 3, the optimization strategy of the sliding window flow training means that when optimizing the hyperparameters, each given set of hyperparameters will be optimized in a sliding window flow manner. WN training and prediction, and WN After the sliding window transfer training is completed, WN The root mean square error between the predicted value and the true value is calculated, and the root mean square error is used as the fitness function required to optimize the hyperparameters. WN is the number of sliding window flow training times.

[0021] Further, in step 5, the online learning module refers to each time the online learning frequency is obtained F OL The latest battery status data before the latest data W 2 The data is used as training input, and the latest data is used as training output. The ISCO-GRU model is further trained to achieve parameter fine-tuning. W 2 is the online learning window size.

[0022] The test data comes from NASA's public lithium battery SOH test data. The experiment selected 166 data from them, of which the first 75 were used as historical data to train the ISCO-GRU model, and the next 91 data were gradually added to the historical data as the latest data, and the model was fine-tuned and predicted through online learning.

[0023] The specific implementation process is as follows: like Figure 1 , an intelligent battery SOH real-time prediction method, comprising the following steps: Step 1: Obtain battery health status (SOH) historical data; Step 2: Set the parameters required for the ISCO-GRU-OL dynamic model; The ISCO-GRU-OL dynamic model is composed of an ISCO-GRU model and an online learning module; the required parameters include parameters required for the improved shark cooperation optimization algorithm, parameters required for building the ISCO-GRU model and parameters required for the online learning module; Among them, the total number of individuals N=20, the proportion of white-tip reef shark individuals to the total individuals P1=0.7, the proportion of shark leaders to the total individuals P2=0.3, and the maximum number of iterations T=20; ISCO-GRU model training window size W1=70, ISCO-GRU model training iteration number EPO1=500, sliding window flow training number WN=5, hyperparameter optimization range upper limit ub=[200,0.1,0.5,0.01], lower limit lb= [20,1e-4,0.0,0.0], hyperparameters correspond to the number of units, learning rate, Dropout rate, and L2 regularization strength of the gated recurrent unit model; the parameters required for the online learning module include the online learning frequency F OL =1, online learning window size W2=50.

[0024] Step 3: Based on the improved shark cooperation optimization algorithm and the historical data of battery health status, an optimization strategy based on sliding window flow training is adopted, and the root mean square error of the predicted value of the gated recurrent unit neural network under sliding window flow training is used as the fitness function to optimize the hyperparameters of the gated recurrent unit neural network; Improve the calculation process of shark cooperation optimization algorithmFigure 2 shown.

[0025] Step 4: Based on the optimized hyperparameters, the ISCO-GRU model is established and trained in combination with historical battery health status data; Step 5: Based on the latest battery health status data and online learning frequency F OL , fine-tune the parameters of the ISCO-GRU model based on the online learning module to obtain the ISCO-GRU model after parameter fine-tuning; Step 6: Based on the latest ISCO-GRU model and the latest battery health status data, predict future battery health status data.

[0026] In order to verify the advantages of the method of the present invention, the prediction results of the GRU model, the ISCO-GRU model and the ISCO-GRU-OL model of the present invention were compared. The number of units, learning rate, dropout rate and L2 regularization strength of the GRU model were 100, 0.01, 0.25 and 0.005 respectively, and the other model parameters involved were consistent.

[0027] Figure 3 The prediction effects of the three models are compared. Figure 3 It can be seen that the ISCO-GRU-OL model of the present invention has obvious advantages over the comparison model.

[0028] In order to further explore the performance of each model, the mean absolute error (MAE), absolute percentage error (MAPE), root mean square error (RMSE) and goodness of fit R were used. 2 There are 4 indicators to quantitatively evaluate each model. Among them, the closer MAE, MAPE and RMSE are to 0, the better the effect is. 2 The closer it is to 1, the more accurate the prediction is. The evaluation results of each model are shown in Table 1:

[0029] As shown in Table 1, the ISCO-GRU-OL model has goodness of fit R 2 The four indicators reached 0.005, 0.648, 0.008, and 0.989 respectively, which are significantly better than the comparison model, fully demonstrating the advantages of the ISCO-GRU-OL model in battery SOH prediction. The present invention can accurately predict SOH in real time and provide a basis for battery health management and remaining service life prediction.

[0030] The above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. For those skilled in the art, the present invention may have various modifications and variations. Any modification, equivalent replacement, improvement, etc. made within the spirit and principle of the present invention shall be included in the protection scope of the present invention.

Claims

1. An intelligent battery SOH real-time prediction method, characterized in that: The following steps are involved: Step 1: Obtain battery health status historical data; Step 2: Set the parameters required for the ISCO-GRU-OL dynamic model; The ISCO-GRU-OL dynamic model is composed of an ISCO-GRU model and an online learning module; the required parameters include parameters required for the improved shark cooperation optimization algorithm, parameters required for building the ISCO-GRU model and parameters required for the online learning module; Step 3: Based on the improved shark cooperation optimization algorithm and the historical data of battery health status, an optimization strategy based on sliding window flow training is adopted, and the root mean square error of the predicted value of the gated recurrent unit neural network under sliding window flow training is used as the fitness function to optimize the hyperparameters of the gated recurrent unit neural network; Step 4: Based on the optimized hyperparameters, the ISCO-GRU model is established and trained in combination with historical battery health status data; Step 5: Based on the latest battery health status data and online learning frequency F OL , fine-tune the parameters of the ISCO-GRU model based on the online learning module to obtain the ISCO-GRU model after parameter fine-tuning; Step 6: Based on the latest ISCO-GRU model and the latest battery health status data, predict future battery health status data.

2. The intelligent battery SOH real-time prediction method according to claim 1, characterized in that: In step 2, the parameters required for the improved shark cooperation optimization algorithm include the total number of individuals N, the proportion of white tip reef shark individuals to the total number of individuals P1 , the proportion of shark leaders to total individuals P2 , maximum number of iterations T ; The parameters required for the ISCO-GRU model include the ISCO-GRU model training window size W 1 , ISCO-GRU model training iterations EPO 1 , Sliding window flow training times WN , upper limit of hyperparameter optimization range ub and lower limit lb ; The parameters required for the online learning module include the online learning frequency F OL , online learning window size W 2 .

3. The intelligent battery SOH real-time prediction method according to claim 1, characterized in that: In step 3, the improved shark cooperation optimization algorithm includes two different types of individuals, white tip reef sharks and grey reef sharks, cooperating with each other, and its search step includes four stages: (1) initialization; (2) team hunting; (3) individual tracking; (4) local reinforcement and global exploration; (1) In the initialization phase, randomly generate N 1 White tip reef shark, N 1 The expression is: N 1 =round( N × P1 ) in N represents the total number of individuals, P1 represents the proportion of white tip reef shark individuals to the total individuals, and the round() function represents rounding the value in the brackets to an integer; The initial position of each white tip reef shark was calculated as follows: M i =lb +( ub - lb )×rand(0,1) in, M i Representative i The location of individual white tip reef sharks, i =1,2,… N1 , lb is the lower limit of the search domain, ub is the upper limit of the search domain, M i , lb , ub Both D dimensional vector, where D represents the dimension of the problem to be solved, and rand(0,1) represents a random number between 0 and 1. D dimensional vector; (2) In the team hunting phase, there are three sub-phases: (i) electing a leader; (ii) generating grey reef sharks and forming a team; (iii) team hunting; (i) Leader election sub-phase, calculation N 1 The fitness values ​​of the white tip reef sharks are selected. P White tip reef shark as leader: P =round( N × P2 ) in, P2 Represents the proportion of shark leaders to the total number of individuals; (ii) Generate grey reef sharks and form a team. Each leader generates a number of S i The leader and the gray reef sharks generated around the leader form a leader team. The number of gray reef sharks depends on the fitness value of the leader and the total number of gray reef sharks. i The number of Grey Reef Sharks that spawn when a Whitetip Reef Shark is the leader S i for: S i = round((1 / F(M i ) / F total × N ×(1- P1 )) in F total It is the sum of the reciprocals of the fitness values ​​of all leader whitetip reef sharks, as follows: ,when i When it belongs to a leader; in F () is the fitness function; After all leaders have generated grey reef sharks, determine the actual total number of grey reef sharks generated∑ S i With the expected Grey Reef Shark spawn total 1- P1 Is it consistent if: ∑ S i > 1- P1 , then the number of gray reef sharks generated by the leader with the best fitness is increased by ∑ S i -(1- P1 ); ∑ S i = 1- P1 , then the number of grey reef sharks remains unchanged; ∑ S i < 1- P1 , then the number of gray reef sharks generated by the leader with the worst fitness is reduced by (1- P1 )-∑ S i , if the number of gray reef sharks generated by the leader with the worst fitness is less than (1- P1 )-∑ S i , then continue to reduce the number of gray reef sharks generated by the second worst fitness leader until ∑ S i = 1- P1 ; The position to spawn a Grey Reef Shark is calculated as follows: M ij = M i +1 / T ×( ub - lb )×(1- t / T ) j =1,2…, S i Among them M ij Indicates leader i The generated j Location of grey reef sharks; T is the maximum number of iterations, t is the current iteration number; if the position of the grey reef shark exceeds the search space boundary in any dimension, the value of that dimension will be adjusted to the nearest boundary value; (iii) Team hunting sub-phase, calculating the fitness value of each grey reef shark and updating the position of the whitetip reef shark to the position of the individual with the best fitness value in the team to which the whitetip reef shark belongs; (3) In the individual tracking phase, there are three sub-phases: (i) selecting the target leader; (ii) calculating the expected update point; (iii) position update; (i) In the target leader selection sub-phase, each white tip reef shark will choose a specific leader to follow. i Probability of being selected R ( i ) is expressed as: ,when i When it belongs to a leader; in, e is the base of natural logarithms; (ii) Calculate the expected update point sub-stage. The expected update point of the whitetip reef shark is determined by three factors: individual tracking term, global tracking term and inertia term. If the whitetip reef shark j Choose a leader i As the target, then its expected update point M pj for: M pj =M j +AG j ×( dk1 ) j +(1-AG j ) ×( dk2 ) j + ( dk3 ) j in, M j White tip reef shark j location, (dk1) j is the target tracking item, (dk2) j is a global tracking item, and (dk3) j is the inertia term, AG j is the iteration factor, and its expressions are: ( dk1 ) j = S max × F ij × AG j ×( M i -M j ) ( dk2 ) j = S max × F jbest × AG j )×( M i -M j ) ( dk3 ) j =0.2×( AG j ×( dk1 ) j +( 1- AG j ) ×( dk2 ) j ) × AG j AG j =e (-t / T) in F ij is the target tracking influence term, F jbest is the global tracking impact term, S max is the maximum step length, and the expressions are: F ij = 1-( F(M j )-F(M i ) ) / ( F(M) max -F(M) min ) F ibest = 1-( F(M j ) - F(M) min ) / ( F(M) max -F(M) min ) S max = ( ub - lb ) / ( N × T ) (1 / 2) in F(M) max and F(M) min Represent the best and worst fitness values ​​among all whitetip reef sharks, respectively; If the expected update point is beyond the search space boundary, the value of this dimension will be adjusted to the nearest boundary value; (iii) Position update sub-stage: first, evaluate the fitness value of the expected update point of each white-tip reef shark; second, for non-leader white-tip reef sharks, directly update their positions to the expected update point; third, for leader white-tip reef sharks, only when the fitness value of the expected update point position is better than the fitness value of the current position, will its position be updated to the expected update point; (4) In the local reinforcement and global exploration phases, for the white-tip reef shark with the worst fitness, its position is updated to the position of the second-best individual in the team of the leader white-tip reef shark with the best fitness; for the white-tip reef shark with the second-worst fitness, a new position is regenerated within the search domain according to the method of the initialization phase; After each iteration, the optimal position and fitness value are recorded and all grey reef shark individuals are deleted. The termination criteria are evaluated. If the criteria are met, the calculation is terminated and the optimal parameters and optimal fitness value are output. Otherwise, the calculations of stages (2) to (4) are repeated.

4. The intelligent battery SOH real-time prediction method according to claim 1, characterized in that: In step three, the gated recurrent unit neural network includes an input layer, a GRU layer including L2 regularization, a dropout layer, a Dense fully connected layer, and a regression layer using an Adam optimizer; the hyperparameters include the number of units of the gated recurrent unit model, the learning rate, the Dropout rate, and the L2 regularization strength.

5. The intelligent battery SOH real-time prediction method according to claim 1, characterized in that: In step 3, the optimization strategy of sliding window flow training refers to that when optimizing hyperparameters, each given set of hyperparameters will be optimized in a sliding window flow manner. WN training and prediction, and WN After the sliding window transfer training is completed, WN The root mean square error between the predicted value and the true value is calculated, and the root mean square error is used as the fitness function required to optimize the hyperparameters. WN is the number of sliding window flow training times.

6. The intelligent battery SOH real-time prediction method according to claim 1, characterized in that: In step 5, the online learning module refers to each time the online learning frequency is obtained F OL The latest battery status data before the latest data W 2 The data is used as training input, and the latest data is used as training output. The ISCO-GRU model is further trained to achieve parameter fine-tuning. W 2 is the online learning window size.

Images