Lithium battery model parameter identification method based on improved balance optimization algorithm

Abstract

The invention discloses a lithium battery model parameter identification method based on an improved balance optimization algorithm. The method comprises the following steps: 1) acquiring actually measured current and actually measured voltage of a lithium battery; 2) establishing a lithium battery model, and taking the root-mean-square error of the estimated terminal voltage and the actually measured voltage of the model as a target function; 3) embedding a dynamic variation quasi-reverse learning strategy and an individual enhancement mechanism into a standard balance optimization algorithm to form an improved balance optimization algorithm; 4) using an improved balance optimization algorithm to minimize the objective function, identifying unknown parameters of the lithium battery model, obtaining optimal parameters and forming a mathematical model; 5) performing current and voltage data fitting according to the obtained optimal parameters and models; the method has the advantages of being high in convergence speed, easy to jump out of local optimum and high in search precision, and the identification precision of the lithium battery model parameters can be effectively improved.

Description

Technical Field

[0001] The present invention belongs to the technical field of new energy, and particularly relates to a method for identifying lithium battery model parameters by improving a balance optimization algorithm. Background Technique

[0002] The lithium battery model is an important research content in the field of battery technology. With the rapid development of the electric vehicle industry, as the preferred supporting power battery, the performance and safety requirements of lithium batteries are getting higher and higher. In order to more accurately reflect the lithium battery circuit model of battery characteristics and improve the energy management system of electric vehicles, it is very meaningful to identify the parameters of the lithium battery model.

[0003] At present, the methods for identifying lithium battery model parameters are roughly divided into three categories: analytical method, numerical calculation method, and optimization algorithm estimation method. The analytical method can achieve fast solution, but the problem of poor solution fitness has not been well solved. Especially, the approximate processing in this method will reduce the accuracy of the solution. The numerical calculation method relies too much on the selection of the initial value. If the error is too large, the initial value has to be reselected each time, and then its convergence is observed. The error in the solution process will also increase with the increase of the identified parameters. The optimization algorithm has the advantages of few constraint conditions and strong ability to identify non-linearity, and has been widely used in the parameter identification of battery models. However, most of the classical intelligent optimization algorithms have problems such as slow convergence speed, low convergence accuracy, and easy to fall into local optimum, resulting in low overall identification accuracy. The balance optimization algorithm is a new type of meta-heuristic optimization algorithm, which realizes the search of the algorithm by simulating the state of dynamic mass balance inside the control volume. However, in non-linear optimization problems, it still has the defects of low optimization efficiency and easy to fall into local optimum. Summary of the Invention

[0004] To solve the problems existing in the prior art, the present invention proposes an improved balance optimization algorithm. It adopts dynamic mutation quasi-opposite learning and individual enhancement mechanism, and applies it to the parameter identification of the lithium battery model to obtain the optimal parameters. The lithium battery model designed by the present invention is closer to the ideal battery model.

[0005] A method for identifying lithium battery model parameters based on an improved balance optimization algorithm according to the present invention includes the following steps:

[0006] 1) Obtain the measured current and measured voltage of the lithium battery;

[0007] 2) Establish a lithium battery model, and take the root mean square error between the model estimated terminal voltage and the measured voltage as the objective function;

[0008] 3) Embed the dynamic mutation quasi-opposite learning strategy and individual enhancement mechanism into the standard balance optimization algorithm to form an improved balance optimization algorithm;

[0009] 4) Use the improved balance optimization algorithm to minimize the objective function, identify the unknown parameters of the lithium battery model, obtain the optimal parameters and form a mathematical model;

[0010] 5) Fit the current and voltage data based on the obtained optimal parameters and model.

[0011] The expression of the lithium battery third-order RC model in the lithium battery model parameter identification method based on the improved balance optimization algorithm is as follows:

[0012]

[0013] In the lithium battery third-order RC model, U k is the model output voltage, I is the model input current, U oc is the open circuit voltage, R0 is the internal resistance, R1, R2 and R3 are the resistances in the RC circuit respectively, C1, C2 and C3 are the capacitances in the RC circuit respectively, and U1, U2 and U3 are the terminal voltages of the RC circuit respectively.

[0014] The objective function is the root mean square error (RMSE) between the measured voltage and the model output voltage. The formula of the root mean square error function RMSE(x) is as follows, where N represents the number of sample data.

[0015]

[0016] F(V,I,x) is calculated using the following formula:

[0017]

[0018] Lithium battery model parameter identification is to find the minimum value of the objective function by finding different parameters.

[0019] The process of finding the optimal parameters of the improved balance optimization algorithm of the lithium battery model parameter identification method based on the improved balance optimization algorithm is as follows:

[0020] The algorithm starts the optimization process by initializing the population, and the particles in the population are randomly initialized in the search space to build an initial concentration.

[0021] The initial concentration of each particle is as follows:

[0022]

[0023] in represents the initial concentration of the i-th particle in the population, Ub and Lb represent the maximum and minimum values of particles in the search space, rand is a random vector in the range [0, 1], and N represents the number of particles.

[0024] To improve the global search ability of the algorithm and avoid falling into low-quality local optimal solutions, after the initialization stage is completed, the particle concentration generated is evaluated and four particles with the smallest fitness values are selected to prepare for the formation of the balance pool.

[0025] The equilibrium state is the ideal convergence state of the algorithm. Therefore, at the beginning of the optimization process, the balance pool and candidate solutions are determined to provide a search pattern for the optimization of particles. The balance pool provides candidate solutions during the algorithm optimization process. It consists of four particles with relatively optimal fitness values and an average particle generated during the current initialization stage. Its definition is as follows:

[0026]

[0027] Among them, C eq_1 , C eq_2 , C eq_3 and C eq_4 represent the four particles with the highest concentrations selected after the completion of the algorithm initialization stage. C eq_ave represents the average particle, and C eq_pool represents the balance pool. In the balance pool, the first four particles help improve the exploration ability of the algorithm, while the average particle plays an important role in the development stage of the algorithm.

[0028] During the iterative process of the algorithm, each particle is selected with equal probability from the five candidate particles in the balance pool, which helps generate the global optimal solution. The mathematical formula is as follows:

[0029] C eq = randi(C eq_pool ) (6)

[0030] The exponential term is an important indicator for the exploration and development capabilities of the balance optimization algorithm. The calculation of the exponential term is described as follows:

[0031] F = a1·sign·(r - 0.5)·[e -λt - 1] (7)

[0032] Among them, a1 is a constant that controls the exploration ability of the algorithm. The higher the value of a1, the better the exploration ability of the algorithm. r and λ represent vectors in the interval [0, 1], and t is a coefficient updated with the number of iterations. The formula is as follows:

[0033]

[0034] Among them, it represents the current iteration number of the algorithm, and T represents the maximum iteration number of the algorithm. a2 is a constant that controls the development ability of the algorithm, and its value is 2. To improve the development ability of the algorithm, an equally important indicator is the generation rate G, and its definition is as follows:

[0035] G = Gep·C eq -λ·C·F (9)

[0036] Where G eq represents the generation rate control parameter vector, and C represents the current particle concentration. After the EO algorithm finishes the concentration update stage, the update formula for each particle is as follows:

[0037]

[0038] Where V represents the unit volume. According to the above description, the update rule of the equilibrium optimization algorithm is to construct the initial concentration of each particle in the initialization stage, select the four particles with the highest concentration from them, and form an equilibrium pool with the four particles and an average particle, which provides candidate solutions for the optimization process of the algorithm. Subsequently, the concentration of each particle is calculated using two important indicators, the exponential term and the generation rate G.

[0039] In the described method for identifying lithium battery model parameters based on an improved equilibrium optimization algorithm, the algorithm often faces problems of individual aggregation and insufficient population diversity during the initialization of the population and the optimization process. This phenomenon not only limits the search ability of the algorithm but also may lead to a decrease in the quality of the solution, thereby affecting the optimization effect. To solve these key problems, a dynamic mutation quasi-opposite learning strategy is used. The core goal of this strategy is to expand the search area of the population and effectively enhance the diversity of the population by incorporating more diverse individuals. The specific formula is as follows:

[0040]

[0041] Where Ub and Lb are the upper and lower bounds of the problem respectively, C is the position of the current individual, and C mqo is the mutated opposite solution. The elite mechanism is used to ensure that the position of the mutated opposite solution is better than the original position.

[0042]

[0043] In addition, an individual enhancement strategy is adopted. This strategy helps each individual move in a better direction in each iteration by enhancing the adaptability of the individual, thereby avoiding falling into local optima. In this way, the solution space can be better explored, making the algorithm more efficient and robust when searching for the global optimal solution. The application of this combined strategy promotes the optimization efficiency and information sharing, thus improving the overall optimization effect. The specific process is as follows:

[0044] Randomly select three solutions: C1, C2, C3 and calculate the average position of the three solutions. The formula is as follows:

[0045]

[0046] Randomly select a solution from the algorithm balance pool. Let C ave and C eq_pool perform crossover on the randomly selected solutions to generate a new crossover solution C p , C p contains the information of the optimal position at a random position and the solution. The formula is as follows:

[0047] C p = r·C ave + (1 - r)·C eq_rand (14)

[0048] Use C p and C ave to generate a new solution C new The formula is as follows:

[0049]

[0050] where r and w are random numbers in [0, 1], is an integer of -1 or 1. BRIEF DESCRIPTION OF THE DRAWINGS

[0051] Figure 1 is the drawing of the abstract of the specification of the present invention;

[0052] Figure 2 is the structure diagram of the three - order RC model of the lithium battery of the present invention;

[0053] Figure 3 is the specific flowchart of the parameter identification of the lithium battery model of the improved balance optimization algorithm of the present invention;

[0054] Figure 4 is the model voltage characteristic curve of the lithium battery model fitting of the improved balance optimization algorithm of the present invention;

[0055] Figure 5 is the error between the model voltage and the measured voltage of the lithium battery model fitting of the improved balance optimization algorithm of the present invention; DETAILED DESCRIPTION OF THE INVENTION

[0056] Next, the technical solutions in the embodiments of the present invention will be clearly and completely described in conjunction with the drawings in the embodiments of the present invention. The described embodiments are only a part of the embodiments of the present invention, rather than all of 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.

[0057] A technical solution provided by the present invention: A method for identifying the parameters of a lithium battery model based on an improved balance optimization algorithm, including the following steps:

[0058] 1) Obtain the lithium-ion battery charge and discharge sampling data under different working conditions through the detection unit, including sampling time, actual current, actual voltage, and test environment temperature;

[0059] 2) Establish an equivalent circuit third-order RC lithium battery model, and use the root mean square function of the terminal voltage of the lithium battery predicted by the third-order RC model and the actually measured terminal voltage as the objective function for optimization;

[0060] 3) Embed the dynamic mutation quasi-oppositional learning strategy and the individual enhancement mechanism into the standard equilibrium optimization algorithm to form an improved equilibrium optimization algorithm;

[0061] 4) Use the improved equilibrium optimization algorithm to minimize the objective function, identify the unknown parameters of the lithium battery model, obtain the optimal parameters and form a mathematical model;

[0062] 5) According to the optimal model parameters obtained, use this model to fit the current and voltage data of the lithium battery.

[0063] The model parameters identified by this method have high accuracy and can better reflect the relationship between the current and the terminal voltage during the charge and discharge of the lithium battery, and have broad application prospects in the field of lithium battery modeling.

Claims

1. A method for identifying the parameters of a lithium battery model based on an improved equilibrium optimization algorithm, characterized by the following steps: 1) Obtain the measured current and the corresponding measured voltage during the charge and discharge of the lithium battery to form sample data; 2) Establish a lithium battery model, and use the root mean square error sum between the model estimated terminal voltage and the measured voltage as the objective function; 3) Embed the dynamic mutation quasi-opposite learning strategy and the individual enhancement mechanism into the standard equilibrium optimization algorithm to form an improved equilibrium optimization algorithm; 4) Use the improved equilibrium optimization algorithm to minimize the objective function, identify the unknown parameters of the lithium battery model, obtain the optimal parameters and form a mathematical model; 5) Predict the terminal voltage under the charge and discharge current (or predict the charge and discharge current of any terminal voltage) according to the obtained optimal parameters and model.

2. The method for identifying the parameters of a lithium battery model based on an improved balance optimization algorithm according to claim 1, wherein For the establishment of the third-order RC model of the lithium battery and the determination of the objective function using the measured data, the specific steps are as follows: 1) Establish a third-order RC model of the lithium battery based on the measured current and measured voltage, and determine the parameters to be identified; 2) By using the measured data of the lithium battery, establish the root mean square error function RMSE(x), and use it as the optimization objective function.

3. A method for identifying lithium battery model parameters based on an improved balance optimization algorithm according to claim 1, characterized in that The expression of the third-order RC model of the lithium battery is: In the described third-order RC model of the lithium battery, U k is the model output voltage, I is the model input current, and U oc is the open-circuit voltage, R0 is the internal resistance, R1, R2, and R3 are the resistances in the RC circuits respectively, C1, C2, and C3 are the capacitances in the RC circuits respectively, U1, U2, and U3 are the terminal voltages of the RC circuits respectively, and U 1k-1 , U 2k-1 , and U 3k-1 are the voltages of U1, U2, and U3 at the previous moment respectively, and I k-1 is the current of I at the previous moment.

4. A method for identifying lithium battery model parameters based on an improved balance optimization algorithm according to claim 1, characterized in that The calculation formula of the root mean square error function RMSE(x) is: Where N represents the number of sample data; the calculation formula of F(V, I, x) is: Among them, the seven parameters of the internal resistance R0, the resistances R1, R2, and R3 in the RC circuit, and the capacitances C1, C2, and C3 are the parameters to be identified.

5. A method for identifying lithium battery model parameters based on an improved balance optimization algorithm according to claim 1, characterized in that The specific steps for improving the equilibrium optimization algorithm using the dynamic mutation quasi-opposite learning strategy and the individual enhancement mechanism are as follows: 1) Set the population size N, the maximum number of iterations T of the algorithm, and the search space of seven parameters to be identified; and initialize the relevant parameters F, t, G according to the formula ep , G; where r and λ represent random vectors within [0, 1], it is the current iteration number, r1 and r2 represent random numbers within [0, 1], and u is a unit vector; 2) Randomly initialize the population individuals; 3) Calculate and update the dynamic mutation opposite solution of the individual according to the formula, and the update formula is: 4) Calculate the fitness value of each individual, select the individual with the best fitness and the average value of four individuals to construct the equilibrium pool; 5) According to the calculated F, t, G ep , and update the position of the individual using the position update formula; 6) Judge whether the individual enhancement mechanism is satisfied, and update the position of the individual according to the individual enhancement mechanism formula; 7) Update parameters F, t, G ep , and G; 8) Judge whether the maximum number of iterations T is reached. If it is reached, go to step 9); otherwise, go to step 3); 9) Output the position of the best individual and its objective function value, which are the parameter identification results of the optimal lithium battery model obtained and the root mean square error between the measured voltage and the model output voltage.

6. A method for identifying lithium battery model parameters based on an improved balance optimization algorithm according to claim 1, characterized in that: The measured current and the measured voltage are acquired through a measurement unit for obtaining the measured current and measured voltage of the lithium battery under different working conditions; and the objective function is optimized through the output current and output voltage, and the optimal parameters of the mathematical model of the lithium battery model are extracted.

Images