A method for estimating and equalizing the state of charge of lithium-ion batteries

Abstract

This invention belongs to the field of battery energy consumption control technology, solving the problems of inconvenient state of charge (SOC) estimation and low energy utilization efficiency in new energy battery vehicles. It provides a method for estimating and equalizing the SOC of lithium-ion batteries. The method establishes an equivalent circuit model of the battery and introduces the state variable of battery SOC. It establishes a battery SOC estimation model based on a BP neural network, outputting the estimated SOC value. It establishes a particle swarm optimization algorithm based on battery energy path optimization, using an inertia weight adjustment strategy based on fitness function values ​​to find the optimal energy transfer path. Finally, it controls the current output through a fuzzy controller to obtain the optimal equalization current. This invention utilizes a BP neural network to estimate the battery SOC, optimizes the battery energy control path through an improved particle swarm optimization algorithm, and adjusts the equalization current using a fuzzy control algorithm, thereby improving the energy utilization rate of new energy battery vehicles and mitigating the problem of battery inconsistency.

Description

Technical Field

[0001] This invention belongs to the field of battery energy consumption control technology, specifically relating to a method for estimating and balancing the state of charge of lithium-ion batteries. Background Technology

[0002] The battery packs of electric vehicles typically consist of multiple individual cells connected in series or parallel to meet the voltage, power, and energy requirements of the vehicle. However, individual cells can exhibit inconsistencies due to differences in manufacturing and usage processes, thus affecting the driving range and safety of the electric vehicle.

[0003] State of Charge (SOC), a key parameter describing the actual usable electricity and evaluating the performance of electric vehicles, is the primary basis for determining the operating state of lithium-ion batteries. The SOC of lithium-ion batteries cannot be directly measured by instruments and is easily affected by internal and external factors such as charge / discharge rate, ambient temperature, and cycle count. It can only be estimated using the main external parameters of lithium-ion batteries, such as current and voltage, but a reasonable and effective method for estimating SOC is lacking. Furthermore, there is a lack of effective control methods for the utilization and transfer of energy in new energy battery vehicles. Summary of the Invention

[0004] In order to solve at least one of the above-mentioned technical problems in the prior art, the present invention provides a method for estimating and equalizing the state of charge of lithium-ion batteries.

[0005] This invention is achieved using the following technical solution: a method for estimating and equalizing the state of charge of a lithium-ion battery, comprising the following steps:

[0006] S1: Establish an equivalent circuit model of the battery and introduce the state variable, battery state of charge.

[0007] S2: Establish a battery state of charge estimation model based on BP neural network, collect physical quantities related to battery state of charge as input to the battery state of charge estimation model, and output the estimated value of battery state of charge.

[0008] S3: Establish a particle swarm optimization algorithm based on battery energy path optimization. Using an inertia weight adjustment strategy based on fitness function values, dynamically adjust the inertia weights of the particle swarm algorithm at different stages to find the optimal path for battery energy transfer; where the fitness function value... The fitness function value for the overall charging energy transfer path of the battery. The expression is:

[0009]

[0010] In the formula, , For the first Battery saving , for The average state of charge of the battery; The times correspond to Section 1, Section 2, Section 3, ..., Section 4 respectively. Save battery;

[0011] S4: The current output is controlled by a fuzzy controller to obtain the optimal equalization current; the input to the fuzzy controller used for equalization current regulation is the average value of the overall state of charge of the battery. The difference between the states of charge of two adjacent batteries The output of the fuzzy controller is the optimal equalization current. .

[0012] Preferably, in step S1, the formula for the battery state of charge is expressed as:

[0013]

[0014] In the formula, This represents the initial value of the battery's state of charge. Indicates battery charge / discharge efficiency. Indicates the battery operating current. Indicates the battery's nominal capacity;

[0015] The initial value of the battery state of charge, the battery charge and discharge efficiency, the battery operating current, and the battery nominal capacity are all used as inputs to the battery state of charge estimation model in step S2. The functional relationship between the input and output of the hidden layer and output layer nodes of the battery state of charge estimation model is the Sigmoid operation function.

[0016] Preferably, step S3 includes the following steps:

[0017] S31: Randomly generate S different particles and set the optimization range of the initial particles;

[0018] S32: Calculate and record the fitness value of each particle, and take the minimum fitness value of each particle as the optimal fitness value for that individual. And record the position of each particle;

[0019] S33: Find the particle with the lowest fitness value in the population and use its fitness value as the global optimal fitness value. The position of this particle is the global optimum; and the average fitness value of particles in the population is calculated. ;

[0020] S34: Improve the inertia weight parameter adjustment formula and calculate the inertia weight;

[0021] S35: Substitute the inertial weight from step S34 into the update formula for particle velocity and position to update the particle velocity and position.

[0022] S36: Determine the velocity and position of the particles during the iteration process to ensure that the particles iterate within a preset range. If the iteration accuracy reaches the preset condition, stop the iteration; otherwise, go to step S32 and continue the iteration.

[0023] S37: After iteration, the optimal solution for the battery energy path is obtained.

[0024] Preferably, in step S34, the improved inertia weight parameter adjustment formula is:

[0025]

[0026] In the formula, Represents inertial parameters, This represents the minimum value of the inertial parameter. This represents the maximum value of the inertial parameter; This represents the average fitness value of the population. This represents the maximum value of the entire battery's state of charge. This represents the average value of the overall state of charge of the battery. This represents the number of iterations in the particle swarm optimization algorithm. This indicates the maximum number of iterations.

[0027] Preferably, in step S35, the update formulas for particle velocity and position are as follows:

[0028]

[0029] In the formula, Indicates being between Random numbers between; Represents inertial parameters, , Indicates the learning factor; Indicates the current position of the particle; This represents the optimal position of a particle in a particle swarm. Indicates the first The velocity of each particle; Indicates the first The position of each particle; Indicates the corresponding time.

[0030] Compared with the prior art, the beneficial effects of the present invention are:

[0031] This invention uses a BP neural network to estimate the battery state of charge, optimizes the battery energy control path by improving the particle swarm optimization algorithm, and adjusts the equalization current by using a fuzzy control algorithm, which can improve the energy utilization rate of new energy battery vehicles and improve the problem of battery inconsistency. Attached Figure Description

[0032] To more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the drawings used in the embodiments will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0033] Figure 1 This is a battery equivalent circuit model diagram of the present invention;

[0034] Figure 2 This is a diagram of the artificial neuron structure and the BP neural network structure of the present invention;

[0035] Figure 3 This is a diagram illustrating the battery state-of-charge estimation process of the present invention;

[0036] Figure 4 This is a flowchart of the particle swarm optimization algorithm of the present invention. Detailed Implementation

[0037] The technical solutions of the embodiments of the present invention will be clearly and completely described with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other implementation methods obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0038] It should be noted that the structures, proportions, sizes, etc., shown in the accompanying drawings of this specification are only for the purpose of assisting those skilled in the art in understanding and reading the content disclosed in the specification, and are not intended to limit the conditions under which the present invention can be implemented. Therefore, they have no substantial technical significance. Any modifications to the structure, changes in the proportional relationships, or adjustments to the size, without affecting the effects and objectives that the present invention can produce, should fall within the scope of the technical content disclosed in the present invention. It should be noted that in this specification, relational terms such as "first" and "second" are only used to distinguish one entity from several other entities, and do not necessarily require or imply any actual relationship or order between these entities.

[0039] This invention provides an embodiment:

[0040] A method for estimating and equalizing the state of charge of a lithium-ion battery includes the following steps:

[0041] S1: Establish an equivalent circuit model of the battery and introduce the state variable, battery state of charge.

[0042] S2: Establish a battery state of charge estimation model based on BP neural network, collect physical quantities related to battery state of charge as input to the battery state of charge estimation model, and output the estimated value of battery state of charge.

[0043] S3: Establish a particle swarm optimization algorithm based on battery energy path optimization, and dynamically adjust the inertia weight of the particle swarm algorithm at different stages based on the fitness function value to find the optimal path for battery energy transfer.

[0044] S4: The current output is controlled by a fuzzy controller to obtain the optimal balanced current.

[0045] In step S1, the equivalent circuit model of the battery is established as follows: Figure 1 As shown in the figure Indicates the open-circuit voltage of the battery This refers to the potential difference between the positive and negative terminals of the battery when it is not powered on. Indicates the ohmic resistance of the battery; and Parallel connection simulates the long-term electrochemical polarization process of a battery; and The parallel simulation of the short-term concentration polarization process of the battery; Indicates the battery's operating voltage; This indicates the battery's operating current.

[0046] Due to the electrical characteristics of the battery equivalent circuit model, a state variable—the battery state of charge—is introduced. The formula is expressed as:

[0047]

[0048] In the formula, This represents the initial value of the battery's state of charge. Indicates battery charge / discharge efficiency. Indicates the battery operating current. This indicates the battery's nominal capacity.

[0049] based on Battery balancing can effectively reduce battery inconsistencies. However, the state of charge (SOC) of a battery cannot be directly obtained through external measurements and can only be estimated using external characteristic parameters. Therefore, a backpropagation (BP) neural network algorithm can be used to estimate the SOC, and the obtained SOC can be used as a balancing variable for battery pack balancing control.

[0050] Training of commonly used neural network models such as backpropagation (BP) neural networks. Figure 2 As shown, the process consists of two steps: 1. Forward propagation of the signal: The input data of the model undergoes a nonlinear transformation to obtain the output of each node, and the output value of the model is calculated. 2. Backward propagation of the error: If the error between the output value of the model and the output signal of the training samples is large and fails to reach the set target value, the error of each node needs to be calculated layer by layer, and the weights and thresholds of each node are adjusted. After multiple iterations of the above two processes, when the error converges to the expected value, the training of the neural network model is completed.

[0051] like Figure 3 As shown, this application uses a BP neural network algorithm to estimate the battery's state of charge (SOC). The input values ​​of the BP neural network are the initial SOC, the battery operating current, the battery charge / discharge efficiency, and the battery's nominal capacity. The Sigmoid function is chosen as the activation function for the neural network system, and the specific formula is as follows:

[0052]

[0053] The training process of a BP neural network is as follows: input training samples, calculate the output of the hidden layers, calculate the output of the output layer, calculate the deviation between the output value and the expected value. If the deviation does not meet the requirements, adjust the weights and thresholds between the neural network layers. By continuously adjusting the weights, the error is minimized, thereby outputting the optimal value. These are all existing technologies and will not be elaborated upon here.

[0054] Particle Swarm Optimization (PSO) is a global optimization algorithm based on a swarm search strategy. It mimics the prey behavior of birds to find the global optimum. PSO uses randomized initial values, where each particle has its own flight direction and velocity. During initialization, PSO uses a set of particles to search within a feasible range, and then all particles follow the overall optimal particle to obtain the optimal solution. In each iteration, each particle updates its tracking of two extreme values, one of which is the current particle's optimal solution. pBest "and another is the overall optimal solution " gBest The particle swarm optimization algorithm is prone to getting stuck in local optima during the iteration process, which can lead to premature convergence of the optimization process.

[0055] This application proposes an improved particle swarm optimization algorithm: based on the current particle fitness function value, the maximum value of the overall particle fitness function and the average value of the particle fitness function, a novel inertia weight adjustment strategy based on the fitness function value is proposed. This strategy enables the particle swarm to dynamically adjust the inertia weight at different stages during iterative optimization, effectively reducing the number of iterations.

[0056] The calculation process for the battery pack's balanced energy path is as follows:

[0057] The average state of charge of n batteries for:

[0058]

[0059] No. Battery saving and average The difference between for:

[0060]

[0061] when When it is positive, the first The battery needs to discharge, when When it is negative, the first The battery needs to be charged.

[0062] No. Battery saving The amount of data transferred is:

[0063]

[0064] fitness function value This represents the overall energy transfer path for battery charging.

[0065] Inertia weight coefficient It is a crucial parameter affecting particle flight speed and regulating algorithm convergence. In the initial stage of particle swarm optimization, With a larger value, it has strong global optimization ability and can quickly approach the optimal solution; in the later stages of iteration, The value is relatively small, and it is slowly adjusted near the optimal value until stability is achieved. Experiments show that dynamic inertia weights can achieve better optimization results than fixed inertia weights.

[0066] The method for adjusting the inertia weight is as follows: Let the population size be... During the iteration process, the single cell The fitness function value is The average fitness value of the population is , This represents the maximum value of the entire battery's state of charge. Therefore, the formula for adjusting the inertia weight parameter is:

[0067]

[0068] In the formula, Represents inertial parameters, This represents the number of iterations in the particle swarm optimization algorithm. Indicates the maximum number of iterations; This represents the minimum value of the inertial parameter. This represents the maximum value of the inertial parameter; This represents the average fitness value of the population. Indicates the first in the population The fitness value of each particle.

[0069] The steps to improve the particle swarm optimization algorithm are as follows: Figure 4 As shown, when the particle swarm optimization algorithm reaches a set accuracy, it achieves the optimal output. The specific steps of the improved particle swarm optimization algorithm are as follows:

[0070] S31: Randomly generate S different particles and set the optimization range of the initial particles;

[0071] S32: Calculate and record the fitness value of each particle, and take the minimum fitness value of each particle as the optimal fitness value for that individual. And record the position of each particle;

[0072] S33: Find the particle with the lowest fitness value in the population and use its fitness value as the global optimal fitness value. The position of this particle is the global optimum; and the average fitness value of particles in the population is calculated. ;

[0073] S34: Improve the inertia weight parameter adjustment formula and calculate the inertia weight;

[0074] S35: Substitute the inertial weight from step S34 into the update formula for particle velocity and position to update the particle velocity and position.

[0075] The formulas for updating particle velocity and position are:

[0076]

[0077] In the formula, Indicates being between Random numbers between, Represents inertial parameters, , Indicates the learning factor; Indicates the current position of the particle; This represents the optimal position of a particle in a particle swarm. Indicates the velocity of the particle; Indicates the position of the particle; Indicates the corresponding time.

[0078] S36: Determine the velocity and position of the particles during the iteration process to ensure that the particles iterate within a preset range. If the iteration accuracy reaches the preset condition, stop the iteration; otherwise, go to step S32 and continue the iteration.

[0079] S37: After iteration, the optimal solution for the battery energy path is obtained.

[0080] A lithium battery model can be equivalent to a combination of individual cells, where a single cell... The values ​​are different, and energy will have different paths when the battery is charging. In order to avoid repeated charging and discharging of the battery and reduce energy loss, we can consider optimizing the charging path of the battery. We can use an improved particle swarm optimization algorithm to optimize the energy path of the battery charging process and find the optimal path for energy transfer.

[0081] Fuzzy control is a control method that uses control laws (control rules) described by fuzzy mathematical language to manipulate the operation of a system. This method first codifies the experience of operators or experts into fuzzy rules, then fuzzifies the real-time signals from sensors, uses the fuzzified signals as input to the fuzzy rules to complete fuzzy inference, and finally adds the output obtained after inference to the actuator. The design of a fuzzy controller for equal current regulation involves the following steps: variable selection, fuzzification, rule base establishment, and defuzzification.

[0082] The balanced current regulation fuzzy controller designed in this application has two input variables and one output variable. The input variable is the average value of the overall state of charge of the battery. The difference in state of charge between two adjacent cells ,pass Determine whether the battery has reached the safety threshold for balance.

[0083] When the average value of the overall state of charge of the battery When the equilibrium target is reached, a high current is used for charging; when the average value of the overall state of charge of the battery is... To avoid overcharging or over-discharging, which could lead to excessively high battery temperatures, thermal imbalance, or even explosion, a smaller current should be selected for battery balancing when the current is too high or too low. When the current is small, a small current should be selected for battery balancing; when the difference in state of charge between two adjacent batteries is small... When the current is large, use a high current for battery balancing to shorten the balancing time.

[0084] The fuzzy controller utilizes the average value of the overall state of charge of the battery. The difference between the states of charge of two adjacent batteries To obtain the optimal equalization current By using fuzzification and establishing a rule base, the equalization current of the battery pack can be adjusted.

[0085] The above description is merely a preferred embodiment of the present invention, but the scope of protection of the present invention is not limited thereto. Any variations or substitutions that can be easily conceived by those skilled in the art within the scope of the technology disclosed in the present invention should be included within the scope of protection of the present invention. Therefore, the scope of protection of the present invention should be determined by the scope of the claims.

Claims

1. A method for estimating and equalizing the state of charge of a lithium-ion battery, characterized in that, Includes the following steps: S1: Establish an equivalent circuit model of the battery and introduce the state variable, battery state of charge. S2: Establish a battery state of charge estimation model based on BP neural network, collect physical quantities related to battery state of charge as input to the battery state of charge estimation model, and output the estimated value of battery state of charge. S3: Establish a particle swarm optimization algorithm based on battery energy path optimization. Using an inertia weight adjustment strategy based on fitness function values, dynamically adjust the inertia weights of the particle swarm algorithm at different stages to find the optimal path for battery energy transfer; where the fitness function value... The fitness function value for the overall charging energy transfer path of the battery. The expression is: In the formula, , For the first Battery saving , for The average state of charge of the battery; The times correspond to Section 1, Section 2, Section 3, ..., Section 4 respectively. Save battery; S4: The current output is controlled by a fuzzy controller to obtain the optimal equalization current; the input to the fuzzy controller used for equalization current regulation is the average value of the overall state of charge of the battery. The difference between the states of charge of two adjacent batteries The output of the fuzzy controller is the optimal equalization current. .

2. The method for estimating and equalizing the state of charge of a lithium-ion battery according to claim 1, characterized in that... In step S1, the formula for the battery state of charge is expressed as follows: In the formula, This represents the initial value of the battery's state of charge. Indicates battery charge / discharge efficiency. Indicates the battery operating current. Indicates the battery's nominal capacity; The initial value of the battery state of charge, the battery charge and discharge efficiency, the battery operating current, and the battery nominal capacity are all used as inputs to the battery state of charge estimation model in step S2. The functional relationship between the input and output of the hidden layer and output layer nodes of the battery state of charge estimation model is the Sigmoid operation function.

3. The method for estimating and equalizing the state of charge of a lithium-ion battery according to claim 2, characterized in that... Step S3 includes the following steps: S31: Randomly generate S different particles and set the optimization range of the initial particles; S32: Calculate and record the fitness value of each particle, and take the minimum fitness value of each particle as the optimal fitness value for that individual. And record the position of each particle; S33: Find the particle with the lowest fitness value in the population and use its fitness value as the global optimal fitness value. The position of this particle is the global optimum; and the average fitness value of particles in the population is calculated. ; S34: Improve the inertia weight parameter adjustment formula and calculate the inertia weight; S35: Substitute the inertial weight from step S34 into the update formula for particle velocity and position to update the particle velocity and position. S36: Determine the velocity and position of the particles during the iteration process to ensure that the particles iterate within a preset range. If the iteration accuracy reaches the preset condition, stop the iteration; otherwise, go to step S32 and continue the iteration. S37: After iteration, the optimal solution for the battery energy path is obtained.

4. The method for estimating and equalizing the state of charge of a lithium-ion battery according to claim 3, characterized in that... In step S34, the improved formula for adjusting the inertia weight parameter is: In the formula, Represents inertial parameters, This represents the minimum value of the inertial parameter. This represents the maximum value of the inertial parameter; This represents the average fitness value of the population. Indicates the first in the population The fitness value of each particle. This represents the maximum value of the entire battery's state of charge. This represents the average value of the overall state of charge of the battery. This represents the number of iterations in the particle swarm optimization algorithm. This indicates the maximum number of iterations.

5. The method for estimating and equalizing the state of charge of a lithium-ion battery according to claim 4, characterized in that... In step S35, the update formulas for particle velocity and position are as follows: In the formula, Indicates being between Random numbers between; Represents inertial parameters, , Indicates the learning factor; Indicates the current position of the particle; This represents the optimal position of a particle in a particle swarm. Indicates the first The velocity of each particle; Indicates the first The position of each particle; Indicates the corresponding time.

Images