Energy storage system SOC calculation method and system based on full life cycle

Abstract

The invention relates to the field of state of charge measurement, and provides an energy storage system SOC calculation method and system based on a full life cycle, and the method comprises the steps: obtaining the cycle index of an energy storage system; according to the cycle index, the life cycle of the energy storage system is determined, and the life cycle comprises an initial use period, a middle use period and a decline period; according to the life cycle, a target model for SOC calculation is determined from a plurality of preset models, input parameters corresponding to the target model are determined from a plurality of preset parameters, and the preset parameters comprise current, voltage, temperature, cycle index, internal resistance and aging attenuation rate; and detecting the input parameters of the energy storage system, and inputting the input parameters into the target model to obtain SOC parameters output by the target model. Therefore, the energy storage system is more accurately divided into different life cycles according to the cycle index, and different target models and input parameters are adopted in different life cycles to calculate the SOC, so that the accuracy of SOC calculation is improved.

Description

Technical Field

[0001] The present application relates to the field of state of charge measurement, and in particular to a method and system for calculating the SOC of an energy storage system based on the entire life cycle. Background Art

[0002] The state of charge (SOC) indicates the available state of the remaining charge in the energy storage system. Currently, the SOC estimation technology of energy storage systems mostly relies on short-term or local operating condition data, and does not fully integrate the multi-dimensional data of the battery's entire life cycle from initial use to capacity decline, resulting in significant deviations in the model due to data limitations in long-term applications. At the same time, the neural network model used to calculate SOC is difficult to dynamically adapt to the dynamic evolution of nonlinear characteristics during battery aging due to fixed hyperparameters. It is easy to fall into local optimality and it is difficult to mine aging laws and time series associations from massive heterogeneous data, further exacerbating the estimation error. This leads to insufficient precision and accuracy in SOC estimation throughout the battery's life cycle, limiting the application effectiveness of energy storage systems in long-cycle scenarios such as electric vehicles and smart grids. Therefore, there is an urgent need for a more accurate SOC calculation method. Summary of the Invention

[0003] The main purpose of this application is to provide a method for calculating the SOC of an energy storage system based on the entire life cycle, characterized in that the method includes:

[0004] Obtaining the number of cycles of the energy storage system;

[0005] Determining the life cycle of the energy storage system according to the number of cycles, wherein the life cycle includes: an initial period of use, a mid-period of use, and a decay period;

[0006] Determining a target model for SOC calculation from a plurality of preset models according to the life cycle, and determining input parameters corresponding to the target model from a plurality of preset parameters, the preset parameters including: current, voltage, temperature, number of cycles, internal resistance, and aging attenuation rate;

[0007] The input parameters of the energy storage system are detected, and the input parameters are input into the target model to obtain the SOC parameters output by the target model.

[0008] In some embodiments, the method further comprises:

[0009] Based on the Raccoon optimization algorithm, determining training parameters corresponding to the first preset model, the second preset model, and the third preset model respectively, wherein the training parameters include: learning rate, number of hidden layer nodes, and regularization coefficient;

[0010] Based on the training parameters, the first preset model, the second preset model, and the third preset model are trained to obtain an initial prediction model, a mid-term prediction model, and a decline prediction model.

[0011] In some embodiments, determining the training parameters corresponding to the first preset model, the second preset model, and the third preset model based on the raccoon optimization algorithm includes:

[0012] Divide the raccoon population into a first raccoon population and a second raccoon population;

[0013] Randomly generate the positions of the first raccoon population and the second raccoon population in the search space to obtain the raccoon population matrix And the objective function matrix Where i = 1, 2, ..., N, j = 1, 2, ..., m, X i represents the position of the i-th raccoon in the search space, x i,j represents the value of the j-th decision variable, N represents the number of raccoon populations, m represents the number of decision variables; F represents the vector of the objective function, F i represents the objective function value obtained by the i-th raccoon;

[0014] Optimizing the training parameters in the search space using the first raccoon population to obtain first training parameters;

[0015] The first training parameters are optimized using the second raccoon population to obtain target training parameters.

[0016] In some embodiments, optimizing the training parameters in the search space using the first raccoon population to obtain the first training parameters includes:

[0017] dividing the first raccoon population into a third raccoon population and a fourth raccoon population;

[0018] The position of raccoons in the third raccoon population is represented based on the following formula: in, represents calculating the new position of the i-th raccoon in the first raccoon population, Indicates the j-dimensionality of the new position, Ig j represents the position of the lizard in the search space, i.e. the position of the optimal solution, j is the dimension, and r represents a random real number in [0, 1];

[0019] The position of raccoons in the fourth raccoon population is represented based on the following formula:

[0020]

[0021] Among them, ub j and lbj are the upper and lower bounds of the jth decision variable, I is an integer randomly selected from the set {1, 2}, and Ig G Indicates the randomly generated position of the lizard on the ground. is its j-dimension, F Ig G is the value of its objective function;

[0022] The positions of the raccoons in the third and fourth raccoon populations are updated based on the following formula: The first training parameter is obtained.

[0023] In some embodiments, optimizing the first training parameters using the second raccoon population to obtain target training parameters includes:

[0024] Generate random locations of raccoons in the second raccoon population near the location of each first training parameter according to the following formula:

[0025]

[0026] in, is the position of the i-th raccoon in the second raccoon population, is its j-dimension, t is the iteration counter, and are the local upper and lower bounds of the j-th decision variable respectively;

[0027] The positions of the raccoons in the second raccoon population are updated according to the following formula:

[0028] Obtaining the target training parameters;

[0029] Among them, F i P2 is the objective function.

[0030] In some embodiments, the first preset model, the second preset model, and the third preset model are trained based on the training parameters to obtain an initial prediction model, a mid-term prediction model, and a decline prediction model, including:

[0031] Based on the current samples, voltage samples, and temperature samples whose cycle times are within the range of [A1, A2], the first preset model is trained to obtain an initial prediction model corresponding to the initial use period;

[0032] Based on the current samples, voltage samples, temperature samples, and cycle number samples whose cycle numbers are within the range of [B1, B2], the second preset model is trained to obtain a medium-term prediction model corresponding to the medium-term use period;

[0033] Based on current samples, voltage samples, temperature samples, cycle number samples, internal resistance samples, and aging attenuation rate samples whose cycle numbers are within the range of [C1, C2], the third preset model is trained to obtain a decay period prediction model corresponding to the decay period.

[0034] In some embodiments, the method further comprises:

[0035] Obtaining a first accuracy rate of the initial prediction model for test samples with different numbers of cycles, a second accuracy rate of the mid-term prediction model for test samples with different numbers of cycles, and a third accuracy rate of the decay period prediction model for test samples with different numbers of cycles;

[0036] Fitting the first accuracy rate, the second accuracy rate, and the third accuracy rate to obtain a first accuracy rate curve, a second accuracy rate curve, and a third accuracy rate curve;

[0037] The first cycle number A is determined according to the intersection of the first accuracy curve and the second accuracy curve as the boundary between the initial prediction model and the medium-term prediction model, and the second cycle number B is determined according to the intersection of the second accuracy curve and the third accuracy curve as the boundary between the medium-term prediction model and the decline period prediction model.

[0038] In some embodiments, determining the life cycle of the energy storage system according to the number of cycles includes:

[0039] When the number of cycles is within the range of (0, A], it is determined that the energy storage system is in the initial stage of its life cycle;

[0040] When the number of cycles is within the range of (A, B], determining that the energy storage system is in the middle of its life cycle;

[0041] When the number of cycles is within the range of (B,∞], it is determined that the life cycle of the energy storage system is in a decay period.

[0042] In some embodiments, determining a target model for SOC calculation from a plurality of preset models according to the life cycle, and determining input parameters corresponding to the target model from a plurality of preset parameters, includes:

[0043] When the energy storage system is in an early stage of its life cycle, an early stage prediction model is determined as the target model, and current, voltage, and temperature are determined as the input parameters;

[0044] When the energy storage system is in the middle of its life cycle, a mid-term prediction model is determined as the target model, and current, voltage, temperature, and number of cycles are determined as the input parameters;

[0045] When the energy storage system is in a decay period in its life cycle, a decay period prediction model is determined as the target model, and current, voltage, temperature, number of cycles, internal resistance, and aging decay rate are determined as the input parameters.

[0046] In a second aspect, an embodiment of the present application further provides a SOC calculation system for an energy storage system based on a full life cycle, the system comprising:

[0047] A cycle number acquisition module, used to obtain the cycle number of the energy storage system;

[0048] A life cycle determination module, configured to determine the life cycle of the energy storage system according to the number of cycles, wherein the life cycle includes: an initial use period, a mid-use period, and a decay period;

[0049] a target model determination module, configured to determine a target model for SOC calculation from a plurality of preset models according to the life cycle, and to determine input parameters corresponding to the target model from a plurality of preset parameters, the preset parameters including: current, voltage, temperature, number of cycles, internal resistance, and aging attenuation rate;

[0050] The SOC prediction module is used to detect the input parameters of the energy storage system and input the input parameters into the target model to obtain the SOC parameters output by the target model.

[0051] The present application provides a method and system for calculating the SOC of an energy storage system based on the entire life cycle. The method includes: obtaining the number of cycles of the energy storage system; determining the life cycle of the energy storage system based on the number of cycles, wherein the life cycle includes: initial use, mid-use, and decay period; determining a target model for SOC calculation from a plurality of preset models based on the life cycle, and determining input parameters corresponding to the target model from a plurality of preset parameters, wherein the preset parameters include: current, voltage, temperature, number of cycles, internal resistance, and aging attenuation rate; detecting the input parameters of the energy storage system, and inputting the input parameters into the target model to obtain the SOC parameters output by the target model. In this way, the energy storage system is more accurately divided into different life cycles based on the number of cycles, and different target models and input parameters are used to calculate the SOC in different life cycles, thereby improving the accuracy of the SOC calculation. BRIEF DESCRIPTION OF THE DRAWINGS

[0052] In order to more clearly illustrate the technical solutions of the embodiments of the present application, the following is a brief introduction to the drawings required for use in the description of the embodiments. Obviously, the drawings described below are some embodiments of the present application. For ordinary technicians in this field, other drawings can be obtained based on these drawings without any creative work.

[0053] Figure 1 A flowchart of a method for calculating the SOC of an energy storage system based on the entire life cycle provided in one embodiment of the present application;

[0054] Figure 2 A schematic block diagram of a scenario of a method for calculating the SOC of an energy storage system based on the entire life cycle provided in one embodiment of the present application;

[0055] Figure 3 A schematic block diagram of a SOC calculation system for an energy storage system based on a full life cycle provided by an embodiment of the present application;

[0056] Figure 4 This is a schematic block diagram of the structure of a computer device involved in one embodiment of the present application. DETAILED DESCRIPTION

[0057] The following will be combined with the drawings in the embodiments of this application to clearly and completely describe the technical solutions in the embodiments of this application. Obviously, the embodiments described are part of the embodiments of this application, not all of them. Based on the embodiments in this application, all other embodiments obtained by ordinary technicians in this field without making creative efforts are within the scope of protection of this application.

[0058] The flowcharts shown in the accompanying drawings are for illustrative purposes only and do not necessarily include all contents and operations / steps, nor must they be executed in the order described. For example, some operations / steps may be decomposed, combined, or partially merged, so the actual execution order may vary depending on the actual situation.

[0059] The embodiments of the present application provide a method and system for calculating the SOC of an energy storage system based on the entire life cycle.

[0060] The following describes some embodiments of the present application in detail with reference to the accompanying drawings. In the absence of conflict, the following embodiments and features therein may be combined with each other.

[0061] Please refer to Figure 1 , Figure 1A flow chart of a method for calculating the SOC of an energy storage system based on the entire life cycle provided for an embodiment of the present application. The method for calculating the SOC of an energy storage system based on the entire life cycle can be used in a terminal or a server to accurately calculate the SOC of the energy storage system based on the life cycle of the energy storage system. The terminal can be an electronic device such as a mobile phone, a tablet computer, a laptop computer, a desktop computer, a personal digital assistant, and a wearable device; the server can be an independent server, a server cluster, or a cloud server that provides basic cloud computing services such as cloud services, cloud databases, cloud computing, cloud functions, cloud storage, network services, cloud communications, middleware services, domain name services, security services, content delivery networks (CDNs), and big data and artificial intelligence platforms.

[0062] like Figure 1 As shown, the energy storage system SOC calculation method based on the entire life cycle includes steps S101 to S104.

[0063] Step S101: Obtain the number of cycles of the energy storage system.

[0064] For example, the number of cycles of an energy storage system is usually limited. Assuming that the rated number of cycles of the energy storage system is about 3,000, it means that the performance of the energy storage system will drop significantly after completing about 3,000 charge and discharge cycles. A charge and discharge cycle refers to the process of charging a battery from a fully discharged state until it is fully charged, and then discharging it until the battery is exhausted. For example, a battery is charged from 0% to 100%, and then discharged from 100% to 0%. This is considered a complete cycle.

[0065] For example, the number of cycles indicates the number of charge and discharge cycles completed after the energy storage system is activated. The number of cycles of the energy storage system can reflect the life cycle of the energy storage system. For example, when the energy storage system is first used, the number of cycles is low. When the number of cycles approaches the rated number of cycles of the energy storage system, it indicates that the energy storage system is close to being scrapped.

[0066] Step S102: Determine the life cycle of the energy storage system according to the number of cycles, where the life cycle includes: an initial use period, a mid-use period, and a decay period.

[0067] For example, at different stages of the life cycle of an energy storage system, the SOC parameters may have different characteristics. Therefore, the number of cycles is used to determine whether the energy storage system is in the early use, middle use, or decline stage of its life cycle, so that different methods can be used to calculate the SOC parameters for different life cycles to improve the accuracy of the SOC parameter calculation.

[0068] In some embodiments, determining the life cycle of the energy storage system according to the number of cycles includes:

[0069] When the number of cycles is within the range of (0, A], it is determined that the energy storage system is in the initial stage of its life cycle;

[0070] When the number of cycles is within the range of (A, B], determining that the energy storage system is in the middle of its life cycle;

[0071] When the number of cycles is within the range of (B,∞], it is determined that the life cycle of the energy storage system is in a decay period.

[0072] For example, the sizes of A and B are pre-set so that the life cycle of the energy storage system can be divided according to the number of cycles and the sizes of A and B. Specifically, in the SOC calculation method for an energy storage system based on the entire life cycle provided in an embodiment of the present application, the size of A can be, for example, 500, and the size of B can be, for example, 1500.

[0073] For example, when the number of cycles is less than or equal to 500, the energy storage system is determined to be in the early stage of its life cycle; when the number of cycles is greater than 500 and less than or equal to 1500, the energy storage system is determined to be in the mid-stage of its life cycle; and when the number of cycles is greater than 1500, the energy storage system is determined to be in the decline stage of its life cycle. Of course, this is not limiting, and the values ​​of A and B can also be other values, which are not limited here.

[0074] Step S103: determining a target model for SOC calculation from a plurality of preset models according to the life cycle, and determining input parameters corresponding to the target model from a plurality of preset parameters, wherein the preset parameters include: current, voltage, temperature, number of cycles, internal resistance, and aging attenuation rate.

[0075] For example, the preset parameters used for SOC calculation at different life cycles may be different. For example, in the initial use, the energy storage system is in good condition and is relatively consistent with the factory setting. At this time, only fewer parameters are needed to accurately calculate the SOC parameters; however, as the energy storage system ages, the state of the energy storage system deviates from the factory state, and the SOC parameters are also prone to drift due to other factors. More preset parameters are required to accurately calculate the SOC parameters.

[0076] In some embodiments, determining a target model for SOC calculation from a plurality of preset models according to the life cycle, and determining input parameters corresponding to the target model from a plurality of preset parameters, includes:

[0077] When the energy storage system is in an early stage of its life cycle, an early stage prediction model is determined as the target model, and current, voltage, and temperature are determined as the input parameters;

[0078] When the energy storage system is in the middle of its life cycle, a mid-term prediction model is determined as the target model, and current, voltage, temperature, and number of cycles are determined as the input parameters;

[0079] When the energy storage system is in a decay period in its life cycle, a decay period prediction model is determined as the target model, and current, voltage, temperature, number of cycles, internal resistance, and aging decay rate are determined as the input parameters.

[0080] For example, in the initial use of the energy storage system, the current, voltage, and temperature are input into the target model to obtain the SOC parameters output by the target model; in the middle use of the energy storage system, the current, voltage, temperature, and number of cycles are input into the target model to obtain the SOC parameters output by the target model; in the decay period of the energy storage system, the current, voltage, temperature, number of cycles, internal resistance, and aging attenuation rate are input into the target model to obtain the SOC parameters output by the target model.

[0081] Step S104 : detecting the input parameters of the energy storage system, and inputting the input parameters into the target model to obtain the SOC parameters output by the target model.

[0082] For example, different target models are used to calculate the SOC parameters of the energy storage system during different life cycles of the energy storage system, thereby improving the accuracy and pertinence of the SOC calculation.

[0083] The target model can be implemented, for example, based on a gated recurrent unit (GRU) neural network. The GRU's reset and update gates control the flow of information. The reset gate determines which information from previous moments is retained at the current moment, while the update gate determines the extent of the current state update. This allows the selection of useful information for SOC calculations, while removing irrelevant or redundant information. This improves computational efficiency and accuracy and mitigates the vanishing gradient problem.

[0084] In some embodiments, the method further comprises:

[0085] Based on the Raccoon optimization algorithm, determining training parameters corresponding to the first preset model, the second preset model, and the third preset model respectively, wherein the training parameters include: learning rate, number of hidden layer nodes, and regularization coefficient;

[0086] Based on the training parameters, the first preset model, the second preset model, and the third preset model are trained to obtain an initial prediction model, a mid-term prediction model, and a decline prediction model.

[0087] Exemplarily, the key parameters that affect the training effect of the GRU neural network model mainly include learning rate, the number of hidden layer neurons and regularization coefficient. Among them, the learning rate controls the update direction and convergence speed of the model, the number of hidden layer neurons determines the nonlinear fitting ability of the model, and the regularization coefficient mainly prevents simulation overfitting. Selecting appropriate hyperparameters of the GRU network can make the model converge quickly and improve the performance and effect of learning. However, the hyperparameters of the GRU neural network are often difficult to determine. The selection of different hyperparameters will have a great impact on the training results, and there may also be problems with multiple local extreme values. Therefore, the method provided in the embodiment of the present application uses the raccoon optimization algorithm to optimize the training parameters of models with different life cycles.

[0088] For example, the Coati Optimization Algorithm (COA) has good global optimization capabilities and convergence speed. The Coati Optimization Algorithm is used to determine the training parameters of models with different life cycles, thereby avoiding the problem of inaccurate initial parameter selection of the GRU neural network, which leads to slow convergence speed and poor results of the training model.

[0089] Please refer to Figure 2 , Figure 2 A schematic block diagram of a scenario of a method for calculating the SOC of an energy storage system based on the entire life cycle provided in one embodiment of the present application.

[0090] like Figure 2 As shown in FIG, after the training parameters of the model are determined through multiple iterations of COA, the optimal parameters are assigned to the hyperparameters of the GRU neural network, thereby training the GRU neural network based on the normalized monitoring data.

[0091] In some embodiments, determining the training parameters corresponding to the first preset model, the second preset model, and the third preset model based on the raccoon optimization algorithm includes:

[0092] Divide the raccoon population into a first raccoon population and a second raccoon population;

[0093] Randomly generate the positions of the first raccoon population and the second raccoon population in the search space to obtain the raccoon population matrix And the objective function matrix Where i = 1, 2, ..., N, j = 1, 2, ..., m, X i represents the position of the i-th raccoon in the search space, x i,jrepresents the value of the j-th decision variable, N represents the number of raccoon populations, m represents the number of decision variables; F represents the vector of the objective function, F i represents the objective function value obtained by the i-th raccoon;

[0094] Optimizing the training parameters in the search space using the first raccoon population to obtain first training parameters;

[0095] The first training parameters are optimized using the second raccoon population to obtain target training parameters.

[0096] For example, the raccoon optimization algorithm (COA) divides optimization into two phases. The first phase is the exploration phase, in which the raccoons' positions are updated by simulating the strategy of a raccoon pack attacking a lizard. In this strategy, a group of raccoons climbs a tree to surround a lizard and scare it. Several other raccoons wait under the tree until the lizard falls. After the lizard falls to the ground, the raccoons on the ground attack and pursue it. This strategy causes the raccoons to move to different locations in the search space, demonstrating the exploration capability of COA in global search for solving spatial problems. The second phase is the exploitation phase, in which the raccoons' positions are updated by simulating the strategy of encountering and escaping from predators. When a predator attacks a raccoon, it flees from its original position. Using this strategy, the raccoons move their current position closer to a safe location, demonstrating the exploitation capability of COA in local search. COA has a high ability to scan the search space at both global and local levels. After identifying the primary optimal area, it can converge to the optimal solution at a very high speed.

[0097] Therefore, the SOC calculation method for the energy storage system based on the entire life cycle provided in the embodiment of the present application randomly generates the position of raccoons in the search space to obtain a raccoon group matrix. The position of the raccoon is the candidate solution. The position of the candidate solution in the decision variable is different, and the evaluation of the objective function value of the problem is also different. Each candidate solution is recorded through the objective function matrix.

[0098] Illustratively, the embodiment of the present application determines a better first training parameter through the exploration phase, and then optimizes the first training parameter through the development phase to determine the target training parameter.

[0099] Exemplarily, the search space is a three-dimensional search space consisting of the learning rate, the number of hidden layer nodes, and the regularization coefficient. The objective function can be determined, for example, by the following formula: F = α MAE + β Training Time + γ Generalization Error. Here, α, β, and γ are weight coefficients, MAE is the mean absolute error of the validation set, and Generalization Error is the cross-stage prediction error.

[0100] In some embodiments, optimizing the training parameters in the search space using the first raccoon population to obtain the first training parameters includes:

[0101] dividing the first raccoon population into a third raccoon population and a fourth raccoon population;

[0102] The position of raccoons in the third raccoon population is represented based on the following formula: in, represents calculating the new position of the i-th raccoon in the first raccoon population, Indicates the j-dimensionality of the new position, Ig j represents the position of the lizard in the search space, i.e. the position of the optimal solution, j is the dimension, and r represents a random real number in [0, 1];

[0103] The position of raccoons in the fourth raccoon population is represented based on the following formula:

[0104]

[0105] Among them, ub j and lb j are the upper and lower bounds of the jth decision variable, I is an integer randomly selected from the set {1, 2}, and Ig G Indicates the randomly generated position of the lizard on the ground. is its j-dimension, F Ig G is the value of its objective function;

[0106] The positions of the raccoons in the third and fourth raccoon populations are updated based on the following formula: The first training parameter is obtained.

[0107] For example, in the exploration phase, the N raccoons in the first raccoon population are divided into two halves to obtain a third raccoon population and a fourth raccoon population, where the number of the third raccoon population and the fourth raccoon population is equal, N / 2 respectively. The third raccoon population is used to simulate raccoons attacking in trees, and the fourth raccoon population is used to simulate raccoons foraging on the ground, thereby generating new hyperparameter combinations in the global space and using initial high-variance data to expand the search range.

[0108] For example, the third raccoon population simulates the raccoons on the tree to maintain their original position or attack the lizard by using the random integer 1 or 2, and the formula Update location; Ig G The new position of the lizard after it falls to the ground is represented by the formula Ig G : and Update location.

[0109] For example, if the new position calculated for each raccoon improves the value of the objective function, the new position is accepted, otherwise the raccoon remains in the previous position. In other words, a greedy selection is performed, as shown in the following formula: Until the number of iterations reaches the first raccoon population size, and the results obtained in the search space through the exploration phase are used as the first training parameters.

[0110] For example, different life cycles correspond to different ub j and lb j For example, the first preset model corresponding to the initial stage has a learning rate of 0.001-0.01, the number of hidden layer nodes is 128-256, and the regularization coefficient is 0.01-0.1, thereby improving the structural complexity through high learning rate and a larger number of hidden layer nodes, expanding the search range, adapting to highly nonlinear data, and quickly fitting dynamic characteristics; the third preset model corresponding to the decay period has a learning rate of 0.0001-0.001, the number of hidden layer nodes is 64-128, and the coefficient regularization is 0.1-0.5, thereby compressing the step size through low learning rate and strong regularization, suppressing noise interference, simplifying the structure through a smaller number of hidden layer nodes, and improving robustness; the training parameters of the second preset model corresponding to the mid-term can be between the first preset model and the third preset model, for example, the learning rate is 0.01-0.1, the number of hidden layer nodes is 64-256, and the regularization coefficient is 0.05-0.25.

[0111] For example, taking the third preset model corresponding to the decay period as an example, the target training parameters finally optimized may be a learning rate of 0.0003, 96 nodes in the hidden layer, and an L2 regularization of 0.3.

[0112] In some embodiments, optimizing the first training parameters using the second raccoon population to obtain target training parameters includes:

[0113] Generate random locations of raccoons in the second raccoon population near each location of the first training parameter according to the following formula:

[0114]

[0115] in, is the position of the i-th raccoon in the second raccoon population, is its j-dimension, t is the iteration counter, and are the local upper and lower bounds of the j-th decision variable respectively;

[0116] The positions of the raccoons in the second raccoon population are updated according to the following formula:

[0117] Obtaining the target training parameters;

[0118] Among them, F i P2 is the objective function.

[0119] For example, in the development phase, a detailed search is performed within the neighborhood of the optimal solution, and for low signal-to-noise ratio data in the decay period, an adaptive local boundary compression search step is used.

[0120] For example, to simulate the behavior of raccoons avoiding predators, a random position is generated near the position of each raccoon (i.e., the first training parameter), as shown in the following formula:

[0121]

[0122] Among them, if the newly calculated position improves the value of the objective function, the position is acceptable and the following formula is used for simulation, that is, a greedy selection is performed again:

[0123] In some embodiments, the training of the first preset model, the second preset model, and the third preset model based on the training parameters to obtain an initial prediction model, a mid-term prediction model, and a decline prediction model includes:

[0124] Based on the current samples, voltage samples, and temperature samples whose cycle times are within the range of [A1, A2], the first preset model is trained to obtain an initial prediction model corresponding to the initial use period;

[0125] Based on the current samples, voltage samples, temperature samples, and cycle number samples whose cycle numbers are within the range of [B1, B2], the second preset model is trained to obtain a medium-term prediction model corresponding to the medium-term use period;

[0126] Based on current samples, voltage samples, temperature samples, cycle number samples, internal resistance samples, and aging attenuation rate samples whose cycle numbers are within the range of [C1, C2], the third preset model is trained to obtain a decay period prediction model corresponding to the decay period.

[0127] Exemplarily, the sample information used to train the model includes at least current samples, voltage samples, temperature samples, cycle count samples, internal resistance samples, and aging decay rate samples. Each sample information corresponds to an SOC tag and is also associated with the cycle count at the time of collection. For example, if a set of sample information including current samples, voltage samples, temperature samples, cycle count samples, internal resistance samples, and aging decay rate samples is collected from an energy storage system with a cycle count of 50, then the associated cycle count is 50, and the corresponding SOC is 80%, and the corresponding SOC tag is 80%.

[0128] For example, since models need to be trained separately for the initial, mid-, and decay phases, it's understandable that the sample information used by different models should come from the initial, mid-, and decay phases, respectively. However, the number of cycles used as the boundary between the initial, mid-, and decay phases may vary for different energy storage systems. Therefore, it's possible to preliminarily determine the approximate number of cycles corresponding to different life cycles, train models corresponding to the initial, mid-, and decay phases, and then determine the specific boundaries for different life cycles based on the accuracy of the different models at different numbers of cycles. For example, the first preset model is trained using sample information with cycle numbers within the range [A1, A2]; the second preset model is trained using sample information with cycle numbers within the range [B1, B2]; and the third preset model is trained using sample information with cycle numbers within the range [C1, C2]. Where A2 < B1, B2 < C1.

[0129] For example, the first preset model can be trained using sample information with a cycle number in the range of [200, 800]; the second preset model can be trained using sample information with a cycle number in the range of [1200, 1800]; and the third preset model can be trained using sample information with a cycle number in the range of [2200, 2800]. Of course, this is not limited to this and is not set forth herein.

[0130] In some embodiments, the method further comprises:

[0131] Obtaining a first accuracy rate of the initial prediction model for test samples with different numbers of cycles, a second accuracy rate of the mid-term prediction model for test samples with different numbers of cycles, and a third accuracy rate of the decay period prediction model for test samples with different numbers of cycles;

[0132] Fitting the first accuracy rate, the second accuracy rate, and the third accuracy rate to obtain a first accuracy rate curve, a second accuracy rate curve, and a third accuracy rate curve;

[0133] The first cycle number A is determined according to the intersection of the first accuracy curve and the second accuracy curve as the boundary between the initial prediction model and the medium-term prediction model, and the second cycle number B is determined according to the intersection of the second accuracy curve and the third accuracy curve as the boundary between the medium-term prediction model and the decline period prediction model.

[0134] For example, the trained initial prediction model, mid-term prediction model, and decay period prediction model are tested on test samples with different numbers of cycles to determine the accuracy of each of the initial prediction model, mid-term prediction model, and decay period prediction model for the test samples with different numbers of cycles. Taking the initial prediction model as an example, the accuracy of the initial prediction model for test samples in the intervals of (600, 700], (700, 800], and (800, 900] can be determined respectively.

[0135] Exemplarily, the first accuracy of the initial prediction model, the second accuracy of the mid-term prediction model, and the third accuracy of the decline prediction model are represented in the same coordinate system, where the horizontal axis of the coordinate system is the number of cycles and the vertical axis is the accuracy. The first accuracy, the second accuracy, and the third accuracy in the coordinate system are fitted respectively to obtain the first accuracy curve, the second accuracy curve, and the third accuracy curve.

[0136] As can be understood, since the initial prediction model is trained based on sample information with a cycle number of [A1, A2], and the mid-term prediction model is trained based on sample information with a cycle number of [B1, B2], the initial prediction model has a higher accuracy for test samples with a cycle number of [A1, A2], and the mid-term prediction model has a higher accuracy for test samples with a cycle number of [B1, B2]. However, when the cycle number is greater than A2 and less than B1, that is, within the range of (A2, B1), the accuracy of both the initial and mid-term prediction models will decrease to a certain extent. As a result, before a certain position in the range of (A2, B1), the first accuracy is greater than the second accuracy, and after this position, the first accuracy is less than the second accuracy. This position is the intersection of the first and second accuracy curves. Therefore, the cycle number at the intersection of the first and second accuracy curves is determined as the boundary between initial and mid-term use. For cycles below this boundary, the SOC is calculated using the initial prediction model, and for cycles above this boundary, the SOC is calculated using the mid-term prediction model. This improves the flexibility of lifecycle segmentation and the accuracy of SOC calculation.

[0137] Please refer to Figure 3 , Figure 3 A schematic block diagram of a SOC calculation system for an energy storage system based on the entire life cycle provided in one embodiment of the present application.

[0138] like Figure 3 As shown, the energy storage system SOC calculation system based on the entire life cycle includes: a cycle number acquisition module 110, a life cycle determination module 120, a target model determination module 130, and an SOC prediction module 140.

[0139] A cycle number acquisition module 110 is used to obtain the cycle number of the energy storage system;

[0140] A life cycle determination module 120 is configured to determine the life cycle of the energy storage system according to the number of cycles, wherein the life cycle includes: an initial use period, a mid-use period, and a decay period;

[0141] a target model determination module 130 for determining a target model for SOC calculation from a plurality of preset models according to the life cycle, and determining input parameters corresponding to the target model from a plurality of preset parameters, the preset parameters including: current, voltage, temperature, number of cycles, internal resistance, and aging attenuation rate;

[0142] The SOC prediction module 140 is configured to detect the input parameters of the energy storage system and input the input parameters into the target model to obtain the SOC parameters output by the target model.

[0143] In some embodiments, the method further comprises:

[0144] Based on the Raccoon optimization algorithm, determining training parameters corresponding to the first preset model, the second preset model, and the third preset model respectively, wherein the training parameters include: learning rate, number of hidden layer nodes, and regularization coefficient;

[0145] Based on the training parameters, the first preset model, the second preset model, and the third preset model are trained to obtain an initial prediction model, a mid-term prediction model, and a decline prediction model.

[0146] In some embodiments, determining the training parameters corresponding to the first preset model, the second preset model, and the third preset model based on the raccoon optimization algorithm includes:

[0147] Divide the raccoon population into a first raccoon population and a second raccoon population;

[0148] Randomly generate the positions of the first raccoon population and the second raccoon population in the search space to obtain the raccoon population matrix And the objective function matrix Where i = 1, 2, ..., N, j = 1, 2, ..., m, X i represents the position of the i-th raccoon in the search space, x i,jrepresents the value of the j-th decision variable, N represents the number of raccoon populations, m represents the number of decision variables; F represents the vector of the objective function, F i represents the objective function value obtained by the i-th raccoon;

[0149] Optimizing the training parameters in the search space using the first raccoon population to obtain first training parameters;

[0150] The first training parameters are optimized using the second raccoon population to obtain target training parameters.

[0151] In some embodiments, optimizing the training parameters in the search space using the first raccoon population to obtain the first training parameters includes:

[0152] dividing the first raccoon population into a third raccoon population and a fourth raccoon population;

[0153] The position of raccoons in the third raccoon population is represented based on the following formula: in, represents calculating the new position of the i-th raccoon in the first raccoon population, Indicates the j-dimensionality of the new position, Ig j represents the position of the lizard in the search space, i.e. the position of the optimal solution, j is the dimension, and r represents a random real number in [0, 1];

[0154] The position of raccoons in the fourth raccoon population is represented based on the following formula:

[0155]

[0156] Among them, ub j and lb j are the upper and lower bounds of the jth decision variable, I is an integer randomly selected from the set {1, 2}, and Ig G Indicates the randomly generated position of the lizard on the ground. is its j-dimension, F Ig G is the value of its objective function;

[0157] The positions of the raccoons in the third and fourth raccoon populations are updated based on the following formula: The first training parameter is obtained.

[0158] In some embodiments, optimizing the first training parameters using the second raccoon population to obtain target training parameters includes:

[0159] Generate random locations of raccoons in the second raccoon population near the location of each first training parameter according to the following formula:

[0160]

[0161] in, is the position of the i-th raccoon in the second raccoon population, is its j-dimension, t is the iteration counter, and are the local upper and lower bounds of the j-th decision variable respectively;

[0162] The positions of the raccoons in the second raccoon population are updated according to the following formula:

[0163] Obtaining the target training parameters;

[0164] Among them, F i P2 is the objective function.

[0165] In some embodiments, the first preset model, the second preset model, and the third preset model are trained based on the training parameters to obtain an initial prediction model, a mid-term prediction model, and a decline prediction model, including:

[0166] Based on the current samples, voltage samples, and temperature samples whose cycle times are within the range of [A1, A2], the first preset model is trained to obtain an initial prediction model corresponding to the initial use period;

[0167] Based on the current samples, voltage samples, temperature samples, and cycle number samples whose cycle numbers are within the range of [B1, B2], the second preset model is trained to obtain a medium-term prediction model corresponding to the medium-term use period;

[0168] Based on current samples, voltage samples, temperature samples, cycle number samples, internal resistance samples, and aging attenuation rate samples whose cycle numbers are within the range of [C1, C2], the third preset model is trained to obtain a decay period prediction model corresponding to the decay period.

[0169] In some embodiments, the method further comprises:

[0170] Obtaining a first accuracy rate of the initial prediction model for test samples with different numbers of cycles, a second accuracy rate of the mid-term prediction model for test samples with different numbers of cycles, and a third accuracy rate of the decay period prediction model for test samples with different numbers of cycles;

[0171] Fitting the first accuracy rate, the second accuracy rate, and the third accuracy rate to obtain a first accuracy rate curve, a second accuracy rate curve, and a third accuracy rate curve;

[0172] The first cycle number A is determined according to the intersection of the first accuracy curve and the second accuracy curve as the boundary between the initial prediction model and the medium-term prediction model, and the second cycle number B is determined according to the intersection of the second accuracy curve and the third accuracy curve as the boundary between the medium-term prediction model and the decline period prediction model.

[0173] In some embodiments, determining the life cycle of the energy storage system according to the number of cycles includes:

[0174] When the number of cycles is within the range of (0, A], it is determined that the energy storage system is in the initial stage of its life cycle;

[0175] When the number of cycles is within the range of (A, B], determining that the energy storage system is in the middle of its life cycle;

[0176] When the number of cycles is within the range of (B,∞], it is determined that the life cycle of the energy storage system is in a decay period.

[0177] In some embodiments, determining a target model for SOC calculation from a plurality of preset models according to the life cycle, and determining input parameters corresponding to the target model from a plurality of preset parameters, includes:

[0178] When the energy storage system is in an early stage of its life cycle, an early stage prediction model is determined as the target model, and current, voltage, and temperature are determined as the input parameters;

[0179] When the energy storage system is in the middle of its life cycle, a mid-term prediction model is determined as the target model, and current, voltage, temperature, and number of cycles are determined as the input parameters;

[0180] When the energy storage system is in a decay period in its life cycle, a decay period prediction model is determined as the target model, and current, voltage, temperature, number of cycles, internal resistance, and aging decay rate are determined as the input parameters.

[0181] For example, the above method and apparatus may be implemented in the form of a computer program. The computer program may be implemented in the form of a computer program. Figure 4 Runs on the computer device shown.

[0182] See also Figure 4 , Figure 4 This is a schematic block diagram of the structure of a computer device provided in an embodiment of the present application. The computer device can be a server or a terminal.

[0183] like Figure 4As shown, the computer device includes a processor, a memory, and a network interface connected via a system bus, wherein the memory may include a storage medium and an internal memory.

[0184] The storage medium can store an operating system and a computer program. The computer program includes program instructions, which, when executed, can enable the processor to execute any one of the energy storage system SOC calculation methods based on the entire life cycle.

[0185] The processor is used to provide computing and control capabilities and support the operation of the entire computer equipment.

[0186] The internal memory provides an environment for the operation of the computer program in the storage medium. When the computer program is executed by the processor, the processor can execute any SOC calculation method of the energy storage system based on the entire life cycle.

[0187] The network interface is used for network communication, such as sending assigned tasks, etc. Those skilled in the art will understand that Figure 4 The structure shown in the figure is only a block diagram of a part of the structure related to the solution of the present application, and does not constitute a limitation on the computer device to which the solution of the present application is applied. The specific computer device may include more or fewer components than shown in the figure, or combine certain components, or have a different component arrangement.

[0188] It should be understood that the processor may be a central processing unit (CPU), or other general-purpose processors, digital signal processors (DSP), application-specific integrated circuits (ASIC), field-programmable gate arrays (FPGA), or other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components, etc. The general-purpose processor may be a microprocessor or any conventional processor, etc.

[0189] In one embodiment, the processor is configured to execute a computer program stored in the memory to implement the following steps:

[0190] Obtaining the number of cycles of the energy storage system;

[0191] Determining the life cycle of the energy storage system according to the number of cycles, wherein the life cycle includes: an initial period of use, a mid-period of use, and a decay period;

[0192] Determining a target model for SOC calculation from a plurality of preset models according to the life cycle, and determining input parameters corresponding to the target model from a plurality of preset parameters, the preset parameters including: current, voltage, temperature, number of cycles, internal resistance, and aging attenuation rate;

[0193] The input parameters of the energy storage system are detected, and the input parameters are input into the target model to obtain the SOC parameters output by the target model.

[0194] It should be understood that the terms used in this specification are only for the purpose of describing specific embodiments and are not intended to limit the present application. As used in this specification and the appended claims, the singular forms "a", "an", and "the" are intended to include the plural forms unless the context clearly indicates otherwise.

[0195] It should also be understood that the term "and / or" used in this specification and the appended claims refers to any combination of one or more of the associated listed items and all possible combinations, including these combinations. It should be noted that, in this article, the terms "include", "comprise" or any other variations thereof are intended to cover non-exclusive inclusion, so that a process, method, article or system that includes a series of elements includes not only those elements, but also other elements that are not explicitly listed, or also includes elements that are inherent to such process, method, article or system. In the absence of further restrictions, an element defined by the sentence "including a..." does not exclude the presence of other identical elements in the process, method, article or system that includes the element.

[0196] The serial numbers of the embodiments of the present application are for description only and do not represent the advantages or disadvantages of the embodiments. The above description is only a specific embodiment of the present application, but the scope of protection of the present application is not limited thereto. Any technician familiar with the technical field can easily think of various equivalent modifications or replacements within the technical scope disclosed in this application, and these modifications or replacements should be included in the scope of protection of this application. Therefore, the scope of protection of this application should be based on the scope of protection of the claims.

Claims

1. A method for calculating the SOC of an energy storage system based on the entire life cycle, characterized in that: The method comprises: Obtaining the number of cycles of the energy storage system; Determining the life cycle of the energy storage system according to the number of cycles, wherein the life cycle includes: an initial period of use, a mid-period of use, and a decay period; Determining a target model for SOC calculation from a plurality of preset models according to the life cycle, and determining input parameters corresponding to the target model from a plurality of preset parameters, the preset parameters including: current, voltage, temperature, number of cycles, internal resistance, and aging attenuation rate; The input parameters of the energy storage system are detected, and the input parameters are input into the target model to obtain the SOC parameters output by the target model.

2. The SOC calculation method for energy storage system based on the whole life cycle according to claim 1 is characterized in that: The method further comprises: Based on the Raccoon optimization algorithm, determining training parameters corresponding to the first preset model, the second preset model, and the third preset model respectively, wherein the training parameters include: learning rate, number of hidden layer nodes, and regularization coefficient; Based on the training parameters, the first preset model, the second preset model, and the third preset model are trained to obtain an initial prediction model, a mid-term prediction model, and a decline prediction model.

3. The SOC calculation method for energy storage system based on the whole life cycle according to claim 2 is characterized in that: The method of determining the training parameters corresponding to the first preset model, the second preset model, and the third preset model based on the raccoon optimization algorithm includes: Divide the raccoon population into a first raccoon population and a second raccoon population; Randomly generate the positions of the first raccoon population and the second raccoon population in the search space to obtain the raccoon population matrix And the objective function matrix Where i = 1, 2, ..., N, j = 1, 2, ..., m, X i represents the position of the i-th raccoon in the search space, x i,j represents the value of the j-th decision variable, N represents the number of raccoon populations, m represents the number of decision variables; F represents the vector of the objective function, F i represents the objective function value obtained by the i-th raccoon; Optimizing the training parameters in the search space using the first raccoon population to obtain first training parameters; The first training parameters are optimized using the second raccoon population to obtain target training parameters.

4. The SOC calculation method for energy storage system based on the whole life cycle according to claim 2 is characterized in that: The step of optimizing the training parameters in the search space using the first raccoon population to obtain first training parameters includes: dividing the first raccoon population into a third raccoon population and a fourth raccoon population; The position of raccoons in the third raccoon population is represented based on the following formula: in, represents calculating the new position of the i-th raccoon in the first raccoon population, Indicates the j-dimensionality of the new position, Ig j represents the position of the lizard in the search space, i.e. the position of the optimal solution, j is the dimension, and r represents a random real number in [0, 1]; The position of raccoons in the fourth raccoon population is represented based on the following formula: Among them, ub j and lb j are the upper and lower bounds of the j-th decision variable, I is a randomly selected integer from the set {1, 2}, and Ig G Indicates the randomly generated position of the lizard on the ground. is its j-dimension, F Ig G is the value of its objective function; The positions of the raccoons in the third and fourth raccoon populations are updated based on the following formula: The first training parameter is obtained.

5. The SOC calculation method for energy storage system based on the whole life cycle according to claim 2 is characterized in that: The step of optimizing the first training parameters by using the second raccoon population to obtain target training parameters includes: Generate random locations of raccoons in the second raccoon population near each location of the first training parameter according to the following formula: in, is the position of the i-th raccoon in the second raccoon population, is its j-dimension, t is the iteration counter, and are the local upper and lower bounds of the j-th decision variable respectively; The positions of the raccoons in the second raccoon population are updated according to the following formula: Obtaining the target training parameters; Among them, F i P2 is the objective function.

6. The SOC calculation method for energy storage system based on the whole life cycle according to claim 2 is characterized in that: The first preset model, the second preset model, and the third preset model are trained based on the training parameters to obtain an initial prediction model, a mid-term prediction model, and a decline prediction model, including: Based on the current samples, voltage samples, and temperature samples whose cycle times are within the range of [A1, A2], the first preset model is trained to obtain an initial prediction model corresponding to the initial use period; Based on the current samples, voltage samples, temperature samples, and cycle number samples whose cycle numbers are within the range of [B1, B2], the second preset model is trained to obtain a medium-term prediction model corresponding to the medium-term use period; Based on current samples, voltage samples, temperature samples, cycle number samples, internal resistance samples, and aging attenuation rate samples whose cycle numbers are within the range of [C1, C2], the third preset model is trained to obtain a decay period prediction model corresponding to the decay period.

7. The SOC calculation method for energy storage system based on the whole life cycle according to claim 6 is characterized in that: The method further comprises: Obtaining a first accuracy rate of the initial prediction model for test samples with different numbers of cycles, a second accuracy rate of the mid-term prediction model for test samples with different numbers of cycles, and a third accuracy rate of the decay period prediction model for test samples with different numbers of cycles; Fitting the first accuracy rate, the second accuracy rate, and the third accuracy rate to obtain a first accuracy rate curve, a second accuracy rate curve, and a third accuracy rate curve; The first cycle number A is determined according to the intersection of the first accuracy curve and the second accuracy curve as the boundary between the initial prediction model and the medium-term prediction model, and the second cycle number B is determined according to the intersection of the second accuracy curve and the third accuracy curve as the boundary between the medium-term prediction model and the decline period prediction model.

8. The SOC calculation method for energy storage system based on the whole life cycle according to claim 1 is characterized in that: Determining the life cycle of the energy storage system according to the number of cycles includes: When the number of cycles is within the range of (0, A], it is determined that the energy storage system is in the initial stage of its life cycle; When the number of cycles is within the range of (A, B], determining that the energy storage system is in the middle of its life cycle; When the number of cycles is within the range of (B,∞], it is determined that the life cycle of the energy storage system is in a decay period.

9. The SOC calculation method for energy storage system based on the whole life cycle according to claim 1 is characterized in that: Determining a target model for SOC calculation from a plurality of preset models according to the life cycle, and determining input parameters corresponding to the target model from a plurality of preset parameters, includes: When the energy storage system is in an early stage of its life cycle, an early stage prediction model is determined as the target model, and current, voltage, and temperature are determined as the input parameters; When the energy storage system is in the middle of its life cycle, a mid-term prediction model is determined as the target model, and current, voltage, temperature, and number of cycles are determined as the input parameters; When the energy storage system is in a decay period in its life cycle, a decay period prediction model is determined as the target model, and current, voltage, temperature, number of cycles, internal resistance, and aging decay rate are determined as the input parameters.

10. A SOC calculation system for energy storage systems based on the entire life cycle, characterized in that: The system comprises: A cycle number acquisition module, used to obtain the cycle number of the energy storage system; A life cycle determination module, configured to determine the life cycle of the energy storage system according to the number of cycles, wherein the life cycle includes: an initial use period, a mid-use period, and a decay period; a target model determination module, configured to determine a target model for SOC calculation from a plurality of preset models according to the life cycle, and to determine input parameters corresponding to the target model from a plurality of preset parameters, the preset parameters including: current, voltage, temperature, number of cycles, internal resistance, and aging attenuation rate; The SOC prediction module is used to detect the input parameters of the energy storage system and input the input parameters into the target model to obtain the SOC parameters output by the target model.

Images