SOC estimation method, device, equipment and medium based on the charging and discharging phase of energy storage system

Abstract

This invention discloses a method, apparatus, device, and medium for estimating the State of Charge (SOC) of an energy storage system during its charging and discharging phases. The method includes: acquiring voltage, current, and temperature data for the constant-current charging, constant-voltage charging, and discharging phases of the energy storage system; determining interaction models for the constant-current charging, constant-voltage charging, and discharging phases based on the voltage, current, and temperature data and the hysteresis characteristics of the energy storage system during charging and discharging; performing unscented Kalman filtering on each interaction model to determine the first state vector estimate at time k for each interaction model, where the state vector estimate reflects the SOC state of the interaction model; and performing interactive multi-model SOC fusion estimation based on the first state vector estimates of each interaction model to determine the target state vector estimate for the energy storage system at time k+1, which reflects the SOC state of the energy storage system. This method of SOC estimation improves the accuracy of the estimation results.

Description

Technical Field

[0001] This invention relates to the field of energy storage system technology, and in particular to a method, apparatus, equipment and medium for estimating the State of Charge (SOC) of an energy storage system during the charging and discharging phases. Background Technology

[0002] The battery charging process involves first constant current charging, then constant voltage charging. During constant current charging, the current remains constant while the voltage continuously increases, transitioning to constant voltage charging once a certain voltage level is reached. During constant voltage charging, the voltage remains constant while the current decreases slowly. During discharge, the battery undergoes dynamic and complex discharge processes based on actual usage requirements. Furthermore, a significant difference exists between the SOC-OCV curves during charging and discharging. This difference is caused by hysteresis voltage, where SOC (state of charge) represents the battery's state of charge, and OCV (open circuit voltage) represents the battery's open circuit voltage. These characteristics necessitate considering the differences in voltage and current during charging and discharging, as well as the hysteresis characteristics present during charging and discharging, when estimating the battery's SOC.

[0003] Existing battery SOC estimation methods only consider the charging or discharging phases of the battery, resulting in inaccurate estimates. Furthermore, most current battery SOC estimation methods do not consider the hysteresis characteristics that occur during battery charging and discharging. This causes the same open-circuit voltage (OCV) to correspond to different SOCs during the charging and discharging phases, which also leads to inaccurate SOC estimations.

[0004] Therefore, the current methods for estimating the SOC (State of Charge) of energy storage batteries still need improvement. Summary of the Invention

[0005] The main objective of this invention is to provide a method, apparatus, device, and medium for estimating the State of Charge (SOC) of an energy storage system during the charging and discharging phases, which can solve the problem of inaccurate SOC estimation in the prior art.

[0006] To achieve the above objectives, the first aspect of the present invention provides a method for estimating the State of Charge (SOC) of an energy storage system during the charging and discharging phases, the method comprising:

[0007] Acquire voltage, current, and temperature data for the constant current charging, constant voltage charging, and discharging phases of the energy storage system;

[0008] Based on the voltage data, current data, temperature data, and the hysteresis characteristics of the energy storage system during charging and discharging, the interaction models for the constant current charging stage, the constant voltage charging stage, and the discharging stage are determined.

[0009] The interaction models are subjected to unscented Kalman filtering to determine the first state vector estimate at time k for each interaction model. The state vector estimate is used to reflect the SOC state of the interaction model.

[0010] Based on the first state vector estimates of each of the interactive models, an interactive multi-model SOC fusion estimation is performed to determine the target state vector estimate of the energy storage system at time k+1. The target state vector estimate is used to reflect the SOC state of the energy storage system.

[0011] In one feasible implementation, the step of performing unscented Kalman filtering on the interaction models to determine the first state vector estimate at time k for each interaction model includes:

[0012] The adaptive forgetting factor recursive least squares method is used to perform parameter identification processing on the interaction model to determine the model parameters of each interaction model;

[0013] Based on the model parameters of each of the interaction models, the state equation of the energy storage system is determined;

[0014] The interaction models are subjected to unscented Kalman filtering using the state equations to determine the first state vector estimate at time k for each interaction model.

[0015] In one feasible implementation, the step of applying unscented Kalman filtering to each interaction model using the state equation to determine the first state vector estimate at time k for each interaction model includes:

[0016] For each interaction model:

[0017] When k=0, the state equation is used to perform the initialization process of the unscented Kalman filter to determine the initial state vector estimate of the interaction model when k=0;

[0018] Let k = k+1 to enter the iterative process of the unscented Kalman filter:

[0019] Determine the first data sample set at time k-1, wherein the data sample set includes several state vectors at time k-1;

[0020] The first data sample set, the estimated value of the second state vector at time k-1, and the first covariance are input into the preset state update model to determine the estimated value of the third state vector at time k and the third covariance.

[0021] The second data sample set is determined by resampling the data at time k using the estimated value of the third state vector and the third covariance at time k.

[0022] Input the second data sample set, the second state vector estimate, and the third state vector estimate into a preset mean and covariance model to determine the mean and covariance of the target state vector at time k.

[0023] Based on the mean of the target state vector, the covariance of the target state vector, and the estimated value of the third state vector, determine the estimated value of the first state vector at time k of the interaction model, let k = k + 1, and return to execute the step of determining the first data sample set at time k-1, wherein the data sample set includes several state vectors at time k-1.

[0024] In one feasible implementation, determining the first state vector estimate of the interaction model at time k based on the target state vector mean, the target state vector covariance, and the third state vector estimate includes:

[0025] The gain matrix of the unscented Kalman filter at time k is determined using the mean of the target state vector, the covariance of the target state vector, and a preset gain matrix algorithm.

[0026] Using the gain matrix and the third state vector estimate, the first state vector estimate of the interaction model at time k is determined.

[0027] In one feasible implementation, the interactive multi-model includes an input interactive model and an output interactive model. The step of performing SOC fusion estimation of the interactive multi-model based on the first state vector estimates of each of the interactive models to determine the target state vector estimate of the energy storage system at time k+1 includes:

[0028] The first state vector estimate of each interaction model is input into the input interaction model for input interaction, and the fourth state vector estimate of each interaction model at time k after input interaction is determined.

[0029] The fourth state vector estimate is subjected to unscented Kalman filtering to determine the fifth state vector estimate, residual, and residual covariance of each interaction model at time k+1.

[0030] Based on the preset model probability algorithm, the residuals, and the residual covariance, the model probability of each interaction model at time k+1 is determined;

[0031] The model probabilities of each interaction model and the estimated value of the fifth state vector are input into the output interaction model for state fusion to determine the estimated value of the target state vector of the energy storage system at time k+1.

[0032] In one feasible implementation, determining the model probability of each interaction model at time k+1 based on a preset model probability algorithm, the residual, and the residual covariance includes:

[0033] The residuals and residual covariance are input into a preset likelihood function algorithm to determine the likelihood function of each interaction model at time k+1.

[0034] The model probability of each interaction model at time k+1 is determined using the likelihood function, the model probability of each interaction model at time k, and the model probability algorithm.

[0035] In one feasible implementation, the step of performing interactive multi-model SOC fusion estimation based on the first state vector estimates of each of the interactive models to determine the target state vector estimate of the energy storage system at time k+1 further includes:

[0036] Using a preset SOC state extraction algorithm and the target state vector estimate, the target state vector estimate of the energy storage system at time k+1 is determined.

[0037] To achieve the above objectives, a second aspect of the present invention provides a State of Charge (SOC) estimation device based on the charging and discharging phases of an energy storage system, the device comprising:

[0038] Data acquisition module: used to acquire voltage data, current data and temperature data of the constant current charging stage, constant voltage charging stage and discharging stage of the energy storage system;

[0039] Model building module: used to determine the interaction model of the constant current charging stage, the interaction model of the constant voltage charging stage, and the interaction model of the discharging stage based on the voltage data, current data, temperature data, and the hysteresis characteristics of the energy storage system during charging and discharging.

[0040] State estimation module: used to perform unscented Kalman filtering on the interaction model respectively, and determine the first state vector estimate of each interaction model at time k. The state vector estimate is used to reflect the SOC state of the interaction model.

[0041] State determination module: used to perform interactive multi-model SOC fusion estimation based on the first state vector estimates of each interactive model, and determine the target state vector estimate of the energy storage system at time k+1. The target state vector estimate is used to reflect the SOC state of the energy storage system.

[0042] To achieve the above objectives, a third aspect of the present invention provides a computer-readable storage medium storing a computer program, which, when executed by a processor, causes the processor to perform the steps shown in the first aspect and any feasible implementation.

[0043] To achieve the above objectives, a fourth aspect of the present invention provides a computer device including a memory and a processor, the memory storing a computer program, which, when executed by the processor, causes the processor to perform the steps shown in the first aspect and any feasible implementation.

[0044] The embodiments of the present invention have the following beneficial effects:

[0045] This invention provides a State of Charge (SOC) estimation method based on the charging and discharging stages of an energy storage system. The method includes: acquiring voltage, current, and temperature data for the constant-current charging, constant-voltage charging, and discharging stages of the energy storage system; determining interaction models for the constant-current charging, constant-voltage charging, and discharging stages based on the voltage, current, and temperature data and the hysteresis characteristics of the energy storage system during charging and discharging; performing unscented Kalman filtering on each interaction model to determine the first state vector estimate at time k for each interaction model, which reflects the SOC state of the interaction model; and performing interactive multi-model SOC fusion estimation based on the first state vector estimates of each interaction model to determine the target state vector estimate of the energy storage system at time k+1, which reflects the SOC state of the energy storage system. Through this method, multiple interaction models of the energy storage system are established, considering different charging and discharging stages and hysteresis characteristics during SOC estimation, which helps improve the accuracy of the estimation results. Furthermore, the accuracy of the estimation results is further improved through subsequent unscented Kalman filtering and interactive multi-model processing. Attached Figure Description

[0046] To more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the drawings used in the description of the embodiments or the prior art 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.

[0047] in:

[0048] Figure 1 This is a flowchart of a SOC estimation method based on the charging and discharging phase of an energy storage system in an embodiment of the present invention;

[0049] Figure 2 This is a schematic diagram of current-voltage variation curves at different charging and discharging stages in an embodiment of the present invention;

[0050] Figure 3 This is a schematic diagram of an equivalent second-order RC circuit considering hysteresis characteristics in an embodiment of the present invention;

[0051] Figure 4 This is a hysteresis curve of charging open-circuit voltage and discharging open-circuit voltage in an embodiment of the present invention;

[0052] Figure 5 This is another flowchart of a SOC estimation method based on the charging and discharging phase of an energy storage system in an embodiment of the present invention;

[0053] Figure 6 The figure shows the AFFRLS parameter identification results of an interaction model for a constant current charging stage in an embodiment of the present invention.

[0054] Figure 7 This is an AFFRLS parameter identification result diagram of an interaction model for a constant voltage charging stage in an embodiment of the present invention;

[0055] Figure 8 This is a diagram showing the AFFRLS parameter identification results of an interaction model for a constant current discharge stage in an embodiment of the present invention.

[0056] Figure 9 An AUKF-based SOC estimation graph is used for the interaction model of the constant current charging stage in an embodiment of the present invention.

[0057] Figure 10 An AUKF-based SOC estimation diagram is used for an interaction model of the constant voltage charging stage in an embodiment of the present invention.

[0058] Figure 11 An AUKF-based SOC estimation diagram is used for an interaction model of the constant current discharge stage in an embodiment of the present invention.

[0059] Figure 12 This is a block diagram of an IMM-AUKF system according to an embodiment of the present invention;

[0060] Figure 13 This is a SOC estimation fusion diagram of IMM-AUKF in an embodiment of the present invention;

[0061] Figure 14 This is a structural block diagram of a SOC estimation device based on the charging and discharging phase of an energy storage system in an embodiment of the present invention.

[0062] Figure 15 This is a structural block diagram of a computer device in an embodiment of the present invention. Detailed Implementation

[0063] The technical solutions of the embodiments of the present invention will be clearly and completely described below 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 embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0064] Please see Figure 1 , Figure 1 This is a flowchart of a SOC estimation method based on the charging and discharging phase of an energy storage system, as described in an embodiment of the present invention. Figure 1 The method shown includes the following steps:

[0065] 101. Obtain voltage data, current data, and temperature data of the constant current charging stage, constant voltage charging stage, and discharge stage of the energy storage system;

[0066] 102. Based on the voltage data, current data, temperature data, and the hysteresis characteristics of the energy storage system during charging and discharging, determine the interaction model for the constant current charging stage, the interaction model for the constant voltage charging stage, and the interaction model for the discharging stage.

[0067] It should be noted that the SOC estimation method based on the charging and discharging stages of the energy storage system presented in this application is for real-time SOC calculation of the energy storage system, so as to dynamically grasp the SOC state of the energy storage system. First, in order to calculate the SOC of the energy storage system, it is necessary to establish an equivalent model of the energy storage system. This application establishes at least three interactive models of equivalent circuits for the energy storage system according to different charging and discharging stages to improve the modeling accuracy of the equivalent model of the energy storage system. Specifically, combining the hysteresis characteristics of the energy storage system during charging and discharging, as well as the voltage, current, and temperature data of the constant current charging stage, constant voltage charging stage, and discharging stage of the energy storage system, a charging interactive model and a discharging interactive model of the energy storage system are established. The charging interactive model can include the interactive model of the constant current charging stage, the interactive model of the constant voltage charging stage, and the interactive model of the discharging stage. The energy storage system can be an energy storage battery, etc. The temperature, voltage, and current data can be obtained by collecting them through sensors installed in the energy storage system, and obtaining the temperature, voltage, and current data for each charging and discharging stage, thereby establishing the interactive model for each charging and discharging stage.

[0068] For example, taking the energy storage system as an energy storage battery, step 101 may specifically include: collecting the voltage and current of the battery during the constant current charging stage, constant voltage charging stage and discharging stage through a voltage and current acquisition device, and collecting the battery terminal temperature at each stage through a temperature sensor.

[0069] This embodiment uses a certain model of lead-carbon battery with a rated capacity of 660Ah as the experimental subject. The battery's rated voltage is 2V, the discharge cut-off voltage is 1.8V, and the maximum charging current is 300A. At 1.8V discharge, the SOC is considered to be 0. Then, it is first charged at a constant current of 0.1C (66A) until the voltage reaches 2.35V, and then charged at a constant voltage of 2.35V until the current is less than 9A, at which point the battery is considered fully charged and charging ends. The fully charged battery is then discharged at a constant current of 0.05C (33A) until the voltage reaches 1.85V, and the discharge stops when the battery SOC reaches 20%. The charging and discharging voltage and current data are collected using a Blue Battery testing system, and the temperature is measured using a DS18B20 adhesive temperature sensor. The room temperature is 25℃. The battery temperature will rise slightly during charging and discharging, but the temperature rise is not significant, so it is considered that the test temperature is constant at 25℃. (See reference...) Figure 2 , Figure 2 This is a schematic diagram of current-voltage variation curves at different charging and discharging stages in an embodiment of the present invention. Figure 2 The diagram shows the voltage and current collected during constant current charging, with the current constant at 66A and the voltage gradually increasing from 1.8V to 2.35V; during constant voltage charging, the voltage is constant at 2.35V, and the current gradually decreases from 66A to 9A, indicating the battery is fully charged; during constant current discharging, the current is constant at -33A, and the voltage gradually decreases from 2.35V to 1.95V, at which point the battery's SOC is 20%, and discharging ends.

[0070] Furthermore, step 102 specifically includes: the battery charging stage consists of constant current charging followed by constant voltage charging; in the constant current charging stage, the current is constant and the voltage continuously increases, switching to constant voltage charging when the voltage reaches 2.35V; in the constant voltage charging stage, the voltage is constant and the current slowly decreases, and the battery open-circuit voltage exhibits hysteresis characteristics during both the charging and discharging stages. Based on the different voltage and current in the constant current and constant voltage charging stages, an equivalent second-order RC circuit considering hysteresis characteristics is used to establish a battery constant current charging interaction model and a constant voltage charging interaction model with hysteresis characteristics. The battery discharging stage may involve short-term constant current, short-term constant voltage, and other complex dynamic discharge conditions; an equivalent second-order RC circuit considering hysteresis characteristics is used to establish a battery discharging interaction model with hysteresis characteristics. Please refer to [link to relevant documentation]. Figure 3 , Figure 3 This is a schematic diagram of an equivalent second-order RC circuit considering hysteresis characteristics in an embodiment of the present invention. Figure 3 An equivalent second-order RC circuit considering hysteresis characteristics based on electrochemical impedance spectroscopy is shown. The circuit consists of the battery equilibrium potential EMF and the hysteresis voltage U. H To form a voltage source to replace the open-circuit voltage U of the battery OCV(SOC,H,T) The ohmic internal resistance of the energy storage battery is replaced by a resistor R0, and the polarization resistance and polarization capacitor of the equivalent circuit are replaced by a second-order RC circuit.

[0071] The state equation and observation equation of the energy storage system are as follows:

[0072]

[0073] In the formula,

[0074]

[0075] τ1=R1C1;

[0076] τ2=R2C2;

[0077] Among them, the open circuit voltage U OCV(SOC,H,T) It can be represented as:

[0078] U OCV(SOC,H,T) =EMF+U H

[0079] In the formula, the hysteresis voltage U H The formula for calculating the equilibrium potential (EMF) is as follows:

[0080]

[0081] In the formula, η takes values ​​from 0.5 to 1, U charge U is the equilibrium terminal voltage of the battery during charging. discharge η is the equilibrium terminal voltage of the battery during discharge; in this embodiment, η is taken as 0.5.

[0082] Please refer to Figure 4 , Figure 4 This is a hysteresis curve of charging open-circuit voltage and discharging open-circuit voltage in an embodiment of the present invention. Figure 4 As can be seen from the charge / discharge hysteresis voltage curves, under the same SOC, the open-circuit voltage of the battery during the charging phase is significantly higher than that during the discharging phase. In other words, for the same open-circuit voltage, the SOC during charging is significantly lower than the SOC during discharging, with the SOC difference reaching up to 30%. This demonstrates that the battery's hysteresis characteristics have a significant impact on the accuracy of SOC estimation. Therefore, hysteresis characteristics must be considered when charging and discharging batteries, requiring separate modeling for charging and discharging, and the use of different SOC-OCV curves for parameter identification and SOC estimation.

[0083] 103. Perform unscented Kalman filtering on the interaction models respectively to determine the first state vector estimate at time k for each interaction model. The state vector estimate is used to reflect the SOC state of the interaction model.

[0084] After obtaining the interaction model for each charging and discharging stage, the State of Charge (SOC) of the interaction model can be estimated using unscented Kalman filtering (AUKF filtering). Specifically, unscented Kalman filtering is applied to each interaction model to determine the first state vector estimate at time k for each interaction model. This state vector estimate reflects the SOC state of the interaction model. Through AUKF filtering, the state vector estimates at time k for the constant voltage charging stage, the constant current charging stage, and the discharging stage can be obtained. Since the state vector is a matrix composed of state estimates for multiple electrical parameters, the state vector X with three electrical parameters is used as an example. k For example, X k It is a three-parameter matrix composed of U1, U2 and SOC. Therefore, the state vector estimate can be used to reflect the SOC state of the interaction model.

[0085] It should be noted that the terms "first," "second," etc., in the specification, claims, and drawings of this application are used to distinguish different objects, rather than to describe a specific order.

[0086] 104. Based on the first state vector estimates of each of the interactive models, perform interactive multi-model SOC fusion estimation to determine the target state vector estimate of the energy storage system at time k+1. The target state vector estimate is used to reflect the SOC state of the energy storage system.

[0087] After obtaining the first state vector estimate of each interaction model at time k, the first state vector estimate of each interaction model and the preset Interacting Multiple Model (IMM) can be used to perform SOC fusion processing of the interactive multiple model to determine the target state vector estimate of the energy storage system at time k+1, that is, the SOC state of the energy storage system at time k+1. In this way, the estimated SOCs of multiple equivalent interaction models can be fused in real time, realizing real-time calculation of the SOC of the energy storage system.

[0088] The core idea of ​​the interactive multi-model control algorithm is based on Bayesian theory, which proposes automatic identification and switching between models: at any tracking moment, real-time maneuvering model detection is performed by setting model filters corresponding to the number of possible target models. Weight coefficients and model update probabilities are set for each filter, and finally, the current optimal estimated state is calculated by weighted calculation, thereby achieving the purpose of adaptive model tracking.

[0089] Therefore, by using the interactive multi-model (IMM) method to fuse and calculate the SOC of multiple interactive models, the real-time SOC of the energy storage battery can be determined with higher accuracy, better extending battery life and better maintaining the safe and stable operation of the power grid.

[0090] This invention provides a State of Charge (SOC) estimation method based on the charging and discharging stages of an energy storage system. The method includes: acquiring voltage, current, and temperature data for the constant-current charging, constant-voltage charging, and discharging stages of the energy storage system; determining interaction models for the constant-current charging, constant-voltage charging, and discharging stages based on the voltage, current, and temperature data and the hysteresis characteristics of the energy storage system during charging and discharging; performing unscented Kalman filtering on each interaction model to determine the first state vector estimate at time k for each interaction model, which reflects the SOC state of the interaction model; and performing interactive multi-model SOC fusion estimation based on the first state vector estimates of each interaction model to determine the target state vector estimate of the energy storage system at time k+1, which reflects the SOC state of the energy storage system. Through this method, multiple interaction models of the energy storage system are established, considering different charging and discharging stages and hysteresis characteristics during SOC estimation, which helps improve the accuracy of the estimation results. Furthermore, the accuracy of the estimation results is further improved through subsequent unscented Kalman filtering and interactive multi-model processing.

[0091] Please see Figure 5 , Figure 5 This is another flowchart of a SOC estimation method based on the charging and discharging phase of an energy storage system, as described in an embodiment of the present invention. Figure 5 The method shown includes the following steps:

[0092] 501. Obtain voltage data, current data, and temperature data of the constant current charging stage, constant voltage charging stage, and discharge stage of the energy storage system;

[0093] 502. Based on the voltage data, current data, temperature data, and the hysteresis characteristics of the energy storage system during charging and discharging, determine the interaction model for the constant current charging stage, the interaction model for the constant voltage charging stage, and the interaction model for the discharging stage.

[0094] It should be noted that the content of steps 501 and 502 is the same as... Figure 1 Steps 101 and 102 in the method shown are similar, and will not be repeated here to avoid repetition. Please refer to the previous section for details. Figure 1 The contents of steps 101 and 102 in the method shown.

[0095] 503. The interaction model is subjected to parameter identification processing using the adaptive forgetting factor recursive least squares method to determine the model parameters of each interaction model;

[0096] It should be noted that before filtering, the model parameters of each interaction model need to be identified. The model parameters of each interaction model are determined by using the adaptive forgetting factor recursive least squares method (AFFRLS).

[0097] For example, using the collected voltage, current, and temperature data, the interactive models for the constant-current charging stage, the constant-voltage charging stage, and the discharging stage of a battery considering hysteresis are established. Adaptive Forgetting Factor Recursive Least Squares (AFFRLS) is then used for real-time parameter identification. The calculation formula can be derived as follows:

[0098]

[0099] In the formula, λ is the adaptive forgetting factor, l and r are adjustment coefficients, and e 2 (k+1) represents the squared value of the estimation error. θ k As input to the parameter identification system, Let y(k+1) be the observed value of the system at time k+1, y(k+1) be the actual feedback value of the system at time k+1, K(k+1) be the system gain, P(k+1) be the system covariance matrix, E be the identity matrix, and T be the transpose. In this embodiment, l is set to 1 and r is set to 0.04.

[0100] Please see Figure 6-8 ,in, Figure 6 The figure shows the AFFRLS parameter identification results of an interaction model for a constant current charging stage in an embodiment of the present invention. Figure 7 This is an AFFRLS parameter identification result diagram of an interaction model for a constant voltage charging stage in an embodiment of the present invention; Figure 8 The figure shows the AFFRLS parameter identification results of an interaction model for a constant current discharge stage in an embodiment of the present invention. A comparison of the three figures shows that the battery internal resistance R0, polarization internal resistance R1, concentration internal resistance R2, polarization capacitance C1, and concentration capacitance C2 are different in each stage of the constant current, constant voltage charging, and discharge stages. The internal resistance at the end of constant current charging is basically close to the internal resistance at the beginning of constant voltage charging.

[0101] 504. Determine the state equation of the energy storage system based on the model parameters of each of the interaction models;

[0102] 505. Using the state equation, perform unscented Kalman filtering on the interaction model respectively to determine the first state vector estimate at time k for each interaction model. The state vector estimate is used to reflect the SOC state of the interaction model.

[0103] After obtaining the model parameters of each interaction model, the SOC state of the interaction model can be estimated. First, the state equation of the energy storage system is determined based on the model parameters of the interaction model. Then, the first state vector estimate of the interaction model at time k is determined by AUKF filtering. The identified model parameters are then used to perform SOC prediction on the constant current charging interaction model with hysteresis, the constant voltage charging interaction model with hysteresis, and the discharging interaction model with hysteresis using AUKF filtering.

[0104] For example, the system's state equations and observation equations are shown below:

[0105]

[0106] X k =[SOC k U 1k U 2k ] T ;

[0107] In the formula, X k SOC represents the estimated state vector at time k. k U is the estimated value of the SOC state at time k. 1k Let U be the state estimate of voltage U1 at time k. 2k Let U be the state estimate of voltage U2 at time k, and T be the transpose. k-1 Let B be the system state transition matrix at time k-1. k-1 W is the system control matrix at time k-1. k-1 Let C be the system process noise at time k-1. k Let D be the observation matrix of the system at time k. k Let V be the direct matrix at time k. k Let X be the noise observed at time k. k It can also be viewed as the state variable at time k, u k Input variables at time k.

[0108] Furthermore, for each interaction model, step 505 may include:

[0109] A1. When k=0, use the state equation to perform the initialization process of unscented Kalman filtering to determine the initial state vector estimate of the interaction model when k=0;

[0110] Initialize and set the state variables and covariance matrix in step A1, referring to the following formula:

[0111]

[0112] In the formula, X0 is the state vector when k = 0. Let be the estimated value of the initial state vector when k = 0, and To obtain the expected state vector estimate; P0 is the covariance when k=0, E represents the expectation, and T is the transpose.

[0113] A2. Let k = k + 1 to enter the iterative process of the unscented Kalman filter:

[0114] A3. Determine the first data sample set at time k-1, wherein the data sample set includes several state vectors at time k-1;

[0115] It should be noted that the state estimation at the current time is performed by using the prior probability of the previous time step. Then, the data at time step k-1 needs to be sampled using the estimated value of the second state vector at time step k-1 and the first covariance to determine the first data sample set. This data sample set includes several state vectors at time step k-1, that is, expanding one data point into N data points. This data sample set can be called the sigma (σ) sample, which includes 2n+1 σ points at time step k-1.

[0116] Specifically, using the second state vector estimate at time k-1 and the first covariance P k-1 Calculate the 2n+1 σ points at time k-1, which are the sigma samples.

[0117]

[0118] In the formula: Represents matrix (n+λ)P k-1 The i-th column has a total length of n. This is the estimated state vector value at time k-1. This is the updated state vector estimate at time k-1, where i is the data point number.

[0119] A4. Input the first data sample set, the estimated value of the second state vector at time k-1, and the first covariance into the preset state update model to determine the estimated value of the third state vector and the third covariance at time k.

[0120] Furthermore, the state vector at time k is estimated using data points in the first sample dataset. By inputting the first data sample set, the estimated second state vector, and the first covariance into a preset state update model, the estimated third state vector and the third covariance at time k are determined.

[0121] For example, the estimated value of the third state vector at time k is calculated based on the sampling points of the sample set in step A3. and the third covariance P k / k-1 One step of predicting model values, complete and P k / k-1 The update process, wherein the state update model is based on the following formula:

[0122]

[0123] In the formula: Q k-1 This is the process noise variance matrix; The expected weights, The variance is the weight.

[0124]

[0125]

[0126] Where: ε i For voltage information, H k Let L be the voltage information covariance matrix at time k, and L be the time window size of the information covariance.

[0127] A5. Use the estimated value of the third state vector at time k and the third covariance to resample the data at time k and determine the second data sample set;

[0128] Finally, the sample set can be updated by resampling the samples using the estimate at time k. Specifically, this is done using the estimate of the third state vector at time k. and the third covariance P k / k-1 Calculate further predictions for the 2n+1 σ points at time k.

[0129]

[0130] A6. Input the second data sample set, the second state vector estimate, and the third state vector estimate into the preset mean and covariance model to determine the mean and covariance of the target state vector at time k.

[0131] The second data sample set, the second state vector estimate, and the third state vector estimate are input into a preset mean and covariance model. The mean and covariance matrix are calculated by weighting the σ points predicted in step A5. The preset mean and covariance model is based on the following formula:

[0132]

[0133]

[0134]

[0135] In the formula, as well as Both are covariance matrices at time k. as well as All are the mean values ​​at time k. Let i be the state vectors at time k; Let k be the state vector at time k; Let i be the observation vectors at time k; Let be the observation vector at time k, and T be the transpose.

[0136] A7. Based on the mean of the target state vector, the covariance of the target state vector, and the estimated value of the third state vector, determine the estimated value of the first state vector at time k of the interaction model, let k = k + 1, and return to execute the step of determining the first data sample set at time k-1, wherein the data sample set includes several state vectors at time k-1.

[0137] Furthermore, based on the mean of the target state vector, the covariance of the target state vector, and the estimated value of the third state vector, the estimated value of the first state vector at time k of the interaction model is determined. For example, this may include steps B1-B2:

[0138] B1. Using the mean of the target state vector, the covariance of the target state vector, and a preset gain matrix algorithm, determine the gain matrix of the unscented Kalman filter at time k.

[0139] B2. Using the gain matrix and the third state vector estimate, determine the first state vector estimate of the interaction model at time k.

[0140] That is, the AUKF gain matrix is ​​calculated using the mean of the target state vector, the covariance of the target state vector, and the estimated value of the third state vector, and the state vector is updated and iterated using the gain matrix.

[0141] For example, the gain matrix algorithm is as follows:

[0142]

[0143] In the formula, K k Let be the gain matrix at time k. as well as All are covariance matrices.

[0144] Finally, the battery state is updated and the estimated covariance is updated, as shown in the following formula:

[0145]

[0146] In the formula, P is the estimated value of the first state vector at time k. k Let be the covariance estimate at time k.

[0147] Please refer to 9-11. Figure 9 An AUKF-based SOC estimation graph is used for the interaction model of the constant current charging stage in an embodiment of the present invention. Figure 10 An AUKF-based SOC estimation diagram is used for an interaction model of the constant voltage charging stage in an embodiment of the present invention. Figure 11 In this embodiment of the invention, an interactive model for the constant current discharge stage is used to perform SOC estimation using AUKF. In this embodiment, n is selected as 2 and L as 4. SOC estimation is performed on the three interactive models respectively. Figure 9 , Figure 10 and Figure 11 As can be seen from the figure, the adaptive unscented Kalman filter algorithm has a smaller error in estimating the SOC of the battery than the ampere-hour integration method, and their errors are all within ±0.005, indicating higher estimation accuracy.

[0148] 506. Based on the first state vector estimates of each of the interactive models, perform interactive multi-model SOC fusion estimation to determine the target state vector estimate of the energy storage system at time k+1. The target state vector estimate is used to reflect the SOC state of the energy storage system.

[0149] It should be noted that step 506 and Figure 1 Step 104 in the method shown is similar, and to avoid repetition, it will not be repeated here. For details, please refer to the aforementioned text. Figure 1 The content of step 104 in the method shown.

[0150] Please see Figure 12 , Figure 12 This is a block diagram of an IMM-AUKF system in an embodiment of the present invention. In this system, an interactive multi-model (IMM) is used to perform SOC fusion estimation on a battery constant-current charging interaction model with hysteresis, a constant-voltage charging interaction model with hysteresis, and a discharging interaction model with hysteresis, to obtain the fused estimated SOC. The interactive multi-model includes an input interaction model and an output interaction model. Therefore, step 506 may include steps C1-C4:

[0151] C1. Input the first state vector estimate of each interaction model into the input interaction model to perform input interaction, and determine the fourth state vector estimate of each interaction model at time k after input interaction;

[0152] Step 1: Input interaction, which involves inputting the state vector into the input interaction model to initialize or reinitialize the model conditions, thereby obtaining the state vectors and covariance matrices of each AUKF filter input at the current time step. For model j with period k, its input interaction model is as follows:

[0153]

[0154]

[0155]

[0156]

[0157] In the formula, π represents the state estimate of model j at time k; lj P represents the probability of switching from model j to model l. 0j (k|k) represents the value of the covariance matrix of model j at time k; u j (k) represents the model probability of model j at time k, and m represents the total number of models.

[0158] C2. Perform unscented Kalman filtering on the estimated fourth state vector to determine the estimated fifth state vector, residual, and residual covariance of each interaction model at time k+1.

[0159] Furthermore, for the at least three models established in step 502, let j = 1, ..., m, and use the battery voltage, current, terminal temperature, and the fourth state vector estimate of model j at time k. and covariance P 0j (k|k) is used as input for AUKF filtering, utilizing the time step k. and P 0j (k|k) yields an estimate of the state vector at the next time step k+1. and its covariance estimator P 0j (k+1|k+1), residual v j (k+1) and residual covariance S j (k+1).

[0160] C3. Determine the model probability of each interaction model at time k+1 based on the preset model probability algorithm, the residual, and the residual covariance.

[0161] Step C3 may include steps D1-D2:

[0162] D1. Input the residuals and residual covariance into a preset likelihood function algorithm to determine the likelihood function of each interaction model at time k+1.

[0163] D2. Using the likelihood function, the model probability of each interaction model at time k, and the model probability algorithm, determine the model probability of each interaction model at time k+1.

[0164] For example, the model probability update refers to the following:

[0165] The likelihood function of the j-th model can be expressed as:

[0166]

[0167] In the formula:

[0168] residual

[0169] Residual Covariance

[0170] In the formula, Z k+1 Let be the observation value (observation vector) at time k+1. P is the state vector of model j at time k; j (k|k) represents the covariance of model j at time k, and T is the transpose.

[0171] Specifically, the mixture probability (model probability) of model j at time k+1 is calculated as follows:

[0172]

[0173]

[0174] In the formula, u j (k+1) represents the model probability of model j at time k+1, Λ j (k+1) is the likelihood function of model j at time k+1; u j (k) represents the model probability of model j at time k.

[0175] C4. Input the model probabilities of each interaction model and the estimated value of the fifth state vector into the output interaction model to perform state fusion and determine the estimated value of the target state vector of the energy storage system at time k+1.

[0176] Finally, the output interaction is performed, where the state vector is input into the output interaction model, and the model is fused with its corresponding probabilities to calculate the state estimate at time k+1. With covariance P(k+1), where the output interaction model is given by the following formula:

[0177]

[0178]

[0179] Furthermore, the target state vector estimate of the energy storage system at time k+1 is determined, and the SOC state can then be extracted from the state vector. Therefore, the target state vector estimate of the energy storage system at time k+1 can also be determined using a preset SOC state extraction algorithm and the target state vector estimate.

[0180] For example, the following SOC state extraction algorithm outputs the fused SOC:

[0181]

[0182] In the formula, SOC ksum For the merged SOC, This is the estimated value of the target state vector.

[0183] In this embodiment, model j is 3. The IMM-AUKF method is used to fuse and estimate the SOC of the battery constant-current charging interaction model with hysteresis, the constant-voltage charging interaction model with hysteresis, and the discharging interaction model with hysteresis. The SOC waveform and error output after fusion are as follows: Figure 13 As shown, Figure 13 This is a SOC estimation fusion map of IMM-AUKF in an embodiment of the present invention, such as... Figure 13 The SOC output after fusion is shown to be basically consistent with the true value. The error curve shows that the error of the IMM-AUKF fusion output is less than ±0.001, and its SOC estimation is very accurate.

[0184] This invention provides a State of Charge (SOC) estimation method based on the charging and discharging stages of an energy storage system. The method offers the following advantages in battery SOC estimation: For the battery charging stage, which involves constant current charging followed by constant voltage charging, and the discharging stage, which involves complex discharging based on actual needs, the open-circuit voltage of the battery exhibits hysteresis characteristics during charging and discharging. An equivalent second-order RC circuit with hysteresis is used to establish hysteresis-based constant current charging and constant voltage charging interaction models, and a hysteresis-based discharging interaction model is established for the discharging stage. Establishing different interaction models allows for a better reflection of the battery's state. Then, the adaptive forgetting factor recursive least squares (AFFRLS) method is used to identify the parameters of each model. Next, an AUKF filter is used to estimate the SOC of the three interaction models. AUKF uses the idea of ​​probability density distribution to solve the problem of nonlinear systems, offering higher accuracy compared to the ampere-hour integral method. Finally, an interactive multiple model (IMM) is used to fuse the SOC estimation of the hysteresis-based constant current charging interaction model, the hysteresis-based constant voltage charging interaction model, and the hysteresis-based discharging interaction model, yielding the fused SOC. This will allow for accurate estimation of the State of Charge (SOC) during both charging and discharging phases of the battery.

[0185] Please see Figure 14 , Figure 14 This is a structural block diagram of a SOC estimation device based on the charging and discharging phase of an energy storage system, as described in an embodiment of the present invention. Figure 14 The apparatus shown includes:

[0186] Data acquisition module 1401: used to acquire voltage data, current data and temperature data of the constant current charging stage, constant voltage charging stage and discharging stage of the energy storage system;

[0187] Model building module 1402: used to determine the interaction model of the constant current charging stage, the interaction model of the constant voltage charging stage, and the interaction model of the discharging stage based on the voltage data, current data, temperature data, and the hysteresis characteristics of the energy storage system during charging and discharging.

[0188] State estimation module 1403: used to perform unscented Kalman filtering on the interaction models respectively, and determine the first state vector estimate value of each interaction model at time k, wherein the state vector estimate value is used to reflect the SOC state of the interaction model;

[0189] State determination module 1404: used to perform interactive multi-model SOC fusion estimation based on the first state vector estimates of each interactive model, and determine the target state vector estimate of the energy storage system at time k+1. The target state vector estimate is used to reflect the SOC state of the energy storage system.

[0190] It should be noted that, Figure 14The functions of each module in the device shown are as follows: Figure 1 The steps in the method shown are similar, and will not be repeated here to avoid repetition. Please refer to the previous sections for details. Figure 1 The steps in the method shown are as follows.

[0191] This invention provides a State of Charge (SOC) estimation device based on the charging and discharging stages of an energy storage system. The device includes: a data acquisition module for acquiring voltage, current, and temperature data during the constant-current charging, constant-voltage charging, and discharging stages of the energy storage system; a model building module for determining interaction models for the constant-current charging, constant-voltage charging, and discharging stages based on the voltage, current, and temperature data and the hysteresis characteristics of the energy storage system during charging and discharging; a state estimation module for performing unscented Kalman filtering on the interaction models to determine the first state vector estimate at time k for each interaction model, the state vector estimate reflecting the SOC state of the interaction model; and a state determination module for performing interactive multi-model SOC fusion estimation based on the first state vector estimates of each interaction model to determine the target state vector estimate of the energy storage system at time k+1, the target state vector estimate reflecting the SOC state of the energy storage system. Through the above methods, multiple interactive models of the energy storage system were established. Different charging and discharging stages and hysteresis characteristics were considered in the SOC estimation process, which helps to improve the accuracy of the estimation results. Furthermore, the accuracy of the estimation results was further improved through subsequent unscented Kalman filtering and interactive multi-model processing.

[0192] Figure 15 An internal structural diagram of a computer device in one embodiment is shown. This computer device can specifically be a terminal or a server. Figure 15 As shown, the computer device includes a processor, memory, and a network interface connected via a system bus. The memory includes a non-volatile storage medium and internal memory. The non-volatile storage medium stores an operating system and may also store a computer program, which, when executed by the processor, causes the processor to perform the aforementioned methods. The internal memory may also store a computer program, which, when executed by the processor, causes the processor to perform the aforementioned methods. Those skilled in the art will understand that… Figure 15 The structure shown is merely a block diagram of a portion of the structure related to the present application and does not constitute a limitation on the computer device to which the present application is applied. Specific computer devices may include more or fewer components than those shown in the figure, or combine certain components, or have different component arrangements.

[0193] In one embodiment, a computer device is provided, including a memory and a processor, the memory storing a computer program that, when executed by the processor, causes the processor to perform actions such as... Figure 1 or Figure 5 The steps of the method shown.

[0194] In one embodiment, a computer-readable storage medium is provided storing a computer program that, when executed by a processor, causes the processor to perform the following actions: Figure 1 or Figure 5 The steps of the method shown.

[0195] Those skilled in the art will understand that all or part of the processes in the above embodiments can be implemented by a computer program instructing related hardware. The program can be stored in a non-volatile computer-readable storage medium, and when executed, it can include the processes of the embodiments described above. Any references to memory, storage, databases, or other media used in the embodiments provided in this application can include non-volatile and / or volatile memory. Non-volatile memory can include read-only memory (ROM), programmable ROM (PROM), electrically programmable ROM (EPROM), electrically erasable programmable ROM (EEPROM), or flash memory. Volatile memory can include random access memory (RAM) or external cache memory. By way of illustration and not limitation, RAM is available in various forms, such as static RAM (SRAM), dynamic RAM (DRAM), synchronous DRAM (SDRAM), dual data rate SDRAM (DDRSDRAM), enhanced SDRAM (ESDRAM), synchronous link DRAM (SLDRAM), RAMbus direct RAM (RDRAM), direct memory bus dynamic RAM (DRDRAM), and RAMbus dynamic RAM (RDRAM), etc.

[0196] The technical features of the above embodiments can be combined in any way. For the sake of brevity, not all possible combinations of the technical features in the above embodiments are described. However, as long as there is no contradiction in the combination of these technical features, they should be considered to be within the scope of this specification.

[0197] The embodiments described above are merely illustrative of several implementation methods of this application, and while the descriptions are specific and detailed, they should not be construed as limiting the scope of this patent application. It should be noted that those skilled in the art can make various modifications and improvements without departing from the concept of this application, and these all fall within the protection scope of this application. Therefore, the protection scope of this patent application should be determined by the appended claims.

Claims

1. A SOC estimation method based on the charge and discharge stages of an energy storage system, characterized in that, The method comprises: obtaining voltage data, current data and temperature data of the energy storage system in a constant-current charging stage, a constant-voltage charging stage and a discharging stage; determining an interaction model of the constant-current charging stage, an interaction model of the constant-voltage charging stage and an interaction model of the discharging stage based on the voltage data, the current data, the temperature data and the hysteresis characteristics of the energy storage system during charging and discharging; respectively, and a first state vector estimation value of each of the interaction models at the moment is determined, the first state vector estimation value being used to reflect the SOC state of the interaction model. k respectively, and a first state vector estimation value of each of the interaction models at the moment is determined, the first state vector estimation value being used to reflect the SOC state of the interaction model. The SOC fusion estimation of the interactive multiple model is performed according to the first state vector estimation value of each interactive model, and it is determined k a target state vector estimation value of the energy storage system at the time t+1, the target state vector estimation value being used to reflect the SOC state of the energy storage system; The method comprises: performing unscented Kalman filtering on the interaction models respectively to determine first state vector estimation values of the interaction models at each moment, and determining the interaction model with the minimum first state vector estimation value as the interaction model with the minimum state vector estimation value at the moment. k The method further comprises: determining the interaction model with the minimum first state vector estimation value as the interaction model with the minimum state vector estimation value at the moment. performing parameter identification processing on the interaction models by using an adaptive forgetting factor recursive least square method to determine model parameters of each of the interaction models; determining a state equation of the energy storage system according to the model parameters of each of the interaction models; respectively, to determine a first state vector estimation value of each of the interaction models at the time instant k t. The interactive multi-model includes an input interaction model and an output interaction model, and the SOC fusion estimation of the interactive multi-model according to the first state vector estimation value of each interaction model is determined k a target state vector estimation value of the energy storage system at the first moment, comprising: inputting the first state vector estimation value of each of the interaction models into the input interaction model for input interaction, determining the fourth state vector estimation value of each of the interaction models after input interaction k the moment performing unscented Kalman filtering on the fourth state vector estimation value to determine k a fifth state vector estimation value, a residual and a residual covariance of each of the interaction models at the time point According to the preset model probability algorithm, the residual error and the residual error covariance, determine k The model probability of each interaction model at the time t+1. The model probability of each of the interaction models and the fifth state vector estimation value are input into the output interaction model for state fusion to determine k The target state vector estimation value of the energy storage system at the +1 moment.

2. The method of claim 1, wherein, determining a first state vector estimation value of each of the interaction models at the time instant using the state equation respectively, comprises: k determining a first state vector estimation value of each of the interaction models at the time instant using the state equation respectively, comprises: for each interaction model: In k an initialization process of the unscented Kalman filter using the state equation when t = 0, determining k an initial state vector estimate of the interaction model when t = 0; Let k = k +1 into the iteration process of the unscented Kalman filter: determining k a first data sample set at time 1, the data sample set comprising a number of k a state vector at time 1; The first data sample set, k The second state vector estimation value at the first moment and the first covariance input a preset state update model to determine k The third state vector estimation value at the first moment and the third covariance Utilizing k a third state vector estimate and a third covariance at the time instant k resampling the data, a second data sample set is determined; inputting the second data sample set, the second state vector estimation value and the third state vector estimation value into a preset mean and covariance model, determining k a target state vector mean and a target state vector covariance at the time instant determining the interaction model based on the target state vector mean, the target state vector covariance, and a third state vector estimate k the first state vector estimate at time instant k = k +1, and returning to perform the determining k the first data sample set at time instant k the state vectors at time instant 3. The method of claim 2, wherein, determining the interaction model according to the target state vector mean value, the target state vector covariance and the third state vector estimation value k the first state vector estimation value at the moment, comprising: The gain matrix of the unscented Kalman filter at the moment is determined by using the target state vector mean value, the target state vector covariance and a preset gain matrix algorithm. k the gain matrix of the unscented Kalman filter at the moment determining a first state vector estimate at the time instant using the gain matrix and a third state vector estimate. k determining a first state vector estimate at the time instant using the gain matrix and a third state vector estimate.

4. The method of claim 1, wherein, The model probability of each interaction model at the time t is determined according to the preset model probability algorithm, the residual, and the residual covariance. k +1 inputting the residual and the residual covariance into a preset likelihood function algorithm to determine k the likelihood function of each interaction model at the time point t using the likelihood function, k the model probabilities of each of the interaction models at the time instant and the model probability algorithm, determine k the model probabilities of each of the interaction models at the time instant.

5. The method according to claim 1, wherein the SOC fusion estimation of interactive multi-models is performed based on the first state vector estimates of each of the interactive models to determine... k The target state vector estimate of the energy storage system at time +1 is then included, followed by: Using a pre-defined SOC state extraction algorithm and the estimated target state vector, determine k The target state vector estimate of the energy storage system at time +1.

6. A SOC estimation device based on the charge and discharge phases of an energy storage system, characterized by, The device comprises: a data acquisition module configured to obtain voltage data, current data and temperature data of the energy storage system in a constant-current charging stage, a constant-voltage charging stage and a discharging stage; a model establishment module configured to determine an interaction model of the constant-current charging stage, an interaction model of the constant-voltage charging stage and an interaction model of the discharging stage based on the voltage data, the current data, the temperature data and the hysteresis characteristics of the energy storage system during charging and discharging; The state estimation module is configured to respectively perform unscented Kalman filtering on the interaction models to determine first state vector estimation values of the interaction models at the moment, the first state vector estimation values being used to reflect SOC states of the interaction models. k The state estimation module is configured to respectively perform unscented Kalman filtering on the interaction models to determine first state vector estimation values of the interaction models at the moment, the first state vector estimation values being used to reflect SOC states of the interaction models. The state determination module is configured to determine a state of the energy storage system according to a SOC fusion estimation of the interactive multiple models based on the first state vector estimation value of each of the interactive models. k a target state vector estimation value of the energy storage system at the first moment, the target state vector estimation value being used to reflect a SOC state of the energy storage system. The state estimation module is specifically configured to: perform parameter identification processing on the interaction models by using an adaptive forgetting factor recursive least square method, to determine model parameters of the interaction models; determine state equations of the energy storage system according to the model parameters of the interaction models; and perform unscented Kalman filtering on the interaction models respectively by using the state equations, to determine first state vector estimation values of the interaction models at the time instant. k the time instant. The interactive multi-models include input interaction models and output interaction models, and the state determination module is specifically configured to: input the first state vector estimation value of each interaction model into the input interaction model for input interaction, determine the fourth state vector estimation value of each interaction model at time t+1 after input interaction; k perform unscented Kalman filtering processing on the fourth state vector estimation value, determine the fifth state vector estimation value, the residual and the residual covariance of each interaction model at time t+1; k perform unscented Kalman filtering processing on the fourth state vector estimation value, determine the fifth state vector estimation value, the residual and the residual covariance of each interaction model at time t+1; k determine the model probability of each interaction model at time t+1 according to a preset model probability algorithm, the residual and the residual covariance; and input the model probability and the fifth state vector estimation value of each interaction model into the output interaction model for state fusion, and determine the target state vector estimation value of the energy storage system at time t+1. k determine the model probability of each interaction model at time t+1 according to a preset model probability algorithm, the residual and the residual covariance; and input the model probability and the fifth state vector estimation value of each interaction model into the output interaction model for state fusion, and determine the target state vector estimation value of the energy storage system at time t+1.

7. A computer readable storage medium storing a computer program, characterized in that, The computer program is executed by the processor to enable the processor to perform the steps of the method according to any one of claims 1 to 5. 8.A computer device, comprising a memory and a processor, and characterized in that, The memory stores a computer program, and the computer program is executed by the processor to enable the processor to perform the steps of the method according to any one of claims 1 to 5.

Images