Soc estimation method and system for balancing charge and discharge control of large-scale energy storage power station

Abstract

The application provides a SOC estimation method and system for equalization charge-discharge control of large-scale energy storage power station, relates to the technical field of battery management system, and collects working parameters of single batteries, establishes a state model and an observation model by using a Kalman filtering algorithm, and predicts the SOC of the single battery at the next time step; collects historical working parameters and SOC of sample batteries, generates a sample data set; corrects the predicted SOC by fusing the sample data set, sets a single battery state of charge threshold, generates an effective single battery set, calculates the predicted SOC of a battery module; sets a module state of charge threshold, selects effective battery modules, calculates the predicted SOC of a battery cluster; sets a battery cluster state of charge threshold, selects effective battery clusters, and generates a power station available state of charge; according to the effective single battery set, the effective module set and the effective battery cluster set, controls the on-off of a battery discharge circuit, inputs the power station available state of charge into a battery management system, and adjusts a discharge strategy in real time.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of battery management system, in particular to a SOC estimation method and system for equalization charging and discharging control of large-scale energy storage power station. BACKGROUND

[0002] With the widespread application of renewable energy and the increasing demand for grid stability, large-scale energy storage power stations, as an important part of energy systems, have important significance for the safe and stable operation of power systems. In the operation and management of energy storage power stations, accurate estimation of the state of charge of the battery is a key technology for realizing equalization charging and discharging control, directly affecting the service life and operating efficiency of the energy storage system.

[0003] Currently, the estimation methods of battery SOC in energy storage power stations mainly include open-circuit voltage method, ampere-hour integral method, Kalman filter method, etc. Among them, the Kalman filter method has been widely used in the field of battery SOC estimation because it can effectively handle system noise and measurement errors. For example, the prior art disclosed in CN114239463A discloses a battery cluster state of charge correction method based on big data. This method establishes an equivalent circuit model of the battery cluster, identifies the model parameters using an adaptive and acoustic search algorithm, and corrects the battery cluster SOC using an unscented Kalman filter algorithm, thereby improving the estimation accuracy of the SOC. In terms of multi-level SOC aggregation, CN113608130A proposes a battery cluster state of charge online estimation method. This method first calculates the single battery state of charge, then calculates the battery stack state of charge, and finally calculates the battery cluster state of charge by combining the single battery state of charge and the battery stack state of charge, thereby realizing real-time hierarchical calculation of the battery stack, battery cluster and single battery state of charge. This hierarchical estimation method helps to improve the calculation accuracy of the battery cluster state of charge, so that the operating state of the battery stack and the battery cluster can follow the charging and discharging state changes of the single battery. However, the SOC estimation methods in existing battery management systems still have the following problems: the single battery SOC estimation accuracy is insufficient, the existing methods often rely on a single estimation algorithm such as Kalman filter or ampere-hour integral method, which is difficult to adapt to the dynamic characteristic changes of the battery under different working conditions, resulting in deviation between the estimated results and the actual power state. Especially in large-scale energy storage power stations, the number of single batteries is large, and the consistency difference between each single battery is more obvious, further increasing the difficulty of SOC estimation. Secondly, there is a lack of effective multi-level SOC aggregation mechanism. Although there are hierarchical estimation methods in existing technologies, most of them only consider simple arithmetic mean or weighted mean, without fully considering the differences between battery units, resulting in inaccurate calculation of the overall SOC of the battery module. Especially in large-scale energy storage power stations, how to effectively handle the information transmission and state fusion between different levels in the multi-level SOC aggregation process from single battery to module, battery cluster to entire power station is still a challenge. Thirdly, the existing methods do not fully consider the differences and dynamic changes between battery units. Due to factors such as manufacturing process, use environment and aging degree, the battery units in the same energy storage power station often have performance differences, which will dynamically change with the charging and discharging cycles. The existing methods lack effective identification and processing mechanism for such differences, which may easily lead to overcharging or overdischarging of some battery units, affecting the safety and service life of the system. Finally, there is a lack of SOC correction mechanism based on historical data. The existing SOC estimation methods are mostly based on real-time measurement data and theoretical models, without fully utilizing the battery characteristic information contained in the historical operation data, which is difficult to meet the SOC estimation needs under complex working conditions such as battery aging, affecting the accuracy and reliability of the estimation results. SUMMARY

[0004] The application aims to provide a SOC estimation method and system for balanced charging and discharging control of large-scale energy storage power stations to solve the problems in the background art.

[0005] To achieve the above-mentioned purpose, the application provides the following technical solutions.

[0006] A SOC estimation method for balanced charging and discharging control of large-scale energy storage power stations, the specific steps include:

[0007] S1: Collect the real-time working parameters of each single battery inside the energy storage power station to be estimated, and establish a state model and an observation model based on the Kalman filtering algorithm to predict the predicted SOC value of each single battery at the next time step;

[0008] S2: Obtain the historical working parameters of each single battery in the energy storage power station at each time and the corresponding SOC data at the next time step, and map them to form a group of reference vectors. All reference vectors form a mapping relationship library, a similarity criterion is defined, each single battery is matched in the mapping relationship library according to the collected real-time working parameters, the reference vectors meeting the similarity criterion are screened, the SOC data in each reference vector is extracted to form a sample data set;

[0009] S3: The SOC data in the sample data set corresponding to each single battery is used to correct the predicted SOC at the next time step, the corrected SOC value of each single battery at the next time step is obtained, and the reference value is calibrated. The threshold value of the state of charge of the single battery is set, and the effective single battery set is screened according to the relationship between the reference value and the threshold value. The SOC prediction value of the battery module to which it belongs is calculated by weighting the rated capacity;

[0010] S4: Based on the predicted SOC of the battery module, the module state of charge threshold value is set, the effective battery module set is screened in the battery cluster, and the SOC prediction value of the battery cluster to which it belongs is calculated by weighting the module capacity. Based on the predicted SOC of the battery cluster, the battery cluster state of charge threshold value is set, the effective battery cluster set is screened at the energy storage power station level, and the SOC prediction value of the power station is calculated by weighting the cluster capacity;

[0011] S5: According to the effective single battery set, the effective module set and the effective battery cluster set, three-level discharge loop control is performed; the SOC prediction value of the power station is input into the battery management system, the discharge strategy is adjusted in real time to ensure that the low-power unit stops discharging and the high-power unit continues to supply power.

[0012] Further, the real-time working parameters of each single battery inside the energy storage power station to be estimated include the initial state of charge, current, voltage, ohmic resistance and polarization resistance, and the state model is established based on the Kalman filtering algorithm as follows:

[0013] ;

[0014] wherein, is the SOC prediction value of the single battery at time t, is the polarization voltage prediction value of the single battery at time t, is the rated capacity of the single battery to be estimated, is the time step, is the polarization time constant, is the coulomb efficiency, is the polarization resistance of the single battery to be estimated, is the current of the single battery at time t, is the process noise; the observation equation is established, and the physical relationship of the state equation and the observation data is determined, and the formula is as follows: ;

[0015] ;

[0016] wherein, is the actual measured voltage between the positive and negative electrodes of the battery, is the slope of the open-circuit voltage of the single battery to SOC, is the ohmic internal resistance, is the observation noise. Further, the historical working parameters of each time of the single battery in the energy storage power station and the SOC data corresponding to the next time step are obtained, and after mapping, a reference vector is formed, and all reference vectors are formed into a mapping relationship library, wherein the historical working parameters are

[0017] , is the discharge current value at time k, is the terminal voltage between the positive and negative electrodes at time k, is the working temperature of the battery at time k, is the state of charge at time k; a similarity criterion is defined, and the collected real-time working parameters of each single battery are matched in the mapping relationship library, and the reference vectors satisfying the similarity criterion are screened out, and the SOC data in each reference vector is extracted to form a sample data set; the similarity criterion needs to satisfy the following conditions between the discharge working parameters of the sample battery and the single battery to be measured:

[0018] ;

[0019] wherein, is the discharge current of the sample battery, is the single battery to be measured, is the single battery to be measured, ​​​discharge current, for the sample battery rated capacity, sample battery current operating temperature, monocell to be measured operating temperature, sample battery current time SOC, monocell to be measured current time SOC; screening out the sample data set in the mapping relationship library by using the similarity criterion wherein battery reference vector satisfying the similarity criterion.

[0020] Further, the SOC data in the sample data set corresponding to each monocell is used to correct the predicted SOC of the next time step, to generate a corrected battery state vector, and the corrected SOC value of each monocell at the next time step is obtained and marked as a reference value:

[0021] ;

[0022] wherein, is the SOC reference value of the monocell at the time t, is the corrected polarization voltage prediction value of the monocell at the time t, is the Kalman gain matrix of the monocell at the time t, is the observation matrix, is the gain matrix between the batteries in the data set satisfying the similarity criterion, is the voltage observation value of the battery in the data set satisfying the similarity criterion, is the observation matrix for the sample battery is the SOC value of the sample battery at the time t, is the polarization voltage value of the sample battery at the time t. Further, a threshold value of the state of charge of the monocell is set, and an effective monocell set is screened out according to the relationship between the reference value and the threshold value, and the SOC prediction value of the battery module to which it belongs is calculated by weighting the rated capacity,

[0023]

[0024] ;​​

[0025] wherein, is the set of effective battery modules, is the single battery state of charge threshold value, weighted by the rated capacity of the battery module it belongs to, to calculate the SOC prediction value of the battery cluster it belongs to:

[0026] ;

[0027] wherein, is the SOC of the battery cluster at time t, is the rated capacity of the single battery, is the SOC prediction value of the battery module, is the rated capacity of the single battery, is the SOC prediction value of the battery cluster. Further, based on the battery module prediction SOC, the module state of charge threshold value is set, the set of effective battery modules is screened within the battery cluster, and the SOC prediction value of the battery cluster it belongs to is calculated by weighting the module capacity:

[0028]

[0029] ; wherein,

[0030] is the set of effective battery modules, is the battery module state of charge threshold value for battery cluster SOC calculation:

[0031] ; wherein,

[0032] is the SOC of the battery cluster at time t, is the rated capacity of the single battery, is the SOC prediction value of the battery cluster, is the effective rated capacity of the single battery it belongs to, is the SOC prediction value of the battery cluster, is the set of effective battery clusters, is the battery cluster state of charge threshold value, is the effective rated capacity of the battery cluster it belongs to.

[0033] ;

[0034] wherein, is the set of effective battery clusters, is the SOC of the energy storage power station at time t, is the battery cluster state of charge threshold value, is the effective rated capacity of the battery cluster it belongs to.

[0035] ​​Further, according to the effective monomer battery set, the effective module set and the effective battery cluster set, a three-level discharge loop control is performed: the monomer discharge loop is cut off , the module contactor is isolated , and the cluster PCS converter is closed ; the power station available state of charge and the battery cluster, battery module and monomer battery state of charge are input into the battery management system, the discharge strategy is adjusted in real time, and the low power unit is stopped from discharging while the high power unit continues to supply power.

[0036] The application also provides an SOC estimation system for balanced charge and discharge control of a large-scale energy storage power station, which is used to implement the above-mentioned SOC estimation method for balanced charge and discharge control of a large-scale energy storage power station, and specifically comprises:

[0037] A monomer battery SOC prediction module is configured to collect real-time working parameters of each monomer battery inside the energy storage power station to be estimated, and establish a state model and an observation model based on a Kalman filtering algorithm to predict a predicted SOC value of each monomer battery at a next time step;

[0038] A sample data collection module is configured to obtain historical working parameters of each monomer battery in the energy storage power station at each time and SOC data at a corresponding next time step, and map the historical working parameters and the SOC data to form a set of reference vectors, form a mapping relationship library by using all the reference vectors, define a similarity criterion, match each monomer battery in the mapping relationship library according to real-time working parameters of each monomer battery, filter reference vectors satisfying the similarity criterion, extract SOC data in each reference vector to form a sample data set, and perform the following steps.

[0039] An SOC correction and estimation module is configured to correct a predicted SOC at a next time step by using SOC data in a sample data set corresponding to each monomer battery, obtain a corrected SOC value of each monomer battery at the next time step, and calibrate the corrected SOC value as a reference value; set a threshold of a monomer battery state of charge, and filter an effective monomer battery set according to a relationship between the reference value and the threshold; and calculate a predicted SOC value of a battery module to which the effective monomer battery set belongs according to a rated capacity.

[0040] An energy storage power station SOC estimation module is configured to predict a SOC of a battery module, set a threshold of a module state of charge, filter an effective battery module set in a battery cluster, and calculate a predicted SOC value of a battery cluster to which the effective battery module set belongs according to a module capacity; predict a SOC of a battery cluster, set a threshold of a cluster state of charge, filter an effective battery cluster set at an energy storage power station level, and calculate a predicted SOC value of the energy storage power station according to a cluster capacity.

[0041] Discharge control and strategy adjustment module: according to the effective monomer battery set, the effective module set and the effective battery cluster set, the three-level discharge loop control is executed; the power station SOC prediction value is input into the battery management system, the discharge strategy is adjusted in real time, and the low power unit is stopped discharging, and the high power unit continues to supply power.

[0042] Compared with the prior art, the beneficial effects of the present application are:

[0043] Through the Kalman filter and historical sample data fusion correction mechanism, the monomer battery SOC prediction accuracy is significantly improved, and the accuracy is improved compared with the traditional method; a four-level SOC calculation system of monomer-module-cluster-power station is established, which ensures that the SOC calculation at each level accurately reflects the available energy, and solves the problem of lack of multi-level aggregation in the traditional method; through the 10% SOC threshold threshold throughout the monomer, module, cluster and step-by-step shutdown mechanism, the battery over-discharge damage is effectively avoided, and the battery service life is prolonged; the strategy of preferentially using high power units prolongs the overall service life of the battery; the three-level shutdown mechanism ensures that the system is safe and controllable under abnormal conditions, prevents safety accidents, and improves the overall safety of the system. BRIEF DESCRIPTION OF DRAWINGS

[0044] Figure 1 It is the overall method flowchart of the present application;

[0045] Figure 2 It is the SOC estimation experimental data comparison chart of the present application;

[0046] Figure 3 It is the schematic diagram of the SOC estimation system in the present application. DETAILED DESCRIPTION

[0047] In order to make the purpose, technical scheme and advantages of the present application clearer and more apparent, the present application will be further described in detail below in combination with specific embodiments.

[0048] It should be noted that, unless otherwise defined, the technical terms or scientific terms used in the present application should be understood as the usual meaning understood by those skilled in the art to which the present application belongs. The "first", "second" and similar words used in the present application do not represent any order, quantity or importance, but are only used to distinguish different components. "Include" or "contain" and similar words mean that the elements or objects before the word cover the elements or objects listed after the word and their equivalents, without excluding other elements or objects. "Connected" or "connected" and similar words are not limited to physical or mechanical connection, but can include electrical connection, whether direct or indirect. "Up", "down", "left", "right" and the like are only used to represent relative positional relationship, when the absolute position of the described object changes, the relative positional relationship may also change accordingly.

[0049] Embodiment:

[0050] Please refer to Figures 1-2 The present application provides a technical solution:

[0051] A SOC estimation method for balancing charge and discharge control of large-scale energy storage power stations, the specific steps comprising:

[0052] S1: Collect the real-time working parameters of each single battery inside the energy storage power station to be estimated, and establish a state model and an observation model based on Kalman filtering algorithm to predict the predicted SOC value of each single battery at the next time step;

[0053] In this embodiment, the real-time working parameters of each single battery inside the energy storage power station to be estimated are collected, including initial state of charge, current, voltage, ohmic internal resistance and polarization resistance, and the state model is established based on Kalman filtering algorithm as follows:

[0054] ;

[0055] Among them, is the SOC prediction value of the single battery at the moment, is the polarization voltage prediction value of the single battery at the moment, is the rated capacity of the single battery to be estimated, is the time step, is the polarization time constant, is the coulomb efficiency, is the polarization resistance of the single battery to be estimated, is the current of the single battery at the moment, is the process noise; the observation equation is established to determine the physical relationship between the state equation and the observation data, and the formula is as follows: ; Among them, is the actual measured voltage between the positive and negative electrodes of the battery,

[0056] is the slope of the open circuit voltage of the single battery to SOC, is the ohmic internal resistance, is the observation noise. The traditional ampere-hour integral method is easily affected by the cumulative error of current noise, and the present scheme significantly reduces the noise interference through bidirectional correction of the state space model and the observation equation; wherein

[0057] ​​​​​​Process noise is a random vector representing the uncertainty and unmodeled dynamic characteristics of the state prediction model itself. It encompasses all interfering factors that affect state (SOC and polarization voltage) changes but cannot be precisely described by deterministic equations. Most schemes use only the single-state variable SOC, neglecting the polarization voltage hysteresis effect, and introduce state equations... , = + It explicitly describes the charging and discharging process of the electric double-layer capacitor, accurately characterizes the voltage relaxation effect during current steps, and reduces the SOC estimation error under discharge conditions; the observation equation is embedded... By amplifying the SOC difference through the slope term, the problem of insufficient voltage resolution in the low / high SOC region is solved; the "next time step" in the predicted SOC value of each individual cell refers to the current time step. After a Discrete time points after It is based on time, not absolute physical time; the time step unifies the calculation cycle of all units, avoiding timing chaos caused by acquisition delay;

[0058] In this embodiment, two sets of discharge simulation experiments were conducted using the same energy storage power station. The technical solution was compared with the traditional SOC estimation solution. The data simulated a 3MWh energy storage power station (containing 40 battery clusters, 20 modules per cluster, and 24 NMC ternary lithium batteries per module). The experimental data for the individual cells within battery module B12 at 115 minutes are shown in the table below.

[0059] Table 1: Kalman Filter Predicted Individual SOC Data Table

[0060]

[0061] S2: Obtain the historical operating parameters of each individual battery in the energy storage power station at each moment and the corresponding SOC data for the next time step, and map them to form a set of reference vectors. Construct a mapping relationship library from all reference vectors, define similarity criteria, match the real-time operating parameters of each individual battery in the mapping relationship library, filter the reference vectors that meet the similarity criteria, extract the SOC data from each reference vector, and construct a sample dataset.

[0062] In this embodiment, the historical operating parameters of each individual battery in the energy storage power station at each moment and the corresponding SOC data for the next time step are obtained, mapped to form a set of reference vectors, and all reference vectors are used to form a mapping relationship library, wherein the historical operating parameters are... , Let k be the discharge current value. Let be the terminal voltage across the positive and negative terminals at time k. Let k be the operating temperature of the battery. SOCk is the state of charge at time k; define a similarity criterion to match the real-time working parameters of each single battery collected in the mapping relationship library, filter out the reference vectors that meet the similarity criterion, extract the SOC data in each reference vector to form a sample data set; the similarity criterion needs to meet the following conditions between the sample battery and the single battery to be tested:

[0063] ;

[0064] wherein, is the sample battery discharge current, is the single battery to be tested discharge current, is the sample battery rated capacity, is the sample battery current working temperature, is the single battery to be tested working temperature, is the sample battery current time SOC, is the single battery to be tested current time SOC; the sample data set is filtered out in the mapping relationship library using the similarity criterion , wherein is the battery reference vector that meets the similarity criterion;

[0065] The similarity criterion is defined using the physical basis of three-dimensional similarity. The current constraint requires that the discharge current difference between the historical sample battery and the target battery be less than 10% of the rated capacity, so as to avoid the polarization effect caused by the difference in large current working conditions; the temperature constraint requires that the working temperature of the historical sample battery and the target battery be within ±2℃, covering the nonlinear interval of lithium ion diffusion rate and reaction activity; the SOC constraint requires that the SOC of the historical sample battery and the target battery be less than 5% of the window, so as to ensure that the working point is in the similar slope section of the OCV-SOC curve; compared with the neural network directly trained without similarity screening, the prediction error is reduced at the boundary working condition, and the similarity criterion between the sample battery and the target battery is the core condition for screening historical sample data, which is used to correct the initial SOC prediction value of the Kalman filter algorithm;

[0066] Take the target battery B1203 as an example, whose parameters at 115min are , , The selected sample datasets are shown in the table below:

[0067] Table 2: Data Table of Battery B1203 Sample Data Set

[0068]

[0069] S3: Use the SOC data in the sample dataset corresponding to each individual cell to correct the predicted SOC for the next time step, obtain the corrected SOC value of each individual cell for the next time step and mark it as a reference value, set the threshold for the state of charge of the individual cell and filter out the effective set of individual cells according to the relationship between the reference value and the threshold, and calculate the predicted SOC value of the battery module to which it belongs by weighting according to the rated capacity.

[0070] In this embodiment, the SOC data from the sample dataset corresponding to each individual battery cell is used to correct the predicted SOC for the next time step, generating the corrected battery cell. The state vector is used to obtain the corrected SOC value for each individual cell at the next time step and calibrate it as a reference value.

[0071] ;

[0072] in, for The SOC reference value of a single cell after time correction. for Time-based single cell battery The corrected polarization voltage prediction value, For time-corrected single cell The Kalman gain matrix, For the observation matrix, , Batteries in the dataset that satisfy the similarity criterion The gain matrix between them Batteries in a dataset that satisfy the similarity criterion Voltage observations, For use in sample batteries The observation matrix for Time Sample Battery SOC value, for Time Sample Battery The polarization voltage value.

[0073] Correct current battery Self-observation residuals resolve the uncertainties in the current battery model; ,in Battery With sample The historical covariance of SOC sample Historical variance of SOC; reflecting the battery With sample battery The interrelationship between states is essentially an amplification factor, which amplifies the influence of the samples. Observation residuals Apply proportionally to the target battery The correction term is used to compensate for the bias of the Kalman filter algorithm using the sample dataset; in this embodiment, the corrected data is shown in the table below:

[0074] Table 3: Revised Individual SOC Data Table

[0075]

[0076] By integrating real-time observations with historical data under similar operating conditions, the model mismatch problem of Kalman filtering during battery aging and sudden temperature changes is overcome. The initial estimate can be corrected using actual sample data, while avoiding misjudgment of low-charge batteries, such as when the SOC is close to 10%, ensuring accurate disconnection of invalid cells. The sample data is continuously updated and automatically tracks battery performance degradation without the need for repeated calibration of model parameters.

[0077] In this embodiment, a threshold value for the state of charge of a single battery cell is set to generate a valid set of single battery cells.

[0078] ;

[0079] in, As an effective collection of individual cells, The threshold for the state of charge of a single battery cell is used to calculate a weighted average of the effective single batteries within each module:

[0080] ;

[0081] in, for Time Battery Module SOC, For single cell batteries Rated capacitor, to obtain the module Predict SOC.

[0082] The design concept of this solution represents the energy storage power station's process of excluding invalid individual cells during discharge, and not including them in the estimated SOC. If they were included in the estimated SOC, the total number of individual cells in the battery module would be greater than the actual number of discharging cells, which could easily lead to over-discharge, shorten battery life, and affect the working efficiency of the energy storage power station. At the same time, it also reflects the principle of large-capacity batteries, avoids the impact of abnormal values ​​of small-capacity batteries on the overall SOC, and also avoids deep discharge of individual cells.

[0083] S4: Based on the predicted SOC of battery modules, a threshold value for the state of charge of modules is set. Within each battery cluster, a set of effective battery modules is selected, and the predicted SOC value of the battery cluster is calculated by weighting the module capacity. Based on the predicted SOC of battery clusters, a threshold value for the state of charge of battery clusters is set. At the energy storage power station level, a set of effective battery clusters is selected, and the predicted SOC value of the power station is calculated by weighting the cluster capacity.

[0084] In this embodiment, based on the predicted SOC of the battery module, a threshold value for the module's state of charge is set, and a set of valid battery modules is selected within the battery cluster. The predicted SOC value of the battery cluster to which the module belongs is calculated by weighting the module capacity.

[0085] ;

[0086] in, For an efficient battery module assembly, To determine the state of charge (SOC) threshold for the battery module, calculate the SOC of the battery cluster:

[0087] ;

[0088] in, for Time Battery Cluster SOC, For effective single cell Belonging Module Rated capacitance; this formula strictly follows the law of conservation of charge, the total usable charge of the battery cluster. Total capacity This method accurately reflects energy reserves rather than using numerical averages. In contrast to traditional methods, when there are batteries with capacity degradation within a battery cluster, their SOC is artificially inflated due to increased internal resistance. The arithmetic average of traditional methods leads to serious distortion in SOC estimation, resulting in even higher errors. This method calculates the SOC of the battery cluster by considering only the effective battery modules and sums the values ​​only for the dischargeable battery modules.

[0089] In this embodiment, based on the predicted SOC of battery clusters, a threshold value for the state of charge of battery clusters is set. At the energy storage power station level, a set of effective battery clusters is selected, and the predicted SOC value of the power station is calculated by weighting the cluster capacity.

[0090] ;

[0091] wherein, is the effective battery cluster set, is the SOC of the energy storage power station at the moment, is the battery cluster state of charge threshold, is the effective module the battery cluster belongs to the effective rated capacity.

[0092] As with single-point battery and battery module screening, the "gradual attenuation" available capacity management from single cell to power station is realized, and invalid battery levels are abandoned;

[0093] In this embodiment, two discharge simulation experiments are carried out on the same energy storage power station, that is, the technical solution is compared with the traditional SOC estimation scheme; the following experimental data is obtained: as shown in the following table:

[0094] Table 4: Experimental data table

[0095]

[0096] The experimental data shows that at the end of the experiment, the number of effective single cells is 6610, accounting for 34.4% of the total capacity, and still working, and the average SOC of the 12,590 isolated batteries is only 7.2%. The technical solution releases 21.5% more energy than the final discharge instruction of the traditional SOC estimation scheme. The prediction value of the traditional scheme leads to an exponential error in the later stage of discharge, resulting in the system terminating discharge in advance. The technical solution controls the error within ±3% through Kalman filtering and sample fusion. Due to the active isolation of low-capacity units, the remaining high-SOC units continue to discharge, although it shows a negative error (-12.29% at 150 min), but this is the additional capacity released by the system reconstruction. The available SOC interval is expanded from 23% in the traditional scheme to 31.5%, which releases 44.1% more energy.

[0097] Referring to Figure 2 , Figure 2The deviation and absolute error between the SOC estimated value of the technical scheme and the actual value of the energy storage power station in Table 4 are clearly shown. It can be seen that the deviation of the technical scheme from the actual value is smaller than that of the traditional technical scheme. In the experimental process, the process of effective battery level screening of the technical scheme is accurate due to the estimation, so that the discharge capacity of the energy storage power station is also larger and safer; through the 10% SOC threshold threshold throughout the single cell, module and cluster level shutdown mechanism, the battery over-discharge damage is effectively avoided, and the battery service life is prolonged; the strategy of preferentially using high power units prolongs the overall service life of the battery; the three-level shutdown mechanism ensures the safety and controllability of the system under abnormal conditions.

[0098] S5: According to the effective single cell battery set, the effective module set and the effective battery cluster set, a three-level discharge loop control is performed; the SOC prediction value of the power station is input into the battery management system, and the discharge strategy is adjusted in real time to ensure that the low power unit stops discharging and the high power unit continues to supply power.

[0099] In this embodiment, according to the effective single cell battery set, the effective module set and the effective battery cluster set, a three-level discharge loop control is performed: the single cell discharge loop is cut off , the module contactor is isolated , and the cluster PCS inverter is turned off ; the available state of charge of the power station, and the state of charge of each battery cluster, battery module and single cell battery are input into the battery management system, and the discharge strategy is adjusted in real time to ensure that the low power unit stops discharging and the high power unit continues to supply power. The available SOC of the power station is input into the energy management system to realize adaptive adjustment of the discharge power; intelligent scheduling of high power clusters preferentially discharges to prolong the system endurance time; three-level hardware shutdown and 10% SOC hard threshold threshold completely eliminate the risk of thermal runaway caused by over-discharge; the available capacity utilization rate is improved;

[0100] Please refer to Figure 3 , the application also provides a SOC estimation system for equalizing charge and discharge control of a large-scale energy storage power station, which is used to realize the SOC estimation method for equalizing charge and discharge control of a large-scale energy storage power station, and specifically comprises:

[0101] A single cell SOC prediction module is used to collect the real-time working parameters of each single cell in the energy storage power station to be estimated, and a state model and an observation model are established based on a Kalman filtering algorithm to predict the predicted SOC value of each single cell at the next time step;

[0102] The sample data acquisition module is configured to acquire historical working parameters of each single battery in the energy storage power station at each time and SOC data corresponding to a next time step, map the historical working parameters and the SOC data to form a set of reference vectors, form a mapping relationship library by using all the reference vectors, define a similarity criterion, match each single battery in real time according to the working parameters collected in the mapping relationship library, filter the reference vectors meeting the similarity criterion, extract the SOC data in each reference vector, and form a sample data set;

[0103] The SOC correction and estimation module is configured to correct a predicted SOC of a next time step by using the SOC data in the sample data set corresponding to each single battery, acquire a corrected SOC value of the next time step of each single battery, and calibrate the corrected SOC value as a reference value, set a threshold of a state of charge of the single battery, and filter an effective single battery set according to a relationship between the reference value and the threshold, and calculate a predicted SOC value of a battery module to which the single battery belongs according to a rated capacity.

[0104] The energy storage power station SOC estimation module is configured to predict a SOC of a battery module, set a threshold of a state of charge of the battery module, filter an effective battery module set in a battery cluster, calculate a predicted SOC value of a battery cluster to which the battery module belongs according to a capacity of the battery module, predict a SOC of the battery cluster, set a threshold of a state of charge of the battery cluster, filter an effective battery cluster set at an energy storage power station level, and calculate a predicted SOC value of the energy storage power station according to a capacity of the battery cluster.

[0105] The discharge control and strategy adjustment module is configured to perform three-level discharge loop control according to the effective single battery set, the effective battery module set and the effective battery cluster set, input the predicted SOC value of the energy storage power station into a battery management system, and adjust a discharge strategy in real time to ensure that a low-power unit stops discharging and a high-power unit continuously supplies power.

[0106] The above formulas are all dimensionless values calculated, the formulas are obtained by software simulation of a large amount of data to obtain a formula of a nearest real situation, and preset parameters in the formulas are set by a person skilled in the art according to actual conditions.

[0107] The above embodiments can be realized wholly or partially by software, hardware, firmware or any combination thereof. When realized by software, the above embodiments can be realized in the form of a computer program product wholly or partially. A person skilled in the art can realize that units and algorithm steps of the examples described in combination with the embodiments disclosed herein can be realized by electronic hardware or a combination of computer software and electronic hardware. Whether the functions are realized by hardware or software methods depends on specific application and design constraints of the technical solutions.

[0108] The units described as separate components may or may not be physically separate, and the components displayed as units may or may not be physical units, and may be located in one place, or distributed on multiple network units. Part or all of the units can be selected to achieve the purpose of the embodiment of the present application according to actual needs.

[0109] The above is merely specific implementation of the present application, but the protection scope of the present application is not limited thereto, any person skilled in the art can easily think of changes or replacements within the technical range disclosed by the present application, which shall be covered within the protection scope of the present application.

Claims

1. A SOC estimation method for balancing charge and discharge control of a large-scale energy storage power station, characterized in that, The specific steps include: S1: Collecting real-time working parameters of each single battery inside the energy storage power station to be estimated, and establishing a state model and an observation model based on Kalman filtering algorithm to predict the predicted SOC value of each single battery at the next time step; S2: Obtaining historical working parameters of each single battery in the energy storage power station at each time and the corresponding SOC data at the next time step, mapping to form a set of reference vectors, constructing a mapping relationship library with all the reference vectors, defining a similarity criterion, matching in the mapping relationship library according to the collected real-time working parameters of each single battery, screening the reference vectors satisfying the similarity criterion, extracting the SOC data in each reference vector to form a sample data set; S3: Using the SOC data in the sample data set corresponding to each single battery to correct the predicted SOC at the next time step, obtaining the corrected SOC value of each single battery at the next time step, and setting a single battery state of charge threshold and screening an effective single battery set according to the relationship between the reference value and the threshold; S4: Based on the battery module predicted SOC, setting the module state of charge threshold, screening the effective battery module set in the battery cluster, and calculating the SOC prediction value of the battery cluster to which it belongs according to the module capacity weighting; S5: According to the effective single battery set, the effective module set and the effective battery cluster set, executing three-level discharge loop control; inputting the predicted SOC value of the power station into the battery management system to adjust the discharge strategy in real time, ensuring that the low power unit stops discharging and the high power unit continues to supply power; wherein the SOC data in the sample data set corresponding to each monomer battery is used to correct the predicted SOC of the next time step, to generate a corrected battery state vector, and the corrected SOC value of each monomer battery at the next time step is obtained and marked as a reference value: in, for The SOC reference value of a single cell after time correction. for Time-based single cell battery The corrected polarization voltage prediction value, for Time-based single cell battery SOC prediction value, for Time-based single cell battery The predicted polarization voltage For time-corrected single cell The Kalman gain matrix, For the observation matrix, Batteries in the dataset that satisfy the similarity criterion The gain matrix between them Batteries in the dataset that satisfy the similarity criterion Voltage observations, For use in sample batteries The observation matrix for Time Sample Battery SOC value, for Time Sample Battery The polarization voltage value; S3: Setting a single battery state of charge threshold and screening an effective single battery set according to the relationship between the reference value and the threshold, and calculating the SOC prediction value of the battery module to which it belongs according to the rated capacity weighting: wherein, is the effective monobloc battery set, is the monobloc battery state of charge threshold, weighted by the rated capacity of the battery module to which it belongs, of the SOC prediction value: wherein, is the SOC of the battery module at the moment, is the rated capacitance of the single battery, the predicted SOC;​​ S4: Based on the battery module predicted SOC, setting the module state of charge threshold, screening the effective battery module set in the battery cluster, and calculating the SOC prediction value of the battery cluster to which it belongs according to the module capacity weighting: wherein, for an effective battery module set, battery module state of charge threshold value, for battery cluster SOC calculation: wherein, is the SOC of the battery cluster at the moment, is the effective monobloc battery the module belongs to rated capacitance;​ S4: Based on the battery cluster predicted SOC, setting the battery cluster state of charge threshold, screening the effective battery cluster set at the energy storage power station level, and calculating the SOC prediction value of the power station according to the cluster capacity weighting: wherein, is an effective battery cluster set, is is the SOC of the energy storage plant at the moment, is a battery cluster state of charge threshold, is an effective module the battery cluster to which the battery belongs effective rated capacitance. 2.The SOC estimation method for balancing charge and discharge control of a large-scale energy storage power station according to claim 1, characterized in that: Collecting real-time working parameters of each single battery inside the energy storage power station to be estimated, including initial state of charge, current, voltage, ohmic resistance and polarization resistance, and establishing a state model based on Kalman filtering algorithm as follows: wherein, is the rated capacity of the battery cell to be estimated, is the time step, is the polarization time constant, is the coulombic efficiency, is the polarization resistance of the battery cell to be estimated, is the battery cell is the current at time instant, is the process noise; the observation equation is established, and the physical relationship of the state equation and the observation data is determined, and the formula is as follows: wherein, is the battery is the actual measured voltage between the positive and negative electrode, is the slope of the open circuit voltage versus SOC for the single cell, is the ohmic internal resistance, is the observation noise. 3.The SOC estimation method for balancing charge and discharge control of large-scale energy storage power station according to claim 1, characterized in that: The historical working parameters of each moment of the single battery in the energy storage power station and the SOC data of the corresponding next time step are acquired, mapped, and constitute a set of reference vectors, and all the reference vectors constitute a mapping relationship library, wherein the historical working parameters are , is the discharge current value at k moment, is the terminal voltage between the positive and negative electrodes at k moment, is the working temperature of the battery at k moment, is the state of charge at k moment; The similarity criterion is defined, and each single battery is matched in the mapping relationship library according to the real-time working parameters collected, the reference vectors meeting the similarity criterion are screened out, the SOC data in each reference vector is extracted, and the sample data set is constituted; the similarity criterion needs the sample battery The similarity criterion is defined, and each single battery is matched in the mapping relationship library according to the real-time working parameters collected, the reference vectors meeting the similarity criterion are screened out, the SOC data in each reference vector is extracted, and the sample data set is constituted; the similarity criterion needs the sample battery wherein, sample battery discharge current, battery under test discharge current, sample battery rated capacitance, sample battery current operating temperature, battery under test operating temperature, sample battery current time SOC, battery under test current time SOC; selecting a sample data set from the mapping relationship library using a similarity criterion wherein, battery reference vector satisfying the similarity criterion. 4.The SOC estimation method for balancing charge and discharge control of a large-scale energy storage power station according to claim 1, characterized in that: According to the effective monomer battery set, the effective module set and the effective battery cluster set, three-level discharge loop control is performed: cut off the monomer discharge loop, isolate the module contactor, close the cluster PCS converter; input the power plant available state of charge, and the state of charge of each battery cluster, battery module and monomer battery into the battery management system, adjust the discharge strategy in real time, and ensure that low power units stop discharging while high power units continue to supply power.

5. A SOC estimation system for balancing charge and discharge control of a large-scale energy storage power station, characterized in that: The SOC estimation system for balanced charging and discharging control of large-scale energy storage power station is used to execute the SOC estimation method of any one of claims 1-4, comprising: The single battery SOC prediction module is used to collect real-time working parameters of each single battery inside the energy storage power station to be estimated, and establish a state model and an observation model based on Kalman filtering algorithm to predict the predicted SOC value of each single battery at the next time step; The sample data acquisition module is configured to acquire historical working parameters of each single battery in the energy storage power station at each time and SOC data corresponding to a next time step, map the historical working parameters and the SOC data to form a set of reference vectors, form a mapping relationship library by using all the reference vectors, define a similarity criterion, match each single battery in real time according to the working parameters collected in the mapping relationship library, filter reference vectors meeting the similarity criterion, extract SOC data in each reference vector, and form a sample data set; The SOC correction and estimation module is configured to correct a predicted SOC of a next time step by using SOC data in the sample data set corresponding to each single battery, acquire an SOC value of each single battery after correction of the next time step, and calibrate the SOC value as a reference value, set a threshold of a state of charge of the single battery, and filter an effective single battery set according to a relationship between the reference value and the threshold, and calculate an SOC prediction value of a battery module to which the single battery belongs according to a rated capacity; The battery cluster SOC estimation module is configured to set a threshold of a state of charge of a battery module, filter an effective battery module set in a battery cluster, and calculate an SOC prediction value of the battery cluster to which the battery module belongs according to a module capacity; The energy storage power station SOC estimation module is configured to set a threshold of a state of charge of a battery cluster, filter an effective battery cluster set at an energy storage power station level, and calculate an SOC prediction value of the energy storage power station according to a cluster capacity; The discharge control and strategy adjustment module is configured to execute three-level discharge loop control according to the effective single battery set, the effective module set and the effective battery cluster set, input the SOC prediction value of the energy storage power station into a battery management system, and adjust a discharge strategy in real time to ensure that a low-power unit stops discharging and a high-power unit continues to supply power.

Images