Energy storage system power scheduling method and energy storage system

Abstract

The present application relates to the technical field of energy storage system, and particularly relates to a power scheduling method of an energy storage system and the energy storage system, comprising the following steps: deploying an edge computing node and building a cloud global optimization engine to provide data support for subsequent accurate regulation and control, and guaranteeing the timeliness and integrity of data acquisition; analyzing the energy storage equipment capacity and power grid demand by means of an intelligent algorithm, setting a reasonable response priority and dynamically allocating tasks to ensure efficient use of resources and accurate execution of tasks; adopting an MPC algorithm to take the power grid frequency deviation as an optimization target, rolling optimization of output instruction and correction of prediction error to realize the regulation of the energy storage system on the power grid frequency fluctuation and enhance the stability of the power grid; and taking the minimization of the whole life cycle cost as a target to optimize the long-time energy storage charging and discharging plan, effectively improving the stability of the power grid and the economy of the energy storage equipment operation.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of energy storage systems, in particular to a power scheduling method of an energy storage system and the energy storage system. BACKGROUND

[0002] In the current field of energy storage systems, power scheduling still faces multiple challenges: the traditional energy storage system data acquisition and control architecture is mostly centralized, cloud processing delay is high, and edge side data processing capability is weak, resulting in device state monitoring lag; existing fixed priority or threshold-based scheduling strategies cannot adapt to dynamic changes in grid demand and device state, resulting in mismatched or overused energy storage device charging and discharging; single time scale optimization leads to response conflicts, and there is a lack of dynamic optimization mechanism for multi-type energy storage collaboration, resulting in low resource utilization. Therefore, the present application proposes a power scheduling method of an energy storage system and the energy storage system. SUMMARY

[0003] The purpose of the present application is to solve the problems in the background art, and a power scheduling method of an energy storage system and the energy storage system are proposed.

[0004] In order to achieve the above-mentioned purpose, the present application adopts the following technical solutions:

[0005] The power scheduling method of the energy storage system comprises:

[0006] S1, deploy an edge computing node, integrate data acquisition and power execution functions, and complete real-time acquisition of energy storage device state data; at the same time, build a cloud global optimization engine, integrate an energy storage device capability feature library and a power price acquisition unit, and complete energy storage device data analysis and feature extraction;

[0007] S2, according to the acquired energy storage device state data, analyze the capacity of the energy storage device and the grid demand through intelligent algorithm, set the response priority of each energy storage device, and dynamically allocate tasks;

[0008] S3, according to the set task allocation and priority, cooperates with the edge layer second-level control and the cloud minute-level optimization to build a multi-time scale optimization scheduling strategy; specifically:

[0009] In the edge computing layer, MPC algorithm is used to take the grid frequency deviation as the optimization target to rollingly optimize the future Np second output instruction; the response of the energy storage system to the grid frequency fluctuation within the future Np seconds is predicted, and the prediction error is corrected by Kalman filtering to adjust the dynamic response of the energy storage system to the grid frequency fluctuation;

[0010] In the cloud global optimization engine, super short-term power prediction and time-of-use price signals are fused to minimize the total life cycle cost as the target to optimize the long-time energy storage charging and discharging plan;

[0011] The second-level MPC control result is combined with the minute-level economic dispatch result to form a multi-time scale optimal dispatch strategy.

[0012] Further, the process of real-time collection and feature extraction of the energy storage device state data includes:

[0013] An edge computing node is deployed at the energy storage device end, which integrates a real-time data acquisition unit and a power execution unit; wherein high-precision sensors are configured in the real-time data acquisition unit to obtain comprehensive energy storage device state data;

[0014] A bidirectional DC / AC converter is deployed in the power execution unit to realize bidirectional conversion of alternating current grid power and battery direct current, and to collect real-time grid frequency data, wherein the collected frequency data is stored in the local storage of the edge device in the form of time series;

[0015] Sensor data is transmitted to the edge computing node through a low-latency bus; the edge computing node performs preprocessing operations on the collected raw data; the preprocessed data is encapsulated and encrypted; the encrypted data packets are uploaded to the cloud global optimization engine in real time through a 5G slice network or an industrial optical fiber;

[0016] A cloud global optimization engine is built, integrating an energy storage device capability feature library and a power price acquisition unit; the cloud global optimization engine receives the encrypted data packets uploaded by the edge layer and performs decryption processing; the key features of the devices are extracted from the decrypted data and labeled as energy storage device capability features to construct the energy storage device capability feature library; wherein the energy storage device capability feature library stores the key features of various energy storage devices; the power price acquisition unit is used to collect ultra-short-term power prediction data and time-of-use price signals; a communication protocol based on MQTT over TLS is defined, and the edge computing node and the cloud global optimization engine realize data interaction through subscription or publication mode.

[0017] Further, according to the collected energy storage device state data, the ability of the energy storage device and the demand of the power grid are analyzed through intelligent algorithm, the response priority of each energy storage device is set, and the process of dynamically allocating tasks includes:

[0018] Real-time demand data and energy storage device state data provided by the power grid dispatch center are obtained, and the power grid demand is classified into high-frequency response demand, long-time regulation demand and economic regulation demand; the demand type is converted into a quantitative feature vector;

[0019] For a group of energy storage devices, a multi-dimensional capability feature library is constructed; the feature vector of each type of energy storage device is mapped to a high-dimensional space to form a device capability portrait;

[0020] Based on the formed energy storage device capability image and the grid demand quantification feature vector, the energy storage device capability features are associated and matched with the grid demand features: the matching degree of the energy storage device capability feature vector and the grid demand feature vector is calculated through cosine similarity; based on the matching degree, initial priority weights are assigned to different types of energy storage devices; according to the matching degree and the initial priority weights, the association rules between the energy storage devices and the grid demand are constructed, and the association matrix of the energy storage devices and the grid demand is generated using the association rules;

[0021] Based on the generated association matrix of the energy storage devices and the grid demand, the dynamic priority allocation of the energy storage devices is adapted using a reinforcement learning algorithm, and the dynamic priority weights of each type of energy storage device are obtained.

[0022] A multi-dimensional evaluation system is established in combination with the multi-dimensional capability feature library; based on the dynamic priority weights of the energy storage devices, the entropy weight method is used to weight and sum the scores of each dimension, and the dynamic priority score of the energy storage device is calculated; the energy storage devices are ranked in descending order according to the priority score, thereby forming a candidate response sequence;

[0023] The current operating state of the grid is analyzed in depth through a graph neural network to identify the grid fluctuation type.

[0024] The energy storage response power is taken as the optimization variable to construct an objective function; meanwhile, an equality constraint condition is established; for different grid fluctuation types, complementary constraints are set respectively; the Lagrange multiplier is dynamically adjusted according to the grid fluctuation type identification result and the candidate response sequence of the energy storage device; the dynamically adjusted Lagrange multiplier is combined with the above complementary constraint conditions to form a final constraint equation set.

[0025] After the constraint equation set is constructed, the interior point method is used to solve the constructed dynamic Lagrange multiplier method model to obtain the output instructions of each energy storage device; the energy storage devices are assigned to different types of tasks according to the complementary constraints, including high-frequency response tasks, long-time regulation tasks and economic regulation tasks; the real-time output instructions of the multi-type energy storage devices after the task type assignment are sent to the control end of the energy storage device.

[0026] Further, the feature vector of each type of energy storage device is mapped to a high-dimensional space, and the process of forming the device capability image includes:

[0027] The inner product operation in the original low-dimensional feature space is converted into the inner product operation in the high-dimensional feature space through a kernel function, thereby realizing the high-dimensional mapping of the energy storage device feature vector; in the energy storage device capability feature mapping, the kernel parameters are adjusted according to different types of energy storage devices and device operating scenarios; the K-means clustering algorithm is used for clustering analysis of the energy storage device capability feature vectors mapped to the high-dimensional space; the energy storage device capability clusters obtained by clustering are associated and matched with the grid demand types.

[0028] Further, the process of constructing the association rules between the energy storage devices and the grid demand and generating the association matrix of the energy storage devices and the grid demand comprises:

[0029] setting a minimum support threshold and a minimum confidence threshold; wherein the minimum support is used to measure the frequency of the energy storage device-grid demand combination appearing in the entire dataset; the minimum confidence reflects the association strength between the energy storage device-grid demand combination; using the Apriori algorithm to mine the frequently occurring energy storage device-grid demand combination; in the mining process, according to the set minimum support threshold, the frequent item set satisfying the minimum confidence condition is screened out; based on the frequent item set, the first association rule is generated according to the minimum confidence threshold;

[0030] using the decision tree algorithm, taking the energy storage device capability characteristics and the grid demand characteristics as the input variables, and taking the priority weight of the energy storage device as the output variable, a decision tree model is constructed; the second association rule is extracted from the decision tree model: starting from the root node, along the branch path of the decision tree, the splitting feature and the splitting threshold of each node are combined into the condition part of the rule, and the priority weight adjustment direction and amplitude AW of the leaf node relative to the root node or other reference nodes are taken as the conclusion part of the rule; according to the constructed second association rule, the association matrix of the energy storage device and the grid demand is generated, wherein the row of the association matrix represents the energy storage device, the column represents the grid demand type, and the dimension of the matrix is determined according to the types of the energy storage device and the types of the grid demand; the first association rule and the second association rule are integrated; according to the integrated association rule, the priority weight adjustment value between the energy storage device and the grid demand is calculated and taken as the element in the association matrix; for each association rule, if the rule condition is satisfied, the priority weight of the energy storage device is adjusted according to the amplitude AW determined in the rule conclusion; the calculated priority weight adjustment value is filled into the corresponding position of the association matrix, and the association matrix is updated; the initial priority weight and the priority weight adjustment value in the association matrix are assigned weights, and the influence degree of both on the final association matrix is integrated; using the weighted summation method, the initial priority weight and the priority weight adjustment value in the association matrix are comprehensively calculated to obtain the final energy storage device-grid demand association matrix.

[0031] Further, the MPC algorithm is used to take the grid frequency deviation as the optimization target to rollingly optimize the output instruction in the future Np seconds; the process of predicting the response of the energy storage system to the grid frequency fluctuation within the future Np seconds, combining the Kalman filter to correct the prediction error, and adjusting the dynamic response of the energy storage system to the grid frequency fluctuation comprises:

[0032] correlatively analyzing the collected real-time grid frequency data, the historical grid frequency data stored locally by the edge computing layer, and the energy storage system operation data;

[0033] establish a dynamic response model of the energy storage system;

[0034] According to the change of the grid frequency and the response speed of the energy storage system, determine the prediction time domain Np of the MPC algorithm; set the control time domain Nc of the MPC algorithm, and the control time domain Nc is less than or equal to the prediction time domain Np;

[0035] Using the established dynamic response model of the energy storage system and the current grid frequency and the operation data of the energy storage system, the change of the grid frequency in the future Np seconds and the response of the energy storage system under the grid frequency fluctuation are predicted;

[0036] Taking the predicted grid frequency deviation as the optimization target, the output instruction of the energy storage system in the future Np seconds is optimized within the control time domain Nc, wherein the optimization process is realized by searching the possible combination of the charging and discharging power of the energy storage system, so that the preset metric of the grid frequency deviation in the prediction time domain is minimized;

[0037] The output instruction of the energy storage system at the first time obtained by optimization is sent to the control end of the energy storage system, so that the charging and discharging power is adjusted according to the instruction, and the rapid response of the energy storage system to the grid frequency fluctuation is realized;

[0038] In each control cycle, when the actual grid frequency data and the operation data of the energy storage system are updated, the error between the actual frequency and the predicted frequency of the MPC algorithm, and the error between the actual charging and discharging power of the energy storage system and the predicted power are calculated;

[0039] Using Kalman filtering algorithm, according to the prediction error at the current time and the prior estimation of Kalman filtering, the state estimation and covariance matrix of Kalman filter are updated;

[0040] According to the updated state estimation of the Kalman filter, the dynamic response model of the energy storage system is corrected.

[0041] Further, the super short-term power prediction and the time-of-use price signal are fused, and the process of optimizing the long-time energy storage charging and discharging plan with the minimum life cycle cost as the target includes:

[0042] All the state data of the energy storage equipment uploaded on the edge computing layer are converged to form a global state data set of the energy storage system;

[0043] Obtain a super short-term power prediction data set and a time-of-use price signal data set;

[0044] Integrate the obtained super short-term power prediction data set, time-of-use price signal data set and global state data set of the energy storage system to form a complete data set for minute-level optimization;

[0045] Based on the integrated minute-level optimization complete data set, the charge and discharge plan of the energy storage system is made: for the selection of the charging period, the large power generation period of new energy power generation is identified in combination with the ultra-short-term power prediction data set; for the selection of the discharging period, the power consumption peak period is locked according to the time-of-use price signal data set; meanwhile, the whole life cycle cost items related to the charge and discharge plan of the energy storage system are identified, including the initial investment cost, operation and maintenance cost, charge and discharge loss cost, battery aging cost and opportunity cost of the energy storage equipment;

[0046] Based on the identified cost items, a whole life cycle cost function is constructed in a weighted summation manner;

[0047] Taking the minimization of the whole life cycle cost as the optimization target, the genetic algorithm is used to find the optimal charge and discharge plan of the energy storage system, so that the whole life cycle cost is minimized;

[0048] The charge and discharge plan of the energy storage system obtained through minute-level optimization is transmitted to the MPC controller of the edge computing layer as reference information; the MPC controller comprehensively considers the minute-level economic dispatching result, the real-time state of the current power grid and the real-time operation data of the energy storage system when rolling optimizing the output instruction of the energy storage system, and forms a multi-time scale optimization scheduling strategy.

[0049] The energy storage system comprises:

[0050] The hierarchical architecture is deployed and the data acquisition module is acquired: the edge computing node is deployed, the data acquisition and power execution functions are integrated, and the real-time acquisition of the energy storage equipment state data is completed; meanwhile, the cloud end global optimization engine is built, the energy storage equipment capability feature library and the power price acquisition unit are integrated, and the energy storage equipment data analysis and feature extraction are completed;

[0051] The dynamic task adaptive allocation module: according to the collected energy storage equipment state data, the capability of the energy storage equipment and the power grid demand are analyzed through an intelligent algorithm, the response priority of each energy storage equipment is set, and the task is dynamically allocated;

[0052] The multi-time scale optimization scheduling module: according to the set task allocation and priority, the edge layer second-level control and the cloud end minute-level optimization are cooperated to construct a multi-time scale optimization scheduling strategy.

[0053] Compared with the existing technology, the power dispatching method based on the energy storage system and the energy storage system provided by the application have the advantages that:

[0054] The application sets the response priority of the energy storage device through an intelligent algorithm and dynamically allocates tasks, ensures the rationality of task allocation, constructs a multi-time scale optimal scheduling strategy, the edge layer uses an MPC algorithm for second-level control, responds to frequency fluctuation quickly with the power grid frequency deviation as the target, combines Kalman filter correction to correct prediction error, enhances dynamic response capability, and guarantees power grid frequency stability, the cloud global optimization engine integrates ultra-short-term power prediction and time-of-use price signals, optimizes long-time energy storage charging and discharging plan at the minute level, reduces the life cycle cost, combines the second-level and minute-level results to form a multi-time scale optimal scheduling strategy that takes into account real-time and economy, and improves the operation efficiency of the energy storage system and the power supply quality of the power grid. BRIEF DESCRIPTION OF DRAWINGS

[0055] Figure 1 The flowchart of the power dispatching method of the energy storage system provided by the application.

[0056] Figure 2 The module diagram of the energy storage system provided by the application. DETAILED DESCRIPTION

[0057] The technical solutions in the embodiments of the application will be described clearly and completely in the embodiments of the application in combination with the drawings. Obviously, the described embodiments are only part of the embodiments of the application, not all the embodiments. Based on the embodiments in the application, all other embodiments obtained by those skilled in the art without creative labor are within the protection scope of the application.

[0058] Please refer to Figure 1 The application provides a power dispatching method of an energy storage system, which comprises the following steps:

[0059] S1, deploy an edge computing node, integrate data acquisition and power execution functions, and complete real-time acquisition of energy storage device state data; at the same time, build a cloud global optimization engine, integrate an energy storage device capability feature library and a power price acquisition unit, and complete energy storage device data analysis and feature extraction;

[0060] S2, according to the acquired energy storage device state data, analyze the capability of the energy storage device and the power grid demand through an intelligent algorithm, set the response priority of each energy storage device, and dynamically allocate tasks;

[0061] S3, according to the set task allocation and priority, cooperates with the edge layer second-level control and the cloud minute-level optimization to construct a multi-time scale optimal scheduling strategy; specifically:

[0062] At the edge computing layer, the MPC algorithm is adopted to take the power grid frequency deviation as the optimization target to rollingly optimize the output instruction in the future Np seconds; the response of the energy storage system to the power grid frequency fluctuation in the future Np seconds is predicted, and the prediction error is corrected by Kalman filtering to adjust the dynamic response of the energy storage system to the power grid frequency fluctuation;

[0063] In the cloud global optimization engine, the ultra-short-term power prediction and time-of-use price signals are fused to optimize the long-time energy storage charging and discharging plan with the minimum life cycle cost as the target;

[0064] The second-level MPC control result and the minute-level economic dispatch result are combined to form a multi-time scale optimization dispatch strategy.

[0065] It should be further explained that in the specific implementation process, the edge computing node is deployed, the data acquisition and power execution functions are integrated, and the real-time acquisition of the energy storage device state data is completed; at the same time, the cloud global optimization engine is built, the energy storage device capability feature library and the power price acquisition unit are integrated, and the process of energy storage device data analysis and feature extraction is completed as follows:

[0066] The edge computing node is deployed at the energy storage device end, which integrates the real-time data acquisition unit and the power execution unit; wherein the real-time data acquisition unit is configured with high-precision sensors (temperature sensors, SOC / SOH monitoring modules, etc.) to monitor the battery's state of charge SOC, state of health SOH, real-time power, and temperature and other key parameters in real time, and obtain comprehensive energy storage device state data;

[0067] The bidirectional DC / AC converter is deployed in the power execution unit to realize the bidirectional conversion (charging or discharging) of alternating power grid energy and battery direct current, and integrate the frequency sensor to acquire the power grid frequency data in real time, wherein the acquired frequency data is stored in the local storage of the edge device in the form of time series; specifically, the power execution unit is used to convert the alternating power grid energy into direct current stored in the battery, and convert the battery direct current into alternating current to feed back to the grid or load;

[0068] The sensor data is transmitted to the edge computing node through a low-latency bus (such as CAN bus); the edge computing node performs preprocessing operations such as filtering, normalization, and outlier detection on the collected raw data; the preprocessed data is packaged into a standard format and encrypted using a symmetric encryption algorithm; the encrypted data packet is uploaded to the cloud global optimization engine in real time through the 5G slice network or industrial optical fiber;

[0069] The cloud global optimization engine is built, and the energy storage device capability feature library and the power price collection unit are integrated. The cloud global optimization engine receives the encrypted data packet uploaded by the edge layer, decrypts it using the corresponding key, and restores it to the original data format. The decrypted data is cleaned again (redundant data is removed, missing values are filled), the key features of the device (such as SOC, SOH, etc.) are extracted and marked as energy storage device capability features, and the energy storage device capability feature library is constructed. The energy storage device capability feature library stores the key features of various energy storage devices, such as power-capacity curves and other key data. The power price collection unit is used to collect ultra-short-term power prediction data and time-of-use price signals. It should be understood that the collection method of ultra-short-term power prediction data is to obtain renewable energy power prediction data, load power prediction data, and overall power demand prediction data for a future period (such as 15 minutes to 1 hour) from professional meteorological departments or power trading centers. These data are based on weather forecasts, historical load data, etc. and are predicted using machine learning methods to provide a basis for developing a reasonable charging and discharging plan for the energy storage system. The collection method of time-of-use price signals is to obtain time-of-use price curves, including peak and valley period prices and auxiliary service compensation prices, by connecting with the power market trading platform. The time-of-use price signal reflects the supply and demand relationship and cost changes of the power market, and the energy storage system can optimize the charging and discharging plan according to the price signal to achieve economic operation. A communication protocol based on MQTT over TLS is defined, and the edge computing node and the cloud global optimization engine realize data interaction through subscription or publication mode.

[0070] It should be further explained that, in the specific implementation process, based on the collected energy storage device state data, the capacity of the energy storage device and the demand of the power grid are analyzed by intelligent algorithm, the response priority of each energy storage device is set, and the process of dynamically allocating tasks is as follows:

[0071] Real-time demand data (such as frequency fluctuation, power shortage prediction) provided by the power grid dispatching center and energy storage device state data (SOC, SOH, etc.) are obtained, and the power grid demand is classified into high-frequency response demand (i.e. frequency disturbance scenario, requiring millisecond-level response), long-time regulation demand (i.e. power shortage scenario, requiring minute-level or above continuous support), and economic regulation demand (i.e. peak-valley arbitrage, auxiliary service market scenario); the demand type is converted into a quantitative feature vector, including frequency fluctuation amplitude, power shortage prediction value, and price difference;

[0072] For the energy storage device group, a multi-dimensional capability feature library is constructed. The feature vector of each type of energy storage device (such as flywheel, lithium battery, and flow battery) is mapped to a high-dimensional space to form a device capability portrait. The process of mapping the feature vector of each type of energy storage device to a high-dimensional space to form a device capability portrait includes:

[0073] The inner product operation in the original low-dimensional feature space is converted into the inner product operation in the high-dimensional feature space by a kernel function, so as to realize high-dimensional mapping of the energy storage device feature vector; in the energy storage device capability feature mapping, the kernel parameters are adjusted according to different types of energy storage devices and device operation scenarios; the K-means clustering algorithm is used to perform clustering analysis on the energy storage device capability feature vectors mapped to the high-dimensional space; the energy storage device capability clusters obtained by clustering are associated and matched with the grid demand types; for example, the high-frequency response demand of the grid usually requires the energy storage device to have fast power regulation capability, so the high-power energy storage device cluster is associated with the high-frequency response demand; the long-time regulation demand focuses on the energy storage capacity of the energy storage device, so the high-energy energy storage device cluster is associated with the long-time regulation demand; the economic regulation demand pays attention to the operation cost and benefit of the energy storage device, so the economic energy storage device cluster is associated with the economic regulation demand;

[0074] Based on the formed energy storage device capability portrait and the grid demand quantification feature vector, the energy storage device capability features are associated and matched with the grid demand features: the matching degree of the energy storage device capability feature vector and the grid demand feature vector is calculated through cosine similarity; based on the matching degree, initial priority weights are assigned to different types of energy storage devices; according to the matching degree and the initial priority weights, the association rules between the energy storage devices and the grid demands are constructed, and the association matrix of the energy storage devices and the grid demands is generated by using these association rules, which is used as the input of subsequent dynamic priority calculation; wherein, the process of constructing the association rules between the energy storage devices and the grid demands and generating the association matrix of the energy storage devices and the grid demands includes:

[0075] The minimum support threshold and the minimum confidence threshold are set; wherein, the minimum support is used to measure the frequency of the energy storage device-grid demand combination in the entire data set; the minimum confidence reflects the association strength between the energy storage device-grid demand combination, that is, the probability that a specific energy storage device is called when the grid demand occurs; the Apriori algorithm is used to mine the frequently occurring energy storage device-grid demand combinations; in the mining process, the frequent item sets that meet the minimum confidence condition are selected according to the set minimum support threshold; based on the frequent item sets, the first association rules are generated according to the minimum confidence threshold;

[0076] The decision tree algorithm is used to construct a decision tree model, with the energy storage device capability features and the power grid demand features as input variables, and the priority weight of the energy storage device as an output variable; a second association rule is extracted from the decision tree model: starting from the root node, along the branch path of the decision tree, the split feature and the split threshold of each node are combined into the condition part of the rule, and the priority weight adjustment direction (increase or decrease) and the amplitude AW of the leaf node relative to the root node or other reference nodes are taken as the conclusion part of the rule; according to the constructed second association rule, an association matrix of the energy storage device and the power grid demand is generated, wherein the rows of the association matrix represent the energy storage devices, the columns represent the power grid demand types, and the dimension of the matrix is determined according to the types of the energy storage devices and the types of the power grid demands; when constructing the association matrix, if no association rule (including the first association rule and the second association rule) is mined between the energy storage device and the power grid demand type, the value of the corresponding position in the initial association matrix is 0 or a very small number, indicating that there is no association or weak association between them; the first association rule and the second association rule are integrated: for the same energy storage device-power grid demand combination involved in the two rules, the final association rule is determined; according to the integrated association rule, the priority weight adjustment value between the energy storage device and the power grid demand is calculated and taken as an element in the association matrix: if the association rule indicates that the energy storage device i has a higher association degree with the power grid demand j, the value of the corresponding position (i, j) in the association matrix is larger; for each association rule, if the rule condition is met, the priority weight of the energy storage device is adjusted according to the amplitude AW determined in the rule conclusion; specifically, the rule condition includes the satisfaction condition of adjusting the priority weight based on the first association rule, the satisfaction condition of adjusting the priority weight based on the second association rule, and the satisfaction condition of adjusting the priority weight comprehensively based on the two rules; wherein the condition part of the first association rule is composed of specific attributes or states of the energy storage device-power grid demand combination, for example, if the power grid demand type is the peak load demand and the energy storage device type is the lithium battery, the priority weight of the energy storage device is increased by AW1, only when the actual power grid monitoring data shows that the current power grid demand type is the peak load demand, and the actual called energy storage device type is the lithium battery, the rule condition is met, and the priority weight of the lithium battery energy storage device is increased by AW1 according to the rule; the second association rule is extracted from the decision tree model, and the rule condition corresponds to the split feature and the split threshold of each node on the path from the root node to the leaf node in the decision tree, for example, a branch rule of the decision tree is: if the power grid demand type is the high-frequency response demand and the power response time of the energy storage device is less than t milliseconds, the priority weight of the energy storage device is increased by AW2, only when the actual power grid demand type is the high-frequency response demand, and the power response time of the energy storage device is less than t milliseconds, the rule condition is met, and the priority weight of the energy storage device is increased by AW2 according to the rule.If the combination of the energy storage device and the grid demand satisfies both the first association rule and the second association rule, the rule priority is preset, for example, assuming that the priority of the first association rule is higher than that of the second association rule, when both rules are satisfied, the priority weight adjustment is performed according to the first association rule with higher priority; if the priorities of the two rules are the same, further judgment is made according to factors such as the confidence level of the rule; the calculated priority weight adjustment value is filled in the corresponding position of the association matrix, and the association matrix is updated: for the case where there are multiple association rules, the weighted average or maximum value method is used to integrate the calculation results of multiple association rules to obtain the final association matrix element value; the initial priority weight and the priority weight adjustment value in the association matrix are assigned weights, and the influence degrees of both on the final association matrix are integrated; the initial priority weight and the priority weight adjustment value in the association matrix are integrated by using the weighted summation method to obtain the final energy storage device-grid demand association matrix;

[0077] Based on the generated energy storage device-grid demand association matrix, the dynamic priority allocation of the energy storage device is adapted using a reinforcement learning algorithm: the energy storage device capability characteristics and the grid demand characteristics are input as the state space; the priority weight of each type of energy storage device is defined as the action space; a reward function is constructed: the response speed, the regulation accuracy and the economy of the energy storage device are rewarded or punished for each action; wherein the response speed is the frequency recovery time and the power shortage filling time, the regulation accuracy is the frequency deviation suppression rate and the power shortage filling error, and the economy includes the charging and discharging cost and the auxiliary service income; the current energy storage device state and the grid demand are input into the trained reinforcement learning model, and the dynamic priority weight of each type of energy storage device is output;

[0078] A multi-dimensional evaluation system including response speed (millisecond level / second level), energy density (kWh), cycle life (number of charge and discharge times), SOC state (percentage of charge), etc. is established in combination with the multi-dimensional capability characteristic library; based on the dynamic priority weight of the energy storage device, the entropy weight method is used to weight and sum the scores of each dimension to calculate the dynamic priority score of the energy storage device; specifically, the dynamic priority weight reflects the importance of different types of energy storage devices under the current grid demand, and the entropy weight method determines the objective weight of each dimension in the comprehensive score according to the dispersion degree of the score of each dimension, and then adjusts the objective weight in combination with the dynamic priority weight to obtain the final comprehensive score; the energy storage devices are arranged in descending order according to the priority score, thereby forming a candidate response sequence;

[0079] The current operation state of the power grid is analyzed in depth by the graph neural network to identify the type of power grid fluctuation: the power grid is abstracted as a dynamic graph structure, with power generation nodes as vertices and power transmission lines as edges to construct a power grid topology relationship network; real-time power grid operation data (such as node frequency deviation) is embedded in the graph vertices, and edge attributes are associated with line impedance and other physical parameters; the graph structure is aggregated by graph convolution to automatically extract topological dependency features (such as local frequency propagation path and power shortage diffusion mode), output the power grid fluctuation type classification result (i.e. distinguish frequency disturbance or power shortage, where for frequency disturbance, identify the node cluster with abnormal frequency change rate and mark it as high frequency response demand; for power shortage, identify the area with insufficient power injection and mark it as long-time adjustment demand), and generate a power grid fluctuation type label for subsequent constraint generation;

[0080] The response power of the energy storage is taken as the optimization variable to construct an objective function (such as minimizing frequency deviation / power shortage) that considers multiple factors such as power grid stability, energy storage device life, and operating cost to maximize comprehensive benefits; at the same time, equality constraint conditions (including global power grid constraints: power generation + energy storage output = load demand, and energy storage body constraints: SOC charging and discharging upper and lower limits, maximum charging and discharging power) are established to ensure that the energy storage device meets the basic power balance requirements of the power grid during operation;

[0081] For different types of power grid fluctuations, complementary constraints are set respectively: for frequency disturbance, according to the characteristics of power grid frequency disturbance, a complementary constraint is set that the response of energy storage is negatively related to the frequency change rate: when the power grid frequency drops, the energy storage device is required to discharge and inject power into the power grid to support the frequency rise; when the power grid frequency rises, the energy storage device charges to absorb excess power from the power grid, suppressing the further rise of the frequency, thereby achieving rapid response and adjustment to frequency disturbance; for power shortage, a complementary constraint is set that the response of energy storage is positively related to the amount of power shortage, i.e. the greater the power shortage of the power grid, the higher the power output of the energy storage device, which ensures the stable operation of the power grid by timely making up the power gap;

[0082] According to the identification result of the power grid fluctuation type and the candidate response sequence of the energy storage device, the Lagrange multiplier is dynamically adjusted (for high-frequency disturbance, the multiplier value of the flywheel energy storage is increased (to respond to frequency first); for power shortage, the multiplier value of the battery energy storage is increased (to make up power first)), to strengthen the dominant role of high-priority devices in the complementary constraint; the dynamically adjusted Lagrange multiplier is combined with the above complementary constraint conditions to form the final constraint equation set, ensuring that the energy storage response and the power grid fluctuation characteristics form a complementary match;

[0083] After the constraint equation set is completed, the constructed dynamic Lagrange multiplier method model is solved using the interior point method to obtain the output instruction of each energy storage device. The energy storage devices are assigned to tasks according to the complementary constraints. Specifically, the tasks are divided into high-frequency response tasks, long-time adjustment tasks, and economic adjustment tasks (for example, high-priority-weight energy storage devices (such as flywheels) are assigned high-frequency response tasks, and the output instruction is adjusted in real time according to the frequency fluctuations (for example, updated every second); low-priority-weight energy storage devices (such as lithium batteries) are assigned long-time adjustment tasks, and the output instruction is adjusted smoothly according to the power shortage prediction (for example, updated every minute); for economic adjustment tasks, low-cost energy storage devices are assigned according to the price difference); the real-time output instructions of the multi-type energy storage devices after task type allocation are sent to the control end of the energy storage device.

[0084] It can be understood that the interior point method is an optimization algorithm that iteratively approaches the optimal solution within the feasible region. By introducing a barrier function, the constraint conditions are integrated into the objective function. In this embodiment, the interior point method is used to solve the nonlinear programming problem of the dynamic Lagrange multiplier model of the energy storage system.

[0085] It should be further explained that, in the specific implementation process, according to the set task allocation and priority, the edge layer second-level control and the cloud minute-level optimization are coordinated to construct a multi-time scale optimization scheduling strategy.

[0086] The collected real-time power grid frequency data, historical power grid frequency data stored locally by the edge computing layer, and energy storage system operation data are analyzed to mine the potential relationship between power grid frequency changes and energy storage system charging and discharging. The energy storage system operation data are obtained by integrating the long-term performance indicators (such as cycle life decay) and scheduling instruction archive data stored in the cloud time series historical database.

[0087] A dynamic response model of the energy storage system is established. The model is used to reflect the dynamic relationship between the power change, power output, and power grid frequency during the charging and discharging process of the energy storage system. The model is described in the form of a state space equation, and considers the charging and discharging efficiency, self-discharge rate, and other characteristics of the energy storage device.

[0088] According to the power grid frequency change and the response speed of the energy storage system, the prediction time domain N of the MPC algorithm is determined. The control time domain of the MPC algorithm is set, and the control time domain is less than or equal to the prediction time domain.

[0089] The established dynamic response model of the energy storage system and the current power grid frequency and energy storage system operation data are used to predict the power grid frequency change in the future N seconds and the response of the energy storage system under the power grid frequency fluctuation.

[0090] Taking the predicted grid frequency deviation as the optimization objective, the output instruction of the energy storage system in the next N seconds is optimized in the control time domain, wherein the optimization process searches for possible combinations of charging and discharging power of the energy storage system, so that the preset metric (such as sum of squares) of the grid frequency deviation in the prediction time domain is minimized;

[0091] The optimized energy storage system output instruction at the first time is sent to the control end of the energy storage system, so that the charging and discharging power is adjusted according to the instruction, and the rapid response of the energy storage system to the grid frequency fluctuation is realized;

[0092] In each control cycle, when the actual grid frequency data and the energy storage system operation data are updated, the error between the actual frequency and the predicted frequency by the MPC algorithm, and the error between the actual charging and discharging power of the energy storage system and the predicted power are calculated;

[0093] Using the Kalman filtering algorithm, the state estimation and covariance matrix of the Kalman filter are updated according to the prediction error at the current time and the prior estimation of the Kalman filter; wherein the Kalman filtering algorithm continuously uses new measurement data to correct the estimation of the state of the energy storage system, thereby improving the prediction accuracy of the grid frequency change and the operation state of the energy storage system;

[0094] According to the updated state estimation of the Kalman filter, the dynamic response model of the energy storage system is corrected; wherein the correction method is to adjust the parameters in the dynamic response model of the energy storage system, such as the charging and discharging efficiency parameters, response time parameters, etc. of the energy storage device;

[0095] All energy storage device state data uploaded on the edge computing layer are converged to form a global state data set of the energy storage system;

[0096] Obtain the ultra-short-term power prediction data set and the time-of-use price signal data set;

[0097] Integrate the obtained ultra-short-term power prediction data set, time-of-use price signal data set and global state data set of the energy storage system to form a complete data set for minute-level optimization;

[0098] Based on the integrated minute-level optimization complete data set, a charging and discharging plan of the energy storage system is made: for charging period selection, in combination with the ultra-short-term power prediction data set, the large generation period of new energy generation (such as wind power, photovoltaic) is identified; for discharging period selection, according to the time-of-use price signal data set, the peak electricity consumption period (such as the evening peak of residential electricity consumption, the concentrated period of industrial production) is locked; at the same time, the life cycle cost items related to the charging and discharging plan of the energy storage system are identified, including the initial investment cost, operation and maintenance cost, charging and discharging loss cost, battery aging cost and opportunity cost of the energy storage device;

[0099] Based on the identified cost items, a full life cycle cost function is constructed in a weighted summation manner, wherein the weights are adjusted according to the importance of different cost items and the actual operation situation, for example, in the initial stage of the life cycle of the energy storage system, the initial investment cost weight is larger, and with the increase of the running time of the equipment, the weights of the operation and maintenance cost and the battery aging cost gradually increase;

[0100] With the optimization goal of minimizing the whole life cycle cost, the genetic algorithm is used to find the optimal charging and discharging plan of the energy storage system, so that the whole life cycle cost is minimized. The process of using the genetic algorithm to find the optimal charging and discharging plan of the energy storage system is as follows: J1, the charging and discharging power of the energy storage system in each optimization period (such as minute level) is taken as the core decision variable: assuming that the optimization period is 24 hours, each hour is divided into 60 periods, each energy storage device has three states of charging, discharging or inaction in each period, which is discretized into different power levels (such as charging power -100kW, -50kW, 0kW, discharging power 50kW, 100kW, etc.); J2, determine which period the energy storage device charges or discharges, represented by binary coding: real number coding is used for charging and discharging power (continuous variable, such as -100kW to 100kW with 0.1kW precision); integer coding is used for charging and discharging period allocation (discrete variable, such as 0 = inaction, 1 = charging, 2 = discharging), a vector with the same number of optimization periods is constructed to describe the charging and discharging period allocation scheme of the energy storage device; J3, within the range of charging and discharging power, randomly generate the initial population according to uniform distribution, for example, for real number coding, if the charging and discharging power range is -100kW to 100kW, a real number between -100 and 100 is randomly generated in each optimization period as the charging and discharging power of the period, and the process is repeated to generate a sufficient number of individuals (such as 100) to form the initial population; based on the constraints of the charging and discharging power limit of the energy storage system, the battery capacity limit, etc., randomly generate the initial population within the range that meets the constraints, for example, according to the maximum charging and discharging power of the energy storage device and the current power state, determine the available range of charging and discharging power in each period, and then randomly generate the initial individual within the range; J4, according to the identified whole life cycle cost items, including the initial investment cost of the energy storage device, the operation and maintenance cost, the charging and discharging loss cost, the battery aging cost and the opportunity cost, calculate the cost of each individual, for example, the initial investment cost: directly calculated according to the capacity and procurement price of the energy storage device; operation and maintenance cost: calculated according to the charging and discharging power and the power change (the higher the power and the more frequent the change, the higher the operation and maintenance cost); charging and discharging loss cost: calculated according to the charging and discharging power and the energy conversion efficiency; battery aging cost: based on the number of battery cycles and the charging and discharging depth (such as 1 cycle for deep discharge 1 time, shallow charging and discharging 0.5, a cycle is calculated once), combined with the battery replacement cost allocation; opportunity cost: the difference between the actual discharge benefit and the theoretical benefit during the peak period of electricity price; J5, the costs are weighted and summed according to the preset weight to obtain the life cycle cost of each individual; the reciprocal of the life cycle cost is taken as the fitness value; the life cycle cost is linearly transformed to map it to a suitable fitness value range; J6, the probability of being selected according to the fitness value of the individual; for real number coding, a crossover point is randomly selected, and the genes of the two parent individuals after the crossover point are exchanged to generate two offspring individuals, for example, for two parent individuals represented by 24-dimensional real number vectors, a time period (such as the 12th time period) is randomly selected as the crossover point, and the charge and discharge power values of the two parent individuals after the 12th time period are exchanged to obtain two offspring individuals; for integer coding, single-point crossover is used, but attention should be paid to whether the individual after crossover meets the constraint condition; for real number coding, each gene is mutated with a certain mutation probability Pm, and a new value is randomly generated within the value range of the gene according to uniform distribution to replace the original gene value, for example, if the original value of a gene is 50kW, a new value between-100kW and 100kW is randomly generated after mutation; for integer coding, a new gear is randomly selected in the discretized power gear to replace the original gear; J7, for individuals that do not meet the constraint condition, a penalty term is added to their fitness value, for example, if the charge and discharge power of the energy storage device exceeds the maximum charge and discharge power limit, or the battery power does not reach the expected range at the end of the optimization period, a corresponding penalty term is designed according to the degree of violation of the constraint; J8, set a maximum number of iterations, when the number of iterations of the algorithm reaches the preset maximum number of iterations, the algorithm terminates, for example, the maximum number of iterations is set to 500; when the algorithm terminates, the optimal individual in the current population is output, that is, the optimal energy storage system charge and discharge plan, wherein the charge and discharge plan includes the charge and discharge power value of each optimization time period and the charge and discharge time period allocation.

[0101] The energy storage system charge and discharge plan obtained by minute-level optimization is transmitted as reference information to the MPC controller of the edge computing layer; the MPC controller integrates the minute-level economic dispatch result, the real-time state of the current power grid and the real-time operation data of the energy storage system to form a multi-time scale optimization dispatch strategy when rolling optimizing the energy storage system output instruction.

[0102] Please refer to Figure 2 The application provides an energy storage system, comprising:

[0103] Hierarchical architecture deployment and data collection module: deploy edge computing nodes, integrate data collection and power execution functions, and complete real-time collection of energy storage device state data; at the same time, build a cloud-based global optimization engine, integrate energy storage device capability feature library and power price collection unit, complete energy storage device data analysis and feature extraction;

[0104] Dynamic task adaptive allocation module: according to the collected energy storage device state data, the ability and grid demand of the energy storage device are analyzed through intelligent algorithm, the response priority of each energy storage device is set, and the task is dynamically allocated;

[0105] Multi-time scale optimization scheduling module: according to the set task allocation and priority, the edge layer second-level control and cloud minute-level optimization are coordinated to build a multi-time scale optimization scheduling strategy.

[0106] It should be understood that the size of the sequence number of each process described above in various embodiments of the present application does not mean the order of execution, and the execution order of each process should be determined according to its function and internal logic, and should not constitute any limitation on the implementation process of the embodiments of the present application.

[0107] It should be understood that determining B according to A does not mean that B is determined only according to A, but B can also be determined according to A and / or other information.

[0108] The above is only a specific embodiment of the present application, but the protection scope of the present application is not limited thereto, and any skilled person in the art can easily think of changes or replacements within the technical scope disclosed in the present application, which should be included in the protection scope of the present application. Therefore, the protection scope of the present application should be subject to the protection scope of the claims.

[0109] Finally: the above is only the preferred embodiment of the present application and does not limit the present application, any modification, equivalent replacement, improvement, etc. within the spirit and principles of the present application should be included in the protection scope of the present application.

Claims

1. A power dispatching method for an energy storage system, characterized in that: S1. Deploy edge computing nodes to integrate data acquisition and power execution functions to complete real-time acquisition of energy storage device status data; at the same time, build a cloud-based global optimization engine to integrate energy storage device capability feature library and power price acquisition unit to complete energy storage device data parsing and feature extraction. S2. Based on the collected energy storage device status data, the system analyzes the capacity of the energy storage devices and the grid demand using intelligent algorithms, sets the response priority of each energy storage device, and dynamically allocates tasks. The process of analyzing the capacity of the energy storage devices and the grid demand using intelligent algorithms based on the collected energy storage device status data, setting the response priority of each energy storage device, and dynamically allocating tasks includes: Acquire real-time demand data and energy storage device status data provided by the power grid dispatch center, and classify power grid demand into high-frequency response demand, long-term regulation demand, and economic regulation demand; convert the demand types into quantitative feature vectors. For energy storage equipment clusters, a multi-dimensional capability feature library is constructed; the feature vectors of each type of energy storage equipment are mapped to a high-dimensional space to form a capability profile of the equipment. Based on the established capability profiles of energy storage devices and the quantitative feature vectors of grid demand, the capability features of energy storage devices are correlated and matched with the features of grid demand: the matching degree between the capability feature vectors of energy storage devices and the feature vectors of grid demand is calculated using cosine similarity; based on the matching degree, initial priority weights are assigned to different types of energy storage devices; based on the matching degree and the initial priority weights, association rules between energy storage devices and grid demand are constructed, and these association rules are used to generate an association matrix between energy storage devices and grid demand. Based on the generated correlation matrix between energy storage devices and grid demand, a reinforcement learning algorithm is used to adapt the dynamic priority allocation of energy storage devices and obtain the dynamic priority weights of each type of energy storage device. A multi-dimensional evaluation system is established by combining a multi-dimensional capability feature library; based on the dynamic priority weight of energy storage equipment, the scores of each dimension are weighted and summed using the entropy weight method to calculate the dynamic priority score of the energy storage equipment; the energy storage equipment is ranked in descending order according to the priority score to form a candidate response sequence; By using graph neural networks, we can perform in-depth analysis of the current operating status of the power grid and identify the types of power grid fluctuations. Using the energy storage response power as the optimization variable, an objective function is constructed; simultaneously, equality constraints are established; complementary constraints are set for different grid fluctuation types; the Lagrange multiplier is dynamically adjusted based on the grid fluctuation type identification results and the candidate response sequence of the energy storage device; the dynamically adjusted Lagrange multiplier is combined with the above complementary constraints to form the final set of constraint equations. After the constraint equations are constructed, the interior point method is used to solve the dynamic Lagrange multiplier model to obtain the output commands of each energy storage device. The energy storage devices are assigned task types according to complementary constraints, specifically high-frequency response tasks, long-term regulation tasks, and economic regulation tasks. The real-time output commands of the various types of energy storage devices after task type assignment are sent to the energy storage device control terminal. S3. Based on the set task allocation and priority, coordinate edge layer second-level control and cloud minute-level optimization to construct a multi-time-scale optimization scheduling strategy; specifically: At the edge computing layer, the MPC algorithm is used to optimize the output command for the next Np seconds with the grid frequency deviation as the optimization target. The response of the energy storage system to grid frequency fluctuations within the next Np seconds is predicted, and the prediction error is corrected by Kalman filtering to adjust the dynamic response of the energy storage system to grid frequency fluctuations. In the cloud-based global optimization engine, ultra-short-term power forecasting and time-of-use pricing signals are integrated to optimize long-term energy storage charging and discharging plans with the goal of minimizing the total lifecycle cost. By combining the second-level MPC control results with the minute-level economic scheduling results, a multi-time-scale optimized scheduling strategy is formed.

2. The power dispatching method for an energy storage system according to claim 1, characterized in that: The process of real-time acquisition and feature extraction of energy storage device status data includes: An edge computing node is deployed at the energy storage device. This node integrates a real-time data acquisition unit and a power execution unit. The real-time data acquisition unit is equipped with high-precision sensors to acquire comprehensive energy storage device status data. A bidirectional DC / AC converter is deployed in the power execution unit to realize bidirectional conversion between AC grid power and battery DC power, and to collect grid frequency data in real time. The collected frequency data is stored in the local memory of the edge device in the form of time series. Sensor data is transmitted to edge computing nodes via a low-latency bus; the edge computing nodes preprocess the collected raw data; the preprocessed data is encapsulated and encrypted; and the encrypted data packets are uploaded to the cloud-based global optimization engine in real time via a 5G slicing network or industrial fiber optic cable. A cloud-based global optimization engine is built, integrating an energy storage device capability feature library and a power price acquisition unit. The cloud-based global optimization engine receives encrypted data packets uploaded from the edge layer and decrypts them. Key device features are extracted from the decrypted data and marked as energy storage device capability features, thus constructing an energy storage device capability feature library. The energy storage device capability feature library stores the key features of various types of energy storage devices. The power price acquisition unit is used to collect ultra-short-term power prediction data and time-of-use pricing signals. A communication protocol based on MQTT over TLS is defined, and edge computing nodes and the cloud-based global optimization engine interact with each other through a subscription or publish-subscribe pattern.

3. The power dispatching method for an energy storage system according to claim 1, characterized in that: The process of mapping the feature vectors of each type of energy storage device to a high-dimensional space to form a device capability profile includes: The kernel function transforms the inner product operation in the original low-dimensional feature space into an inner product operation in the high-dimensional feature space, thereby realizing the high-dimensional mapping of the energy storage device feature vector. In the energy storage device capability feature mapping, the kernel parameters are adjusted according to different energy storage device types and operating scenarios. The K-means clustering algorithm is used to perform cluster analysis on the energy storage device capability feature vector mapped to the high-dimensional space. The clustered energy storage device capability clusters are then matched with the grid demand type.

4. The power dispatching method for an energy storage system according to claim 1, characterized in that: The process of constructing the association rules between energy storage devices and grid demand, and generating the association matrix between energy storage devices and grid demand, includes: Minimum support threshold and minimum confidence threshold are set. Minimum support measures the frequency of energy storage device-grid demand combinations in the entire dataset. Minimum confidence reflects the strength of the association between energy storage device-grid demand combinations. The Apriori algorithm is used to mine frequently occurring energy storage device-grid demand combinations. During the mining process, frequent itemsets that meet the minimum confidence condition are selected based on the set minimum support threshold. Based on the frequent itemsets, the first association rule is generated according to the minimum confidence threshold. Using a decision tree algorithm, a decision tree model is constructed with energy storage device capability characteristics and grid demand characteristics as input variables and the priority weights of energy storage devices as output variables. A second association rule is extracted from the decision tree model: starting from the root node, along the branch paths of the decision tree, the splitting characteristics and splitting thresholds of each node are combined into the condition part of the rule, and the adjustment direction and magnitude ΔW of the priority weights of leaf nodes relative to the root node or other reference nodes are used as the conclusion part of the rule. Based on the constructed second association rule, an association matrix between energy storage devices and grid demand is generated, where the rows of the association matrix represent energy storage devices, the columns represent grid demand types, and the dimensions of the matrix are determined according to the type of energy storage device and the type of grid demand. The first association rule... Then, it is integrated with the second association rule; based on the integrated association rule, the priority weight adjustment value between energy storage devices and grid demand is calculated and used as an element in the association matrix; for each association rule, if the rule condition is met, the priority weight of the energy storage device is adjusted according to the magnitude ΔW determined in the rule conclusion; the calculated priority weight adjustment value is filled into the corresponding position in the association matrix, and the association matrix is ​​updated; weights are assigned to the initial priority weight and the priority weight adjustment value in the association matrix, and the influence of both on the final association matrix is ​​considered; using a weighted summation method, the initial priority weight and the priority weight adjustment value in the association matrix are comprehensively calculated to obtain the final association matrix between energy storage devices and grid demand.

5. The power dispatching method for an energy storage system according to claim 1, characterized in that: The MPC algorithm is used to optimize the power output command for the next Np seconds using the grid frequency deviation as the optimization objective. The process of predicting the energy storage system's response to grid frequency fluctuations within the next Np seconds, and then correcting the prediction error using Kalman filtering, to adjust the dynamic response of the energy storage system to grid frequency fluctuations includes: The collected real-time power grid frequency data, historical power grid frequency data stored locally at the edge computing layer, and energy storage system operation data are correlated and analyzed. Establish a dynamic response model for the energy storage system; Based on the grid frequency variation and the energy storage system response speed, determine the prediction time domain Np of the MPC algorithm; set the control time domain Nc of the MPC algorithm, where the control time domain Nc is less than or equal to the prediction time domain Np; Using the established dynamic response model of the energy storage system and the current grid frequency and energy storage system operation data, the grid frequency change and the response of the energy storage system under grid frequency fluctuations are predicted within the next Np seconds. Using the predicted grid frequency deviation as the optimization objective, the output command of the energy storage system in the future Np seconds is optimized in the control time domain Nc. The optimization process minimizes the preset metric of grid frequency deviation in the prediction time domain by searching for possible combinations of charging and discharging power of the energy storage system. The optimized first-moment output command of the energy storage system is sent to the control terminal of the energy storage system, so that it adjusts the charging and discharging power according to the command, thereby realizing the rapid response of the energy storage system to grid frequency fluctuations; Within each control cycle, after the actual grid frequency data and energy storage system operation data are updated, the error between the actual frequency and the frequency predicted by the MPC algorithm, as well as the error between the actual charging and discharging power of the energy storage system and the predicted power, are calculated. Using the Kalman filter algorithm, the state estimate and covariance matrix of the Kalman filter are updated based on the prediction error at the current time and the prior estimate of the Kalman filter. The dynamic response model of the energy storage system is corrected based on the updated state estimate obtained from the Kalman filter.

6. The power dispatching method for an energy storage system according to claim 1, characterized in that: The process of integrating ultra-short-term power forecasting and time-of-use pricing signals to optimize long-term energy storage charging and discharging plans with the goal of minimizing total lifecycle costs includes: The system aggregates the status data of all energy storage devices uploaded from the edge computing layer to form a global status dataset for the energy storage system. Acquire ultra-short-term power prediction datasets and time-of-use electricity price signal datasets; The acquired ultra-short-term power prediction dataset, time-of-use electricity price signal dataset, and energy storage system global state dataset are integrated to form a complete dataset for minute-level optimization. Based on an integrated minute-level optimized complete dataset, a charging and discharging plan for the energy storage system is formulated: for the selection of charging periods, the peak periods of new energy power generation are identified by combining ultra-short-term power prediction datasets; for the selection of discharging periods, peak electricity consumption periods are locked based on time-of-use price signal datasets; at the same time, the full life cycle cost items related to the charging and discharging plan of the energy storage system are identified, including the initial investment cost of energy storage equipment, operation and maintenance costs, charging and discharging loss costs, battery aging costs, and opportunity costs. Based on the identified cost items, a full life cycle cost function is constructed using a weighted summation method. With the goal of minimizing the total life cycle cost, a genetic algorithm is used to find the optimal charging and discharging plan for the energy storage system, so as to minimize the total life cycle cost. The energy storage system's charge and discharge plan, optimized at the minute level, is passed to the MPC controller at the edge computing layer as reference information. When the MPC controller is continuously optimizing the energy storage system's output commands, it integrates the minute-level economic dispatch results, the current real-time status of the power grid, and the real-time operating data of the energy storage system to form a multi-time-scale optimization dispatch strategy.

7. An energy storage system, characterized in that, A power dispatching method applied to an energy storage system as described in any one of claims 1-6, the system comprising: Layered architecture deployment and data acquisition module: Deploy edge computing nodes to integrate data acquisition and power execution functions to complete real-time acquisition of energy storage device status data; at the same time, build a cloud-based global optimization engine to integrate energy storage device capability feature library and power price acquisition unit to complete energy storage device data parsing and feature extraction; Dynamic task adaptive allocation module: Based on the collected energy storage device status data, it analyzes the capacity of the energy storage devices and the grid demand through intelligent algorithms, sets the response priority of each energy storage device, and dynamically allocates tasks; Multi-timescale optimization scheduling module: Based on the set task allocation and priority, it coordinates the edge layer's second-level control and the cloud's minute-level optimization to build a multi-timescale optimization scheduling strategy.

Images