Park energy storage charging and discharging strategy generation method, device and equipment

Abstract

The embodiment of the invention discloses a park energy storage charging and discharging strategy generation method, device and equipment. The method comprises the following steps: firstly, acquiring distributed energy storage data and a distributed energy storage energy conversion form of current energy storage charging and discharging of a park; then selecting a mathematical model and an operation constraint condition of distributed energy storage; then building a distributed energy storage charge and discharge optimization model covering peak regulation capability and photovoltaic output prediction deviation; based on a preset dynamic game model of a leader and a follower, optimizing the distributed energy storage charging and discharging optimization model to obtain a target optimization model; based on a variable neighborhood search and multi-target optimization combined algorithm, solving the target optimization model to obtain a Pareto optimal solution set; and finally, based on a preset target preference and a Pareto optimal solution set, generating a park energy storage charging and discharging optimal strategy. According to the method, the energy storage charging and discharging strategy of the park can be generated, and multi-target overall optimal scheduling of the park is realized by executing the strategy.

Description

TECHNICAL FIELD

[0001] The embodiment of the application belongs to the technical field of energy storage, and particularly relates to a method, device and equipment for generating a park energy storage charging and discharging strategy. BACKGROUND

[0002] In a park (such as an industrial park, a science and technology park, etc.), a high concentration of loads is a main feature - numerous production devices and process systems are densely laid out, and meanwhile, photovoltaic power, wind power and other renewable energy sources are widely connected, so that the energy demand of the park shows large-scale and highly dynamic changes. Such a complex demand structure brings great challenges to traditional energy management and dispatching: on the one hand, the start and stop of devices, the fluctuation of process flows and the real-time adjustment of enterprise production plans can easily cause rapid changes in loads; on the other hand, the output of distributed photovoltaic power and other intermittent renewable energy sources is affected by weather conditions, further increasing the uncertainty of supply and demand.

[0003] With the continuous deepening of green energy transformation, the large-scale deployment of distributed energy storage systems in parks is gradually becoming a reality. Distributed energy storage systems greatly enhance the flexibility of park energy system regulation and the ability to cope with uncertainties, and become an important link to support the construction of new power systems.

[0004] On this basis, the park gradually evolves into an energy platform jointly operated by multiple subjects. Multiple enterprises in the park build a closely linked energy ecosystem by sharing energy infrastructure. However, due to differences and even conflicts in the goals of each subject (for example, some pursue the lowest cost, and some place more emphasis on maximizing benefits), this makes the energy dispatching problem of the park more complex. On the one hand, it is necessary to balance the interests of different parties; on the other hand, it is also necessary to cope with multiple challenges such as real-time price fluctuations, prediction deviations of photovoltaic output, etc.

[0005] At present, there is no energy storage charging and discharging strategy that can simultaneously achieve the optimal coordination of these multiple goals, including balancing the interests of different subjects, improving peak shaving capacity and coping with the uncertainty of photovoltaic output. SUMMARY

[0006] The technical problem to be solved by the present application is to solve the above-mentioned deficiencies of the prior art, and a method, device and equipment for generating a park energy storage charging and discharging strategy are provided. The method can generate an energy storage charging and discharging strategy for the park, and by executing the strategy, the overall optimal dispatching of multiple goals of the park can be achieved.

[0007] In a first aspect, the present application provides a method for generating a park energy storage charging and discharging strategy, which comprises the following steps:

[0008] obtain distributed energy storage data of current energy storage charging and discharging of the park; and obtain a distributed energy storage energy conversion form of current energy storage charging and discharging of the park;

[0009] select a mathematical model and an operation constraint condition of the distributed energy storage based on the distributed energy storage energy conversion form;

[0010] build a distributed energy storage charging and discharging optimization model covering peak regulation capacity and photovoltaic output prediction deviation based on the distributed energy storage data, the mathematical model and the operation constraint condition;

[0011] optimize the distributed energy storage charging and discharging optimization model based on a preset dynamic game model of leaders and followers to obtain a target optimization model;

[0012] solve the target optimization model based on a variable neighborhood search and multi-objective optimization combination algorithm to obtain a Pareto optimal solution set;

[0013] generate an optimal strategy for energy storage charging and discharging of the park based on a preset target preference and the Pareto optimal solution set.

[0014] Further, the distributed energy storage energy conversion form includes electrochemical energy storage, and / or, physical energy storage, and / or, electromagnetic energy storage;

[0015] If the distributed energy storage energy conversion form is electrochemical energy storage, the mathematical model includes a distributed energy storage charging and discharging efficiency model, a distributed energy storage power capacity model, a distributed energy storage state of charge model, and a distributed energy storage aggregation model, and the operation constraint condition includes a distributed energy storage charging and discharging power constraint condition, a maximum demand constraint condition, an anti-flow constraint condition, an electric quantity balance constraint condition, and a battery state of charge constraint condition.

[0016] Further, the distributed energy storage charging and discharging optimization model includes a peak regulation capacity objective function and a photovoltaic output prediction deviation objective function.

[0017] The peak regulation capacity objective function quantifies the peak regulation target through a load peak value, an average load, and a change in load after peak regulation.

[0018] The peak regulation target includes a minimum load curve standard deviation, and / or, a minimum net load fluctuation, and / or, a minimum grid fluctuation, and / or, a minimum peak-valley difference.

[0019] The photovoltaic output prediction deviation objective function is used to minimize the deviation between a predicted value and an actual value of photovoltaic output.

[0020] Further, the dynamic game model of leaders and followers is specifically:

[0021] The Multi-Follower-Stackelberg game model is composed of a leader objective function, a follower objective function and a game constraint condition;

[0022] The Multi-Follower-Stackelberg game model is implemented through the following steps:

[0023] The leader and the follower in the Multi-Follower-Stackelberg game model are determined, and the objective of the leader, the objective of the follower and the game constraint condition are determined;

[0024] The game constraint condition includes that the total discharge amount of the energy storage is equal to the total charge amount of the energy storage.

[0025] Based on the game constraint condition and according to the objective of the leader and the objective of the follower, a Stackelberg equilibrium is determined, and the Multi-Follower-Stackelberg game model is obtained.

[0026] Further, based on the preset target preference and the Pareto optimal solution set, the optimal energy storage charging and discharging strategy of the park is generated, which includes the following steps:

[0027] Step A1: The specific values of the target preference and various target indicators in the Pareto optimal solution set are uniformly standardized;

[0028] Step A2: The covariance matrix between all target indicators is calculated by using the standardized target indicators;

[0029] Step A3: The eigenvalue matrix is obtained by performing eigenvalue decomposition on the obtained covariance matrix;

[0030] Step A4: The characteristic information contribution rate in the eigenvalue matrix is calculated, and the principal component with a larger characteristic information contribution rate is selected; and based on the selected principal component, the comprehensive score of each eigenvalue in the principal component space is calculated;

[0031] The characteristic information contribution rate is the proportion of each eigenvalue in the total sum of all eigenvalues;

[0032] Step A5: The optimal energy storage charging and discharging strategy of the park is obtained by comparing the comprehensive scores.

[0033] Further, the variable neighborhood search and multi-objective optimization combined algorithm is used to solve the target optimization model to obtain the Pareto optimal solution set, which includes the following steps:

[0034] Step B1: The initial parameters and the maximum number of iterations of the variable neighborhood search and multi-objective optimization combined algorithm are set, and the initial universe population and the initial candidate combination are constructed;

[0035] The initial parameters of the variable neighborhood search combined with multi-objective optimization algorithm include the number of universes, the field structure set, the wormhole existence probability, and the travel distance rate.

[0036] Step B2: Based on the initial parameters of the variable neighborhood search combined with multi-objective optimization algorithm and the initial universe population, local field search and multi-universe interaction are performed to obtain an updated candidate combination, updated parameters of the variable neighborhood search combined with multi-objective optimization algorithm, and an updated universe population.

[0037] Step B3: The current Pareto front is selected from the updated candidate combination.

[0038] Step B4: Based on the updated parameters of the variable neighborhood search combined with multi-objective optimization algorithm and the updated universe population, local field search and multi-universe interaction are performed to obtain a further updated candidate combination, further updated parameters of the variable neighborhood search combined with multi-objective optimization algorithm, and a further updated universe population.

[0039] Step B5: The current Pareto front is selected from the further updated candidate combination.

[0040] Step B6: Steps B4 to B5 are repeated until the maximum number of iterations is met.

[0041] Step B7: The Pareto front of the last iteration is taken as the Pareto optimal solution set.

[0042] In a second aspect, the present application provides a device for generating a park energy storage charging and discharging strategy, which comprises:

[0043] An acquisition unit is configured to acquire distributed energy storage data of current park energy storage charging and discharging, and to acquire a distributed energy storage energy conversion form of current park energy storage charging and discharging.

[0044] A selection unit is connected to the acquisition unit and is configured to select a mathematical model and operating constraints of the distributed energy storage based on the distributed energy storage energy conversion form.

[0045] A construction unit is connected to the acquisition unit and the selection unit and is configured to build a distributed energy storage charging and discharging optimization model covering peak shaving capacity and photovoltaic output prediction deviation based on the distributed energy storage data, the mathematical model, and the operating constraints.

[0046] An optimization unit is connected to the construction unit and is configured to optimize the distributed energy storage charging and discharging optimization model based on a preset leader-follower dynamic game model to obtain a target optimization model.

[0047] The solving unit, connected with the optimization unit, is configured to solve the target optimization model based on a variable neighborhood search and multi-objective optimization combination algorithm to obtain a Pareto optimal solution set;

[0048] The generating unit, connected with the solving unit, is configured to generate the park energy storage charging and discharging optimal strategy based on the preset target preference and the Pareto optimal solution set.

[0049] Further, the generating unit comprises:

[0050] The standardization processing module is configured to perform unified standardization processing on specific values of various target indicators in the target preference and the Pareto optimal solution set;

[0051] The first calculation module, connected with the standardization processing module, is configured to calculate a covariance matrix among all target indicators by using the standardized target indicators;

[0052] The decomposition module, connected with the first calculation module, is configured to perform eigenvalue decomposition on the obtained covariance matrix to obtain an eigenvalue matrix;

[0053] The second calculation module, connected with the decomposition module, is configured to calculate a characteristic information contribution rate in the eigenvalue matrix;

[0054] The selection module, connected with the second calculation module, is configured to select a principal component with a larger characteristic information contribution rate according to the characteristic information contribution rate;

[0055] The third calculation module, connected with the selection module, is configured to calculate a comprehensive score of each eigenvalue in the eigenvalue matrix in the principal component space based on the selected principal component;

[0056] The comparison module, connected with the third calculation module, is configured to obtain the park energy storage charging and discharging optimal strategy by comparing the comprehensive scores.

[0057] Further, the solving unit comprises:

[0058] The initialization module is configured to set initial parameters and a maximum number of iterations of the variable neighborhood search and multi-objective optimization combination algorithm, and to construct an initial universe population and an initial candidate combination;

[0059] The initial parameters of the variable neighborhood search and multi-objective optimization combination algorithm include a number of universes, a set of field structures, a wormhole existence probability, and a travel distance rate;

[0060] The first processing module, connected with the initialization module, is configured to perform local field search and multi-universe interaction based on the initial parameters of the variable neighborhood search and multi-objective optimization combination algorithm and the initial universe population to obtain an updated candidate combination, updated parameters of the variable neighborhood search and multi-objective optimization combination algorithm, and an updated universe population;

[0061] The first screening module is connected with the first processing module, and is used for screening the current Pareto frontier from the updated candidate combination according to the processing result of the first processing module.

[0062] The second processing module is connected with the first processing module, and is used for performing local field search and multi-universe interaction based on the updated variable neighborhood search and multi-objective optimization combined algorithm parameters and the updated universe group, to obtain the again updated candidate combination, the again updated variable neighborhood search and multi-objective optimization combined algorithm parameters and the again updated universe group, until the maximum iteration number is met.

[0063] The second screening module is connected with the second processing module, and is used for continuously screening the current Pareto frontier from the again updated candidate combination according to the processing result of the second processing module, until the second processing module has no processing result.

[0064] The summary module is connected with the first screening module and the second screening module respectively, and is used for taking the Pareto frontier of the last iteration as the Pareto optimal solution set.

[0065] In a third aspect, the present application provides an electronic device, which comprises a processor and a memory, and the memory stores a computer program, when the processor runs the computer program stored in the memory, the processor executes the generation method of the park energy storage charging and discharging strategy according to the first aspect.

[0066] By adopting the multi-objective optimization method, combining the dynamic game model and the heuristic search technology, the present application can systematically generate the energy storage charging and discharging strategy of the park, and realize the overall optimal scheduling of the park multi-objective by effectively executing the strategy. The specific beneficial effects are as follows:

[0067] 1. Improve the scientificity and rationality of energy scheduling: the present application uses the dynamic game model of leader-follower, fully considers the interests of each subject in the park, ensures that the scheduling strategy meets the actual operation demand and has theoretical rationality, and effectively avoids the single and rigidity of the traditional rule-oriented strategy.

[0068] 2. Enhance the accommodation capacity and peak regulation capacity of new energy: based on the constructed charging and discharging optimization model, the present application fully considers the prediction deviation of photovoltaic renewable energy, intelligently adjusts the charging and discharging time of energy storage equipment, and improves the utilization efficiency of photovoltaic power generation.

[0069] 3. Improve the flexibility and adaptability of scheduling: the present application combines the variable neighborhood search and multi-objective optimization algorithm, can quickly find the optimal strategy meeting the multi-objective demand in the Pareto optimal solution set, realizes the rapid response in the dynamic change environment of load, electricity price and renewable energy output, and improves the adaptability of scheduling.

[0070] 4. Achieve the optimal balance of multi-target and personalized customization: the application screens the charging and discharging strategy most suitable for the actual needs of the park from a large number of Pareto optimal schemes according to the preset target preference, meets the diversified targets of different enterprises and managers, and provides strong technical support for the park to achieve the win-win of green low carbon and economic benefit. BRIEF DESCRIPTION OF DRAWINGS

[0071] Figure 1 The generation method of the park energy storage charging and discharging strategy in the embodiment of the application is shown in the schematic diagram.

[0072] Figure 2 The flowchart of generating the park energy storage charging and discharging strategy in the embodiment of the application is shown in the schematic diagram.

[0073] Figure 3 The optical storage park topology in the embodiment of the application is shown in the schematic diagram.

[0074] Figure 4 The DES charging and discharging optimization model in the embodiment of the application is shown in the schematic diagram.

[0075] Figure 5 The charging and discharging optimization model optimization based on Multi-Follower-Stackelberg game in the embodiment of the application is shown in the schematic diagram.

[0076] Figure 6 The VNS-MVO algorithm flowchart in the embodiment of the application is shown in the schematic diagram.

[0077] Figure 7 The principal component analysis flowchart in the embodiment of the application is shown in the schematic diagram.

[0078] Figure 8 The generation device of the park energy storage charging and discharging strategy in the embodiment of the application is shown in the schematic diagram.

[0079] Figure 9 The architecture of the electronic device in the embodiment of the application is shown in the schematic diagram.

[0080] Reference signs: 10, acquisition unit, 20, selection unit, 30, construction unit, 40, optimization unit, 50, solving unit, 60, generation unit, 100, processor, 200, memory. DETAILED DESCRIPTION

[0081] In order for those skilled in the art to better understand the technical solutions of the application, the embodiments of the application will be further described in detail below with reference to the drawings.

[0082] It can be understood that the specific embodiments and drawings described herein are only used to explain the application, but not to limit the application.

[0083] It can be understood that the embodiments in the present application and the features in the embodiments can be combined with each other without conflict.

[0084] It can be understood that, for the convenience of description, only parts related to the present application are shown in the drawings of the present application, and parts irrelevant to the present application are not shown in the drawings.

[0085] It can be understood that each unit and module involved in the embodiments of the present application can correspond to only one entity structure, or can be composed of multiple entity structures, or multiple units and modules can be integrated into one entity structure.

[0086] It can be understood that the functions and steps marked in the flowcharts and block diagrams of the present application can occur in an order different from that marked in the drawings without conflict.

[0087] It can be understood that in the flowcharts and block diagrams of the present application, the architecture, functions and operations of possible implementations of the system, device, equipment and method according to the embodiments of the present application are shown. Each block in the flowchart or block diagram can represent a unit, module, program segment, code, which contains executable instructions for realizing the specified function. Moreover, each block or combination of blocks in the block diagram and flowchart can be realized by a hardware-based system for realizing the specified function, or by a combination of hardware and computer instructions.

[0088] It can be understood that the units and modules involved in the embodiments of the present application can be realized by software or by hardware, for example, the units and modules can be located in a processor.

[0089] Embodiment 1

[0090] The embodiment provides a method for generating a park energy storage charging and discharging strategy. The method is suitable for scenarios such as scheduling management of an intelligent light storage energy park, optimization of new energy consumption, support for power peak shaving and frequency modulation, and improvement of overall operation efficiency of the system, especially in a complex background of multiple enterprises, multiple devices and multiple targets. Through dynamic game and multi-objective optimization, the embodiment realizes collaborative scheduling and benefit balance of energy storage devices, effectively deals with fluctuations and deviations of photovoltaic power generation, guarantees the continuity and stability of power supply, promotes the development of green and low-carbon energy, and provides strong technical support for intelligent power grids and new energy demonstration areas.

[0091] As shown in the drawing, Figure 1 The method for generating the park energy storage charging and discharging strategy specifically includes the following steps:

[0092] Step D1: acquiring distributed energy storage data of current energy storage charging and discharging of the park; and acquiring a distributed energy storage energy conversion form of current energy storage charging and discharging of the park.

[0093] In the process of implementing the park energy storage charging and discharging strategy, it is necessary to first obtain the distributed energy storage data of the system, covering a wide range of information. The data required for model optimization mainly includes: distributed photovoltaic system data such as actual power generation, predicted power generation, grid electricity, and corresponding unit price; distributed energy storage system data including battery charging and discharging status, charging and discharging capacity, and charging capacity of city power and green power, charging and discharging efficiency, maximum charging and discharging power, rated capacity, and energy storage price; park electricity data, including not only city power and green power electricity consumption, but also time-of-use electricity price information; and enterprise electricity data, such as enterprise energy storage capacity and city power consumption. These data are collected through smart meters and sensors under real-time monitoring, transmitted remotely using Internet of Things technology and gateway devices to ensure the timeliness and accuracy of the data, and finally uploaded to the digital twin platform for storage and management.

[0094] In the data processing stage, first, the raw data collected is cleaned to remove invalid information such as outliers and duplicates, improving the reliability of the data. Then, data from different sources is integrated, standardized, and stored in a unified structure, using advanced technologies such as distributed storage and time-series databases to handle the storage and management of massive amounts of data. This series of processing steps ensures the effectiveness of the data in subsequent analysis and decision-making, providing a solid data foundation for multi-objective optimization, model training, and strategy formulation.

[0095] In this embodiment, distributed energy storage can be referred to as DES (DES is the abbreviation of Distributed Energy Storage).

[0096] DES has various energy conversion forms, including electrochemical energy storage such as various battery technologies, physical energy storage such as inertial energy storage, pumped storage, or electromagnetic energy storage such as superconducting magnetic storage. These different energy conversion forms have their own advantages and can be flexibly selected according to specific application requirements, enabling efficient and reliable energy storage and dispatch.

[0097] In constructing the operating constraints of the park energy storage system, for electrochemical energy storage (such as battery systems), it is necessary to ensure energy balance and safe operation, including capacity limitations for energy storage, ensuring that the battery's stored energy does not exceed its rated capacity and is within a safe range throughout the charging and discharging process, combined with maximum charging and discharging current and voltage limits to ensure equipment safety. At the same time, it is necessary to meet the maximum charging and discharging power and rate limitations of the battery to avoid damage due to rapid charging and discharging, and to consider the changes in the battery's state (such as state of charge SOC) to ensure it remains within a safe operating range (such as upper and lower limits of SOC), in addition to considering charging and discharging efficiency and energy loss to ensure the authenticity and feasibility of energy input-output relationships.

[0098] For physical energy storage devices such as pumped hydro or inertial energy storage systems, it is necessary to ensure that the energy storage is within the maximum capacity of the device and does not exceed the maximum capacity limit to prevent device damage or safety hazards. At the same time, the dynamic characteristics of physical energy storage such as the energy conversion speed and efficiency of water pumps, gravity or inertial devices need to be considered to limit the rate of energy flow to meet the operating range of the device. In addition, the operating conditions and safety requirements of the device need to be met to ensure that the mechanical, hydraulic or other physical systems operate in accordance with the design standards and avoid safety problems caused by improper operation.

[0099] For electromagnetic energy storage devices such as superconducting magnetic storage systems, it is necessary to determine the maximum capacity of the magnetic energy storage to ensure that the charging and discharging operations are within a safe and stable range to avoid damage to the device or interference caused by excessive magnetic field strength. Electromagnetic energy storage systems must also comply with magnetic field strength limits to ensure safe operation of the device within the magnetic field parameter range and avoid damage to the device due to excessive magnetic field. In addition, the conversion rate and efficiency of magnetic energy should be ensured to meet the system design requirements, and the energy flow during charging and discharging should be reasonably controlled to achieve efficient and stable energy storage and release.

[0100] Therefore, for different types of energy storage, constructing corresponding operating constraints is a key step to achieve safe and efficient operation of the park energy storage system. These constraints collectively ensure that various energy storage systems work together within a normal, safe and economic range to provide stable and reliable energy dispatching support for the park and meet the needs of multi-objective optimization and intelligent dispatching.

[0101] Step D2: Based on the distributed energy storage energy conversion form, select the mathematical model and operating constraints of the distributed energy storage.

[0102] The DES energy conversion form of this embodiment includes one or more of electrochemical energy storage, physical energy storage, and electromagnetic energy storage. If the distributed energy storage energy conversion form is electrochemical energy storage, the mathematical model includes a distributed energy storage charging and discharging efficiency model, a distributed energy storage power capacity model, a distributed energy storage state of charge model, a distributed energy storage aggregation model, and operating constraints include distributed energy storage charging and discharging power constraints, maximum demand constraints, anti-backflow constraints, power balance constraints, and battery state of charge constraints.

[0103] This embodiment mainly studies the case where the DES energy conversion form is electrochemical energy storage.

[0104] When the DES energy conversion form is electrochemical energy storage, the mathematical model includes a DES charging and discharging efficiency model, a DES power capacity model, a DES state of charge model, a DES aggregation model, and a DES mathematical basis model of equivalent aggregation parameters;

[0105] The calculation formula of the DES charging and discharging efficiency model is as follows:

[0106]

[0107] wherein,

[0108] η ch represents the energy storage charging efficiency;

[0109] η dis represents the energy storage discharging efficiency;

[0110] E es represents the energy storage rated capacity, which is 2 MWh in the embodiment;

[0111] k represents the battery characteristic coefficient, which is 0.9 in the embodiment;

[0112] c represents the charging and discharging rate, which is 1C in the embodiment; that is, the energy storage device can charge or discharge an electric quantity equal to its rated capacity 1 times per hour;

[0113] E loss,ch represents the charging loss electric quantity;

[0114] E loss,dis represents the discharging loss electric quantity;

[0115] r dis represents the self-discharge coefficient;

[0116] The DES charging and discharging efficiency model is combined with the problem that the increase of the charging and discharging rate causes the polarization effect and leads to the efficiency reduction, and the self-discharge coefficient r dis = 0.08% is introduced, and the following is obtained.

[0117] The calculation formula of the charging loss electric quantity E loss,ch and the discharging loss electric quantity E loss,dis is as follows:

[0118]

[0119] wherein, N represents the number of battery groups; N = 100 in the embodiment;

[0120] P loss,ch represents the charging loss power;

[0121] P loss,dis represents the discharging loss power;

[0122] end-ch,t represents the charging end time;

[0123] start-ch,t represents the charging start time;

[0124] end-dis,t represents the discharging end time;

[0125] start-dis, t represents the discharge start time;

[0126] The DES power capacity model includes a charging power model, a discharging power model, and an electric quantity change calculation model;

[0127] The charging power model is as follows:

[0128]

[0129] wherein, P es,ch,t represents the charging power of the energy storage at time t;

[0130] P es,t represents the rated power of the energy storage at time t;

[0131] The discharging power model is as follows:

[0132]

[0133] A power value less than 0 represents charging, and a power value greater than 0 represents discharging;

[0134] wherein, P es,dis,t represents the discharging power of the energy storage at time t;

[0135] The electric quantity change calculation model is as follows:

[0136]

[0137] wherein, E(t+1) represents the electric quantity at time t+1;

[0138] E(t) represents the electric quantity at time t;

[0139] The calculation formula of the DES state of charge model is as follows:

[0140]

[0141] wherein,

[0142] S SOC,t represents the SOC at time t;

[0143] S SOC,t-1 represents the SOC at time t-1;

[0144] The calculation formula of the DES aggregation model is as follows:

[0145]

[0146] The model considers energy balance of the energy storage system, operating state of the energy storage device, economy of the operating strategy, and other factors, and designs a power distribution method to make only one energy storage variable in the same power interval and ensure relative balance of the SOC of each energy storage unit.

[0147] wherein p iess represents the rated power of the battery pack i;

[0148] p ic represents the charging power of the battery pack i;

[0149] p id represents the discharging power of the battery pack i;

[0150] p jhc represents the total charging power of the scheduling demand;

[0151] p jhd represents the total discharging power of the scheduling demand;

[0152] f c (SOC i ) represents the charging SOC function of the i-th battery pack;

[0153] f d (SOC i ) represents the discharging SOC function of the i-th battery pack;

[0154] f c (SOC j ) represents the charging SOC function of the j-th battery pack;

[0155] f d (SOC j ) represents the discharging SOC function of the j-th battery pack;

[0156] The calculation formula of the charging and discharging SOC function is as follows:

[0157]

[0158] wherein e -20(x-0.5) represents the -20(x-0.5) power of natural logarithm e;

[0159] The expression of the equivalent aggregation parameter is as follows:

[0160]

[0161] wherein,

[0162] M represents the total number of distributed energy storages, that is, the number of battery packs, which is the same as N above, both representing the number of battery packs, but M is valued at 10 here;

[0163] P es denotes the rated power of the equivalent aggregated energy storage;

[0164] E es denotes the rated capacity of the equivalent aggregated energy storage;

[0165] η ic denotes the charging efficiency of the i-th distributed energy storage;

[0166] η id denotes the discharging efficiency of the i-th distributed energy storage;

[0167] η jhc denotes the equivalent centralized charging efficiency;

[0168] η jhd denotes the equivalent centralized discharging efficiency.

[0169] The function of the charge-discharge SOC function is to reduce the charging power of high SOC and the discharging power of low SOC, and to increase the discharging power of high SOC and the charging power of low SOC, so as to make the power distribution more reasonable.

[0170] The embodiment can optimize the operation of the distributed energy storage system in the pure photovoltaic charging mode and the hybrid charging mode, and construct a DES charge-discharge real-time correction model with the objectives of maximizing the peak shaving capacity and maximizing the compensation of photovoltaic output prediction deviation. In the model, the decision variable is the charge-discharge power of the DES every 15 minutes within a day, and dynamic regulation is achieved through optimization. The constraint conditions of the model include the charge-discharge power limit, capacity constraint, maximum demand limit, anti-flow requirement, power balance, and SOC (state of charge) constraint of the DES, to ensure the safety and reliability of the operation. At the same time, the topological structure of the photovoltaic storage park is as shown in Figure 3 , and the specific framework of the entire DES charge-discharge optimization model is as shown in Figure 4 , which provides a theoretical basis and technical solution for efficient and stable energy scheduling.

[0171] Figure 3 The power system topology of a photovoltaic storage park is shown, which includes the grid, photovoltaic power generation system, and distributed energy storage system (DES). The grid and photovoltaic system serve as the main source of electricity, supplying power to the energy storage system. The energy storage system, as the core component, is responsible for storing and regulating electrical energy to optimize energy use and improve power supply reliability. The energy storage system supplies power to enterprises 1, 2, and 3, etc. through charge and discharge, and these enterprises contain lighting, electric vehicles, factory air conditioning, and other electrical equipment. The entire system design aims to integrate photovoltaic power generation and energy storage technology to achieve efficient use of energy, reduce costs, and support peak shaving, grid scheduling, and carbon emission reduction, among other goals.

[0172] AsFigure 4 As shown, the distributed energy storage charging and discharging optimization model aims to optimize the charging and discharging power decision of the energy storage system by comprehensively considering the data inputs such as time-of-use electricity price, on-grid electricity price, photovoltaic predicted power, actual power and energy storage system parameters, and the uncertainty of photovoltaic output. The goal of the model is to improve the peak shaving capability of the system and compensate for the photovoltaic output prediction deviation, while meeting various constraint conditions including charging and discharging power constraints, maximum demand constraints, anti-backflow constraints, power balance constraints and SOC constraints, to achieve cost-effectiveness maximization and energy utilization efficiency improvement.

[0173] As a specific implementation, the DES energy conversion form is electrochemical energy storage, and the operating constraints of the distributed energy storage specifically include:

[0174] DES charging and discharging power constraints, maximum demand constraints, anti-backflow constraints, power balance constraints, and SOC constraints;

[0175] The calculation formula of the DES charging and discharging power constraint is as follows:

[0176]

[0177] Wherein,

[0178] P es,ch (t) represents the energy storage charging power at time t;

[0179] P es,dis (t) represents the energy storage discharging power at time t;

[0180] P es,ch,max represents the maximum charging power of the energy storage;

[0181] P es,dis,max represents the maximum discharging power of the energy storage;

[0182] The calculation formula of the maximum demand constraint is as follows:

[0183]

[0184] Wherein, σ represents the maximum load rate of the transformer;

[0185] P B represents the rated power of the transformer;

[0186] P load,ch,avg represents the average load of the charging period;

[0187] P load,dis,avg represents the average load of the discharging period;

[0188] T ch represents the charging period;

[0189] T dis denotes a discharging period;

[0190] The calculation formula of the anti-flow constraint is as follows:

[0191] P es,dis (t) denotes a load at time t; load (t)≤0;

[0192] wherein,

[0193] P load (t) denotes a load at time t;

[0194] The calculation formula of the power balance constraint is as follows:

[0195] E(t)=E(t-1)+η ch ·P esch (t) denotes a load at time t; esdis (t) / η dis ;

[0196] wherein,

[0197] E(t) denotes an energy at time t;

[0198] E(t-1) denotes an energy at time t-1;

[0199] η ch denotes a storage charging efficiency;

[0200] η dis denotes a storage discharging efficiency;

[0201] The calculation formula of the SOC constraint is as follows:

[0202]

[0203] SOC SOC,start (t) denotes an SOC at a starting time of storage;

[0204] SOC SOC,end (t) denotes an SOC at an ending time of storage;

[0205] E es denotes a rated capacity of an equivalent aggregated storage.

[0206] Step D3: According to the distributed storage data, the mathematical model and the operation constraint condition, a distributed storage charging and discharging optimization model covering the peak shaving capacity and the photovoltaic output prediction deviation is built.

[0207] The distributed storage charging and discharging optimization model includes a peak shaving capacity objective function and a photovoltaic output prediction deviation objective function;

[0208] The peak regulation capability objective function quantifies the peak regulation target through the load peak value, the average load, and the change of the load after peak regulation.

[0209] The peak regulation target includes the minimum standard deviation of the load curve, the minimum net load fluctuation, the minimum grid fluctuation, and the minimum peak-valley difference.

[0210] The photovoltaic output prediction deviation objective function is used to minimize the deviation between the predicted value and the actual value of the photovoltaic output.

[0211] As a specific embodiment, the distributed energy storage charging and discharging optimization model (DES charging and discharging optimization model) specifically includes the peak regulation capability objective function and the photovoltaic output prediction deviation objective function.

[0212] The peak regulation target function is the minimum standard deviation of the load curve, and the peak regulation target function is determined based on the load power after peak regulation, the average load power after peak regulation, and the peak regulation action period, and the cumulative number of peak regulation times.

[0213] The calculation formula of the peak regulation capability objective function is as follows:

[0214]

[0215] Wherein, f2 represents the load fluctuation size;

[0216] P load,e (t) represents the load power after peak regulation;

[0217] P load,e,avg represents the average load power after peak regulation;

[0218] P load,f represents the load power peak value;

[0219] start,t represents the start time;

[0220] end,t represents the end time;

[0221] Δt represents the interval time between the start time and the end time;

[0222] P es,dis (t) represents the energy storage discharging power at time t;

[0223] P es,ch (t) represents the energy storage charging power at time t;

[0224] The photovoltaic output prediction deviation objective function is as follows:

[0225]

[0226] Wherein,

[0227] f3 represents a photovoltaic output prediction deviation;

[0228] P PV,yc (t) represents a photovoltaic power prediction value at time t;

[0229] P PV,sj (t) represents a photovoltaic power actual value at time t;

[0230] wherein the uncertainty of the photovoltaic output is described by a budget uncertainty set, the photovoltaic output prediction deviation f3 is expressed as:

[0231] The photovoltaic power set is expressed as follows:

[0232]

[0233] wherein the response deviation of the photovoltaic output is subject to a probability distribution with an expectation of 0 and a covariance of ω;

[0234] represents a photovoltaic output power set;

[0235] P PV,i represents a photovoltaic output power;

[0236] ΔP PV,i represents a photovoltaic output power deviation;

[0237] ω represents a covariance;

[0238] ε ηi represents a confidence constant when the confidence of the photovoltaic output power is η, there are n confidence constants, and n is a natural number greater than 1.

[0239] The budget uncertainty set can cover more effective data, eliminate more invalid areas, improve the precision of the description of uncertainty, compress the uncertainty measure, and has linear adjustable ability, and is more suitable for practical application.

[0240] Step D4: based on the preset dynamic game model of leaders and followers, the distributed energy storage charging and discharging optimization model is optimized to obtain a target optimization model.

[0241] As a specific implementation, the dynamic game model of leaders and followers is specifically a Multi-Follower-Stackelberg game model.

[0242] The Multi-Follower-Stackelberg game model includes the following steps in the implementation process:

[0243] determining the leader and the follower in the Multi-Follower-Stackelberg game model, and determining the target of the leader, the target of the follower and the game constraint condition;

[0244] wherein the game constraint condition comprises that the total discharging capacity of the energy storage is equal to the total charging capacity of the energy storage;

[0245] based on the game constraint condition, and according to the target of the leader and the target of the follower, determining the Stackelberg equilibrium, to obtain the Multi-Follower-Stackelberg game model.

[0246] Specifically, the Multi-Follower-Stackelberg game model comprises a game model of the leader, a game model of the follower and a game constraint condition;

[0247] The calculation formula of the game model of the leader is as follows:

[0248]

[0249] wherein,

[0250] f 1 represents the target income of the first enterprise;

[0251] P buy,i (t) represents the electric power purchased by the i-th enterprise from the DES, 2≤i≤K+1;

[0252] P buy,1 (t) represents the electric power purchased by the first enterprise from the DES;

[0253] c des,sell represents the specified energy storage price;

[0254] E es,chsd (t) represents the charging capacity of the DES at time t;

[0255] η dis represents the discharging efficiency;

[0256] η ch represents the charging efficiency;

[0257] c price represents the charging price;

[0258] E es,chpv (t) represents the discharging capacity of the DES at time t;

[0259] c load represents the load price;

[0260] The calculation formula of the game model of the follower is as follows:

[0261]

[0262] where f i represents the cost of the i-th enterprise, 2≤i≤K+1;

[0263] P buy,i,grid (t) represents the electrical power purchased by the i-th enterprise from the grid, 2≤i≤K+1;

[0264] The calculation formula of the game constraint condition is as follows:

[0265]

[0266] Where K+1 enterprises in the light storage park, the first one is the leader, the second to the K+1 one is the follower, the sampling step is 15 minutes, and the optimization time is 24 hours.

[0267] As Figure 5 shown, in this light storage park energy management strategy based on the Multi-Follower-Stackelberg game model, enterprise 1 as the leader is responsible for the construction and management of distributed energy storage system (DES), the goal is to maximize its benefit in a day, which may be through the sale of electricity to the follower enterprise 2 and enterprise 3 and profit from it. Enterprise 2 and enterprise 3 as the follower, the goal is to minimize the respective electricity cost, which involves the price of purchasing electricity from enterprise 1. The model coordinates the interests of the three parties by setting the energy storage price, ensures that enterprise 1 can optimize its energy storage strategy to maximize the benefit, while enterprise 2 and enterprise 3 can minimize the electricity cost. In addition, the model also considers the uncertainty of photovoltaic output, quantifies through the prediction of uncertainty index, and sets a variety of constraint conditions including charge and discharge power constraint, maximum demand constraint, anti-flow constraint, power balance constraint and SOC constraint, to realize the efficient use and cost optimization of energy in the light storage park.

[0268] Step D5: based on the variable neighborhood search and multi-objective optimization combined algorithm, the target optimization model is solved to obtain the Pareto optimal solution set.

[0269] In this embodiment, the algorithm combining variable neighborhood search and multi-objective optimization can be simply referred to as VNS-MVO algorithm (Variable Neighborhood Search-Multi-Objective Optimization Algorithm).

[0270] As a specific implementation, based on the VNS-MVO algorithm, the target optimization model is solved to obtain the Pareto optimal solution set, which specifically includes the following steps:

[0271] Step B1: setting variable neighborhood search and multi-objective optimization combined algorithm initial parameters and maximum iteration number; and constructing initial universe population and initial candidate combination;

[0272] Among them, the initial parameters of the variable neighborhood search and multi-objective optimization combined algorithm include the number of universes, the set of field structures, the wormhole existence probability, and the travel distance rate.

[0273] Step B2: based on the initial parameters of the variable neighborhood search and multi-objective optimization combined algorithm and the initial universe population, performing local field search and multi-universe interaction to obtain updated candidate combination, updated variable neighborhood search and multi-objective optimization combined algorithm parameters, and updated universe population;

[0274] Step B3: screening the current Pareto front from the updated candidate combination;

[0275] Step B4: based on the updated variable neighborhood search and multi-objective optimization combined algorithm parameters and the updated universe population, performing local field search and multi-universe interaction to obtain again updated candidate combination, again updated variable neighborhood search and multi-objective optimization combined algorithm parameters, and again updated universe population;

[0276] Step B5: screening the current Pareto front from the again updated candidate combination;

[0277] Step B6: repeating steps B4 to B5 until the maximum iteration number is met;

[0278] Step B7: taking the Pareto front of the last iteration as the Pareto optimal solution set.

[0279] The local field search is specifically: in the neighborhood of the current solution, fine-tuning is performed on the target variables using predefined neighborhood structures, and by changing the dimensions or parameters of the solution, a better local solution is found.

[0280] The multi-universe interaction is specifically: in the entire population, using the black hole mechanism, the better performing universes are selected according to the roulette algorithm for information transmission and resource exchange.

[0281] The calculation process of the VNS-MVO algorithm is as follows Figure 6As shown, the algorithm nests the neighborhood disturbance mechanism of VNS (Variable Neighborhood Search) in the cosmic migration strategy of the multi-objective optimization algorithm MVO (Multi-Objective Variable Neighborhood Search Algorithm) to form a double-layer disturbance mechanism to enhance the search ability. In the local search stage, the quality of the solution is evaluated by the dominance relationship in the objective space, and the fine-grained neighborhood disturbance technology of VNS is combined to disturb the current solution in different neighborhood structures at the microscopic level through structured neighborhood disturbance, such as randomly selecting two different dimensional values for exchange. When the search falls into local optimum, the neighborhood structure switching is performed in a Gaussian disturbance manner to break the local restriction and reasonably constrain the boundary of the solution to generate new candidate solutions (cosmos). At the macroscopic level, the intercosmic matter exchange mechanism of MVO is combined to transfer matter in each cosmos through the "black hole" mechanism, select the cosmos with better performance based on the roulette selection, perform global migration in a wormhole jumping manner, and adjust the migration amplitude of different dimensions according to the travel distance rate, so as to generate a diverse solution set after jumping. In the iteration process, the probability of wormhole existence α and the travel distance rate β are adjusted after each dynamic update to balance the efficiency of global exploration and local search. After each iteration, the new and old solution sets are fused, and the non-dominated sorting and crowded distance screening are performed to select the Pareto optimal solution, and the search convergence speed and diversity distribution are gradually improved by updating the parameters α and β. The whole process continues until the preset maximum iteration number MAX is reached, and finally a set of Pareto optimal solution set is output. The fusion framework effectively combines the fine-grained local search of VNS and the global exploration ability of MVO, can efficiently solve complex multi-objective optimization problems, ensure fast convergence speed, and also obtain a well-distributed and diverse Pareto frontier, which well balances the optimization effect and search efficiency.

[0282] Step D6: Based on the preset target preference and the Pareto optimal solution set, the park energy storage charging and discharging optimal strategy is generated.

[0283] As a specific implementation, the park energy storage charging and discharging strategy is generated according to the principal component analysis method. Figure 7A flowchart of the principal component analysis (PCA) method is shown, and each step of extracting main features from raw data to simplify the data structure while retaining key information is described in detail. First, the raw data needs to be standardized to convert the value of each indicator to a standard normal distribution with a mean of 0 and a standard deviation of 1, so as to eliminate the dimensional differences between indicators and ensure the comparability of each indicator. Then, the correlation coefficient between each indicator in the standardized data set is calculated to form a correlation coefficient matrix, which is used to measure the linear correlation between different indicators. Then, eigenvalue decomposition is performed on the correlation coefficient matrix to obtain the eigenvalues and corresponding eigenvectors, and the eigenvalues reflect the importance of each principal component, while the eigenvectors indicate the direction of main data change. Then, according to the size of the eigenvalues, select several features with the largest contribution rate as the principal components, usually select the total contribution rate of the principal components close to 100% and not more than 5, to ensure that the main information of the data is included. Finally, use the selected principal components and their corresponding eigenvectors to calculate the score of each sample in the principal component space, which is simple and clear, and provides a concise and efficient feature representation for subsequent pattern recognition, data analysis and decision-making.

[0284] As a specific implementation, the park energy storage charging and discharging strategy is generated according to the principal component analysis method. Based on the preset target preference and the Pareto optimal solution set, the park energy storage charging and discharging strategy is generated, which specifically includes the following steps:

[0285] Step A1: The specific values of the target preference and various target indicators in the Pareto optimal solution set are uniformly standardized;

[0286] Standardized indicators The calculation formula is as follows:

[0287]

[0288] Wherein, x ij represents the original value of the jth indicator of the ith sample;

[0289] μ j represents the mean of the jth indicator;

[0290] s j represents the standard deviation of the jth indicator;

[0291] Step A2: Calculate the covariance matrix between all target indicators using the standardized target indicators;

[0292] The calculation formula of the covariance matrix X 方差矩阵 is as follows:

[0293]

[0294] Wherein, I is the total number of indicators of sample K;

[0295] is the covariance matrix of sample K and indicator i;

[0296] is the covariance matrix of sample K and indicator j;

[0297] Step A3: eigenvalue decomposition is performed on the obtained covariance matrix to obtain an eigenvalue matrix;

[0298] The eigenvalue of the covariance matrix is calculated according to the following formula:

[0299]

[0300] Wherein, Y j represents the eigenvalue of the jth principal component;

[0301] represents the first covariance matrix;

[0302] represents the second covariance matrix;

[0303] represents the third covariance matrix;

[0304] represents the nth covariance matrix;

[0305] u1j represents the eigenvector of the first covariance matrix;

[0306] u2j represents the eigenvector of the second covariance matrix;

[0307] u3j represents the eigenvector of the third covariance matrix;

[0308] u nj represents the eigenvector of the nth covariance matrix;

[0309] Step A4: calculate the characteristic information contribution rate in the eigenvalue matrix, and select the principal component with larger characteristic information contribution rate; and based on the selected principal component, calculate the comprehensive score of each eigenvalue in the principal component space in the eigenvalue matrix;

[0310] Wherein, the characteristic information contribution rate is the proportion of each eigenvalue in the total sum of all eigenvalues;

[0311] The calculation formula of the cumulative contribution rate is as follows:

[0312]

[0313] Wherein, β represents the cumulative contribution rate;

[0314] λ k λ represents the eigenvalue of the jth principal component; P represents the number of principal components; and m represents the total number of components;

[0315] The calculation formula of the comprehensive score Z is as follows:

[0316]

[0317] wherein b j λ represents the eigenvalue of the jth principal component; P represents the number of principal components; and m represents the total number of components;

[0318] Y j λ represents the eigenvalue of the jth principal component; P represents the number of principal components; and m represents the total number of components;

[0319] The calculation formula of the characteristic information contribution rate is as follows:

[0320]

[0321] wherein,

[0322] λ j λ represents the eigenvalue of the jth principal component; P represents the number of principal components; and m represents the total number of components;

[0323] Step A5: Obtain the optimal strategy of energy storage charging and discharging in the park by comparing the comprehensive scores.

[0324] The embodiment can also first establish a mathematical model, as shown in the following formula: Figure 2 As shown in the following formula, a mathematical model is first needed to describe the behavior and characteristics of the distributed energy storage system, which helps to analyze and predict the performance of the energy storage system under different working conditions. Subsequently, a DES charging and discharging optimization model considering photovoltaic uncertainty is constructed, and methods such as probability analysis, stochastic process or scenario analysis are used to solve the uncertainty brought by photovoltaic power generation, and to determine the charging and discharging strategy under different modes. Then, the optimization model is modified (or optimized) based on the Multi-Follower-Stackelberg game theory, which can coordinate the interests of multiple decision makers (such as different enterprises or users), so as to realize the overall optimization of the system. After that, the data in the photovoltaic storage park are collected and processed, including photovoltaic power generation, load demand and electricity price information, etc., to provide input and verification basis for the optimization model. Subsequently, with the aid of variable neighborhood search (VNS) and multi-objective optimization (MVO) algorithm, the charging and discharging strategy of DES is solved, and the heuristic search ability of VNS is combined with the multi-objective optimization characteristics of MVO to effectively explore the solution space to find the optimal solution. Finally, the principal component analysis (PCA) technology is used to extract the key features from multiple indexes, simplify the decision-making process, and determine the optimal charging and discharging strategy of the park, so as to realize efficient and scientific energy dispatching management.

[0325] To illustrate, assume that there are 150 different energy storage charging and discharging strategies in the optimal Pareto solution set. In order to further screen the optimal scheme, principal component analysis (PCA) is used to analyze the 150 strategies. First, the standard deviation of each strategy on different indicators is calculated, and a standard deviation table is arranged; then, a covariance matrix between indicators is established to evaluate the correlation between indicators. Next, eigenvalue decomposition is performed to obtain the eigenvalues and corresponding eigenvectors (eigenvalue vectors) of each indicator, and the contribution rate of each indicator (indicating the proportion of each principal component to the overall variation) is calculated. Finally, combined with the contribution rate, the indicators are weighted to calculate the comprehensive score of each strategy, as shown in the following five tables.

[0326] Table 1: Standard Deviation Table:

[0327]

[0328] Table 2: Covariance Matrix Table:

[0329]

[0330] Table 3: Eigenvalue Vector Table:

[0331]

[0332] Table 4: Contribution Rate Table:

[0333]

[0334] Table 5: Top Five Comprehensive Score Table:

[0335] Strategy Strategy No. 56 Strategy No. 121 Strategy No. 41 Strategy No. 96 Strategy No. 5 Composite Score 1.76 1.75 1.55 1.51 1.37

[0336] After sorting the comprehensive scores of all strategies, the strategy with the highest score is selected as the optimal charging and discharging scheme. The analysis results show that strategy No. 56 has the highest comprehensive score and is therefore determined as the optimal charging and discharging strategy for the system. Through this method, the most representative and optimized operation scheme can be systematically and scientifically screened.

[0337] The embodiment proposes a complete park energy storage charging and discharging strategy generation method, which is suitable for scheduling management, new energy consumption and system optimization of intelligent energy park. Especially in the complex environment of multiple enterprises, multiple devices and multiple targets, combined with dynamic game and multi-objective optimization technology, the collaborative scheduling and benefit balance of energy storage devices are realized, the photovoltaic fluctuation is effectively dealt with, the power supply stability is guaranteed, and the green energy development is promoted. The specific process includes: first, collecting and cleaning multi-source data, establishing operation constraints considering different energy conversion modes (electrochemical, physical and electromagnetic); In the model, the capacity, safety, power, SOC and other restrictions are integrated to maximize the peak shaving capacity and reduce the photovoltaic output deviation. Then, a multi-leader-follower game model is used to coordinate the interests of enterprises, and a VNS-MVO algorithm is used for multi-objective optimization to obtain a Pareto optimal solution set. On this basis, principal component analysis (PCA) is used to reduce the dimension and extract the features of 150 optimization solutions, and the optimal strategy No. 56 is selected according to the comprehensive score, which effectively realizes scientific decision-making under the condition of complex, multi-index and multi-objective, and provides strong technical support for intelligent power grid and new energy demonstration area.

[0338] Embodiment 2:

[0339] As shown in Figure 8 The embodiment provides a park energy storage charging and discharging strategy generation device, which comprises:

[0340] The acquisition unit 10 is used for acquiring the distributed energy storage data of the current energy storage charging and discharging of the park, and is also used for acquiring the energy conversion form of the distributed energy storage of the current energy storage charging and discharging of the park;

[0341] The selection unit 20 is connected with the acquisition unit 10, and is used for selecting the mathematical model and the operation constraint condition of the distributed energy storage based on the energy conversion form of the distributed energy storage;

[0342] The construction unit 30 is connected with the acquisition unit 10 and the selection unit 20 respectively, and is used for building a distributed energy storage charging and discharging optimization model covering the peak shaving capacity and the photovoltaic output prediction deviation according to the distributed energy storage data, the mathematical model and the operation constraint condition;

[0343] The optimization unit 40 is connected with the construction unit 30, and is used for optimizing the distributed energy storage charging and discharging optimization model based on a preset dynamic game model of leaders and followers to obtain a target optimization model;

[0344] The solving unit 50 is connected with the optimization unit 40, and is used for solving the target optimization model based on a variable neighborhood search and multi-objective optimization combination algorithm to obtain a Pareto optimal solution set;

[0345] The generating unit 60 is connected with the solving unit 50, and is configured to generate the park energy storage charging and discharging optimal strategy based on the preset target preference and the Pareto optimal solution set.

[0346] As a specific implementation, the generating unit 60 comprises:

[0347] a standardization processing module configured to perform unified standardization processing on specific values of various target indicators in the target preference and the Pareto optimal solution set;

[0348] a first calculation module connected with the standardization processing module and configured to calculate a covariance matrix among all target indicators by using the standardized target indicators;

[0349] a decomposition module connected with the first calculation module and configured to perform eigenvalue decomposition on the obtained covariance matrix to obtain an eigenvalue matrix;

[0350] a second calculation module connected with the decomposition module and configured to calculate a characteristic information contribution rate in the eigenvalue matrix;

[0351] a selection module connected with the second calculation module and configured to select principal components with a larger characteristic information contribution rate according to the characteristic information contribution rate;

[0352] a third calculation module connected with the selection module and configured to calculate a comprehensive score of each eigenvalue in the eigenvalue matrix in the principal component space based on the selected principal components;

[0353] a comparison module connected with the third calculation module and configured to obtain the park energy storage charging and discharging optimal strategy by comparing the comprehensive scores.

[0354] As a specific implementation, the solving unit 50 comprises:

[0355] an initialization module configured to set initial parameters and a maximum number of iterations of a variable neighborhood search and multi-objective optimization combined algorithm, and to construct an initial universe population and an initial candidate combination;

[0356] The initial parameters of the variable neighborhood search and multi-objective optimization combined algorithm include a number of universes, a set of field structures, a wormhole existence probability, and a travel distance rate.

[0357] a first processing module connected with the initialization module and configured to perform local field search and multi-universe interaction based on the initial parameters of the variable neighborhood search and multi-objective optimization combined algorithm and the initial universe population, to obtain an updated candidate combination, updated parameters of the variable neighborhood search and multi-objective optimization combined algorithm, and an updated universe population;

[0358] The first screening module is connected with the first processing module, and is configured to screen the current Pareto front from the updated candidate combination according to a processing result of the first processing module.

[0359] The second processing module is connected with the first processing module, and is configured to perform local field search and multi-universe interaction based on the updated variable neighborhood search and multi-objective optimization combined algorithm parameters and the updated universe population, to obtain again updated candidate combination, again updated variable neighborhood search and multi-objective optimization combined algorithm parameters and again updated universe population, until a maximum iteration number is met.

[0360] The second screening module is connected with the second processing module, and is configured to constantly screen the current Pareto front from the again updated candidate combination according to a processing result of the second processing module, until the second processing module has no processing result.

[0361] The summary module is connected with the first screening module and the second screening module respectively, and is configured to take the Pareto front of the last iteration as a Pareto optimal solution set.

[0362] The device in the embodiment can execute the method in Embodiment 1.

[0363] Embodiment 3

[0364] As shown in Figure 9 The embodiment provides an electronic device, which comprises a processor 100 and a memory 200. The memory 200 stores a computer program. When the processor 100 runs the computer program stored in the memory 200, the processor 100 executes the generation method of the park energy storage charging and discharging strategy according to Embodiment 1.

[0365] It can be understood that the above embodiments are only exemplary embodiments for illustrating the principles of the present application, and the present application is not limited thereto. Various modifications and improvements can be made by those skilled in the art without departing from the spirit and essence of the present application, and these modifications and improvements are also considered to be within the protection scope of the present application.

Claims

1. A method for generating a park energy storage charging and discharging strategy, characterized in that, The method comprises the following steps: obtaining distributed energy storage data of current energy storage charging and discharging of the park; and obtaining a distributed energy storage energy conversion form of current energy storage charging and discharging of the park; based on the distributed energy storage energy conversion form, selecting a mathematical model of distributed energy storage and an operating constraint condition; based on the distributed energy storage data, the mathematical model and the operating constraint condition, building a distributed energy storage charging and discharging optimization model covering peak shaving capacity and photovoltaic output prediction deviation; based on a preset dynamic game model of leaders and followers, optimizing the distributed energy storage charging and discharging optimization model to obtain a target optimization model; based on a variable neighborhood search and multi-objective optimization combination algorithm, solving the target optimization model to obtain a Pareto optimal solution set; based on a preset target preference and the Pareto optimal solution set, generating an optimal strategy for energy storage charging and discharging of the park.

2. The method of claim 1, wherein the distributed energy storage energy conversion form comprises electrochemical energy storage, and / or physical energy storage, and / or electromagnetic energy storage; if the distributed energy storage energy conversion form is electrochemical energy storage, the mathematical model comprises a distributed energy storage charging and discharging efficiency model, a distributed energy storage power capacity model, a distributed energy storage state of charge model, and a distributed energy storage aggregation model, and the operating constraint condition comprises a distributed energy storage charging and discharging power constraint condition, a maximum demand constraint condition, an anti-flow constraint condition, an electric quantity balance constraint condition, and a battery state of charge constraint condition.

3. The method of claim 1, wherein the distributed energy storage charging and discharging optimization model comprises a peak shaving capacity objective function and a photovoltaic output prediction deviation objective function; the peak shaving capacity objective function quantifies a peak shaving target through a load peak value, an average load, and a change in load after peak shaving; wherein the peak shaving target comprises a minimum standard deviation of load curve, and / or a minimum net load fluctuation, and / or a minimum grid fluctuation, and / or a minimum peak-valley difference; the photovoltaic output prediction deviation objective function is used to minimize the deviation between a predicted value and an actual value of photovoltaic output.

4. The method of claim 1, wherein the dynamic game model of leaders and followers is specifically: a Multi-Follower-Stackelberg game model composed of a leader objective function, a follower objective function, and a game constraint condition; the implementation process of the Multi-Follower-Stackelberg game model specifically comprises the following steps: determining the leader and the follower in the Multi-Follower-Stackelberg game model, and determining the target of the leader, the target of the follower, and the game constraint condition; wherein the game constraint condition comprises that the total discharging amount of energy storage is equal to the total charging amount of energy storage. ​ ​ ​ A Stackelberg equilibrium is determined based on a game constraint condition and according to a target of the leader and a target of the follower, to obtain a Multi-Follower-Stackelberg game model.

5. The method of claim 1, wherein the optimal charging and discharging strategy of the park energy storage is generated based on the preset target preference and the Pareto optimal solution set, and specifically comprises the following steps: Step A1: standardizing the specific values of the target preference and the various target indicators in the Pareto optimal solution set; Step A2: calculating the covariance matrix between all the target indicators by using the standardized target indicators; Step A3: performing eigenvalue decomposition on the obtained covariance matrix to obtain an eigenvalue matrix; Step A4: calculating the characteristic information contribution rate in the eigenvalue matrix and selecting the principal component with a larger characteristic information contribution rate; and calculating the comprehensive score of each eigenvalue in the principal component space based on the selected principal component; wherein the characteristic information contribution rate is the proportion of each eigenvalue in the total sum of all eigenvalues; Step A5: obtaining the optimal charging and discharging strategy of the park energy storage by comparing the comprehensive scores.

6. The method of claim 1 or 2 or 3 or 4 or 5, wherein the target optimization model is solved based on a variable neighborhood search and multi-objective optimization combined algorithm to obtain a Pareto optimal solution set, and specifically comprises the following steps: Step B1: setting initial parameters and a maximum number of iterations of the variable neighborhood search and multi-objective optimization combined algorithm, and constructing an initial universe population and an initial candidate combination; wherein the initial parameters of the variable neighborhood search and multi-objective optimization combined algorithm include the number of universes, the set of field structures, the wormhole existence probability, and the travel distance rate; Step B2: performing local field search and multi-universe interaction based on the initial parameters of the variable neighborhood search and multi-objective optimization combined algorithm and the initial universe population to obtain an updated candidate combination, updated parameters of the variable neighborhood search and multi-objective optimization combined algorithm, and an updated universe population; Step B3: selecting the current Pareto frontier from the updated candidate combination; Step B4: performing local field search and multi-universe interaction based on the updated parameters of the variable neighborhood search and multi-objective optimization combined algorithm and the updated universe population to obtain a re-updated candidate combination, re-updated parameters of the variable neighborhood search and multi-objective optimization combined algorithm, and a re-updated universe population; Step B5: selecting the current Pareto frontier from the re-updated candidate combination; Step B6: repeating steps B4 to B5 until the maximum number of iterations is met; Step B7: taking the Pareto frontier of the last iteration as the Pareto optimal solution set. The method comprises the following steps: An acquisition unit is configured to acquire distributed energy storage data of current charging and discharging of the park energy storage, and to acquire energy conversion forms of the current charging and discharging of the park energy storage.

7. A park energy storage charging and discharging strategy generation device, characterized in that, ​ ​ The selection unit is connected with the acquisition unit and is configured to select a mathematical model of the distributed energy storage and operation constraint conditions based on the distributed energy storage energy conversion form; The construction unit is connected with the acquisition unit and the selection unit respectively, and is configured to build a distributed energy storage charging and discharging optimization model covering peak regulation capacity and photovoltaic output prediction deviation according to the distributed energy storage data, the mathematical model and the operation constraint conditions; The optimization unit is connected with the construction unit and is configured to optimize the distributed energy storage charging and discharging optimization model based on a preset dynamic game model of leaders and followers to obtain a target optimization model; The solving unit is connected with the optimization unit and is configured to solve the target optimization model based on a variable neighborhood search and multi-objective optimization combination algorithm to obtain a Pareto optimal solution set; The generation unit is connected with the solving unit and is configured to generate a park energy storage charging and discharging optimal strategy based on a preset target preference and the Pareto optimal solution set.

8. The park energy storage charging and discharging strategy generation device according to claim 7, wherein the generation unit comprises: a standardization processing module configured to perform uniform standardization processing on specific values of various target indicators in the target preference and the Pareto optimal solution set; a first calculation module connected with the standardization processing module and configured to calculate a covariance matrix among all target indicators by using the standardized target indicators; a decomposition module connected with the first calculation module and configured to perform eigenvalue decomposition on the obtained covariance matrix to obtain an eigenvalue matrix; a second calculation module connected with the decomposition module and configured to calculate a characteristic information contribution rate in the eigenvalue matrix; a selection module connected with the second calculation module and configured to select principal components with a larger characteristic information contribution rate according to the characteristic information contribution rate; a third calculation module connected with the selection module and configured to calculate a comprehensive score of each eigenvalue in the eigenvalue matrix in the principal component space based on the selected principal components; a comparison module connected with the third calculation module and configured to obtain the park energy storage charging and discharging optimal strategy by comparing the comprehensive scores.

9. The park energy storage charging and discharging strategy generation device according to claim 7 or 8, wherein the solving unit comprises: an initialization module configured to set initial parameters and a maximum number of iterations of a variable neighborhood search and multi-objective optimization combination algorithm, and to construct an initial universe population and an initial candidate combination; wherein the initial parameters of the variable neighborhood search and multi-objective optimization combination algorithm include a number of universes, a set of field structures, a wormhole existence probability and a travel distance rate; a first processing module connected with the initialization module and configured to perform local field search and multi-universe interaction based on the initial parameters of the variable neighborhood search and multi-objective optimization combination algorithm and the initial universe population to obtain an updated candidate combination, updated parameters of the variable neighborhood search and multi-objective optimization combination algorithm and an updated universe population; ​ ​ The first screening module is connected with the first processing module and is configured to screen a current Pareto front from the updated candidate combination according to a processing result of the first processing module. The second processing module is connected with the first processing module and is configured to perform local field search and multi-universe interaction based on the updated variable neighborhood search and multi-objective optimization combined algorithm parameters and the updated universe population, to obtain again updated candidate combination, again updated variable neighborhood search and multi-objective optimization combined algorithm parameters and again updated universe population, until a maximum iteration number is met. The second screening module is connected with the second processing module and is configured to continuously screen a current Pareto front from the again updated candidate combination according to a processing result of the second processing module, until the second processing module has no processing result. The summary module is connected with the first screening module and the second screening module respectively, and is configured to take a Pareto front of a last iteration as a Pareto optimal solution set.

10. An electronic device, comprising: The device comprises a processor and a memory, and the memory stores a computer program. When the processor runs the computer program stored in the memory, the processor executes the generation method of the park energy storage charging and discharging strategy according to any one of claims 1 to 6.

Images