Virtual power plant optical storage cluster collaborative optimization method, device, equipment and medium

Abstract

The invention belongs to the field of electric power planning, and particularly relates to a cluster collaborative optimization method and device for virtual power plant optical storage, equipment and a medium. The method comprises the steps that the adjustable capacity of a virtual power plant distributed optical storage cluster is analyzed, an adjustable capacity prediction model is established, target functions and constraint conditions are set, and the target functions specifically comprise an operation economy optimization target, a user satisfaction optimization target and a photovoltaic consumption optimization target; according to the method, a distributed optical storage cluster multi-objective optimization scheduling model is established, detailed simulation analysis is performed through actual data, power output and operation cost comparison under different objective functions is performed on three different scenes, and the influence of the user response degree on the system operation cost is also discussed at the same time. Through example analysis of actual data, the system operation cost, the user satisfaction and the photovoltaic consumption quantitative index can reach the optimal or close to the optimal solution in different scenes.

Description

Technical Field

[0001] The present invention belongs to the field of power planning, and in particular relates to a method, device, equipment, and medium for collaborative optimization of a virtual power plant photovoltaic-storage-load cluster based on model-data dual drive. Background Art

[0002] A virtual power plant is an intelligent power grid technology that participates in the operation and dispatching of the power grid through a distributed power management system. It aggregates one or more controllable resources such as adjustable loads, energy storage, microgrids, electric vehicles, and distributed power sources in different spaces to achieve autonomous coordinated optimization control and participate in the operation of the power system and power market transactions. The development of virtual power plant technology provides technical support and solutions for the optimal scheduling of distributed photovoltaic-storage-load clusters.

[0003] In the context of the large-scale access of distributed energy, how to achieve the optimal scheduling of these resources, improve energy utilization efficiency, and reduce energy costs has become an urgent problem to be solved. Optimization scheduling technologies, including coordinated control technologies, optimization algorithms, etc., are the key to achieving this goal. Through real-time or near-real-time data collection, monitoring, and control, as well as response to market price signals or system operating states, a virtual power plant can perform optimal scheduling on a distributed photovoltaic-storage-load cluster to maximize resource utilization efficiency.

[0004] Currently, there are problems such as large environmental impact on photovoltaic output and flexible load power, and difficult prediction of baseline power; it is difficult to quantify the adjustable capacity of distributed photovoltaic, energy storage, and flexible load clusters; it is difficult to accurately generate optimal control strategies in different scenarios such as photovoltaic power consumption, power supply guarantee, and flexible regulation interaction.

[0005] Regarding the problems of aggregated modeling and adjustable capacity evaluation of distributed optical storage clusters for virtual power plants, Xu Xiaochun et al. divided the clusters from two aspects of structure and function, and proposed an equivalent aggregation model of distributed generation clusters participating in active distribution networks. Chen Can et al. proposed a method combining data-driven and deep belief networks to evaluate the adjustable capacity of air-conditioning temperature control load clusters participating in distributed photovoltaic accommodation, and could predict the adjustable capacity of temperature control load clusters in real time. Regarding the optimal scheduling problem of distributed optical storage clusters for virtual power plants, it mainly focuses on aspects such as the operating cost, photovoltaic accommodation level, and user satisfaction of the clusters under different scenarios. To strengthen the photovoltaic accommodation capacity of the power grid, Wang Feng et al. proposed a day-ahead and intra-day optimal scheduling strategy for regional power grids including distributed energy storage, solved it using an improved particle swarm optimization algorithm, and achieved the optimization of operating costs and correction costs. Cai Qinqin et al. established a bi-objective function model aiming to minimize building electricity bills and maximize local photovoltaic accommodation, solved it using an improved discrete binary particle swarm optimization algorithm, and verified the effectiveness and economy of the model. Xu Tongqing et al. proposed a photovoltaic accommodation evaluation model considering flexible load response from the perspective of improving photovoltaic accommodation and reducing power grid operating costs, and solved the joint optimal scheduling problem of optical storage charging and distribution power grids through an improved particle swarm optimization algorithm. In terms of demand response, Lim et al. studied the multi-level optimal scheduling problem of distributed optical storage considering user satisfaction and demand response, and achieved the optimal allocation of ES / PV for different users using a genetic algorithm. Lu Qing et al. explored the optimal scheduling problem of household appliances for smart homes participating in demand response, and studied the influence of user satisfaction and different optical storage system capacity configurations on the scheduling results. In terms of peak shaving optimal scheduling, Zhu Hao et al. proposed an optimal scheduling strategy for energy storage power stations participating in peak shaving considering photovoltaic accommodation. Wang Ting et al. proposed a coordinated peak shaving optimal scheduling model for optical storage systems to minimize peak shaving operating costs and used the response characteristics of adjustable loads to relieve peak shaving pressure. Yang Qian et al. also aimed at minimizing the system peak shaving cost, proposed an optimal peak shaving and matching strategy for the system considering different photovoltaic penetration rates. The results showed that as the photovoltaic penetration rate increased, the peak shaving demand of the system increased. By setting an appropriate system peak shaving margin, the photovoltaic penetration rate could be matched with the peak shaving capacity of the power system, thereby reducing the system operating cost.

[0006] The existing optimal scheduling of distributed optical storage clusters mainly focuses on a single objective function, and there is little research on the scheduling operation problems under different objective functions. In addition, existing research often focuses on the operating environment of a single scenario, and there is less research on the optimal scheduling strategies for multiple different scenarios. The accuracy of existing prediction models is not high, which in turn leads to inaccurate and unreliable scheduling decisions.

[0007] Therefore, the present invention studies the system scheduling operation problem by comparing the power output and operating cost under different objective functions in three different scenarios, and explores the impact of user response degree on the economy of system scheduling operation. Summary of the Invention

[0008] To achieve the efficient operation and optimized management of the virtual power plant distributed optical storage and utilization cluster, the present invention proposes a collaborative optimization method, device, equipment, and medium for the virtual power plant optical storage and utilization cluster. Based on model-data dual drive, the present invention establishes a more accurate prediction model to improve the accuracy and reliability of scheduling decisions.

[0009] The object of the present invention can be achieved by the following technical solutions:

[0010] A collaborative optimization method for a virtual power plant optical storage and utilization cluster, characterized by including the following:

[0011] Select adjustment characteristic indicators, establish an adjustable capacity prediction model containing reference power, power boundary, and energy boundary, and use a random forest algorithm optimized by the sparrow search optimization algorithm (SSA) to solve the adjustable function prediction model;

[0012] Set the objective function and constraint conditions, and establish a multi-objective optimization scheduling model for the distributed optical storage and utilization cluster based on the set objective function and constraint conditions;

[0013] Case analysis, detailed simulation analysis is carried out through actual data, and the power output and operating cost under different objective functions are compared for different scenarios.

[0014] The collaborative optimization method for the virtual power plant optical storage and utilization cluster selects adjustment characteristic indicators and establishes an adjustable capacity prediction model. Among them, the adjustment characteristic indicators include one of the seven types: maximum adjustable capacity, response time, adjustment rate adjustment gradient, adjustment accuracy, adjustment load rate, and adjustment average passivity rate. A physical model of reference power, power boundary, and energy boundary is established, and a random forest algorithm optimized by the sparrow search optimization algorithm (SSA) is used to more accurately predict the reference power, power boundary, and energy boundary of the cluster, providing a theoretical support for the optimized control strategy of the distributed optical storage and utilization cluster to actively support the power grid.

[0015] The described virtual power plant cluster collaborative optimization method for photovoltaic and energy storage systems, the objective function specifically includes one or more of the following: the optimization objective of operating economy, the optimization objective of user satisfaction, and the optimization objective of photovoltaic utilization rate; establish a system operating economy calculation model, a user satisfaction calculation model, and a system photovoltaic utilization rate calculation model for the three objective functions; considering the operating economy and stability of the regional power grid, set the importance ranking of the optimization indicators as operating economy > photovoltaic utilization rate > user satisfaction, normalize the objective function, normalization can eliminate the direct impact brought by different objective dimensions, and then use the weighted method for multi-objective optimization to make the weight distribution more reasonable and effective.

[0016] The described constraint conditions specifically include one or more of the following: power balance constraint, photovoltaic unit constraint, energy storage constraint, intelligent air conditioner constraint, intelligent water heater constraint, and tie-line power constraint.

[0017] The described virtual power plant cluster collaborative optimization method for photovoltaic and energy storage systems, establish a system operating economy calculation model;

[0018] The operating cost of the distributed photovoltaic and energy storage cluster optimization scheduling is mainly composed of the photovoltaic power generation cost, the cost of buying and selling electricity interacting with the external power grid, the energy storage cost, and the operating costs of intelligent air conditioner equipment and intelligent water heater equipment. The objective function is as follows:

[0019]

[0020] Where:

[0021] T: Time period; The active power of the i-th photovoltaic device at time t; c PV : Photovoltaic device unit power generation cost coefficient; The active power of the k-th energy storage device at time t; c cs : Energy storage device unit power scheduling cost coefficient; The active power of the f-th intelligent air conditioner device at time t; c ac : Intelligent air conditioner device unit power scheduling cost coefficient; The active power of the m-th intelligent water heater device at time t; c h : Intelligent water heater device unit power scheduling cost coefficient; The active power input and output of the first external network tie-line; c x+- Unit power purchase and sale price.

[0022] The described virtual power plant cluster collaborative optimization method for photovoltaic and energy storage systems, establish a user satisfaction calculation model;

[0023] Within the unit scheduling period, the user satisfaction is closely related to the change value of power consumption before and after the demand response service, and its objective function is as follows:

[0024]

[0025] In the formula:

[0026] T: Time period; Q t : Power consumption at time t before the demand response service; Q' t : Power consumption at time t after the demand response service; The sum of the change values of power consumption in each time period when participating in the demand response service; The sum of power consumption in each time period before participating in the demand response service.

[0027] For the virtual power plant photovoltaic-storage-load cluster collaborative optimization method described above, a system photovoltaic utilization rate calculation model is established;

[0028] Taking the minimum net load of the system as the calculation target of the system photovoltaic utilization rate, the objective function is:

[0029]

[0030] In the formula:

[0031] T: Time period; The load power after scheduling at time t; P pV (t): The power of the photovoltaic cluster at time t; λ: The percentage of photovoltaic accommodation.

[0032] For the virtual power plant photovoltaic-storage-load cluster collaborative optimization method described above, the power balance constraint is:

[0033] Ensuring that the power supply meets the power supply demand is the most basic power generation principle. The power balance constraint of the distributed photovoltaic-storage-load cluster can be written as:

[0034]

[0035] In the formula:

[0036] The active power of the kth energy storage device at time t; The active power input or output by the first external network connection line; P g,j : The active power of the gth conventional load device at time t;

[0037] The photovoltaic unit constraint is:

[0038] The power characteristics of photovoltaic power generation are susceptible to the influence of light intensity and weather factors, and have strong randomness. The actual power of photovoltaic power generation usually cannot reach the rated power. Therefore, the actual power of photovoltaic power generation is set to be less than the rated power as a constraint condition:

[0039] 0 ≤ Pi i PV,s ≤ P i PV，ratete (5),

[0040] In the formula:

[0041] P i PV，ratete : The rated power (kW) of the i-th photovoltaic device;

[0042] The energy storage constraint is:

[0043] The energy storage battery can achieve the purpose of suppressing load fluctuations through charge and discharge. The output power of the energy storage battery must be within the rated power to ensure stable operation. Its output constraint can be expressed as:

[0044] ,

[0045] In the formula:

[0046] The energy storage power of the i-th energy storage battery device; The rated power of the i-th energy storage battery device; The reserved upward power; The reserved downward power; The maximum charge and discharge power (kW); The state of charge; The minimum and maximum state of charge: η esch : The charge and discharge efficiency; The battery capacity (kWh);

[0047] The intelligent air conditioner constraint is:

[0048] The intelligent air conditioner load needs to meet the constraint that the active power is less than or equal to the rated power:

[0049]

[0050] In the formula:

[0051] The rated power of the i-th intelligent air conditioner device;

[0052] Secondly, avoid the intelligent air conditioner load exceeding the maximum power and the maximum allowable duration of the allowable cut load during curtailment, otherwise it will cause unnecessary losses;

[0053] ΔP i acc，t ≤ΔP i ac，max (12)

[0054] t ac ≤t ac,max (13),

[0055] Wherein:

[0056] ΔP i ac,t : The active power cut off by the i-th intelligent air conditioner device during the t period; ΔP i ac，ma : The maximum allowable power that can be cut off by the i-th intelligent air conditioner device; t ac : The sustainable time for the intelligent air conditioner device to reduce; t ac,max : The maximum allowable sustainable time for the intelligent air conditioner device to reduce;

[0057] The constraints of the described intelligent water heater are:

[0058] The intelligent water heater device needs to meet the constraint that the active power is less than or equal to the rated power:

[0059] 0≤P i h，t ≤P i h, rate (14),

[0060] Wherein:

[0061] P i h，rate : The rated power of the i-th intelligent water heater device;

[0062] Secondly, avoid the load of the intelligent water heater device exceeding the maximum power and the maximum allowable duration of the allowable cut-off load during reduction, otherwise unnecessary losses will be caused:

[0063]

[0064] t h ≤t h,max (16),

[0065] Wherein:

[0066] ΔP i h，j : The active power cut off by the i-th intelligent water heater device during the t period; The maximum allowable power that can be cut off by the i-th intelligent water heater device; t h : The sustainable time for the intelligent water heater device to reduce; t h,max: The maximum allowable duration reduced by the intelligent water heater device;

[0067] The power constraint of the tie line is as follows:

[0068] The exchanged power of the tie line within the distributed optical storage cluster should meet the rated power limit. To avoid safety accidents, the exchanged power of the tie line should meet the following constraints:

[0069]

[0070] In the formula:

[0071] The maximum value of the exchanged power of the first external network tie line.

[0072] The described collaborative optimization method for the virtual power plant optical storage cluster uses the existing improved bat algorithm as the multi-objective optimization solution algorithm.

[0073] Case study. Through detailed simulation analysis with actual data, the power output and operating costs under different objective functions are compared for three different scenarios. The specific scenarios are as follows: Scenario 1: Under normal conditions, flexible resources in the system participate in demand response. Scenario 2: When the load output is excessive, the system participates in demand response. Scenario 3: When the photovoltaic output is excessive, the system participates in demand response.

[0074] Perform a case study on the multi-objective optimal scheduling model of the distributed optical storage cluster. Through simulation analysis with real data, time-of-use electricity prices, case study parameters, and data of a typical day are set. The power output and operating costs under different objective functions are compared for three different scenarios, and the optimal solution of the comprehensive benefit is obtained.

[0075] A collaborative optimization device for a virtual power plant optical storage cluster, including:

[0076] A construction module, used to construct an adjustable capacity prediction model containing benchmark power, power boundary, and energy boundary, and to construct a multi-objective optimal scheduling model for the distributed optical storage cluster;

[0077] An analysis module, which conducts a case study. Through detailed simulation analysis with the actual data of the virtual power plant, the power output and operating costs under different objective functions are compared for different scenarios, and the feasibility of the constructed adjustable capacity prediction model and the multi-objective optimal scheduling model of the distributed optical storage cluster is verified.

[0078] A computer device, including a memory, a processor, and a computer program stored on the memory. The processor, when executing the computer program, implements the collaborative optimization method for the virtual power plant optical storage cluster of the present invention.

[0079] A computer-readable storage medium, characterized in that a computer program is stored on the computer-readable storage medium, and the computer program is used to cause a computer to execute the virtual power plant optical storage cluster collaborative optimization method described in the present invention.

[0080] The present invention relates to a model-data dual-driven virtual power plant optical storage cluster collaborative optimization method, device, equipment, and medium, analyzes the adjustable capacity of the distributed optical storage cluster of the virtual power plant, establishes an adjustable capacity prediction model, sets the objective function and constraint conditions, establishes a multi-objective optimization scheduling model for the distributed optical storage cluster, and conducts detailed simulation analysis through actual data, compares the power output and operating costs under different objective functions for three different scenarios, and also explores the impact of user response degree on the system operating cost.

[0081] Compared with the prior art, the present invention has the following advantages:

[0082] The purpose of the present invention is to provide a model-data dual-driven virtual power plant optical storage cluster collaborative optimization method, device, equipment, and medium to achieve the efficient operation and optimized management of the distributed optical storage cluster of the virtual power plant, which is of great significance for the scheduling operation problem of the distributed optical storage cluster actively supporting the power grid.

[0083] The features and advantages of the present invention are as follows:

[0084] 1. The present invention proposes an evaluation method for the adjustable capacity of the distributed optical storage cluster of the virtual power plant. By selecting appropriate adjustment characteristic indicators and establishing a prediction model, and using the random forest algorithm optimized by SSA for solution, it effectively predicts the benchmark power of the cluster and its power and energy boundaries. The simulation results show that the SSA-optimized random forest algorithm has a significant improvement in prediction accuracy compared with the traditional random forest algorithm, and successfully predicts the benchmark power, power boundary, and energy boundary of the cluster, verifying the effectiveness of the method.

[0085] 2. The present invention constructs a multi-objective optimization scheduling model with the goals of operation economy, user satisfaction, and optimal photovoltaic accommodation, and uses the improved bat algorithm to solve it. Through the case analysis of actual data, under different scenarios, the system operating cost, user satisfaction, and photovoltaic accommodation quantification indicators can reach the optimal or near-optimal solutions. Description of the Drawings

[0086] Figure 1 It is a flowchart for model construction.

[0087] Figure 2 It is a flowchart for the case.

[0088] Figure 3 It is the power prediction value for Scenario 1.

[0089] Figure 4 The power prediction value for Scenario 2.

[0090] Figure 5 The power prediction value for Scenario 3.

[0091] Figure 6 The comparison of power outputs under different objective functions for Scenario 1.

[0092] Figure 7 The comparison of power outputs under different objective functions for Scenario 2.

[0093] Figure 8 The comparison of power outputs under different objective functions for Scenario 3. Specific implementation manners

[0094] The present invention will be described in detail below in conjunction with the accompanying drawings and specific embodiments.

[0095] In view of the insufficient local consumption capacity in the context of large-scale photovoltaic access, the present invention considers the multi-objective optimal scheduling problem of distributed photovoltaic-storage-load clusters, constructs a multi-objective optimal scheduling model for distributed photovoltaic-storage-load clusters, sets relevant constraints, linearly processes the multi-objective function and solves the model by using relevant algorithms, effectively predicts the benchmark power and its power and energy boundaries of the cluster, verifies the scheduling plan results through numerical examples, analyzes the system operation cost, user satisfaction and photovoltaic consumption quantification index under different scenarios to reach the optimal comprehensive benefit solution, and verifies the effectiveness of the method.

[0096] See Figure 1 A collaborative optimization method for a virtual power plant's photovoltaic-storage-load cluster, based on model-data double drive, the method includes the following contents:

[0097] Select regulation characteristic indexes, establish an adjustable capacity prediction model containing benchmark power, power boundary and energy boundary, and solve the adjustable capacity prediction model by using a random forest algorithm optimized by a sparrow search optimization algorithm (SSA);

[0098] Set the objective function and constraint conditions, and establish a multi-objective optimal scheduling model for distributed photovoltaic-storage-load clusters based on the set objective function and constraint conditions;

[0099] Numerical example analysis, through detailed simulation analysis with actual data, compare the power outputs and operation costs under different objective functions for different scenarios, and verify the feasibility of the established adjustable capacity prediction model and the multi-objective optimal scheduling model for distributed photovoltaic-storage-load clusters.

[0100] First, analyze the adjustable capacity of the virtual power plant's distributed photovoltaic and energy storage cluster, establish an adjustable capacity prediction model, use K-means clustering analysis, based on historical load data, obtain a set of typical load curves, construct a simulated energy storage system cluster, and evaluate the adjustable capacity of the distributed photovoltaic and energy storage cluster according to the specific operation mode; by selecting adjustment characteristic indicators and establishing an adjustable capacity prediction model, establish physical models of benchmark power, power boundary, and energy boundary, and use the random forest algorithm optimized by SSA to analyze the benchmark power, power boundary, and energy boundary of different types of resources, and obtain the prediction curves of photovoltaic cluster power, energy storage cluster power, and temperature control load cluster power.

[0101] Set the objective function and constraint conditions, and establish a multi-objective optimal scheduling model for the distributed photovoltaic and energy storage cluster;

[0102] The objective functions specifically include the optimization objective of operation economy, the optimization objective of user satisfaction, and the optimization objective of photovoltaic utilization rate; establish a system operation economy calculation model, a user satisfaction calculation model, and a system photovoltaic utilization rate calculation model for the three objective functions; set the importance ranking of the optimization indicators as the optimization objective of operation economy > the optimization objective of photovoltaic utilization rate > the optimization objective of user satisfaction. Normalize the objective function and then use the weighted method for multi-objective optimization.

[0103] The adjustment characteristic indicators include the maximum adjustable capacity, response time, adjustment rate, adjustment gradient, adjustment accuracy, adjustment load rate, and adjustment average passivity rate.

[0104] The constraint conditions specifically include power balance constraints, photovoltaic unit constraints, energy storage constraints, intelligent air conditioner constraints, intelligent water heater constraints, and tie-line power constraints;

[0105] Case study, conduct detailed simulation analysis through actual data, compare the power output and operation cost under different objective functions for different scenarios, and explore the impact of user response degree on the system operation cost.

[0106] Analyze and solve the model with the input power prediction curve and time-of-use electricity price. Analyze the prediction curves of photovoltaic cluster power, energy storage cluster power, and temperature control load cluster power to provide a basis for subsequent optimal scheduling. At the same time, the model considers the purchase price and selling price at different times for model analysis.

[0107] Solve the constructed model based on the existing improved bat algorithm to obtain the objective function value, algorithm optimization result, and system optimization result.

[0108] The specific calculation models are as follows:

[0109] (1) System operation economy calculation model

[0110] The operating cost of the distributed optical storage cluster optimization scheduling mainly consists of the photovoltaic power generation cost, the cost of buying and selling electricity interacting with the external power grid, the energy storage cost, and the operating costs of intelligent air conditioning equipment and intelligent water heater equipment. The objective function is as follows:

[0111]

[0112] In the formula:

[0113] T: Time period; The active power (kW) of the i-th photovoltaic device at time t; c PV : Photovoltaic device unit power generation cost coefficient (yuan / kW); The active power (kW) of the k-th energy storage device at time t; c cs : Energy storage device unit power scheduling cost coefficient (yuan / kW); The active power (kW) of the f-th intelligent air conditioning device at time t; c ac : Intelligent air conditioning device unit power scheduling cost coefficient (yuan / kW); The active power (kW) of the m-th intelligent water heater device at time t; c h : Intelligent water heater device unit power scheduling cost coefficient (yuan / kW): The active power (kW) input and output by the first external network connection line; c x+- Unit power purchase and sale price (yuan / kW).

[0114] (2) Establish a user satisfaction calculation model

[0115] During a unit scheduling period, user satisfaction is closely related to the change in electricity consumption before and after the demand response service. The objective function is as follows:

[0116]

[0117] In the formula:

[0118] T: Time period; Q t : Electricity consumption (kWh) at time t before the demand response service: Q' t : Electricity consumption (kWh) at time t after the demand response service; The sum of the changes in electricity consumption in each period when participating in the demand response service (kWh); The sum of electricity consumption in each period before participating in the demand response service (kWh).

[0119] (3) System photovoltaic utilization rate calculation model

[0120] In order to make full use of photovoltaic power generation, during the system optimization and dispatching process, it should be considered to consume as much photovoltaic power as possible with the new load curve. Considering the cost of long-distance power transmission, local consumption should be given priority. Based on this, the minimum net load of the system is taken as the calculation target for the system's photovoltaic utilization rate, and the objective function is:

[0121]

[0122] In the formula:

[0123] T: Time period; The load power (kW) after dispatching at time t; P pV (t): The power (kW) of the photovoltaic cluster at time t; λ: The percentage of photovoltaic power consumption.

[0124] The specific constraint conditions include power balance constraint, photovoltaic unit constraint, energy storage constraint, intelligent air conditioner constraint, intelligent water heater constraint, and tie-line power constraint.

[0125] (1) Power balance constraint

[0126] Ensuring that the power supply meets the power supply demand is the most basic power generation principle. The power balance constraint of the distributed photovoltaic-storage-usage cluster can be written as:

[0127]

[0128] In the formula:

[0129] The active power (kW) of the kth energy storage device at time t; The active power (kW) input or output by the first external network tie-line; P g,j : The active power (kW) of the gth conventional load device at time t.

[0130] (2) Photovoltaic unit constraint

[0131] The power generation power characteristics of photovoltaic power generation are easily affected by light intensity and weather factors, and have strong randomness. The actual power of photovoltaic power generation usually cannot reach the rated power. Therefore, setting the actual power of photovoltaic power generation to be less than the rated power as a constraint condition:

[0132] 0 ≤ P i PV，s ≤ P i PV，ratete (5),

[0133] In the formula:

[0134] P i PV，ratete : The rated power (kW) of the ith photovoltaic device.

[0135] (3) Energy storage constraint

[0136] Energy storage batteries can achieve the purpose of suppressing load fluctuations through charge and discharge. The output power of energy storage batteries must be within the rated power to ensure stable operation. Its output constraint can be expressed as:

[0137]

[0138] ,

[0139] In the formula:

[0140] The energy storage power (kW) of the i-th energy storage battery device; The rated power (kW) of the i-th energy storage battery device; The reserved upward regulation power (kW); The reserved downward regulation power (kW); The maximum charge and discharge power (kW); State of charge; Minimum and maximum state of charge: η esch : Charge and discharge efficiency; Battery capacity (kWh).

[0141] (4) Smart air conditioner constraint

[0142] As a flexible regulation object, when the power grid is short of power or fails, the pressure on the regional power grid can be relieved by reducing or cutting off the smart air conditioner load. First, the smart air conditioner load needs to meet the constraint that the active power is less than or equal to the rated power:

[0143]

[0144] In the formula:

[0145] The rated power (kW) of the i-th smart air conditioner device.

[0146] Secondly, avoid the smart air conditioner load exceeding the maximum power and maximum allowable duration of the allowable cut-off load during reduction, otherwise unnecessary losses will be caused.

[0147] ΔP i ac，t ≤ΔP i ac，max (12)

[0148] t ac ≤t ac,max (13),

[0149] In the formula:

[0150] ΔP i ac，t : The active power (kW) cut off by the i-th intelligent air conditioner device during the t period;

[0151] ΔP i ac，ma : The maximum allowable cut-off power (kW) of the i-th intelligent air conditioner device; t ac : The sustainable time (min) of the intelligent air conditioner device reduction; t ac,max : The maximum allowable sustainable time (min) of the intelligent air conditioner device reduction.

[0152] (5) Intelligent water heater constraint

[0153] Similarly, first, the intelligent water heater device needs to satisfy the constraint that the active power is less than or equal to the rated power:

[0154] 0 ≤ P i h，t ≤ P i h，rate (14),

[0155] In the formula:

[0156] P i h，rate : The rated power (kW) of the i-th intelligent water heater device.

[0157] Secondly, avoid the load of the intelligent water heater device exceeding the maximum power and the maximum allowable sustainable time of the allowable cut-off load during reduction, otherwise unnecessary losses will be caused.

[0158]

[0159] t h ≤ t h,max (16),

[0160] In the formula:

[0161] ΔP i h，j : The active power (kW) cut off by the i-th intelligent water heater device during the t period; The maximum allowable cut-off power (kW) of the i-th intelligent water heater device; t h : The sustainable time (min) of the intelligent water heater device reduction; t h,max : The maximum allowable sustainable time (min) of the intelligent water heater device reduction.

[0162] (6) Tie-line power constraint

[0163] The exchange power of the interconnection lines within the cluster for distributed optical storage should meet the rated power limit. To avoid safety accidents, the exchange power of the interconnection lines should meet the following constraints:

[0164]

[0165] In the formula:

[0166] The maximum value (kW) of the exchange power of the first external network interconnection line.

[0167] The existing improved bat algorithm is used as the multi-objective optimization solution algorithm.

[0168] Perform a case study on the multi-objective optimal scheduling of the distributed optical storage cluster for the virtual power plant. Through the simulation analysis of the real data of a rural area in Zhejiang, the time-of-use electricity price, case parameters, and data of a typical day in summer are set. The power output and operating costs under different objective functions are compared for three different scenarios, and the optimal solution of the comprehensive benefit is obtained. Moreover, the influence of the user response degree on the system operating cost is also explored.

[0169] A virtual power plant optical storage cluster collaborative optimization device, including:

[0170] A construction module for constructing an adjustable capacity prediction model containing the reference power, power boundary, and energy boundary, and constructing a multi-objective optimal scheduling model for the distributed optical storage cluster;

[0171] An analysis module for performing a case study, conducting a detailed simulation analysis through the actual data of the virtual power plant, comparing the power output and operating costs under different objective functions for different scenarios, and verifying the feasibility of the constructed adjustable capacity prediction model and the multi-objective optimal scheduling model for the distributed optical storage cluster.

[0172] A computer device, including a memory, a processor, and a computer program stored on the memory. When the processor executes the computer program, the virtual power plant optical storage cluster collaborative optimization method is implemented.

[0173] A computer-readable storage medium, on which a computer program is stored. The computer program is used to make a computer execute the virtual power plant optical storage cluster collaborative optimization method.

[0174] Example 1

[0175] A virtual power plant optical storage cluster collaborative optimization method proposed by the present invention, based on model-data dual-driving, the logic is as Figure 2 shown. Taking a certain area in Zhejiang as an example, a case study is carried out in combination with the above steps.

[0176] (1) Based on the photovoltaic energy storage and user electricity consumption data in a certain area of Zhejiang, a simulation example is made. A typical day in summer in this area is selected as the optimization background. The time-of-use electricity price in Zhejiang Province includes four types: peak, valley, flat, and shoulder, and the electricity price per degree is 1.254 yuan, 0.931 yuan, 0.716 yuan, and 0.380 yuan respectively. The selling electricity price is selected as 50% of the purchase electricity price, as shown in Table 1. The various parameters of the example are set as follows: The system includes photovoltaic, energy storage, smart air conditioners, and electric water heaters. The internal resources and parameters in this area are set as shown in Table 2.

[0177] Table 1 System purchase and sale electricity prices

[0178]

[0179]

[0180] Table 2 Internal resource parameters

[0181]

[0182] (2) Conduct a typical day analysis. In order to meet the grid regulation requirements and facilitate the demonstration of the active support of the virtual power plant's distributed photovoltaic energy storage cluster for grid regulation under different scenarios, three scenarios are set in the present invention:

[0183] Scenario 1: Under normal conditions, flexible resources in the system participate in demand response.

[0184] Scenario 2: When the load output is excessive, the system participates in demand response.

[0185] Scenario 3: When the photovoltaic output is excessive, the system participates in demand response.

[0186] Since the photovoltaic output, residential load, air conditioner load, and water heater load are greatly affected by environmental factors, they are generally predicted through historical data or meteorological prediction data. The present invention uses the SSA-optimized random forest algorithm to predict the output power of photovoltaic power generation, residential load, air conditioner load, and water heater load. The predicted values of the typical day load in the three scenarios are as Figure 3 、 Figure 4 and Figure 5 shown.

[0187] (3) The objective function of the scheduling model proposed by the present invention is a multi-objective equilibrium solution, including the operating economy, user satisfaction, and photovoltaic utilization rate of the distributed photovoltaic energy storage cluster.

[0188] Figure 6It can be seen that in Scenario 1: Charging is carried out when the electricity price is low and discharging is carried out when the electricity price is high, effectively reducing the system operation cost. At the same time, since the photovoltaic power generation is not enough to cover the system demand, the system needs to purchase electricity from the main grid to meet the load. Combining with Table 3, it can be known that when the economic optimization is selected as the objective function, the system operation cost is 695,042.48 yuan, and the system reduces the air-conditioning cluster load and water heater cluster load during the peak electricity price period to the greatest extent. The regulation amount of the air-conditioning cluster and the water heater cluster is the largest, and at this time, the user satisfaction and the photovoltaic power consumption are the lowest. When the highest user satisfaction is selected as the objective function, the system operation cost is 741,566.68 yuan, and the regulation amount of the system air-conditioning cluster and the water heater cluster is the smallest, and the user comfort is the highest. At this time, the system operation cost is the highest. When the optimal photovoltaic utilization rate is selected as the objective function, the system operation cost is 737,695.78 yuan, and the system preferentially uses the output power of the photovoltaic units. When the three-objective equilibrium solution is selected as the objective function, the system operation cost is 709,984.68 yuan. After comprehensively considering the operation cost, the satisfaction quantification index and the photovoltaic power consumption quantification index, the photovoltaic utilization rate, the air-conditioning cluster regulation amount and the water heater cluster regulation amount basically reach the optimal values, and the comprehensive benefit is the highest.

[0189] Table 3 Comparison of operation costs under different objective functions in Scenario 1

[0190] Objective function Operating cost (yuan) Satisfaction quantification Photovoltaic accommodation quantification Economically optimal 695042.48 0.09 218183.21 Highest user satisfaction 741566.68 0.00 221091.94 Optimal photovoltaic utilization rate 737695.78 0.06 143296.70 Equilibrium solution of three objectives 709984.68 0.04 192865.21

[0191] Figure 7It can be seen that in Scenario 2: The low - valley electricity price period from 1:00 to 2:00 and the peak - electricity - price period from 12:00 to 13:00 are two time periods with sudden increases in load. At this time, the energy storage, as an auxiliary regulation function, releases the electric energy stored in itself, alleviating the operation pressure of the power grid. The air - conditioner cluster and the water - heater cluster also reduce their own loads to the maximum extent. Since the photovoltaic power generation is mainly concentrated in the period from 8:00 to 19:00 and the power far cannot meet the load demand of the system, the system can only purchase electricity from the main grid to meet the load demand. When the economic - optimization is selected as the objective function, the system operation cost is 362,065.90 yuan. The system reduces the loads of the air - conditioner cluster and the water - heater cluster in the peak - electricity - price period to the maximum extent. The regulation amount of the air - conditioner cluster and the regulation amount of the water - heater cluster are the largest, and at this time, the user satisfaction and the photovoltaic accommodation index are the lowest. When the highest user satisfaction is selected as the objective function, the system operation cost is 401,817.97 yuan. The regulation amount of the system air - conditioner cluster and the regulation amount of the water - heater cluster are the smallest, the user comfort is the highest, and at this time, the system operation cost is the highest. When the optimal photovoltaic utilization rate is selected as the objective function, the system operation cost is 370,598.90 yuan. The system preferentially uses the output power of the photovoltaic units. When the three - target equilibrium solution is selected as the objective function, the system operation cost is 364,545.20 yuan. After comprehensively considering the operation cost, the satisfaction quantification index, and the photovoltaic accommodation quantification index, the photovoltaic utilization rate, the regulation amount of the air - conditioner cluster, and the regulation amount of the water - heater cluster basically reach the optimal values, and the comprehensive benefit is the highest.

[0192] Table 4 Comparison of operation costs under different objective functions in Scenario 2

[0193]

[0194] Figure 8It can be seen that in Scenario 3: The peak power generation period is mainly concentrated in the period from 11:00 to 17:00. Among them, the period from 11:00 to 13:00 is the low electricity price period, and the period from 13:00 to 17:00 is the high electricity price period. At this time, the energy storage plays an auxiliary regulation role, and the purpose of consuming photovoltaic power is achieved by storing electric energy. The air-conditioning cluster and the water heater cluster also increase a certain load during this period to consume photovoltaic power. When the economic optimization is selected as the objective function, the system operation cost is 644,475.91 yuan. The system reduces the load of the air-conditioning cluster and the water heater cluster during the peak electricity price period to the greatest extent. The regulation amount of the air-conditioning cluster and the regulation amount of the water heater cluster are the largest. At this time, the user satisfaction and the photovoltaic consumption index are the lowest. When the highest user satisfaction is selected as the objective function, the system operation cost is 727,288.72 yuan. The regulation amount of the system air-conditioning cluster and the regulation amount of the water heater cluster are the smallest, and the user comfort is the highest. At this time, the system operation cost is the highest. When the optimal photovoltaic utilization rate is selected as the objective function, the system operation cost is 719,484.04 yuan. The system preferentially uses the output power of the photovoltaic unit. When the three-objective equilibrium solution is selected as the objective function, the system operation cost is 658,213.49 yuan. After comprehensively considering the operation cost, the satisfaction quantification index and the photovoltaic consumption quantification index, the photovoltaic utilization rate, the regulation amount of the air-conditioning cluster and the regulation amount of the water heater cluster basically reach the optimal values, and the comprehensive benefit is the highest.

[0195] Table 5 Comparison of operation costs under different objective functions in Scenario 3

[0196] Objective function Operating cost (yuan) Satisfaction quantification Photovoltaic accommodation quantification Economically optimal 644475.91 0.09 163275.38 Highest user satisfaction 727288.72 0.00 143169.10 Optimal photovoltaic utilization rate 719484.04 0.03 93991.52 Equilibrium solution of three objectives 658213.49 0.01 110047.51

[0197] A model-data dual-driven collaborative optimization method for virtual power plant photovoltaic energy storage and utilization clusters proposed by the present invention improves the volatility of optimized photovoltaic output and the overall regulation ability of the system, can better solve the dispatching operation problem of distributed photovoltaic energy storage and utilization clusters actively supporting the power grid, and provides strong support for the reliability, economy and environmental friendliness of the new power system.

Claims

1. A collaborative optimization method for a virtual power plant's photovoltaic and energy storage cluster, characterized in that It includes the following contents: Select adjustment characteristic indexes, establish an adjustable capacity prediction model containing reference power, power boundary and energy boundary, and use the random forest algorithm optimized by the sparrow search optimization algorithm (SSA) to solve the adjustable capacity prediction model; Set the objective function and constraint conditions, and establish a multi-objective optimal scheduling model for distributed photovoltaic-energy storage-load clusters based on the set objective function and constraint conditions; Case analysis: Through detailed simulation analysis with actual data, compare the power output and operating costs under different objective functions for different scenarios, and verify the feasibility of the established adjustable capacity prediction model and the multi-objective optimal scheduling model for distributed photovoltaic-energy storage-load clusters.

2. The collaborative optimization method for photovoltaic-energy storage-load clusters of a virtual power plant according to claim 1, characterized in that: The adjustment characteristic indexes include at least one of the following: maximum adjustable capacity, response time, adjustment rate, adjustment gradient, adjustment accuracy, adjustment load rate, adjustment average passive rate.

3. The collaborative optimization method for photovoltaic-energy storage-load clusters of a virtual power plant according to claim 1, characterized in that: The specific objective functions include one or more of the following: the optimal operation economy objective, the optimal user satisfaction objective, the optimal photovoltaic utilization rate objective; establish a system operation economy calculation model, a user satisfaction calculation model and a system photovoltaic utilization rate calculation model for the three objective functions; Set the importance ranking of the optimization indexes as operation economy > photovoltaic utilization rate > user satisfaction, normalize the objective functions, and then use the weighted method for multi-objective optimization; The specific constraint conditions include one or more of the following: power balance constraint, photovoltaic unit constraint, energy storage constraint, intelligent air conditioner constraint, intelligent water heater constraint and tie-line power constraint.

4. The collaborative optimization method for photovoltaic-energy storage-load clusters of a virtual power plant according to claim 3, characterized in that: Establish a system operation economy calculation model; The operating cost of the distributed photovoltaic-energy storage-load cluster optimization scheduling mainly consists of the photovoltaic power generation cost, the cost of buying and selling electricity interacting with the external power grid, the energy storage cost, and the operating costs of intelligent air conditioner equipment and intelligent water heater equipment. The objective function is as follows: Where: T: Time period; Active power of the i-th photovoltaic device in the t-th period; c PV : Generation cost coefficient per unit of electricity of the photovoltaic device; Active power of the k-th energy storage device in the t-th period; c cs : Scheduling cost coefficient per unit of electricity of the energy storage device; Active power of the f-th intelligent air conditioner device in the t-th period; c ac : Scheduling cost coefficient per unit of electricity of the intelligent air conditioner device; Active power of the m-th intelligent water heater device in the t-th period; c h : Scheduling cost coefficient per unit of electricity of the intelligent water heater device; Active power input and output of the first external network connection line; c x+- Purchase and sale electricity price per unit of electricity.

5. The collaborative optimization method for photovoltaic-energy storage-load clusters of a virtual power plant according to claim 3, characterized in that: Establish a user satisfaction calculation model; Within a unit scheduling period, the user satisfaction is closely related to the change value of the electricity consumption before and after the demand response service. The objective function is as follows: Where: T: Time period; Q t : Electricity consumption during time period t before the demand response service; Q' t : Electricity consumption during time period t after the demand response service; : Sum of the change values of electricity consumption in each time period when participating in the demand response service; : Sum of electricity consumption in each time period before participating in the demand response service.

6. The collaborative optimization method for photovoltaic-energy storage-load clusters of a virtual power plant according to claim 3, characterized in that: Establish a system photovoltaic utilization rate calculation model; Taking the minimum net load of the system as the calculation target of the system photovoltaic utilization rate, the objective function is: Where: T: Time period; The load power after scheduling in the t time period; P pV (t): The power of the photovoltaic cluster in the t time period; λ: The percentage ratio of photovoltaic accommodation.

7. The collaborative optimization method for photovoltaic-energy storage-load clusters of a virtual power plant according to claim 3, characterized in that: The power balance constraint is: Ensuring that the power supply meets the power supply demand is the most basic power generation principle. The power balance constraint of the distributed photovoltaic-energy storage-load cluster can be written as: Where: Active power of the kth energy storage device in the t time period; Active power input or output by the first external network connection line; P g,j : Active power of the gth conventional load device in the t time period; The photovoltaic unit constraint is: The power characteristics of photovoltaic power generation are vulnerable to the influence of light intensity and weather factors, and have strong randomness; the actual power of photovoltaic power generation usually cannot reach the rated power. Therefore, the actual power of photovoltaic power generation is set to be less than the rated power as a constraint condition: 0 ≤ P i PV，s ≤ P i PV，ratete (5), In the formula: P i PV，ratete : The rated power of the i-th photovoltaic device; The energy storage constraint is: The energy storage battery can achieve the purpose of suppressing load fluctuations through charging and discharging. The output power of the energy storage battery must be within the rated power to ensure stable operation. Its output constraint can be expressed as: ， In the formula: Energy storage power of the i-th energy storage battery device Rated power of the i-th energy storage battery device Reserved upward regulation power Reserved downward regulation power Maximum charge and discharge power State of charge Minimum and maximum state of charge: η esch : Charge and discharge efficiency Battery capacity The intelligent air conditioner constraint is: The intelligent air conditioner load needs to meet the constraint that the active power is less than or equal to the rated power: In the formula: Rated power of the i-th intelligent air conditioner device; Secondly, avoid the intelligent air conditioner load exceeding the maximum power and the maximum allowable duration of the load that can be cut during curtailment, otherwise unnecessary losses will be caused; ΔP i ac，t ≤ΔP i PV，ratete (12) t ac ≤t ac,max (13), In the formula: ΔP i ac，t : The active power cut off by the i-th intelligent air conditioner device during the t period; ΔP i ax，ma : The maximum allowable cut-off power of the i-th intelligent air conditioner device; t ac : Sustainable time reduced by the intelligent air conditioning device; t ac,max : Maximum allowable duration of reduction by the intelligent air conditioning device The intelligent water heater constraint is: The intelligent water heater equipment needs to meet the constraint that the active power is less than or equal to the rated power: 0 ≤ P i h，t ≤ P i h，rate (14), In the formula: P i h，rate : The rated power of the i-th intelligent water heater device; Secondly, avoid the intelligent water heater equipment load exceeding the maximum power and the maximum allowable duration of the load that can be cut during curtailment, otherwise unnecessary losses will be caused: t h ≤t h,max (16), In the formula: ΔP i h,j : The active power cut off by the i-th intelligent water heater device during the t period; The maximum allowable power cut off for the i-th intelligent water heater device; t h : The sustainable time for the intelligent water heater device to cut down; t h,max : The maximum allowable sustainable time for the intelligent water heater device to cut down; The tie-line power constraint is: The tie-line exchange power in the distributed optical storage and utilization cluster should meet the rated power limit. To avoid safety accidents, the tie-line exchange power should meet the following constraints: In the formula: The maximum value of the exchange power of the first external network connection line.

8. The virtual power plant optical storage cluster collaborative optimization method according to claim 1, characterized in that The existing improved bat algorithm is used as the multi-objective optimization solution algorithm.

9. The virtual power plant optical storage cluster collaborative optimization method according to claim 1, characterized in that Perform a case study on the multi-objective optimization scheduling model of the distributed optical storage and utilization cluster. Through simulation analysis with real data, time-of-use electricity prices, case parameters, and data of a certain typical day are set. The power output and operating costs under different objective functions are compared for three different scenarios. The specific scenarios are: Scenario 1: Under normal conditions, flexible resources in the system participate in demand response; Scenario 2: When the load output is excessive, the system participates in demand response; Scenario 3: When the photovoltaic output is excessive, the system participates in demand response; and the optimal solution of comprehensive benefits is obtained.

10. A cluster collaborative optimization device for virtual power plants with optical storage, characterized in that, Including: A construction module for constructing an adjustable capacity prediction model containing benchmark power, power boundaries, and energy boundaries, and constructing a multi-objective optimization scheduling model for a distributed optical storage and utilization cluster; An analysis module for performing case analysis, conducting detailed simulation analysis through the actual data of a virtual power plant, comparing the power output and operating costs under different objective functions for different scenarios, and verifying the feasibility of the constructed adjustable capacity prediction model and the multi-objective optimization scheduling model for a distributed optical storage and utilization cluster.

11. A computer device, comprising a memory, a processor, and a computer program stored on the memory, characterized in that When the processor executes the computer program, it implements the method according to any one of claims 1-9.

12. A computer-readable storage medium, characterized in that, A computer program is stored on the computer-readable storage medium, and the computer program is used to cause a computer to execute the method according to any one of claims 1 to 9.

Images