Economic and auxiliary service optimization method for multi-energy virtual power plant

Abstract

The invention relates to a multi-energy virtual power plant economic and auxiliary service optimization method, which belongs to the technical field of virtual power plant scheduling, and comprises the following steps: considering the uncertainty of renewable energy sources, and constructing a virtual power plant economic and auxiliary service joint optimization model; and solving the joint optimization model by adopting a hybrid optimization strategy fusing an improved particle swarm optimization algorithm and a snow melting optimizer to obtain an optimal scheduling scheme. According to the invention, collaborative balance between operation cost minimization and auxiliary service revenue maximization is realized; the optimization method has the advantages of few parameters, fast convergence, high precision and the like, and the solution performance of the model and the optimization quality of the scheduling scheme are remarkably enhanced.

Description

Technical Field

[0001] The present invention belongs to the technical field of virtual power plant scheduling, and in particular relates to a method for optimizing the economy and auxiliary services of a multi-energy virtual power plant. Background Art

[0002] In recent years, my country has vigorously promoted the green and low-carbon transformation of its energy sector, achieving new breakthroughs in renewable energy development and entering a new stage of large-scale, high-quality leapfrogging development. Domestic renewable energy project construction is maintaining a strong momentum and is expected to maintain a high level of commissioning in the future.

[0003] However, renewable energy is inherently intermittent, volatile, and random. For example, photovoltaic power generation is highly concentrated and constrained by distributed layouts and grid dispatch mechanisms. This has led to an increasingly prominent problem: "generated, delivered, and used." The concentrated output of renewable energy has reshaped the traditional load curve, creating "valleys" in the grid's net load during certain periods, creating an urgent need for load shifting and peak-to-valley shaving. These problems will be further exacerbated by the large-scale integration of renewable energy, leading to grid frequency fluctuations and voltage instability, seriously impacting power quality and system security.

[0004] To address these challenges, virtual power plants (VPPs) have emerged as a new aggregation and coordination mechanism. By integrating distributed power sources, controllable loads, and energy storage devices, VPPs enable unified monitoring and optimized dispatch, enhancing the controllability and market adaptability of renewable energy. Virtual power plants are becoming a crucial component in ensuring the safe operation of power grids and increasing the utilization of renewable energy.

[0005] At the same time, with the continuous advancement of new power system construction, my country's power ancillary services market is also rapidly developing. The system's demand for regulating resources such as frequency regulation, backup, and rapid ramping has increased significantly. With their resource aggregation and rapid response capabilities, virtual power plants (VPPs) are playing an increasingly important role in the ancillary services market. Their ability to coordinate renewable energy output with grid demand is becoming a key path to promoting the green, safe, and efficient development of the energy system. However, current VPP resource scheduling approaches fail to consider the power ancillary services market.

[0006] Furthermore, the collaborative scheduling model for virtual power plants is highly nonlinear, strongly coupled, and subject to multiple constraints. Traditional analytical methods are difficult to solve directly, and linearization often loses the underlying structure of the problem. Therefore, intelligent optimization algorithms are required to perform a global search across the multidimensional variable space to achieve the optimal scheduling strategy. However, during this global search, the algorithm is prone to becoming trapped in a local maximum solution. Summary of the Invention

[0007] The purpose of the present invention is to provide a multi-energy virtual power plant economic and ancillary service optimization method to solve the problems that the existing virtual power plant resource scheduling algorithm does not take the power ancillary service market into consideration and the algorithm is prone to fall into local maximum solutions.

[0008] In order to achieve the above object, the technical solution of the present invention is as follows: The present invention relates to a multi-energy virtual power plant economic and ancillary service optimization method, which comprises the following steps: S1. Considering the uncertainty of renewable energy, a joint optimization model of virtual power plant economics and ancillary services is constructed; S2. A hybrid optimization strategy combining the improved particle swarm optimization algorithm and the snowmelt optimizer is used to solve the joint optimization model and obtain the optimal scheduling solution.

[0009] Preferably, the S1 construction of a joint optimization model of virtual power plant economy and ancillary services includes: unified modeling of various resources within the virtual power plant under a typical intraday scheduling scenario, and construction of ancillary service participation mechanism and benefit evaluation model.

[0010] Preferably, the unified modeling of multiple resources in the virtual power plant in S1 specifically includes: establishing an objective function for minimizing the operating cost of the virtual power plant and establishing constraints for the virtual power plant.

[0011] Preferably, the expression of the objective function of minimizing the virtual power plant operating cost is: (1), (2), (3), in, represents minimizing the operating cost of the virtual power plant, t Indicates the t scheduling periods, T represents the set sum of scheduling periods, 、 、 、 、 、 The power operators, wind turbines, photovoltaic modules, fuel cell systems, micro gas turbines and energy storage devices during the dispatch period are t energy costs; 、 、 、 、 、 The power operators, wind turbines, photovoltaic modules, fuel cell systems, micro gas turbines and energy storage devices during the dispatch period are tPower injection; is the flexible load adjustment variable, indicating that t The amount of power adjustment achieved by delaying or reducing loads; 、 、 、 They are the operating status of wind turbine, photovoltaic module, fuel cell system, and micro gas turbine in the scheduling period t, respectively. "1" indicates operation and "0" indicates shutdown; and Respectively represent the operating status of wind turbines, photovoltaic modules, fuel cell systems, and micro gas turbines in the scheduling period t and the scheduling period t-1, with "1" indicating operation and "0" indicating shutdown. represents the operating cost of the energy storage equipment, and They represent the operating status of the energy storage device, and are also set to 0 / 1, which is used to calculate the start and stop cost of the energy storage unit; 、 Respectively i distributed generation units and j The start-up and shutdown costs of each energy storage device; 、 are the number of distributed generation units and energy storage devices, respectively; 、 Respectively b During the dispatch period t resistance and current; 、 、 They are the original loss, final loss and new loss after optimization of the system; 、 They are unit loss cost and power difference before and after optimization respectively; To sell electrical power to the grid, is the electricity selling price, represents the flexible load adjustment cost, It represents the total number of grid branches in the power system, that is, the number of transmission lines in the entire distribution network that participate in the line power loss calculation.

[0012] Preferably, the virtual power plant constraints include: The power balance constraint is expressed as: (4), in, 、 Energy storage equipment t The discharge and charging power of each scheduling period, For the tLine loss power during each scheduling period, No. t Load demand during each dispatch period; The output power constraint of the wind turbine generator set is expressed as: (5), in, 、 are the maximum and minimum power output of the wind turbine generator set respectively; The output power constraint of photovoltaic modules is expressed as: (6), in, 、 are the maximum and minimum power output of the photovoltaic modules respectively; The output limit of micro gas turbine is expressed as: (7), in, 、 are the maximum and minimum power output of the micro gas turbine respectively; The fuel cell system output limit is expressed as: (8), in, 、 are the maximum and minimum power output of the fuel cell system respectively; The power purchase restriction from power operators is expressed as: (9), in, 、 are the maximum and minimum power purchased from power operators respectively; Adjustable load output limit, its expression is: (10), in, 、 are the maximum and minimum power of the adjustable load of the energy storage device respectively; Flexible load regulation limit, its expression is: (11), in, It is the upper limit of flexible load adjustment; The charging and discharging limits of energy storage equipment are expressed as: (12), (13), (14), (15), (16), in, 、 For energy storage equipment t The charging and discharging power of each scheduling period, 、 Energy storage equipment t The maximum charge and discharge power of each scheduling period, 、 is a binary state variable, For the t The energy storage capacity of each scheduling period, is the energy storage capacity at the previous moment; Energy storage charging efficiency, Energy storage discharge efficiency; The power consumption constraint is expressed as follows: (17), (18), (19), in, Indicates the amount of electricity sold. is the remaining power, is the actual load, is a binary variable, M is a constant.

[0013] Preferably, the construction of the auxiliary service participation mechanism and benefit evaluation model in S1 specifically includes: Construct the service response power allocation constraint, which is expressed as: (20), in, For resources r In the t The power generation during each dispatch period, , The reserved frequency regulation and standby service response power are respectively, For resources r Maximum output capacity; Construct the supply and demand matching constraint, whose expression is: (twenty one), (twenty two), in, 、 Respectively represent frequency regulation and standby requirements; Construct the service revenue objective function, which is expressed as follows: (twenty three), in, 、 Respectively t The market price of frequency regulation and standby services during each dispatch period, is the total ancillary service revenue of the virtual power plant during the entire dispatch cycle.

[0014] Preferably, the specific steps of S2 using a hybrid optimization strategy combining an improved particle swarm optimization algorithm and a snowmelt optimizer to solve the joint optimization model are as follows: S2.1. Use the improved particle swarm optimization algorithm to perform a global search and obtain a scheduling solution; S2.2. Use the fitness function to evaluate the scheduling plan and determine whether it meets the iteration and convergence conditions. If all the iteration and convergence conditions are met, the scheduling plan is selected as the optimal scheduling plan and output. If any of the iteration and convergence conditions are not met, proceed to S2.3. S2.3. Combine the hybrid optimization strategy of the snowmelt optimizer to perform local fine perturbation and convergence control to optimize the scheduling plan.

[0015] Preferably, the specific method of obtaining the scheduling solution in S2.1 is: using the improved particle swarm optimization algorithm to perform a global search, updating the speed and position of the particles, and gradually obtaining the scheduling solution. The updating formula of the particle speed and position is: (25), (26), in, g represents the number of iterations, and represent the velocity and position of the particle respectively, For particles p The best historical position, is the current global optimal position, and is a random vector, represents the Hadamard product, is the linear decrease result of inertia weight, and is the learning factor.

[0016] Preferably, the fitness function in S2.2 is: (27), in, represents fitness, represents the operating cost of the virtual power plant, represents the weight coefficient, is the total revenue of ancillary services provided by the virtual power plant during the entire dispatch cycle; Determine whether the following convergence conditions are met: Condition 1. Reach the maximum number of iterations ; Condition 2. Continuous The change in the optimal fitness function value within a generation is less than the set threshold ,Right now: (28).

[0017] Preferably, the specific steps of performing local fine perturbation and convergence control by combining the hybrid optimization strategy of the snowmelt optimizer in S2.3 include: S2.3.1. Based on the solution to the scheduling solution obtained in S2.1, generate a random set in matrix form and the upper and lower bounds of the solution. The expression for generating the random set in matrix form is: (29), The calculation formula for the upper and lower limits is: (30), in, N and n denote the number of particles in the population and the dimension of the solution, respectively. m represents the dimension index in the matrix, m ∈ n , U 、 L are the upper and lower limits of the value, is a random number in [0,1]; S2.3.2. Update the particle position by introducing Brownian motion. The update formula of the particle position is: (31), in, For the p The position of the particle, g is the current iteration number, is a random individual in the elite particle collection, A random number vector generated by Gaussian distribution for Brownian motion, To multiply by row, are individuals randomly selected from several elite groups in the representative group. is the center of mass of the entire particle position; S2.3.3. Gradually converge through the snowmelt process to find the optimal solution. The position update formula at this stage is: (32), (33), Where: is a random number in [-1,1], P For the snowmelt model, is the maximum number of iterations, is a random number in [0,1], Indicates the The position or index of a randomly selected individual in a generation.

[0018] Compared with the prior art, the technical solution provided by the present invention has the following beneficial effects: 1. The present invention involves a multi-energy virtual power plant economic and ancillary service optimization method, which constructs a virtual power plant economic and ancillary service joint optimization model. The optimal scheduling scheme obtained after solving the model achieves a synergistic balance between minimizing operating costs and maximizing ancillary service benefits.

[0019] 2. This invention relates to a multi-energy virtual power plant economic and ancillary service optimization method that employs a hybrid optimization strategy that integrates an improved particle swarm optimization algorithm and a snowmelt optimizer to solve a joint optimization model and obtain the optimal scheduling solution. The improved particle swarm optimization algorithm is first used to globally explore the solution space, and then the snowmelt optimizer is used to perform localized fine-tuning of candidate solutions, balancing global search capability with local convergence accuracy. This method boasts advantages such as a low number of parameters, fast convergence, and high accuracy, significantly enhancing the model's solution performance and the optimization quality of the scheduling solution. BRIEF DESCRIPTION OF THE DRAWINGS

[0020] Figure 1 It is the overall flow chart of the multi-energy virtual power plant economic and ancillary service optimization method; Figure 2 It is a structural diagram of a virtual power plant model with adjustable load and energy storage participation; Figure 3 It is a flow chart of the hybrid optimization strategy algorithm that integrates the improved particle swarm optimization algorithm and the snowmelt optimizer; Figure 4 It is a 24-hour load and available energy output diagram; Figure 5 It is the dispatch output diagram of the comprehensive energy system after optimization; Figure 6 It is a system dispatch diagram without the participation of auxiliary services; Figure 7 It is the system scheduling diagram in the scenario without adjustable load. DETAILED DESCRIPTION

[0021] In order to further understand the content of the present invention, the present invention is described in detail with reference to the examples. The following examples are used to illustrate the present invention but are not used to limit the scope of the present invention.

[0022] Refer to the attached Figure 2 As shown, the virtual power plant model involved in this embodiment is based on the day-ahead market and uses the predicted power demand of local loads within 24 hours as a scheduling reference, such as Figure 4 As shown in the figure, local distributed generation (DG, including wind power, photovoltaic power generation, and micro-gas turbines) is prioritized to meet load demand. When DG output is insufficient or costs are high, the system operator (i.e., the energy management system) communicates with the utility grid to purchase electricity to fill the shortfall. Conversely, if DG output is sufficient, excess electricity is converted into heat for storage or stored in fuel cells (FCs) and sold to the grid at a predetermined profit margin. Furthermore, to enhance the flexibility and economy of system scheduling, flexible load resources (adjustable loads) are introduced. By adjusting some electricity consumption within the permitted timeframe, dynamic load response is achieved and auxiliary regulation is implemented to balance power, further optimizing local energy utilization and reducing electricity purchase costs.

[0023] Refer to the attached Figure 1 As shown, the present invention relates to a multi-energy virtual power plant economic and ancillary service optimization method, which includes the following steps: S1. Initialize the auxiliary service parameters of the virtual power plant (such as the power output of each device, the SOC of the energy storage device, the input load, the photovoltaic power generation, the wind power generation, the electricity price and other information), consider the uncertainty of renewable energy, and build a joint optimization model of the virtual power plant economy and auxiliary services. Specifically, it includes: using a multi-source collaborative modeling module to uniformly model the various resources in the virtual power plant under a typical daily scheduling scenario, that is, to uniformly model distributed resources, flexible loads, and auxiliary services, and to build an auxiliary service participation mechanism and benefit evaluation model based on market mechanisms and benefits.

[0024] The unified modeling of multiple resources within the virtual power plant specifically includes: establishing an objective function for minimizing the operating costs of the virtual power plant and establishing virtual power plant constraints.

[0025] The expression of the objective function for minimizing the operating cost of the virtual power plant is: (1), (2), (3), in, represents minimizing the operating cost of the virtual power plant, t Indicates the t scheduling periods, Trepresents the set sum of scheduling periods, 、 、 、 、 、 The power operators, wind turbines, photovoltaic modules, fuel cell systems, micro gas turbines and energy storage devices during the dispatch period are t energy costs; 、 、 、 、 、 The power operators, wind turbines, photovoltaic modules, fuel cell systems, micro gas turbines and energy storage devices during the dispatch period are t Power injection; is the flexible load adjustment variable, indicating that t The amount of power adjustment achieved by delaying or reducing loads; 、 、 、 They are the operating status of wind turbine, photovoltaic module, fuel cell system, and micro gas turbine in the scheduling period t, respectively. "1" indicates operation and "0" indicates shutdown; and Respectively represent the operating status of wind turbines, photovoltaic modules, fuel cell systems, and micro gas turbines in the scheduling period t and the scheduling period t-1, with "1" indicating operation and "0" indicating shutdown. represents the operating cost of the energy storage equipment, and They represent the operating status of the energy storage device, and are also set to 0 / 1, which is used to calculate the start and stop cost of the energy storage unit; 、 Respectively i distributed generation units and j The start-up and shutdown costs of each energy storage device; 、 are the number of distributed generation units and energy storage devices, respectively; 、 Respectively b During the dispatch period t resistance and current; 、 、 They are the original loss, final loss and new loss after optimization of the system; 、 They are unit loss cost and power difference before and after optimization respectively; To sell electrical power to the grid, is the electricity selling price, represents the flexible load adjustment cost, It represents the total number of grid branches in the power system, that is, the number of transmission lines in the entire distribution network that participate in the line power loss calculation.

[0026] Virtual power plant constraints include: Power balance constraints. In this phase, only the modeling power constraints of the virtual power plant itself are considered. To ensure a balanced supply and demand of system energy, the virtual power plant must maintain a balance between total generated power and total load demand within each dispatch cycle (typically 24 hours). This balance takes into account the output and consumption of resources such as distributed generation, grid power purchases, and energy storage systems. The power balance constraint is expressed as: (4), in, 、 Energy storage equipment t The discharge and charging power of each scheduling period, For the t Line loss power during each scheduling period, No. t Load demand during each dispatch period; The output power constraint of the wind turbine generator set is expressed as: (5), in, 、 are the maximum and minimum power output of the wind turbine generator set respectively; The output power constraint of photovoltaic modules is expressed as: (6), in, 、 are the maximum and minimum power output of the photovoltaic modules respectively; The output limit of micro gas turbine is expressed as: (7), in, 、 are the maximum and minimum power output of the micro gas turbine respectively; The fuel cell system output limit is expressed as: (8), in, 、 are the maximum and minimum power output of the fuel cell system respectively; The power purchase restriction from power operators is expressed as: (9), in, 、 are the maximum and minimum power purchased from power operators respectively; Adjustable load output limit, its expression is: (10), in, 、 are the maximum and minimum power of the adjustable load of the energy storage device respectively; Flexible load regulation limit, its expression is: (11), in, It is the upper limit of flexible load adjustment; The charging and discharging limits of energy storage equipment are expressed as: (12), (13), (14), (15), (16), in, 、 For energy storage equipment t The charging and discharging power of each scheduling period, 、 Energy storage equipment t The maximum charge and discharge power of each scheduling period, 、 is a binary state variable, For the t The energy storage capacity of each scheduling period, is the energy storage capacity at the previous moment; Energy storage charging efficiency, Energy storage discharge efficiency; The power consumption constraint is expressed as follows: (17), (18), (19), in, represents the amount of electricity sold. Formula (17) indicates that the amount of electricity sold cannot be negative. is the remaining power, To ensure that the electricity sold can only be used for the actual load, and no overselling is allowed. is a binary variable, M is a constant, and formula (19) indicates that conflicts between electricity sales and electricity purchases are avoided.

[0027] According to the system ancillary services market rules, the services that virtual power plants can participate in mainly include frequency regulation and backup services. The response power required for ancillary services is "reserved" from the maximum output of each resource unit. That is, the equipment must reserve sufficient capacity for ancillary services while completing the original scheduling task. Resource units with regulation capabilities (such as fuel cells, micro gas turbines, and battery energy storage) are mapped as adjustable resources, each with a certain degree of frequency regulation or backup response capabilities. The following service participation mapping table is established: Table 1 Service participation mapping table

[0028] Establish ancillary service participation mechanism and benefit evaluation model, including: To ensure the feasibility and resource controllability of ancillary services, it is necessary to introduce a service response power allocation constraint based on the original output power. Therefore, a service response power allocation constraint is constructed, and its expression is: (20), in, For resources r In the t The power generation during each dispatch period, , The reserved frequency regulation and standby service response power are respectively, For resources r Maximum output capacity; Construct the supply and demand matching constraint, whose expression is: (twenty one), (twenty two), in, 、 Respectively represent frequency regulation and standby requirements; Construct the service revenue objective function, which is expressed as follows: (twenty three), in, 、 Respectively t The market price of frequency regulation and standby services during each dispatch period, is the total revenue of ancillary services provided by the virtual power plant during the entire dispatch cycle.

[0029] This benefit will be optimized together with the electricity market cost in the overall objective function to achieve the "dual market benefit maximization" strategy.

[0030] Based on the above steps, the dispatch behavior of the virtual power plant in the electricity market and the ancillary service market is uniformly modeled to construct the objective function of "cost-benefit collaborative optimization". The goal is to minimize the net cost between the total operating cost of the virtual power plant and maximize the ancillary service market revenue. The specific form is: (twenty four), in, minJ represents the minimum net cost, α To adjust the factors, the weight of auxiliary services in the overall objectives is controlled to support sensitivity analysis or multi-scenario trade-off decisions.

[0031] S2. The virtual power plant collaborative scheduling model constructed by the present invention has highly nonlinear, strongly coupled and multi-constrained characteristics. Traditional analytical methods are difficult to solve directly, and linear processing is prone to losing the essential structure of the problem. Therefore, it is necessary to use an intelligent optimization algorithm to conduct a global search of the multidimensional variable space to obtain the optimal scheduling strategy. However, in the process of global search, the algorithm is prone to fall into the local maximum solution; therefore, the present invention introduces a global-local algorithm that combines an improved particle swarm method with a snowmelt optimizer SAO. The improved particle swarm method's initial search for global solutions explores (multi-directions) to quickly expand the distribution of solutions, and the SAO locally refines, improves the accuracy of the solution, and compensates for the particle convergence error. This algorithm has better balancing ability, search efficiency, and adaptability than other optimization algorithms when dealing with complex optimization problems, especially when dealing with multi-peak and high-dimensional problems.

[0032] This paper adopts a hybrid optimization strategy that integrates the improved particle swarm optimization algorithm and the snowmelt optimizer to solve the joint optimization model and obtain the optimal scheduling solution. The specific steps are as follows: Figure 3 Shown, including: S2.1. Use the improved particle swarm optimization algorithm to perform a global search and obtain a scheduling solution. This algorithm uses the improved particle swarm optimization algorithm to perform a global search, update the speed and position of the particles, and gradually obtain a scheduling solution. The update formula for the particle speed and position is: The above S2.1 uses the improved particle swarm optimization algorithm to perform global search, update the speed and position of the particles, and gradually obtain the scheduling plan. The update formula of the particle speed and position is: (25), (26), in, g represents the number of iterations, and represent the velocity and position of the particle respectively, For particles p The best historical position, is the current global optimal position, and is a random vector, represents the Hadamard product, is the linear decrease result of inertia weight, and is the learning factor.

[0033] S2.2. Use the fitness function to evaluate each generation of scheduling solutions: (27), in, represents fitness, represents the operating cost of the virtual power plant, Represents the weight coefficient, which is used to balance the relationship between operating costs and ancillary service revenue. is the total revenue of ancillary services provided by the virtual power plant during the entire dispatch cycle; Determine whether the following convergence conditions are met: Condition 1. Reach the maximum number of iterations ; Condition 2. Continuous The change in the optimal fitness function value within a generation is less than the set threshold ,Right now: (28).

[0034] If the convergence conditions are met, the current optimal solution is output as the final scheduling solution; otherwise, it enters the local optimization stage, that is, S2.3 combines the hybrid optimization strategy of the snowmelt optimizer to perform local fine perturbation and convergence control. S2.3. Combine the hybrid optimization strategy of the snowmelt optimizer to perform local fine perturbation and convergence control. The specific steps include: S2.3.1. Based on the solution to the scheduling solution obtained in S2.1, generate a random set in matrix form and the upper and lower bounds of the solution. The expression for generating the random set in matrix form is: (29), The calculation formula for the upper and lower limits is: (30), in, N and n denote the number of particles in the population and the dimension of the solution, respectively. m represents the dimension index in the matrix, m ∈ n , used to represent the particle in the population at m Dimensional location; U 、 L are the upper and lower limits of the value, is a random number in [0,1]; S2.3.2. Update the particle position by introducing Brownian motion. The update formula of the particle position is: (31), in, For the p The position of the particle, g is the current iteration number, is a random individual in the elite particle collection, A random number vector generated by Gaussian distribution for Brownian motion, To multiply by row, are individuals randomly selected from several elite groups in the representative group. is the center of mass of the entire particle position; S2.3.3. Gradually converge through the snowmelt process to find the optimal solution. The position update formula at this stage is: (32), (33), Where: is a random number in [-1,1], P For the snowmelt model, is the maximum number of iterations, is a random number in [0,1], Indicates the The position or index of a randomly selected individual in a generation.

[0035] Example: This example uses a small electronics assembly plant as an example. This plant primarily performs tasks such as electronic component soldering, module testing, and assembly line assembly. Its daily load fluctuates regularly, with lows in the morning and evening and peaks at noon, with a maximum daily load of approximately 155 kW. To address electricity price fluctuations and achieve energy conservation and emission reduction, the plant deployed a distributed integrated energy system. Using a given-classification algorithm to optimize the scheduling solution, the resulting cost results are shown in Table 2: Table 2 Cost results after optimization of various algorithms

[0036] At the same time, the solution involved in the present invention is simulated, and the simulation results are described as follows: Figure 5This is the 24-hour dispatch output curve of the integrated energy system after the introduction of the flexible load regulation mechanism and the auxiliary service response mechanism. It can be observed from the figure that the system effectively achieves supply and demand balance during high-load periods (such as 11, 14, 21, and 22) by mobilizing the charging and discharging of the energy storage system, flexible load response, and calling on gas units and grid resources. At the same time, during some time periods, photovoltaic and wind power fluctuated significantly (such as the 10th to 13th hours), and the system activated auxiliary services (such as frequency regulation and standby) in a timely manner, providing stable output compensation, thereby improving the safety and stability of system operation. In comparison, Figure 6 The demonstration demonstrates the dispatch strategy when the ancillary services mechanism is disabled. Because energy storage and fuel cells cannot be used for frequency regulation and backup, the system exhibits significant imbalances during periods of significant wind and solar fluctuations. This necessitates greater reliance on grid power purchases or increased gas-fired generator load, increasing operating costs and reducing the system's adaptability to uncertainty. Figure 7 As a result of system operation without the introduction of flexible load regulation, the system is unable to respond to demand-side peak loads and fill valleys, resulting in the need to drastically mobilize gas units and energy storage systems during peak hours in the morning and evening, further increasing the burden of energy storage. At the same time, some photovoltaic power generation is surplus and cannot be utilized, affecting the system's economic efficiency.

[0037] From a cost perspective, Figure 5 The results of various mainstream optimization algorithms for scheduling this integrated energy system are compared. After considering flexible loads and ancillary services, the proposed improved algorithm, iPSO+SAO, achieved the best scheduling results, with an optimal operating cost of 198.45 yuan and an average cost of 199.38 yuan, significantly outperforming other algorithms (e.g., GA with an average cost of 339.94 yuan and AOA with an average cost of 232.42 yuan). This result validates the advantages of the proposed algorithm in search accuracy and global optimization capabilities, and also demonstrates the effective support of flexible strategies and ancillary service mechanisms in improving system economic efficiency.

[0038] The present invention has been described in detail above with reference to the embodiments. However, the contents described are only preferred embodiments of the present invention and should not be considered as limiting the scope of the present invention. All equivalent changes and improvements made within the scope of the present invention should still fall within the scope of the patent coverage of the present invention.

Claims

1. A multi-energy virtual power plant economic and ancillary service optimization method, characterized by: It includes the following steps: S1. Considering the uncertainty of renewable energy, a joint optimization model of virtual power plant economics and ancillary services is constructed; S2. A hybrid optimization strategy combining the improved particle swarm optimization algorithm and the snowmelt optimizer is used to solve the joint optimization model and obtain the optimal scheduling solution.

2. The multi-energy virtual power plant economic and ancillary service optimization method according to claim 1, characterized in that: The S1 construction of a joint optimization model of virtual power plant economy and ancillary services includes: unified modeling of various resources within the virtual power plant under a typical intraday scheduling scenario, and construction of ancillary service participation mechanism and benefit evaluation model.

3. The multi-energy virtual power plant economic and ancillary service optimization method according to claim 2, characterized in that: The unified modeling of multiple resources within the virtual power plant in S1 specifically includes: establishing an objective function for minimizing the operating costs of the virtual power plant and establishing constraints on the virtual power plant.

4. The multi-energy virtual power plant economic and ancillary service optimization method according to claim 3, characterized in that: The expression of the objective function for minimizing the virtual power plant operating cost is: （1）， （2）， （3）， in, represents minimizing the operating cost of the virtual power plant, t Indicates the t scheduling periods, T represents the set sum of scheduling periods, 、 、 、 、 、 The power operators, wind turbines, photovoltaic modules, fuel cell systems, micro gas turbines and energy storage devices during the dispatch period are t energy costs; 、 、 、 、 、 The power operators, wind turbines, photovoltaic modules, fuel cell systems, micro gas turbines and energy storage devices during the dispatch period are t Power injection; is the flexible load adjustment variable, indicating that t The amount of power adjustment achieved by delaying or reducing loads; 、 、 、 are the operating status of wind turbine, photovoltaic module, fuel cell system, and micro gas turbine in the scheduling period t, respectively. "1" indicates operation and "0" indicates shutdown; and Respectively represent the operating status of wind turbine generator set, photovoltaic module, fuel cell system and micro gas turbine in scheduling period t and scheduling period t-1, with "1" representing operation and "0" representing shutdown. represents the operating cost of the energy storage equipment, and They represent the operating status of the energy storage device, and are also set to 0 / 1, which is used to calculate the start and stop cost of the energy storage unit; 、 Respectively i distributed generation units and j The start-up and shutdown costs of each energy storage device; 、 are the number of distributed generation units and energy storage devices, respectively; 、 Respectively b During the dispatch period t resistance and current; 、 、 They are the original loss, final loss and new loss after optimization of the system; 、 They are unit loss cost and power difference before and after optimization respectively; To sell electrical power to the grid, is the electricity selling price, represents the flexible load adjustment cost, It represents the total number of grid branches in the power system, that is, the number of transmission lines in the entire distribution network that participate in the line power loss calculation.

5. The multi-energy virtual power plant economic and ancillary service optimization method according to claim 4, characterized in that: The virtual power plant constraints include: The power balance constraint is expressed as: （4）， in, 、 Energy storage equipment t The discharge and charging power of each scheduling period, For the t The line loss power in each dispatch period is: No. t Load demand during each dispatch period; The output power constraint of the wind turbine generator set is expressed as: （5）， in, 、 are the maximum and minimum power output of the wind turbine generator set respectively; The output power constraint of photovoltaic modules is expressed as: （6）， in, 、 are the maximum and minimum power output of the photovoltaic modules respectively; The output limit of micro gas turbine is expressed as: （7）， in, 、 are the maximum and minimum power output of the micro gas turbine respectively; The fuel cell system output limit is expressed as: （8）， in, 、 are the maximum and minimum power output of the fuel cell system respectively; The power purchase restriction from power operators is expressed as: （9）， in, 、 are the maximum and minimum power purchased from power operators respectively; Adjustable load output limit, its expression is: （10）， in, 、 are the maximum and minimum power of the adjustable load of the energy storage device respectively; Flexible load regulation limit, its expression is: （11）， in, It is the upper limit of flexible load adjustment; The charging and discharging limits of energy storage equipment are expressed as: （12）， （13）， （14）， （15）， （16）， in, 、 For energy storage equipment t The charging and discharging power of each scheduling period, 、 Energy storage equipment t The maximum charge and discharge power of each scheduling period, 、 is a binary state variable, For the t The energy storage capacity of each scheduling period, is the energy storage capacity at the previous moment; Energy storage charging efficiency, Energy storage discharge efficiency; The power consumption constraint is expressed as follows: （17）， （18）， （19）， in, Indicates the amount of electricity sold. is the remaining power, is the actual load, is a binary variable, M is a constant.

6. The multi-energy virtual power plant economic and ancillary service optimization method according to claim 2, characterized in that: The construction of the auxiliary service participation mechanism and benefit evaluation model in S1 specifically includes: Construct the service response power allocation constraint, which is expressed as: （20）， in, For resources r In the t The power generation during each dispatch period, , The reserved frequency regulation and standby service response power are respectively, For resources r Maximum output capacity; Construct the supply and demand matching constraint, whose expression is: （21）， （22）， in, 、 Respectively represent frequency regulation and standby requirements; Construct the service revenue objective function, which is expressed as follows: （23）， in, 、 Respectively t The market price of frequency regulation and standby services during each dispatch period, is the total ancillary service revenue of the virtual power plant during the entire dispatch cycle.

7. The multi-energy virtual power plant economic and ancillary service optimization method according to claim 1, characterized in that: The specific steps of solving the joint optimization model by S2 using the hybrid optimization strategy of integrating the improved particle swarm optimization algorithm and the snowmelt optimizer are as follows: S2.

1. Use the improved particle swarm optimization algorithm to perform a global search and obtain a scheduling solution; S2.

2. Use the fitness function to evaluate the scheduling plan and determine whether it meets the iteration and convergence conditions. If all the iteration and convergence conditions are met, the scheduling plan is selected as the optimal scheduling plan and output. If any of the iteration and convergence conditions are not met, proceed to S2.

3. S2.

3. Combine the hybrid optimization strategy of the snowmelt optimizer to perform local fine perturbation and convergence control to optimize the scheduling plan.

8. The multi-energy virtual power plant economic and ancillary service optimization method according to claim 7, characterized in that: The specific method of obtaining the scheduling plan in S2.1 is: using the improved particle swarm optimization algorithm to perform a global search, update the speed and position of the particles, and gradually obtain the scheduling plan. The update formula of the particle speed and position is: （25）， （26）， in, g represents the number of iterations, and represent the velocity and position of the particle respectively, For particles p The best historical position, is the current global optimal position, and is a random vector, represents the Hadamard product, is the linear decrease result of inertia weight, and is the learning factor.

9. The multi-energy virtual power plant economic and ancillary service optimization method according to claim 7, characterized in that: The fitness function in S2.2 is: (27)， in, represents fitness, represents the operating cost of the virtual power plant, represents the weight coefficient, is the total revenue of ancillary services provided by the virtual power plant during the entire dispatch cycle; Determine whether the following convergence conditions are met: Condition 1. Reach the maximum number of iterations ; Condition 2. Continuous The change in the optimal fitness function value within a generation is less than the set threshold ,Right now: （28）。 10. The multi-energy virtual power plant economic and ancillary service optimization method according to claim 7, characterized in that: The specific steps of S2.3 combining the hybrid optimization strategy of the snowmelt optimizer to perform local fine perturbation and convergence control include: S2.3.

1. Based on the solution to the scheduling solution obtained in S2.1, generate a random set in matrix form and the upper and lower bounds of the solution. The expression for generating the random set in matrix form is: （29）， The calculation formula for the upper and lower limits is: （30）， in, N and n denote the number of particles in the population and the dimension of the solution, respectively. m represents the dimension index in the matrix, m ∈ n , U 、 L are the upper and lower limits of the value, is a random number in [0,1]; S2.3.

2. Update the particle position by introducing Brownian motion. The update formula of the particle position is: （31）， in, For the p The position of the particle, g is the current iteration number, is a random individual in the elite particle collection, A random number vector generated by Gaussian distribution for Brownian motion, To multiply by row, are individuals randomly selected from several elite groups in the representative group. is the center of mass of the entire particle position; S2.3.

3. Gradually converge through the snowmelt process to find the optimal solution. The position update formula at this stage is: （32）， （33）， Where: is a random number in [-1,1], P For the snowmelt model, is the maximum number of iterations, is a random number in [0,1], Indicates the The position or index of a randomly selected individual in a generation.

Images