Virtual power plant optimization scheduling method considering uncertain factors

Abstract

The invention relates to a virtual power plant optimal scheduling method considering uncertain factors. The method comprises the following steps: step 1, generating a wind and light uncertainty scene by adopting a Monte Carlo simulation method based on Latin hypercube sampling; step 2, reducing the wind and light uncertainty scene generated by simulation in the step 1; 3, constructing a virtual power plant optimization scheduling model considering uncertainty factors on the basis of the wind and light uncertainty scene cut down in the step 2; and step 4, based on the virtual power plant optimization scheduling model considering the uncertainty factors established in the step 3, designing a plurality of solving scenes of the virtual power plant optimization scheduling model considering the uncertainty factors, performing solving, and finally selecting an optimal scheme of virtual power plant optimization scheduling. According to the method, the optimal scheduling of the operation of the virtual power plant can be realized, the stability of power supply is ensured, and the problem of output uncertainty of different resources is solved.

Description

Technical Field

[0001] The present invention relates to the technical field of power system optimization and scheduling, and in particular to a virtual power plant optimization and scheduling method that can adapt to the uncertainty of distributed energy output such as wind power and photovoltaic power. Background Art

[0002] As an emerging form of power resource management, Virtual Power Plant (VPP) integrates decentralized power generation units, energy storage equipment, controllable loads and other resources through information and communication technology, and participates in the operation and scheduling of the power grid as a whole to improve the operation efficiency of the power grid.

[0003] The core advantage of virtual power plants lies in their flexibility and scalability, which can adapt to the ever-changing electricity market demand and supply conditions. However, the distributed generation resources in virtual power plants, especially renewable energy such as wind power and photovoltaic power, have significant randomness and volatility.

[0004] These characteristics make virtual power plants face many uncertain challenges in actual operation. For example, the unpredictability of wind speed and solar radiation leads to unstable output of wind and photovoltaic power generation; there is also uncertainty in changes in load demand and the operating status of power generation equipment. These uncertain factors have a significant impact on the stable operation of virtual power plants.

[0005] In the existing technology, the optimization scheduling methods of virtual power plants often rely on deterministic models, which assume that all input parameters are known, or use simplified probabilistic models to describe uncertainty. However, this method often cannot provide sufficiently flexible and robust scheduling strategies when facing complex and changing power markets and highly uncertain renewable energy output.

[0006] Therefore, the present invention provides a virtual power plant optimization scheduling method that can comprehensively consider various uncertain factors, which is of great significance for improving the system stability of the virtual power plant.

[0007] After searching, no public documents of the prior art identical or similar to the present invention were found. Summary of the invention

[0008] The purpose of the present invention is to overcome the shortcomings of the prior art and propose a virtual power plant optimization scheduling method taking into account uncertain factors. By comprehensively considering the random fluctuations and prediction errors of various energy sources in the system, the optimal scheduling of the virtual power plant operation is achieved, the stability of the power supply is ensured, and the uncertainty problem of the output of different resources is solved.

[0009] The present invention solves the practical problem by adopting the following technical solutions:

[0010] Step 1: Generate wind and solar uncertainty scenarios using the Monte Carlo simulation method based on Latin hypercube sampling to characterize the uncertainty characteristics of distributed wind power and photovoltaic output in the operation of virtual power plants;

[0011] Step 2: Use the fast previous generation elimination technology based on probability distance to reduce the wind and solar uncertainty scenario simulated and generated in step 1;

[0012] Step 3: Based on the uncertainty scenario of wind and solar power after curtailment in step 2, a virtual power plant optimization scheduling model considering uncertainty factors is constructed;

[0013] Step 4: Based on the virtual power plant optimization scheduling model considering uncertainty factors established in step 3, design a variety of solution scenarios for the virtual power plant optimization scheduling model considering uncertainty factors, solve them, and finally select the best solution for virtual power plant optimization scheduling.

[0014] Advantages and beneficial effects of the present invention:

[0015] 1. The present invention proposes a virtual power plant optimization scheduling method taking into account uncertain factors, taking into account the uncertain factors of each unit of the virtual power plant, and designing relevant strategies for combining wind and solar resources and adjustable resources. First, based on the Monte Carlo simulation method and the scene distance reduction technology of Latin hypercube sampling, the overall space of wind and solar samples is fully covered. On this basis, other uncertainties of wind and solar except for its own output volatility are considered. Secondly, taking the minimum comprehensive cost of virtual power plant operation as the objective function, the present invention establishes an optimization scheduling model taking into account uncertain factors such as load forecasting, distributed new energy and generator power generation, component state operation, power market electricity price and load response, and designs 6 mode scenarios to study the impact of the access capacity and number differences of different internal resources on the virtual power plant. The simulation results of the present invention show that under the condition of considering the above-mentioned uncertain factors, when the virtual power plant is connected to 10 70KW distributed wind power equipment and 50 6kW distributed photovoltaic units, and the ratio of energy storage, gas turbine and wind and solar resources is kept at 15% and 25%, the virtual power plant can achieve peak shaving and valley filling and stable operation.

[0016] 2. The virtual power plant optimization scheduling method of the present invention takes into account uncertain factors such as load forecasting, distributed renewable energy power generation, component status operation, electricity market price and load response, thereby improving the flexibility and robustness of the scheduling strategy.

[0017] 3. The present invention can more effectively utilize various resources within the virtual power plant, such as wind power, photovoltaics, energy storage equipment, gas turbines and controllable loads, by designing a strategy that combines wind and solar resources with adjustable resources to achieve optimal resource allocation. It also helps to better integrate and utilize new energy sources such as wind power and photovoltaics, and promote the large-scale consumption of new energy.

[0018] 4. The present invention uses advanced technologies and tools such as Latin hypercube sampling, scene distance reduction technology, and mixed integer linear programming to ensure the scientificity and effectiveness of the optimization process. By considering uncertainty factors, it helps to improve the stable operation capability of virtual power plants in the face of renewable energy volatility and market changes; through strategies such as peak shaving and valley filling, the present invention can enable virtual power plants to better adapt to changes in grid demand and improve the overall operation efficiency of the grid. BRIEF DESCRIPTION OF THE DRAWINGS

[0019] Figure 1 is a process flow chart of the present invention;

[0020] Figure 2 It is a load day-ahead forecast diagram, which is a day-ahead forecast result diagram of the next day's load within the virtual power plant dispatching range based on historical data;

[0021] Figure 3 This is a cost comparison chart of the virtual power plant optimization results considering the uncertainty of wind power;

[0022] Figure 4 This is the result diagram of power optimization dispatch considering the uncertainty of wind power;

[0023] Figure 5 This is the result diagram of energy storage optimization scheduling considering the uncertainty of wind power;

[0024] Figure 6 This is a cost comparison chart of the virtual power plant optimization results considering PV uncertainty;

[0025] Figure 7 This is the result diagram of power optimization scheduling considering photovoltaic uncertainty;

[0026] Figure 8 This is the result diagram of energy storage optimization scheduling considering photovoltaic uncertainty;

[0027] Fig. 9 This is a cost comparison chart of the virtual power plant optimization results considering the uncertainty of wind and solar power and different wind and solar power ratios;

[0028] Fig.10 This is the result diagram of power optimization dispatch considering the uncertainty of wind and solar power and different ratios of wind and solar power;

[0029] Fig.11 This is the result diagram of energy storage optimization scheduling considering the uncertainty of wind and solar power and different ratios of wind and solar power;

[0030] Fig.12 This is the result diagram of power optimization dispatch considering the uncertainty of wind and solar power and the difference in the number of wind power accesses;

[0031] Fig.13This is the result diagram of energy storage optimization scheduling considering the uncertainty of wind and solar power and the difference in the number of wind power connections;

[0032] Fig.14 This is the result diagram of power optimization dispatch considering the uncertainty of wind and solar power and the difference in the number of wind and solar power accesses;

[0033] Fig.15 This is the result diagram of energy storage optimization scheduling that takes into account the uncertainty of wind and solar power and the difference in the number of wind and solar power connections;

[0034] Fig.16 It is the result diagram of power optimization dispatch considering the uncertainty of wind, solar and controllable load;

[0035] Fig.17 This is a diagram of the optimal dispatching results of energy storage considering the uncertainty of wind, solar and controllable loads. DETAILED DESCRIPTION

[0036] The embodiments of the present invention are further described in detail below with reference to the accompanying drawings:

[0037] A virtual power plant optimization scheduling method considering uncertain factors, such as Figure 1 As shown, the following steps are included:

[0038] Step 1: Generate wind and solar uncertainty scenarios using the Monte Carlo simulation method based on Latin hypercube sampling to characterize the uncertainty characteristics of distributed wind power and photovoltaic output in the operation of virtual power plants;

[0039] After stratified sampling of the cumulative probability curve using the Latin hypercube sampling method, sample data is obtained to ensure full coverage of the overall spatial sample, and the uncertainty of wind and solar power is processed through a multi-scenario generation method.

[0040] The Latin hypercube sampling method is a method that can process random variables. The present invention assumes that the distributed wind power and photovoltaic output as random variables obey the historical distribution. Therefore, the present invention adopts the Latin hypercube sampling method to simulate and generate scenarios to characterize the uncertainty characteristics of distributed wind power and photovoltaic output in the operation of virtual power plants, and uses the probability-based fast previous generation elimination technology to obtain typical scenarios and corresponding probabilities taking wind turbines as examples.

[0041] The specific steps of step 1 include:

[0042] (1) First, collect sampling data of wind power scenarios, that is, determine the random variables such as wind speed and wind direction at different times required for wind power scenario generation, determine the uncertainty parameters of wind power and photovoltaic output, such as wind speed and solar radiation intensity, and define a reasonable parameter range for each parameter;

[0043] (2) According to the parameter range defined in step (1), it is divided into several equally probable self-controlled years or "layers" and sample points are generated. That is, according to the characteristics of the wind power scenario sampling data, the matrix of random numbers, sampling value probabilities and variable space is generated in sequence according to the Monte Carlo method, and sampling is performed according to the Latin hypercube sampling method to ensure that each layer is evenly sampled for constructing the wind power scenario;

[0044] (3) Perform correlation verification on the random numbers, sampling value probabilities and variable space matrices generated in step (2) to ensure that the generated data conforms to the correlation of the actual wind power data; if the correlation condition is met, proceed to step (4); if not, return to step (2) and regenerate;

[0045] (4) The newly generated data set in step (3) that has been verified for correlation is used to form a wind power scenario sampling value matrix. This matrix is ​​a set of sample points generated by the Latin hypercube sampling method. Each row of the matrix represents a complete wind power output scenario, and each column represents a specific parameter. The matrix is ​​merged with the existing wind power scenario sampling value matrix to enrich the diversity of wind power scenarios.

[0046] (5) Check whether the merged wind power scenario sampling value matrix meets the convergence conditions; if so, proceed to step (6); if not, return to step (3) to continue to expand the sampling, indicating that the distribution and diversity of the sample points are not sufficient to fully cover the entire parameter space, and it is necessary to regenerate the sample points and verify the correlation until the convergence conditions are met.

[0047] (6) Stop expanding and generating the wind power scenario sampling value matrix, complete the wind and solar uncertainty scenario simulation generation, and organize the wind and solar uncertainty scenario data.

[0048] Step 2: Use the fast previous generation elimination technology based on probability distance to reduce the wind and solar uncertainty scenario simulated and generated in step 1.

[0049] The increase in the size of the scene set will increase the amount of calculation required to solve the optimization model. Considering the trade-off between calculation accuracy and amount of calculation, a fast predecessor elimination technique based on probability distance is used to reduce the scenes.

[0050] The specific steps of step 2 include:

[0051] (1) Set the wind and solar uncertainty scenario set simulated and generated in step 1 as Y, and calculate the geometric distance between each pair of scenarios in the large-scale scenario set Y of wind turbine output and photovoltaic output;

[0052] (2) Compare the geometric distances between the scenarios and select the wind and solar power output uncertainty scenario y with the smallest sum of probability distances to the remaining scenarios;

[0053] (3) Select the scene in Y that is closest to scene y in geometric distance, set it as r, and make it replace scene y. Add the probability of y to the probability of scene r, eliminate d, and form a new s';

[0054] (4) Determine whether the number of remaining wind and solar power uncertain output scenarios meets the requirement. If it does, the scenario reduction is terminated. If it does not, repeat steps (1) to (3) to ultimately achieve the purpose of scenario reduction.

[0055] Step 3: Based on the uncertainty scenario of wind and solar power after curtailment in step 2, a virtual power plant optimization scheduling model considering uncertainty factors is constructed;

[0056] Step 3 mainly discusses the optimization scheduling problem of virtual power plants considering uncertain factors. Virtual power plants actively encourage new energy sources such as wind and solar power to optimize their output, and adjustable resources to flexibly assist in operation. Users regulate electricity consumption based on dynamic electricity prices, and reduce the uncertainty pressure that the power system may face by shaving peaks and filling valleys.

[0057] The specific steps of step 3 include:

[0058] (1) Setting the objective function

[0059] In the virtual power plant, 24 hours is set as one calculation cycle, and the minimum comprehensive cost of the virtual power plant operation within a certain period of time is taken as the objective function of the current model, including wind power operation and maintenance costs, photovoltaic operation and maintenance costs, gas turbine costs, energy storage equipment costs, total costs of purchasing and selling electricity, and controllable load costs. In addition, the operating conditions under each set scenario are also included, including electricity prices, wind and solar output forecasts, etc.

[0060] The objective function of the virtual power plant optimization scheduling problem taking into account uncertainties is as follows:

[0061]

[0062] In the formula, C Total is the comprehensive cost of the virtual power plant during a certain period of operation, T is the total number of periods in the entire dispatch cycle, and C WP,n (t), C PV,n (t), C GT,n (t), C ES,n (t), C buy (t) are the wind turbine operation and maintenance costs, photovoltaic operation and maintenance costs, gas turbine operation and maintenance costs, energy storage operation and maintenance costs and electricity purchase costs during period t.

[0063] a. Fan operation and maintenance costs

[0064] C WP,n (t) represents the cost required for the virtual power plant to call for wind power operation and maintenance during period t, and the calculation formula is:

[0065] C WP,n (t) = η WP ×P WP,n (t) (2)

[0066] Where η WP is the wind power operation coefficient, P WP,n (t) is the output forecast value of n units in the wind farm at time t; at the same time, since the output of new energy at each moment is affected by uncertain factors such as weather, there will be a certain output uncertainty. The present invention uses the mean and random deviation to represent the uncertainty of the output of new energy, which is expressed as:

[0067]

[0068] In the formula, is the mean output of wind power, ξ WP,t is the random deviation of wind power and photovoltaic power output.

[0069] Combining formula (3), the cost of wind power operation is:

[0070]

[0071] b. Photovoltaic operation and maintenance costs

[0072] C PV,n (t) represents the cost required for the virtual power plant to call for photovoltaic operation during period t, and the calculation formula is:

[0073] C PV,n (t) = δ PV ×P PV,n (t) (5)

[0074] In the formula, δ PV is the photovoltaic operation coefficient, P PV,n (t) is the output forecast value of n units at the photovoltaic station at time t. In addition, the volatility caused by uncertain factors must also be considered. Combining formula (6), we can get:

[0075]

[0076] In the formula, is the average output of photovoltaic power, ξ PV,t is the random deviation of photovoltaic output.

[0077] The cost of photovoltaic operation is

[0078]

[0079] c. Cost of gas turbine

[0080] CGT The cost of the gas turbine of the virtual power plant in period t is expressed as follows:

[0081]

[0082] In the formula, γ is the fixed startup cost, s GT,open is the startup state of the VPP gas turbine during period t, is the piecewise linearization cost, P GT,t is the engine output during period t, s GT,close In shutdown state, C GT,stop is the start-stop fee; n is the number of connected devices. The volatility caused by uncertain factors must also be considered.

[0083]

[0084] In the formula, is the mean output of the generator, ξ GT,t It is the random deviation of generator output.

[0085] Combining formula (9), the cost of gas turbine operation is

[0086]

[0087] d. Cost of energy storage equipment

[0088] C ES,n (t) represents the cost of calling the energy storage equipment for the virtual power plant during period t, and the calculation formula is:

[0089]

[0090] In the formula, p ES,sell (t) is the electricity price of the energy storage device at time t, λ out is the discharge efficiency of the energy storage device. μ(t) represents the charging and discharging state of the energy storage device. It is a 0-1 variable. μ(t) = 0 means that the energy storage device is charging, and the cost is 0 at this time. μ(t) = 1 means that the energy storage device is discharging. At this time, is the discharge power.

[0091] e. Total cost of electricity purchase and sale

[0092] C buy (t) represents the total cost of electricity purchase and sale by the virtual power plant in the day-ahead market during period t, and the calculation formula is:

[0093] C buy (t) = C buy,t P buy,t (12)

[0094] In the formula, C buy,trepresents the electricity purchase and sale price of the virtual power plant in period t, P buy,t Represents the electricity purchase amount during period t.

[0095] f. Controllable load cost

[0096] C RL (t) and C TF (t) represents the cost of load reduction and load transfer in period t, and the calculation formula is:

[0097] C RL (t) = C RL,t P RL,t (13)

[0098] C TF (t) = C TF,t P TF,t (14)

[0099] In the formula, C RL,t and C TF,t represents the unit cost of load that can be reduced and transferred at time t, P RL,t and P TF,t Represents the amount of load that can be reduced and transferred at time t. Considering the volatility caused by uncertain factors, the load output can be reduced and transferable load output The following conditions are met: the mean is μ and the standard deviation is σ 2 The normal distribution of

[0100]

[0101] (2) Constructing constraints

[0102] a. Power balance constraints

[0103] The calculation formula of the electric energy balance constraint of the model of the present invention is:

[0104]

[0105] Where P PV (t), P wp (t), P buy (t), P GT (t) covers the uncertainty interval, L(t) is the electric load power at time t, and the specific calculation formula is:

[0106] L(t)=Fload(t)+L RL (t)+L TF (t) (18)

[0107] In the formula, the load power L(t) consists of three parts: the uncontrollable load Fload(t), the reducible load L RL (t) and transferable load L TF (t), the latter two take into account the uncertainty of controllable load.

[0108] b. Fan output related constraints

[0109] P WP (t) is the wind power output. In the previous research work, the Monte Carlo method was used to generate scenarios for wind power output, and the fast predecessor elimination technology based on probability distance was used to reduce the scenarios.

[0110] The fan-related constraints are:

[0111] P WP (t) min ≤P WP (t)≤P WP (t) max (19)

[0112] Where P WP (t) max , P WP (t) min They are the corresponding maximum and minimum power values ​​of the wind turbines respectively.

[0113] c. Constraints on photovoltaic output

[0114] P PV (t) is the photovoltaic output power. In the previous research work, the Monte Carlo method was used to generate scenarios for photovoltaic output, and the fast predecessor elimination technology based on probability distance was used to reduce the scenarios.

[0115] The relevant constraints of the photovoltaic unit are:

[0116] P PV (t) min ≤P PV (t)≤P PV (t) max (20)

[0117] Where P PV (t) max , P PV (t) min They are the corresponding maximum and minimum power values ​​of the photovoltaic system respectively.

[0118] d. Gas turbine related constraints

[0119] i) Operational constraints

[0120] The operating constraints of the gas turbine are:

[0121]

[0122] In the formula, and P GT,t are the corresponding maximum and minimum power values ​​of the gas turbine, T onGT,t-1 is the continuous operation time of the gas turbine, t onGT is the shortest startup time, T offGT,t-1 is the continuous downtime, t offGT Minimum downtime.

[0123] ii) Hill climbing constraints

[0124] The gas turbine unit is also subject to the constraint of ramp power, which is calculated as follows:

[0125]

[0126] In the formula, are the upward and downward climbing rates of the gas turbine unit, respectively.

[0127] e. Energy storage charging and discharging constraints

[0128] The constraints related to energy storage charging and discharging are:

[0129]

[0130] In the formula, P ES,t , They are the minimum and maximum values ​​of the energy storage battery capacity respectively;

[0131] is the discharge power, is the charging power;

[0132] r min 、r max It is the minimum and maximum value of the corresponding climbing power when charging and discharging the energy of the storage;

[0133] μ(t) represents the charging and discharging state of the energy storage device and is a 0-1 variable. μ(t)=0 represents charging of the energy storage device, and μ(t)=1 represents discharging of the energy storage device.

[0134] f. Controllable load constraints

[0135] P RL,min ≤P RL,t ≤P RL,max (twenty four)

[0136] P TF,min ≤P TF,t ≤P TF,max (25)

[0137] Where P RL,min , P RL,max They are the minimum and maximum values ​​of load that can be reduced respectively;

[0138] P TF,min , P TF,max are the minimum and maximum values ​​of the transferable load respectively.

[0139] Step 4: Based on the virtual power plant optimization scheduling model considering uncertainty factors established in step 3, design a variety of solution scenarios for the virtual power plant optimization scheduling model considering uncertainty factors, solve them, and finally select the best solution for virtual power plant optimization scheduling;

[0140] The specific method of step 4 is:

[0141] The optimization problem of the virtual power plant optimal scheduling model considering uncertain factors is a 0-1 type mixed integer linear programming problem, and the standard form of its solution model is:

[0142]

[0143] In the formula, the optimization variable x includes the load forecast, distributed renewable energy power generation, generator output, electricity price in the power market, power purchase and sales, load response and component operation status taking into account uncertain factors; the equality constraint is the power balance constraint; the inequality constraint is the operation constraint of each device; p represents the variable subject to each inequality constraint in the planning problem, including wind turbine output constraint, photovoltaic output constraint, gas turbine operation and climbing constraint, energy storage charging and discharging constraint. Controllable load constraint; q represents the constraint condition that some unknown numbers in the planning problem are integers, including the start and stop of each unit. Based on this, the present invention adopts Gurobi solver to solve in MATLAB.

[0144] Six model scenarios taking into account uncertainty are designed, as shown in Table 1. By comparing the comprehensive costs of virtual power plants under the six scenarios, the effectiveness of virtual power plants in peak shaving and valley filling can be reflected, and the best solution for optimizing the scheduling of virtual power plants can be finally selected.

[0145] Table 1 Scene settings

[0146]

[0147] The solution scenarios of the virtual power plant optimization scheduling model considering uncertainty factors in step 4 include:

[0148] Scenario 1: Optimal dispatch of virtual power plants considering wind power uncertainty

[0149] In this scenario, the optimal dispatch of the virtual power plant considering the uncertainty of wind power is input with different wind power capacity values ​​ranging from 200kW to 900kW.

[0150] Scenario 2: Optimal dispatch of virtual power plants considering PV uncertainty

[0151] In this scenario, the optimal dispatch of the virtual power plant considering PV uncertainty is input with different PV capacity values ​​ranging from 200kW to 900kW.

[0152] Scenario 3: Optimal dispatch of virtual power plants considering the uncertainty of wind power and photovoltaic power and different ratios of wind power and photovoltaic power

[0153] In this scenario, the virtual power plant optimizes the scheduling of wind power and photovoltaic power with comprehensive consideration of the uncertainties, with different capacities and ratios of wind power and photovoltaic power input ranging from 200kW to 800kW.

[0154] Scenario 4: Optimal scheduling of virtual power plants considering the uncertainty of wind and solar power and the difference in the number of wind power connections

[0155] In this scenario, the optimal scheduling of virtual power plants with comprehensive consideration of the uncertainty of wind and solar power is studied, with the input of 1 to 50 distributed wind turbines of the same capacity, to study the impact of the difference in the number of wind power access on the economic efficiency of virtual power plants.

[0156] Scenario 5: Optimal scheduling of virtual power plants considering the uncertainty of wind and solar power and the difference in the number of wind and solar power connections

[0157] In this scenario, the uncertainty of wind and solar power and the optimal scheduling of virtual power plants with the number of wind turbines connected are comprehensively considered. With 1 to 50 distributed photovoltaic units of the same capacity as input, the impact of differences in the number of photovoltaic connections on the economic efficiency of virtual power plants is studied.

[0158] Scenario 6: Optimal scheduling of virtual power plants considering the uncertainty of wind, solar and controllable loads

[0159] In this scenario, the optimal scheduling of virtual power plants, which comprehensively considers the uncertainties of wind power and photovoltaic power (including capacity and number), will include controllable load inputs of curtailable loads and transferable loads.

[0160] The working principle of the present invention is:

[0161] First, the Monte Carlo method using Latin hypercube sampling is used to generate renewable energy output scenarios. Second, the scenario reduction is performed based on the fast previous generation elimination technology based on probability distance to obtain the predicted wind and light output of the virtual power plant for the next day. In addition, the optimal objective function of the virtual power plant is set based on the operating characteristics of wind power, photovoltaic power, gas turbine, energy storage equipment and controllable load, and the constraints of each device taking into account uncertainty factors are input. The different impacts of different operating strategies of wind and solar operation and adjustable resources under different conditions are analyzed, and 6 model scenarios taking into account uncertainty are designed; finally, the virtual power plant optimization scheduling problem is transformed into a mixed integer linear programming model, and the six model scenarios are solved using the Gurobi solver, and finally the optimal result of the virtual power plant's next-day optimization scheduling is obtained.

[0162] The present invention is further described below by means of specific examples:

[0163] This example uses Matlab R2019b version to write the program, the computer processor model is AMD Ryzen 74800H with Radeon Graphics, the machine has 16GB of RAM, and Gurobi Optimizer version10.0.2build v10.0.2rc0 (win64) is used. In view of the instability caused by the uncertainty of wind and solar energy output, this example uses energy storage, gas turbines and controllable loads as adjustable resources to ensure the safety, stability and reliability of the virtual power plant.

[0164] 1. Simulation parameter setting

[0165] The composition of the virtual power plant includes distributed wind turbines, distributed photovoltaic units, small gas turbine power stations, distributed energy storage equipment and controllable loads. The specific access capacity and number refer to the above six scenarios. The total installed capacity of distributed wind turbines is between 200-900kW, the total installed capacity of distributed photovoltaic units is between 200-900kW, the power supply range of gas turbines is set between 50-120kW, the maximum charging power of distributed energy storage equipment is -1000kW, and the maximum discharge power is 1000kW. The time-of-use electricity price for the purchase of electricity by the virtual power plant is set at 0.63 yuan / kW, 1.02 yuan / kW, and 1.52 yuan / kW, and the corresponding time-of-use electricity price for electricity sales is 1.5 times the corresponding time-of-use electricity price for electricity purchase. The operating cost coefficients of wind power and photovoltaic power are both 360 and 300, the upper limit of a single wind power output is set at 600kW, and the upper limit of a single photovoltaic output is set at 2200kW.

[0166] The time-of-use electricity price period division and electricity price information used in this example are shown in Table 2, the unit parameter information is shown in Table 3, the wind power and photovoltaic related information is shown in Table 4, and the operating parameters of the energy storage battery are shown in Table 5.

[0167] Table 2 Time-of-use electricity price period division and electricity price information

[0168]

[0169] Table 3 Wind power and photovoltaic parameter information

[0170]

[0171]

[0172] Table 4 Gas unit parameter information

[0173]

[0174] Table 5 Energy storage unit parameter information

[0175]

[0176] Based on historical data, the day-ahead forecast results of the next day’s load within the virtual power plant dispatch range are as follows: Figure 2 shown.

[0177] Depend on Figure 2 It can be seen that there are two small peaks in electricity load the next day, the first at 10 am and the second from 19:00 to 20:00; the valley period is from 23:00 to 5:00 the next day, with the lowest electricity load.

[0178] In addition, the distributed wind power and distributed photovoltaic output in the undispatched units have strong volatility and uncertainty, which can be used as random variables in the present invention. The uncertainty of wind power and photovoltaic output is processed using the scenario analysis method.

[0179] The Latin hypercube sampling method is a method that can process random variables. The present invention assumes that the distributed wind power and photovoltaic output as random variables obey the historical distribution. Therefore, the present invention adopts the Latin hypercube sampling method to simulate and generate scenarios to characterize the uncertainty characteristics of distributed wind power and photovoltaic output in the operation of virtual power plants, and uses the probability-based fast previous generation elimination technology to obtain typical scenarios and corresponding probabilities of distributed wind power and photovoltaic output.

[0180] The virtual power plant predicts that the wind speed of wind power will remain in a fluctuating range during the day, with the peak mainly in the evening to before sunrise the next day; for photovoltaics, the photovoltaic output is predicted to be highest between 10:00 and 14:00, as it is noon and the energy of sunlight is the highest, so the photovoltaic output is also the highest; between 18:00 and 6:00 the next day, there is no sunlight after the sun sets, and there is no photovoltaic output at this time.

[0181] 2. Analysis of simulation results in different scenarios

[0182] The present invention divides the optimization scheduling mode scenario settings of the virtual power plant into 6 types, compares the comprehensive costs of the virtual power plant under the 6 scenarios, and finally determines the most economical optimization scheduling plan for the internal resources of the virtual power plant.

[0183] (1) Scenario 1: Optimal scheduling of virtual power plants considering wind power uncertainty

[0184] The rated powers of the three DGs, photovoltaic, energy storage and gas turbine, are set to 200, 50 and 50 kW respectively. Wind power with different capacities ranging from 200 kW to 900 kW is connected, with a sampling interval of 100 kW, and is divided into 8 different scenarios. The scenarios considering the uncertainty of wind power are shown in Table 6.

[0185] Table 6 Scenarios considering wind power uncertainty

[0186]

[0187]

[0188] The model was built for solving 8 different scenarios, with the goal of minimizing the cost of the virtual power plant. The cost comparison of the virtual power plant optimization results considering wind power uncertainty is shown in Figure 2. Figure 3 shown.

[0189] The optimization results show that as the wind power capacity connected to the virtual power plant increases, the overall cost decreases, and the overall cost reduction trend gradually slows down. In scenario 8, when the ratio of wind and solar resources is 900:200, the overall cost of the virtual power plant is the lowest, at 15,260.3799 yuan, which is 37.46% lower than scenario 1, when the ratio of wind and solar resources is 200:200.

[0190] The model considers the optimal dispatch of virtual power plants with uncertainties in wind power, with the minimum comprehensive cost of virtual power plants in the corresponding period as the objective function, and the ultimate goal is to maximize the benefits of virtual power plants. For scenario 8, the results of power optimization dispatch are as follows: Figure 4 As shown in Figure 2, the optimal dispatch results of energy storage considering wind power uncertainty are as follows: Figure 5 shown.

[0191] As can be seen from the results, the peak period occurred twice. The first time was from 9:00 to 11:00, when photovoltaic power began to play a role, but at this time, the wind and solar resources were insufficient, so the energy storage began to discharge, alleviating the problem of increased electricity consumption during the morning working hours. At noon, as photovoltaic power generation began to reach its peak, the energy storage would switch to a charging state. The second time occurred from 14:00 to 21:00, when photovoltaic power was in a state of reduced output and the purchase price of electricity was relatively high. Since the wind power access capacity was as high as 900kW, the wind power output in this stage could meet the power demand of the virtual power plant, and even store the electricity, such as during the period from 17:00 to 20:00.

[0192] During off-peak hours, wind power generation and gas turbines generate electricity load, and the excess electricity load is stored in batteries to absorb the excess wind power. Since the cost of purchasing electricity is lower than that of generating electricity, purchasing electricity can be chosen instead of generating electricity, and the excess electricity can be stored to meet the electricity demand during peak hours.

[0193] During normal times, wind turbines continue to output power, and the 900kW wind power output is sufficient to maintain basic electricity demand. The gas turbine output can be used to supplement it, and the excess electricity can be stored.

[0194] (2) Scenario 2: Optimal scheduling of virtual power plants considering PV uncertainty

[0195] The rated powers of wind power, energy storage and gas turbine are set to 200, 50 and 50 kW respectively. Photovoltaics with different capacities ranging from 200 kW to 900 kW are connected, with a sampling interval of 100 kW, and divided into 8 different scenarios. The scenarios considering the uncertainty of photovoltaics are shown in Table 7.

[0196] Table 7 Scenarios considering PV uncertainty

[0197]

[0198]

[0199] The model was built for solving 8 different scenarios, with the goal of minimizing the cost of the virtual power plant. The cost comparison of the virtual power plant optimization results considering photovoltaic uncertainty is shown in Figure 2. Figure 6 shown.

[0200] The optimization results show that as the photovoltaic capacity connected to the virtual power plant increases, the overall cost decreases, and the overall cost reduction trend gradually slows down. In scenario 8, when the ratio of wind and solar resources is 200:900, the overall cost of the virtual power plant is the lowest, at 24,424.3825 yuan, which is 8.10% lower than scenario 1, when the ratio of wind and solar resources is 200:200.

[0201] The model takes into account the optimal scheduling of virtual power plants with PV uncertainty, with the minimum comprehensive cost of the virtual power plant in the corresponding period as the objective function, and the ultimate goal is to maximize the benefits of the virtual power plant. For scenario 8, the results of power optimization scheduling are as follows: Figure 7 As shown in Figure 2, the optimal dispatch results of energy storage considering photovoltaic uncertainty are as follows: Figure 8 shown.

[0202] From the results, we can see that the peak period appeared twice. The first time was from 9:00 to 11:00, when photovoltaic power generation gradually reached its peak. At 8:00 on the same day, the working hours began and the electricity load increased. At this time, the photovoltaic output was not obvious. Therefore, under the condition of normal fluctuation of wind power output, energy storage discharge maintained a balance and the gas turbine operated at full load. Until 9:00, 900kW of photovoltaic access greatly eased the pressure on electricity consumption. The virtual power plant stopped purchasing electricity and began to call on photovoltaic resources. At the same time, due to the time-of-use electricity price, the grid price was high at this stage. During the peak period of electricity consumption, the battery was in a discharge state, and the released electricity was sold to the grid, which could save a certain amount of cost. The second time appeared from 18:00 to 20:00, when photovoltaics were almost not working and the electricity price was relatively high. At this time, the gas turbines and other equipment were running at maximum power and the battery was in a state of energy discharge. If the wind and solar output could not meet the system needs at this time, the insufficient electricity needed to be purchased from the grid to meet the needs, and the purchase amount reached the maximum value.

[0203] During off-peak hours, wind power generation and gas turbines generate electricity load, and the excess electricity load is stored in batteries to absorb the excess wind power. Since the cost of purchasing electricity is lower than that of generating electricity, purchasing electricity can be chosen instead of generating electricity, and the excess electricity can be stored to meet the electricity demand during peak hours.

[0204] During normal times, photovoltaic output is large, gas turbine output is sufficient to meet the electricity load demand, and excess electricity can be stored.

[0205] (3) Scenario 3: Optimal scheduling of virtual power plants considering wind and solar uncertainty and different wind and solar ratios

[0206] While keeping the ratio of energy storage, gas turbines and wind and solar resources at 15% and 25%, scenario 3 studies the optimal scheduling of the access capacity ratio of wind turbines and photovoltaic units. The rated power of energy storage and gas turbines is set to 150kW and 250kW respectively. While ensuring the total wind and solar resources of 1000kW, wind and solar units with different capacity ratios ranging from 100kW to 900kW are connected. The sampling interval is 100kW and divided into 6 sub-scenarios. The scenarios considering the uncertainty of wind and solar are shown in Table 8.

[0207] Table 8 Scenarios considering wind and solar uncertainty

[0208]

[0209] The model was built for solving six different scenarios, with the goal of minimizing the cost of the virtual power plant. The cost comparison of the virtual power plant optimization results considering the uncertainty of wind and solar power and different wind and solar power ratios is shown in the figure. Fig. 9 shown.

[0210] The optimization results show that scenario 5 has the lowest overall cost, that is, when the wind power access capacity is 700kW and the photovoltaic access capacity is 300kW, the overall cost of the virtual power plant is 15,073.1307 yuan, which is 32.81% lower than scenario 1, that is, when the ratio of wind and solar resources is 200:800.

[0211] The model considers the optimal dispatch of virtual power plants with uncertainties in wind power and photovoltaic power, with the minimum comprehensive cost of the virtual power plant in the corresponding period as the objective function, and the ultimate goal is to maximize the benefits of the virtual power plant. For scenario 5, the results of power optimization dispatch are as follows: Fig.10 As shown in the figure, the optimal dispatch results of energy storage considering the uncertainty of wind and solar power and different ratios of wind and solar power are as follows: Fig.11 shown.

[0212] As can be seen from the results, the peak period appeared twice. The first time was from 9:00 to 11:00, when photovoltaic power began to play a role. When the wind and solar power were combined, the load generated not only met the peak power demand during working hours, but also carried out a short energy storage charging at 9 o'clock and energy storage discharge at 10 o'clock, greatly reducing the cost of purchasing electricity. The second time was from 18:00 to 20:00, during which the photovoltaic output gradually dropped to 0. Since the wind power output connected to scenario 4 reached 700kW, the output of the wind turbine during this period can meet the peak power demand at night. At the same time, the output of photovoltaic power during the day was charged with energy storage. Therefore, the peak at night can also be relieved by energy storage discharge, without the need to purchase electricity to meet the power demand.

[0213] During the off-peak period, wind power generation and gas turbines are mainly used to generate electricity load. When there is a power shortage during the off-peak period, the energy storage will be discharged; when there is excess wind power, the energy storage will be consumed in real time. Since the cost of purchasing electricity during the off-peak period is low, in this scenario, the virtual power plant only purchases and stores electricity in three time periods in the early morning to reduce the cost of electricity during peak hours.

[0214] During normal times, wind turbines and photovoltaics continue to output power. The 700kW wind power output at night is sufficient to maintain basic electricity demand. During the day, 300kW photovoltaic power combined with 700kW wind turbine output is sufficient to maintain basic electricity demand. The gas turbine output can be supplemented and the excess electricity can be stored.

[0215] (4) Scenario 4: Optimal scheduling of virtual power plants considering the uncertainty of wind and solar power and the difference in the number of wind power connections

[0216] Based on the results of scenario 3, the optimal dispatch of the virtual power plant is most economical when the wind power access capacity is 700kW and the photovoltaic access capacity is 300kW. When the number of wind turbines connected is 10, the uncertainty from the wind turbines that the virtual power plant needs to bear is the lowest. Therefore, when the access capacity of wind turbines remains unchanged, the economic feasibility of connecting one 700kW wind turbine (case 1) and connecting ten 70kW distributed wind turbines (case 2) should also be studied.

[0217] According to the MATLAB simulation results, the cost of case 1 is 15073.1307 yuan, and the cost of case 2 is 14905.5639 yuan. It is more economical to connect a virtual power plant with 10 70kW distributed wind turbines. The results of power optimization scheduling considering the uncertainty of wind and solar power and the difference in the number of wind power connected are as follows: Fig.12 shown.

[0218] As the number of distributed wind power connected increases, the uncertainty from wind turbines that the virtual power plant needs to bear is weaker. The energy storage optimization scheduling results considering the uncertainty of wind and solar power and the difference in the number of wind power connected are as follows: Fig.13 shown.

[0219] From the results, we can see that Fig.13 and Fig.11 In comparison, wind power generation during peak hours can be more fully dispatched. In the two time periods of 6:00-10:00 and 18:00-20:00, wind turbines have higher output. However, due to the volatility of wind power output, energy storage needs to play a greater role in this scenario, with more discharge time and capacity in the T period.

[0220] (5) Scenario 5: Optimal scheduling of virtual power plants considering the uncertainty of wind and solar power and the difference in the number of wind and solar power connections

[0221] Based on the results of scenario 3, the most economical situation is when the wind power access capacity is 700kW and the photovoltaic access capacity is 300kW; based on the results of scenario 4, when the number of wind turbines connected is 10, the uncertainty from wind turbines that the virtual power plant needs to bear is the lowest; when the number of 500kW photovoltaic units connected is 10 and 50, the uncertainty from photovoltaic units that the virtual power plant needs to bear is relatively low. Therefore, under the condition that the access capacity of wind and solar units and the number of wind power remain unchanged, the economic feasibility of connecting 10 30kW photovoltaic units (scenario 3) and connecting 50 6kW distributed wind turbines (scenario 4) should also be studied.

[0222] According to the MATLAB simulation results, the cost of case 3 is 14816.8444 yuan, and the cost of case 4 is 14815.7005 yuan, which shows that the virtual power plant with 50 6kW distributed photovoltaic units is more economical. The results of power optimization scheduling considering the uncertainty of wind and solar power and the difference in the number of wind and solar power units connected are as follows: Fig.14 As shown in the figure, the optimal dispatch result of energy storage considering the uncertainty of wind and solar power and the difference in the number of wind and solar power connected is as follows: Fig.15 shown.

[0223] As the number of distributed photovoltaic connections increases, the uncertainty from photovoltaic units that the virtual power plant needs to bear is weaker. From the results, it can be seen that Fig.15 and Fig.13 In comparison, the photovoltaic power generation at noon can be more fully dispatched, that is, in the two time periods of 11:00-12:00 and 13:00-14:00, the photovoltaic units are more motivated to participate in the virtual power plant dispatch. Similar to scenario 4, due to the volatility of photovoltaic output, energy storage also needs to play a big role in this scenario. The storage frequency is reduced in the T period, while the discharge frequency is higher.

[0224] (6) Scenario 6: Optimal scheduling of virtual power plants considering the uncertainty of wind, solar and controllable loads

[0225] Based on the results of sub-scenarios 3-5, when the virtual power plant is connected to 10 70KW distributed wind power equipment and 50 6kW distributed photovoltaic units, the virtual power plant has the lowest comprehensive cost and is the most economical. In this scenario, the impact of load-side uncertainty on the economic efficiency of the virtual power plant is further studied while keeping the access capacity and number of wind and solar units unchanged. The controllable loads in this scenario include curtailable loads and transferable loads. Curtailable loads include but are not limited to temperature-controlled loads such as air conditioners and water heaters in residents' homes, and transferable loads include but are not limited to loads with energy storage properties such as electric vehicles. The virtual power plant takes incentive measures for controllable loads. Users will receive electricity discounts during valley hours, while they will bear high electricity prices during peak hours. In this mode, the optimal scheduling of virtual power plants considering the uncertainty of wind, solar and controllable loads is studied. From the simulation results, it can be obtained that the comprehensive cost of the virtual power plant is 14406.3538 yuan, which shows that the incentive for controllable loads can adjust the user's electricity consumption habits and shift the peak electricity consumption. At the same time, the virtual power plant can achieve more economical operation. The results of power optimization dispatch considering the uncertainty of wind, solar and controllable loads are as follows: Fig.16 As shown in Figure 2, the energy storage optimization scheduling results are as follows: Fig.17 shown.

[0226] From the results, due to the existence of incentive policies, the transferable load and the curtailable load are mainly connected from 22:00 to 6:00 the next day, and the rest of the time is maintained at a stable output stage. Compared with the previous five scenarios, the nighttime load demand in this scenario increases, and the lack of photovoltaic output at night leads to more energy storage discharge during this period, and more electricity needs to be purchased for supplement. The peak period occurs twice a day, and the second peak occurs from 17:00 to 21:00. During this period, energy storage is charged when wind power output increases to meet the power demand of controllable loads at night.

[0227] In summary, in view of the uncertainty of different resource outputs, the virtual power plant in the present invention takes into account the uncertainty factors of each unit and designs relevant strategies for combining wind and solar resources and adjustable resources. First, the Monte Carlo simulation method based on Latin hypercube sampling and the scene distance reduction technology method are used to ensure full coverage of the overall space of wind and solar samples. On this basis, other uncertainties of wind and solar in addition to its own output volatility are studied. Secondly, taking the minimum comprehensive cost of virtual power plant operation as the objective function, an optimization scheduling model is established that takes into account uncertainty factors such as load forecasting, distributed new energy and generator power generation, component state operation, power market electricity price and load response, and 6 mode scenarios are designed to study the impact of the access capacity and number differences of different internal resources on the virtual power plant. The simulation results show that under the condition of considering the above-mentioned uncertainty factors, the virtual power plant is connected to 10 70KW distributed wind power equipment and 50 6kW distributed photovoltaic units, and the ratio of energy storage, gas turbines and wind and solar resources is kept at 15% and 25%, which can achieve the minimum comprehensive cost of the virtual power plant, thereby obtaining the best solution for the optimal scheduling of the virtual power plant.

[0228] It should be emphasized that the embodiments of the present invention are illustrative rather than restrictive. Therefore, the present invention includes but is not limited to the embodiments described in the specific implementation modes. Any other implementation modes derived by those skilled in the art based on the technical solutions of the present invention also fall within the scope of protection of the present invention.

Claims

1. A virtual power plant optimization scheduling method considering uncertain factors, Features: The following steps are involved: Step 1: Generate wind and solar uncertainty scenarios using the Monte Carlo simulation method based on Latin hypercube sampling to characterize the uncertainty characteristics of distributed wind power and photovoltaic output in the operation of virtual power plants; Step 2: Use the fast previous generation elimination technology based on probability distance to reduce the wind and solar uncertainty scenario simulated and generated in step 1; Step 3: Based on the uncertainty scenario of wind and solar power after curtailment in step 2, a virtual power plant optimization scheduling model considering uncertainty factors is constructed; Step 4: Based on the virtual power plant optimization scheduling model considering uncertainty factors established in step 3, design a variety of solution scenarios for the virtual power plant optimization scheduling model considering uncertainty factors, solve them, and finally select the best solution for virtual power plant optimization scheduling.

2. According to claim 1, a virtual power plant optimization scheduling method taking into account uncertain factors, Features: The specific steps of step 1 include: (1) First, collect wind power scenario sampling data, that is, determine the random variables such as wind speed and wind direction at different times required for wind power scenario generation, determine the uncertainty parameters of wind power and photovoltaic output, and define a reasonable parameter range for each parameter; (2) According to the parameter range defined in step (1), the data is divided into several equally probable self-controlled years or "layers" and sample points are generated. That is, according to the characteristics of the wind power scenario sampling data, the matrix of random numbers, sampling value probabilities and variable space is generated in sequence according to the Monte Carlo method, and sampling is performed according to the Latin hypercube sampling method to ensure that each layer is evenly sampled for constructing the wind power scenario; (3) Perform correlation verification on the random numbers, sampling value probabilities and variable space matrices generated in step (2) to ensure that the generated data conforms to the correlation of the actual wind power data; if the correlation condition is met, proceed to step (4); if not, return to step (2) and regenerate; (4) The newly generated data set in step (3) after correlation verification is formed into a wind power scenario sampling value matrix. This matrix is ​​a set of sample points generated by the Latin hypercube sampling method. Each row of the matrix represents a complete wind power output scenario, and each column represents a specific parameter. The matrix is ​​merged with the existing wind power scenario sampling value matrix to enrich the diversity of wind power scenarios. (5) Check whether the merged wind power scenario sampling value matrix meets the convergence condition; if so, proceed to step (6); if not, return to step (3) to continue to expand the sampling, indicating that the distribution and diversity of the sample points are not sufficient to fully cover the entire parameter space, and it is necessary to regenerate the sample points and verify the correlation until the convergence condition is met; (6) Stop expanding and generating the wind power scenario sampling value matrix, complete the wind and solar uncertainty scenario simulation generation, and organize the wind and solar uncertainty scenario data.

3. According to claim 1, a virtual power plant optimization scheduling method taking into account uncertain factors, Features: The specific steps of step 2 include: (1) Set the wind and solar uncertainty scenario set simulated and generated in step 1 as Y, and calculate the geometric distance between each pair of scenarios in the large-scale scenario set Y of wind turbine output and photovoltaic output; (2) Compare the geometric distances between the scenarios and select the wind and solar power output uncertainty scenario y with the smallest sum of probability distances to the remaining scenarios; (3) Select the scene in Y that is closest to scene y in geometric distance, set it as r, and make it replace scene y. Add the probability of y to the probability of scene r, eliminate d, and form a new s'; (4) Determine whether the number of remaining wind and solar power uncertain output scenarios meets the requirement. If it does, the scenario reduction is terminated. If it does not, repeat steps (1) to (3) to ultimately achieve the purpose of scenario reduction.

4. According to claim 1, a virtual power plant optimization scheduling method taking into account uncertain factors, Features: The specific steps of step 3 include: (1) Setting the objective function The objective function of the virtual power plant optimization scheduling problem taking into account uncertainties is as follows: In the formula, C Total is the comprehensive cost of the virtual power plant during a certain period of operation, T is the total number of periods in the entire dispatch cycle, and C WP,n (t), C PV,n (t), C GT,n (t), C ES,n (t), C buy (t) are wind turbine operation and maintenance costs, photovoltaic operation and maintenance costs, gas turbine operation and maintenance costs, energy storage operation and maintenance costs, and electricity purchase costs during period t; a. Fan operation and maintenance costs C WP,n (t) represents the cost required for the virtual power plant to call for wind power operation and maintenance during period t, and the calculation formula is: C WP,n (t)=η WP ×P WP,n (t) (2) Where η WP is the wind power operation coefficient, P WP,n (t) is the output forecast value of n units in the wind farm at time t; at the same time, since the output of new energy at each moment is affected by uncertain factors such as weather, there will be a certain output uncertainty. The present invention uses the mean and random deviation to represent the uncertainty of the output of new energy, which is expressed as: In the formula, is the mean output of wind power, ξ WP,t is the random deviation of wind power and photovoltaic power output; Combining formula (3), the cost of wind power operation is: b. Photovoltaic operation and maintenance costs C PV,n (t) represents the cost required for the virtual power plant to call for photovoltaic operation during period t, and the calculation formula is: C PV,n (t)=δ PV ×P PV,n (t) (5) In the formula, δ PV is the photovoltaic operation coefficient, P PV,n (t) is the output forecast value of n units at the photovoltaic station at time t; in addition, the volatility caused by uncertain factors must also be considered. Combining formula (6), we can get: In the formula, is the average output of photovoltaic power, ξ PV,t is the random deviation of photovoltaic output; The cost of photovoltaic operation is c. Cost of gas turbine C GT The cost of the gas turbine of the virtual power plant in period t is expressed as follows: In the formula, γ is the fixed startup cost, s GT,open is the startup state of the VPP gas turbine during period t, is the piecewise linearization cost, P GT,t is the engine output during period t, s GT,close In shutdown state, C GT,stop is the start-stop fee; n is the number of connected devices; the volatility caused by uncertain factors must also be considered; In the formula, is the mean output of the generator, ξ GT,t It is the random deviation of generator output; Combining formula (9), the cost of gas turbine operation is d. Cost of energy storage equipment C ES,n (t) represents the cost of calling the energy storage equipment for the virtual power plant during period t, and the calculation formula is: In the formula, p ES,sell (t) is the electricity price of the energy storage device at time t, λ out is the discharge efficiency of the energy storage device. μ(t) represents the charging and discharging state of the energy storage device. It is a 0-1 variable. μ(t) = 0 means that the energy storage device is charging, and the cost is 0 at this time. μ(t) = 1 means that the energy storage device is discharging. At this time, is the discharge power; e. Total cost of electricity purchase and sale C buy (t) represents the total cost of electricity purchase and sale by the virtual power plant in the day-ahead market during period t, and the calculation formula is: C buy (t)=C buy,t P buy,t (12) In the formula, C buy,t represents the electricity purchase and sale price of the virtual power plant in period t, P buy,t represents the amount of electricity purchased during period t; f. Controllable load cost C RL (t) and C TF (t) represents the cost of load reduction and load transfer in period t, and the calculation formula is: C RL (t)=C RL,t P RL,t (13) C TF (t)=C TF,t P TF,t (14) In the formula, C RL,t and C TF,t represents the unit cost of load that can be reduced and transferred at time t, P RL,t and P TF,t Represents the amount of load that can be reduced and transferred at time t; considering the volatility caused by uncertain factors, the load output can be reduced and transferable load output The following conditions are met: the mean is μ and the standard deviation is σ 2 The normal distribution of (2) Constructing constraints a. Power balance constraints The calculation formula of the electric energy balance constraint of the model of the present invention is: Where P PV (t), P wp (t), P buy (t), P GT (t) covers the uncertainty interval, L(t) is the electric load power at time t, and the specific calculation formula is: L(t)=Fload(t)+L RL (t)+L TF (t) (18) In the formula, the load power L(t) consists of three parts: the uncontrollable load Fload(t), the reducible load L RL (t) and transferable load L TF (t), the latter two take into account the uncertainty of controllable load; b. Fan output related constraints P WP (t) is the wind power output. In the previous research work, the Monte Carlo method was used to generate scenarios for wind power output, and the fast predecessor elimination technology based on probability distance was used to reduce the scenarios; The fan-related constraints are: P WP (t) min ≤P WP (t)≤P WP (t) max (19) Where P WP (t) max , P WP (t) min are the corresponding maximum and minimum power values ​​of the wind turbines respectively; c. Constraints on photovoltaic output P PV (t) is the photovoltaic output power. In the previous research work, the Monte Carlo method was used to generate scenarios for photovoltaic output, and the fast previous generation elimination technology based on probability distance was used to reduce the scenarios; The relevant constraints of the photovoltaic unit are: P PV (t) min ≤P PV (t)≤P PV (t) max (20) Where P PV (t) max , P PV (t) min They are the corresponding maximum and minimum power values ​​of the photovoltaic unit; d. Gas turbine related constraints i) Operational constraints The operating constraints of the gas turbine are: In the formula, and P GT,t are the corresponding maximum and minimum power values ​​of the gas turbine, T onGT,t-1 is the continuous operation time of the gas turbine, t onGT is the shortest startup time, T offGT,t-1 is the continuous downtime, t offGT Minimum downtime; ii) Hill climbing constraints The gas turbine unit is also subject to the constraint of ramp power, which is calculated as follows: In the formula, are the upward and downward climbing rates of the gas turbine unit, respectively; e. Energy storage charging and discharging constraints The constraints related to energy storage charging and discharging are: Where P ES,t , They are the minimum and maximum values ​​of the energy storage battery capacity respectively; is the discharge power, is the charging power; r min 、r max It is the minimum and maximum value of the corresponding climbing power when charging and discharging the energy of the storage; μ(t) represents the charging and discharging state of the energy storage device, which is a 0-1 variable. μ(t) = 0 represents the charging of the energy storage device, and μ(t) = 1 represents the discharging of the energy storage device. f. Controllable load constraints P RL,min ≤P RL,t ≤P RL,max (24) P TF,min ≤P TF,t ≤P TF,max (25) Where P RL,min , P RL,max They are the minimum and maximum values ​​of load that can be reduced respectively; P TF,min , P TF,max are the minimum and maximum values ​​of the transferable load respectively.

5. According to claim 1, a virtual power plant optimization scheduling method taking into account uncertain factors, Features: The specific method of step 4 is: The optimization problem of the virtual power plant optimal scheduling model considering uncertain factors is a 0-1 type mixed integer linear programming problem, and the standard form of its solution model is: In the formula, the optimization variable x includes the load forecast amount taking into account the uncertainty factors, the distributed renewable energy power generation, the generator output, the electricity price in the power market, the power purchase and sale, the load response and the component operation status; the equality constraint is the power balance constraint; the inequality constraint is the operation constraint of each device; p represents the variable subject to each inequality constraint in the planning problem, including the wind turbine output constraint, the photovoltaic output constraint, the operation and climbing constraint of the gas turbine, the energy storage charging and discharging constraint; the controllable load constraint; q represents the constraint condition that some unknown numbers in the planning problem are integers, including the start and stop of each unit; based on this, the present invention adopts the Gurobi solver to solve in MATLAB; Six model scenarios taking uncertainty into account are designed. By comparing the comprehensive costs of virtual power plants under the six scenarios, the effectiveness of virtual power plants in peak shaving and valley filling is reflected, and finally the best solution for optimizing the scheduling of virtual power plants is selected.

6. A virtual power plant optimization scheduling method taking uncertainty factors into account according to claim 5, Features: The solution scenarios of the virtual power plant optimization scheduling model considering uncertainty factors in step 4 include: Scenario 1: Optimal dispatch of virtual power plants considering wind power uncertainty In this scenario, the optimal dispatch of the virtual power plant considering the uncertainty of wind power is input with different wind power capacity values ​​ranging from 200kW to 900kW; Scenario 2: Optimal dispatch of virtual power plants considering PV uncertainty In this scenario, the optimal dispatch of the virtual power plant considering PV uncertainty is input with different PV capacity values ​​ranging from 200kW to 900kW; Scenario 3: Optimal dispatch of virtual power plants considering the uncertainty of wind power and photovoltaic power and different ratios of wind power and photovoltaic power In this scenario, the virtual power plant optimizes the dispatching of wind power and photovoltaic power with comprehensive consideration of the uncertainty, with different capacities and ratios of wind power and photovoltaic power ranging from 200kW to 800kW; Scenario 4: Optimal scheduling of virtual power plants considering the uncertainty of wind and solar power and the difference in the number of wind power connections In this scenario, the optimal dispatch of virtual power plants with comprehensive consideration of wind and solar uncertainty is studied, with 1 to 50 distributed wind turbines of the same capacity input, to study the impact of differences in the number of wind power access on the economic efficiency of virtual power plants; Scenario 5: Optimal scheduling of virtual power plants considering the uncertainty of wind and solar power and the difference in the number of wind and solar power connections In this scenario, the optimal scheduling of virtual power plants is considered by comprehensively considering the uncertainty of wind and solar power and the number of wind turbines connected. The impact of the difference in the number of photovoltaic connections on the economic efficiency of virtual power plants is studied by inputting 1 to 50 distributed photovoltaic units of the same capacity. Scenario 6: Optimal scheduling of virtual power plants considering the uncertainty of wind, solar and controllable loads In this scenario, the optimal scheduling of the virtual power plant, which takes into account the uncertainties of wind power and photovoltaic power, will include controllable load inputs that can be curtailed and transferred.

Images