Multi-time scale optimization scheduling method and system for virtual power plants

Abstract

The present invention discloses a multi-time scale optimization scheduling method for a virtual power plant, including obtaining data information of a target power system and a virtual power plant; modeling an incentive price-type demand response; processing the uncertainty of intraday wind power generation, intraday photovoltaic power generation, intraday electricity price and intraday user load; constructing a multi-time scale optimization scheduling model for the day before and within the day of the virtual power plant and solving it to obtain a virtual power plant scheduling plan; controlling the operation of the virtual power plant according to the virtual power plant scheduling plan to complete the multi-time scale optimization scheduling for the virtual power plant. The present invention also discloses a system for implementing the multi-time scale optimization scheduling method for a virtual power plant. The present invention establishes an optimization scheduling model for a virtual power plant, combines it with an incentive price-type demand response, constructs a final multi-time scale optimization scheduling model and solves it, thereby completing the multi-time scale optimization scheduling of the virtual power plant; therefore, the present invention has high reliability, good stability and high accuracy.

Description

Technical Field

[0001] The present invention belongs to the field of electrical automation technology, and in particular relates to a multi-time scale optimization scheduling method and system for a virtual power plant. Background Art

[0002] With the development of economy and technology and the improvement of people's living standards, electricity has become an indispensable secondary energy source in people's production and life, bringing endless convenience to people's production and life. Therefore, ensuring the stable and reliable supply of electricity has become one of the most important tasks of the power system.

[0003] At present, more and more renewable energy power generation systems are connected to the power grid. The strong volatility and geographical dispersion of renewable energy power generation systems have brought great challenges to the stable and reliable operation of the existing power system. After the renewable energy power generation systems are connected to the power grid, the proportion of traditional regulating power sources in the power system decreases, which in turn increases the gap in regulating resources in the power system, increases the regulation cost of the power system, and makes the regulation of the power system more difficult.

[0004] The introduction of Virtual Power Plant (VPP) technology provides a solution for the operation of the power system after the large-scale renewable energy power generation system is connected to the power grid. As a bridge between the power system and distributed energy, the virtual power plant can ultimately achieve resource integration and peak load shifting of the power system through the control of flexible resources, electric vehicle loads, controllable loads and energy storage equipment in the power system, thereby ensuring the stable and reliable operation of the power system.

[0005] However, the current control schemes for virtual power plants still have defects such as poor control accuracy and many uncertain factors in the control process. This causes many problems in the actual application of the current virtual power plant technology, which in turn affects the stable and reliable operation of the power system. Summary of the invention

[0006] One of the purposes of the present invention is to provide a multi-time scale optimization scheduling method for virtual power plants with high reliability, good stability and high accuracy.

[0007] A second objective of the present invention is to provide a system for implementing the multi-time scale optimization scheduling method for virtual power plants.

[0008] The multi-time scale optimization scheduling method for a virtual power plant provided by the present invention comprises the following steps:

[0009] S1. Obtain data information of the target power system and virtual power plant;

[0010] S2. Modeling of incentive electricity price demand response based on the data information obtained in step S1;

[0011] S3. According to the data information obtained in step S1, the uncertainty of intraday wind power generation, intraday photovoltaic power generation, intraday electricity price and intraday user load is processed;

[0012] S4. Based on the data information obtained in step S2 and step S3, a multi-time scale optimization scheduling model of the virtual power plant is constructed;

[0013] S5. Solve the multi-time scale optimization scheduling model constructed in step S4 to obtain the final virtual power plant scheduling plan;

[0014] S6. Use the virtual power plant dispatching plan obtained in step S5 to control the operation of the virtual power plant and complete the multi-time scale optimization dispatching for the virtual power plant.

[0015] The step S2 of modeling the incentive electricity price demand response based on the data information obtained in step S1 specifically includes the following steps:

[0016] In the intraday stage, the incentive electricity price mechanism is used on the user side, and the electricity price elasticity matrix is ​​introduced for modeling, which is expressed as Where L is the electricity price elasticity coefficient, ΔP load_pdr is the load reduction amount of price-based demand response, c e_0 is a fixed electricity price, P load_p is the load before response, Δc e is the difference between the electricity price after demand response and the fixed electricity price;

[0017] According to the incentive electricity price mechanism, the following electricity quantity and price elasticity matrix L is constructed: ep :

[0018]

[0019] The electricity price elasticity matrix L ep The element L in ij Expressed as

[0020]

[0021] Where ΔP load_pdri is the load reduction at time i, c e_0i is the fixed electricity price at time i, P load_pi is the original load at time i, Δc ei is the change in electricity price at time i after demand response, c e_0j is the fixed electricity price at time j, Δc ej is the change in electricity price at time j after demand response; the electricity quantity and price elasticity matrix Lep The element L in ii represents the self-elastic coefficient at time i, L ij represents the cross elasticity coefficient between time i and time j;

[0022] Load reduction after demand response Expressed as

[0023] Step S3, based on the data information obtained in step S1, processes the uncertainty of intra-day wind power generation, intra-day photovoltaic power generation, intra-day electricity price and intra-day user load, and specifically includes the following steps:

[0024] For intraday wind power generation, intraday photovoltaic power generation, intraday electricity price and intraday user load, the following steps are used to handle uncertainty:

[0025] Data expansion: In a scheduling cycle, there are 24 data in the day-ahead phase and 96 data in the intraday phase; therefore, the following formula is used for expansion:

[0026] D pre_q (i) = {D pre_n (4i-3),D pre_n (4i-2),D pre_n (4i-1),D pre_n (4i)}

[0027] Where D pre_q (i) is the i-th initial data in the day-ahead data; D pre_n (4i) is the 4ith initial data in the intraday data, i is a positive integer ranging from 1 to 24; there are 4*24 initial data after expansion;

[0028] Monte Carlo method is used to select the specific standard deviation s within the day n , generate a standard normal distribution σ of random numbers, and then generate 250 specific original scenes to be cut;

[0029] The specific original scenes to be cut are cut using the Manhattan probability distance to obtain 10 scenes after the cut and the corresponding scene probabilities;

[0030] According to the output value and corresponding probability of the reduced scenario, the following formula is used to combine and obtain the final output curve:

[0031] S pre_n =S pre_n1 ×P pre_n1 +S pre_n2 ×P pre_n2 +....+S pre_n10 ×P pre_n10

[0032] Where S pre_n It is the output curve generated after the intraday stage processing; S pre_n1 ~S pre_n10 is the output value of the first to tenth scenes after reduction; P pre_n1 ~P pre_n10 The probabilities corresponding to the first to tenth scenes after reduction.

[0033] According to the data information obtained in step S2 and step S3, step S4 constructs a multi-time scale optimization scheduling model of the virtual power plant on the day before and within the day, which specifically includes the following steps:

[0034] Day-ahead stage: Dispatch is carried out 24 hours in advance, with the dispatch time scale set to 1 hour. Based on wind and solar power output, load forecast, energy storage adjustable output, incentive-based demand response parameters and system parameters, the virtual power plant dispatch plan for the next 24 hours is formulated with the minimum total cost of the virtual power plant as the objective function, combined with the purchase and sale electricity price and market carbon price;

[0035] Intraday stage: scheduling is carried out 4 hours in advance, with a time scale of 15 minutes. The calculation is rolled over every 15 minutes, and a scheduling plan for the next 4 hours is formed each time. The operating status of the gas turbine is determined by the result of the day-ahead optimization. Taking the minimum sum of the penalty caused by the intraday and day-ahead power deviation and the total cost of the virtual power plant as the objective function, formulate the daily output of each unit, daily power purchase and sale, and daily energy storage charging and discharging plans.

[0036] The objective function is to minimize the sum of the penalty caused by the intraday and day-ahead power deviations and the total cost of the virtual power plant, and specifically includes the following steps:

[0037] The following formula is used as the objective function:

[0038]

[0039] In the formula It is the sum of the penalty caused by the intraday and day-ahead power deviation and the total cost of the virtual power plant; Operation and maintenance costs for wind power, photovoltaic power and energy storage; For gas turbine operating costs; The cost of purchasing and selling electricity for virtual power plants; for the cost of carbon emissions; is the demand response cost of the intraday incentive electricity price, and T is the scheduling period, is the electricity price before demand response, is the load after demand response at time t and is the load before demand response, is the load reduction amount, is the change in electricity price, Δt in is the intraday time scale; is the penalty cost for output deviation, and is the deviation of the gas turbine daily dispatch value compared with the day-ahead plan and is the value of the i-th gas turbine at time t within the day, P Gi (t) is the optimization result of the i-th gas turbine at time t a day ago, is the deviation of the battery’s daily dispatch value compared to the day-ahead plan and is the value of the ith battery at time t within the day, P bat (t) is the optimization result of the i-th battery at time t, is the deviation of the daily dispatch value of the power grid transaction power compared with the day-ahead plan and is the value of the i-th power grid transaction electricity at time t within the day, P net (t) is the optimization result of the i-th power grid transaction electricity at time t a day ago, Γ GT is the gas turbine deviation penalty coefficient, Γ bat is the battery deviation penalty coefficient, Γ net It is the penalty coefficient for the deviation of power purchase and sale transaction;

[0040] The following formula is used as the constraint condition:

[0041] Power balance constraints: In the formula It is the intraday rolling optimization value of the power grid transaction electricity; It is the intraday rolling optimization value of wind power generation; The intraday rolling optimization value for solar power generation; It is the intraday rolling optimization value of the gas turbine; is the load after demand response at time t; It is the daily rolling optimization value of the battery;

[0042] Gas turbine start and stop constraints:

[0043]

[0044] In the formula is the startup state of the i-th gas turbine at time t within the day, δ Gsi (t) is the startup status of the i-th gas turbine at time t, is the stop state of the i-th gas turbine at time t within the day, δ Gti (t) is the stop state of the i-th gas turbine at time t.

[0045] Step S5 solves the multi-time scale optimization scheduling model constructed in step S4, specifically including the following steps:

[0046] The YALMIP toolbox in matlabR2018a is used to build the model, and the CPLEX12.8.0 solver is used to solve the problem. The specific steps are as follows:

[0047] Data input: input the acquired data information into the solver;

[0048] The hourly data is processed and combined with the input data information to build a multi-time scale optimization scheduling model for the virtual power plant on the day before and within the day;

[0049] Solve the day-ahead phase of the multi-time-scale optimization dispatch model of the constructed virtual power plant to obtain the day-ahead output plan;

[0050] In the intraday stage, ultra-short-term forecasts are made for wind and solar power output, electricity prices, and user load data to obtain intraday optimized dispatch data, which is updated at the set time.

[0051] Combine the day-ahead dispatch reference value and intraday constraints, add price-based demand response, perform rolling optimization according to the set update time, and use the obtained results as the correction of the day-ahead dispatch plan;

[0052] Repeat the above steps to solve the constructed multi-time scale optimization scheduling model in real time.

[0053] The present invention also provides a system for implementing the multi-time scale optimization scheduling method for virtual power plants, including a data acquisition module, a demand response modeling module, an uncertainty processing module, an optimization scheduling modeling module, a model solving module and an operation module; the data acquisition module, the demand response modeling module, the uncertainty processing module, the optimization scheduling modeling module, the model solving module and the operation module are connected in series in sequence; the data acquisition module is used to acquire data information of the target power system and the virtual power plant, and upload the data to the demand response modeling module; the demand response modeling module is used to model the incentive electricity price type demand response according to the received data, and upload the data to the uncertainty processing module; the uncertainty processing module is used to The module processes the uncertainties of intraday wind power generation, intraday photovoltaic power generation, intraday electricity price and intraday user load according to the received data, and uploads the data to the optimization scheduling modeling module; the optimization scheduling modeling module is used to construct a multi-time scale optimization scheduling model of the virtual power plant within the day before according to the received data, and uploads the data to the model solving module; the model solving module is used to solve the constructed multi-time scale optimization scheduling model according to the received data, obtain the final virtual power plant scheduling plan, and upload the data to the operation module; the operation module is used to control the operation of the virtual power plant according to the received data and adopt the obtained virtual power plant scheduling plan to complete the multi-time scale optimization scheduling for the virtual power plant.

[0054] The multi-time-scale optimization scheduling method and system for virtual power plants provided by the present invention establish an optimization scheduling model for the virtual power plant. On the basis of ensuring the operational reliability and safety of the virtual power plant, combined with incentive electricity price-type demand response, the final multi-time-scale optimization scheduling model of the virtual power plant is constructed and solved, and finally the multi-time-scale optimization scheduling of the virtual power plant is completed. Therefore, the present invention not only realizes the multi-time-scale optimization scheduling of the virtual power plant, but also has high reliability, good stability and high accuracy. BRIEF DESCRIPTION OF THE DRAWINGS

[0055] Figure 1 The figure is a schematic diagram of the method flow of the present invention.

[0056] Figure 2 A schematic diagram is generated for an intraday scenario in the method of the present invention.

[0057] Figure 3 It is a schematic diagram of intra-day scene reduction in the method of the present invention.

[0058] Figure 4 Schematic diagram of the probability corresponding to the intraday reduction scenario in the method of the present invention.

[0059] Figure 5 This is a schematic diagram of the scenario-based daytime wind and solar output and user prediction according to an embodiment of the method of the present invention.

[0060] Figure 6 This is a schematic diagram of the prediction of the purchase and sale electricity price of a virtual power plant within a day according to the scenario method of an embodiment of the method of the present invention.

[0061] Figure 7 This is a schematic diagram of comparing the optimized gas turbine outputs within the day before according to an embodiment of the method of the present invention.

[0062] Figure 8 Schematic diagram of comparison of optimized energy storage output within the day before according to an embodiment of the method of the present invention.

[0063] Fig. 9 This is a schematic diagram showing a comparison of the amount of electricity purchased and sold by a virtual power plant optimized within the day before according to an embodiment of the method of the present invention.

[0064] Fig.10 Schematic diagram of the functional modules of the system of the present invention. DETAILED DESCRIPTION

[0065] like Figure 1 The method flow diagram of the method of the present invention is shown as follows: The multi-time scale optimization scheduling method for a virtual power plant provided by the present invention comprises the following steps:

[0066] S1. Obtain data information of the target power system and virtual power plant;

[0067] S2. Modeling of incentive electricity price demand response is performed based on the data information obtained in step S1; specifically, the steps include:

[0068] In the intraday stage, the amount of user data is greater than that in the day-ahead stage, and the accuracy is also higher, so the uncertainty is greater. It is more suitable to adopt incentive electricity price demand response for user load regulation. This method belongs to price-based demand response. On the user side, the incentive electricity price mechanism is used to introduce the electricity price elasticity matrix for modeling.

[0069] On the user side, the incentive electricity price mechanism is used and the electricity price elasticity matrix is ​​introduced for modeling, which is expressed as Where L is the electricity price elasticity coefficient, ΔP load_pdr is the load reduction amount of price-based demand response, c e_0 is a fixed electricity price, P load_p is the load before response, Δc e is the difference between the electricity price after demand response and the fixed electricity price;

[0070] Since the incentive electricity price is introduced, the user-side electricity price after demand response is the incentive electricity price. According to the incentive electricity price mechanism, the following electricity quantity and price elasticity matrix L is constructed: ep , used to describe the relationship between electricity quantity and electricity price:

[0071]

[0072] The electricity price elasticity matrix L ep The element L in ij Expressed as

[0073]

[0074] Where ΔP load_pdri is the load reduction at time i, c e_0i is the fixed electricity price at time i, P load_pi is the original load at time i, Δc ei is the change in electricity price at time i after demand response, c e_0j is the fixed electricity price at time j, Δc ej is the change in electricity price at time j after demand response; the electricity quantity and price elasticity matrix L ep The element L in ii represents the self-elastic coefficient at time i, L ij represents the cross elasticity coefficient between time i and time j;

[0075] Load reduction after demand response Expressed as

[0076] S3. According to the data information obtained in step S1, the uncertainty of intraday wind power generation, intraday photovoltaic power generation, intraday electricity price and intraday user load is processed; specifically, the following steps are included:

[0077] The initial data processing for the day before and within the day adopts the scenario generation and reduction method based on Monte Carlo and Manhattan probability distance. The difference is that the initial data dimension in the intraday optimization scheduling is different from that of the day before, and the data accuracy requirement for the day is higher. Therefore, improvements need to be made when using the scenario generation and reduction method;

[0078] For intraday wind power generation, intraday photovoltaic power generation, intraday electricity price and intraday user load, the following steps are used to handle uncertainty (the following steps take wind power output as an example):

[0079] Data expansion: In a scheduling cycle, there are 24 data in the day-ahead phase and 96 data in the day-ahead phase; therefore, each data in the day-ahead phase is equivalent to the value of 4 data in the day; the following formula is used for expansion:

[0080] D pre_q (i) = {D pre_n (4i-3),D pre_n (4i-2),D pre_n (4i-1),D pre_n (4i)}

[0081] Where D pre_q(i) is the i-th initial data in the day-ahead data; D pre_n (4i) is the 4ith initial data in the intraday data, i is a positive integer ranging from 1 to 24; there are 4*24 initial data after expansion;

[0082] Monte Carlo method is used to select the specific standard deviation s within the day n , generate a standard normal distribution σ of random numbers, and then generate 250 specific original scenes to be cut, such as Figure 2 As shown;

[0083] The specific original scenes to be cut are cut using the Manhattan probability distance to obtain 10 scenes after cutting and the corresponding scene probabilities, as shown in Figure 3 and Figure 4 As shown;

[0084] According to the output value and corresponding probability of the reduced scenario, the following formula is used to combine and obtain the final output curve:

[0085] S pre_n =S pre_n1 ×P pre_n1 +S pre_n2 ×P pre_n2 +....+S pre_n10 ×P pre_n10

[0086] Where S pre_n It is the output curve generated after the intraday stage processing; S pre_n1 ~S pre_n10 is the output value of the first to tenth scenes after reduction; P pre_n1 ~P pre_n10 The probabilities corresponding to the first to tenth scenes after reduction;

[0087] S4. Based on the data information obtained in step S2 and step S3, a multi-time scale optimization scheduling model of the virtual power plant is constructed; specifically, the following steps are included:

[0088] In the multi-time scale optimization dispatch model of virtual power plants, virtual power plants aggregate wind turbines, gas turbines, energy storage and user-side flexible loads, and also participate in electricity carbon market transactions. The difference is that the virtual power plant coordination control center adopts price-based demand response to reduce user loads. The uncertainty of wind, solar and electricity prices is expanded to 96 data when generating and reducing scenarios. The carbon price and demand response basic data take the middle value of the interval number. In order to establish the long-time scale optimization dispatch of virtual power plants and the short-time scale rolling optimization dispatch of virtual power plants within the day, the reasonable application of the forecast information within the day and different time scales should be divided into the following two stages:

[0089] Day-ahead stage: Dispatch is carried out 24 hours in advance, with the dispatch time scale set to 1 hour. Based on wind and solar power output, load forecast, energy storage adjustable output, incentive-based demand response parameters and system parameters, the virtual power plant dispatch plan for the next 24 hours is formulated with the minimum total cost of the virtual power plant as the objective function, combined with the purchase and sale electricity price and market carbon price;

[0090] Intraday stage: dispatching is carried out 4 hours in advance, with a time scale of 15 minutes, and rolling calculation is carried out every 15 minutes. Each time, a dispatching plan for the next 4 hours is formed. The operating status of the gas turbine is determined by the result of the day-ahead optimization. The objective function is to minimize the sum of the penalty caused by the intraday and day-ahead power deviation and the total cost of the virtual power plant, and formulate the daily output of each unit, the daily purchase and sale of electricity, and the daily energy storage charging and discharging plan. In specific implementation, the following formula is used as the objective function:

[0091]

[0092] In the formula It is the sum of the penalty caused by the intraday and day-ahead power deviation and the total cost of the virtual power plant; Operation and maintenance costs for wind power, photovoltaic power and energy storage; For gas turbine operating costs; The cost of purchasing and selling electricity for virtual power plants; for the cost of carbon emissions; is the demand response cost of the intraday incentive electricity price, and T is the scheduling period, is the electricity price before demand response, is the load after demand response at time t and is the load before demand response, is the load reduction amount, is the change in electricity price, Δt in is the intraday time scale; is the penalty cost for output deviation, and is the deviation of the gas turbine daily dispatch value compared with the day-ahead plan and is the value of the i-th gas turbine at time t within the day, P Gi (t) is the optimization result of the i-th gas turbine at time t a day ago, is the deviation of the battery’s daily dispatch value compared to the day-ahead plan and is the value of the ith battery at time t within the day, P bat (t) is the optimization result of the i-th battery at time t, is the deviation of the daily dispatch value of the power grid transaction power compared with the day-ahead plan and is the value of the i-th power grid transaction electricity at time t within the day, P net (t) is the optimization result of the i-th power grid transaction electricity at time t a day ago, Γ GT is the gas turbine deviation penalty coefficient, Γ bat is the battery deviation penalty coefficient, Γ net It is the penalty coefficient for the deviation of power purchase and sale transaction;

[0093] In the specific implementation, all parameters within the day are deterministic parameters, and the interval variables of the day-ahead take the midpoint value of the interval within the day, and there are no interval variables; therefore, all variables in the day-ahead function and constraints are deterministic values ​​in the intraday optimization scheduling;

[0094] The following formula is used as the constraint condition:

[0095] The constraints of the intraday rolling optimization scheduling include gas turbine constraints, power balance constraints, virtual power plant power purchase and sales constraints, and energy storage constraints. Since the day-ahead scheduling has determined the gas turbine start and stop plan, the day-ahead schedule is followed. Since the demand response adopts a price-based approach, its power balance constraints have changed.

[0096] Power balance constraints: In the formula It is the intraday rolling optimization value of the power grid transaction electricity; It is the intraday rolling optimization value of wind power generation; The intraday rolling optimization value for solar power generation; It is the intraday rolling optimization value of the gas turbine; is the load after demand response at time t; It is the daily rolling optimization value of the battery;

[0097] Gas turbine start and stop constraints:

[0098]

[0099] In the formula is the startup state of the i-th gas turbine at time t within the day, δ Gsi (t) is the startup status of the i-th gas turbine at time t, is the stop state of the i-th gas turbine at time t within the day, δ Gti (t) is the stop state of the i-th gas turbine at time t on the previous day;

[0100] S5. Solve the multi-time scale optimization scheduling model constructed in step S4 to obtain the final virtual power plant scheduling plan; specifically, the steps include:

[0101] Perform rolling optimization on multiple time scales: The short-time rolling optimization on a daily basis is different from the long-time optimization on a day-ahead basis. The long-time optimization is based on the overall optimum and can obtain the result by calculating and solving the mathematical model once. However, the consequence is that the actual situation cannot be considered during the scheduling process, and the result will be biased. The rolling optimization on multiple time scales on a daily basis is based on the results of some variables determined in the day-ahead scheduling plan and is continuously updated on a short-time scale. Each rolling is repeated with the updated source-load information to obtain the optimal solution. The system is dynamically optimized and the previous poor solution is adjusted in real-time tracking to ensure that the scheduling strategy is presented in the best form. The scheduling result has better reliability and adaptability than the result of one-time optimization scheduling, and can reduce the uncertainty caused by system modeling, external interference, prediction error, etc. and its impact on the scheduling result.

[0102] The YALMIP toolbox in matlabR2018a is used to build the model, and the CPLEX12.8.0 solver is called for solving the problem. The solution process consists of two stages, namely the long-time scale model solution of the day-ahead and the short-time scale rolling optimization solution of the day. The results of the day-ahead scheduling are used as guidance for the day, and the day-ahead scheduling is used to correct the day-ahead scheduling results through real-time forecast data. The specific steps are as follows:

[0103] Data input: input the acquired data information into the solver;

[0104] The hourly data is processed and combined with the input data information to build a multi-time scale optimization scheduling model for the virtual power plant on the day before and within the day;

[0105] Solve the day-ahead phase of the multi-time-scale optimization dispatch model of the constructed virtual power plant to obtain the day-ahead output plan;

[0106] In the intraday stage, ultra-short-term forecasts are made for wind and solar power output, electricity prices, and user load data to obtain intraday optimized dispatch data, which is updated at the set time.

[0107] Combine the day-ahead dispatch reference value and intraday constraints, add price-based demand response, perform rolling optimization according to the set update time, and use the obtained results as the correction of the day-ahead dispatch plan;

[0108] Repeat the above steps to solve the constructed multi-time scale optimization scheduling model in real time;

[0109] S6. Use the virtual power plant dispatching plan obtained in step S5 to control the operation of the virtual power plant and complete the multi-time scale optimization dispatching for the virtual power plant.

[0110] The method of the present invention is further described below in conjunction with an embodiment:

[0111] A small virtual power plant system in East China is taken as the research object. The virtual power plant includes gas turbines, wind turbines, energy storage and user-side flexible loads. The main scheduling parameters of the example on the date are shown in Table 1:

[0112] Table 1 Schematic diagram of scheduling technical parameters within the previous day

[0113]

[0114] In addition, when the intraday deviation is compensated by the purchase and sale of electricity by the virtual power plant, the intraday real-time electricity price settlement is adopted, and the fixed electricity price c before the price-based demand response is executed e_0 Take 0.78 yuan / kwh, the standard deviation of wind and solar fluctuations in the scene generation and reduction method used is s ws_in Take 25%, the standard deviation of electricity price fluctuation s pr_in Take 55% and generate a forecast data every 15 minutes; the system incentive electricity price demand response and penalty parameters are shown in Table 2;

[0115] Table 2 Schematic diagram of demand response and penalty parameters

[0116] parameter Numeric parameter Numeric <![CDATA[L ii ]]> -0.1 <![CDATA[L ij ]]> 0.03 <![CDATA[Γ GT ]]> 1.5 <![CDATA[Γ bat ]]> 1.2 <![CDATA[Γ net ]]> 1.8

[0117] The load-side incentive electricity price parameters are shown in Table 3:

[0118] Table 3 Load power supply incentive electricity price diagram

[0119] Time Electricity price / (yuan / kWh) 00:00-7:00,22:00-24:00 0.35 7:00-9:00,12:00-17:00,21:00-22:00 0.66 9:00-12:00,17:00-21:00 1.1

[0120] The wind and solar power prices and electricity purchase and sales prices generated by the daily load forecast are as follows Figure 5 and Figure 6 As shown;

[0121] Analysis of the advantages of scene generation and reduction method:

[0122] In order to verify the advantages of the scene generation and reduction method based on Monte Carlo and Manhattan probability distance proposed in this paper in dealing with uncertainty, this section introduces the root mean square error RMSE and mean absolute percentage error MAPE to quantify the data prediction effect. The specific meaning of RMSE is In the formula, n is the amount of data, m is i is the predicted value, g(z i ) is true;

[0123] Based on the processing results of wind turbine output, the method proposed in the present invention is compared with the traditional k-means method and Wasserstain probability distance method through corresponding indicators. The results are shown in Table 4:

[0124] Table 4 Data processing deviation comparison diagram

[0125]

[0126] From the data analysis in the table, it can be seen that in the day-ahead stage, the time scale is 1h, each group has 24 data, and the RMSE and MAPE corresponding to the three methods are not much different in value. The maximum difference in RMSE for day-ahead data processing is 5.185%, and the maximum difference in MAPE is 4.930%, indicating that the three methods have equivalent performance in processing uncertainty in day-ahead data processing, and can achieve the expected effect in actual use; but in the intraday stage, the time scale is 15min, each group has 96 data, and the RMSE of the method proposed in the present invention is 6.012, which is 9.81% and 8.11% higher than the k-means method and Wasserstain probability distance method; MAPE is 3.534%, which is 8.11% and 11.34% higher than the k-means method and Wasserstain probability distance method. Therefore, for intraday data processing, the method proposed in the present invention has better performance than the other two methods. In summary, it can be inferred that the scene generation and reduction method proposed in the present invention is more suitable for situations with large amounts of data and higher requirements for data.

[0127] Analysis of virtual power plant intraday rolling optimization results:

[0128] The virtual power plant's intraday optimization scheduling is updated every 15 minutes. In order to reflect the optimization results under different time scales and the correction effect of the intraday on the day before, the day-ahead scheduling results are compared. The results are as follows: Figure 7 to Figure 9 As shown;

[0129] The day-ahead optimization dispatch results are too ideal because they are based on hourly time scales, which may make the system unable to respond to fluctuations caused by renewable energy output and users in a timely manner. Due to the shortened time scale within the day, the use of rolling optimization dispatch can further reduce the system errors caused by the uncertainty of wind, solar, electricity prices and user loads, making the dispatch results more practical. Figure 7 to Figure 9It can be seen that the overall changing trends of gas turbine output, battery charging and discharging, and virtual power plant power purchase and sales remain unchanged during the intraday scheduling stage, especially the virtual power plant power purchase and sales and gas turbine output. The deviation basically occurs at the output peak and valley value. Combined with the changes in user load, it means that when the system bears a large load, the virtual power plant will make real-time corrections and adjustments to the output of each unit during the day according to the overall system situation to meet the overall operation goals of the system; and the overall daily adjustment of energy storage is larger than that of gas turbines and power purchase and sales, and its main adjustment periods are 6:00, 12:00 and 20:00, which verifies that energy storage, as a resource with rapid response capabilities, can give full play to its flexibility when facing the actual situation during the day, respond to changes in wind, light and load in real time during the intraday stage, smooth out the fluctuations of the entire virtual power plant system, and ensure the smooth operation of the system.

[0130] Comparative analysis of different strategies:

[0131] In order to verify the advantages of the multi-time scale scheduling strategy considering price-based demand response and output deviation proposed in the present invention, the following scenarios are established for comparative analysis:

[0132] Scenario 1: Only the day-ahead optimization scheduling strategy is used. All optimized results are expressed as the middle value of the interval number, without considering the deviation caused by the actual situation during the day.

[0133] Scenario 2: DA-P strategy: The intraday dispatch deviation caused by the forecast error is completely borne by the power grid purchase and sales, and the output of other units in the virtual power plant is not corrected.

[0134] Scenario 3: (The method of the present invention) Based on the day-ahead plan, price-based demand response and output deviation are considered, and the output of the internal units of the virtual power plant is continuously optimized and corrected on a fixed time scale.

[0135] The comparison results of different scenarios are shown in Table 5:

[0136] Table 5 Cost comparison in different scenarios

[0137]

[0138] Comparing the results of scenario 1 and scenario 3, it can be seen that due to the consideration of the penalty cost of output deviation in intraday optimization, the total cost of the day is slightly increased relative to the total cost of the day-ahead, by about 1.28%. However, since only the day-ahead optimization scheduling is adopted, many deviations caused by actual conditions are ignored, and the model accuracy is insufficient. Therefore, for the actual conditions, the multi-time scale optimization scheduling strategy proposed in the present invention improves the accuracy of the model and enhances the practical significance of the model while only increasing the cost by 1.28%, thus proving the advantages of the scheduling strategy proposed in the present invention; comparing scenario 2 and scenario 3, it can be seen that if the output of other units is not corrected in a rolling manner during the intraday stage, relying solely on electricity trading to maintain power balance will increase costs by about 2.8%, and the economic performance will deteriorate. At the same time, relying solely on electricity trading cannot give full play to the flexible adjustment function of the energy storage unit. Therefore, the advantages of the method of the present invention are verified.

[0139] like Fig.10 The figure shows a schematic diagram of the functional modules of the system of the present invention: the system disclosed in the present invention for realizing the multi-time scale optimization scheduling method for virtual power plants comprises a data acquisition module, a demand response modeling module, an uncertainty processing module, an optimization scheduling modeling module, a model solving module and an operation module; the data acquisition module, the demand response modeling module, the uncertainty processing module, the optimization scheduling modeling module, the model solving module and the operation module are connected in series in sequence; the data acquisition module is used to acquire data information of the target power system and the virtual power plant, and upload the data to the demand response modeling module; the demand response modeling module is used to model the incentive electricity price demand response according to the received data, and upload the data to the uncertainty processing module; The deterministic processing module is used to process the uncertainty of intraday wind power generation, intraday photovoltaic power generation, intraday electricity price and intraday user load according to the received data, and upload the data to the optimization scheduling modeling module; the optimization scheduling modeling module is used to construct a multi-time scale optimization scheduling model of the virtual power plant within the day according to the received data, and upload the data to the model solving module; the model solving module is used to solve the constructed multi-time scale optimization scheduling model according to the received data, obtain the final virtual power plant scheduling plan, and upload the data to the operation module; the operation module is used to control the operation of the virtual power plant according to the received data, using the obtained virtual power plant scheduling plan, and complete the multi-time scale optimization scheduling for the virtual power plant.

Claims

1. A multi-time scale optimization scheduling method for a virtual power plant, comprising the following steps: S1. Obtain data information of the target power system and virtual power plant; S2. Modeling of incentive electricity price demand response is performed based on the data information obtained in step S1; specifically, the steps include: In the intraday stage, the incentive electricity price mechanism is used on the user side, and the electricity price elasticity matrix is ​​introduced for modeling, which is expressed as Where L is the electricity price elasticity coefficient, ΔP load_pdr is the load reduction amount of price-based demand response, c e_0 is a fixed electricity price, P load_p is the load before response, Δc e is the difference between the electricity price after demand response and the fixed electricity price; According to the incentive electricity price mechanism, the following electricity quantity and price elasticity matrix L is constructed: ep : The electricity price elasticity matrix L ep The element L in ij Expressed as Where ΔP load_pdri is the load reduction at time i, c e_0i is the fixed electricity price at time i, P load_pi is the original load at time i, Δc ei is the change in electricity price at time i after demand response, c e_0j is the fixed electricity price at time j, Δc ej is the change in electricity price at time j after demand response; the electricity quantity and price elasticity matrix L ep The element L in ii represents the self-elastic coefficient at time i, L ij represents the cross elasticity coefficient between time i and time j; Load reduction after demand response Expressed as S3. According to the data information obtained in step S1, the uncertainty of intra-day wind power generation, intra-day photovoltaic power generation, intra-day electricity price and intra-day user load is processed; specifically, the following steps are included: For intraday wind power generation, intraday photovoltaic power generation, intraday electricity price and intraday user load, the following steps are used to handle uncertainty: Data expansion: In a scheduling cycle, there are 24 data in the day-ahead phase and 96 data in the intraday phase; therefore, the following formula is used for expansion: <h2 style=";text-align:left;direction:ltr">D<h2 style=";text-align:left;direction:ltr"> pre_q <h2 style=";text-align:left;direction:ltr"> (i)={D<h2 style=";text-align:left;direction:ltr"> pre_n <h2 style=";text-align:left;direction:ltr"> (4i-3),D<h2 style=";text-align:left;direction:ltr"> pre_n <h2 style=";text-align:left;direction:ltr"> (4i-2),D<h2 style=";text-align:left;direction:ltr"> pre_n <h2 style=";text-align:left;direction:ltr"> (4i-1),D<h2 style=";text-align:left;direction:ltr"> pre_n <h2 style=";text-align:left;direction:ltr"> (4i)} Where D pre_q (i) is the i-th initial data in the day-ahead data; D pre_n (4i) is the 4ith initial data in the intraday data, i is a positive integer ranging from 1 to 24; the expanded initial data has 4*24; Monte Carlo method is used to select the specific standard deviation s within the day n , generate a standard normal distribution σ of random numbers, and then generate 250 specific original scenes to be cut; The specific original scenes to be cut are cut using the Manhattan probability distance to obtain 10 scenes after the cut and the corresponding scene probabilities; According to the output value and corresponding probability of the reduced scenario, the following formula is used to combine and obtain the final output curve: S pre_n =S pre_n1 ×P pre_n1 +S pre_n2 ×P pre_n2 +....+S pre_n10 ×P pre_n10 Where S pre_n It is the output curve generated after the intraday stage processing; S pre_n1 ~S pre_n10 is the output value of the first to tenth scenes after reduction; P pre_n1 ~P pre_n10 The probabilities corresponding to the first to tenth scenes after reduction; S4. Based on the data information obtained in step S2 and step S3, a multi-time scale optimization scheduling model of the virtual power plant is constructed; specifically, the following steps are included: Day-ahead stage: Dispatch is carried out 24 hours in advance, with the dispatch time scale set to 1 hour. Based on wind and solar power output, load forecast, energy storage adjustable output, incentive-based demand response parameters and system parameters, the virtual power plant dispatch plan for the next 24 hours is formulated with the minimum total cost of the virtual power plant as the objective function, combined with the purchase and sale electricity price and market carbon price; Intraday stage: Dispatch is carried out 4 hours in advance, with a time scale of 15 minutes. The calculation is rolled out every 15 minutes, and a dispatch plan for the next 4 hours is formed each time. The operating status of the gas turbine is determined by the result of the day-ahead optimization. The objective function is to minimize the sum of the penalty caused by the intraday and day-ahead power deviation and the total cost of the virtual power plant, and formulate the daily output of each unit, the daily purchase and sale of electricity, and the daily energy storage charging and discharging plan; The objective function is to minimize the sum of the penalty caused by the intraday and day-ahead power deviation and the total cost of the virtual power plant, and specifically includes the following steps: The following formula is used as the objective function: In the formula It is the sum of the penalty caused by the intraday and day-ahead power deviation and the total cost of the virtual power plant; Operation and maintenance costs for wind power, photovoltaic power and energy storage; For gas turbine operating costs; The cost of purchasing and selling electricity for virtual power plants; for the cost of carbon emissions; is the demand response cost of the intraday incentive electricity price, and T is the scheduling period, is the electricity price before demand response, is the load after demand response at time t and is the load before demand response, is the load reduction amount, is the change in electricity price, Δt in is the intraday time scale; is the penalty cost for output deviation, and is the deviation of the gas turbine daily dispatch value compared with the day-ahead plan and is the value of the i-th gas turbine at time t within the day, P Gi (t) is the optimization result of the i-th gas turbine at time t a day ago, is the deviation of the battery’s daily dispatch value compared to the day-ahead plan and is the value of the ith battery at time t within the day, P bat (t) is the optimization result of the i-th battery at time t, is the deviation of the daily dispatch value of the power grid transaction power compared with the day-ahead plan and is the value of the i-th power grid transaction electricity at time t within the day, P net (t) is the optimization result of the i-th power grid transaction electricity at time t a day ago, Γ GT is the gas turbine deviation penalty coefficient, Γ bat is the battery deviation penalty coefficient, Γ net It is the penalty coefficient for the deviation of power purchase and sale transaction; The following formula is used as the constraint condition: Power balance constraints: In the formula It is the intraday rolling optimization value of the power grid transaction electricity; It is the intraday rolling optimization value of wind power generation; The intraday rolling optimization value for solar power generation; It is the intraday rolling optimization value of the gas turbine; is the load after demand response at time t; It is the daily rolling optimization value of the battery; Gas turbine start and stop constraints: In the formula is the startup state of the i-th gas turbine at time t within the day, δ Gsi (t) is the startup status of the i-th gas turbine at time t, is the stop state of the i-th gas turbine at time t within the day, δ Gti (t) is the stop state of the i-th gas turbine at time t on the previous day; S5. Solve the multi-time scale optimization scheduling model constructed in step S4 to obtain the final virtual power plant scheduling plan; specifically, the steps include: The YALMIP toolbox in matlabR2018a is used to build the model, and the CPLEX12.8.0 solver is used to solve the problem. The specific steps are as follows: Data input: input the acquired data information into the solver; The hourly data is processed and combined with the input data information to build a multi-time scale optimization scheduling model for the virtual power plant on the day before and within the day; Solve the day-ahead phase of the multi-time-scale optimization dispatch model of the constructed virtual power plant to obtain the day-ahead output plan; In the intraday stage, ultra-short-term forecasts are made for wind and solar power output, electricity prices, and user load data to obtain intraday optimized dispatch data, which is updated at the set time. Combine the day-ahead dispatch reference value and intraday constraints, add price-based demand response, perform rolling optimization according to the set update time, and use the obtained results as the correction of the day-ahead dispatch plan; Repeat the above steps to solve the constructed multi-time scale optimization scheduling model in real time; S6. Use the virtual power plant dispatching plan obtained in step S5 to control the operation of the virtual power plant and complete the multi-time scale optimization dispatching for the virtual power plant.

2. A system for implementing the multi-time scale optimization scheduling method for virtual power plants according to claim 1, characterized in that It includes a data acquisition module, a demand response modeling module, an uncertainty processing module, an optimization scheduling modeling module, a model solving module and an operation module; the data acquisition module, the demand response modeling module, the uncertainty processing module, the optimization scheduling modeling module, the model solving module and the operation module are connected in series in sequence; the data acquisition module is used to obtain data information of the target power system and the virtual power plant, and upload the data to the demand response modeling module; the demand response modeling module is used to model the incentive electricity price type demand response according to the received data, and upload the data to the uncertainty processing module; the uncertainty processing module is used to process the uncertainty of intraday wind power generation, intraday photovoltaic power generation, intraday electricity price and intraday user load according to the received data, and upload the data to the optimization scheduling modeling module; The optimization scheduling modeling module is used to construct a multi-time scale optimization scheduling model for the virtual power plant based on the received data, and upload the data to the model solving module; the model solving module is used to solve the constructed multi-time scale optimization scheduling model based on the received data, obtain the final virtual power plant scheduling plan, and upload the data to the operation module; the operation module is used to control the operation of the virtual power plant based on the received data and adopt the obtained virtual power plant scheduling plan to complete the multi-time scale optimization scheduling for the virtual power plant.

Images