Multi-energy virtual power plant optimal scheduling method considering source-load cooperation

Abstract

The invention discloses a multi-energy virtual power plant optimal scheduling method considering source-load cooperation. The method comprises the steps of data acquisition, data preprocessing, establishment of a wind turbine generator, a photovoltaic unit, a gas turbine, distributed energy storage and a demand side response model, establishment of a double-layer optimal scheduling model, solving of the double-layer optimal scheduling model and obtaining of an optimal scheduling result. According to the method, a double-layer optimization scheduling model is designed, the upper-layer optimization scheduling model takes power plant income maximization as a target function, demand response is considered to carry out optimization scheduling on the load side, the effect of peak clipping and valley filling is achieved for the controllable load, and the lower-layer optimization scheduling model considers the output cost of wind, light and fuel storage to carry out optimization scheduling on the power supply side. The established multi-energy virtual power plant double-layer optimization scheduling method considering source-load cooperation is more reasonable, the utilization rate of wind power and photovoltaic power can be improved through source-load cooperative scheduling, and the economical efficiency of overall operation of the system is improved.

Description

Technical Field

[0001] The present invention relates to the field of virtual power plant scheduling and operation, and in particular to a multi-energy virtual power plant optimization scheduling method considering source-load coordination. Background Art

[0002] The proposal of virtual power plants provides a solution for the centralized management of distributed energy sources with dispersed geographical locations and small individual unit capacity, and effectively resolves the huge impact of distributed energy units directly connected to the grid. Virtual power plants are called virtual power plants in English, abbreviated as VPP in English, and distributed energy resources in English are called distributed energy resources in English, abbreviated as DER in English. Virtual power plants include both power generation units and power loads, but the optimization scheduling methods in the prior art that focus on the optimization scheduling of multi-energy virtual power plants mainly focus on the single optimization scheduling on the power supply side or the load side, which has a poor effect on improving the absorption capacity of clean energy such as wind power and photovoltaics, and is likely to affect the stable operation of the power system. Summary of the invention

[0003] The purpose of the present invention is to solve the technical problem that the existing optimization scheduling method has a poor effect on improving the absorption capacity of clean energy such as wind power and photovoltaics.

[0004] In order to achieve the above object, the present invention provides a multi-energy virtual power plant optimization scheduling method considering source-load coordination, which includes:

[0005] S1: Collect the load sample data and meteorological sample data of the previous day. The load sample data includes the load that can be reduced, the load that can be shifted, and the load that can be transferred intermittently; the meteorological sample data includes wind speed, light, and temperature;

[0006] S2: Preprocessing the collected sample data, and obtaining the next day's predicted load and predicted meteorological data based on the sample data, which specifically includes:

[0007] S201: using the box plot theory to determine the data outliers, that is, sorting the sample data and dividing them into four equal parts, the endpoints of the four equal parts are Q0, Q1, Q2, Q3, and Q4, where Q1 and Q3 are the lower quartile and the upper quartile, respectively, and the interquartile range is IOR=Q3-Q1. It is assumed that the data falling within the range of (Q1-1.5×IQR, Q3+1.5×IQR) is a normal value, and the data outside the range is an outlier;

[0008] S202: using mean interpolation method to correct missing values ​​and outliers, that is, according to the time when the missing values ​​and outliers occurred, the weighted average of the data at the same time in the past n days of the same type is calculated to correct the data. The expression of the data correction is:

[0009]

[0010] In the formula, x m,t is the data at time t on the mth day, x m-1,t is the data at time t of the previous day, x m-2,t is the data at time t two days ago, x m-n,t is the data at time t n days ago, a k is the kth weight coefficient, a1 is the weight coefficient of the previous day, a2 is the weight coefficient of the previous 2 days, and a n is the weight coefficient of the previous n days, and a is the smoothing coefficient that determines the weight coefficient;

[0011] S203: Divide the corrected sample data into a training set and a test set, build a BP neural network, input the training set into the BP neural network for training, terminate the training after reaching a preset number of training times or the prediction accuracy meets the requirements, use the test set for testing, and after the training is completed, select the data of the previous three days and input them into the BP neural network for prediction to obtain the predicted load and predicted meteorological data for the next day;

[0012] S3: Establish a virtual power plant wind turbine model, a photovoltaic unit model, a gas turbine model, a distributed energy storage model, and a demand-side response model, and input the predicted meteorological data into the wind turbine model and the photovoltaic unit model to obtain the predicted output of the wind turbine and the predicted output of the photovoltaic unit;

[0013] In the wind turbine model, the relationship between wind speed and wind turbine output power is:

[0014]

[0015] Where P t,WT is the predicted output of the wind turbine generator set at time t, v i is the cut-in speed of the wind turbine, v o is the cut-out speed of the wind turbine, v r is the rated speed of the wind turbine, m, n, p, q are all the output coefficients of the wind turbine; p r is the rated power;

[0016] The power generation cost expression of the wind turbine is as follows:

[0017]

[0018] In the formula, Ft,WT is the output cost of the wind turbine at time t, k WT To reflect the correlation coefficient between output cost and output size, is the actual output of the wind turbine at time t, c SM represents the real-time market electricity price, ΔP t,WT is the wind abandonment volume at time t;

[0019] The relationship between photovoltaic power generation and light radiation intensity in the photovoltaic unit model is expressed as follows:

[0020]

[0021] Where f(θ) is the photovoltaic power generation, ξ and ψ are shape parameters, μ is the mean photovoltaic radiation intensity, Γ is the Gamma function, θ is the photovoltaic radiation intensity, and σ is the standard deviation of the photovoltaic intensity;

[0022] The total cost of power generation in the photovoltaic unit model is expressed as:

[0023]

[0024] In the formula, F t,PV is the output cost of the photovoltaic unit at time t, k PV To reflect the correlation coefficient between the output cost and output size of photovoltaic units, is the actual output of the photovoltaic unit at time t, c SM represents the real-time market electricity price, ΔP t,PV is the amount of abandoned light at time t;

[0025] The power generation cost expression of the gas turbine in the gas turbine model is:

[0026] F t，j,MT =a j,MT P t，j,M 2 +b j,MT P t，j,M +c j,MT

[0027] In the formula, F t，j,MT is the power generation cost of the jth gas turbine at time t, P t，j,M is the output of the jth gas turbine at time t, a j,MT , b j,MT 、c j,MT is the unit operating cost coefficient of the jth gas turbine;

[0028] The expression of the charging and discharging model in the distributed energy storage model is:

[0029]

[0030] In the formula, SOC t is the state of charge of the battery at time t, SOC t-1 represents the state of charge at time t-1, P t,ess is the charge and discharge power of the battery at time t, P t,ess When it is a positive number, the battery is in a discharging state, P t,ess When it is a negative number, the battery is in a charging state, E ess is the battery capacity, η ch is the battery charging efficiency, η dis is the discharge efficiency of the battery;

[0031] The power generation cost of the energy storage unit in the distributed energy storage model is expressed as:

[0032] F t,ess =a ess P t,ess 2 +b ess |P t,ess |+c ess

[0033] In the formula, F t,ess is the power generation cost of the energy storage unit at time t, P t,ess is the charge and discharge power of the battery at time t, a ess , b ess 、c ess is the power parameter of the battery;

[0034] The demand-side response model uses elasticity coefficient to represent the impact of electricity price change rate on load response rate. The impact of electricity price change at time t on electricity demand is expressed as:

[0035]

[0036] In the formula, is the response rate of the load at time t, φ Δp,t is the rate of change of electricity price at time t, ε j is the self-elasticity coefficient in the elasticity matrix, j is the scheduling time;

[0037] The economic compensation expression given to various controllable loads at time t in the demand-side response model is:

[0038]

[0039] In the formula, F t,cut F is the economic compensation obtained after load reduction and dispatch. t,tr is the economic compensation obtained after the movable load participates in the dispatch, F t,in The economic compensation obtained after intermittent load transfer and dispatch;t,cut 、x t,tr 、x t,in is a 0-1 variable, 1 represents participation in scheduling, and 0 represents non-participation in scheduling; a cut is the unit compensation price for load reduction, a tr is the unit compensation price of the movable load, a in It is divided into unit compensation price for intermittent transfer load; P t,cut is the load transfer amount that can reduce the load, P t,tr is the load transfer of the translatable load, P t,in The load transfer amount for intermittent load transfer;

[0040] S4: Establish a two-layer optimization scheduling model for a virtual power plant considering source-load coordination, wherein the two-layer optimization scheduling model for a virtual power plant considering source-load coordination includes an upper-layer optimization scheduling model and a lower-layer optimization scheduling model, and the expression of the upper-layer optimization scheduling model is:

[0041]

[0042] In the formula, F t,up is the total revenue of the virtual power plant, T is the number of time periods in a scheduling cycle, ρ t,use is the real-time electricity price of the virtual power plant, P t,use is the electricity consumption of the user at time t under price-based demand response, ρ t,CT and P t,CT are the electricity price and electricity quantity purchased by the virtual power plant from the power supplier at time t, ρ t,g and P t,g are the electricity price and electricity volume traded between the virtual power plant and the external power market at time t. If the virtual power plant purchases electricity from the outside, then P t,g is a positive number. If the virtual power plant sells electricity to the outside, then P t,g is a negative number, ΔT is the duration of each scheduling period, F t,down is the power generation cost of the virtual power plant;

[0043] The power consumption expression of the user at time t under the price-based demand response is:

[0044]

[0045] Where P t,L is the predicted user load at time t;

[0046] The expression of the lower-level optimization scheduling model is:

[0047]

[0048] In the formula, F t,down is the power generation cost of the virtual power plant, Ft,WT is the output cost of the wind turbine at time t, F t,PV is the output cost of the photovoltaic unit at time t, T is the number of time periods in a scheduling cycle, and F t，j,MT is the power generation cost of the jth gas turbine at time t, M is the total number of gas turbines, F t,ess is the power generation cost of the energy storage unit at time t;

[0049] S5: Based on the predicted output of wind turbines and photovoltaic units, the commercial optimization software CPLEX is called in Matlab software through the YALMIP toolkit to optimize the double-layer optimization scheduling model and obtain the optimization scheduling results, thus realizing the optimization scheduling of multi-energy virtual power plants considering source-load coordination. The specific steps include:

[0050] S501: Input the predicted output of wind turbine generators and photovoltaic power generation into the upper optimization scheduling model, and the upper optimization scheduling model obtains the total output of each generator set of the optimized VPP, the optimization scheduling result, and the economic compensation F obtained after the load that can be cut, the load that can be shifted, and the intermittent transfer load participate in the scheduling. t,cut 、F t,tr 、F t,in Optimize scheduling results;

[0051] S502: The obtained total output optimization scheduling result of each generator set of the VPP is input into the lower optimization model, and the lower optimization model obtains the output optimization scheduling result of each generator set, and transmits the output optimization scheduling result of each generator set to the upper optimization scheduling model, and the wind turbine set, photovoltaic unit, gas turbine, and distributed energy storage are optimized and scheduled according to their respective output optimization scheduling results;

[0052] S503: The upper-level optimization scheduling model compares the output of each generator set transmitted by the lower-level optimization scheduling model with the predicted load. If it is greater than the predicted load, the virtual power plant sells electricity to the power grid; if it is less than the predicted load, the virtual power plant purchases electricity from the power grid.

[0053] In a preferred manner, the constraint condition of the upper-layer optimization scheduling model includes a power balance constraint, and the expression of the power balance constraint is:

[0054] P t,dowm +P t,g +P t,CT =P t,use -P t,cut -P t,in -P t,tr

[0055] Where P t,dowm is the total output of wind, solar, fuel and storage units in the virtual power plant, P t,gis the amount of electricity traded between the virtual power plant and the external power market at time t, P t,CT is the amount of electricity purchased by the virtual power plant from the power supplier at time t, P t,use is the electricity consumption of the user at time t under price-based demand response, P t,cut is the load transfer amount that can reduce the load, P t,tr is the load transfer of the translatable load, P t,in The load transfer amount for intermittent load transfer;

[0056] The constraint condition of the upper-level optimization scheduling model also includes a time-of-use electricity price constraint, and the expression of the time-of-use electricity price constraint is:

[0057]

[0058] In the formula, is the average electricity price traded between virtual power plants and users within a dispatch cycle, The lower limit of the time-of-use electricity price set for virtual power plants, The upper limit of the time-of-use electricity price set for the virtual power plant, P t,use is the electricity consumption of users at time t under price-based demand response;

[0059] The demand response load constraint in the constraint conditions of the upper-level optimization scheduling model includes a curtailable load, and the expression of the curtailable load is:

[0060]

[0061] Where N is the upper limit of the number of load reductions that the user can accept within a period of time, P t,cut To reduce the load, is the minimum value of load reduction, is the maximum value of load reduction, T is the number of time periods in a scheduling cycle;

[0062] The demand response load constraint in the constraint conditions of the upper-level optimization scheduling model also includes a shiftable load, and the expression of the shiftable load is:

[0063]

[0064] In the formula, t1 and t2 are the time when the load starts to use electricity and stops using electricity before translation; are the time when the load starts to use electricity and stops using electricity after translation; are the lower and upper limits of the load translation amount respectively;

[0065] The demand response load constraint in the constraint conditions of the upper-level optimization scheduling model also includes an intermittent transfer load, and the expression of the intermittent transfer load is:

[0066]

[0067] In the formula, t3 is the time when the load starts to use electricity before the intermittent transfer, and t4 is the time when the load stops using electricity before the intermittent transfer; The time when the load starts to use electricity after intermittent transfer. ΔT is the time when the load stops using electricity after intermittent transfer; min The minimum working time for intermittent load transfer; x t-1,in Indicates whether it participates in scheduling at time t-1, 1 means it participates in scheduling, and 0 means it does not participate in scheduling; is the minimum value of load intermittent transfer, It is the maximum value of intermittent load transfer.

[0068] In a preferred manner, the constraint condition of the lower-layer optimization scheduling model includes a power balance constraint, and the expression of the power balance constraint is:

[0069]

[0070] Where P t,dowm is the total output of wind, solar, fuel and storage units in the virtual power plant, P t,WT is the predicted output of the wind turbine at time t, P t,PV is the predicted output of the photovoltaic unit, P t,j,MT is the output of the jth gas turbine at time t, P t,ess is the charge and discharge power of the battery at time t;

[0071] The constraint conditions of the lower-level optimization scheduling model also include the output constraint of the new energy unit. The expression of the output constraint of the new energy unit is:

[0072]

[0073] In the formula, is the wind turbine output at time t, and They are the minimum and maximum output of the wind turbine at time t respectively; is the output of the photovoltaic unit at time t, and They are the minimum and maximum power of the photovoltaic unit at time t respectively;

[0074] The constraint conditions of the lower-level optimization scheduling model also include thermal power unit climbing power constraint and output constraint. The expressions of the thermal power unit climbing power constraint and output constraint are:

[0075]

[0076] Where P t,j,MT is the output of the jth gas turbine at time t, P t-1,j,MT is the power generation of the jth gas turbine at time t-1, are the maximum rising power and maximum falling power of the jth gas turbine respectively; are the upper and lower limits of the output of the jth gas turbine respectively;

[0077] The constraints of the lower-level optimization scheduling model also include energy storage battery capacity constraints and charge and discharge constraints. The expressions of the energy storage battery capacity constraints and charge and discharge constraints are:

[0078]

[0079] In the formula, S min , S max are the lower and upper limits of the battery capacity respectively; S0 is the original capacity of the battery; T1 is the battery charging cut-off time; They are the upper limit of battery charging power and the upper limit of battery discharging power respectively.

[0080] The beneficial effects of the present invention are as follows: the present invention designs a double-layer optimization scheduling model, the upper-layer optimization scheduling model takes the maximization of power plant revenue as the objective function, considers the demand response to perform optimization scheduling on the load side, and achieves the effect of peak shaving and valley filling for controllable loads, and the lower-layer optimization scheduling model considers the output cost of wind, solar, fuel and storage to perform optimization scheduling on the power supply side, so that the established double-layer optimization scheduling method of a multi-energy virtual power plant considering source-load coordination is more reasonable, and the utilization rate of wind power and photovoltaic power can be improved through the coordinated scheduling of source and load, thereby improving the economy of the overall operation of the system; basic unit modules of wind turbine model, photovoltaic unit model, gas turbine model, distributed energy storage model and demand-side response model are established, and these geographically dispersed distributed energy sources with small single unit capacity are aggregated into a virtual power plant for centralized management, thereby resolving the huge impact of the direct grid-connected operation of distributed energy units on the power grid; by collecting the load data and meteorological data of the day before and preprocessing the data, the accuracy of load data and wind and solar output prediction can be effectively improved. BRIEF DESCRIPTION OF THE DRAWINGS

[0081] Figure 1 is a flow chart of the optimization scheduling method of the present invention;

[0082] Figure 2 It is an optimization scheduling flow chart of the double-layer optimization scheduling model of the present invention;

[0083] Figure 3is a load data, wind turbine generator set predicted output, and photovoltaic generator set predicted output prediction curve diagram in the embodiment of the present invention;

[0084] Figure 4 is a power diagram of each controllable load before optimization scheduling in an embodiment of the present invention;

[0085] Figure 5 is a power diagram of each controllable load after optimized scheduling in an embodiment of the present invention;

[0086] Figure 6 It is the output optimization scheduling result of each generator set in the embodiment of the present invention. DETAILED DESCRIPTION Specific embodiment:

[0088] like Figure 1 , 2 As shown, the present invention provides a multi-energy virtual power plant optimization scheduling method considering source-load coordination, comprising the following steps:

[0089] S1: Collect the load sample data and meteorological sample data of the previous day. The load sample data includes the load that can be reduced, the load that can be shifted, and the load that can be transferred intermittently; the meteorological sample data includes wind speed, light, and temperature;

[0090] Table 1 Controllable load parameters involved in virtual power plant dispatch

[0091]

[0092]

[0093] S2: Preprocess the collected sample data and obtain the next day's forecast load and forecast meteorological data based on the sample data, such as Figure 3 As shown;

[0094] The preprocessing of sample data includes the following steps:

[0095] S2: Preprocessing the collected sample data, and obtaining the next day's predicted load and predicted meteorological data based on the sample data, which specifically includes:

[0096] S201: using the box plot theory to determine the data outliers, that is, sorting the sample data and dividing them into four equal parts, the endpoints of the four equal parts are Q0, Q1, Q2, Q3, and Q4, where Q1 and Q3 are the lower quartile and the upper quartile, respectively, and the interquartile range is IOR=Q3-Q1. It is assumed that the data falling within the range of (Q1-1.5×IQR, Q3+1.5×IQR) is a normal value, and the data outside the range is an outlier;

[0097] S202: using mean interpolation method to correct missing values ​​and outliers, that is, according to the time when the missing values ​​and outliers occurred, the weighted average of the data at the same time in the past n days of the same type is calculated to correct the data. The expression of the data correction is:

[0098]

[0099] In the formula, x m,t is the data at time t on the mth day, x m-1,t is the data at time t of the previous day, x m-2,t is the data at time t two days ago, x m-n,t is the data at time t n days ago, a k is the kth weight coefficient, a1 is the weight coefficient of the previous day, a2 is the weight coefficient of the previous 2 days, and a n is the weight coefficient of the previous n days, and a is the smoothing coefficient that determines the weight coefficient;

[0100] S203: Divide the corrected sample data into a training set and a test set, build a BP neural network, input the training set into the BP neural network for training, terminate the training after reaching a preset number of training times or the prediction accuracy meets the requirements, use the test set for testing, and after the training is completed, select the data of the previous three days and input them into the BP neural network for prediction to obtain the predicted load and predicted meteorological data for the next day;

[0101] S3: Establish a virtual power plant wind turbine model, a photovoltaic unit model, a gas turbine model, a distributed energy storage model, and a demand-side response model, and input the predicted meteorological data into the wind turbine model and the photovoltaic unit model to obtain the predicted output of the wind turbine and the predicted output of the photovoltaic unit;

[0102] In the wind turbine model, the relationship between wind speed and wind turbine output power is:

[0103]

[0104] Where P t,WT is the predicted output of the wind turbine generator set at time t, v i is the cut-in speed of the wind turbine, v o is the cut-out speed of the wind turbine, v r is the rated speed of the wind turbine, m, n, p, q are all the output coefficients of the wind turbine; p r is the rated power;

[0105] The power generation cost expression of the wind turbine is as follows:

[0106]

[0107] In the formula, Ft,WT is the output cost of the wind turbine at time t, k WT To reflect the correlation coefficient between output cost and output size, is the actual output of the wind turbine at time t, c SM represents the real-time market electricity price, ΔP t,WT is the wind abandonment volume at time t;

[0108] The relationship between photovoltaic power generation and light radiation intensity in the photovoltaic unit model is expressed as follows:

[0109]

[0110] Where f(θ) is the photovoltaic power generation, ξ and ψ are shape parameters, μ is the mean photovoltaic radiation intensity, Γ is the Gamma function, θ is the photovoltaic radiation intensity, and σ is the standard deviation of the photovoltaic intensity;

[0111] The total cost of power generation in the photovoltaic unit model is expressed as:

[0112]

[0113] In the formula, F t,PV is the output cost of the photovoltaic unit at time t, k PV To reflect the correlation coefficient between the output cost and output size of photovoltaic units, is the actual output of the photovoltaic unit at time t, c SM represents the real-time market electricity price, ΔP t,PV is the amount of abandoned light at time t;

[0114] The power generation cost expression of the gas turbine in the gas turbine model is:

[0115] F t，j,MT =a j,MT P t，j,M 2 +b j,MT P t，j,M +c j,MT

[0116] In the formula, F t，j,MT is the power generation cost of the jth gas turbine at time t, P t，j,M is the output of the jth gas turbine at time t, a j,MT , b j,MT 、c j,MT is the unit operating cost coefficient of the jth gas turbine;

[0117] The expression of the charging and discharging model in the distributed energy storage model is:

[0118]

[0119] In the formula, SOC t is the state of charge of the battery at time t, SOC t-1 represents the state of charge at time t-1, P t,ess is the charge and discharge power of the battery at time t, P t,ess When it is a positive number, the battery is in a discharging state, P t,ess When it is a negative number, the battery is in a charging state, E ess is the battery capacity, η ch is the battery charging efficiency, η dis is the discharge efficiency of the battery;

[0120] The power generation cost of the energy storage unit in the distributed energy storage model is expressed as:

[0121] F t,ess =a ess P t,ess 2 +b ess |P t,ess |+c ess

[0122] In the formula, F t,ess is the power generation cost of the energy storage unit at time t, P t,ess is the charge and discharge power of the battery at time t, a ess , b ess 、c ess is the power parameter of the battery;

[0123] The demand-side response model uses elasticity coefficient to represent the impact of electricity price change rate on load response rate. The impact of electricity price change at time t on electricity demand is expressed as:

[0124]

[0125] In the formula, is the response rate of the load at time t, φ Δp,t is the rate of change of electricity price at time t, ε j is the self-elasticity coefficient in the elasticity matrix, j is the scheduling time;

[0126] The economic compensation expression given to various controllable loads at time t in the demand-side response model is:

[0127]

[0128] In the formula, F t,cut F is the economic compensation obtained after load reduction and dispatch. t,tr is the economic compensation obtained after the movable load participates in the dispatch, F t,in The economic compensation obtained after intermittent load transfer and dispatch;t,cut 、x t,tr 、x t,in is a 0-1 variable, 1 represents participation in scheduling, and 0 represents non-participation in scheduling; a cut is the unit compensation price for load reduction, a tr is the unit compensation price of the movable load, a in It is divided into unit compensation price for intermittent transfer load; P t,cut is the load transfer amount that can reduce the load, P t,tr is the load transfer of the translatable load, P t,in The load transfer amount for intermittent load transfer;

[0129] S4: Establish a two-layer optimization scheduling model for a virtual power plant considering source-load coordination, wherein the two-layer optimization scheduling model for a virtual power plant considering source-load coordination includes an upper-layer optimization scheduling model and a lower-layer optimization scheduling model, and the expression of the upper-layer optimization scheduling model is:

[0130]

[0131] In the formula, F t,up is the total revenue of the virtual power plant, T is the number of time periods in a scheduling cycle, ρ t,use is the real-time electricity price of the virtual power plant, P t,use is the electricity consumption of the user at time t under price-based demand response, ρ t,CT and P t,CT are the electricity price and electricity quantity purchased by the virtual power plant from the power supplier at time t, ρ t,g and P t,g are the electricity price and electricity volume traded between the virtual power plant and the external power market at time t. If the virtual power plant purchases electricity from the outside, then P t,g is a positive number. If the virtual power plant sells electricity to the outside, then P t,g is a negative number, ΔT is the duration of each scheduling period, F t,down is the power generation cost of the virtual power plant;

[0132] The power consumption expression of the user at time t under the price-based demand response is:

[0133]

[0134] Where P t,L is the predicted user load at time t;

[0135] The expression of the lower-level optimization scheduling model is:

[0136]

[0137] In the formula, F t,down is the power generation cost of the virtual power plant, Ft,WT is the output cost of the wind turbine at time t, F t,PV is the output cost of the photovoltaic unit at time t, T is the number of time periods in a scheduling cycle, and F t，j,MT is the power generation cost of the jth gas turbine at time t, M is the total number of gas turbines, F t,ess is the power generation cost of the energy storage unit at time t;

[0138] S5: Based on the predicted output of wind turbines and photovoltaic units, the commercial optimization software CPLEX is called in Matlab software through the YALMIP toolkit to optimize the double-layer optimization scheduling model and obtain the optimization scheduling results, thus realizing the optimization scheduling of multi-energy virtual power plants considering source-load coordination. The specific steps include:

[0139] S501: Input the predicted output of wind turbine generators and photovoltaic power generation into the upper optimization scheduling model, and the upper optimization scheduling model obtains the total output of each generator set of the optimized VPP, the optimization scheduling result, and the economic compensation F obtained after the load that can be cut, the load that can be shifted, and the intermittent transfer load participate in the scheduling. t,cut 、F t,tr 、F t,in Optimize scheduling results; Figure 4 is the power of each controllable load before optimization, Figure 5 is the optimized controllable load power, Figure 4 and Figure 5 It can be seen that for the load that can be reduced, the power consumption is reduced at 12:00-15:00 and 18:00-21:00, reducing the burden on the power generation system during peak hours; for the load that can be shifted, the power consumption period is shifted from 18:00-22:00 to 5:00-9:00 as a whole, effectively avoiding the peak power consumption period; for the intermittent transfer load, the load from 11:00-15:00 is dispersed and transferred to 8:00-11:00 and 15:00-17:00, and the peak load is effectively reduced by multiple decentralized transfers. It can be seen that the optimization scheduling of various controllable loads can play a role in peak shaving and valley filling;

[0140] S502: The obtained total output optimization scheduling result of each generator set of the VPP is input into the lower optimization model. The lower optimization model obtains the output optimization scheduling result of each generator set, and transmits the output optimization scheduling result of each generator set to the upper optimization scheduling model. The wind turbine set, photovoltaic unit, gas turbine, and distributed energy storage are optimized and scheduled according to their respective output optimization scheduling results. The result is as follows: Figure 6 As shown;

[0141] S503: The upper-level optimization scheduling model compares the output of each generator set transmitted by the lower-level optimization scheduling model with the predicted load. If it is greater than the predicted load, the virtual power plant sells electricity to the power grid; if it is less than the predicted load, the virtual power plant purchases electricity from the power grid.

[0142] The above description is only a preferred specific implementation manner of the present invention, but the protection scope of the present invention is not limited thereto. Any technician familiar with the technical field can make equivalent replacements or changes according to the technical scheme and inventive concept of the present invention within the technical scope disclosed by the present invention, which should be covered by the protection scope of the present invention.

Claims

1. A multi-energy virtual power plant optimization scheduling method considering source-load coordination, characterized in that: It includes: S1: Collect the load sample data and meteorological sample data of the previous day, wherein the load sample data includes the load that can be cut, the load that can be shifted, and the load that can be transferred intermittently; Meteorological sample data include wind speed, light, and temperature; S2: Preprocessing the collected sample data, and obtaining the next day's predicted load and predicted meteorological data based on the sample data, which specifically includes: S201: using the box plot theory to determine the data outliers, that is, sorting the sample data and dividing them into four equal parts, the endpoints of the four equal parts are Q0, Q1, Q2, Q3, and Q4, where Q1 and Q3 are the lower quartile and the upper quartile, respectively, and the interquartile range is IOR=Q3-Q1. It is assumed that the data falling within the range of (Q1-1.5×IQR, Q3+1.5×IQR) is a normal value, and the data outside the range is an outlier; S202: using mean interpolation method to correct missing values ​​and outliers, that is, according to the time when the missing values ​​and outliers occurred, the weighted average of the data at the same time in the past n days of the same type is calculated to correct the data. The expression of the data correction is: In the formula, x m,t is the data at time t on the mth day, x m-1,t is the data at time t of the previous day, x m-2,t is the data at time t two days ago, x m-n,t is the data at time t n days ago, a k is the kth weight coefficient, a1 is the weight coefficient of the previous day, a2 is the weight coefficient of the previous 2 days, and a n is the weight coefficient of the previous n days, and a is the smoothing coefficient that determines the weight coefficient; S203: Divide the corrected sample data into a training set and a test set, build a BP neural network, input the training set into the BP neural network for training, terminate the training after reaching a preset number of training times or the prediction accuracy meets the requirements, use the test set for testing, and after the training is completed, select the data of the previous three days and input them into the BP neural network for prediction to obtain the predicted load and predicted meteorological data for the next day; S3: Establish a virtual power plant wind turbine model, a photovoltaic unit model, a gas turbine model, a distributed energy storage model, and a demand-side response model, and input the predicted meteorological data into the wind turbine model and the photovoltaic unit model to obtain the predicted output of the wind turbine and the predicted output of the photovoltaic unit; In the wind turbine model, the relationship between wind speed and wind turbine output power is: Where P t,WT is the predicted output of the wind turbine generator set at time t, v i is the cut-in speed of the wind turbine, v o is the cut-out speed of the wind turbine, v r is the rated speed of the wind turbine, m, n, p, q are all the output coefficients of the wind turbine; p r is the rated power; The power generation cost expression of the wind turbine is as follows: In the formula, F t,WT is the output cost of the wind turbine at time t, k WT To reflect the correlation coefficient between output cost and output size, is the actual output of the wind turbine at time t, c SM represents the real-time market electricity price, ΔP t,WT is the wind abandonment volume at time t; The relationship between photovoltaic power generation and light radiation intensity in the photovoltaic unit model is expressed as follows: Where f(θ) is the photovoltaic power generation, ξ and ψ are shape parameters, μ is the mean photovoltaic radiation intensity, Γ is the Gamma function, θ is the photovoltaic radiation intensity, and σ is the standard deviation of the photovoltaic intensity; The total cost of power generation in the photovoltaic unit model is expressed as: In the formula, F t,PV is the output cost of the photovoltaic unit at time t, k PV To reflect the correlation coefficient between the output cost and output size of photovoltaic units, is the actual output of the photovoltaic unit at time t, c SM represents the real-time market electricity price, ΔP t,PV is the amount of abandoned light at time t; The power generation cost expression of the gas turbine in the gas turbine model is: F t，j,MT =a j,MT P t，j,M 2 +b j,MT P t，j,M +c j,MT In the formula, F t，j,MT is the power generation cost of the jth gas turbine at time t, P t，j,M is the output of the jth gas turbine at time t, a j,MT 、b j,MT 、c j,MT is the unit operating cost coefficient of the jth gas turbine; The expression of the charging and discharging model in the distributed energy storage model is: In the formula, SOC t is the state of charge of the battery at time t, SOC t-1 represents the state of charge at time t-1, P t,ess is the charge and discharge power of the battery at time t, P t,ess When it is a positive number, the battery is in a discharging state, P t,ess When it is a negative number, the battery is in a charging state, E ess is the battery capacity, η ch is the charging efficiency of the battery, η dis is the discharge efficiency of the battery; The power generation cost of the energy storage unit in the distributed energy storage model is expressed as: F t,ess =a ess P t,ess 2 +b ess |P t,ess |+c ess In the formula, F t,ess is the power generation cost of the energy storage unit at time t, P t,ess is the charge and discharge power of the battery at time t, a ess , b ess 、c ess is the power parameter of the battery; The demand-side response model uses elasticity coefficient to represent the impact of electricity price change rate on load response rate. The impact of electricity price change at time t on electricity demand is expressed as: In the formula, is the response rate of the load at time t, φ Δp,t is the rate of change of electricity price at time t, ε j is the self-elasticity coefficient in the elasticity matrix, j is the scheduling time; The economic compensation expression given to various controllable loads at time t in the demand-side response model is: In the formula, F t,cut F is the economic compensation obtained after load reduction and dispatch. t,tr is the economic compensation obtained after the movable load participates in the dispatch, F t,in The economic compensation obtained after intermittent load transfer and dispatch; t,cut 、x t,tr 、x t,in is a 0-1 variable, 1 represents participation in scheduling, and 0 represents non-participation in scheduling; a cut is the unit compensation price for load reduction, a tr is the unit compensation price of the movable load, a in It is divided into unit compensation price for intermittent transfer load; P t,cut is the load transfer amount that can reduce the load, P t,tr is the load transfer of the translatable load, P t,in The load transfer amount for intermittent load transfer; S4: Establish a two-layer optimization scheduling model for a virtual power plant considering source-load coordination, wherein the two-layer optimization scheduling model for a virtual power plant considering source-load coordination includes an upper-layer optimization scheduling model and a lower-layer optimization scheduling model, and the expression of the upper-layer optimization scheduling model is: In the formula, F t,up is the total revenue of the virtual power plant, T is the number of time periods in a scheduling cycle, ρ t,use is the real-time electricity price of the virtual power plant, P t,use is the electricity consumption of the user at time t under price-based demand response, ρ t,CT and P t,CT are the electricity price and electricity quantity purchased by the virtual power plant from the power supplier at time t, ρ t,g and P t,g are the electricity price and electricity volume traded between the virtual power plant and the external power market at time t. If the virtual power plant purchases electricity from the outside, then P t,g is a positive number. If the virtual power plant sells electricity to the outside, then P t,g is a negative number, ΔT is the duration of each scheduling period, F t,down is the power generation cost of the virtual power plant; The power consumption expression of the user at time t under the price-based demand response is: Where P t,L is the predicted user load at time t; The expression of the lower-level optimization scheduling model is: In the formula, F t,down is the power generation cost of the virtual power plant, F t,WT is the output cost of the wind turbine at time t, F t,PV is the output cost of the photovoltaic unit at time t, T is the number of time periods in a scheduling cycle, and F t，j,MT is the power generation cost of the jth gas turbine at time t, M is the total number of gas turbines, F t,ess is the power generation cost of the energy storage unit at time t; S5: Based on the predicted output of wind turbines and photovoltaic units, the commercial optimization software CPLEX is called in Matlab software through the YALMIP toolkit to optimize the double-layer optimization scheduling model and obtain the optimization scheduling results, thus realizing the optimization scheduling of multi-energy virtual power plants considering source-load coordination. The specific steps include: S501: Input the predicted output of wind turbine generators and photovoltaic power generation into the upper optimization scheduling model, and the upper optimization scheduling model obtains the total output of each generator set of the optimized VPP, the optimization scheduling result, and the economic compensation F obtained after the load that can be cut, the load that can be shifted, and the intermittent transfer load participate in the scheduling. t,cut 、F t,tr 、F t,in Optimize scheduling results; S502: The obtained total output optimization scheduling result of each generator set of the VPP is input into the lower optimization model, and the lower optimization model obtains the output optimization scheduling result of each generator set, and transmits the output optimization scheduling result of each generator set to the upper optimization scheduling model, and the wind turbine set, photovoltaic unit, gas turbine, and distributed energy storage are optimized and scheduled according to their respective output optimization scheduling results; S503: The upper-level optimization scheduling model compares the output of each generator set transmitted by the lower-level optimization scheduling model with the predicted load. If it is greater than the predicted load, the virtual power plant sells electricity to the power grid; if it is less than the predicted load, the virtual power plant purchases electricity from the power grid.

2. The multi-energy virtual power plant optimization scheduling method considering source-load coordination according to claim 1 is characterized in that: The constraint condition of the upper layer optimization scheduling model includes a power balance constraint, and the expression of the power balance constraint is: P t,dowm +P t,g +P t,CT =P t,use -P t,cut -P t,in -P t,tr Where P t,dowm is the total output of wind, solar, fuel and storage units in the virtual power plant, P t,g is the amount of electricity traded between the virtual power plant and the external power market at time t, P t,CT is the amount of electricity purchased by the virtual power plant from the power supplier at time t, P t,use is the electricity consumption of the user at time t under price-based demand response, P t,cut is the load transfer amount that can reduce the load, P t,tr is the load transfer of the translatable load, P t,in The load transfer amount for intermittent load transfer; The constraint condition of the upper-level optimization scheduling model also includes a time-of-use electricity price constraint, and the expression of the time-of-use electricity price constraint is: In the formula, is the average electricity price traded between virtual power plants and users within a dispatch cycle, The lower limit of the time-of-use electricity price set for virtual power plants, The upper limit of the time-of-use electricity price set for the virtual power plant, P t,use is the electricity consumption of users at time t under price-based demand response; The demand response load constraint in the constraint conditions of the upper-level optimization scheduling model includes a curtailable load, and the expression of the curtailable load is: Where N is the upper limit of the number of load reductions that the user can accept within a period of time, P t,cut To reduce the load, is the minimum value of load reduction, is the maximum value of load reduction, T is the number of time periods in a scheduling cycle; The demand response load constraint in the constraint conditions of the upper-level optimization scheduling model also includes a shiftable load, and the expression of the shiftable load is: In the formula, t1 and t2 are the time when the load starts to use electricity and stops using electricity before translation; are the time when the load starts to use electricity and stops using electricity after translation; are the lower and upper limits of the load translation amount respectively; The demand response load constraint in the constraint conditions of the upper-level optimization scheduling model also includes an intermittent transfer load, and the expression of the intermittent transfer load is: In the formula, t3 is the time when the load starts to use electricity before the intermittent transfer, and t4 is the time when the load stops using electricity before the intermittent transfer; The time when the load starts to use electricity after intermittent transfer. ΔT is the time when the load stops using electricity after intermittent transfer; min The minimum working time for intermittent load transfer; x t-1,in Indicates whether it participates in scheduling at time t-1, 1 means it participates in scheduling, and 0 means it does not participate in scheduling; is the minimum value of load intermittent transfer, It is the maximum value of intermittent load transfer.

3. The multi-energy virtual power plant optimization scheduling method considering source-load coordination according to claim 1 is characterized in that: The constraint condition of the lower-layer optimization scheduling model includes a power balance constraint, and the expression of the power balance constraint is: Where P t,dowm is the total output of wind, solar, fuel and storage units in the virtual power plant, P t,WT is the predicted output of the wind turbine at time t, P t,PV is the predicted output of the photovoltaic unit, P t,j,MT is the output of the jth gas turbine at time t, P t,ess is the charge and discharge power of the battery at time t; The constraint conditions of the lower-level optimization scheduling model also include the output constraint of the new energy unit. The expression of the output constraint of the new energy unit is: In the formula, is the wind turbine output at time t, and They are the minimum and maximum output of the wind turbine at time t respectively; is the output of the photovoltaic unit at time t, and They are the minimum and maximum power of the photovoltaic unit at time t respectively; The constraint conditions of the lower-level optimization scheduling model also include thermal power unit climbing power constraint and output constraint. The expressions of the thermal power unit climbing power constraint and output constraint are: Where P t,j,MT is the output of the jth gas turbine at time t, P t-1,j,MT is the power generation of the jth gas turbine at time t-1, are the maximum rising power and maximum falling power of the jth gas turbine respectively; are the upper and lower limits of the output of the jth gas turbine respectively; The constraints of the lower-level optimization scheduling model also include energy storage battery capacity constraints and charge and discharge constraints. The expressions of the energy storage battery capacity constraints and charge and discharge constraints are: In the formula, S min , S max are the lower and upper limits of battery capacity respectively; S0 is the original capacity of the battery; T1 is the battery charging cut-off time; They are the upper limit of battery charging power and the upper limit of battery discharging power respectively.

Images