Multi-time scale scheduling method considering large-scale renewable energy integration

Abstract

The present invention provides a multi-time scale scheduling method considering large-scale renewable energy access, comprising the following steps: S1, predicting the upper and lower limits of the day-ahead output of renewable energy through the probability distribution of the day-ahead prediction error of renewable energy, and constructing a robust uncertainty set; S2, constructing a two-stage robust scheduling model of a distribution network and a multi-microgrid system, comprising a day-ahead stage model and a regulation stage model; S3, based on the day-ahead stage model, splitting the day-ahead robust scheduling problem into main sub-problems through a column constraint generation algorithm, and iteratively solving to obtain the optimal day-ahead scheduling decision of the distribution network and the multi-microgrid system under the worst scenario; S4, based on the regulation stage model, on the basis of the optimal day-ahead scheduling decision, adopting an intra-day rolling optimization algorithm to perform rolling optimization on the intra-day scheduling of the distribution network and the multi-microgrid system, so as to realize multi-time scale scheduling of the distribution network and the multi-microgrid system.

Description

Technical Field

[0001] The present invention relates to the technical field of power dispatching, and in particular to a multi-time scale dispatching method considering large-scale renewable energy access. Background Art

[0002] With the rapid development and extensive application of renewable energy distributed generation technology, microgrids have attracted widespread attention as an effective access method for distributed power sources. Microgrid groups coordinate the energy management of adjacent distributed units, microgrids and loads in a region, which can improve the penetration rate of distributed power sources and achieve efficient utilization of renewable energy. Due to the volatility of renewable energy output, the large-scale access of distributed renewable energy will bring challenges to the operation and scheduling of distribution networks and multi-microgrid systems. At present, in the scheduling research of distribution networks and multi-microgrids considering the uncertainty of renewable energy, stochastic optimization is difficult to cope with extreme scenarios; the chance constraint method requires a large amount of data and is difficult to solve; robust optimization can achieve robustness of scheduling, but its day-ahead decision is too conservative. In addition, the above scheduling models are all for day-ahead scheduling. Since the prediction error of renewable energy on the day-ahead is larger than that of intraday prediction, the above scheduling models are less accurate in actual scheduling and need to be adjusted based on day-ahead scheduling. Summary of the invention

[0003] The purpose of the present invention is to provide a multi-time scale scheduling method that takes into account the access of large-scale renewable energy, and to perform day-ahead and intraday scheduling of large-scale renewable energy connected to the distribution network and multi-microgrid system. In day-ahead scheduling, the distribution network and multi-microgrid system predict the output of renewable energy in the next 24 hours, and the system dispatcher arranges the system day-ahead scheduling plan based on the day-ahead prediction information; due to the large error of day-ahead prediction, in order to accurately reflect the fluctuation of renewable energy output, intraday real-time scheduling is performed on the basis of day-ahead scheduling. Intraday scheduling rolls and updates the predicted value of renewable energy output in the future several time periods through ultra-short-term (usually less than 1 hour) prediction. Based on the predicted value of intraday output, the dispatch center performs economic scheduling of the real-time output of each controllable unit at the next moment.

[0004] In order to achieve the above object, the present invention provides a multi-time scale scheduling method considering large-scale renewable energy access, which is applicable to distribution networks and multi-microgrid systems, and comprises the following steps:

[0005] S1. Predict the upper and lower limits of the day-ahead output of renewable energy through the probability distribution of the day-ahead forecast error of renewable energy, and construct a robust uncertainty set;

[0006] S2. Construct a two-stage robust dispatch model of the distribution network and the multi-microgrid system; the two-stage robust dispatch model includes a day-ahead stage model and a control stage model; the day-ahead stage model makes a day-ahead dispatch decision based on the day-ahead predicted renewable energy output; the control stage model identifies the worst scenario of renewable energy output based on the day-ahead dispatch decision, and modifies the day-ahead dispatch decision according to the worst scenario;

[0007] S3. Split the day-ahead robust dispatch problem into main sub-problems through the column constraint generation algorithm, and iteratively solve the optimal day-ahead dispatch decision for the distribution network and multi-microgrid system under the worst scenario;

[0008] S4. Based on the optimal day-ahead dispatch decision, a daily rolling optimization algorithm is used to perform rolling optimization on the daily dispatch of the distribution network and the multi-microgrid system, so as to realize multi-time scale dispatch of the distribution network and the multi-microgrid system.

[0009] Optionally, step S1 includes:

[0010] S11. Order is the maximum positive and negative prediction error of the output of the nth renewable energy unit, is the day-ahead output forecast error of the nth renewable energy unit; The probability distribution of is obtained by taking the λ quantile and 1-λ quantile of the cumulative probability density function respectively. Where λ is the preset value;

[0011] S12. Express the uncertainty of renewable energy output as:

[0012]

[0013]

[0014] in is the actual output of the nth renewable energy unit; The day-ahead predicted output of the nth renewable energy unit;

[0015] S13. Order is the robust uncertainty set of the nth renewable energy unit.

[0016] Optionally, the objective function of the day-ahead model is:

[0017]

[0018]

[0019]

[0020] Where i and j represent the distribution network node numbers; T is the total dispatch period; t represents the current dispatch time; ij is the distribution network line between the i-th distribution network node and the j-th distribution network node; f da1 is the day-ahead dispatching cost of the distribution network and multi-microgrid system; da2 is the control cost of the distribution network and multi-microgrid system; x is the day-ahead dispatch decision variable; y is the control decision variable; ξ is the uncertain variable; k is the microgrid number; N MG is the number of microgrids; C tr , C loss , C IDR are the unit power purchase cost of the distribution network from the main grid, the network loss cost of the distribution network, and the load demand response cost of the distribution network node; C re , C es They are the penalty coefficient for abandoning renewable energy in microgrid and the penalty coefficient for energy storage loss cost; Purchase power from the main grid for the distribution network; I is the output of the gas turbine in the kth microgrid; k,t is a variable in [0,1], representing the output state variable of the gas turbine unit in the kth microgrid; D j,t is the shiftable load power of the jth distribution network node at the scheduling time t before demand response; is the shiftable load power of the jth distribution network node at the scheduling time t after demand response; is the abandoned renewable energy power of the kth microgrid; is the square of the distribution network line current; r ij is the line resistance; Ω PE ,Ω TR They are the distribution network line set and the distribution network transformer node set respectively; are the charging power and discharging power of energy storage respectively; a k , b k 、c k are the gas turbine output cost coefficients respectively;

[0021] The objective function of the control phase model is: Under a certain day-ahead dispatch decision, the control cost f of the distribution network and multi-microgrid system under the worst scenario of renewable energy output is achieved. da2 Minimum.

[0022] Optionally, the constraints of the day-ahead model include: main grid output constraints, gas turbine minimum startup / shutdown time constraints, and tie line power transmission constraints;

[0023] The main network output constraints include:

[0024]

[0025]

[0026] in Transmit reactive power to the main grid; They are the upper and lower limits of reactive power transmitted by the main grid respectively; Transmits the upper and lower limits of active power for the main network.

[0027] The gas turbine minimum startup / shutdown time constraint includes:

[0028]

[0029]

[0030]

[0031]

[0032] in, It is the minimum start-up and shutdown time of the unit.

[0033] The tie line power transmission constraints include

[0034]

[0035] in, is the transmission power from the distribution network to the kth microgrid; are the lower and upper limits of the power transmitted between the microgrid and the distribution network, respectively.

[0036] Optional, objective function f of the intraday rolling optimization algorithm di for:

[0037]

[0038]

[0039] Among them, t in is the step size of intraday rolling optimization; PF is the energy storage penalty term; pf is the penalty coefficient; It is the energy storage charge capacity obtained by the day-ahead dispatch.

[0040] Optionally, the constraints of the intraday rolling optimization algorithm are the same as those of the control stage model, including: distribution network flow constraints, distribution network security constraints, static VAR compensator constraints, gas turbine output constraints, microgrid power balance constraints, energy storage charging and discharging constraints, renewable energy abandonment constraints, and demand response load constraints;

[0041] The power flow constraints of the distribution network include:

[0042]

[0043]

[0044]

[0045]

[0046] Among them, i, j, h, k are the node numbers of the distribution network; δ j is the set of lines with the jth distribution network node as the head end; π j is the set of lines ending at the jth distribution network node; Ω PB is the set of power grid nodes; Ω SVC is the reactive compensator set; P ij,t , Q ij,t are the active and reactive power of line ij respectively; x ij is the reactance value of line ij; are the active and reactive load demands of the j-th distribution network node respectively; is the reactive power of the static VAR compensator; is the square of the voltage of the i-th distribution network node and the j-th distribution network node at time t; P jh,t Indicates the active power of line jh; Ω MG represents the microgrid set; Ω TR represents the set of transformer nodes; Q jk,t represents the reactive power of line jk;

[0047] The distribution network security constraints include:

[0048]

[0049]

[0050] in, is the upper limit of the square of the line current; are the upper and lower limits of the node voltage square;

[0051] The static VAR compensator constraints include:

[0052]

[0053] in, The upper and lower limits of the reactive power output of the static VAR compensator installed at the distribution network node j.

[0054] The gas turbine output constraint includes:

[0055]

[0056]

[0057] Among them, RUg ,RD g is the maximum up and down ramp rate of the gas turbine unit; They are the upper and lower limits of the active power output of the gas turbine unit respectively;

[0058] The microgrid power balance constraint includes:

[0059]

[0060] in, The output of renewable energy is uncertain variable; is the load of the kth microgrid;

[0061] The energy storage charging and discharging constraints include:

[0062]

[0063]

[0064]

[0065]

[0066] E k,0 =E k,T

[0067]

[0068] in, are variables in [0,1], representing the energy storage charging state and discharging state respectively; are the charging power and discharging power of energy storage respectively; are the charging power and discharging power of the energy storage of the kth distribution network at time t-1 respectively; They are the upper and lower limits of energy storage charging power and the upper and lower limits of discharging power respectively; E is the charging efficiency and discharging efficiency respectively; k,t is the energy storage charge capacity; E k , are the upper and lower limits of energy storage capacity; Δt represents the length of the dispatch period; E k,0 Indicates the initial charge capacity of energy storage; E k,T Indicates the energy storage charge capacity at time T;

[0069] The renewable energy curtailment constraints include:

[0070]

[0071] in is the maximum power of abandoned renewable energy;

[0072] The demand response load constraint includes:

[0073]

[0074] Among them, D j,t is the shiftable load power of the jth distribution network node in time period t before demand response; is the shiftable load power of the jth distribution network node in time period t after demand response; are the maximum and minimum electricity demands of the demand response load in period t respectively.

[0075] Optionally, step S3 includes:

[0076] S31. Write the day-ahead robust scheduling problem in the following form:

[0077]

[0078] Where x is the day-ahead dispatch decision variable; u is the day-ahead output forecast value of renewable energy; u min 、u max are the upper and lower limits of the forecast value of renewable energy output on the day before; y is the control decision variable such as line power and voltage; hx is the day-ahead dispatch cost; ay is the control cost; h is the day-ahead dispatch cost coefficient matrix; a is the control cost coefficient matrix; A and b are the day-ahead dispatch constraint condition correlation coefficient matrix; C, D, E, G, g, f are the control constraint cost correlation coefficient matrix;

[0079] S32, dividing the day-ahead robust scheduling problem into a main problem and sub-problems;

[0080] The mathematical model of the main problem is:

[0081]

[0082] Among them, f MP is the main problem expression; n is the maximum number of iterations of the column constraint generation algorithm; m is the number of iterations of the column constraint generation algorithm; u m is the predicted value of renewable energy output in the worst scenario of the mth iteration; y m The control decision variable in the worst scenario of the mth iteration; η is the maximum control cost of the main problem in the worst scenario;

[0083] The mathematical model of the sub-problem is:

[0084]

[0085] Among them, f SP is the sub-problem expression; is the solution result of the day-ahead scheduling decision variables of the main problem at the mth iteration.

[0086] Optionally, iteratively solving the optimal day-ahead dispatching scheme of the distribution network and the multi-microgrid system under the severe scenario includes the following steps:

[0087] S34. Set the lower bound L B =-∞, upper bound U B =+∞, the number of iterations of the column constraint generation algorithm m=1.

[0088] S35. Solve the main problem and obtain the optimal solution for the day-ahead scheduling decision Day-ahead dispatch cost And the maximum control cost under the worst scenario of the main problem And update the lower bound

[0089] S36, Fixed day-ahead scheduling decision variables Solve the subproblem; use the Lagrange dual method to dualize the subproblem to obtain the dual problem of the subproblem; solve the dual problem to obtain the day-ahead output value u of renewable energy in the worst scenario in the mth iteration m , and update the upper bound For the dual problem, enter S37;

[0090] S37, if U B -L B <ε, get the optimal solution for day-ahead scheduling decision The iteration ends; if U B -L B ≥ε, enter S38;

[0091] S38. Update the predicted value u of renewable energy output under the worst scenario of the main problem m , and add the constraints of the control phase model of the current worst scenario to the main problem, m=m+1, and return to step S35.

[0092] Optionally, step S4 includes:

[0093] S41, setting the initial scheduling time of the day t=1;

[0094] S42, based on the intraday rolling optimization algorithm to obtain t+1~t+t in The daily renewable energy output forecast value at time t in Step size optimized for intraday rolling;

[0095] S43, based on the day-ahead scheduling decision, based on the objective function and constraints of the intraday rolling optimization algorithm, optimize and solve the time from t+1 to t+t inThe intraday dispatch control variables at time t+1 are used to perform intraday real-time dispatch of the distribution network and the multi-microgrid system; let t=t+1, and repeat steps S42 and S43.

[0096] Optionally, step S42 includes:

[0097] S421, according to the upper and lower limits of the renewable energy output forecast value, the renewable energy output forecast value is divided into the first to nth intervals at equal intervals, corresponding to the discretized n states S1 to S2 respectively. n ;

[0098] S422. Establish a one-step state transfer matrix Π(t, t+1) based on the Markov chain:

[0099]

[0100] Among them, t represents the current time, P a,b Indicates that from state S a Transfer to state S b The probability of , a, b∈[1,n];

[0101] S423, according to the current t The predicted output value of renewable energy at the moment, and the corresponding state S i , i∈[1,n];

[0102] Find the maximum probability P of the i-th row in the one-step state transfer matrix ij ; then in the current t Time state S i In the case of ij Transition to state S j ; Let the average value corresponding to the jth interval be the daily renewable energy output forecast value at time t+1;

[0103] S424, let t = t + 1, keep rolling the time domain, and repeat step S423 until t + t is obtained. in The predicted output value of renewable energy within the day; where t in Step size optimized for intraday rolling.

[0104] Compared with the prior art, the present invention has the following beneficial effects:

[0105] The multi-time scale scheduling method considering large-scale renewable energy access of the present invention can coordinate the day-ahead scheduling and intra-day scheduling, reduce the impact of renewable energy uncertainty on the scheduling of distribution networks and multi-microgrid systems, and improve the scheduling accuracy compared with a single day-ahead scheduling method, thereby effectively improving the ability of distribution networks and multi-microgrid systems to cope with renewable energy uncertainty.

[0106] The present invention provides a scientific scheduling basis for the dispatchers of distribution network and multi-microgrid system by establishing a two-stage robust scheduling model with multiple time scales (including day-ahead scale and intraday scale), thereby realizing the coordinated scheduling of distribution network and multi-microgrid system with multiple time scales. BRIEF DESCRIPTION OF THE DRAWINGS

[0107] In order to more clearly illustrate the technical solution of the present invention, the following briefly introduces the drawings required for the description. Obviously, the drawings described below are an embodiment of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative work:

[0108] Figure 1 A multi-timescale dispatch framework for distribution networks and multi-microgrid systems;

[0109] Figure 2 is a flow chart of the multi-time scale scheduling method of the present invention;

[0110] Figure 3 Solve the flow chart for the day-ahead robust dispatch of distribution networks and multi-microgrid systems;

[0111] Figure 4 Solve the flow chart for the intraday rolling dispatch of distribution network and multi-microgrid system. DETAILED DESCRIPTION

[0112] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0113] The multi-time scale scheduling method of the present invention first establishes a multi-time scale scheduling framework, and divides the scheduling of the distribution network and multi-microgrid system into two stages: day-ahead and intra-day. Figure 1 As shown, the details are as follows:

[0114] Phase 1: Day-ahead dispatch. In day-ahead dispatch, the distribution network and multi-microgrid system predict the output of renewable energy in the next 24 hours. The system dispatcher arranges the system day-ahead dispatch plan based on the day-ahead forecast information. In this phase, the forecast information includes not only the forecast value of renewable energy output in the next 24 hours, but also its forecast upper and lower limits. Considering the large error of day-ahead forecast, robust optimization is used to construct the uncertain set of renewable energy output, and the day-ahead stage model of the distribution network and multi-microgrid system is established. In order to meet the real-time power balance of the system, the day-ahead dispatch decision in this phase only includes the day-ahead start and shutdown plan of the gas turbine, the day-ahead power purchase plan of the distribution network, the day-ahead interconnection line transmission power of the distribution network and multi-microgrid, and other dispatch strategies that need to be formulated one day in advance.

[0115] Phase 2: Intraday short-term dispatch. Due to the large forecast range and low forecast accuracy, the forecast results of renewable energy output one day in advance often cannot meet the requirements of power balance. In order to accurately reflect the fluctuations in renewable energy output and reduce the deviation between advance dispatch and real-time dispatch, intraday short-term dispatch is introduced. In intraday dispatch, ultra-short-term forecasts are made through the control stage model to obtain the forecast values ​​of renewable energy output in the next few time periods. Based on the day-ahead dispatch, the dispatch center will economically dispatch the output of each controllable unit and formulate an output plan for each unit in the next time period. This stage is generally completed 15 minutes to 1 hour in advance, and the intraday dispatch plan is revised and updated according to the ultra-short-term forecast, so it is a rolling dispatch.

[0116] The present invention provides a multi-time scale scheduling method considering large-scale renewable energy access, which is applicable to distribution networks and multi-microgrid systems, such as Figure 2 As shown, it includes the steps:

[0117] S1. Predict the upper and lower limits of the day-ahead output of renewable energy through the probability distribution of the day-ahead forecast error of renewable energy, and construct a robust uncertainty set;

[0118] Step S1 comprises:

[0119] S11. Order is the maximum positive and negative prediction error of the output of the nth renewable energy unit, is the day-ahead output forecast error of the nth renewable energy unit; The probability distribution of is obtained by taking the λ quantile and 1-λ quantile of the cumulative probability density function respectively. Where λ is a preset value; λ can be selected according to the system scheduling requirements. The smaller λ is, the higher the scheduling robustness is.

[0120] Taking the wind and solar power unit as an example, the prediction error of its output satisfies the third-order mixed Gaussian distribution, that is, the function form of the probability density of the prediction error of the wind turbine and photovoltaic maximum output is:

[0121]

[0122] Among them, x is the prediction error value; μ, δ, and α are the expectation, standard deviation, and weight phasor of each group of Gaussian distributions, respectively. i is the weight of the i-th group of Gaussian distribution in the mixed Gaussian distribution; μ i is the expectation of the i-th group of Gaussian distribution of mixed Gaussian distribution; δ i The standard deviation of the ith Gaussian distribution of the mixed Gaussian distribution. i=1,2,3.

[0123] S12. Express the uncertainty of renewable energy output as:

[0124]

[0125]

[0126] in is the actual output of the nth renewable energy unit; The day-ahead predicted output of the nth renewable energy unit;

[0127] S13. Order is the robust uncertainty set of the nth renewable energy unit.

[0128] S2. Construct a two-stage robust dispatching model of the distribution network and the multi-microgrid system; the two-stage robust dispatching model includes a day-ahead stage model and a regulation stage model; the day-ahead stage model formulates a day-ahead dispatching decision based on the day-ahead predicted renewable energy output; the regulation stage model identifies the worst scenario of renewable energy output based on the day-ahead dispatching decision, and modifies the day-ahead dispatching decision according to the worst scenario.

[0129] The worst scenario in scheduling refers to the scenario of renewable energy uncertain variables that maximizes the system scheduling cost under the current day-ahead scheduling decision. According to convex optimization theory, the worst scenario is generally at a vertex in the polyhedron solution space, that is, the scenario where the renewable energy output takes an extreme value.

[0130] The day-ahead-scale scheduling is divided into the day-ahead stage (based on the day-ahead stage model) and the regulation stage (based on the regulation stage model). The objective function of the two-stage robust scheduling model mainly includes the day-ahead scheduling cost and the regulation cost of the distribution network and multi-microgrids.

[0131] The objective function of the day-ahead model mainly includes the day-ahead dispatching cost and regulation cost of the distribution network and multi-microgrid. The objective function of the day-ahead model is:

[0132]

[0133]

[0134]

[0135] Where i and j represent the distribution network node numbers; T is the total dispatch period; t represents the current dispatch time; ij is the distribution network line between the i-th distribution network node and the j-th distribution network node; f da1 is the day-ahead dispatching cost of the distribution network and multi-microgrid system; da2 is the control cost of the distribution network and multi-microgrid system; x is the day-ahead dispatch decision variable; y is the control decision variable; ξ is the uncertain variable; k is the microgrid number; N MG is the number of microgrids; C tr , C loss , C IDR are the unit power purchase cost of the distribution network from the main grid, the network loss cost of the distribution network, and the load demand response cost of the distribution network node; C re , C es They are the penalty coefficient for abandoning renewable energy in microgrid and the penalty coefficient for energy storage loss cost; Purchase power from the main grid for the distribution network; I is the output of the gas turbine in the kth microgrid; k,t is a variable in [0,1], representing the output state variable of the gas turbine unit in the kth microgrid; D j,t is the shiftable load power of the jth distribution network node at the scheduling time t before demand response; is the shiftable load power of the jth distribution network node at the scheduling time t after demand response; is the abandoned renewable energy power of the kth microgrid; is the square of the distribution network line current; r ij is the line resistance; Ω PE ,Ω TR They are the distribution network line set and the distribution network transformer node set respectively; are the charging power and discharging power of energy storage respectively; a k , b k 、c k are the gas turbine output cost coefficients respectively;

[0136] The constraints of the day-ahead model include: main grid output constraints, gas turbine minimum startup / shutdown time constraints, and tie-line power transmission constraints;

[0137] The main network output constraints include:

[0138]

[0139]

[0140] in Transmit reactive power to the main grid; They are the upper and lower limits of reactive power transmitted by the main grid respectively; Transmits the upper and lower limits of active power for the main network.

[0141] The gas turbine minimum startup / shutdown time constraint includes:

[0142]

[0143]

[0144]

[0145]

[0146] in, It is the minimum start-up and shutdown time of the unit.

[0147] The tie line power transmission constraints include

[0148]

[0149] in, is the transmission power from the distribution network to the kth microgrid; are the lower and upper limits of the power transmitted between the microgrid and the distribution network, respectively.

[0150] The objective function of the control phase model is: Under a certain day-ahead dispatch decision, the control cost f of the distribution network and multi-microgrid system under the worst scenario of renewable energy output is achieved. da2 Minimum.

[0151] The constraints of the control stage model include: distribution network flow constraints, distribution network security constraints, static VAR compensator constraints, gas turbine output constraints, microgrid power balance constraints, energy storage charging and discharging constraints, renewable energy abandonment constraints, and demand response load constraints;

[0152] The power flow constraints of the distribution network include:

[0153]

[0154]

[0155]

[0156]

[0157] in, i , j, h, k are the node numbers of the distribution network; δj is the set of lines with the jth distribution network node as the head end; π j is the set of lines ending at the jth distribution network node; Ω PB is the set of power grid nodes; Ω SVC is the reactive power compensator set; P ij,t ,Q ij,t are the active and reactive power of line ij respectively; x ij is the reactance value of line ij; are the active and reactive load demands of the j-th distribution network node respectively; is the reactive power of the static VAR compensator; is the square of the voltage of the i-th distribution network node and the j-th distribution network node at time t; P jh,t Indicates the active power of line jh; Ω MG represents the microgrid set; Ω TR represents the transformer node set; Q jk,t represents the reactive power of line jk;

[0158] The distribution network security constraints include:

[0159]

[0160]

[0161] in, is the upper limit of the square of the line current; are the upper and lower limits of the node voltage square;

[0162] The static VAR compensator constraints include:

[0163]

[0164] in, The upper and lower limits of reactive power output of the static VAR compensator installed at distribution network node j.

[0165] The gas turbine output constraint includes:

[0166]

[0167]

[0168] Among them, RU g ,RD g is the maximum up and down ramp rate of the gas turbine unit; They are the upper and lower limits of the active power output of the gas turbine unit respectively;

[0169] The microgrid power balance constraint includes:

[0170]

[0171] in, The output of renewable energy is uncertain variable; is the load of the kth microgrid;

[0172] The energy storage charging and discharging constraints include:

[0173]

[0174]

[0175]

[0176]

[0177] E k,0 =E k,T

[0178]

[0179] in, are variables in [0,1], representing the energy storage charging state and discharging state respectively; are the charging power and discharging power of energy storage respectively; are the charging power and discharging power of the energy storage of the kth distribution network at time t-1 respectively; They are the upper and lower limits of energy storage charging power and the upper and lower limits of discharging power respectively; E is the charging efficiency and discharging efficiency respectively; k,t is the energy storage charge capacity; E k , are the upper and lower limits of energy storage capacity; Δt represents the length of the dispatch period; E k,0 Indicates the initial charge capacity of energy storage; E k,T Indicates the energy storage charge capacity at time T;

[0180] The renewable energy curtailment constraints include:

[0181]

[0182] in is the maximum power of abandoned renewable energy;

[0183] The demand response load constraints include:

[0184]

[0185] Among them, D j,t is the shiftable load power of the jth distribution network node in time period t before demand response; is the shiftable load power of the jth distribution network node in time period t after demand response; are the maximum and minimum electricity demands of the demand response load in period t respectively.

[0186] S3: The day-ahead robust scheduling problem is split into main and sub-problems through the column-and-constraint generation (C&CG) algorithm, and the optimal day-ahead scheduling decision of the distribution network and multi-microgrid system under the worst scenario is obtained through iterative solution; the main problem solves the optimal day-ahead scheduling plan under the worst scenario and passes it to the sub-problem, which is solved by the duality principle to obtain the control cost under the current worst scenario and the output of renewable energy under the worst scenario, thereby identifying the worst scenario under the day-ahead scheduling plan. By adding the control phase constraints under the worst scenario to the main problem, the decision of the main problem is affected, and finally the optimal solution of the day-ahead robust scheduling problem is obtained through the iteration of the main and sub-problems. The objective function of the main problem includes the day-ahead scheduling cost and the control cost under the worst scenario, which is the same as the day-ahead scheduling model, but the constraints of the main problem include not only the constraints of the day-ahead stage model, but also the constraints of the control phase model under the m worst scenarios generated in the iteration.

[0187] The objective function of the sub-problem is the control cost, which is the same as the objective function of the model in the control stage. The constraints of the sub-problem are the constraints of the model in the control stage. The control cost in the worst scenario under the day-ahead decision of the current main problem is obtained through dual solution.

[0188] Step S3 comprises:

[0189] S31. Based on the objective function and constraints of the day-ahead model, the day-ahead robust scheduling problem is written as follows:

[0190]

[0191] Where x is the day-ahead dispatch decision variable; u is the day-ahead output forecast value of renewable energy; u min 、u max are the upper and lower limits of the forecast value of renewable energy output on the day before; y is the control decision variable such as line power and voltage; hx is the day-ahead dispatch cost; ay is the control cost; h is the day-ahead dispatch cost coefficient matrix; a is the control cost coefficient matrix; A and b are the day-ahead dispatch constraint condition correlation coefficient matrix; C, D, E, G, g, f are the control constraint cost correlation coefficient matrix;

[0192] S32, dividing the day-ahead robust scheduling problem into a main problem and sub-problems;

[0193] The mathematical model of the main problem is:

[0194]

[0195] Among them, f MP is the main problem expression; n is the maximum number of iterations of the column constraint generation algorithm; m is the number of iterations of the column constraint generation algorithm; u m is the predicted value of renewable energy output in the worst scenario of the mth iteration; y m The control decision variable in the worst scenario of the mth iteration; η is the maximum control cost of the main problem in the worst scenario;

[0196] The mathematical model of the sub-problem is:

[0197]

[0198] Among them, f SP is the sub-problem expression; is the solution result of the day-ahead scheduling decision variables of the main problem at the mth iteration.

[0199] like Figure 3 As shown in Figure 1, the iterative solution to obtain the optimal day-ahead dispatching scheme for the distribution network and multi-microgrid system under severe scenarios includes the following steps:

[0200] S34. Set the lower bound L B =-∞, upper bound U B =+∞, the number of iterations of the column constraint generation algorithm m=1.

[0201] S35. Solve the main problem and obtain the optimal solution for the day-ahead scheduling decision Day-ahead dispatch cost And the maximum control cost under the worst scenario of the main problem And update the lower bound

[0202] S36, Fixed day-ahead scheduling decision variables Solve the subproblem; use the Lagrange dual method to dualize the subproblem to obtain the dual problem of the subproblem; solve the dual problem, and if there is a feasible solution, obtain the predicted value u of the renewable energy output in the worst scenario in the mth iteration m , and update the upper bound is the dual problem, and the process goes to S37; if there is no feasible solution to the dual problem, the predicted value u of the renewable energy output in the worst scenario in the mth iteration is obtained. m , enter S38;

[0203] In this embodiment, the dual problem of the subproblem is expressed as:

[0204]

[0205] C T π1+D T π2+E T π3≤a

[0206] π1≤0

[0207] u min ≤u≤u max

[0208] Among them, π1, π3, and π2 are introduced dual variables. According to the duality principle, the solution of the dual problem is the same as that of the original problem, so the subproblem is solved directly by the dual problem. The subproblem can be solved by the dual algorithm to obtain the regulation cost under the current worst scenario and the renewable energy output under the worst scenario. The renewable energy output under the worst scenario can be obtained to identify the worst scenario.

[0209] S37, if U B -L B <ε, get the optimal solution for day-ahead scheduling decision The iteration ends; if U B -L B ≥ε, enter S38;

[0210] S38. Update the predicted value u of renewable energy output under the worst scenario of the main problem m , and add the constraints of the control phase model of the current worst scenario to the main problem, m=m+1, and return to step S35.

[0211] S4: On the basis of the optimal day-ahead dispatching decision, an intraday rolling optimization algorithm is used to perform rolling optimization on the intraday dispatching of the distribution network and the multi-microgrid system, and the intraday mobilization decision of the distribution network and the multi-microgrid system is obtained (in this embodiment, the intraday mobilization decision is used for the dispatching in the next 15 minutes to 1 hour), so as to realize the multi-time scale dispatching of the distribution network and the multi-microgrid system.

[0212] Considering that the prediction accuracy of renewable energy output decreases with the increase of prediction time scale, the two-stage robust scheduling model based on the day-ahead prediction results is difficult to meet the needs of real-time scheduling, so it is necessary to use the intraday rolling optimization algorithm to make adjustments based on the day-ahead scheduling decision.

[0213] The intraday rolling optimization algorithm follows the day-ahead energy storage charging and discharging status, unit start-up and shutdown plan, and interconnection line power obtained in step S3, and adjusts intraday decision variables such as unit output according to the short-term forecast of intraday renewable energy fluctuations.

[0214] The objective function f of the intraday rolling optimization algorithm di for:

[0215]

[0216]

[0217] Among them, t in is the step size of intraday rolling optimization; PF is the energy storage penalty term; pf is the penalty coefficient; It is the energy storage charge capacity obtained by the day-ahead dispatch.

[0218] The constraints of the intraday rolling optimization algorithm are the same as those of the regulation phase model.

[0219] like Figure 4 As shown, step S4 comprises:

[0220] S41, setting the initial scheduling time of the day t=1;

[0221] S42, based on the intraday rolling optimization algorithm to obtain t+1~t+t in The daily renewable energy output forecast value at time t in Step size optimized for intraday rolling;

[0222] S43, based on the day-ahead scheduling decision, based on the objective function and constraints of the intraday rolling optimization algorithm, optimize and solve the time from t+1 to t+t in The intraday dispatch control variables at time t+1 are used to perform intraday real-time dispatch of the distribution network and the multi-microgrid system; let t=t+1, and repeat steps S42 and S43.

[0223] Step S42 comprises:

[0224] S421, according to the upper and lower limits of the renewable energy output forecast value, the renewable energy output forecast value is divided into the first to nth intervals at equal intervals, corresponding to the discretized n states S1 to S2 respectively. n ;

[0225] S422. Establish a one-step state transfer matrix Π(t, t+1) based on the Markov chain:

[0226]

[0227] Among them, t represents the current time, P a,b Indicates that from state S a Transfer to state S b The probability of , a, b∈[1,n];

[0228] S423, according to the current t The predicted output value of renewable energy at the moment, and the corresponding state S i , i∈[1,n];

[0229] Find the maximum probability P of the i-th row in the one-step state transfer matrix ij ; then in the current t Time state S i In the case of ij Transition to state S j ; Let the average value corresponding to the jth interval be the daily renewable energy output forecast value at time t+1;

[0230] S424, let t = t + 1, keep rolling the time domain, and repeat step S423 until t + t is obtained. in The predicted output value of renewable energy within the day; where t in Step size optimized for intraday rolling.

[0231] The multi-time scale scheduling method considering large-scale renewable energy access of the present invention can coordinate the day-ahead scheduling and intra-day scheduling, reduce the impact of renewable energy uncertainty on the scheduling of distribution networks and multi-microgrid systems, and improve the scheduling accuracy compared with a single day-ahead scheduling method, thereby effectively improving the ability of distribution networks and multi-microgrid systems to cope with renewable energy uncertainty.

[0232] The present invention provides a scientific scheduling basis for the dispatchers of distribution network and multi-microgrid system by establishing a two-stage robust scheduling model with multiple time scales (including day-ahead scale and intraday scale), thereby realizing the coordinated scheduling of distribution network and multi-microgrid system with multiple time scales.

[0233] The above is only a specific embodiment of the present invention, but the protection scope of the present invention is not limited thereto. Any technician familiar with the technical field can easily think of various equivalent modifications or replacements within the technical scope disclosed by the present invention, and these modifications or replacements should be included in the protection scope of the present invention. Therefore, the protection scope of the present invention shall be based on the protection scope of the claims.

Claims

1. A multi-time scale scheduling method considering large-scale renewable energy access, applicable to distribution networks and multi-microgrid systems, characterized in that: Contains steps: S1. Predict the upper and lower limits of the day-ahead output of renewable energy through the probability distribution of the day-ahead forecast error of renewable energy, and construct a robust uncertainty set; Step S1 comprises: S11. Order is the maximum positive and negative prediction error of the output of the nth renewable energy unit, is the day-ahead output forecast error of the nth renewable energy unit; The probability distribution of is obtained by taking the λ quantile and 1-λ quantile of the cumulative probability density function respectively. Where λ is the preset value; S12. Express the uncertainty of renewable energy output as: in is the actual output of the nth renewable energy unit; The day-ahead predicted output of the nth renewable energy unit; S13. Order is the robust uncertainty set of the nth renewable energy unit; S2. Construct a two-stage robust dispatching model of the distribution network and the multi-microgrid system; the two-stage robust dispatching model includes a day-ahead stage model and a control stage model; the day-ahead stage model makes a day-ahead dispatching decision based on the predicted day-ahead output of renewable energy; the control stage model is based on the day-ahead dispatching decision, obtains the control cost under the current worst scenario and the output of renewable energy under the worst scenario by solving the control stage model, thereby identifying the worst scenario of renewable energy output, and corrects the day-ahead dispatching decision according to the worst scenario; S3: The day-ahead robust scheduling problem is split into main sub-problems through the column constraint generation algorithm, and the optimal day-ahead scheduling decision of the distribution network and multi-microgrid system under the worst scenario is obtained by iterative solution; S4. Based on the optimal day-ahead dispatch decision, a day-by-day rolling optimization algorithm is used to perform rolling optimization on the day-by-day dispatch of the distribution network and the multi-microgrid system, so as to realize multi-time scale dispatch of the distribution network and the multi-microgrid system; Step S4 comprises: S41, setting the initial scheduling time t=1 within the day; S42, based on the intraday rolling optimization algorithm to obtain t+1~t+t in The daily renewable energy output forecast value at time t in Step size optimized for intraday rolling; Step S42 comprises: S421, according to the upper and lower limits of the renewable energy output forecast value, the renewable energy output forecast value is divided into the first to nth intervals at equal intervals, corresponding to the discretized n states S1 to S2 respectively. n ; S422. Establish a one-step state transfer matrix Π(t, t+1) based on the Markov chain: Among them, t represents the current time, P a,b Indicates that from state S a Transfer to state S b The probability of , a, b∈[1,n]; S423. According to the predicted output value of renewable energy at the current time t, the corresponding state S is obtained. i , i∈[1,n]; Find the maximum probability P of the i-th row in the one-step state transfer matrix ij ; then at the current time t, the state S i In the case of ij Transition to state S j ; Let the average value corresponding to the jth interval be the daily renewable energy output forecast value at time t+1; S424, let t = t + 1, keep rolling the time domain, and repeat step S423 until t + t is obtained. in The predicted output value of renewable energy within the day; where t in Step size optimized for intraday rolling; S43, based on the day-ahead scheduling decision, based on the objective function and constraints of the intraday rolling optimization algorithm, optimize and solve the time from t+1 to t+t in The intraday dispatch control variables at time t+1 are used to perform intraday real-time dispatch of the distribution network and the multi-microgrid system; let t=t+1, and repeat steps S42 and S43.

2. The multi-time scale scheduling method considering large-scale renewable energy access as claimed in claim 1, characterized in that: The objective function of the day-ahead model is: Where i and j represent the distribution network node numbers; T is the total dispatch period; t represents the current dispatch time; ij is the distribution network line between the i-th distribution network node and the j-th distribution network node; f da1 is the day-ahead dispatching cost of the distribution network and multi-microgrid system; da2 is the control cost of the distribution network and multi-microgrid system; x is the day-ahead dispatch decision variable; y is the control decision variable; ξ is the uncertain variable; k is the microgrid number; N MG is the number of microgrids; C tr , C loss , C IDR are the unit power purchase cost of the distribution network from the main grid, the network loss cost of the distribution network, and the load demand response cost of the distribution network node; C re , C es They are the penalty coefficient for abandoning renewable energy in microgrid and the penalty coefficient for energy storage loss cost; Purchase power from the main grid for the distribution network; I is the output of the gas turbine in the kth microgrid; k,t is a variable in [0,1], representing the output state variable of the gas turbine unit in the kth microgrid; D j,t is the shiftable load power of the jth distribution network node at the scheduling time t before demand response; is the shiftable load power of the jth distribution network node at the scheduling time t after demand response; is the abandoned renewable energy power of the kth microgrid; is the square of the distribution network line current; r ij is the line resistance; Ω PE ,Ω TR They are the distribution network line set and the distribution network transformer node set respectively; are the charging power and discharging power of energy storage respectively; a k 、b k 、c k are the gas turbine output cost coefficients respectively; The objective function of the control phase model is: Under a certain day-ahead dispatch decision, the control cost f of the distribution network and multi-microgrid system under the worst scenario of renewable energy output is achieved. da2 Minimum.

3. The multi-time scale scheduling method considering large-scale renewable energy access as claimed in claim 2, characterized in that: The constraints of the day-ahead model include: main grid output constraints, gas turbine minimum startup / shutdown time constraints, and tie-line power transmission constraints; The main network output constraints include: in Transmit reactive power to the main grid; They are the upper and lower limits of reactive power transmitted by the main grid respectively; Transmit active power upper and lower limits for the main network; The gas turbine minimum startup / shutdown time constraint includes: in, The minimum start-up and shutdown time of the unit; The tie line power transmission constraints include in, is the transmission power from the distribution network to the kth microgrid; are the lower and upper limits of the power transmitted between the microgrid and the distribution network, respectively.

4. The multi-time scale scheduling method considering large-scale renewable energy access as claimed in claim 3, characterized in that: The objective function f of the intraday rolling optimization algorithm di for: Among them, t in is the step size of intraday rolling optimization; PF is the energy storage penalty term; pf is the penalty coefficient; It is the energy storage charge capacity obtained by the day-ahead dispatch.

5. The multi-time scale scheduling method considering large-scale renewable energy access as claimed in claim 4, characterized in that: The constraints of the intraday rolling optimization algorithm are the same as those of the control stage model, including: distribution network flow constraints, distribution network security constraints, static VAR compensator constraints, gas turbine output constraints, microgrid power balance constraints, energy storage charging and discharging constraints, renewable energy abandonment constraints, and demand response load constraints. The power flow constraints of the distribution network include: Among them, i, j, h, k are the node numbers of the distribution network; δ j is the set of lines with the jth distribution network node as the head end; π j is the set of lines ending at the jth distribution network node; Ω PB is the set of power grid nodes; Ω SVC is the reactive power compensator set; P ij,t ,Q ij,t are the active and reactive power of line ij respectively; x ij is the reactance value of line ij; are the active and reactive load demands of the j-th distribution network node respectively; is the reactive power of the static VAR compensator; is the square of the voltage of the i-th distribution network node and the j-th distribution network node at time t; P jh,t Indicates the active power of line jh; Ω MG represents the microgrid set; Ω TR represents the transformer node set; Q jk,t represents the reactive power of line jk; The distribution network security constraints include: in, is the upper limit of the square of the line current; are the upper and lower limits of the node voltage square; The static VAR compensator constraints include: in, The upper and lower limits of reactive power output of the static VAR compensator installed at node j of the distribution network; The gas turbine output constraint includes: Among them, RU g ,RD g is the maximum up and down ramp rate of the gas turbine unit; They are the upper and lower limits of the active power output of the gas turbine unit respectively; The microgrid power balance constraint includes: in, The output of renewable energy is uncertain variable; is the load of the kth microgrid; The energy storage charging and discharging constraints include: AND k,0 =And k,T in, are variables in [0,1], representing the energy storage charging state and discharging state respectively; are the charging power and discharging power of energy storage respectively; are the charging power and discharging power of the energy storage of the kth distribution network at time t-1 respectively; They are the upper and lower limits of energy storage charging power and the upper and lower limits of discharging power respectively; E is the charging efficiency and discharging efficiency respectively; k,t is the energy storage charge capacity; E k , are the upper and lower limits of energy storage capacity; Δt represents the length of the dispatch period; E k,0 Indicates the initial charge capacity of energy storage; E k,T Indicates the energy storage charge capacity at time T; The renewable energy curtailment constraints include: in is the maximum power of abandoned renewable energy; The demand response load constraints include: in, Dj,t is the shiftable load power of the jth distribution network node in time period t before demand response; is the shiftable load power of the jth distribution network node in time period t after demand response; are the maximum and minimum electricity demands of the demand response load in period t respectively.

6. The multi-time scale scheduling method considering large-scale renewable energy access as claimed in claim 5, characterized in that: Step S3 comprises: S31. Based on the objective function and constraints of the day-ahead model, the day-ahead robust scheduling problem is written as follows: Where x is the day-ahead dispatch decision variable; u is the day-ahead output forecast value of renewable energy; u min 、u max are the upper and lower limits of the forecast value of renewable energy output on the day before; y is the control decision variable such as line power and voltage; hx is the day-ahead dispatch cost; ay is the control cost; h is the day-ahead dispatch cost coefficient matrix; a is the control cost coefficient matrix; A and b are the day-ahead dispatch constraint condition correlation coefficient matrix; C, D, E, G, g, f are the control constraint cost correlation coefficient matrix; S32, dividing the day-ahead robust scheduling problem into a main problem and sub-problems; The mathematical model of the main problem is: Among them, f MP is the main problem expression; n is the maximum number of iterations of the column constraint generation algorithm; m is the number of iterations of the column constraint generation algorithm; u m is the predicted value of renewable energy output in the worst scenario of the mth iteration; y m The control decision variable in the worst scenario of the mth iteration; η is the maximum control cost of the main problem in the worst scenario; The mathematical model of the sub-problem is: Among them, f SP is the sub-problem expression; is the solution result of the day-ahead scheduling decision variables of the main problem at the mth iteration.

7. The multi-time scale scheduling method considering large-scale renewable energy access as claimed in claim 6, characterized in that: The iterative solution in step S3 to obtain the optimal day-ahead dispatching scheme of the distribution network and the multi-microgrid system under the severe scenario includes the following steps: S34. Set the lower bound L B =-∞, upper bound U B = +∞, the number of iterations of the column constraint generation algorithm m = 1; S35. Solve the main problem and obtain the optimal solution for the day-ahead scheduling decision Day-ahead dispatch cost And the maximum control cost under the worst scenario of the main problem And update the lower bound S36, Fixed day-ahead scheduling decision variables Solve the subproblem; use the Lagrange dual method to dualize the subproblem to obtain the dual problem of the subproblem; solve the dual problem to obtain the day-ahead output value u of renewable energy in the worst scenario in the mth iteration m , and update the upper bound For the dual problem, enter S37; S37, if U B -L B <ε, get the optimal solution for day-ahead scheduling decision The iteration ends; if U B -L B ≥ε, enter S38; S38. Update the predicted value u of renewable energy output under the worst scenario of the main problem m , and add the control phase model constraints under the current worst scenario to the main problem, m=m+1, and return to step S35.

Images