Security constraint unit combination composite decision optimization method for coping with typhoon weather

Abstract

A safety constraint unit combination composite decision optimization method for coping with typhoon weather comprises the following steps: step 1, a typhoon weather scene uncertainty set is constructed based on a state sampling method, and the typhoon weather scene uncertainty set comprises fan output and a line fault state; 2, establishing an SCUC composite decision dynamic game robust optimization model by adopting the wind weather scene uncertainty set obtained in the step 1; and 3, solving the SCUC composite decision dynamic game robust optimization model obtained in the step 2 based on an iterative algorithm to obtain an optimal composite decision. According to the method, through overall consideration of the dynamic interaction relation between different stages and different strategies, the problem that an existing multi-stage SCUC model cannot give consideration to decision economy and safety at the same time is solved, the overall load loss in the extreme weather process can be reduced to the maximum extent at the lowest cost, the recovery speed of system performance is increased, and the system performance recovery efficiency is improved. Therefore, the toughness level of the power system in the face of extreme disasters is effectively improved, and the method provided by the invention is more scientific.

Description

Technical Field

[0001] The present invention relates to the technical field of power systems, and in particular to an optimization dispatching technology for power systems with most equipment failures under typhoon weather, and specifically to a composite decision-making optimization model and method for safety-constrained unit combinations in response to typhoon weather. Background Art

[0002] With the intensification of global climate change and greenhouse effect, the frequency of typhoon weather events has increased significantly, causing serious impacts on the power system. These typhoon weather events not only increase the operating risks of the power system, but also put forward higher requirements on the stability and reliability of power supply. In this context, SCUC, as the cornerstone of power system day-ahead dispatch, faces unprecedented challenges. Traditional SCUC methods cannot fully cope with the uncertainty and complexity brought by typhoon weather. Therefore, it is urgent to study and develop SCUC decision-making methods that can adapt to typhoon weather events.

[0003] In order to improve the disaster resistance of the power grid, reduce the scope of power outages, and reduce the degree of power grid load loss, certain resilience enhancement strategies must be adopted to ensure normal power supply during the disaster. According to the failure and recovery process of the power grid in disaster events, most existing studies divide typhoon weather events into the pre-event prevention stage, the in-event response stage, and the post-event recovery stage, and adopt corresponding response strategies for each stage. The prevention stage refers to the comprehensive analysis of disaster forecast information and the current operating status of the power system before the natural disaster, and then timely adjust the operation mode of the power grid through strategies such as SCUC to avoid power grid failures and reduce fault propagation as much as possible; the response stage refers to the active removal of some loads in natural disasters according to the fault conditions of power grid equipment to minimize the scope of the fault impact; the recovery stage refers to the power grid company restoring normal power supply to power users during or after the disaster by repairing equipment and adjusting the operation mode of the power grid.

[0004] The paper with the document number 10.1109 / TSG.2020.3048234 divides typhoon weather events into multiple fixed stages and proposes a multi-stage SCUC decision-making model. However, this approach ignores the dynamic interaction process between different stages, so that the response strategies such as SCUC, load shedding and fault recovery are regarded as independent decisions in sequence. As a systematic decision-making system, the SCUC for typhoon weather needs to comprehensively consider the dynamic coupling relationship between different stages and different response strategies, and reasonably allocate limited resources to maximize the resilience of the power system. In addition, the multi-stage model based on the penalty function with the cost optimization as the goal is difficult to take into account both the cost target and the load shedding target in the SCUC decision-making process. Summary of the invention

[0005] The purpose of the present invention is to solve the technical problem pointed out in the background technology that the existing multi-stage SCUC model in the prior art literature cannot take into account both the cost target and the load reduction target in the decision-making process, and to provide a safety constraint unit combination composite decision optimization technology for dealing with typhoon weather.

[0006] In order to solve the above technical problems, the technical solution adopted by the present invention is:

[0007] A composite decision optimization method for safety-constrained unit commitment in response to typhoon weather comprises the following steps:

[0008] Step 1: Construct a typhoon weather scene uncertainty set based on the state sampling method. The typhoon weather scene uncertainty set includes wind turbine output and line fault status;

[0009] Step 2: Use the uncertain set of wind weather scenarios obtained in step 1 to establish a SCUC composite decision-making dynamic game robust optimization model;

[0010] Step 3: Solve the SCUC compound decision dynamic game robust optimization model obtained in step 2 based on the iterative algorithm to obtain the optimal compound decision.

[0011] In step 1, the following sub-steps are included:

[0012] Step 1.1: Construct a Rankine-based wind field model to fit the wind speed of each transmission line section and wind turbine at any time;

[0013] Step 1.2: Establish a vulnerability model of line failure probability and a wind turbine output model respectively, and use the wind speed obtained in step 1.1 to solve the uncertainty set of each line failure probability and each wind turbine output at different times;

[0014] Step 1.3: Using the failure probability of each line at different times obtained in step 1.2, a line fault state scenario set is established based on the state sampling method;

[0015] The uncertainty set of wind turbine output at different times obtained in step 1.2 and the line fault state scenario set selected in step 1.3 are the typhoon weather scenario uncertainty set.

[0016] In step 1.1, under typhoon disaster events, the uncertainty of grid component failures is a random event related to the spatiotemporal changes and impacts of typhoons; transmission lines face the risk of failures caused by high wind speeds during typhoons, which leads to line power flow fluctuations; therefore, it is necessary to consider the impact of typhoons on wind turbines and transmission lines. The wind speed V in each region can be calculated using the improved Rankine wind field model:

[0017]

[0018] Where V represents the wind speed in each area; T typ represents the adjustment parameter of wind speed distribution; r represents the distance from the typhoon center; V max and R max Respectively represent the maximum wind speed and maximum wind speed radius of the typhoon; K V represents the correction factor;

[0019] In step 1.2, the output of the wind turbine generator depends on its rated power and the real-time wind speed; when considering the cut-in and cut-out scenarios of the wind turbine generator, the output of the generator can be approximated by the following piecewise linearization method:

[0020]

[0021] Where P W′ Indicates the active output power of the fan; C w is the output coefficient of the fan; Represents the maximum output power of the fan; V in 、V out and V e They are the cut-in wind speed, cut-out wind speed and rated wind speed of the fan;

[0022] Combined with formula (2), we can get the uncertainty set of each fan output:

[0023]

[0024] Where P W represents the actual output power of the fan; α1 and α2 represent the upper and lower limit coefficients of the uncertainty interval of the fan output respectively;

[0025] Under the influence of typhoons, the power grid needs to consider transmission line failures. The relationship between line failure rate and typhoon wind speed is expressed by the vulnerability curve, and its specific expression is as follows:

[0026]

[0027] In the formula, p f represents the failure rate of the transmission line; v represents the maximum wind speed that the line can withstand; β represents the coefficient of the fragility curve.

[0028] In step 1.3, assume that the system consists of m lines, S k represents the state of line k, then z=(S1,...,S k ,...,S m ); the system state z depends on the state combination of the lines, so as long as the state of each line is determined, the system state x is also determined accordingly;

[0029] The line status can be obtained by randomly sampling the probability characteristics of the line, and the probability characteristics of the line can be described by a uniform distribution between [0,1]. Assuming that each line has only two states, fault and operation, and that line faults are independent events, a uniformly distributed random number U is drawn with a value interval of [0,1]. k , combined with formula (4), we have:

[0030]

[0031] Draw m random numbers U1,...,U k ,...,U m , we can determine a system state z=(S1,…,,S k ,…,,S m ), repeat the above process N times, and we can get a scene set Z={z1,z2,…,z N}.

[0032] In step 2, the following sub-steps are included:

[0033] Step 2.1: Establish the SCUC decision-making subject model and the line restoration and load shedding slave models respectively;

[0034] Step 2.2: Build a DAD model to search the scene uncertainty set obtained in step 1 to obtain the worst failure scenario;

[0035] Step 2.3: Use the various master-slave models obtained in step 2.1 and the worst failure scenario obtained in step 2.2 to establish a SCUC composite decision-making dynamic game robust optimization model.

[0036] In step 2.1: Establish the SCUC decision-making subject model:

[0037] The objective function of the SCUC decision-making subject model is to minimize the total cost C, which includes the start-up and shutdown cost C of the unit. onoff and operating cost C ope and the recovery cost C rec ;

[0038]

[0039] in:

[0040]

[0041] Where: represents the active output of unit i at time t; represents the start and stop status of unit i at time t; N G is the total number of system units; NL is the total number of system lines; (a i ,b i ,c i ) is the operating cost parameter of unit i; are the start-up and shutdown costs of unit i respectively; (V i,t ,W i,t ) are the start-up and shutdown states of unit i at time t, respectively; h l,t represents the maintenance status of line l at time t;

[0042] The constraints of the SCUC strategy model mainly include power balance constraints, line flow constraints, unit output constraints, spinning reserve constraints, unit ramp constraints and minimum start and stop time constraints.

[0043] 1) Power balance constraints

[0044]

[0045] Where: P t W represents the output of the fan at time t; represents the load of node d at time t; represents the load shedding amount of node d at time t; N D is the total number of system nodes.

[0046] 2) Line flow constraints

[0047]

[0048] Where: S l,t is the state of line l at time t; K W,SF and Respectively represent the network transfer factors of thermal power units, wind power units and loads; P lmax is the upper limit of the power flow of line l;

[0049] 3) Unit output constraints

[0050]

[0051] Where: and They represent the lower and upper limits of the output of unit i respectively;

[0052] 4) Spinning reserve constraints

[0053]

[0054] Where: The system positive spinning reserve demand caused by load forecast error; and They are the positive and negative spinning reserve demands of the system caused by the wind turbine output prediction error;

[0055] 5) Unit climbing constraints

[0056]

[0057] Where: △P i G,up and △P i G,down They represent the upward climbing and downward climbing constraints of unit i respectively;

[0058] 6) Minimum start and stop time constraints

[0059]

[0060] Where: and Respectively represent the continuous start and stop time of unit i; T i G,UP and T i G,DOWN They represent the minimum continuous start and stop time of unit i respectively;

[0061] In step 2.1: Establish the line recovery slave model:

[0062] The main task of the power system restoration phase is to repair the faulty transmission lines and restore the load supply as quickly and efficiently as possible. For the repair of transmission line faults, in order to ensure the safety of maintenance personnel, it is stipulated that the repair work can only be started after the typhoon no longer affects the outage line; the objective function of the lower-level restoration model is to maximize the load supply, and its constraints include power balance, line flow, and restoration resource constraints;

[0063]

[0064] Where: represents the load supply capacity of node d at time t.

[0065] 1) Power balance constraints

[0066]

[0067] 2) Line flow constraints

[0068]

[0069] 3) Restore resource constraints

[0070]

[0071] Where: H maxis the maximum number of lines that can be repaired simultaneously; r l,t 1 means that line l starts to be repaired at time t; T REP The time required for line repair.

[0072] Among them, since the line repair is done within a certain period of time T REP Completed within, so r l,t and h l,t The relationship is shown in formula (22):

[0073]

[0074] At the same time, the system state S l,t And the fault state z l,t With r l,t The relationship is shown in formula (23):

[0075]

[0076] In step 2.1: Establish the load shedding slave model:

[0077] During the fault evolution process, the power system needs to reallocate power generation tasks or perform emergency load shedding after the fault occurs to meet the static safety constraints of the system; in the risk assessment of the power system, the optimal power flow model based on DC power flow is usually used to quickly calculate the load shedding amount. In this patent, another sub-problem with the goal of minimizing the load shedding amount is established at the lower level;

[0078]

[0079] The constraints of the load shedding model include power balance constraints, line flow constraints, unit output constraints, spinning reserve constraints, and unit ramp constraints as shown in equations (10)-(14), and the load shedding constraints are shown in the following equation:

[0080]

[0081] In step 2.2, the most severe failure scenario is searched using the classic DAD model based on robust optimization without considering additional measures to enhance source-load flexibility, which is described as follows:

[0082]

[0083] Where ξ1,ξ2 are the feasible domains of decision variables before and during the disaster, VOLL d is the load shedding penalty cost of node d.

[0084] In step 2.3: Establish the SCUC compound decision dynamic game robust optimization model:

[0085] Based on the above description of the SCUC composite decision model, line restoration and load shedding are both based on the start-stop plan decided by SCUC, and the decision plans of line restoration and load shedding in turn affect SCUC; this decision-making process conforms to the situation of leader-follower game; therefore, by taking SCUC as the leader and line restoration and load shedding as different followers, a Stackelberg game model with a single leader and multiple followers can be established, and its game relationship is as follows:

[0086] 1) SCUC, as the leader, from the upper strategy space Ω SCUC Choose a decision strategy Based on information such as network parameters, load, and distributed renewable energy output, it is then passed on to the lower layers;

[0087] 2) Line restoration and load shedding As a follower, adjust the restoration and load shedding plan within a certain range. Based on the published strategy, from the lower strategy space Ω line,load Select the best response strategy

[0088] 3) SCUC obtains the most cost-optimal decision based on the response strategy of feedback. This iterative game process continues until equilibrium is reached;

[0089] The above model can be expressed in the standard form of Stackelberg game as:

[0090] G={M;U G ; {h,△P D};C;{H,L}} (27)

[0091] Where M is the set of three participants in the game.

[0092] In step 3, the details are as follows:

[0093] Step 3-1: Initialize the original data and parameters, including the expected accident information, network parameters, initial start-stop plan and other necessary parameters;

[0094] Step 3-2: Calculate the optimal line recovery strategy under the worst line outage scenario through the unit start-stop plan, and update the line scenario to solve the optimal load shedding plan;

[0095] Step 3-3: Determine whether the wheel load shedding scheme is better than the optimal load shedding scheme. If so, replace the optimal load shedding scheme with the wheel load shedding scheme.

[0096] Step 3-4: Based on the line scenario and load shedding plan obtained in step 3-2, solve the SCUC model to obtain the optimal start-stop plan;

[0097] Step 3-5: Determine whether the equilibrium state is reached. If so, go to step 3-6, otherwise go to step 3-2;

[0098] Step 3-6: Output the equilibrium solution and obtain the optimal solution of SCUC composite decision;

[0099] When all participants cannot gain greater benefits by changing their strategies, it indicates that the game has reached the Stackelberg equilibrium; at this time, the equilibrium solution of the SCUC compound decision-making master-slave game model is It can be expressed as:

[0100]

[0101] A method for constructing a SCUC composite decision-making dynamic game robust optimization model includes the following sub-steps:

[0102] Step S1: Establish the SCUC decision-making subject model and the line restoration and load shedding slave models respectively;

[0103] Step S2: construct a DAD model to search the scene uncertainty set and obtain the worst failure scenario;

[0104] Step S3: Using various master-slave models obtained in step S1 and the worst failure scenario obtained in step S2, a SCUC composite decision-making dynamic game robust optimization model is established.

[0105] In step S1: Establish the SCUC decision-making subject model:

[0106] The objective function of the SCUC decision-making subject model is to minimize the total cost C, which includes the start-up and shutdown cost C of the unit. onoff and operating cost C ope and the recovery cost C rec ;

[0107]

[0108] in:

[0109]

[0110] Where: represents the active output of unit i at time t; represents the start and stop status of unit i at time t; N G is the total number of system units; N L is the total number of system lines; (a i ,b i ,c i ) is the operating cost parameter of unit i; are the start-up and shutdown costs of unit i respectively; (Vi,t ,W i,t ) are the start-up and shutdown states of unit i at time t, respectively; h l,t represents the maintenance status of line l at time t;

[0111] The constraints of the SCUC strategy model mainly include power balance constraints, line flow constraints, unit output constraints, spinning reserve constraints, unit ramp constraints and minimum start and stop time constraints.

[0112] 1) Power balance constraints

[0113]

[0114] Where: P t W represents the output of the fan at time t; represents the load of node d at time t; represents the load shedding amount of node d at time t; N D is the total number of system nodes.

[0115] 2) Line flow constraints

[0116]

[0117] Where: S l,t is the state of line l at time t; K W,SF and Respectively represent the network transfer factors of thermal power units, wind power units and loads; P lmax is the upper limit of the power flow of line l;

[0118] 3) Unit output constraints

[0119]

[0120] Where: and They represent the lower and upper limits of the output of unit i respectively;

[0121] 4) Spinning reserve constraints

[0122]

[0123] Where: The system positive spinning reserve demand caused by load forecast error; and They are the positive and negative spinning reserve demands of the system caused by the wind turbine output prediction error;

[0124] 5) Unit climbing constraints

[0125]

[0126] Where: △P i G,up and △P i G,down They represent the upward climbing and downward climbing constraints of unit i respectively;

[0127] 6) Minimum start and stop time constraints

[0128]

[0129] Where: and Respectively represent the continuous start and stop time of unit i; T i G,UP and T i G,DOWN They represent the minimum continuous start and stop time of unit i respectively;

[0130] In step S1: establish a line recovery slave model:

[0131] The main task of the power system restoration phase is to repair the faulty transmission lines and restore the load supply as quickly and efficiently as possible. For the repair of transmission line faults, in order to ensure the safety of maintenance personnel, it is stipulated that the repair work can only be started after the typhoon no longer affects the outage line; the objective function of the lower-level restoration model is to maximize the load supply, and its constraints include power balance, line flow, and restoration resource constraints;

[0132]

[0133] Where: represents the load supply capacity of node d at time t.

[0134] 1) Power balance constraints

[0135]

[0136] 2) Line flow constraints

[0137]

[0138] 3) Restore resource constraints

[0139]

[0140] Where: H max is the maximum number of lines that can be repaired simultaneously; r l,t 1 means that line l starts to be repaired at time t; T REP The time required for line repair.

[0141] Among them, since the line repair is done within a certain period of time T REP Completed within, so r l,t and h l,t The relationship is shown in formula (22):

[0142]

[0143] At the same time, the system state S l,t And the fault state z l,t With r l,t The relationship is shown in formula (23):

[0144]

[0145] In step S1: establish a load shedding slave model:

[0146] During the fault evolution process, the power system needs to reallocate power generation tasks or perform emergency load shedding after the fault occurs to meet the static safety constraints of the system; in the risk assessment of the power system, the optimal power flow model based on DC power flow is usually used to quickly calculate the load shedding amount. In this patent, another sub-problem with the goal of minimizing the load shedding amount is established at the lower level;

[0147]

[0148] The constraints of the load shedding model include power balance constraints, line flow constraints, unit output constraints, spinning reserve constraints, and unit ramp constraints as shown in equations (10)-(14), and the load shedding constraints are shown in the following equation:

[0149]

[0150] In step S2, the most severe failure scenario is searched using the classic DAD model based on robust optimization without considering additional measures to enhance source-load flexibility, which is described as follows:

[0151]

[0152] Where ξ1,ξ2 are the feasible domains of decision variables before and during the disaster, VOLL d is the load shedding penalty cost of node d.

[0153] In step S3: Establish the SCUC composite decision-making dynamic game robust optimization model:

[0154] Based on the above description of the SCUC composite decision model, line restoration and load shedding are both based on the start-stop plan decided by SCUC, and the decision plans of line restoration and load shedding in turn affect SCUC; this decision-making process conforms to the situation of leader-follower game; therefore, by taking SCUC as the leader and line restoration and load shedding as different followers, a Stackelberg game model with a single leader and multiple followers can be established, and its game relationship is as follows:

[0155] 1) SCUC, as the leader, from the upper strategy space Ω SCUC Choose a decision strategy Based on information such as network parameters, load, and distributed renewable energy output, it is then passed on to the lower layers;

[0156] 2) Line restoration and load shedding As a follower, adjust the restoration and load shedding plan within a certain range. Based on the published strategy, from the lower strategy space Ω line,load Select the best response strategy

[0157] 3) SCUC obtains the most cost-optimal decision based on the response strategy of feedback. This iterative game process continues until equilibrium is reached;

[0158] The above model can be expressed in the standard form of Stackelberg game as:

[0159] G={M;U G ; {h,△P D};C;{H,L}} (27)

[0160] Where M is the set of three participants in the game.

[0161] Compared with the prior art, the present invention has the following technical effects:

[0162] 1) The present invention can effectively take into account the dynamic coupling relationship between different stages and different response strategies under typhoon weather, and reduces the load loss by 62.9% compared with the prior art literature pointed out in the background technology, so it has higher practicality.

[0163] 2) The line restoration slave model proposed in the present invention can fully tap the response potential of the multiple resources of the power system, maximize the value of various resources and obtain the overall optimal restoration plan, which greatly improves the recovery capacity of the power system after typhoon disaster events compared with traditional methods.

[0164] 3) The SCUC main model and load shedding slave model proposed in the present invention can take into account the cost target and load shedding target in the SCUC decision-making process through dynamic game, optimize system resource allocation, thereby reducing overall load loss and improving the flexibility and resilience of the system in the face of typhoon weather events.

[0165] 4) The SCUC composite decision-making dynamic game robust optimization model proposed in this invention takes into account the dynamic coupling relationship between different strategies. Compared with the existing multi-stage SCUC model with a single response strategy, it can minimize the overall load loss, speed up the recovery of system performance, and improve the resilience of the power system under typhoon events. BRIEF DESCRIPTION OF THE DRAWINGS

[0166] The present invention will be further described below in conjunction with the accompanying drawings and embodiments:

[0167] Figure 1 This is a flow chart of the composite decision-making optimization method for safety constraint unit combination in response to typhoon weather according to the present invention;

[0168] Figure 2 It is a flow chart of solving the robust optimization model of SCUC composite decision-making dynamic game based on iterative algorithm of the present invention;

[0169] Figure 3 This is a comparison diagram of the toughness curves simulated in the embodiment of the present invention;

[0170] Figure 4 This is a comparison diagram of the start-stop scheme of the simulated unit according to an embodiment of the present invention;

[0171] Figure 5 This is a comparison diagram of the simulation line recovery solutions according to the embodiments of the present invention. DETAILED DESCRIPTION

[0172] A composite decision optimization method for safety-constrained unit commitment in response to typhoon weather comprises the following steps:

[0173] Step 1: Construct a typhoon weather scene uncertainty set based on the state sampling method. The typhoon weather scene uncertainty set includes wind turbine output and line fault status;

[0174] Step 2: Use the uncertain set of wind weather scenarios obtained in step 1 to establish a SCUC composite decision-making dynamic game robust optimization model;

[0175] Step 3: Solve the SCUC compound decision dynamic game robust optimization model obtained in step 2 based on the iterative algorithm to obtain the optimal compound decision.

[0176] In step 1, the following sub-steps are included:

[0177] Step 1.1: Construct a Rankine-based wind field model to fit the wind speed of each transmission line section and wind turbine at any time;

[0178] Step 1.2: Establish a vulnerability model of line failure probability and a wind turbine output model respectively, and use the wind speed obtained in step 1.1 to solve the uncertainty set of each line failure probability and each wind turbine output at different times;

[0179] Step 1.3: Using the failure probability of each line at different times obtained in step 1.2, a line fault state scenario set is established based on the state sampling method;

[0180] The uncertainty set of wind turbine output at different times obtained in step 1.2 and the line fault state scenario set selected in step 1.3 are the typhoon weather scenario uncertainty set.

[0181] In step 1.1, under typhoon disaster events, the uncertainty of grid component failures is a random event related to the spatiotemporal changes and impacts of typhoons; transmission lines face the risk of failures caused by high wind speeds during typhoons, which leads to line power flow fluctuations; therefore, it is necessary to consider the impact of typhoons on wind turbines and transmission lines. The wind speed V in each region can be calculated using the improved Rankine wind field model:

[0182]

[0183] Where, T typ represents the adjustment parameter of wind speed distribution; r represents the distance from the typhoon center; V max and R max Respectively represent the maximum wind speed and maximum wind speed radius of the typhoon; K V represents the correction factor;

[0184] In step 1.2, the output of the wind turbine generator depends on its rated power and the real-time wind speed; when considering the cut-in and cut-out scenarios of the wind turbine generator, the output of the generator can be approximated by the following piecewise linearization method:

[0185]

[0186] Where P W ′ represents the active output power of the fan; C w is the output coefficient of the fan; Represents the maximum output power of the fan; V in 、V out and V e They are the cut-in wind speed, cut-out wind speed and rated wind speed of the fan;

[0187] Combined with formula (2), we can get the uncertainty set of each fan output:

[0188]

[0189] Where P W represents the actual output power of the fan; α1 and α2 represent the upper and lower limit coefficients of the uncertainty interval of the fan output respectively;

[0190] Under the influence of typhoons, the power grid needs to consider transmission line failures. The relationship between line failure rate and typhoon wind speed is expressed by the vulnerability curve, and its specific expression is as follows:

[0191]

[0192] In the formula, p f represents the failure rate of the transmission line; v represents the maximum wind speed that the line can withstand; β represents the coefficient of the fragility curve. In step 1.3, assuming that the system consists of m lines, S k represents the state of line k, then z=(S1,...,S k ,...,S m ); the system state z depends on the state combination of the lines, so as long as the state of each line is determined, the system state x is also determined accordingly;

[0193] The line status can be obtained by randomly sampling the probability characteristics of the line, and the probability characteristics of the line can be described by a uniform distribution between [0,1]. Assuming that each line has only two states, fault and operation, and that line faults are independent events, a uniformly distributed random number U is drawn with a value interval of [0,1]. k , combined with formula (4), we have:

[0194]

[0195] Draw m random numbers U1,...,U k ,...,U m , we can determine a system state z=(S1,…,S k ,…,S m ), repeat the above process N times, and we can get a scene set Z={z1,z2,…,z N}.

[0196] In step 2, the following sub-steps are included:

[0197] Step 2.1: Establish the SCUC decision-making subject model and the line restoration and load shedding slave models respectively;

[0198] Step 2.2: Build a DAD model to search the scene uncertainty set obtained in step 1 to obtain the worst failure scenario;

[0199] Step 2.3: Use the various master-slave models obtained in step 2.1 and the worst failure scenario obtained in step 2.2 to establish a SCUC composite decision-making dynamic game robust optimization model.

[0200] In step 2.1: Establish the SCUC decision-making subject model:

[0201] The objective function of the SCUC decision-making subject model is to minimize the total cost C, which includes the start-up and shutdown cost C of the unit. onoff and operating cost C ope and the recovery cost C rec ;

[0202]

[0203] in:

[0204]

[0205] Where: represents the active output of unit i at time t; represents the start and stop status of unit i at time t; N G is the total number of system units; N L is the total number of system lines; (a i ,b i ,c i ) is the operating cost parameter of unit i; are the start-up and shutdown costs of unit i respectively; (V i,t ,W i,t ) are the start-up and shutdown states of unit i at time t, respectively; h l,t represents the maintenance status of line l at time t;

[0206] The constraints of the SCUC strategy model mainly include power balance constraints, line flow constraints, unit output constraints, spinning reserve constraints, unit ramp constraints and minimum start and stop time constraints.

[0207] 1) Power balance constraints

[0208]

[0209] Where: P t W represents the output of the fan at time t; represents the load of node d at time t; represents the load shedding amount of node d at time t; N D is the total number of system nodes.

[0210] 2) Line flow constraints

[0211]

[0212] Where: S l,t is the state of line l at time t; K W,SF and Respectively represent the network transfer factors of thermal power units, wind power units and loads; P lmax is the upper limit of the power flow of line l;

[0213] 3) Unit output constraints

[0214]

[0215] Where: and They represent the lower and upper limits of the output of unit i respectively;

[0216] 4) Spinning reserve constraints

[0217]

[0218] Where: S t D The system positive spinning reserve demand caused by load forecast error; and They are the positive and negative spinning reserve demands of the system caused by the wind turbine output prediction error;

[0219] 5) Unit climbing constraints

[0220]

[0221] Where: △P i G,up and △P i G,down They represent the upward climbing and downward climbing constraints of unit i respectively;

[0222] 6) Minimum start and stop time constraints

[0223]

[0224] Where: and Respectively represent the continuous start and stop time of unit i; T i G,UP and T i G,DOWN They represent the minimum continuous start and stop time of unit i respectively;

[0225] In step 2.1: Establish the line recovery slave model:

[0226] The main task of the power system restoration phase is to repair the faulty transmission lines and restore the load supply as quickly and efficiently as possible. For the repair of transmission line faults, in order to ensure the safety of maintenance personnel, it is stipulated that the repair work can only be started after the typhoon no longer affects the outage line; the objective function of the lower-level restoration model is to maximize the load supply, and its constraints include power balance, line flow, and restoration resource constraints;

[0227]

[0228] Where: represents the load supply capacity of node d at time t.

[0229] 1) Power balance constraints

[0230]

[0231] 2) Line flow constraints

[0232]

[0233] 3) Restore resource constraints

[0234]

[0235] Where: H max is the maximum number of lines that can be repaired simultaneously; r l,t 1 means that line l starts to be repaired at time t; T REP The time required for line repair.

[0236] Among them, since the line repair is done within a certain period of time T REP Completed within, so r l,t and h l,t The relationship is shown in formula (22):

[0237]

[0238] At the same time, the system state S l,t And the fault state z l,t With r l,t The relationship is shown in formula (23):

[0239]

[0240] In step 2.1: Establish the load shedding slave model:

[0241] During the fault evolution process, the power system needs to reallocate power generation tasks or perform emergency load shedding after the fault occurs to meet the static safety constraints of the system; in the risk assessment of the power system, the optimal power flow model based on DC power flow is usually used to quickly calculate the load shedding amount. In this patent, another sub-problem with the goal of minimizing the load shedding amount is established at the lower level;

[0242]

[0243] The constraints of the load shedding model include power balance constraints, line flow constraints, unit output constraints, spinning reserve constraints, and unit ramp constraints as shown in equations (10)-(14), and the load shedding constraints are shown in the following equation:

[0244]

[0245] In step 2.2, the most severe failure scenario is searched using the classic DAD model based on robust optimization without considering additional measures to enhance source-load flexibility, which is described as follows:

[0246]

[0247] Where ξ1,ξ2 are the feasible domains of decision variables before and during the disaster, VOLL d is the load shedding penalty cost of node d.

[0248] In step 2.3: Establish the SCUC compound decision dynamic game robust optimization model:

[0249] Based on the above description of the SCUC composite decision model, line restoration and load shedding are both based on the start-stop plan decided by SCUC, and the decision plans of line restoration and load shedding in turn affect SCUC; this decision-making process conforms to the situation of leader-follower game; therefore, by taking SCUC as the leader and line restoration and load shedding as different followers, a Stackelberg game model with a single leader and multiple followers can be established, and its game relationship is as follows:

[0250] 1) SCUC, as the leader, from the upper strategy space Ω SCUC Choose a decision strategy Based on information such as network parameters, load, and distributed renewable energy output, it is then passed on to the lower layers;

[0251] 2) Line restoration and load shedding As a follower, adjust the restoration and load shedding plan within a certain range. Based on the published strategy, from the lower strategy space Ω line,load Select the best response strategy

[0252] 3) SCUC obtains the most cost-optimal decision based on the response strategy of feedback. This iterative game process continues until equilibrium is reached;

[0253] The above model can be expressed in the standard form of Stackelberg game as:

[0254] G={M;U G ; {h,△P D};C;{H,L}} (27)

[0255] Where M is the set of three participants in the game.

[0256] In step 3, the details are as follows:

[0257] Step 3-1: Initialize the original data and parameters, including the expected accident information, network parameters, initial start-stop plan and other necessary parameters;

[0258] Step 3-2: Calculate the optimal line recovery strategy under the worst line outage scenario through the unit start-stop plan, and update the line scenario to solve the optimal load shedding plan;

[0259] Step 3-3: Determine whether the wheel load shedding scheme is better than the optimal load shedding scheme. If so, replace the optimal load shedding scheme with the wheel load shedding scheme.

[0260] Step 3-4: Based on the line scenario and load shedding plan obtained in step 3-2, solve the SCUC model to obtain the optimal start-stop plan;

[0261] Step 3-5: Determine whether the equilibrium state is reached. If so, go to step 3-6, otherwise go to step 3-2;

[0262] Step 3-6: Output the equilibrium solution and obtain the optimal solution of SCUC composite decision;

[0263] When all participants cannot gain greater benefits by changing their strategies, it indicates that the game has reached the Stackelberg equilibrium; at this time, the equilibrium solution of the SCUC compound decision-making master-slave game model is It can be expressed as:

[0264]

[0265] The present invention also includes a method for constructing a SCUC composite decision-making dynamic game robust optimization model.

[0266] This technology can solve the technical problem that the existing technology cannot take into account both the cost target and the load reduction target in the decision-making process when making SCUC decisions in typhoon weather;

[0267] It includes the following sub-steps:

[0268] Step S1: Establish the SCUC decision-making subject model and the line restoration and load shedding slave models respectively;

[0269] Step S2: construct a DAD model to search the scene uncertainty set and obtain the worst failure scenario;

[0270] Step S3: Using various master-slave models obtained in step S1 and the worst failure scenario obtained in step S2, a SCUC composite decision-making dynamic game robust optimization model is established.

[0271] In step S1: Establish the SCUC decision-making subject model:

[0272] The objective function of the SCUC decision-making subject model is to minimize the total cost C, which includes the start-up and shutdown cost C of the unit. onoff and operating cost C ope and the recovery cost C rec ;

[0273]

[0274] in:

[0275]

[0276] Where: represents the active output of unit i at time t; represents the start and stop status of unit i at time t; N G is the total number of system units; N L is the total number of system lines; (a i ,b i ,c i ) is the operating cost parameter of unit i; are the start-up and shutdown costs of unit i respectively; (V i,t ,W i,t ) are the start-up and shutdown states of unit i at time t, respectively; h l,t represents the maintenance status of line l at time t;

[0277] The constraints of the SCUC strategy model mainly include power balance constraints, line flow constraints, unit output constraints, spinning reserve constraints, unit ramp constraints and minimum start and stop time constraints.

[0278] 1) Power balance constraints

[0279]

[0280] Where: P t W represents the output of the fan at time t; represents the load of node d at time t; represents the load shedding amount of node d at time t; N D is the total number of system nodes.

[0281] 2) Line flow constraints

[0282]

[0283] Where: S l,t is the state of line l at time t; K W,SF and Respectively represent the network transfer factors of thermal power units, wind power units and loads; P lmax is the upper limit of the power flow of line l;

[0284] 3) Unit output constraints

[0285]

[0286] Where: and They represent the lower and upper limits of the output of unit i respectively;

[0287] 4) Spinning reserve constraints

[0288]

[0289] Where: The system positive spinning reserve demand caused by load forecast error; and They are the positive and negative spinning reserve demands of the system caused by the wind turbine output prediction error;

[0290] 5) Unit climbing constraints

[0291]

[0292] Where: △P i G,up and △P i G,down They represent the upward climbing and downward climbing constraints of unit i respectively;

[0293] 6) Minimum start and stop time constraints

[0294]

[0295] Where: and Respectively represent the continuous start and stop time of unit i; T i G,UP and T iG,DOWN They represent the minimum continuous start and stop time of unit i respectively;

[0296] In step S1: establish a line recovery slave model:

[0297] The main task of the power system restoration phase is to repair the faulty transmission lines and restore the load supply as quickly and efficiently as possible. For the repair of transmission line faults, in order to ensure the safety of maintenance personnel, it is stipulated that the repair work can only be started after the typhoon no longer affects the outage line; the objective function of the lower-level restoration model is to maximize the load supply, and its constraints include power balance, line flow, and restoration resource constraints;

[0298]

[0299] Where: represents the load supply capacity of node d at time t.

[0300] 1) Power balance constraints

[0301]

[0302] 2) Line flow constraints

[0303]

[0304] 3) Restore resource constraints

[0305]

[0306] Where: H max is the maximum number of lines that can be repaired simultaneously; r l,t 1 means that line l starts to be repaired at time t; T REP The time required for line repair.

[0307] Among them, since the line repair is done within a certain period of time T REP Completed within, so r l,t and h l,t The relationship is shown in formula (22):

[0308]

[0309] At the same time, the system state S l,t And the fault state z l,t With r l,t The relationship is shown in formula (23):

[0310]

[0311] In step S1: establish a load shedding slave model:

[0312] During the fault evolution process, the power system needs to reallocate power generation tasks or perform emergency load shedding after the fault occurs to meet the static safety constraints of the system; in the risk assessment of the power system, the optimal power flow model based on DC power flow is usually used to quickly calculate the load shedding amount. In this patent, another sub-problem with the goal of minimizing the load shedding amount is established at the lower level;

[0313]

[0314] The constraints of the load shedding model include power balance constraints, line flow constraints, unit output constraints, spinning reserve constraints, and unit ramp constraints as shown in equations (10)-(14), and the load shedding constraints are shown in the following equation:

[0315]

[0316] In step S2, the most severe failure scenario is searched using the classic DAD model based on robust optimization without considering additional measures to enhance source-load flexibility, which is described as follows:

[0317]

[0318] Where ξ1,ξ2 are the feasible domains of decision variables before and during the disaster, VOLL d is the load shedding penalty cost of node d.

[0319] In step S3: Establish the SCUC composite decision-making dynamic game robust optimization model:

[0320] Based on the above description of the SCUC composite decision model, line restoration and load shedding are both based on the start-stop plan decided by SCUC, and the decision plans of line restoration and load shedding in turn affect SCUC; this decision-making process conforms to the situation of leader-follower game; therefore, by taking SCUC as the leader and line restoration and load shedding as different followers, a Stackelberg game model with a single leader and multiple followers can be established, and its game relationship is as follows:

[0321] 1) SCUC, as the leader, from the upper strategy space Ω SCUC Choose a decision strategy Based on information such as network parameters, load, and distributed renewable energy output, it is then passed on to the lower layers;

[0322] 2) Line restoration and load shedding As a follower, adjust the restoration and load shedding plan within a certain range. Based on the published strategy, from the lower strategy space Ω line,load Select the best response strategy

[0323] 3) SCUC obtains the most cost-effective decision based on the response strategy of feedback. This iterative game process continues until equilibrium is reached;

[0324] The above model can be expressed in the standard form of Stackelberg game as:

[0325] G={M;U G ;{h,△P D};C;{H,L}} (27)

[0326] Where M is the set of three participants in the game.

[0327] Example

[0328] This paper uses the modified IEEE118 node system as an example, and the typhoon wind circle and trajectory data are based on the historical data of Typhoon Lekima that landed in 2019 to verify the effectiveness and correctness of the composite decision-making method proposed in this paper. The simulation experiments are all implemented in Matlab environment.

[0329] In order to verify the correctness and effectiveness of this method, three methods are used to simulate the examples in this paper:

[0330] Method 1: The traditional three-stage SCUC robust optimization model without game theory, i.e. the technical method with reference number 10.1109 / TSG.2020.3048234;

[0331] Method 2: A zero-sum game model based on the SCUC decision model in document number 10.1109 / TSG.2020.3048234 and the line restoration model proposed in this invention;

[0332] Method 3: The composite decision-making master-slave game model proposed in the present invention takes SCUC as the leader and load shedding and line restoration as followers.

[0333] The present invention uses the ΦΛΕΠ indicator to evaluate the decision-making effects of the three methods, wherein the meaning and calculation formula of each indicator are shown in Table 1.

[0334] Table 1 Meaning and calculation formula of ΦΛΕΠ toughness index

[0335]

[0336] Where R0 is the load level under normal system operation; R pe The lowest load level after the system is derated due to typhoon weather; t e is the moment when the system begins to be attacked by typhoon weather; pe The moment when the typhoon weather attack ends; rThe time when the system starts to recover performance; t pr The time when the system returns to normal operating state.

[0337] Comparison of toughness curves of the three methods is shown in the attached figure. Figure 3 As shown in the figure, it can be seen that after the typhoon started at 03:00, the system load began to drop rapidly and reached the lowest point at 06:00-07:00. During this period, the load supply of method 3 was always higher than that of method 1 and method 2. In the load recovery stage, the system load of method 2 and method 3 recovered rapidly under the coordination of line repair and other response strategies and reached normal levels at 20:00 and 17:00 respectively. At the same time, we can notice that although method 2 recovered faster than method 3 in the initial stage, there was a load drop at 10:00, while method 1 did not consider the impact of system dynamic changes on line recovery and repaired key lines in time, resulting in slow recovery and finally reaching normal levels at 21:00. The specific resilience indicators of the three methods are shown in Table 2.

[0338] Table 2 Comparison of toughness indicators of different methods

[0339]

[0340] It can be found from Table 2 that: compared with the multi-stage model of independent decision-making in Method 1, the load drop rate Φ and amplitude Λ of Method 2 are reduced by 23.6% and 21.2% respectively, and the derating operation time Ε is reduced by 1h. At the same time, since the difference amplitude of Λ is much larger than Ε, the load recovery rate of Method 2 is reduced by 9.9% compared with Method 1, indicating that compared with the independent decision of line restoration, Method 2 considering the dynamic interaction between line restoration and SCUC has a greater effect on improving the level of grid resilience; and the resilience index of Method 3 has been comprehensively improved compared with Methods 1 and 2, indicating that the composite decision-making model proposed in this paper that comprehensively considers the dynamic interaction process of load shedding, line restoration and SCUC can further improve the comprehensive level of grid resilience compared with Method 2 that only considers the dynamic interaction between line restoration and SCUC.

[0341] In order to further analyze the mechanism of this method to improve the resilience of the power system, the start-stop plan of the unit and the line status are taken as examples for further analysis. The results are shown in the attached Figure 4 , 5 .

[0342] It can be seen from the figure that compared with Method 2 and Method 3, the line restoration scheme of Method 1 is too centralized. Some island nodes with higher loads (such as 75 and 106) only need one line to restore power supply, but because the load of the node is shared by multiple adjacent lines in normal state, the importance of these lines is higher than that of other lines, which leads to the waste of restoration resources in Method 1. For example, at 04:00, the three lines LN-167, 169 and 172 adjacent to node 106 were disconnected, resulting in an island, while Method 1 prioritized the centralized restoration of the lines adjacent to large-capacity units (such as TG-35 and 36) to meet the load supply of surrounding nodes, resulting in node 106 remaining in an island state until 16:00, resulting in a total unnecessary load loss of 458.5MW during the 04:00-16:00 period. Method 2 considers the dynamic interaction between SCUC and line restoration, completely shuts down TG-36 and transfers the supply of its surrounding nodes to TG-44, 45, 46 and 54, while ensuring its load supply and using the excess recovery resources to repair node 106.

[0343] Method 2 chooses to repair LN-172 among the three lines because its main model aims to minimize the total cost, so that TG-49 with the lowest cost can supply the load of node 106 alone until 14:00. However, since the capacity of TG-49 is only 20MW, it cannot fully support the island load, resulting in a total load loss of 176.2MW during the period of 04:00-14:00, which is 61.6% less than that of Method 1.

[0344] Method 3 takes into account the dynamic interaction among SCUC, load shedding and line restoration. By repairing LN-167 and 170 at the same time, node 106 is connected to the system while TG-49 is connected to the island system composed of nodes 103-105 and 108-112, so that the high-load node 106 is separated from the island system and its load supply pressure is reduced. Finally, node 106 only loses 93.9MW of load during the island period from 04:00 to 07:00, and the load loss of the island system is also reduced from 132.4MW in method 2 to 44MW. In addition, the comparison of the cost and total load shedding of the three methods during the entire typhoon weather process is shown in Table 3.

[0345] Table 3 Comparison of simulation results of different methods

[0346]

[0347] As can be seen from the table, this method considers the dynamic interaction of various response strategies during typhoon weather, reasonably allocates limited resources to minimize load losses at the lowest possible cost, and compared with method 1 and method 2, which are simply based on cost, it reduces unnecessary load shedding by 62.9% and 19.4% at the cost of increasing the total cost by 33.4% and 12.9%, respectively. It can be concluded that the composite decision-making method proposed in the present invention can not only effectively make decisions on the fault recovery sequence and obtain the optimal recovery plan, but also optimize the performance of the power system during the entire typhoon weather process while taking into account the cost target in the decision-making process, further improving the resilience level of the power grid when encountering typhoon disasters.

Claims

1. A composite decision optimization method for safety-constrained unit commitment in response to typhoon weather, characterized in that: The following steps are involved: Step 1: Construct a typhoon weather scene uncertainty set based on the state sampling method. The typhoon weather scene uncertainty set includes wind turbine output and line fault status; Step 2: Use the uncertain set of wind weather scenarios obtained in step 1 to establish a SCUC composite decision-making dynamic game robust optimization model; Step 3: Solve the SCUC compound decision dynamic game robust optimization model obtained in step 2 based on the iterative algorithm to obtain the optimal compound decision.

2. The method according to claim 1, characterized in that In step 1, the following sub-steps are included: Step 1.1: Construct a Rankine-based wind field model to fit the wind speed of each transmission line section and wind turbine at any time; Step 1.2: Establish a vulnerability model of line failure probability and a wind turbine output model respectively, and use the wind speed obtained in step 1.1 to solve the uncertainty set of each line failure probability and each wind turbine output at different times; Step 1.3: Using the failure probability of each line at different times obtained in step 1.2, a line fault state scenario set is established based on the state sampling method; The uncertainty set of wind turbine output at different times obtained in step 1.2 and the line fault state scenario set selected in step 1.3 are the typhoon weather scenario uncertainty set.

3. The method according to claim 2, characterized in that In step 1.1, under typhoon disaster events, the uncertainty of grid component failures is a random event related to the spatiotemporal changes and impacts of typhoons; transmission lines face the risk of failures caused by high wind speeds during typhoons, which leads to line power flow fluctuations; therefore, it is necessary to consider the impact of typhoons on wind turbines and transmission lines. The wind speed V in each region can be calculated using the improved Rankine wind field model: Where, T typ represents the adjustment parameter of wind speed distribution; r represents the distance from the typhoon center; V max and R max Respectively represent the maximum wind speed and maximum wind speed radius of the typhoon; K V Represents the correction factor.

4. The method according to claim 2, characterized in that: In step 1.2, the output of the wind turbine generator depends on its rated power and the real-time wind speed; when considering the cut-in and cut-out scenarios of the wind turbine generator, the output of the generator can be approximated by the following piecewise linearization method: Where P W ′ represents the active output power of the fan; C w is the output coefficient of the fan; Represents the maximum output power of the fan; V in 、V out and V e They are the cut-in wind speed, cut-out wind speed and rated wind speed of the fan; Combined with formula (2), we can get the uncertainty set of each fan output: Where P W represents the actual output power of the fan; α1 and α2 represent the upper and lower limit coefficients of the uncertainty interval of the fan output respectively; Under the influence of typhoons, the power grid needs to consider transmission line failures. The relationship between line failure rate and typhoon wind speed is expressed by the vulnerability curve, and its specific expression is as follows: In the formula, p f represents the failure rate of the transmission line; v represents the maximum wind speed that the line can withstand; β represents the coefficient of the fragility curve.

5. The method according to claim 2, characterized in that: In step 1.3, assume that the system consists of m lines, S k represents the state of line k, then z=(S1,…,S k ,…,S m ); the system state z depends on the state combination of the lines, so as long as the state of each line is determined, the system state x is also determined accordingly; The line status can be obtained by randomly sampling the probability characteristics of the line, and the probability characteristics of the line can be described by a uniform distribution between [0,1]. Assuming that each line has only two states, fault and operation, and that line faults are independent events, a uniformly distributed random number U is drawn with a value interval of [0,1]. k , combined with formula (4), we have: Draw m random numbers U1,…,U k ,…,U m , we can determine a system state z=(S1,…,S k ,…,S m ), repeat the above process N times, and we can get a scene set Z={z1,z2,…,z N }.

6. The method according to claim 1, characterized in that In step 2, the following sub-steps are included: Step 2.1: Establish the SCUC decision-making subject model and the line restoration and load shedding slave models respectively; Step 2.2: Build a DAD model to search the scene uncertainty set obtained in step 1 to obtain the worst failure scenario; Step 2.3: Use the various master-slave models obtained in step 2.1 and the worst failure scenario obtained in step 2.2 to establish a SCUC composite decision-making dynamic game robust optimization model.

7. The method according to claim 6, characterized in that In step 2.1: Establish the SCUC decision-making subject model: The objective function of the SCUC decision-making subject model is to minimize the total cost C, which includes the start-up and shutdown cost C of the unit. onoff and operating cost C ope and the recovery cost C rec ; in: Where: represents the active output of unit i at time t; represents the start / stop status of unit i at time t; N G is the total number of system units; N L is the total number of system lines; (a i ,b i ,c i ) is the operating cost parameter of unit i; are the start-up and shutdown costs of unit i respectively; (V i,t ,W i,t ) are the start-up and shutdown states of unit i at time t, respectively; hl,t represents the maintenance state of line l at time t; The constraints of the SCUC strategy model mainly include power balance constraints, line flow constraints, unit output constraints, spinning reserve constraints, unit ramp constraints, and minimum start-stop time constraints; 1) Power balance constraints Where: represents the output of the fan at time t; represents the load of node d at time t; represents the load shedding amount of node d at time t; N D is the total number of system nodes; 2) Line flow constraints Where: S l,t is the state of line l at time t; K W,SF and Respectively represent the network transfer factors of thermal power units, wind power units and loads; P lmax is the upper limit of the power flow of line l; 3) Unit output constraints Where: and They represent the lower and upper limits of the output of unit i respectively; 4) Spinning reserve constraints Where: The system positive spinning reserve demand caused by load forecast error; and They are the positive and negative spinning reserve demands of the system caused by the wind turbine output prediction error; 5) Unit climbing constraints Where: and They represent the upward climbing and downward climbing constraints of unit i respectively; 6) Minimum start and stop time constraints Where: and Respectively represent the continuous start and stop time of unit i; and They represent the minimum continuous start and stop time of unit i respectively.

8. The method according to claim 6, characterized in that In step 2.1: Establish the line recovery slave model: The main task of the power system restoration phase is to repair the faulty transmission lines and restore load supply as quickly and efficiently as possible. For the repair of transmission line faults, in order to ensure the safety of maintenance personnel, it is stipulated that the repair work can only be started after the typhoon no longer affects the suspended lines. The objective function of the lower restoration model is to maximize load supply, and its constraints include power balance, line flow, and restoration resource constraints; Where: represents the load supply capacity of node d at time t; 1) Power balance constraints 2) Line flow constraints 3) Restore resource constraints Where: H max is the maximum number of lines that can be repaired simultaneously; r l,t 1 means that line l starts to be repaired at time t; T REP The time required for line repair; Among them, since the line repair is done within a certain period of time T REP Completed within, so r l,t and h l,t The relationship is shown in formula (22): At the same time, the system state S l,t And the fault state z l,t With r l,t The relationship is shown in formula (23): In step 2.1: Establish the load shedding slave model: During the fault evolution process, the power system needs to reallocate power generation tasks or perform emergency load shedding after a fault occurs to meet the static safety constraints of the system; In power system risk assessment, the optimal power flow model based on DC power flow is usually used to quickly calculate the load shedding amount; therefore, another sub-problem with the goal of minimizing the load shedding amount is established at the lower level; The constraints of the load shedding model include power balance constraints, line flow constraints, unit output constraints, spinning reserve constraints, and unit ramp constraints as shown in equations (10)-(14), and the load shedding constraints are shown in the following equation:

9. The method according to claim 6, characterized in that In step 2.2, the most severe failure scenario is searched using the classic DAD model based on robust optimization without considering additional measures to enhance source-load flexibility, which is described as follows: Where ξ1,ξ2 are the feasible domains of decision variables before and during the disaster, VOLL d is the load shedding penalty cost of node d.

10. The method according to claim 6, characterized in that In step 2.3: Establish the SCUC compound decision dynamic game robust optimization model: Based on the above description of the SCUC composite decision model, line restoration and load shedding are both based on the start-stop plan decided by SCUC, and the decision plans of line restoration and load shedding in turn affect SCUC; this decision-making process conforms to the situation of leader-follower game; therefore, by taking SCUC as the leader and line restoration and load shedding as different followers, a Stackelberg game model with a single leader and multiple followers can be established, and its game relationship is as follows: 1) SCUC, as the leader, from the upper strategy space Ω SCUC Choose a decision strategy Based on information such as network parameters, load, and distributed renewable energy output, it is then passed on to the lower layers; 2) Line restoration and load shedding As a follower, adjust the restoration and load shedding plan within a certain range; based on the published strategy, from the lower strategy space Ω line,load Select the best response strategy 3) SCUC obtains the most cost-optimal decision based on the response strategy of feedback; this iterative game process continues until equilibrium is reached; The above model can be expressed in the standard form of Stackelberg game as: G={M;U G ;{h,△P D };C;{H,L}} (27) Where M is the set of three participants in the game.

Images