Multi-time-scale optimization scheduling method considering key section power flow safety

Abstract

The invention relates to the technical field of power system optimization scheduling, and discloses a multi-time-scale optimization scheduling method considering key section power flow safety, which comprises the following steps: firstly, identifying a system partition map based on an adaptive GN algorithm, and establishing an optimization model considering power grid safety operation constraint conditions such as node power balance and line power flow balance; optimizing to obtain a key section tidal current safety limit value; establishing a day-ahead deterministic optimization scheduling model by taking the minimization of the total operation cost of the power system as a target, and further establishing a date two-stage distribution robust optimization scheduling model considering the wind power uncertainty and solving the date two-stage distribution robust optimization scheduling model; and considering a refined time scale, carrying out refined modeling on an intra-day load and a wind power output predicted value, establishing an intra-day rolling optimization scheduling model based on a model prediction control theory, and solving the intra-day rolling optimization scheduling model. According to the method, the robustness of a power system for dealing with short-term internal source load rapid fluctuation is improved, and the safety and reliability of power grid operation are remarkably improved.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of power system optimal scheduling, in particular to a multi-time scale optimal scheduling method considering key section power flow safety. BACKGROUND

[0002] In recent years, new energy power generation represented by wind power and photovoltaic power continues to increase in the power system. It is estimated that by 2030, the installed capacity of new energy will reach 1.05 billion kilowatts, accounting for 36% of the total installed capacity. The low-carbon transformation of the power system structure has become an inevitable trend. However, the multi-element uncertainty of new energy output and load fluctuation, randomness and intermittency brings a series of challenges to the safe and economic dispatching of the power system. The multiple blackouts that have occurred at home and abroad in recent years have exposed the vulnerability of modern large interconnected power grids. In order to improve the reliability of power grid operation, it is of great theoretical significance and practical value to assess the overall operation risk of the power system and the potential risk of the internal equipment operation.

[0003] At present, the research on improving the economic efficiency of power dispatching based on new energy power prediction technology has been relatively mature, but most of the existing prediction models have high dependence on scenarios and general accuracy. Therefore, it is necessary to gradually refine the dispatching time scale, improve the robustness of the power system in response to short-term source and load rapid fluctuations, and consider the day-ahead rolling on the basis of day-ahead dispatching. The introduction of model predictive control theory into the power system dispatching problem to realize the update of parameter state and prediction data has been widely applied. SUMMARY

[0004] The present application proposes a multi-time scale optimal scheduling method considering key section power flow safety to solve the problems of the prior art. Based on the linear programming theory, the key section power flow limit value is determined, a day-ahead two-stage distribution robust optimization scheduling model based on multiple discrete scenarios is established, and the time scale is refined. The model predictive control theory is introduced to realize day-ahead-intra-day coordinated rolling optimization, which realizes the accurate identification of the key transmission channel of the power grid, significantly improves the ability of the power system to respond to short-term source and load rapid fluctuations, and provides effective technical support and practical path for the safe dispatching of the power system with high proportion of new energy access. The technical scheme is as follows:

[0005] A multi-time scale optimal scheduling method considering key section power flow safety, comprising the following steps:

[0006] Step 1: Based on the adaptive GN algorithm, the power system partition graph is identified, the linear programming theory is adopted, the optimization model of the power grid safety operation constraint condition considering node power balance and line power flow balance is established, and the key section power flow safety limit value is obtained by optimization;

[0007] Step 2: A deterministic day-ahead optimal scheduling model considering output constraints, ramping constraints, system power balance constraints and key section power flow safety constraints is established with the objective of minimizing the total cost of power system operation;

[0008] Step 3: Based on the deterministic day-ahead optimal scheduling model established in step 2, a day-ahead two-stage distribution robust optimal scheduling model considering wind power uncertainty is established; the first stage minimizes the total cost of system operation; the second stage considers uncertainty based on discrete wind power output scenarios, and minimizes the system adjustment cost;

[0009] Step 4: Based on the day-ahead two-stage distribution robust optimal scheduling model established in step 3, the day-ahead rolling optimal scheduling model based on model predictive control theory is established by considering the refined time scale and modeling the refined day-ahead load and wind power output prediction values, so as to improve the robustness of the power system in response to short-term rapid fluctuations in sources and loads;

[0010] Step 5: The day-ahead two-stage distribution robust optimal scheduling model considering wind power uncertainty established in step 3 and the day-ahead rolling optimal scheduling model based on model predictive control theory established in step 4 are solved by using an optimization software combined with a column and constraint generation algorithm.

[0011] The application has the following advantages and beneficial effects:

[0012] (1) The linear programming-based key section power flow safety limit analysis method proposed in the application establishes a key section and card mouth fault identification model based on an adaptive GN partitioning algorithm, and realizes accurate identification of the key transmission channel of the power grid. On this basis, combined with the linear programming method, a safety limit calculation model is established with the maximum sum of section power flow as the objective. Compared with the prior art, the scheme realizes the full-process connection from power grid topology analysis to electrical safety quantitative evaluation, effectively analyzes the weak links and transmission section carrying capacity in the operation of the power grid, effectively enhances the scientificity of the power grid operation mode arrangement and the pertinence of the preventive control strategy, and balances the safety and operation efficiency of the power grid.

[0013] (2) The multi-time scale optimal scheduling method considering key section power flow safety constraints proposed in the application considers wind power output uncertainty in the day-ahead stage, and establishes a day-ahead optimal scheduling model with the minimum total cost of system operation as the objective function. In the day-ahead refined time scale, the day-ahead rolling optimal scheduling model based on MPC is proposed, and the adjustment amount of each unit in the system is minimized. In addition, compared with the prior art, the scheme realizes the modeling of the optimal scheduling method considering the power flow safety of the transmission section in the multi-time scale, improves the robustness of the power system in response to short-term rapid fluctuations in sources and loads, and balances the economy and power supply reliability of the power system. BRIEF DESCRIPTION OF DRAWINGS

[0014] Figure 1 Flow chart of multi-time scale optimal scheduling method considering key section power flow safety of the present application.

[0015] Figure 2 IEEE-39 node system partition diagram and section power flow limit value of the present application.

[0016] Figure 3 Day-ahead scheduling layer load and wind power output prediction value of the present application.

[0017] Figure 4 Thermal power unit output of the present application without considering section power flow safety constraint.

[0018] Figure 5 Wind power output and its initial probability distribution under discrete scenario of the present application.

[0019] Figure 6 Thermal power unit reserve calling situation without considering wind power uncertainty of the present application.

[0020] Figure 7 Key section power flow situation before and after not considering section power flow safety constraint of the present application.

[0021] Figure 8 Thermal power unit output considering section power flow safety constraint of the present application.

[0022] Figure 9 Key section power flow situation before and after considering section power flow safety constraint of the present application.

[0023] Figure 10(a) is the day-ahead-intra-day prediction value of wind power output of the present application.

[0024] Figure 10(b) is the day-ahead-intra-day prediction value of load of the present application.

[0025] Figure 11(a) is the day-ahead-intra-day output comparison chart of wind power unit considering section power flow safety constraint of the present application.

[0026] Figure 11(b) is the day-ahead-intra-day output comparison chart of thermal power unit considering section power flow safety constraint of the present application. DETAILED DESCRIPTION

[0027] The present application will be further described below in combination with the drawings and specific embodiments.

[0028] The present application proposes a multi-time scale optimal scheduling method considering key section power flow safety, determines key section power flow limit value based on linear programming theory, establishes a day-ahead two-stage distribution robust optimal scheduling model based on multiple discrete scenarios, and refines time scale, introduces model predictive control theory to realize day-ahead-intra-day coordinated rolling optimization. The flow chart of the method of the present application is as follows Figure 1As shown, specifically comprising the following steps:

[0029] Step 1: Based on the adaptive GN algorithm, an optimization model considering the power balance of nodes and the power flow balance of lines is established by using linear programming theory, and the key section power flow safety limit is obtained by optimization.

[0030] Based on the complex network theory, the adaptive GN partition algorithm is proposed to identify the key section and the key section fault. The line importance index is used to reflect the importance of the transmission line:

[0031] ;

[0032] In the formula, is the topological betweenness of line l; is the number of shortest paths between node i and node j passing through line l; is the number of shortest paths between node i and node j; V is the set of nodes in the power grid.

[0033] B e The larger the value of B e (l), the higher the importance of the transmission line l, and the transmission line with high B

[0034] GN algorithm is proposed by Girvan and Newman, which is a standard algorithm for community structure analysis. The basic idea is to remove the edge with the largest betweenness from the network, thereby obtaining the best community structure. The edge betweenness is defined as the number of shortest paths passing through each edge in the network.

[0035] The power grid is partitioned, and after removing the transmission line with the largest B e (l) value, the importance index of the remaining transmission lines needs to be recalculated. After the partition of the power grid is completed, the best partition scheme of the power grid needs to be determined, and the partition tie line needs to be searched.

[0036] The modularity index based on the community network theory is used to evaluate the partition scheme of the power grid to determine the best partition scheme. The definition of the modularity index is as follows:

[0037] ;

[0038] In the formula, Q is the modularity of the power grid partition scheme; m is the number of transmission lines in the power grid; A vw is the adjacency matrix of the power grid; k v is the node degree of node v; k w is the node degree of node w; and denote the community number to which nodes v and w belong, respectively; denote the function that takes value 1 when nodes v and w belong to the same sub-region, and 0 otherwise.

[0039] ;

[0040] ;

[0041] ;

[0042] If the power grid is divided into h sub-regions, then the h*h symmetric matrix E = e dn The modularity index Q is calculated. The element e dn denotes the proportion of the transmission line connecting the nodes in sub-region d and sub-region n in the total number of transmission lines in the power grid. In summary, the calculation method of modularity Q can be simplified as:

[0043] ;

[0044] ;

[0045] ;

[0046] In the formula, is the weight of the number of edges connected to the nodes in sub-region d in the total number of the whole network; is the function that takes value 1 when node v belongs to sub-region d, and 0 otherwise; denotes 1 when node w belongs to sub-region n, and 0 otherwise; is the diagonal element of the symmetric matrix E.

[0047] The maximum value of modularity Q is 1, and the closer the value of Q is to 1, the more obvious the community structure of the power grid is, that is, the more reasonable the partition scheme is. By calculating the modularity of the scheme obtained in the partition process, the partition scheme with the Q value closest to 1 is selected as the best partition scheme of the power grid, and the partition tie lines between the sub-regions of the power grid are obtained.

[0048] The power grid partition process based on the adaptive GN algorithm is as follows:

[0049] (1) Simplify the topology structure of the power grid: convert the generator, substation, transformer and load node into node in the topology graph; convert the transmission line into tie line in the topology graph, ignore the transmission line in the power plant; merge the double-circuit or three-circuit transmission line.

[0050] (2) Calculate the importance index B e (l) of all transmission lines in the power grid.

[0051] (3) Remove B from the power grid e (l) The transmission line with the largest value is selected, and a depth-first search algorithm is used to determine whether a new partition has been generated.

[0052] (4) Calculate the modularity Q value of the network in each round, and store the modularity Q value and the community segmentation in the current network.

[0053] (5) If a new partition is generated, the removed transmission lines and partition results are saved, and the modularity Q value of the current network is calculated. The difference between the Q value and the original modularity is compared. If the current Q value and the community partitioning in the network are stored and updated, then the next network partitioning is performed.

[0054] (6) Search for whether there are still removable lines. If there are, execute (2); if not, execute (7).

[0055] (7) Stop removing transmission lines in the power grid, and search for a single node that constitutes a sub-region of the power grid. If it exists, merge the sub-region formed by the single node with its nearest sub-region. If it does not exist, execute (8).

[0056] (8) Output the power grid partitioning results saved in (4) and return the current modularity Q value.

[0057] Transmission sections serve as energy transfer channels between different areas of the power grid and have always been a key focus of power grid dispatching and operation. This invention proposes a method for determining power flow safety limits for critical sections based on linear programming. The optimization model for power flow limits of critical transmission sections based on linear programming theory is as follows:

[0058] Objective function:

[0059] ;

[0060] In the formula, For the power flow of all transmission lines within the cross section The sum; Let be the power flow of transmission line l at time t; S be the set of all transmission lines that make up this section.

[0061] Constraints:

[0062] Power balance constraints:

[0063] ;

[0064] In the formula, This represents the total number of thermal power units. This represents the total number of wind turbine units. The total number of transmission lines l; is the active power output of the thermal power unit at time t; is the active power output of the wind power unit at time t; s(l) and r(l) are the sending bus (node) and the receiving bus (node) of the transmission line l, respectively; is the system load forecast value at time t; is the transmission power flow of the transmission line l.

[0065] Thermal power unit output constraint:

[0066] ;

[0067] wherein, and are the upper and lower limits of the thermal power unit output, respectively.

[0068] Interconnection line power flow constraint:

[0069] ;

[0070] wherein, and are the voltage phase angles of the sending bus and the receiving bus of the transmission line l at time t, respectively; is the reactance value of the transmission line l; and are the upper and lower limits of the transmission power flow of the transmission line l, respectively.

[0071] Node phase angle constraint:

[0072] ;

[0073] wherein, and are the upper and lower limits of the voltage phase angle , respectively; is the voltage phase angle of the equilibrium node.

[0074] The application proposes a key transmission section power flow safety limit value determination method based on linear programming on the basis of the transmission section identification based on the GN adaptive algorithm, and uses an optimization solver to calculate the above model, and respectively obtains the power flow limit value of the key section identified based on the adaptive GN algorithm as shown in formula (1). Figure 2 .

[0075] Step 2: Taking the minimization of the total cost of the power system operation as the target, a day-ahead deterministic optimization scheduling model considering the output constraint, the climbing constraint, the system power balance constraint and the key section power flow safety constraint is established.

[0076] The deterministic day-ahead optimization scheduling model is as follows:

[0077] Objective function:

[0078] Economic benefit is an important influencing factor in dispatching decision, and a target function of minimizing total system operation cost is established, and the target function of day-ahead dispatching is:

[0079] ;

[0080] wherein, is total dispatching duration; a, b, c are operation cost coefficients of thermal power units; and are respectively start / stop cost of thermal power unit g; is penalty factor of abandoned wind power; is predicted value of wind power unit at time t.

[0081] The constraint conditions of the day-ahead optimal dispatching model include system power balance constraint, line flow safety constraint, start / stop cost constraint of thermal power unit, unit output constraint, ramp constraint, minimum start / stop time constraint and wind power output constraint.

[0082] System power balance constraint:

[0083] ;

[0084] Thermal power unit output constraint:

[0085] ;

[0086] wherein, and are respectively start / stop state of thermal power unit g at time t and time t-1, and start is 1 and otherwise is 0; and are respectively upward and downward reserve capacity called by thermal power unit at time t.

[0087] Start / stop cost constraint of thermal power unit:

[0088] ;

[0089] wherein, and are respectively start / stop cost coefficients.

[0090] Ramp constraint of thermal power unit:

[0091] ;

[0092] wherein, and are respectively maximum upward and downward ramp rate of thermal power unit.

[0093] Minimum start / stop time constraint of thermal power unit:

[0094]

[0095] In the formula, and respectively, the operation and shutdown time of the unit g at t time; and respectively, the minimum start-up and shutdown time of the unit g.

[0096] Wind power output constraint:

[0097]

[0098] Key cross-section flow safety constraint:

[0099]

[0100] In the formula, and respectively, the upper and lower limits of the flow of the key power transmission section S.

[0101] Step 3: Based on the deterministic day-ahead optimization scheduling model established in step 2, a deterministic two-stage distribution robust optimization scheduling model considering wind power uncertainty is established, the first stage minimizes the total system operation cost; the second stage considers uncertainty based on discrete wind power output scenarios, and minimizes the system adjustment cost.

[0102] The application adopts distribution robustness, establishes a two-stage distribution robust optimization scheduling model considering wind power uncertainty based on a deterministic optimization scheduling model. The objective function is:

[0103]

[0104] In the formula, Ω is the feasible region of the second stage operation scenario; p k is the probability value of each discrete scenario; and respectively, the upward and downward reserve capacity called by the thermal power unit under the kth discrete scenario; and respectively, the predicted value and actual output of the wind power unit under the scenario k; is the unit cost required for upward and downward adjustment of the reserve capacity of the thermal power unit.

[0105] Second stage constraint condition:

[0106] In order to cope with the uncertainty of wind power output, the second stage needs to call the reserve capacity of thermal power, and the related constraints are as follows:

[0107] System power balance constraint:

[0108]

[0109] ​​​​​In the formula, is the transmission flow of the power transmission line l under the discrete scenario k;

[0110] Wind power output constraint:

[0111] ;

[0112] Thermal power reserve capacity up and down constraint:

[0113] ;

[0114] In the formula, is the maximum limit of the thermal power unit reserve capacity.

[0115] Critical section flow safety constraint:

[0116] ;

[0117] Based on the N A wind power output discrete scenario set obtained by clustering M historical data, the initial probability value of each discrete scenario is obtained, considering the uncertainty of the probability distribution of each discrete scenario, the probability distribution value of the multiple discrete wind power scenarios is further constrained by taking the initial probability distribution value as the center and including the comprehensive norm constraint condition of 1-norm and ∞-norm, and the feasible region is Ω1 and Ω ∞ , respectively.

[0118] ;

[0119] In the formula, is the initial probability value of the kth discrete scenario obtained according to historical data; is the 1-norm constraint condition; is the ∞-norm constraint condition; and are the maximum allowable deviation of the probability of 1-norm and ∞-norm, respectively.

[0120] satisfy the following confidence:

[0121] ;

[0122] In the formula, Pr{·} represents the probability of the event being true.

[0123] Let the right side of the above inequality be the confidence α1 and α ∞ of the probability distribution value, respectively. The above formula can be converted to:

[0124] ;

[0125] Because of the absolute value term in the 1-norm constraint and the ∞-norm constraint, 0-1 auxiliary variables are introduced and The 1-norm constraint is processed, and 0-1 auxiliary variables are introduced and The ∞-norm constraint is processed. After linearization, it is expressed as:

[0126] ;

[0127] ;

[0128] In the formula, and are the positive and negative offset values of the probability value p k relative to .

[0129] Step 4: Based on the two-stage distribution robust optimization scheduling model established in step 3, the intra-day load and wind power output prediction value is refined, and an intra-day rolling optimization scheduling model based on model predictive control theory is established to improve the robustness of the power system in response to short-term rapid fluctuations in sources and loads.

[0130] Considering the refined modeling of intra-day load prediction value and wind power under multiple time scales, an intra-day rolling optimization scheduling model based on model predictive control theory is established, and the intra-day rolling optimization scheduling model is as follows:

[0131] The source / load value at the next moment is predicted through historical data to provide a reference for scheduling decisions. There is a certain error between the actual state and the optimization scheduling decision made, which will deviate from the actual situation to a certain extent, so it is necessary to correct and rolling optimize the day-ahead stage. Model predictive control (MPC) theory is a feedback control optimization strategy with constraints, mainly composed of a prediction model, rolling optimization and feedback correction. The intra-day optimization scheduling mainly combines MPC theory to establish a mixed integer linear programming model.

[0132] The intra-day scheduling establishes an MPC rolling optimization model with a 2h prediction period and a 15min control period. According to the latest prediction results, the thermal power and wind power output in the future 2h prediction window is arranged, but only the scheduling plan for the first 15 minutes is executed, and the output of the system equipment in the next period is updated, and the above process is repeated constantly to continuously rolling correct and optimize until the optimization in the entire scheduling period is completed.

[0133] The objective function is as follows:

[0134] ;

[0135] In the formula, t s is the current scheduling time; T s is the intraday control time domain; and are the intraday outputs of thermal power units and wind power respectively; and are the day-ahead output of thermal power units and the intraday predicted output of wind power respectively.

[0136] Constraint conditions:

[0137] The intraday constraints are the same as the day-ahead constraints, including thermal power output constraints, system power balance constraints, wind power output constraints and power flow safety constraints.

[0138] Thermal power unit ramping constraints:

[0139] ;

[0140] Critical section power flow safety constraints:

[0141] .

[0142] Step 5: The mature optimization software is combined with the C&CG (Column and Constraint Generation) algorithm to solve the multi-time scale optimization scheduling model established in steps 3 and 4, and an example analysis is carried out by using improved IEEE-39 node data, and the effectiveness of the model is verified.

[0143] The mature optimization software is combined with the C&CG (Column and Constraint Generation) algorithm to solve the multi-time scale optimization scheduling model established in steps 3 and 4 under MATLAB2022b.

[0144] The solving process of the day-ahead two-stage distributed robust optimization scheduling model is as follows:

[0145] The day-ahead two-stage distributed robust model can be represented as a three-layer two-stage problem of "min-max-min":

[0146] ;

[0147] In the formula, a T and b T are the cost coefficient matrices of the first-stage decision variable x and the second-stage decision variable y k respectively; c T is the second-stage wind power curtailment penalty cost coefficient; 、 and are x, y k , ξ kcorresponding coefficient matrix is a constant matrix; ξ k is the wind power output under the kth discrete scenario is the occurrence probability of the kth discrete scenario is the first-stage decision variable set is the second-stage decision variable set related to the discrete scenario k.

[0148] (1) The main problem is to solve the optimal solution that satisfies the constraint condition under the known poor probability distribution, and provide a lower bound value for the objective function:

[0149] ;

[0150] In the formula, N B is the maximum number of iterations; y0 is the second-stage decision variable under a given set of initial values; ξ0 is the initial value of wind power output; m is the iteration number; is the introduced continuous auxiliary variable.

[0151] (2) The sub-problem is to search for the current worst wind power probability distribution under the decision variable x made by the main problem, and generate the corresponding constraint containing the two-stage recourse variable and add it to the main problem:

[0152] ;

[0153] The sub-problem is a double-layer optimization problem, and each scenario in the inner minimization problem is independent and can be solved in parallel. First, solve the inner minimum value problem, and then solve the outer maximum value.

[0154] Let The above formula is transformed into:

[0155] .

[0156] In summary, the two-stage model is iterated and updated through the main and sub-problems until the given precision value is reached, and the iteration is stopped.

[0157] In the solution process, the MPC-based intraday rolling optimization scheduling model is solved as follows:

[0158] Based on the idea of multi-time scale, the intraday load and wind power prediction value are modeled in detail, the minimum adjustment amount of each unit in the system is taken as the target, the robustness of the power system in response to short-term rapid fluctuations in sources and loads is improved, and the MPC-based intraday rolling optimization scheduling model is solved by using mature optimization software.

[0159] The data for solving include: wind power output prediction data, system operation parameters, power grid parameters, optimization model parameters, algorithm parameters.

[0160] Simulation example:

[0161] The improved IEEE-39 node system is taken as an example to verify the effectiveness of the method, including 39 nodes and 46 transmission lines, and a wind farm is connected to 16 nodes. The transmission section identification process based on the adaptive GN algorithm is as follows:

[0162] (1) Calculate the importance index B of the transmission line e (l) The importance index of the 46 transmission lines of the entire power grid is calculated, and the lines are sorted according to the importance, and the power grid is partitioned by removing the transmission lines with the highest importance in turn;

[0163] (2) During the partitioning process, after removing the transmission line with the maximum B e (l) value, the B e (l) value of the remaining transmission lines in the power grid needs to be recalculated. The termination condition of the power grid partitioning process is set to the number of broken lines being 1 / 2 of the total number of transmission lines in the power grid, that is, the maximum number of removable transmission lines is 23;

[0164] (3) The Karger algorithm is used to determine whether the system forms a partition, and the modularity corresponding to the system partition at this time is recorded;

[0165] (4) Select the partition scheme with the highest modularity to obtain the system partition at this time, and the tie line between the system partition at this time can be regarded as the initial section.

[0166] As Figure 2 The IEEE-39 node system partitioning diagram based on the adaptive GN algorithm is shown in Figure 1, which is divided into 6 sub-regions. The composition of the transmission section is shown in Table 1.

[0167] Table 1 Composition of transmission section

[0168] .

[0169] The transmission section identification method identifies transmission sections that are all key channels of active power flow transmission of the power grid, and reflect the vulnerable links of the power grid. The first transmission section (transmission line 16-19) is a power flow sending line of the generator node 33 and the generator node 34, and once the section is disconnected, the entire power grid will be out of balance, so the section is a vulnerable link of the 39-node system. The second transmission section (transmission line 15-16 and transmission line 16-17) plays a role in connecting the sub-regions 4 / 5 and other sub-regions, and the electrical distance of the transmission line 15-16 and the transmission line 16-17 is very short, and once any one of the lines is disconnected due to failure, a large-scale cascading failure will be caused. The third transmission section (transmission line 16-17, transmission line 5-6, transmission line 4-14 and transmission line 6-7) contains power flow sending lines of the generator node 32 to the generator node 38. The fourth transmission section (transmission line 2-25 and transmission line 26-17) also contains power flow sending lines of the generator node 37 and the generator node 38. The fifth transmission section (transmission line 2-3, transmission line 8-9 and transmission line 26-27) is a power flow sending channel of the generator node 30, the generator node 37, the generator node 38 and the generator node 39. The above transmission sections are all key links and vulnerable links of the power grid, and directly affect the stability of the 39-node system. The critical transmission section flow limit value is shown in Table 2.

[0170] Table 2 Critical transmission section flow limit value

[0171] .

[0172] The critical section flow safety constraint obtained based on the above method is added to the day-ahead-intra-day rolling optimization scheduling model proposed in the application to establish a multi-time scale optimization scheduling model considering the critical section flow safety constraint. The example settings are shown in Table 3. Example 1 establishes a day-ahead optimization scheduling model without considering wind power uncertainty; Example 2 considers wind power uncertainty and re-formulates the day-ahead scheduling decision; Example 3 adds the section flow safety constraint on the basis of Example 2; and Example 4 considers intra-day rolling optimization on the basis of Example 3 to establish a multi-time scale optimization scheduling model.

[0173] Table 3 Example scene settings

[0174] .

[0175] As Figure 3 The day-ahead scheduling layer load and wind power output prediction value are shown in Table 4.

[0176] Table 4 Running cost of Examples 1-4

[0177] .

[0178] The thermal power unit output in Example 1 is as shown in Figure 4 The total system operation cost is 272.53 million yuan. From the dispatching result, it can be known that during the period of 10:00-18:00 when the wind power output is low, the load is at the peak and the load level is high, the thermal power unit output is increased to meet the load demand, and through fine coordination of the thermal power unit output and the wind power output, the precise matching of the load demand is realized, and the full consumption of the wind power resources is ensured, so that the efficient operation target of zero wind curtailment is achieved. The thermal power unit plays its flexible adjustment advantage and is complementary to the fluctuation characteristics of the wind power output, so that the clean energy is maximized in use under the premise of ensuring the safe and stable operation of the system, and the dual optimization of economy and environmental protection is realized.

[0179] In Example 2, the wind power output uncertainty is considered on the basis of Example 1, Figure 5 the wind power output and the initial probability distribution in the discrete scene, the system operation cost is 385.25 yuan, which increases by 41.36%. The standby calling condition is as shown in Figure 6 After considering the uncertainty, the standby capacity of the thermal power unit is called to cope with the wind power output uncertainty to meet the system power balance constraint.

[0180] After the decision is made by the two-stage distribution robust optimization dispatching based on multiple discrete scenes, the key section flow condition is as shown in Figure 7 It can be known from Figure 7 that the key section flow condition, the section 1 is not over limit during the dispatching time; the section 2 is over limit during the time of 2:00-7:00, 9:00, 12:00, 14:00, 17:00-18:00 and 21:00-23:00; the section 3 is over limit during the time of 8:00, 11:00, 13:00, 16:00, 19:00-20:00 and 24:00; the section 4 is over limit during the time of 1:00, 3:00-5:00 and 7:00-24:00; and the section 5 is over limit during the time of 1:00-7:00 and 11:00-24:00.

[0181] In order to ensure the safety of the key section flow, the key section flow safety constraint is considered in Example 3 on the basis of Example 2, the total system operation cost is 385.68 million yuan, and due to the adjustment of the output of each thermal power unit, the operation cost increases by 0.43 million yuan compared with Example 2. The thermal power unit output condition considering the section safety constraint is as shown in Figure 8 It can be known from Figure 8 and Figure 4 that in order to meet the key section flow safety constraint, the output of each unit is adjusted, the smooth output condition is adjusted to adjust the system flow to adapt to the system operation safety constraint. The key section flow condition is asFigure 9 As shown in the figure, all the cross-section tidal flows meet the safety constraints, and the optimal scheduling decision is obtained through the key cross-section tidal flow safety check.

[0182] Since the day-ahead data prediction cannot completely accurately reflect the actual situation in the future, the accuracy and effectiveness of the scheduling strategy cannot be guaranteed. In order to guarantee the accuracy and effectiveness of the scheduling strategy, in Example 4, the day-ahead scheduling is introduced on the basis of Example 3 to cope with the uncertainty in the short term. It is assumed in the application that the wind power output and the load prediction error obey the normal distribution, and a certain random error is superimposed on the basis of the day-ahead prediction value of the wind power output and the load to simulate the day-ahead prediction data of the wind power and the load. The day-ahead and day-ahead prediction values of the wind power output and the load are shown in Figs. 10 (a) and 10 (b).

[0183] The day-ahead scheduling decision is made by rolling optimization through MPC until the last prediction time domain. The comparison between the day-ahead and day-ahead actual outputs of the wind power and the thermal power is shown in Figs. 11 (a) and 11 (b). As can be seen from the scheduling result, the day-ahead and day-ahead trends of the wind power and the thermal power unit output are basically consistent, only with partial deviation. The day-ahead rolling optimization continuously adjusts the thermal power and the wind power output on the basis of the day-ahead scheduling plan, reduces the power deviation in operation, realizes more fine scheduling, and ensures more stable and efficient operation of the unit.

[0184] Firstly, the application identifies the weak links and key transmission sections in the operation of the power system based on the adaptive GN partition algorithm, and proposes a key section flow safety constraint analysis method based on linear programming theory. On this basis, a multi-time scale optimal scheduling model considering the key transmission section flow safety is established. The two-stage distributed robust optimization scheduling model is established with the minimum total system operation cost as the target in the day-ahead, and the rolling optimization is realized based on the MPC theory in the day-ahead. The improved IEEE-39 node system is used for example analysis, and the following conclusions are drawn:

[0185] (1) The key section flow safety limit determination method based on the adaptive GN partition algorithm and linear programming proposed in the application can effectively identify the weak links and key transmission channels of the power grid, provide an important safety boundary basis for multi-time scale optimal scheduling, and significantly improve the scientificity of the power grid operation mode and the effectiveness of the preventive control.

[0186] (2) The constructed day-ahead two-stage distributed robust optimization model balances the system economy and robustness by considering the wind power uncertainty through discrete scene and probability distribution constraints, effectively reduces the sensitivity of the scheduling decision to the prediction error, and improves the ability of the system to cope with the source and load fluctuations.

[0187] (3) The rolling optimization mechanism of model predictive control (MPC) theory is introduced, and through multi-time scale coordination and feedback correction, dynamic modification and fine adjustment of the day-ahead plan are realized, which further enhances the response capability and operation reliability of the system in short time scale, ensures the practical feasibility and adaptability of the dispatching strategy, and provides effective technical support and practical path for safe dispatching of power systems with high proportion of new energy access.

Claims

1. A multi-time scale optimal scheduling method considering key section power flow security, characterized in that, The method comprises the following steps: Step 1: identifying the power system partition map based on an adaptive GN algorithm, using linear programming theory to establish an optimization model considering the power balance of nodes and the power flow balance of lines as the safe operation constraint conditions of the power grid, and optimizing to obtain the critical section power flow safety limit; Step 2: establishing a deterministic day-ahead optimal scheduling model considering output constraints, ramping constraints, system power balance constraints and critical section power flow safety constraints, with the goal of minimizing the total cost of power system operation; Step 3: based on the deterministic day-ahead optimal scheduling model established in step 2, a day-ahead two-stage distribution robust optimal scheduling model considering wind power uncertainty is established; the first stage minimizes the total cost of system operation; the second stage considers uncertainty based on discrete wind power output scenarios, and minimizes the system adjustment cost; Step 4: based on the day-ahead two-stage distribution robust optimal scheduling model established in step 3, considering the refined time scale, the day-ahead load and wind power output prediction value are modeled in detail, and a day-ahead rolling optimal scheduling model based on model predictive control theory is established to improve the robustness of the power system in response to short-term rapid fluctuations in sources and loads; Step 5: using an optimization software and a constraint generation algorithm to solve the day-ahead two-stage distribution robust optimal scheduling model considering wind power uncertainty established in step 3, and the day-ahead rolling optimal scheduling model based on model predictive control theory established in step 4.

2. The multi-time scale optimal scheduling method considering key section power flow security according to claim 1, characterized in that, In step 1, the identification of the power system partition map based on the adaptive GN algorithm comprises: The topological betweenness of the line is used as an important degree index of the line to reflect the importance of the power transmission line, and is calculated as follows: ； wherein is the topological betweenness of the transmission line l; is the number of shortest paths between node i and node j that pass through transmission line l; is the number of shortest paths between node i and node j; V is the set of nodes of the power grid. partitioning the power grid, removing the transmission line with the largest value each time, recalculating the importance index of the remaining transmission lines; after the power grid is partitioned, determining the best partition scheme of the power grid and searching for the partition tie line; The modularity index based on the community network theory is used to evaluate the partition scheme of the power grid to determine the optimal partition scheme; the definition of the modularity index is as follows: ； where Q is the modularity of the grid partitioning scheme; m is the number of transmission lines in the grid; A vw is the adjacency matrix of the grid; k v is the node degree of node v; k w is the node degree of node w; and denote the community number to which nodes v and w belong, respectively; denotes the function that takes the value 1 when nodes v and w belong to the same sub-region and 0 otherwise. ； ； ； If the grid is divided into h sub-areas, then by defining a h*h symmetric matrix E = e dn The module Q is calculated; the element e dn represents the proportion of the total number of power lines in the grid that connects the nodes in sub-area d and sub-area n. The calculation method of the modularity Q is simplified as: ； ； ； wherein is the weight of the number of edges connected to the nodes in the sub-area d relative to the total number of edges in the network; is is the function that takes the value 1 when the node v belongs to the sub-area d and 0 otherwise; is the function that takes the value 1 when the node w belongs to the sub-area n and 0 otherwise; is the diagonal element of the symmetric matrix E; By calculating the modularity of the scheme obtained in the partition process, the partition scheme with the value of the modularity Q closest to 1 is selected as the optimal partition scheme of the power grid, and the partition tie lines between the sub-regions of the power grid are obtained.

3. The multi-time scale optimal scheduling method considering key-section flow safety according to claim 2, characterized in that, In step 1, the power grid partition process based on the adaptive GN algorithm is as follows: Step 1.1: simplify the topology structure of the power grid: convert the generator, substation, transformer and load node into a node in the topology graph; convert the power transmission line into a tie line in the topology graph, and ignore the power transmission line in the power plant; merge double-circuit or three-circuit power transmission lines; Step 1.2: Calculate the importance index of all transmission lines in the power grid, i.e. topological betweenness B e (l); Step 1.3: B is removed from the grid e (l) the transmission line with the largest value and based on a depth-first algorithm search whether a new partition is created; Step 1.4: calculate the modularity Q of each network, and store the modularity Q and the community partition in the current network; Step 1.5: if a new partition is generated, save the removed power transmission line and the partition result, and calculate the modularity Q of the current network; compare the difference between the modularity Q and the original modularity; if the difference is less than the original modularity, store and update the modularity Q and the community partition in the network, otherwise, perform the next network partition; Step 1.6: search whether there is a removable power transmission line; If there is, execute step 1.2, if not, execute step 1.7; Step 1.7: stop removing the transmission line in the power grid, search whether there is a single node constituting the sub-region of the power grid, if yes, merge the single node constituting the sub-region with the nearest sub-region, if not, execute step 1.8; Step 1.8: output the saved power grid partition result of step 1.4: save the current modularity Q.

4. The multi-time scale optimal scheduling method considering key-section flow safety according to claim 3, characterized in that, In step 1, in the process of determining the key section flow safety limit value, a key transmission section flow limit value optimization model based on linear programming theory is established, and the objective function of the key transmission section flow limit value optimization model is: ； In the formula, For the power flow of all transmission lines within the cross section The sum; Let be the power flow of transmission line l at time t; S be the set of all transmission lines that make up this section.

5. The multi-time scale optimal scheduling method considering key-section flow safety according to claim 4, characterized in that, The constraint conditions of the key transmission section flow limit value optimization model include: Power balance constraint: ； wherein, is the total number of thermal generating units g; is the total number of wind generating units; is the total number of transmission lines l; is the active power output of thermal generating unit g at time t; is the active power output of wind generating unit s at time t; s(l) and r(l) are the sending bus and receiving bus of transmission line l, respectively; is the system load forecast value at time t; is the transmission power flow of transmission line l; Thermal power unit output constraint: ； In the formula, and are the upper and lower limits of the thermal power unit output, respectively. Interconnection line flow constraint: ； wherein, and are the voltage phase angles of the bus at time t injected and injected out of the bus by transmission line l, respectively; is the reactance value of transmission line l; and are the upper and lower limits of the power flow transmitted by transmission line l, respectively. Node phase angle constraint: ； wherein and are the upper and lower limits, respectively, of the voltage phase angle of the voltage phase angle is the balanced node voltage phase angle.

6. The multi-time scale optimal scheduling method considering key-section flow safety according to claim 5, characterized in that, In step 2, the objective function of the established deterministic day-ahead optimal scheduling model is: ； In the formula, is the total scheduling time; a, b, c are the operation cost coefficients of the thermal power unit; are respectively the start / stop cost of the thermal power unit g; is the wind power penalty factor; is the predicted value of the wind power unit at time t.​ 7. The multi-time scale optimal scheduling method considering key-section flow safety according to claim 6, characterized in that, The constraint conditions of the deterministic day-ahead optimal scheduling model include system power balance constraint, line flow safety constraint, thermal power unit start / stop cost constraint, unit output constraint, climbing constraint, minimum start / stop time constraint, and wind power output constraint, which are as follows: System power balance constraint: ； Thermal power unit output constraint: ； In the formula, and are the start-stop states of the thermal power unit g at the tth and (t-1)th moments, and 1 indicates start-up, otherwise 0; and are the upward and downward standby capacities called at the tth moment of the thermal power unit. Thermal power unit start / stop cost constraint: ； wherein and are start / stop machine cost factors, respectively; Thermal power unit climbing constraint: ； In the formula, and are the maximum uphill and downhill ramp rates of the thermal power unit, respectively. Thermal power unit minimum start / stop time constraint: ； wherein, and respectively are the on and off time of the unit g at time t; and respectively are the minimum on and off time of the unit g; Wind power output constraint: ； Key section flow safety constraint: ； In the formula, and are the upper and lower limits of the power flow of the critical transmission section S, respectively.

8. The multi-time scale optimal scheduling method considering key-section power flow security according to claim 7, characterized in that, In step 3, when establishing the day-ahead two-stage distribution robust optimal scheduling model, the deterministic day-ahead optimal scheduling model is expressed in a linear matrix form: x represents the continuous variable related to output adjustment; and the objective function of the day-ahead two-stage distribution robust optimal scheduling model is: ； wherein, is the feasible region of the second stage operation scenario; p k is the probability value of each discrete scenario; and are the up and down reserve capacities called by the thermal power unit g in the kth discrete scenario, respectively; and are the predicted and actual outputs of the wind power unit in the discrete scenario k, respectively; K is the total number of discrete scenarios; is the unit cost required for up and down adjustment of the thermal power unit reserve capacity.

9. The multi-time scale optimal scheduling method considering key-section power flow security according to claim 8, characterized in that, The second stage constraint condition in the day-ahead two-stage distribution robust optimal scheduling model is determined as follows: In order to cope with the uncertainty of wind power output, the second stage needs to call the thermal power reserve capacity, and the related constraints include: System power balance constraint: ； wherein is the transmission flow of the transmission line / for the discrete scenario k; Wind power output constraint: ； Thermal power reserve capacity up / down constraint: ； In the formula, is the maximum limit of the standby capacity called by the thermal power unit; Key section flow safety constraint: ； Based on the N A discrete wind power output scenario set obtained after clustering of M historical data, an initial probability value of each discrete scenario is obtained. Considering the uncertainty of the probability distribution of each discrete scenario, the probability distribution value of the multiple discrete wind power scenarios is further constrained with the initial probability distribution value as the center, comprehensive norm constraint conditions containing 1-norm and ∞-norm, and the feasible region is Ω ∞ 1 and Ω The feasible region Ω of the second stage operation scenario is: ； wherein is an initial probability value for the kth discrete scenario obtained from historical data; is a 1-norm constraint condition; is an ∞-norm constraint condition; and are the maximum allowable deviations of the probabilities for the 1-norm and the ∞-norm, respectively. satisfies a confidence ； In the formula, Pr{·} represents the probability that the event is true; Let the right side of the inequality above be the confidence of the probability distribution value α1 and α ∞ The above equation is converted to: ； Introduce 0-1 auxiliary variables and Handling 1-norm constraints, introduce 0-1 auxiliary variables and Handling ∞-norm constraints, linearized as follows: ； ； wherein and are probability values p k relative positive and negative offsets.

10. The multi-time scale optimal scheduling method considering key-section power flow security according to claim 9, characterized in that, In step 4, the day-ahead rolling optimal scheduling model is as follows: The prediction time domain and control time domain of the day-ahead scheduling are determined, a model predictive control rolling optimization model is established, the latest prediction result is used to arrange the thermal power and wind power output in the prediction window of the future prediction time domain, but only the scheduling plan in the front control time domain is executed, the output of the system equipment in the next period is updated, and the above process is repeatedly repeated until the optimization in the entire scheduling period is completed; The objective function is as follows: ； In the formula, t s is the current scheduling time; T s is the intraday control time domain; and are the intraday outputs of the thermal power unit and the wind power, respectively; and are the day-ahead output of the thermal power unit and the intraday predicted output of the wind power, respectively. Constraint condition: The constraint conditions of the day-ahead rolling optimal scheduling model include the same thermal power unit output constraint, system power balance constraint, wind power output constraint and key section flow safety constraint as the constraint conditions of the deterministic day-ahead optimal scheduling model; in addition, the following constraints are included: Thermal power unit climbing constraint: 。 11. The multi-time scale optimal scheduling method considering key-section power flow security according to claim 10, characterized in that, In step 5, the solving process of the day-ahead two-stage distribution robust optimal scheduling model considering the uncertainty of wind power is as follows: The day-ahead two-stage distribution robust model is expressed as a "min-max-min" three-layer two-stage problem, as shown below: ； wherein, and are the first-stage decision variables x and the second-stage decision variables y, respectively k is the cost coefficient matrix of the objective function; is the cost coefficient of the second-stage wind curtailment penalty; , and are the coefficient matrices of x, y k , ξ k , respectively; is the constant matrix; ξ k is the wind power output under the kth discrete scenario; is the occurrence probability of the kth discrete scenario; is the set of first-stage decision variables; is the set of second-stage decision variables associated with the discrete scenario k; (1) The main problem is to solve the optimal solution that meets the constraint condition under the known bad probability distribution, and provide a lower bound value for the objective function: ； In the formula, N B is the maximum number of iterations; y0is the second-stage decision variable under a given set of initial values; ξ0is the initial value of wind power output; m is the iteration number; is the introduced continuous auxiliary variable; is the maximum number of iterations; (2) The sub-problem is to search for the current worst wind power output probability distribution under the decision variable x made by the main problem, and generate the corresponding constraint containing the two-stage recourse variable to be added to the main problem: ； Let The above equation is transformed into: ； The two-stage model iterates and updates variables between the main and sub-problems until a given precision value is reached, stopping iteration.

12. The multi-time scale optimal scheduling method considering key-section flow safety according to claim 11, characterized in that, In step 5, when solving the intraday rolling optimization scheduling model based on model predictive control theory, a multi-time scale idea is used to model the intraday load and wind power prediction values in detail, with the minimum adjustment amount of each unit in the system as the target, to improve the robustness of the power system in dealing with short-term rapid fluctuations in sources and loads.

Images