Distributed stochastic optimization and load transfer method for hybrid power transmission and distribution system

Abstract

The invention relates to a distributed stochastic optimization and load transfer method for a hybrid power transmission and distribution system. The method comprises the following steps: step 1, constructing a distributed optimization planning model of the power transmission and distribution system; 2, distributed optimization solution is carried out, wherein an ATC-based distributed optimization algorithm is adopted to solve the joint optimization planning model; step 3, establishing a load transfer mathematical model, wherein the control variable is the load cut-off amount of the switch on the line and the load point in the power-losing area; 4, adopting a heuristic algorithm to quickly search a solution when the load does not need to be cut off; step 5, if the loads in the power-losing area cannot be completely recovered, carrying out load removal by adopting particle swarm optimization random optimization so as to ensure that the system meets security constraints; the method has the advantages that decomposition optimization modeling is utilized, the second-order cone programming model is constructed, the spanning tree and the loop matrix are combined, and controllability is considered during load removal.

Description

Technical Field

[0001] The present invention belongs to the technical field of power transmission and distribution systems, and particularly relates to a distributed random optimization and load transfer method for a hybrid power transmission and distribution system. Background Art

[0002] The transmission and distribution system is a hybrid physical system with the spatiotemporal coupling characteristics of "source-grid-load". The information exchange interface between the transmission and distribution systems is realized through substation nodes. Under the traditional power flow, the transmission system can be regarded as a virtual power plant relative to the distribution system, and the distribution system can be regarded as a virtual load relative to the transmission system. In recent years, with the large-scale access of distributed power sources (DG) at the distribution system level, the distribution system has achieved a transformation from passive to active. However, the existing independent operation planning method is difficult to solve the system operation problems such as voltage limit exceeding, DG absorption, and line congestion caused by the high proportion of renewable energy access. At the same time, due to the time-varying characteristics of DG output and load access, in order to ensure the safe and reliable operation of the system, the planning problem needs to fully take into account the above uncertainties. The impact on system operation, therefore, it is necessary to comprehensively utilize system synergy theory and stochastic programming to achieve global optimization of transmission and distribution system operation; in addition, with the development of the national economy, the scale of power grids is expanding and becoming more complex, and the probability of power grid failures is also increasing accordingly. When a distribution network fails, how to restore power supply to as many loads as possible in the power-off area as soon as possible after the failure occurs through operations such as opening and closing switches and shedding loads while ensuring that the network meets safety constraints, so as to minimize losses has become an urgent problem to be solved; therefore, it is very necessary to provide a hybrid transmission and distribution system distributed stochastic optimization and load transfer method that utilizes decomposition optimization modeling, constructs a second-order cone programming model, combines spanning tree and loop matrix, and considers controllability when shedding loads. Summary of the Invention

[0003] The purpose of the present invention is to overcome the shortcomings of the existing technology and provide a distributed stochastic optimization and load transfer method for a hybrid power transmission and distribution system by utilizing decomposition optimization modeling, constructing a second-order cone programming model, combining a spanning tree and a loop matrix, and considering controllability when load shedding.

[0004] The object of the present invention is achieved as follows: a method for distributed random optimization and load transfer of a hybrid power transmission and distribution system, comprising the following steps:

[0005] Step 1: Construct a distributed optimization planning model for the transmission and distribution system;

[0006] Step 2: Distributed optimization solution: Use the ATC-based distributed optimization algorithm to solve the joint optimization planning model;

[0007] Step 3: Establish a mathematical model for load transfer: the control variables are the switches on the line and the load shedding amount of the load points in the power outage area;

[0008] Step 4: Use heuristic algorithm to quickly search for the solution when load shedding is not required;

[0009] Step 5: If it is impossible to restore all the loads in the power-off area, the particle swarm algorithm is used for random optimization to remove the loads to ensure that the system meets the safety constraints.

[0010] The construction of the distributed optimization planning model of the power transmission and distribution system in step 1 includes the following steps:

[0011] Step 1.1: Decomposition optimization modeling;

[0012] Step 1.2: Shared variable modeling: Select the active power exchanged between the transmission and distribution systems as the shared variable. Assuming that the positive direction of the exchanged power flows from the transmission system to the distribution system, it can be regarded as a virtual power plant relative to the distribution system and a virtual load for the transmission system. When the shared variable is negative, it means that power is transferred from the distribution system to the adjacent transmission system. Therefore, the shared variable can be expressed as: t 2q,s =P T,q,s (5), r 2q,s =P D,q,s (6), where P T,q,s 、P D,q,s are the exchanged active power between the transmission system and the qth distribution system on the transmission system side and the distribution system side under scenario s, respectively;

[0013] Step 1.3: Transmission system planning sub-model;

[0014] Step 1.4: Distribution system planning sub-model.

[0015] The transmission system planning sub-model in step 1.3 is as follows: for the transmission system, the decision variable related to investment is the new line, and the decision variable related to operation is the generator output and the exchange power connected to each distribution system. Its optimization objective is:

[0016] Where, is a binary variable; is the number of new lines on corridor ij; is the unit investment cost of new lines on corridor ij; l ij is the length of corridor ij; C G,i (·) is the power generation cost function of the generator at node i; P G,i,s is the active power output of the generator at node i in scenario s; ρ T,qThe unit price of electricity purchased by the transmission system from the qth connected distribution system; Ω C ,Ω G ,Ω D They are respectively a set of corridors for newly built lines, a set of generator installation nodes, and a set of distribution systems; the constraints of the transmission system planning submodel include constraints on the number of newly built lines, power balance constraints on transmission system nodes, DC power flow constraints on existing lines, DC power flow constraints on newly built lines, capacity constraints on existing lines, capacity constraints on newly built lines, generator output constraints, and shared variable constraints on the transmission system side.

[0017] The distribution system planning sub-model in step 1.4 is as follows: for distribution system q, the decision variables related to investment are line reconstruction, new line construction, and substation expansion, while the decision variables related to operation are DG output and exchange power with the transmission system. Its optimization objective is: Where, is a binary variable; u ss Both are binary variables; C ss are the unit investment cost of the line reconstruction on the corridor protection ij and the substation ss expansion cost respectively; C DG,i (·) is the DG power generation cost function at node i; P DG,i,s is the active power output of the DG at node i in scenario s; ρ D,q The unit price of electricity supplied by the distribution system q to the transmission system; Ω s ,Ω DG are the existing substation set and the DG installation node set respectively; the constraints of the distribution system planning submodel include the number of transformation lines constraint, distribution system node power balance constraint, line flow equation, line capacity constraint, node voltage constraint, DG output constraint, radial constraint and distribution system side shared variable constraint.

[0018] The distributed optimization solution in step 2 includes the following steps:

[0019] Step 2.1: Second-order cone relaxation of the distribution system planning sub-model;

[0020] Step 2.2: Use the ATC-based distributed optimization algorithm to solve the joint optimization planning model.

[0021] The second-order cone relaxation distribution system planning sub-model in step 2.1 is specifically as follows: the non-convex optimization problem containing cross terms and quadratic nonlinear terms in equations (18)-(21) and (27) is converted into a convex optimization problem of the second-order cone programming model by comprehensively utilizing the large M method and the second-order cone relaxation. and a sufficiently large positive number M, equations (18-(23) and (27) can be rewritten as:

[0022] (V i min ) 2 ≤U i,s ≤(V i max ) 2 (35), Formula (29) can be relaxed to a second-order cone constraint, namely: After the above transformation, the original distribution system planning sub-problem is modeled as a convex second-order cone programming problem.

[0023] The step 2.2 uses the ATC-based distributed optimization algorithm to solve the joint optimization planning model, including the following steps:

[0024] Step 2.21: Set the initial value of the number of inner loop iterations l = 0 and the initial value of the number of outer loop iterations d = 0, and set the consistency constraint penalty function coefficient Weight Shared variables with the transmission system side

[0025] Step 2.22: Let l = l + 1, solve the qth distribution system planning subproblem, and obtain the shared variables on the distribution system side

[0026] Step 2.23: Solve the transmission system planning subproblem and obtain the transmission system side shared variables

[0027] Step 2.24: Determine whether the inner loop has converged. The inner loop convergence criterion is If it does not converge, return to step 2.22, otherwise continue;

[0028] Step 2.25: Determine whether the outer loop has converged. The outer loop convergence criterion is |c d |≤ε2(40), c d -c d-1 |≤ε3(41), if converged, output the optimal solution and exit the loop, otherwise continue;

[0029] Step 2.26: Let d = d + 1, according to the formula Update consistency constraint penalty function coefficients and weights The value of

[0030] Step 2.27: Setup and l=0, return to step 2.22.

[0031] The specific steps of establishing the load transfer mathematical model in step 3 are as follows: ① The load shedding amount is minimal: ②The sum of the number of switch actions is the least: At the same time, the recovery strategy needs to meet the following constraints: ① Topology constraint: the system maintains a radiation state without closed loops and islands; ② Branch current does not exceed the limit: I i ≤I i,N , where I i is the steady-state current flowing through line i; I i,N is the rated current of line i; ③ bus voltage does not exceed the limit: U i,min ≤U i ≤U i,max , where U i,min 、U i,max are the lower and upper limits of the node voltage respectively; ④ Load removal constraint: that is, the load removed from the load point cannot exceed the load of this load point 0≤P icut ≤P i ; ⑤ Load constraint: The load is controllable, uncontrollable or incompletely controllable; the controllable load can change continuously, and any load not exceeding the maximum load of the load point can be cut off as needed; when the uncontrollable load needs to be cut off, all loads can only be cut off; incompletely controllable loads can only cut off a certain amount of load according to a certain load level.

[0032] In step 4, a heuristic algorithm is used to quickly search for a solution when load shedding is not required. Specifically, the objective function in the random method includes: the sum of the load shedding amounts of various types, the number of switch operations, and the over-limit penalty term. The objective function expression is as follows: in, Where, the first term is the sum of the active power removal of each load in the power failure zone; d j is the weight coefficient of the jth load; the second term is the total number of switch actions; the third term is the voltage over-limit penalty term; the fourth term is the power over-limit penalty term; V j is the voltage amplitude at point j; V j,max 、V j,min are the upper and lower limits of the voltage amplitude at point j; S j is the current apparent power of the jth line; S j,max is its power upper limit; p ,λ n ,λ u ,λ s are the weight coefficients of each item in the objective function respectively; the heuristic algorithm system automatically maintains the radial topology during the search process, and the safety constraints that need to be met are: the power flow equation converges, the line is not overloaded, and the voltage does not exceed the limit.

[0033] The heuristic algorithm parameters in step 4 include: ① the remaining capacity of the tie switch IM :I M =min{(I 1,max -I1),(I 2,max -I2),...,(I n,max -I n )}(47), where I1...I n The I of a certain tie switch is the difference between the maximum current allowed by each branch on the path from the tie switch to the power point and the actual current. M Indicates how much load the tie switch can add before an overload occurs; ② The electrical distance Z between the busbar where the voltage overshoot occurs and a tie switch path : Where V ts Indicates the voltage at the voltage crossing busbar; is the sum of all load currents from the busbar to a certain tie switch; ③ The load that can be transferred by the section switch in the power failure zone I SS : I SS Represents the sum of all load currents downstream of the section switch.

[0034] Beneficial effects of the invention: The present invention is a distributed random optimization and load transfer method for a hybrid power transmission and distribution system. During use, the present invention takes into account the uncertainty of renewable energy output and load fluctuations, and utilizes the random programming theory based on the scenario analysis method and the distributed optimization method based on ATC to coordinate planning of the power transmission and distribution system containing DG. Under the premise of meeting the consistency of the interactive information between the power transmission and distribution systems, the joint optimization planning problem of the power transmission and distribution system is decomposed into a transmission system planning sub-problem and a series of distribution system planning sub-problems, realizing parallel computing during the coordinated optimization planning of the power transmission and distribution system. For the distribution system planning sub-problem, based on the distribution system power flow equation and constraint relaxation technology, The invention adopts the technology to construct a second-order cone convex relaxation programming model for distribution system planning, thereby improving iterative convergence; adopts a distributed optimization algorithm with double-layer cyclic iteration to solve the constructed model, which helps to improve the economy of the transmission and distribution system planning scheme and the DG absorption capacity; the invention combines heuristic search with random optimization and adopts a new encoding method in the random optimization process. The generated solution satisfies the radial topology constraint without the need for traditional random algorithm to perform topology verification, thereby improving the solution efficiency, greatly shortening the calculation time, and better realizing load transfer; the invention has the advantages of utilizing decomposition optimization modeling, constructing a second-order cone programming model, combining the spanning tree and the loop matrix, and considering controllability when load shedding. BRIEF DESCRIPTION OF THE DRAWINGS

[0035] Figure 1 Schematic diagram of the two-layer hybrid power transmission and distribution system of the present invention.

[0036] Figure 2Schematic diagram of decomposition of the optimization problem of the present invention.

[0037] Figure 3 This is a flowchart of solving the joint optimization planning of the power transmission and distribution system of the present invention.

[0038] Figure 4 This is a flow chart of the heuristic algorithm of the present invention. DETAILED DESCRIPTION

[0039] The present invention will be further described below with reference to the accompanying drawings.

[0040] Example 1

[0041] like Figure 1-4 As shown, a distributed stochastic optimization and load transfer method for a hybrid power transmission and distribution system includes the following steps:

[0042] Step 1: Construct a distributed optimization planning model for the transmission and distribution system. This model uses multi-scenario technology to convert the effects of load fluctuation uncertainty on grid operation into a series of deterministic scenarios. Each scenario represents a single load level rather than a typical day. The model optimizes the generator output in the transmission system, the DG output in the distribution system, and the shared variables at the boundary nodes between the two.

[0043] In this embodiment, step 1 includes the following steps:

[0044] Step 1.1: Decomposition optimization modeling;

[0045] Specifically: A hybrid transmission and distribution system can be decomposed into a transmission system, a distribution system and a boundary node system connecting the two from a physical perspective. Figure 1 A schematic diagram of a two-layer hybrid transmission and distribution system is given to reflect the above structure. The elements on the first layer represent the transmission system, and the elements on the second layer represent the distribution system. After a transmission system and n distribution systems are interconnected, they can achieve coordinated operation. The traditional centralized optimization planning model of the transmission and distribution system can be expressed as: Where x and y are the decision variable vectors for investment and operation optimization of hybrid transmission and distribution system, respectively; f is the comprehensive fee; f inv 、f ope are investment and operating costs respectively; g and h are inequality constraints and equality constraints respectively; for the model in formula (1), if the transmission and distribution systems are regarded as independent optimization entities, they can be decomposed into a transmission system planning sub-problem and a series of distribution system planning sub-problems, that is, the objective function and constraints in formula (1) can correspond to Figure 1 Each element in each layer is represented independently, so f, g, and h can be equivalently written as f 11 +...+f 2n 、[g 11 …g2n ]、[h 11 …h 2n ];

[0046] In order to model the connections and couplings between the various subsystems in the hybrid transmission and distribution system, it is necessary to divide the optimization variables of each subsystem into two categories: ① local variables, i.e., variables that are only related to the independent optimization of the subsystem and do not participate in the optimization decision of other subsystems; ② shared variables, i.e., variables that are shared with the optimization of other subsystems and participate in the optimization decision of multiple subsystems at the same time;

[0047] for Figure 1 For each element in each layer, local variables include investment-related optimization variables x pq and the optimization variables y related to the operation in each scenario pq,s , shared variables refer to the optimization variables z that interact with other elements in each scenario pq,s ; Then set the shared variable z pq,s Further decomposed into target variable t pq,s and the response variable r pq,s ; where t pq,s is a shared variable processed from the perspective of the transmission system, that is, it is used as an optimization variable in the transmission system planning sub-problem; r pq,s It is a shared variable processed from the perspective of the distribution system, that is, it is used as an optimization variable in the distribution system planning sub-problem. By using the above decomposition method, the optimization problem of each subsystem can be solved independently, such as Figure 2 As shown, the target variable t pq,s and the response variable r pq,s Essentially, they represent the same shared variables. While each subsystem is independently planned, the above variables must be consistent. Therefore, consistency constraints and related variables c must be introduced in the decomposition optimization. pq,s , c pq,s =t pq,s -r pq,s =0(2), for the consistency constraint of shared variables, the penalty function is used to relax in the decomposition optimization method, and the penalty function is defined as π. After the above processing, the original problem in formula (1) can be rewritten as the following stochastic model: Where, c is the vector composed of all the inconsistencies of shared variables, c=[c 21 …c 2n ]; m is the number of scenes; D s is the time corresponding to scene s;

[0048] To ensure the consistency of shared variables, the augmented Lagrangian penalty function is used. For any subsystem, the planning problem can be modeled as: Where, v pq,s、w pq,s are the coefficient vector and weight vector of the penalty function respectively; the symbol represents the Hadamard product, i.e., the multiplication calculation term by term; Equation (4) can be solved independently for each subsystem while ensuring the consistency of shared variables.

[0049] Step 1.2: Shared variable modeling: Select the active power exchanged between the transmission and distribution systems as the shared variable. Assuming that the positive direction of the exchanged power flows from the transmission system to the distribution system, it can be regarded as a virtual power plant relative to the distribution system and a virtual load for the transmission system. When the shared variable is negative, it means that power is transferred from the distribution system to the adjacent transmission system. Therefore, the shared variable can be expressed as: t 2q,s =P T,q,s (5), r 2q,s =P D,q,s (6), where P T,q,s 、P D,q,s are the exchanged active power between the transmission system and the qth distribution system on the transmission system side and the distribution system side under scenario s, respectively.

[0050] Step 1.3: Transmission system planning sub-model;

[0051] Specifically, for the transmission system, the decision variable related to investment is the new line, while the decision variable related to operation is the generator output and the exchange power connected to each distribution system. The optimization objective is: Where, is a binary variable. If a new line is built on corridor ij, Take 1, otherwise Take 0; is the number of new lines on corridor ij; is the unit investment cost of new lines on corridor ij; l ij is the length of corridor ij; C G,i (·) is the power generation cost function of the generator at node i; P G,i,s is the active power output of the generator at node i in scenario s; ρ T,q The unit price of electricity purchased by the transmission system from the qth connected distribution system; Ω C ,Ω G ,Ω Dare the set of corridors for newly built lines, the set of generator installation nodes, and the set of distribution systems, respectively. For formula (7), the first term represents the cost of newly built lines, the second term represents the cost of generator power generation, the third term represents the cost of electricity purchased by the transmission system from the connected distribution system, and the remaining terms represent the penalty costs related to the shared variables on the distribution system side. The constraints of the transmission system planning submodel include the number of newly built lines, the power balance constraint of the transmission system nodes, the DC power flow constraint of the existing lines, the DC power flow constraint of the newly built lines, the capacity constraint of the existing lines, the capacity constraint of the newly built lines, the generator output constraint, and the shared variable constraint on the transmission system side.

[0052] In this embodiment, ① the number of newly created lines is constrained: Where, is the upper limit of new lines on corridor ij;② Transmission system node power balance constraint: Where, is the active power of the kth line on corridor ij under scenario s; P L,i,s is the active load of node i in scenario s; Ω L For all line sets; ③ DC power flow constraints of existing lines: Where, is the kth line susceptance between nodes i and j; is the number of existing lines between nodes i and j; θ i,s ,θ j,s are the phase angles of nodes i and j in scenario s respectively; Ω0 is the existing line set; ④ DC power flow constraints for new lines: Where, is a binary variable. If the kth line is newly built on corridor ij, Take 1, otherwise, Take 0; M ij is a sufficiently large positive number, and its value can be Among them, (p,q) ij ∈Ω0 represents a path from node i to node j through several branches pq in Ω0; ⑤ Existing line capacity constraints: Where, is the capacity of the kth line between nodes i and j; ⑥ New line capacity constraint: ⑦Generator output constraints: Where, are the upper and lower limits of the active power generated by the generator at node i;⑧ shared variable constraints on the transmission system side: Where, are the upper and lower limits of the active power exchanged between the transmission system and the qth distribution system, respectively.

[0053] Step 1.4: Distribution system planning sub-model.

[0054] Specifically, for the distribution system q, the decision variables related to investment are line reconstruction, new line construction, and substation expansion, while the decision variables related to operation are DG output and exchange power with the transmission system. The optimization objective is: Where, is a binary variable. If the kth line on corridor ij is renovated, Take 1, otherwise Take 0;u ss As binary variables, if the substation ss is expanded, u ss Take 1, otherwise u ss Take 0; C ss are the unit investment cost of the line reconstruction on the corridor protection ij and the substation ss expansion cost respectively; C DG,i (·) is the DG power generation cost function at node i; P DG,i,s is the active power output of the DG at node i in scenario s; ρ D,q The unit price of electricity supplied by the distribution system q to the transmission system; Ω s ,Ω DG are the set of existing substations and the set of DG installation nodes respectively; for formula (16), the first three items represent the cost of line transformation, the cost of new line construction and the cost of substation expansion, the fourth item represents the cost of DG power generation, the fifth item represents the revenue from electricity sales from the distribution system to the transmission system, and the remaining items represent the penalty costs related to the shared variables on the transmission system side; it should be noted that P D,q,s The positive direction is the flow from the distribution system q to the transmission system, so the fifth term is negative. The constraints of the distribution system planning submodel include the number of transformed lines, the power balance constraint of the distribution system nodes, the line power flow equation, the line capacity constraint, the node voltage constraint, the DG output constraint, the radial constraint, and the shared variable constraint on the distribution system side.

[0055] In this embodiment, ① the number of modified lines is constrained: 0≤n ref ≤n ref,max (17), where n ref 、n ref,max are the number of transformed lines and their upper limit; ② Power balance constraints of distribution system nodes: Where, is the current amplitude of the kth line on corridor ij under scenario s; V i,s is the voltage amplitude of node ii under scenario s; The work is the resistance and reactance of the kth line on corridor ij; Q DG,i,s , Q L,i,sare the reactive power of the kth line on corridor ij under scenario s, the reactive output of the DG at node i, and the reactive load at node i; ③ Line power flow equation: ④Road capacity constraints: Where, is the current upper limit of the kth line on corridor ij; ⑤Node voltage constraint: V i min ≤V i,s ≤V i max (23), where V i max 、V i min are the upper and lower limits of the voltage amplitude of node i respectively; ⑥DG output constraint: In the formula are the upper and lower limits of the active power emitted by the DG at node i; are the upper and lower limits of the reactive power emitted by the DG at node i; ⑦ Radial constraint: In the formula, N + The number of newly added load nodes; is a binary variable, if the corridor The kth line on the line is closed, Take 1, otherwise Take 0; is the virtual active power of the kth line on corridor ij under scenario s;⑧Shared variable constraints on the distribution system side:

[0056] Step 2: Distributed optimization solution: Use the ATC-based distributed optimization algorithm to solve the joint optimization planning model;

[0057] In this embodiment, step 2 includes the following steps:

[0058] Step 2.1: Second-order cone relaxation of the distribution system planning sub-model;

[0059] Specifically, the large M method and second-order cone relaxation are used to transform the non-convex optimization problems containing cross terms and quadratic nonlinear terms in Equations (18)-(21) and (27) into convex optimization problems of the second-order cone programming model. By introducing two new variables and a sufficiently large positive number M, equations (18-(23) and (27) can be rewritten as:

[0060] (V i min ) 2 ≤U i,s ≤(V imax ) 2 (35), Formula (29) can be relaxed to a second-order cone constraint, namely: After the above transformation, the original distribution system planning sub-problem is modeled as a convex second-order cone programming problem.

[0061] Step 2.2: Use the ATC-based distributed optimization algorithm to solve the joint optimization planning model.

[0062] Specifically, the algorithm is an iterative solution method consisting of two layers of loops, inner and outer, which can be used to solve the target optimization problem in complex engineering systems. In the inner loop, the consistency constraint penalty function coefficient v and weight w are kept unchanged, and the optimization planning problems of each transmission and distribution system are solved alternately; in the outer loop, the consistency constraint penalty function coefficient v and weight w are updated according to the solution results of the inner loop. The solution process of the joint optimization planning model of the transmission and distribution system is as follows: Figure 3 As shown, the following steps are included:

[0063] Step 2.21: Set the initial value of the number of inner loop iterations l = 0 and the initial value of the number of outer loop iterations d = 0, and set the consistency constraint penalty function coefficient Weight Shared variables with the transmission system side

[0064] Step 2.22: Let l = l + 1, solve the qth distribution system planning subproblem, and obtain the shared variables on the distribution system side

[0065] Step 2.23: Solve the transmission system planning subproblem and obtain the transmission system side shared variables

[0066] Step 2.24: Determine whether the inner loop has converged. The inner loop convergence criterion is If it does not converge, return to step 2.22, otherwise continue;

[0067] Step 2.25: Determine whether the outer loop has converged. The outer loop convergence criterion is |c d |≤ε2(40), c d -c d-1 |≤ε3(41), if converged, output the optimal solution and exit the loop, otherwise continue;

[0068] Step 2.26: Let d = d + 1, according to the formula Where γ = 0.25, update consistency constraint penalty function coefficient and weights The value of

[0069] Step 2.27: Setup and l=0, return to step 2.22.

[0070] It should be noted that to improve the convergence speed and accuracy of the iterations, the present invention makes the following improvements to the algorithm: ① Introducing an augmented Lagrangian penalty function; ② Dynamically updating the penalty function coefficients and weights. The addition of a quadratic penalty term to the objective function helps enhance the convexity of the planning model, and the penalty function weight initially exhibits a monotonically increasing trend with the number of iterations, effectively improving the algorithm's convergence characteristics.

[0071] Step 3: Establish a mathematical model for load transfer: the control variables are the switches on the line and the load shedding amount of the load points in the power outage area;

[0072] Step 4: Use heuristic algorithm to quickly search for the solution when load shedding is not required;

[0073] Step 5: If it is impossible to restore all the loads in the power-off area, the particle swarm algorithm is used for random optimization to remove the loads to ensure that the system meets the safety constraints.

[0074] The present invention is a distributed random optimization and load transfer method for a hybrid power transmission and distribution system. In use, the present invention takes into account the uncertainty of renewable energy output and load fluctuations, and utilizes the random programming theory based on the scenario analysis method and the distributed optimization method based on ATC to coordinate the planning of the power transmission and distribution system containing DG. Under the premise of meeting the consistency of the interactive information between the power transmission and distribution systems, the power transmission and distribution system joint optimization planning problem is decomposed into a transmission system planning sub-problem and a series of distribution system planning sub-problems, realizing the parallel operation in the coordinated optimization planning of the power transmission and distribution system. For the distribution system planning sub-problem, a power distribution system power flow equation and constraint relaxation technology are used to construct a The second-order cone convex relaxation programming model for distribution system planning improves iterative convergence; the distributed optimization algorithm with double-layer cyclic iteration is used to solve the constructed model, which helps to improve the economy of the transmission and distribution system planning scheme and the DG absorption capacity; the present invention combines heuristic search with random optimization and adopts a new encoding method in the random optimization process. The generated solution satisfies the radial topology constraint without the need for traditional random algorithms for topology verification, thereby improving the solution efficiency, greatly shortening the calculation time, and better realizing load transfer; the present invention has the advantages of utilizing decomposition optimization modeling, constructing a second-order cone programming model, combining the spanning tree and the loop matrix, and considering controllability when load shedding.

[0075] Example 2

[0076] like Figure 1-4 As shown, a distributed stochastic optimization and load transfer method for a hybrid power transmission and distribution system includes the following steps:

[0077] Step 1: Construct a distributed optimization planning model for the transmission and distribution system;

[0078] Step 2: Distributed optimization solution: Use the ATC-based distributed optimization algorithm to solve the joint optimization planning model;

[0079] Step 3: Establish a mathematical model for load transfer: the control variables are the switches on the line and the load shedding amount of the load points in the power outage area;

[0080] In this embodiment, specifically: ① The load shedding amount is minimal: ②The sum of the number of switch actions is the least: At the same time, the recovery strategy needs to meet the following constraints: ① Topology constraint: the system maintains a radiation state without closed loops and islands; ② Branch current does not exceed the limit: I i ≤I i,N , where I i is the steady-state current flowing through line i; I i,N is the rated current of line i; ③ bus voltage does not exceed the limit: U i,min ≤U i ≤U i,max , where U i,min 、U i,max are the lower and upper limits of the node voltage respectively; ④ Load removal constraint: that is, the load removed from the load point cannot exceed the load of this load point 0≤P icut ≤P i ; ⑤ Load constraint: The load is controllable, uncontrollable or incompletely controllable; the controllable load can change continuously, and any load not exceeding the maximum load of the load point can be cut off as needed; when the uncontrollable load needs to be cut off, all loads can only be cut off; incompletely controllable loads can only cut off a certain amount of load according to a certain load level.

[0081] Step 4: Use heuristic algorithm to quickly search for the solution when load shedding is not required;

[0082] In this embodiment, specifically, the objective function in the random method includes: the sum of the load shedding amounts of various types, the number of switch operations, and the over-limit penalty term. The objective function expression is as follows: in, Where, the first term is the sum of the active power removal of each load in the power failure zone; d j is the weight coefficient of the jth load, which depends on the importance of the load at that point; the second term is the total number of switch actions; the third term is the voltage over-limit penalty term; the fourth term is the power over-limit penalty term; V j is the voltage amplitude at point j; V j,max 、V j,min are the upper and lower limits of the voltage amplitude at point j; S jis the current apparent power of the jth line; S j,max is its power upper limit; p ,λ n ,λ u ,λ s They are respectively the weight coefficients of each item in the objective function, and their values ​​are related to the importance of each item in the objective and the scale of the entire network. In order to eliminate over-limit, the weight value of the penalty item is larger; the smaller the value of the entire objective function, the better. The present invention takes the adaptation value as the reciprocal of C, f=1 / C, that is, the target values ​​are sorted from large to small; the heuristic algorithm system automatically maintains the radial topology during the search process, and the safety constraints that need to be met are: the power flow equation converges, the line is not overloaded, and the voltage does not exceed the limit.

[0083] The relevant parameters of the heuristic algorithm include: ① The remaining capacity of the tie switch I M :I M =min{(I 1,max -I1),(I 2,max -I2),...,(I n,max -I n )}(47), where I1...I n The I of a certain tie switch is the difference between the maximum current allowed by each branch on the path from the tie switch to the power point and the actual current. M Indicates how much load the tie switch can add before an overload occurs; ② The electrical distance Z between the busbar where the voltage overshoot occurs and a tie switch path : Where V ts Indicates the voltage at the voltage crossing busbar; is the sum of all load currents from the busbar to a certain tie switch; ③ The load that can be transferred by the section switch in the power failure zone I SS : I SS Represents the sum of all load currents downstream of the section switch.

[0084] In this embodiment, the heuristic algorithm is used to solve the problem when load shedding is not required. If all loads in the power outage area cannot be restored, the particle swarm optimization process is used to shelve the load. The steps are as follows:

[0085] (1) Initial area division: Based on the known system structure and fault information, a search is performed. At this point, the entire system can be divided into three areas: normal operating area, fault isolation area, and non-fault power loss area. The loads in the non-fault power loss area (abbreviated as "power loss area") are the targets for load transfer. After the fault occurs, the normal area still has normal power supply. After the division of each area, a power flow calculation is performed on the normal area to determine the initial state of the system after the fault occurs and is isolated.

[0086] (2) Searching for available tie switches: For a power outage area, search for tie switches available for this area, that is, tie switches connecting the power outage area and the normal working area, and calculate the remaining capacity of each tie switch; if there is no tie switch available for this power outage area or each tie switch has no remaining capacity (the remaining capacity of the tie switch is compared with a set very small number; if it is less than this value, it is considered that the tie switch has no remaining capacity), then this power outage area cannot be restored and the program ends; the following process assumes that there are multiple available tie switches;

[0087] (3) Close a tie switch to restore the entire power outage area: Sort the tie switches in descending order according to their remaining capacity. If there is a tie switch with a remaining capacity greater than the total load in the power outage area, close the largest switch, perform flow calculations, and check whether there is overload or voltage exceeding the limit. If overload occurs, according to the order of the tie switches, if there is a switch with a capacity greater than the total load, close it, disconnect the switch that was just closed, and recheck. If there is no overload, the tie switch closed at this time can restore the entire power outage area and obtain a transfer plan. If a tie switch is closed, After the power is turned off, there is no overload, but the voltage exceeds the limit. If there are still tie switches with a remaining capacity greater than the sum of the loads in the entire power-off zone, these tie switches are re-sorted according to the electrical distance from the switch to the over-limit bus. If the voltage exceeds the lower limit, the tie switches are sorted from small to large according to the electrical distance to reduce the voltage drop between the bus and the power supply switch. Conversely, if the voltage exceeds the upper limit, the tie switches are sorted from large to small according to the electrical distance to increase the voltage drop between the bus and the power supply switch. If any switch cannot restore the power-off zone, that is, the overload or voltage limit cannot be eliminated, go to step (4).

[0088] (4) Use multiple tie switches to restore power: If the entire power outage area cannot be restored after (3), close the tie switch with the largest remaining capacity, record it as ts1, and record the overload of ts1 at this time; if there is no other available tie switch in this power outage area except ts1, then the heuristic method cannot be restored, end the heuristic method, and switch to the particle swarm algorithm to remove the load; if there are other available tie switches in this power outage area, find a segmented switch in the power outage area that can be paired with other tie switches to form a switch pair; the purpose of finding a switch pair is: since closing only one tie switch ts1 cannot restore the entire area, let other tie switches with remaining capacity provide power to part of the load to reduce the power supply borne by ts1; the method of finding a switch pair for a certain tie switch ts2 is as follows:

[0089] A: Search all switch branches on the path from ts1 to ts2 and calculate the transfer capacity I of each switch branch SS ;

[0090] B: Finding Satisfaction I SS <I M And I SS The largest segment switch is denoted as ss2. If no segment switch that meets the conditions is found, then no segment switch is paired with the tie switch, and the tie switch is unavailable in this stage.

[0091] C: For the switch pair (ts2, ss2) found in B, close ts2 and open ss2. Calculate the power flow and verify if there is any over-limit. If there is, pair a sectionalizing switch downstream of ss2 with ts2 and re-verify until the section powered by the switch pair is free of over-limits, thus ensuring that the section powered by other switches is free of over-limits.

[0092] D: After finding the corresponding switch pairs for all tie switches except ts1, sort the switch pairs from large to small according to the transferable capacity, and operate these switch pairs in sequence. Each time a switch pair is operated, the switch pair is updated. When the load transferred from the ts1 power supply area is greater than the overload, a safety check is performed on the network at this time. If no overload occurs, the search process is stopped and a load restoration plan is obtained. If the overload cannot be eliminated after all switch pairs are operated, the random method is used to shed the load. The flowchart of the entire algorithm is shown below. Figure 4 shown.

[0093] Step 5: If it is impossible to restore all the loads in the power-off area, the particle swarm algorithm is used for random optimization to remove the loads to ensure that the system meets the safety constraints.

[0094] In this embodiment, the particle swarm algorithm is specifically as follows: ① Assume that in a D-dimensional target search space, the position X of the i-th particle (the potential solution of the optimization problem) is (x i1 ,x i2 ,...,x iD ), the flight speed V is (v i1 ,v i2 ,...,v iD ), record the best position P found by the particle itself so far i For (p i1 ,p i2 ,...,p iD ), the best position P found by the group so far g For (p g1 ,p g2 ,...,p gD ), each iteration updates the velocity and position by the following formula: Where ω is the inertia factor, which is generally 0.8-1.2; c1 and c2 are learning factors, which are generally 1.5-2.0; r1 and r2 are random numbers in the interval [0, 1]; v max 、v min Is the speed limit, usually set by the user, during iteration, if but if but

[0095] ② Coding method: First, the ends of the available tie switches that are not in the power-off zone are merged and treated as a virtual power source. All branches are closed. The network formed at this time is called the target network. When there are multiple available tie switches for a certain area to be restored, the target network formed contains multiple rings. When coding, it is necessary to ensure that the network meets the topological constraints and optimize the target. The control variables are the switches in the network and the load shedding amount. Therefore, the coding is divided into two parts: the first part is the branch number to be disconnected in the network, and the second part is the load shedding amount.

[0096] 1) Ensuring the encoding of topological constraints: If there are more than one available tie switches in a region to be restored, loops will be created in the target network. Assuming there are n tie switches, n-1 loops will be formed in the target network. A branch must be disconnected in each loop to maintain the radial network.

[0097] 1.1) Network simplification: This primarily involves operations such as branch deletion and merging. If a branch is not located in any loop in the network, its opening and closing will not affect the network topology and can be deleted during network simplification. After deleting branches that do not affect the network topology, two branches connecting nodes with two outgoing lines can be merged into one branch. These branches are then merged sequentially, converting the branches in the original network into branch groups to obtain a simplified network. If a branch group needs to be disconnected, a single branch in the group is selected for disconnection. The simplified network is topologically equivalent to the original network, but on a smaller scale. Any manipulation of the network topology can be converted into closing or disconnecting branch groups in the simplified network.

[0098] 1.2) Mayeda's method for generating all trees: After selecting a tree in the graph G, this tree is recorded as the reference tree t0, and the basic cut set of the branch e in the reference tree is recorded as S e (t), if a branch e in the reference tree is replaced by another branch to form a new tree, then the replacement branch must belong to S e (t), and the tree is unique. The set of trees generated by this process is recorded as: Among them, T e represents the tree generated by replacing e in the reference tree; e iTo replace the branch of e, e i ∈S e (t0) represents e i Belong to the basic cut set of e in the reference tree t0; for a tree generated by the reference tree Defined as: (where (i1,i2,...,i k ) is a set of (1,2,...,e) arranged in natural order) The spanning tree process is briefly described as follows:

[0099] a. For a graph G, first select an initial tree, called the reference tree, denoted by t0, t0 = {e1, e2, ..., e n}, a total of n branches;

[0100] b. For each branch e in t0 i , use e i The branches in the basic cut set are replaced separately, that is, the selected branches are disconnected, the branches are closed, and the set is generated.

[0101] c.Yes Each spanning tree in k Replace its branches j ,i<j≤n, where

[0102] d. Repeat the above operation. Each time the branch is replaced, the connected branch must belong to the basic cut set of the branch in the reference tree and the cut set of the branch in the current generated tree.

[0103] 1.3) Matrix method for spanning trees: For a graph G, there may be multiple rings. In order to obtain a radial network, a branch must be disconnected in each ring. In this process, if other loops also contain this branch, then the disconnection of this ring will affect other loops. To solve this problem, based on the Mayeda spanning tree, matrix operations are used to implement the exchange of branches to obtain a topological structure that meets the requirements: First, a tree is selected as a reference tree in the simplified network, and a dimension of n (A is established) is established. T ) lines, n(A L ) columns of the matrix M, where A T and A L The number of branches and the number of connected branches are respectively, each row represents a connected branch, each column represents a branch, and the i-th row and j-th column in the matrix L is M ij The values ​​are as follows: Among them, e j represents the jth branch; Indicates that branch l is connected in the reference tree t0 iCorresponding basic circuit; each branch corresponds to a basic circuit. If a branch belongs to the basic circuit corresponding to the branch, the value in the column corresponding to the branch position in the row is 1. If it is 0, it means that the branch is not in the basic circuit corresponding to this branch;

[0104] 1.4) Coding method: For a given network structure, after network simplification and selection of the initial tree, the initial tree (each branch in this tree represents a branch group, which can be merged into a single branch during network simplification) is numbered and these branches are swapped to obtain different topologies. The coding uses small numbers with values ​​of (0, 1), and the number of coding bits is equal to the number of branches.

[0105] 2) Load coding: Since loads are removed during the random optimization process, the load removal amount at the load point is also used as a controllable variable. The loads are divided into three categories according to their importance: Class I loads, Class II loads, and Class III loads. When performing load removal, the removal of loads of different weights will have a much greater impact on the individual target value than the removal of equal amounts of the three classes of loads.

[0106] And these three types of loads are divided into three categories according to their controllability: the first type is controllable load, which can be continuously cut off, and the range of change is from 0 to the maximum load value of the load point. When coding, this article divides this range into 200 points, and the value range is 0-199; the second type of load is uncontrollable load, when the load at this point needs to be cut off, it can only be cut off completely, and the coding value is 0 or 1; the third type of load is incompletely controllable load, which can only be cut off according to several levels of load, and there are five cut-off levels for the load, and the coding value is an integer between 0-4; the above loads are all active power cuts, and the corresponding reactive power cut amount is determined according to the power factor of the load point;

[0107] The particle swarm process first randomly generates an initial population as the reference population according to the above encoding rules, then calculates the fitness value of each individual in the population, records the individual code with the largest fitness value in the reference population, and uses it as the global optimal solution in the first optimization. During the first evolution, the optimal solution of each individual in the reference population is itself, and then the topology and load codes are updated according to the particle swarm process.

[0108] The present invention is a distributed random optimization and load transfer method for a hybrid power transmission and distribution system. In use, the present invention takes into account the uncertainty of renewable energy output and load fluctuations, and utilizes the random programming theory based on the scenario analysis method and the distributed optimization method based on ATC to coordinate the planning of the power transmission and distribution system containing DG. Under the premise of meeting the consistency of the interactive information between the power transmission and distribution systems, the power transmission and distribution system joint optimization planning problem is decomposed into a transmission system planning sub-problem and a series of distribution system planning sub-problems, realizing the parallel operation in the coordinated optimization planning of the power transmission and distribution system. For the distribution system planning sub-problem, a power distribution system power flow equation and constraint relaxation technology are used to construct a The second-order cone convex relaxation programming model for distribution system planning improves iterative convergence; the distributed optimization algorithm with double-layer cyclic iteration is used to solve the constructed model, which helps to improve the economy of the transmission and distribution system planning scheme and the DG absorption capacity; the present invention combines heuristic search with random optimization and adopts a new encoding method in the random optimization process. The generated solution satisfies the radial topology constraint without the need for traditional random algorithms for topology verification, thereby improving the solution efficiency, greatly shortening the calculation time, and better realizing load transfer; the present invention has the advantages of utilizing decomposition optimization modeling, constructing a second-order cone programming model, combining the spanning tree and the loop matrix, and considering controllability when load shedding.

Claims

1. A method for distributed random optimization and load transfer of a hybrid power transmission and distribution system, characterized in that: The method comprises the following steps: Step 1: Construct a distributed optimization planning model for the transmission and distribution system: Use multi-scenario technology to convert the effects of load fluctuation uncertainty on grid operation into a series of certain scenarios for processing. Each scenario is a single load level instead of a traditional typical day, and optimizes the generator output in the transmission system, the DG output in the distribution system, and the shared variables of the boundary nodes between the two. Step 2: Distributed optimization solution: Use the ATC-based distributed optimization algorithm to solve the joint optimization planning model; Step 3: Establish a mathematical model for load transfer: the control variables are the switches on the line and the load shedding amount of the load points in the power failure area; Step 4: Use a heuristic algorithm to quickly search for solutions when load shedding is not required; Step 5: If it is impossible to restore all the loads in the power-off area, the particle swarm algorithm is used to randomly optimize the load removal to ensure that the system meets the safety constraints.

2. A method for distributed random optimization and load transfer of a hybrid power transmission and distribution system as claimed in claim 1, characterized in that: The step 1 of constructing a distributed optimization planning model for the power transmission and distribution system includes the following steps: Step 1.1: Decomposition optimization modeling; Step 1.2: Shared variable modeling: The exchanged active power between the transmission and distribution systems is selected as the shared variable. Assuming that the positive direction of the exchanged power flows from the transmission system to the distribution system, it can be regarded as a virtual power plant relative to the distribution system and a virtual load for the transmission system. When the shared variable is negative, it means that the power is transmitted from the distribution system to the adjacent transmission system. Therefore, the shared variable can be expressed as: t 2q,s =P T,q,s (5), r 2q,s =P D,q,s (6), where P T,q,s , P D,q,s are the exchanged active power between the transmission system and the qth distribution system on the transmission system side and the distribution system side under scenario s, respectively; Step 1.3: Transmission system planning sub-model; Step 1.4: Distribution system planning sub-model.

3. A method for distributed random optimization and load transfer of a hybrid power transmission and distribution system as claimed in claim 2, characterized in that: The transmission system planning sub-model in step 1.3 is specifically as follows: for the transmission system, the decision variable related to investment is the new line, and the decision variable related to operation is the generator output and the exchange power connected to each distribution system. The optimization goal is: In the formula, is a binary variable; is the number of newly built lines on corridor ij; is the unit investment cost of the new line on corridor ij; l ij is the length of corridor ij; C G,i (·) is the power generation cost function of the generator located at node i; P G,i,s is the active power output of the generator at node i in scenario s; ρ T,q The unit price of electricity purchased by the transmission system from the connected qth distribution system; Ω C ,Ω G ,Ω D They are respectively a set of corridors for new lines to be built, a set of generator installation nodes and a set of distribution systems; the constraints of the transmission system planning submodel include constraints on the number of new lines, power balance constraints on transmission system nodes, DC power flow constraints on existing lines, DC power flow constraints on new lines, capacity constraints on existing lines, capacity constraints on new lines, generator output constraints and shared variable constraints on the transmission system side.

4. A method for distributed random optimization and load transfer of a hybrid power transmission and distribution system as claimed in claim 2, characterized in that: The distribution system planning sub-model in step 1.4 is as follows: for the distribution system q, the decision variables related to investment are the transformation of lines, the construction of new lines and the expansion of substations, while the decision variables related to operation are the DG output and the exchange power with the transmission system. The optimization objective is: In the formula, is a binary variable; u ss Both are binary variables; C ss are the unit investment cost of the line transformation on the corridor protection ij and the cost of substation ss expansion; C DG,i (·) is the DG power generation cost function at node i; P DG,i,s is the active power output of the DG at node i in scenario s; ρ D,q The unit price of electricity supplied by the distribution system q to the transmission system; Ω s ,Ω DG They are respectively the existing substation set and the DG installation node set; the constraints of the distribution system planning submodel include the number of transformation lines constraint, distribution system node power balance constraint, line flow equation, line capacity constraint, node voltage constraint, DG output constraint, radial constraint and distribution system side shared variable constraint.

5. A method for distributed random optimization and load transfer of a hybrid power transmission and distribution system as claimed in claim 1, characterized in that: The distributed optimization solution in step 2 includes the following steps: Step 2.1: Second-order cone relaxation of distribution system planning submodel; Step 2.2: Use the ATC-based distributed optimization algorithm to solve the joint optimization planning model.

6. A method for distributed random optimization and load transfer of a hybrid power transmission and distribution system as claimed in claim 5, characterized in that: The second-order cone relaxation distribution system planning submodel in step 2.1 is specifically as follows: the non-convex optimization problem containing cross terms and quadratic nonlinear terms in equations (18)-(21) and (27) is converted into a convex optimization problem of the second-order cone programming model by comprehensively utilizing the large M method and the second-order cone relaxation. and a sufficiently large positive number M, equations (18-(23) and (27) can be rewritten as: Formula (29) can be relaxed to a second-order cone constraint, namely: After the above transformation, the original distribution system planning sub-problem is modeled as a convex second-order cone programming problem.

7. A method for distributed random optimization and load transfer of a hybrid power transmission and distribution system as claimed in claim 5, characterized in that: The step 2.2 uses the ATC-based distributed optimization algorithm to solve the joint optimization planning model, including the following steps: Step 2.21: Set the initial value of the number of inner loop iterations l = 0 and the initial value of the number of outer loop iterations d = 0, and set the consistency constraint penalty function coefficient Weight Shared variables with the transmission system Step 2.22: Let l = l + 1, solve the qth distribution system planning sub-problem, and obtain the shared variables on the distribution system side Step 2.23: Solve the transmission system planning subproblem and obtain the shared variables on the transmission system side Step 2.24: Determine whether the inner loop converges. The inner loop convergence criterion is If it does not converge, return to step 2.22, otherwise continue; Step 2.25: Determine whether the outer loop converges, where the outer loop convergence criterion is |c d |≤ε2(40),|c d -c d-1 |≤ε3(41), if converged, output the optimal solution and exit the loop, otherwise continue; Step 2.26: Let d = d + 1, according to the formula Update consistency constraint penalty function coefficients and weight The value of Step 2.27: Settings And l=0, return to step 2.

22.

8. A method for distributed random optimization and load transfer of a hybrid power transmission and distribution system as claimed in claim 1, characterized in that: The specific steps of establishing the load transfer mathematical model in step 3 are as follows: ① The load shedding amount is the least: ②The sum of the number of switch actions is the least: At the same time, the constraints that the recovery strategy needs to meet are: ① Topological structure constraint: the system maintains a radiation state without closed loops and islands; ② Branch current does not exceed the limit: I i ≤I i,N , where I i is the steady-state current flowing through line i; I i,N is the rated current of line i; ③ The bus voltage does not exceed the limit: U i,min ≤U i ≤U i,max , where U i,min , U i,max are the lower and upper limits of node voltage respectively; ④ Load removal constraint: that is, the load removed from the load point cannot exceed the load of this load point 0≤P icut ≤P i ⑤ Load constraint: The load is controllable, uncontrollable or incompletely controllable; controllable load can change continuously, and any load not exceeding the maximum load of the load point can be cut off as needed; when uncontrollable load needs to be cut off, all loads can only be cut off; incompletely controllable load can only cut off a certain amount of load according to a certain load level.

9. A method for distributed random optimization and load transfer of a hybrid power transmission and distribution system as claimed in claim 1, characterized in that: In step 4, a heuristic algorithm is used to quickly search for a solution when load shedding is not required. Specifically, the objective function in the random method includes: the sum of the shedding amounts of various types of loads, the number of switch actions, and the over-limit penalty term. The objective function expression is as follows: in, In the formula, the first term is the sum of the active power removal of each load in the power failure area; d j is the weight coefficient of the jth load; the second term is the total number of switch actions; the third term is the voltage over-limit penalty term; the fourth term is the power over-limit penalty term; V j is the voltage amplitude at point j; V j,max 、V j,min are the upper and lower limits of the voltage amplitude at point j respectively; S j is the current apparent power of the jth line; S j,max is its power upper limit; p , n , u , s are the weight coefficients of each item in the objective function respectively; the heuristic algorithm system automatically maintains the radial topology during the search process, and the safety constraints that need to be met are: the power flow equation converges, the line is not overloaded, and the voltage does not exceed the limit.

10. A method for distributed random optimization and load transfer of a hybrid power transmission and distribution system as claimed in claim 9, characterized in that: The heuristic algorithm-related parameters in step 4 include: ① The remaining capacity I of the tie switch M :I M =min{(I 1,max -I1),(I 2,max -I2),...,(I n,max -I n )}(47), where I1...I n The I of a certain tie switch is the difference between the maximum current allowed by each branch on the path from the tie switch to the power supply point and the actual current. M Indicates how much load the tie switch can add before an overload occurs; ② The electrical distance Z between the busbar where the voltage crosses the line and a tie switch path : Where V ts Indicates the voltage at the voltage crossing bus; is the sum of all load currents from the bus to a certain tie switch; ③ The load that can be transferred by the section switch in the power failure zone I SS : I SS Represents the sum of all load currents downstream of the section switch.

Images