Medium and low voltage distribution network fault reconstruction method based on particle swarm optimization

Abstract

The invention discloses a medium and low voltage power distribution network fault reconstruction method based on a particle swarm algorithm. The method comprises the following steps: S101, obtaining a target power distribution network management data set; step S102, determining an island division range by taking each DG as a center and taking the capacity of the DG as a constraint condition, dividing load grades into a first-level load, a second-level load and a third-level load, and numbering load nodes in each divided area from high to low; step S103, according to the target power distribution network management data set, using an optimization algorithm based on a particle swarm, taking the minimum network loss as a target, and taking the bearing capacity of the DG as a constraint condition to carry out islanding solving on the DG, and obtaining load points contained in each islanding and a switch required to be disconnected; and S104, fault reconstruction is carried out according to the island division solving result, and a reconstructed network topology structure and a system fault reconstruction result are obtained.

Description

Technical Field

[0001] The present invention relates to the field of power system optimization and intelligent algorithm application, and in particular to a medium and low voltage distribution network fault reconstruction method based on particle swarm algorithm. Background Art

[0002] With the continuous advancement of smart grid technology, fault reconstruction technology for medium and low voltage distribution networks has become a key means to improve grid resilience and power supply reliability. The distribution network fault reconstruction method based on particle swarm algorithm, as an innovative solution, is gradually gaining widespread attention in the industry. This method uses a fault reconstruction method based on particle swarm algorithm to quickly reconfigure the network structure when a distribution network fault occurs to restore power supply to non-faulty areas and optimize the operation status of the power grid. The particle swarm algorithm simulates the foraging behavior of bird flocks and searches for the optimal solution in the solution space in a swarm intelligence manner. It is suitable for solving complex distribution network fault reconstruction problems.

[0003] In the management of medium and low voltage distribution networks, fault reconstruction technology aims to achieve fault isolation and power restoration in non-fault areas by adjusting the switch state, minimizing the scope and duration of power outages. Traditional fault reconstruction methods mainly rely on manual experience or simple rule judgments, and it is difficult to achieve efficient and accurate fault recovery in a complex and changeable power grid environment. The fault reconstruction method based on the particle swarm algorithm can comprehensively consider multiple factors such as network topology, load balance, and the number of switch operations, and automatically search for the optimal or nearly optimal fault recovery plan.

[0004] Although the existing fault reconstruction methods for medium and low voltage distribution networks have improved the speed and efficiency of fault recovery to a certain extent, they still face some challenges. First, when dealing with large-scale and complex distribution networks, traditional methods have large computational complexity and low search efficiency, making it difficult to quickly find the optimal solution. Secondly, in the face of new distribution network elements such as the access of new energy sources and the increase in distributed power sources, traditional methods are often difficult to adapt, which may lead to unreasonable or infeasible fault reconstruction solutions. Finally, most of the existing technologies are based on static or quasi-static power grid models, lack real-time performance, and are difficult to dynamically adjust according to the actual operating status of the power grid, affecting the timeliness and effectiveness of fault reconstruction.

[0005] Therefore, a fault reconstruction method for medium and low voltage distribution networks based on particle swarm optimization is proposed to solve the above problems. This technology improves the search efficiency and accuracy of fault reconstruction by introducing advanced intelligent optimization algorithms, can better adapt to the characteristics of new distribution networks, and achieve fast and intelligent recovery after faults, thereby improving the overall operation level and power supply reliability of medium and low voltage distribution networks.

[0006] However, there are still some problems with the existing distribution network management technology and emergency plan decision-making technology:

[0007] Complex multi-objective optimization problem: The structure of medium and low voltage distribution networks is complex, with many nodes and multiple constraints, such as safe operation, economic benefits, power supply reliability, etc. This makes the distribution network fault reconstruction problem a complex multi-objective optimization problem. Traditional optimization methods are difficult to effectively solve this problem.

[0008] Search efficiency and convergence optimization problem of particle swarm algorithm: Although particle swarm algorithm has strong global search capability, its search efficiency and convergence still need to be further optimized when dealing with large-scale distribution network fault reconstruction problems.

[0009] The problem of ensuring the validity of the state: During the distribution network fault reconstruction process, it is necessary to maintain the radial topology of the distribution network to avoid loops or islands. This requires the use of certain algorithms to ensure the validity of the state, which increases the complexity of technical implementation. Summary of the invention

[0010] In order to overcome the deficiencies in the prior art, the present invention provides a medium and low voltage distribution network fault reconstruction method based on a particle swarm algorithm, which realizes fault reconstruction of a medium and low voltage distribution network through an algorithm technology based on a particle swarm.

[0011] In order to achieve the above-mentioned invention object, the technical solution adopted to solve the technical problem is as follows:

[0012] The present invention discloses a method for fault reconstruction of medium and low voltage distribution networks based on a particle swarm algorithm, comprising the following steps:

[0013] Step S101: Acquire a target distribution network management data set;

[0014] Step S102: Taking each DG as the center and the capacity of the DG as the constraint condition, the range of island division is determined, the load level is divided into primary load, secondary load and tertiary load, and the load nodes in each divided area are numbered from high to low;

[0015] Step S103: according to the target distribution network management data set, using a particle swarm optimization algorithm, with the goal of minimizing network loss and the DG bearing capacity as a constraint, the DG is divided into islands for solution, and the load points contained in each island and the required disconnection switches are obtained;

[0016] Step S104: perform fault reconstruction according to the island division solution result to obtain a reconstructed network topology and a system fault reconstruction result, wherein the system fault reconstruction result includes: disconnected switch information before and after the fault reconstruction, active power loss and system minimum voltage.

[0017] Furthermore, in step S101, the target distribution network management data set includes distribution network topology, DG access information and power grid operation data; wherein the DG access information includes each DG access point of the distribution network, DG capacity, active power and power factor.

[0018] Furthermore, in step S102, the island division criterion is as follows:

[0019] 1) Each node in the network is divided into different levels according to load demand, and important loads are restored to power first;

[0020] 2) The DG in the island is assigned a total load that matches its total power supply capacity. If it is too large, it will harm the system and equipment performance, while if it is too small, the DG will not be fully utilized, resulting in low economic efficiency.

[0021] Further, in step S103, according to the target distribution network data set, an optimization algorithm based on a particle swarm is used to perform island partitioning and solving, and the optimization algorithm based on a particle swarm includes constructing a fitness function;

[0022] The objective function is constructed with the goal of minimizing network power loss, and its expression is:

[0023]

[0024] Where n is the number of distribution network branches; i is the branch number; K i is the state code of the switch. i When it is 1, it is closed, and when it is 0, it is open; R i is the resistance of the i-th branch; P i and Q i is the active and reactive power of the i-th branch; U i is the voltage on the ith node.

[0025] Furthermore, the constraints that the optimization model must follow are as follows:

[0026] (1) Node power flow constraints

[0027]

[0028] Where P DGi With Q DGi are the active power and reactive power injected into node i after DG is connected to the grid; P i With Q i is the active and reactive power provided by the main system at node i; P Li With Q Li is the active and reactive power of the load at node i; m is the number of nodes; G ij is the conductance between node i and node j; B ijis the susceptance between node i and node j; is the phase angle difference between node i and node j;

[0029] (2) Branch power constraints

[0030] S i ≤S imax

[0031] In the formula, S i is the power of the ith node, S imax The power threshold of the i-th node;

[0032] (3) Node voltage constraints

[0033] U imin ≤U i ≤U imax

[0034] Where U imin , U imax The voltage threshold of the i-th node;

[0035] (4) Branch current constraints

[0036] I i ≤I imax

[0037] In the formula, I i is the current of the ith node, I imax Current threshold of the i-th node.

[0038] Furthermore, in step S103, according to the target distribution network data set, an optimization algorithm based on a particle swarm is used to perform island partitioning and solving, and the optimization algorithm based on a particle swarm also includes defining a position and speed update method of particles;

[0039] In the particle swarm algorithm, the population X consisting of n particles is (X 1 ,X 2 ,X 3 ,…,X n ); the i-th particle represents the D-dimensional vector X i =(X i1 ,X i2 ,X i3 ,…,X iD ) T ; The velocity of the i-th particle is V i =(V i1 ,V i2 ,V i3 ,…,V iD ) T ; The individual extreme value is P i =(P i1,P i2 ,P i3 ,…,P iD ) T , the global extreme value is P g =(P g1 ,P g2 ,P g3 ,…,P gD ) T , evaluate the objective function value of each particle in each iteration and determine the optimal position p of each particle at time t best The optimal position g found by the group best , the speed and position update formula of each particle is:

[0040] V ij (t+1)=V ij (t)+c 1 r 1 [p ij -X ij (t)]+c 2 r 2 [p gj -X ij (t)]

[0041] X ij (t+1)=X ij (t)+V ij (t+1), j=1,...,d

[0042] In the formula, c 1 and c 2 is a positive learning factor; r 1 and r 2 is a random number uniformly distributed between 0 and 1.

[0043] Furthermore, the inertia weight coefficient ω is introduced to enable it to have efficient local search capabilities. At this time, the speed and position expressions of each particle are:

[0044]

[0045] X ij (t+1)=X ij (t)+V ij (t+1), j=1,...,d

[0046] The particle's position is defined from a probability perspective by the particle's velocity, so that the particle's position x in each dimension is in The positioning is 0 or 1, and the optimal solution is searched by solving the probability of 0 and 1 in the system. The specific calculation formula is:

[0047]

[0048] Where rand() is a random number uniformly distributed in the interval [0,1] generated after each iteration, representing the latest position of the i-th particle when calculated to the m+1th time;

[0049] Define the Sigmoid function as:

[0050]

[0051] The Sigmoid function graph is S-shaped, and the particle speed is limited to [-4, 4] to prevent the Sigmoid function from reaching saturation due to excessive speed. The corresponding Sigmoid function takes values ​​within [-0.98, 0.98].

[0052] Furthermore, in order to ensure the balance between global and local optimization, a PSO model with a shrinkage factor is constructed, and the velocity and position expressions of each particle can be converted into:

[0053] V ij (t+1)=K{V ij (t)+c 1 r 1 [p ij -X ij (t)]+c 2 r 2 [p gj -X ij (t)]}

[0054] Where K is the shrinkage factor, and its conditional relationship is C=c 1 +c 2 .

[0055] Furthermore, the global optimization and local optimization capabilities of the nonlinear dynamic inertia weight coefficient balancing algorithm are adopted, and the expression is:

[0056]

[0057] In the formula, and for The maximum and minimum values ​​of; f is the objective function value; f avg and f min The average target value and the minimum target value for all particles; inertia weight It decreases linearly with the number of iterations, with a maximum value of 0.9 and a minimum value of 0.4.

[0058] Furthermore, the hybridization concept and natural selection mechanism in the genetic algorithm are used to ensure that each generation of particles participating in the iteration has good performance, so that the algorithm can quickly converge to the global optimal solution. The position of the offspring particles is obtained through the hybridization operation of the parent particles:

[0059] n x =σm x (1)+(1-σ)m x (2)

[0060] In the formula, m x Represents the position of the parent particle; n x represents the position of the offspring particle; σ is a random number between 0 and 1, and the offspring velocity formula is:

[0061]

[0062] In the formula, m v Represents the speed of the parent particle; n v Indicates the velocity of the descendant particles.

[0063] Furthermore, through the hybridization operation, the offspring particles can combine the advantages of the parent particles to further increase the diversity of the search space. The offspring particles will replace the parent particles and become part of the next generation population, continuing to participate in the next iteration. The algorithm after introducing the selection and hybridization mechanism can improve the global search capability and gradually approach the global optimal solution.

[0064] Chaotic mapping is used to detect all solutions generated in the iterative process. After the termination condition is met, this solution is used as the starting point of the next iteration to improve the global search ability of the algorithm. The mapping sequence is controlled by adjusting the control parameters to accelerate the convergence process of the particle swarm algorithm, reduce the number of iterations, and improve the efficiency of the algorithm. The mapping formula is:

[0065] z k+1 =μz k (1-z k ),μ∈(0,4)

[0066] In the formula, μ is the control parameter, z k is the current status.

[0067] Furthermore, the solution steps based on the particle swarm algorithm are as follows:

[0068] 1) Define the fitness function, which is a measure of power loss;

[0069] 2) Define the update method of particle position and velocity, as well as the initialization and termination conditions of the particle swarm;

[0070] 3) Initialize the particle swarm, including the population size, maximum number of iterations, learning factors c1 and c2, and the maximum and minimum values ​​of the inertia weight w;

[0071] 4) Randomly initialize particle speed and position;

[0072] 5) According to the fitness function, calculate the fitness value of each particle, and store the particle position and fitness value in the individual extreme value p besti In the besti The individual position and fitness value of the best fitness value in the global extreme value g best middle;

[0073] 6) Use the formula to update the particle position and velocity;

[0074] 7) Compare the fitness value of each particle with the optimal position. If they are similar, take the current value as the optimal position and compare all p besti With g best , update g best ;

[0075] 8) Sort the population according to the fitness value, replace the worst half of the particles with the best half of the particles, and retain the historical optimal value of the original particles;

[0076] 9) Set the hybridization rate, extract a specified number of particles and put them into the hybridization pool, hybridize to generate a corresponding number of offspring particles, and maintain p besti and g best constant;

[0077] 10) besti and g best Map it to the definition domain of the chaotic equation, obtain the sequence through iteration and map it back to the original solution space, and calculate the fitness value of the optimal solution;

[0078] 11) When the algorithm reaches the termination condition, stop searching and output the optimal solution, otherwise return to step 3) to continue searching;

[0079] After the iteration, the global extreme value g is output best .

[0080] Due to the adoption of the above technical solution, the present invention has the following advantages and positive effects compared with the prior art:

[0081] 1. Improved efficiency and accuracy of distribution network fault reconstruction: The present invention uses a particle swarm algorithm to simulate the foraging behavior of a flock of birds, search in the solution space, and continuously update its position to find the optimal solution. Compared with traditional optimization methods, the particle swarm algorithm has stronger global search capabilities and can converge quickly in complex multi-objective optimization problems, thereby improving the efficiency and accuracy of distribution network fault reconstruction.

[0082] 2. Enhanced adaptability and flexibility of distribution network fault reconstruction: The present invention establishes a mathematical model, defines the fitness function, the update method of position and speed, and the initialization and termination conditions of the particle swarm, so that the particle swarm algorithm can adapt to distribution networks of different scales and structures and has stronger adaptability and flexibility.

[0083] 3. Guaranteeing the validity of the state of distribution network fault reconstruction: In the process of distribution network fault reconstruction, the present invention uses topological constraints to ensure the validity of the state, maintain the radial topological structure of the distribution network, and avoid loops or islands.

[0084] 4. Improved economic benefits of distribution network fault reconstruction: The goal of the present invention is to reduce power loss, improve power supply reliability and economic benefits by optimizing the topology and load information of the distribution network. Compared with traditional optimization methods, the present invention can more effectively balance power supply reliability and economic benefits, thereby improving the economic benefits of distribution network fault reconstruction. BRIEF DESCRIPTION OF THE DRAWINGS

[0085] In order to more clearly illustrate the technical solutions of the embodiments of the present invention, the following briefly introduces the drawings required for describing the embodiments. Obviously, the drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative work. In the drawings:

[0086] Figure 1 A schematic diagram of a process flow of a medium and low voltage distribution network fault reconstruction method based on a particle swarm algorithm proposed by the present invention;

[0087] Figure 2 A flow chart of a hybrid particle swarm optimization algorithm provided in an embodiment of the present application; DETAILED DESCRIPTION

[0088] The technical solution of the present invention will be described clearly and completely below in conjunction with the accompanying drawings. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0089] Embodiment 1

[0090] like Figure 1 As shown, this embodiment discloses a method for fault reconstruction of medium and low voltage distribution networks based on particle swarm algorithm, which includes but is not limited to the following steps:

[0091] Step S101: Acquire a target distribution network management data set;

[0092] Specifically, in step S101, the target distribution network management data set includes distribution network topology, DG access information and power grid operation data; wherein the DG access information includes each DG access point of the distribution network, DG capacity, active power and power factor.

[0093] Step S102: Taking each DG as the center and the capacity of the DG as the constraint condition (division radius) to determine the scope of island division, the load level is divided into primary load, secondary load and tertiary load, and then the load nodes in each divided area are numbered from high to low;

[0094] Specifically, in step S102, the island division criterion is as follows:

[0095] 1) Each node in the network is divided into different levels according to load demand, and important loads are restored to power first;

[0096] 2) The DG in the island is assigned a total load that matches its total power supply capacity. If it is too large, it will harm the system and equipment performance, while if it is too small, the DG will not be fully utilized, resulting in low economic efficiency.

[0097] Step S103: Based on the target distribution network management data set, a particle swarm optimization algorithm is used to divide the DG into islands and solve the problem, with the minimum network loss as the goal and the DG bearing capacity as the constraint condition, to obtain the load points contained in each island and the required disconnection switches. Figure 2 , Figure 2 The hybrid particle swarm optimization algorithm provided in the embodiment of the present application is a flow chart, and the method includes but is not limited to the following steps:

[0098] 1) Constructing fitness function

[0099] The objective function is constructed with the goal of minimizing network power loss, and its expression is:

[0100]

[0101] Where n is the number of distribution network branches; i is the branch number; K i is the state code of the switch. i When it is 1, it is closed, and when it is 0, it is open; R i is the resistance of the i-th branch; P i and Q i is the active and reactive power of the i-th branch; U i is the voltage on the i-th node;

[0102] The constraints that the optimization model must follow are as follows:

[0103] (1) Node power flow constraints

[0104]

[0105] Where P DGi With Q DGi are the active power and reactive power injected into node i after DG is connected to the grid; P i With Q i is the active and reactive power provided by the main system at node i; P Li With Q Li is the active and reactive power of the load at node i; m is the number of nodes; G ij is the conductance between node i and node j; B ij is the susceptance between node i and node j; is the phase angle difference between node i and node j;

[0106] (2) Branch power constraints

[0107] S i ≤S imax

[0108] In the formula, S i is the power of the ith node, S imax The power threshold of the i-th node;

[0109] (3) Node voltage constraints

[0110] U imin ≤U i ≤U imax

[0111] Where U imin , U imax The voltage threshold of the i-th node;

[0112] (4) Branch current constraints

[0113] I i ≤I imax

[0114] In the formula, I i is the current of the ith node, I imax The current threshold of the i-th node;

[0115] 2) Define the particle position and velocity update method

[0116] In the particle swarm algorithm, the population X consisting of n particles is (X 1 ,X 2 ,X 3 ,…,X n ); the i-th particle represents the D-dimensional vector X i =(X i1 ,X i2 ,X i3 ,…,XiD ) T ; The velocity of the i-th particle is V i =(V i1 ,V i2 ,V i3 ,…,V iD ) T ; The individual extreme value is P i =(P i1 ,P i2 ,P i3 ,…,P iD ) T , the global extreme value is P g =(P g1 ,P g2 ,P g3 ,…,P gD ) T , evaluate the objective function value of each particle in each iteration and determine the optimal position p of each particle at time t best The optimal position g found by the group best , the speed and position update formula of each particle is:

[0117] V ij (t+1)=V ij (t)+c 1 r 1 [p ij -X ij (t)]+c 2 r 2 [p gj -X ij (t)]

[0118] X ij (t+1)=X ij (t)+V ij (t+1), j=1,...,d

[0119] In the formula, c 1 and c 2 is a positive learning factor; r 1 and r 2 is a random number uniformly distributed between 0 and 1;

[0120] The inertia weight coefficient ω is introduced to enable it to have efficient local search capabilities. At this time, the speed and position expressions of each particle are:

[0121]

[0122] X ij (t+1)=X ij (t)+V ij (t+1), j=1,...,d

[0123] The particle's position is defined from a probability perspective by the particle's velocity, so that the particle's position x in each dimension is in The positioning is 0 or 1, and the optimal solution is searched by solving the probability of 0 and 1 in the system. The specific calculation formula is:

[0124]

[0125] Where rand() is a random number uniformly distributed in the interval [0,1] generated after each iteration, representing the latest position of the i-th particle when calculated to the m+1th time;

[0126] Define the Sigmoid function as:

[0127]

[0128] The Sigmoid function graph is S-shaped, and the particle speed is limited to the range of [-4,4] to prevent the Sigmoid function from reaching saturation due to excessive speed. The corresponding Sigmoid function takes values ​​within [-0.98,0.98];

[0129] In order to ensure the balance between global and local optimization, a PSO model with a contraction factor is constructed. The velocity and position expressions of each particle can be converted into:

[0130] V ij (t+1)=K{V ij (t)+c 1 r 1 [p ij -X ij (t)]+c 2 r 2 [p gj -X ij (t)]}

[0131] Where K is the shrinkage factor, and its conditional relationship is C=c 1 +c 2 ;

[0132] The global optimization and local optimization capabilities of the nonlinear dynamic inertia weight coefficient balancing algorithm are adopted, and the expression is:

[0133]

[0134] In the formula, and for The maximum and minimum values ​​of; f is the objective function value; f avg and f minis the average target value and minimum target value of all particles, inertia weight It decreases linearly with the number of iterations, with a maximum value of 0.9 and a minimum value of 0.4;

[0135] By drawing on the hybridization concept and natural selection mechanism in genetic algorithms, we ensure that each generation of particles participating in the iteration has good performance, so that the algorithm can quickly converge to the global optimal solution. The position of the offspring particles is obtained through the hybridization operation of the parent particles:

[0136] n x =σm x (1)+(1-σ)m x (2)

[0137] In the formula, m x Represents the position of the parent particle; n x represents the position of the offspring particle; σ is a random number between 0 and 1, and the offspring velocity formula is:

[0138]

[0139] In the formula, m v Represents the speed of the parent particle; n v Represents the velocity of the descendant particles;

[0140] Through the hybridization operation, the offspring particles can combine the advantages of the parent particles to further increase the diversity of the search space. The offspring particles will replace the parent particles and become part of the next generation population, continuing to participate in the next iteration. The algorithm after introducing the selection and hybridization mechanism can improve the global search capability and gradually approach the global optimal solution.

[0141] Chaotic mapping is used to detect all solutions generated in the iterative process. After the termination condition is met, this solution is used as the starting point of the next iteration to improve the global search ability of the algorithm. The mapping sequence is controlled by adjusting the control parameters to accelerate the convergence process of the particle swarm algorithm, reduce the number of iterations, and improve the efficiency of the algorithm. The mapping formula is:

[0142] z k+1 =μz k (1-z k ),μ∈(0,4)

[0143] In the formula, μ is the control parameter, z k is the current status.

[0144] Furthermore, the solution steps of the hybrid particle swarm optimization algorithm are as follows:

[0145] 1) Define the fitness function, which is a measure of power loss;

[0146] 2) Define the update method of particle position and velocity, as well as the initialization and termination conditions of the particle swarm;

[0147] 3) Initialize the particle swarm, including the population size, maximum number of iterations, learning factors c1 and c2, and the maximum and minimum values ​​of the inertia weight w;

[0148] 4) Randomly initialize particle speed and position;

[0149] 5) According to the fitness function, calculate the fitness value of each particle, and store the particle position and fitness value in the individual extreme value p besti In the besti The individual position and fitness value of the best fitness value in the global extreme value g best middle;

[0150] 6) Use the formula to update the particle position and velocity;

[0151] 7) Compare the fitness value of each particle with the optimal position. If they are similar, take the current value as the optimal position and compare all p besti With g best , update g best ;

[0152] 8) Sort the population according to the fitness value, replace the worst half of the particles with the best half of the particles, and retain the historical optimal value of the original particles;

[0153] 9) Set the hybridization rate, extract a specified number of particles and put them into the hybridization pool, hybridize to generate a corresponding number of offspring particles, and maintain p besti and g best constant;

[0154] 10) besti and g best Map it to the definition domain of the chaotic equation, obtain the sequence through iteration and map it back to the original solution space, and calculate the fitness value of the optimal solution;

[0155] 11) When the algorithm reaches the termination condition, stop searching and output the optimal solution, otherwise return to step 3) to continue searching;

[0156] After the iteration, the global extreme value g is output best .

[0157] Step S104: perform fault reconstruction according to the island division solution result to obtain the network topology structure and system fault reconstruction result after fault reconstruction, wherein the system fault reconstruction result includes: disconnected switch information before and after fault reconstruction, active power loss and system minimum voltage.

[0158] The above is only a preferred specific embodiment of the present invention, but the protection scope of the present invention is not limited thereto. Any changes or substitutions that can be easily thought of by a person skilled in the art within the technical scope disclosed by the present invention should be included in the protection scope of the present invention. Therefore, the protection scope of the present invention should be based on the protection scope of the claims.

Claims

1. A fault reconstruction method for medium and low voltage distribution networks based on particle swarm algorithm, comprising the following steps: Step S101: Acquire a target distribution network management data set; Step S102: Taking each DG as the center and the capacity of the DG as the constraint condition, the range of island division is determined, the load level is divided into primary load, secondary load and tertiary load, and then the load nodes in each divided area are numbered from high to low; Step S103: according to the target distribution network management data set, using a particle swarm optimization algorithm, with the goal of minimizing network loss and the DG bearing capacity as a constraint, the DG is divided into islands for solution, and the load points contained in each island and the required disconnection switches are obtained; Step S104: perform fault reconstruction according to the island division solution result to obtain a reconstructed network topology and a system fault reconstruction result, wherein the system fault reconstruction result includes: disconnected switch information before and after the fault reconstruction, active power loss and system minimum voltage.

2. According to the particle swarm algorithm-based medium and low voltage distribution network fault reconstruction method of claim 1, it is characterized in that: In step S101, the target distribution network management data set includes distribution network topology, DG access information and power grid operation data; wherein the DG access information includes each DG access point of the distribution network, DG capacity, active power and power factor.

3. According to the particle swarm algorithm-based medium and low voltage distribution network fault reconstruction method of claim 1, it is characterized in that: In step S102, the island division criteria are as follows: 1) Each node in the network is divided into different levels according to load demand, and important loads are restored to power first; 2) The DG in the island is assigned a total load that matches its total power supply capacity. If it is too large, it will harm the system and equipment performance, while if it is too small, the DG will not be fully utilized, resulting in low economic efficiency.

4. According to the particle swarm algorithm-based medium and low voltage distribution network fault reconstruction method of claim 1, it is characterized in that: In step S103, according to the target distribution network data set, an optimization algorithm based on a particle swarm is used to perform island partitioning and solving, and the optimization algorithm based on a particle swarm includes constructing a fitness function; The objective function is constructed with the goal of minimizing network power loss, and its expression is: Where n is the number of distribution network branches; i is the branch number; K i is the state code of the switch. i When it is 1, it is closed, and when it is 0, it is open; R i is the resistance of the i-th branch; P i and Q i is the active and reactive power of the i-th branch; U i is the voltage on the ith node.

5. A method for fault reconstruction of medium and low voltage distribution networks based on particle swarm algorithm according to claim 4, characterized in that: The constraints that the optimization model must follow are as follows: (1) Node power flow constraints Where P DGi With Q DGi are the active power and reactive power injected into node i after DG is connected to the grid; P i With Q i is the active and reactive power provided by the main system at node i; P Li With Q Li is the active and reactive power of the load at node i; m is the number of nodes; G ij is the conductance between node i and node j; B ij is the susceptance between node i and node j; is the phase angle difference between node i and node j; (2) Branch power constraints S i ≤S imax In the formula, S i is the power of the ith node, S imax The power threshold of the i-th node; (3) Node voltage constraints IN imin ≤U i ≤U imax Where U imin , U imax The voltage threshold of the i-th node; (4) Branch current constraints I i ≤I imax In the formula, I i is the current of the ith node, I imax Current threshold of the i-th node.

6. A method for fault reconstruction of medium and low voltage distribution networks based on particle swarm algorithm according to claim 4, characterized in that: In step S103, according to the target distribution network data set, an optimization algorithm based on a particle swarm is used to solve the island partitioning problem, and the optimization algorithm based on a particle swarm also includes defining a position and speed update method of particles; In the particle swarm algorithm, the population X consisting of n particles is (X1, X2, X3, …, X n ); the i-th particle represents the D-dimensional vector X i =(X i1 ,X i2 ,X i3 ,…,X iD ) T ; The velocity of the i-th particle is V i =(V i1 ,V i2 ,V i3 ,…,V iD ) T ; The individual extreme value is P i =(P i1 ,P i2 ,P i3 ,…,P iD ) T , the global extreme value is P g =(P g1 ,P g2 ,P g3 ,…,P gD ) T , evaluate the objective function value of each particle in each iteration and determine the optimal position p of each particle at time t best The optimal position g found by the group best , the speed and position update formula of each particle is: V ij (t+1)=V ij (t)+c1r1[p ij -X ij (t)]+c2r2[p gj -X ij (t)] X ij (t+1)=X ij (t)+V ij (t+1),j=1,...,d Where c1 and c2 are positive learning factors; r1 and r2 are random numbers uniformly distributed between 0 and 1.

7. A method for fault reconstruction of medium and low voltage distribution networks based on particle swarm algorithm according to claim 6, characterized in that: The inertia weight coefficient ω is introduced to enable it to have efficient local search capabilities. At this time, the speed and position expressions of each particle are: X ij (t+1)=X ij (t)+V ij (t+1),j=1,...,d The particle's position is defined from a probability perspective by the particle's velocity, so that the particle's position x in each dimension is in The positioning is 0 or 1, and the optimal solution is searched by solving the probability of 0 and 1 in the system. The specific calculation formula is: Where rand() is a random number uniformly distributed in the interval [0,1] generated after each iteration, representing the latest position of the i-th particle when calculated to the m+1th time; Define the Sigmoid function as: The Sigmoid function graph is S-shaped, and the particle speed is limited to [-4, 4] to prevent the Sigmoid function from reaching saturation due to excessive speed. The corresponding Sigmoid function takes values ​​within [-0.98, 0.98].

8. A method for fault reconstruction of medium and low voltage distribution networks based on particle swarm algorithm according to claim 7, characterized in that: In order to ensure the balance between global and local optimization, a PSO model with a contraction factor is constructed. The velocity and position expressions of each particle can be converted into: V ij (t+1)=K{V ij (t)+c1r1[p ij -X ij (t)]+c2r2[p gj -X ij (t)]} Where K is the shrinkage factor, and its conditional relationship is C=c1+c2.

9. A method for fault reconstruction of medium and low voltage distribution networks based on particle swarm algorithm according to claim 8, characterized in that: The global optimization and local optimization capabilities of the nonlinear dynamic inertia weight coefficient balancing algorithm are adopted, and the expression is: In the formula, and for The maximum and minimum values ​​of ; f is the objective function value; f avg and f min The average target value and the minimum target value for all particles; inertia weight It decreases linearly with the number of iterations, with a maximum value of 0.9 and a minimum value of 0.

4.

10. A method for fault reconstruction of medium and low voltage distribution networks based on particle swarm algorithm according to claim 9, characterized in that: By drawing on the hybridization concept and natural selection mechanism in genetic algorithms, we ensure that each generation of particles participating in the iteration has good performance, so that the algorithm can quickly converge to the global optimal solution. The position of the offspring particles is obtained through the hybridization operation of the parent particles: n x =σm x (1)+(1-σ)m x (2) In the formula, m x Represents the position of the parent particle; n x represents the position of the offspring particle; σ is a random number between 0 and 1, and the offspring velocity formula is: In the formula, m v Represents the speed of the parent particle; n v Indicates the velocity of the descendant particles.

11. A method for fault reconstruction of medium and low voltage distribution networks based on particle swarm algorithm according to claim 10, characterized in that: Through the hybridization operation, the offspring particles can combine the advantages of the parent particles to further increase the diversity of the search space. The offspring particles will replace the parent particles and become part of the next generation population, continuing to participate in the next iteration. The algorithm after introducing the selection and hybridization mechanism can improve the global search capability and gradually approach the global optimal solution. Chaotic mapping is used to detect all solutions generated in the iterative process. After the termination condition is met, this solution is used as the starting point of the next iteration to improve the global search ability of the algorithm. The mapping sequence is controlled by adjusting the control parameters to accelerate the convergence process of the particle swarm algorithm, reduce the number of iterations, and improve the efficiency of the algorithm. The mapping formula is: z k+1 =μz k (1-z k ),μ∈(0,4) In the formula, μ is the control parameter, z k is the current status.

12. A method for fault reconstruction of medium and low voltage distribution networks based on particle swarm algorithm according to claim 4, characterized in that: The solution steps based on particle swarm algorithm are as follows: 1) Define the fitness function, which is a measure of power loss; 2) Define the update method of particle position and velocity, as well as the initialization and termination conditions of the particle swarm; 3) Initialize the particle swarm, including the population size, maximum number of iterations, learning factors c1 and c2, and the maximum and minimum values ​​of the inertia weight w; 4) Randomly initialize particle speed and position; 5) According to the fitness function, calculate the fitness value of each particle, and store the particle position and fitness value in the individual extreme value p besti In the besti The individual position and fitness value of the best fitness value in the global extreme value g best middle; 6) Use the formula to update the particle position and velocity; 7) Compare the fitness value of each particle with the optimal position. If they are similar, take the current value as the optimal position and compare all p besti With g best , update g best ; 8) Sort the population according to the fitness value, replace the worst half of the particles with the best half of the particles, and retain the historical optimal value of the original particles; 9) Set the hybridization rate, extract a specified number of particles and put them into the hybridization pool, hybridize to generate a corresponding number of offspring particles, and maintain p besti and g best constant; 10) besti and g best Map it to the definition domain of the chaotic equation, obtain the sequence through iteration and map it back to the original solution space, and calculate the fitness value of the optimal solution; 11) When the algorithm reaches the termination condition, stop searching and output the optimal solution, otherwise return to step 3) to continue searching; After the iteration, the global extreme value g is output best .

Images