Distributed feeder fault positioning method based on BN-BPSO algorithm

Abstract

The invention discloses a distributed feeder fault positioning method based on a BN-BPSO algorithm, and belongs to the technical field of feeder fault positioning, and the method comprises the steps: 1, setting and initializing parameters; 2, constructing a switching function and an evaluation function, and inputting fault information; 3, calculating an evaluation function value, and setting an individual optimal position and extreme value and a global optimal position and extreme value according to the initial evaluation function value; 4, sorting the whole particle swarm according to evaluation function values, replacing the worst half of particles with the best half of particles, and keeping the individual and global extreme values unchanged; 5, the replaced particles are extracted according to the hybridization rate and put into a hybridization pool, the particles in the pool are randomly hybridized in pairs to generate filial generation particles with the same number, and the individual and global extreme values are not changed; and 6, judging whether the offspring particles meet a convergence condition, if so, outputting an optimal particle and displaying a fault area, and otherwise, turning to the step 3. According to the method, distributed feeder line fault positioning can be rapidly and accurately realized, and the stability is good.

Description

Technical Field

[0001] The present invention relates to the technical field of feeder fault location, and in particular to a distributed feeder fault location method based on a BN-BPSO algorithm. Background Art

[0002] In the power system, the feeder is a distribution line distributed from the distribution network busbar directly to the load line of the load. The purpose of distribution network feeder automation (FA) is to improve the efficiency of distribution operation and maintenance and improve the safety of power supply. Based on the primary equipment of the distribution network, it uses communication and computer technologies to realize the functions of monitoring the distribution network feeder, quickly discovering faults, isolating fault areas, and restoring power supply to non-fault areas. However, the traditional feeder fault location method can only realize the fault location of the line powered by a single power supply. When used for distributed feeder fault location, the accuracy and stability are poor.

[0003] The patent document with publication number CN 112582987 A discloses a method for intelligent distributed feeder fault judgment and positioning of a distribution network, including the following steps: setting a distribution automation feeder terminal, setting an intelligent substation configuration file, forming a description information relationship table, judging faults in the node area, power direction judgment, fault location and isolation. This invention is aimed at a ring network structure distribution network with multiple power supply paths or large-scale distributed power sources, and adopts overcurrent protection, power direction judgment, intelligent distributed fault location and isolation to quickly determine the fault area, accurately locate and isolate the fault, and then perform fault isolation and recovery of the complex distribution network with a multi-power ring network structure to improve power supply reliability. However, the invention needs to be improved in terms of the speed and stability of distributed feeder fault judgment and positioning. Summary of the invention

[0004] In view of this, the present invention aims at the deficiencies in the prior art and provides a distributed feeder fault location method based on the BN-BPSO algorithm, which can quickly and accurately realize distributed feeder fault location, has good stability, has high fault tolerance, and can be effectively applied to feeder networks with complex branches.

[0005] In order to solve the above technical problems, the technical solution adopted by the present invention is: a distributed feeder fault location method based on the BN-BPSO algorithm, comprising the following steps:

[0006] Step 1: Parameter setting and initialization;

[0007] Step 2: Construct the switch function and evaluation function of the distributed feeder according to the network structure, and input the fault information uploaded by the feeder terminal equipment;

[0008] Step 3: Calculate the evaluation function value, and set the individual optimal position and extreme value, the global optimal position and extreme value according to the initial evaluation function value;

[0009] Step 4: Sort the entire particle swarm by the evaluation function value, replace the worst half of the particles with the best half of the particles, and keep the individual and global extreme values ​​unchanged;

[0010] Step 5: Extract the replaced particles according to the hybridization rate and put them into the hybridization pool. The particles in the pool are randomly hybridized to generate the same number of offspring particles. The individual and global extreme values ​​remain unchanged.

[0011] Step 6: Determine whether the descendant particles meet the convergence conditions. If so, output the optimal particle and display the fault area. Otherwise, go to step 3.

[0012] Furthermore, in step 1, parameter settings include: population size, maximum number of iterations, learning factor and weight coefficient, particle dimension, hybridization rate; and initialization of the position and velocity of the particle swarm.

[0013] Furthermore, in step 2, the switching function of the distributed feeder is:

[0014]

[0015]

[0016]

[0017] In the formula, K u , K d They represent the switching coefficients of the upstream and downstream power supplies of switch j, respectively. When the power supply is put into operation, the value is 1, and when it is turned off, it is 0; M2 and N2 represent the total number of feeder sections in the upper and lower sections of switch j, respectively; M1 and N1 represent the total number of power supplies in the upper and lower halves of switch j, respectively; s j,u 、s j,d are the status values ​​of all feeder sections from switch j to the upper and lower sections, respectively. The value is 1 when a fault occurs and the value is 0 when the section is normal. j,Gu 、s j,Gd The jth switch to the upstream system power supply G u With downstream distributed power supply G d The state value of the feeder section passed on the path; ju (s), I jd (s) represents the switch function of the upstream and downstream lines of the jth switch; I j (s) represents the switch function of the jth switch; П represents the logical OR operation;

[0018] The evaluation function of distributed feeder is:

[0019]

[0020] In the formula, I h Indicates the fault overcurrent information of the hth section; I h * (S) is the switching function of the hth section; H is the total number of sections in the entire distributed feeder network; x(h) is the fault state of the hth section, which is 1 when a fault occurs and 0 otherwise; μ is the weight coefficient, which ranges from 0 to 1;

[0021] The fault information is the equivalent fault vector converted from the overcurrent information collected by the feeder terminal equipment:

[0022] The direction of the main power supply current is defined as the positive direction. When the feeder terminal device detects positive fault current information, the equivalent fault vector is 1; when negative fault current information is detected, the equivalent fault vector is -1; if no fault current flows through the node, the equivalent fault vector is 0.

[0023] Furthermore, the particle dimension is the number of distributed feeder sections.

[0024] Compared with the prior art, the present invention has the following beneficial effects:

[0025] The present invention adopts a distributed feeder fault location method based on the BN-BPSO (hybrid particle swarm of hybridization and natural selection) algorithm, uses an improved feeder encoding method to realize the change from fault information to fault vector, constructs a feeder network switch function and an evaluation function connected to a distributed power source, and inputs them into the BN-BPSO algorithm model. The population dimension is initialized to the number of feeder sections, and then the evaluation function value is calculated. Finally, the optimal position of the particle is output to realize the fault section location of the distribution network.

[0026] The BN-BPSO algorithm of the present invention has a small number of iterations and fast convergence, can quickly obtain the global optimal value in both cases, has good stability, has high fault tolerance, and can be effectively applied to feeder networks with complex branches. BRIEF DESCRIPTION OF THE DRAWINGS

[0027] Figure 1 is a flow chart of the method of the present invention;

[0028] Figure 2 It is a schematic diagram of a 33-node power distribution network in an embodiment of the present invention. DETAILED DESCRIPTION

[0029] In order to make the purpose, technical solution and advantages of the embodiment of the present invention clearer, the technical solution of the embodiment of the present invention will be clearly and completely described below in conjunction with the drawings of the embodiment of the present invention. Obviously, the described embodiment is a part of the embodiment of the present invention, not all of the embodiments. Based on the described embodiment of the present invention, all other embodiments obtained by ordinary technicians in this field belong to the scope of protection of the present invention.

[0030] like Figure 1 As shown, a distributed feeder fault location method based on the BN-BPSO algorithm includes the following steps:

[0031] Step 1: Parameter setting and initialization;

[0032] Step 2: Construct the switch function and evaluation function of the distributed feeder according to the network structure, and input the fault information uploaded by the feeder terminal equipment;

[0033] Step 3: Calculate the evaluation function value, and set the individual optimal position and extreme value, the global optimal position and extreme value according to the initial evaluation function value;

[0034] Step 4: Sort the entire particle swarm by the evaluation function value, replace the worst half of the particles with the best half of the particles, and keep the individual and global extreme values ​​unchanged;

[0035] Step 5: Extract the replaced particles according to the hybridization rate and put them into the hybridization pool. The particles in the pool are randomly hybridized to generate the same number of offspring particles. The individual and global extreme values ​​remain unchanged.

[0036] Step 6: Determine whether the descendant particles meet the convergence conditions. If so, output the optimal particle and display the fault area. Otherwise, go to step 3.

[0037] Specifically, in step 1, parameter settings include: population size, maximum number of iterations, learning factor and weight coefficient, particle dimension, hybridization rate; initialization of the position and velocity of the particle swarm; the population size is the total number of distributed feeder segments, and the particle dimension is the number of distributed feeder segments.

[0038] In step 2, the switching function of the distributed feeder is:

[0039]

[0040]

[0041]

[0042] In the formula, K u , K dThey represent the switching coefficients of the upstream and downstream power supplies of switch j, respectively. When the power supply is put into operation, the value is 1, and when it is turned off, it is 0; M2 and N2 represent the total number of feeder sections in the upper and lower sections of switch j, respectively; M1 and N1 represent the total number of power supplies in the upper and lower halves of switch j, respectively; s j,u 、s j,d are the status values ​​of all feeder sections from switch j to the upper and lower sections, respectively. The value is 1 when a fault occurs and the value is 0 when the section is normal. j,Gu 、s j,Gd The jth switch to the upstream system power supply G u With downstream distributed power supply G d The state value of the feeder section passed on the path; ju (s), I jd (s) represents the switch function of the upstream and downstream lines of the jth switch; I j (s) represents the switch function of the jth switch; ∏ represents the logical OR operation;

[0043] The evaluation function of the distributed feeder is to select the fitness function, and the formula is:

[0044]

[0045] In the formula, I h Indicates the fault overcurrent information of the hth section; I h * (S) is the switching function of the hth section; H is the total number of sections in the entire distributed feeder network; x(h) is the fault state of the hth section, which is 1 when a fault occurs and 0 otherwise; μ is the weight coefficient, which ranges from 0 to 1;

[0046] In step 2, the fault information is an equivalent fault vector converted from the overcurrent information collected by the feeder terminal device, specifically: the main power current direction is defined as the positive direction. When the feeder terminal device detects positive fault current information, the equivalent fault vector is 1; when negative fault current information is detected, the equivalent fault vector is -1; if no fault current flows through the node, the equivalent fault vector is 0.

[0047] The derivation process of the BN-BPSO algorithm in the embodiment of the present invention is as follows:

[0048] 1. Particle Swarm Optimization (PSO) Algorithm

[0049] The basic particle swarm algorithm should first abstract a bird in space as a particle i, and the particle swarm is a population composed of n particles X = (X1, X2, X3, ..., X n ); the i-th particle represents the D-dimensional vector Xi = (X i1 ,X i2 ,Xi3 ,...,X iD )T; the velocity of the ith particle is Vi=(V i1 ,V i2 ,V i3 ,...,V iD )T;P i =(P i1 ,

[0050] P i2 ,...P iD )T is the individual extreme value. These can be regarded as the actions of a single particle itself, and the global extreme value is P g =(P g1 ,P g2 ,...P gD )T, which can be regarded as the collective action of all particles. In each iteration, the objective function of each particle is evaluated to determine the optimal position p of each particle at time t. best The optimal position g found by the group best , by updating these two optimal positions, the speed and position of each particle are updated according to the following formula.

[0051] V i.j (t+1)=V i.j (t)+c1r1[p i.j -X i.j (t)]+c2r2[p g.j -X i.j (t)]

[0052] X i.j (t+1)=X i.j (t)+V i.j (t+1), j=1, ..., d

[0053] Among them, c1 and c2 are positive learning factors, r1 and r2 are random numbers uniformly distributed between 0 and 1. And the speed range of the particle is adjusted [V min ,V max ] and position interval [X min ,X max ], you can limit the movement of particles to achieve the desired effect.

[0054] In order to allow the algorithm to effectively control the flight speed of particles, the inertia weight coefficient ω is introduced into the algorithm model. At this time, the speed and position expressions of each particle become:

[0055] X i.j (t+1)=ωV i.j (t)+c1r1[p i.j -X i.j (t)]+c2r 2[p g.j -X i.j (t)]

[0056] X i.j (t+1)=X i.j (t)+V i.j (t+1), j=1, ..., d

[0057] (II) Binary Particle Swarm Optimization (BPSO) Algorithm

[0058] The position of a particle in BPSO can be defined from a probability perspective by the size of the particle velocity, so that the position x of each dimension of the particle in Positioned as 0 or 1. In the actual distribution network fault location problem, it can be regarded as a binary nonlinear integer optimization problem. The optimal solution is searched by solving the probability of 0 and 1 in the system. The specific calculation formula is as follows:

[0059]

[0060] The rand() in the formula is a random number uniformly distributed in the interval [0,1] generated after each iteration, and the formula represents the latest position of the i-th particle when calculated to the m+1th time. The Sigmoid function is defined as:

[0061]

[0062] From the above formula, we can get that when the speed v id m+1 The smaller it is, the closer the value of the Sigmoid function is to 0. When the speed is larger, the function value is closer to 1. Since the Sigmoid function graph is S-shaped, 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].

[0063] (III) Improved Binary Particle Swarm Optimization

[0064] The learning factors c1 and c2 determine the empirical information of the particles, and they reflect the information exchange of the particle group. The learning factor c1 affects the particles to move closer to the individual optimal position, while the learning factor c2 affects the particles to move closer to the historical optimal position of the group. In order to ensure the balance between global and local optimization, a PSO model with compression factor is constructed:

[0065]

[0066] In the formula is the shrinkage factor, and its conditional relationship is:

[0067] In this embodiment, C=4.1.

[0068] The inertia weight coefficient can affect the local and global optimal capabilities of particles. Adding linear decreasing weights to perform accurate local searches in the later stages of the algorithm greatly facilitates the convergence of the algorithm. The update formula with the number of algorithm iterations is as follows:

[0069]

[0070] In the formula, ω max With ω min represents the maximum and minimum value of ω, t is the current iteration step, t max is the maximum number of iteration steps. In this embodiment, the maximum weight coefficient is set to 0.9 and the minimum value is set to 0.4.

[0071] (IV) Hybrid Particle Swarm Optimization

[0072] The natural selection mechanism is to sort particles according to their fitness values ​​in each iteration, replace the poor half of the particles with the good half, and retain the historical optimal value of each individual memory. This can ensure that all particles participating in the iteration have good performance, which is conducive to improving the fitness value of the entire population, so that the algorithm can converge quickly globally.

[0073] The principle of Breeding is based on the crossover mechanism in genetic algorithms. In each iteration, a specified number of particles are selected according to the crossover rate and placed in the crossover pool. They are randomly crossbred with each other to produce the same number of offspring particles n, and the offspring particles replace the parent particles m. The positions of the offspring particles are obtained by crossing the parent positions:

[0074] nx=i*mx(1)+(1-i)*mx(2)

[0075] Among them, mx represents the position of the parent particle, nx represents the position of the child particle, and i is a random number between 0 and 1.

[0076] The speed formula of offspring is:

[0077]

[0078] Among them, mv represents the velocity of the parent particle, and nv represents the velocity of the child particle.

[0079] In summary, the steps of the hybrid particle swarm algorithm based on hybridization and natural selection (BN-BPSO) are as follows:

[0080] (1) Randomly set the speed and position of each particle.

[0081] (2) Evaluate the fitness value of each particle and store the particle position and fitness value in the particle's individual extreme value p best In the best The individual position and fitness value of the best fitness value in the global extreme value g best middle.

[0082] (3) Update the particle displacement and velocity, and compare the fitness value of each particle with the particle's best position. If they are close, the current value is used as the particle's best position. best With g best , update g best .

[0083] (4) Sort the particle swarm according to the fitness value, replace the worst half of the particles in the swarm with the best half, and retain the historical optimal value remembered by each individual.

[0084] (5) Set the hybridization rate, extract a specified number of particles and put them into the hybridization pool. The particles in the pool are randomly hybridized to generate the same number of offspring particles. best and g best constant.

[0085] (6) When the algorithm reaches the stopping condition, it stops searching and outputs the results; otherwise, it returns to step (3) to continue searching.

[0086] Implementing simulation and analysis

[0087] The BN-BPSO algorithm of the embodiment of the present invention is compared with other algorithms, and distributed power supply and information distortion are used as experimental variables. Figure 2 The standard 33-node distribution network shown in Figure 1 is simulated. G represents the main power source, and three distributed power sources are connected: DG1, DG2 and DG3. 1 to 33 are the switch node numbers, (1) to (33) are the feeder section numbers, and K1 to K3 are the grid-connected coefficients of the distributed power source DG.

[0088] The BN-BPSO algorithm did not make any errors in locating feeder network faults with single faults, multiple faults, and after adding DG, and it still maintained accuracy when facing distorted information under different circumstances. This algorithm has a good fault tolerance rate and strong anti-interference ability.

[0089] The BN-BPSO algorithm has fewer iterations and fast convergence. In both cases, it can quickly obtain the global optimal value, has good stability, and has high fault tolerance. It can be effectively applied to feeder networks with complex branches.

[0090] In view of the increasingly complex fault location problem, the present invention improves the BPSO algorithm by increasing the compression factor and introducing a linearly decreasing inertia weight, and integrates the hybridization and natural selection mechanisms in the genetic algorithm. The BN-BPSO algorithm is used to locate the fault of the distribution network containing DG, and it is verified that this optimization algorithm has good performance and can be applied to complex distribution networks. The comparison with other algorithms shows that the BN-BPSO algorithm has better computational efficiency and fault tolerance, proving its superiority and feasibility in fault location of distribution networks containing distributed power sources.

[0091] The above is a preferred embodiment of the present invention. It should be pointed out that for ordinary technicians in this technical field, several improvements and modifications can be made without departing from the principles of the present invention. These improvements and modifications should also be regarded as the scope of protection of the present invention.

Claims

1. A distributed feeder fault location method based on BN-BPSO algorithm, characterized in that: The following steps are involved: Step 1: Parameter setting and initialization; Step 2: Construct the switch function and evaluation function of the distributed feeder according to the network structure, and input the fault information uploaded by the feeder terminal equipment; Step 3: Calculate the evaluation function value, and set the individual optimal position and extreme value, the global optimal position and extreme value according to the initial evaluation function value; Step 4: Sort the entire particle swarm by the evaluation function value, replace the worst half of the particles with the best half of the particles, and keep the individual and global extreme values ​​unchanged; Step 5: Extract the replaced particles according to the hybridization rate and put them into the hybridization pool. The particles in the pool are randomly hybridized to generate the same number of offspring particles. The individual and global extreme values ​​remain unchanged. Step 6: Determine whether the descendant particles meet the convergence conditions. If so, output the optimal particle and display the fault area. Otherwise, go to step 3.

2. The distributed feeder fault location method based on the BN-BPSO algorithm according to claim 1 is characterized in that: In step 1, the parameter settings include: population size, maximum number of iterations, learning factor and weight coefficient, particle dimension, hybridization rate; and the position and speed of the initialized particle swarm.

3. The distributed feeder fault location method based on the BN-BPSO algorithm according to claim 2 is characterized in that: In step 2, the switching function of the distributed feeder is: In the formula, K u , K d They represent the switching coefficients of the upstream and downstream power supplies of switch j, respectively. When the power supply is put into operation, the value is 1, and when it is turned off, it is 0; M2 and N2 represent the total number of feeder sections in the upper and lower sections of switch j, respectively; M1 and N1 represent the total number of power supplies in the upper and lower halves of switch j, respectively; s j,u 、s j,d are the status values ​​of all feeder sections from switch j to the upper and lower sections, respectively. The value is 1 when a fault occurs and the value is 0 when the section is normal. j,Gu 、s j,Gd The jth switch to the upstream system power supply G u With downstream distributed power supply G d The state value of the feeder section passed on the path; ju (s), I jd (s) represents the switch function of the upstream and downstream lines of the jth switch; I j (s) represents the switch function of the jth switch; Π represents the logical OR operation; The evaluation function of distributed feeder is: In the formula, I h Indicates the fault overcurrent information of the hth section; I h * (S) is the switching function of the hth section; H is the total number of sections in the entire distributed feeder network; x(h) is the fault state of the hth section, which is 1 when a fault occurs and 0 otherwise; μ is the weight coefficient, which ranges from 0 to 1; The fault information is the equivalent fault vector converted from the overcurrent information collected by the feeder terminal equipment: The direction of the main power supply current is defined as the positive direction. When the feeder terminal device detects positive fault current information, the equivalent fault vector is 1; when negative fault current information is detected, the equivalent fault vector is -1; if no fault current flows through the node, the equivalent fault vector is 0.

4. The distributed feeder fault location method based on the BN-BPSO algorithm according to claim 2 is characterized in that: The particle dimension is the number of distributed feeder sections.

Images