CN-116224145-B - Full-line grounding state evaluation method based on improved hybrid genetic algorithm

Abstract

The invention provides a full-line grounding state assessment method based on an improved hybrid genetic algorithm, which belongs to the problem of monitoring of grounding resistance of an electric power system. According to the invention, the ground resistance measurement mode is improved, the ground resistance is monitored in real time, trend analysis and early warning are carried out, the manual measurement workload is reduced, and the working efficiency is improved.

Inventors

• HUANG XINZHANG
• WANG RUOXI

Assignees

• 沈阳工程学院

Dates

Publication Date

20260512

Application Date

20221207

Claims (2)

1. 1. The full-line grounding state evaluation method based on the improved hybrid genetic algorithm is characterized by adopting the improved hybrid genetic algorithm to analyze and predict through a BP neural network so as to realize real-time monitoring of the grounding resistance, and comprising the following steps: Step 1, measuring each grounding resistance; Step 2, establishing an objective function for improving a hybrid genetic algorithm, which specifically comprises the following steps of; Step 2.1, setting population scale M and crossover probability of the traditional genetic algorithm Variation of Probability of Iteration number N; Step 2.2, generating an initial population; step 2.3 regarding the difference between the apparent ground resistance value and the actual measured value The expression is as follows: , wherein: each grounding electrode value obtained by actual measurement; step 2.4 the difference between the apparent ground resistance and the actual measured value according to step 2.3 The smaller the solved resistance value The closer to the true value, the objective function F of the ground resistance, i.e. the fitness of the individual, is obtained as follows: ; Step 3, performing crossing and mutation operations on the reserved individuals; step 4, calculating the center of the population The expression is as follows: , Step 5, connecting As the initial search point of conjugate gradient method, the iteration times are set Executing to obtain the result ; Step 6, connecting Adding the solution into a population, replacing individuals with minimum fitness in the population, turning to step 2, and stopping the algorithm after the set condition is met to obtain a local optimal solution; step 7, constructing Lagrange function by taking the local optimal solution obtained in the step 6 as an initial value of a sequence quadratic programming algorithm; step 8, obtaining Lagrange factors Determining the step length; Step 9, obtaining new iteration points Updating the Hessian matrix; Step 10, judging Whether the iteration termination condition is satisfied, if so, outputting a final junction Fruit, otherwise, let Repeating the steps 7 to 9; Step 11, constructing a BP neural network, wherein the number of layers of the BP neural network is 3, and the BP neural network is an input layer, a hidden layer and an output layer respectively; Step 12, initializing parameters; step 13, performing BP neural network training, which specifically comprises the following steps: Step 13.1, calculating a hidden layer and outputting the hidden layer And input vector Input layer-hidden layer connection weights Hidden layer threshold The relationship of (2) is expressed as follows: , wherein: output for hidden layer; An excitation function for the hidden layer; step 13.2 BP neural network output And hidden layer output Hidden layer-output layer connection weights And output layer threshold The relationship of (2) is expressed as follows: , Step 13.3 calculating error, predicting error And desired output And prediction output The relationship of (2) is expressed as follows: , step 13.4, updating the weight value and utilizing BP prediction error Updating connection weights of a network And The expression is as follows: , , wherein: Is the learning rate; Step 13.5 updating the threshold value, utilizing BP prediction error Updating a threshold of a node And The expression is as follows: , ; step 14, judging whether the error reaches the prescribed requirement, outputting data if the error reaches the prescribed requirement, and re-entering step 13 if the error does not reach the prescribed requirement.
2. 2. The method for evaluating the grounding state of the whole circuit based on the improved hybrid genetic algorithm according to claim 1, wherein the step 1 specifically comprises the following steps: step 1.1 defining the apparent resistance at the 1 st ground electrode in the circuit as The computational expression is expressed as follows: , Wherein u is a measurement voltage; Measuring current for the 1 st ground electrode; The 1 st grounding electrode has a grounding resistance value; step 1.2, measuring the resistance of each grounding electrode loop Obtaining the apparent grounding resistance of each grounding electrode The expression is as follows: , wherein: ; the total parallel resistance value of the rest branch circuits; step 1.3 percentage error with respect to ground resistance measurement The expression is as follows: 。

Description

Full-line grounding state evaluation method based on improved hybrid genetic algorithm Technical Field The invention belongs to the problem of monitoring the grounding resistance of a power system, and particularly relates to a full-line grounding state evaluation method based on an improved hybrid genetic algorithm. Background At present, the domestic power transmission lines are widely distributed, long in distance and crisscross, are in mountain areas with complex topography and changeable in climate, are easy to suffer tripping accidents caused by lightning direct impact in the running process of the power transmission lines, and cause serious damage to a power system, so that the power transmission lines need to have a good drainage channel. The tower grounding device is used as a main component of the drainage channel, and plays an important role in the stable operation of the power transmission line. When a lightning strike accident occurs, lightning current flows into the ground through the grounding device, and if the grounding resistance is too large, the lightning conductor can generate counterattack overvoltage, so that the circuit is tripped. The measurement of the grounding resistance of the transmission line tower is affected by a plurality of factors such as the form of a grounding body, inductive components, the inductance of the tower, the inductance of a lightning conductor, the surrounding environment and the like. Therefore, how to quickly and accurately measure the grounding resistance of the tower of the power transmission line is one of the problems to be solved in the power industry at present. The current measurement of the grounding resistance of the tower mainly adopts a megger method and a clamp meter method. The megger method is accurate in measurement, but is complex in operation, voltage poles and current poles are required to be arranged, the grounding down conductor must be disconnected during measurement, the labor intensity of staff is greatly increased, and the efficiency is low. The clamp meter method is simple and convenient to measure, the ground down wires are clamped by the clamp jaws, when one ground wire is arranged on the pole tower, the ground down wires are not required to be disconnected, when a plurality of ground down wires are arranged, other ground down wires are still required to be disconnected, only the tested wires are reserved, inconvenience is brought to measurement, and the measurement error is large. The two methods have the common defects that manual and periodical field measurement is needed, the grounding down wire is required to be disconnected during measurement, a great amount of manpower and material resources are consumed, and the loosening or poor contact of bolts is easy to cause, so that the grounding resistance of the pole tower is increased. Therefore, an improved ground resistance measurement mode is urgently needed, real-time monitoring of the ground resistance is achieved, trend analysis and early warning are carried out, manual measurement workload is reduced, and working efficiency is improved. Disclosure of Invention The full-line grounding state evaluation method based on the improved hybrid genetic algorithm reduces the manual measurement workload and improves the working efficiency. In order to solve the technical problems, the invention adopts the following technical scheme: An all-line grounding state evaluation method based on an improved hybrid genetic algorithm adopts the improved hybrid genetic algorithm to analyze and predict through a BP neural network so as to realize real-time monitoring of grounding resistance, and comprises the following steps: Step 1, measuring each grounding resistance; Step 2, establishing an objective function for improving a hybrid genetic algorithm; Step 3, performing crossing and mutation operations on the reserved individuals; step 4, calculating the center of the population The expression is as follows: Step 5, connecting Setting iteration times N 0 as initial search point of conjugate gradient method, and executing to obtain result Step 6, connectingAdding the solution into a population, replacing individuals with minimum fitness in the population, turning to step 2, and stopping the algorithm after the set condition is met to obtain a local optimal solution; step 7, constructing Lagrange function by taking the local optimal solution obtained in the step 6 as an initial value of a sequence quadratic programming algorithm; step 8, obtaining Lagrange factor theta and determining the step length; Step 9, obtaining a new iteration point χ θ+1 and updating the Hessian matrix; And 10, judging whether χ θ+1 meets the iteration termination condition, if yes, outputting a final result, otherwise, enabling θ=θ+1, and repeating the steps 7 to 9. Step 11, constructing a BP neural network, wherein the number of layers of the BP neural network is 3, and the BP neural network is an input layer, a hidden