GA optimization serial BP network-based cable joint wire temperature prediction method

Abstract

The invention provides a GA optimization serial BP network-based cable joint wire temperature prediction method. The method comprises the following steps of selecting and normalizing related data; establishing a training model; and performing prediction by using the trained model. The innovation of the invention lies in that two BP neural networks are established in series in the modeling process;the reflection factor data and the prediction result are trained and learned respectively, wherein an output vector of the network 1, namely a reflection factor data prediction result, is used as aninput vector of the network 2; the weights and thresholds of the two networks are optimized by using a genetic algorithm; therefore, under the condition that no reaction factor data exists, the groupof target values can still be accurately predicted; the network 1 is used for training the reflection factor data to obtain the temperature value at the corresponding moment, then the network 2 is used for training the prediction results of the reflection factor data at three continuous moments to obtain the temperature value at the fourth moment, and the whole process of solving the temperaturevalue at the fourth moment does not need the reflection factor data at the moment. The model not only has strong learning ability of the BP neural networks, but also combines excellent global search ability of a genetic algorithm, and meanwhile, serial fusion of the two BP neural networks enables the prediction performance of the model to be more excellent.

Description

technical field [0001] The invention relates to a method for predicting the temperature of a wire at a cable joint in a power transmission system, in particular to a method for predicting the temperature of a wire at a cable joint in a power transmission system based on a genetic algorithm optimized serial BP neural network. Background technique [0002] Genetic Algorithm (GA) is a model based on the natural selection phenomenon of survival of the fittest in the biological world. It was proposed by Professor Holland in the United States in 1975 to solve optimization problems, also known as evolutionary algorithm. The genetic algorithm is simple and easy to operate. It uses group search technology, and the objects it operates and the solution to the problem are populations. A new generation of population is generated by applying a series of genetic operations to the current population, including the selection of preferred strong individuals, the crossover of gene fragments ex...

AI Technical Summary

Problems solved by technology

At this stage, the sensitivity of cable detection equipment is low and the maintenance cost is high. More importantly, once a failure occurs, some power systems will not be able to operate normally or even severely paralyzed.

Images