Study On Transmission Line Galloping Prediction System Based On GPSO-BP Neural Network

Galloping of transmission lines is a frequent safety accident in power grid system.Galloping of transmission lines may cause trip,flashover and even wire break and tower fall,resulting in huge losses.In this paper,an improved BP neural network based on improved particle swarm optimization(GPSO-BP)algorithm is proposed,and a transmission line galloping prediction system is developed,which can realize the prediction of conductor galloping,and has important theoretical significance and engineering application value for the safe and stable operation of power grid.The main research contents and achievements of this paper are as follows:(1)The text mining technology is used to deeply analyze the relevant literature on transmission line galloping research,and the key factors causing galloping are extracted by word frequency and semantic network analysis.It is found that there is a strong correlation between the span,thickness of ice,temperature,wind speed,relative humidity,rainfall and angle between wind direction and line and galloping,and the main parameters for prediction of transmission line galloping are determined.It provides the basis for establishing the prediction model of transmission line galloping.(2)An improved particle swarm optimization(GPSO)algorithm was proposed based on the traditional particle swarm optimization(PSO)algorithm by adopting linear decreasing inertia weight and adding simple mutation operator.Improved particle swarm optimization(GPSO)algorithm was used to optimize the initial weights and threshold value of BP neural network,and the number of hidden layer nodes of BP neural network was determined by comparative optimization method.GPSO-BP algorithm was proposed by combining improved particle swarm optimization algorithm with BP neural network,this algorithm has stronger global and local search ability,make up for the deficiency of the particle swarm optimization(PSO)algorithm and BP neural network.(3)GPSO-BP algorithm was used to simulate the complex nonlinear mapping relationship between galloping and galloping factors of transmission lines,and a galloping prediction model of BP neural network optimized by improved particle swarm optimization(GPSO-BP)was proposed.The GPSO-BP neural network galloping prediction model was verified by the galloping example of xiezhuang transmission line in Henan province in 2009.The results show that the reliability and effectiveness of GPSO-BP neural network galloping prediction model meet the requirements.(4)APP Designer was used to design the interface of the established GPSO-BP neural network galloping prediction model,and the transmission lines galloping prediction software based on galloping causes was developed.This paper introduced the function of the software and the realization process of the function in detail,and introduced the operation flow of the software through the interface design.