High Voltage Submersible Motor Insulation Life Prediction Based On BP Neural Network

High voltage submersible pump with high voltage submersible motors arewidely used in rivers, lakes, agricultural irrigation, sewage disposal in cities orfactories, offshore oil platforms to extract seawater and other important occasionsof the economy, especially in the drainage, emergency rescue and disaster relief,submersible motors play an important role. A high voltage submersible motorworks in the deep water all the year, and its operating insulation performancedeteriorates in the complex environment. Besides, due to the special installationenvironments of the motor, it can not be maintained in time. Consequently, it is ofgreat significance to predict the insulation life expectancy and reduce the lossescaused by the motor deterioration.This article describes the factors which influence the insulation aging of thesubmersible motor, analyzes the mechanism of partial discharge of the submersiblemotor, gives the partial discharge model of the motor, and especially studies theinfluence of voltage on the partial discharge of the submersible motor. Thesubmersible motor losses during operation and the motor’s heat-producing processare introduced. The impact of temperature on the submersible motor insulation lifeis investigated. And the boundary conditions in the heat transfer process and theheat transfer model are given; at last, finite element method is used to analyze thetemperature field of the submersible motor stator. According to the structuralcharacteristics of the submersible motor, the thesis analyzes the influence of thewater pressure on the submersible motor. Since the phenomenon of water is in thesubmersible motor insulation air gap, the change of insulation resistance anddielectric loss factor which affect the motor insulation life is studied, thus the BPneural network structure and algorithm characteristics are introduced according tothe principle of life-accelerating test. At the same time, the ways of using BP neuralnetwork to predict the insulation life-expectancy of the high voltage submersiblemotor are proposed. It is proved by the accelerated life experiment that actualrequirements are obtained using BP neural network to predict the motor insulationlife-expectancy.