Research And Implementation Of Intelligent Detection System For Vacuum Based On Neural Network Algorithm

As the core component of the high voltage switch,the vacuum interrupter ushered in a very broad market prospect with the increasing application of medium and high voltage power industry in life and production.The vacuum degree is related to the core parameters of vacuum interrupter such as the insulation degree and the switching performance.A vacuum interrupter with unqualified vacuum degree will lead to an unsuccessful interruption of the high voltage circuit,which will cause a major safety accident.The measurement of vacuum degree of vacuum interrupter is a key issue taking this into account.This paper is a set of vacuum measurement system for vacuum interrupters based on the actual testing requirements of a domestic production company of vacuum interrupters.This paper first introduces the status of vacuum measurement in vacuum interrupters both at home and abroad,and focuses on the introduction magnetically controlled discharge method,do the theoretical analysis of the detection process based on the Thomson gas discharge theory,and proves the rationality of getting vacuum degree by surface fitting of leakage current and ionization current.Then,the BP neural network is introduced,and the feasibility of fitting the surface through the neural network is expounded.Taking the actual requirements of the detection system into account,the input layer,the output layer,the number of hidden layers and the number of nodes are selected for the neural network.The design of neural network is completed,and the fitting function of neural network is verified by several common functions.Then the hardware of the detection system is divided into different modules in accordance with the function,and the design and implementation of each hardware module are completed in turn.This paper mainly introduces the pulse high voltage generating module,the magnetic field current generating module,the signal input detection module,the human-machine interaction module and the control core of the detection terminal.The pulse high voltage generator module provides the detection system with the controlled pulse high voltage,and provides the detection system with the security guarantee by feedback control.The magnetic field current generatingmodule can charge and discharge the capacitor stack.When the capacitor is discharged,it will provide the excitation coil with exciting current,and the coil will produce a stable magnetic field under the action of current.The signal input detection module can detect the weak current signal by the multistage amplifier circuit.On this basis,the software of the detection system is designed in this paper.The software mainly includes the program of the microcomputer and the software realization of the upper computer.The microcomputer program controls STM32 to complete the detection process and program through the C language.The host computer can communicate with the micro controller through serial port,control the operation of the detection terminal,and complete the preservation of historical data and the training of neural network.The programming language is Java.Finally,the test platform is set up to debug each module,by witch the function of the detection system is verified.