Research On Fault Diagnosis Of High Voltage Disconnector Mechanism Based On Motor Stator Current Characteristics

As the most widely used high-voltage switchgear in traction power supply system,high voltage disconnector is exposed to extremely harsh outdoor environment all year round.It is affected by dust,wind,rain and snow.It is very easy to cause mechanism failure problems such as mechanism corrosion,spring failure and mechanism fatigue,which leads to jam,not in place,spring failure and other problems during opening and closing.This is strict It is a serious threat to the safe and stable operation of traction power supply system.Many methods have been put forward for the diagnosis of the faults of the mechanism,but there are still some shortcomings such as the insufficient accuracy and the low intelligence.Therefore,this paper studies the fault diagnosis of high voltage isolation switch mechanism in traction power supply system,and proposes a fault diagnosis method of high voltage isolation switch mechanism based on stator current characteristics.The stator current feature of motor is used as the input of diagnosis algorithm,so as to judge the fault type,and the proposed method provides a new way for fault diagnosis of high voltage switchgear.The paper describes the motor fault signal,sampling filter design,feature extraction,diagnosis algorithm and system building,and the main contents are as follows:1)Analyze and study the characteristics of motor stator current caused by mechanism failure.According to the derived relationship between the operating torque of the isolating switch and the motor current,the functional relationship between the operating torque and the motor stator current is obtained.After analyzing the characteristics of the motor stator current signal,the relationship between the current fluctuation caused by the torque change and the fundamental wave and the side frequency component of frequency is established when the mechanism fails.This provides a theoretical basis for using the fundamental wave and45Hz～55Hz sideband components of the stator current as the characteristic quantities for fault diagnosis.2)The sampling and extracting method of motor stator current characteristic quantity are studied.In view of the large amount of harmonic current in stator current,a combination of low-pass Butterworth filter and digital FIR band-pass filter is designed,which can reduce sampling interference and filter out a large number of harmonic components of current,improve the robustness of diagnosis and analysis.The 45-55 Hz amplitude frequency characteristics of current extracted by fast Fourier transform and root mean square method are used to calculate the stator current The effective value of the fundamental current time t relation of the signal is compared and analyzed in this paper.3)The accuracy of fault diagnosis based on the fusion of current characteristic quantity and different algorithms is analyzed and compared.Taking the normalized stator current’s 45 Hz ～ 55 Hz sideband component as the input of the algorithm,the accuracy of K-means clustering analysis,SVM classification model and BP neural network for fault diagnosis is compared.The results show that BP neural network has the highest accuracy of 97.9%,which is higher than 92.5% of k-means clustering analysis and 95.8% of SVM fault classification model,It shows that BP neural network has stronger ability to process and predict large sample data.In the four kinds of faults,the diagnosis accuracy of the mechanism jam is low,because the slight jam can not be clearly shown in the stator current characteristics.4)The fault diagnosis device of disconnector mechanism based on Lab VIEW and DSP is designed.This paper introduces the design of peripheral circuit such as sampling filter unit,power supply unit and other peripheral circuit design and DSP board software development environment configuration from two aspects: hardware part and upper computer software part.And Lab VIEW is used to design the upper computer software.In the fault diagnosis method,BP neural network with higher precision is used to realize the display of current signal and characteristic quantity by using upper computer Indicates and alarms for mechanism failure.