Mechanical Fault Diagnosis Of GIS Equipment Based On Joint Analysis Of Vibration And Sound

With the comprehensive development of UHV project in our country,the SF6 GIS Gas Insulated Switchgear(GIS)has been widely used.Compared with traditional power equipment,GIS has numerous advantages such as small size,small footprint,high performance and low failure rate.However,GIS also has some disadvantages.For example,when the fault occurs,it is difficult to determine the fault type and locate the fault due to its full closure,leading to more serious accidents and huge losses.Moreover,it is not easy to identify some of the micro mechanical defects of the GIS during the installation,acceptance and early operational stages.Therefore,it is very important to set up a real-time monitoring system for GIS equipment.The main contents of this thesis are as follows:A joint vibration-sound signal analysis based mechanical fault diagnosis system for GIS equipment is proposed,which can realize a real time and all-weather monitoring.The system exploits the advantages of both sound and vibration signals.Through analysis based on the electromotive force theory as well as mechanical faulty simulation experiments for GIS isolating switch,time-frequency characteristics of the sound and vibration signals have been acquired,which can be further regarded as an evidence to fault diagnose.The noise type and power spectrum characteristics of each type of noise around high-voltage substation are analyzed,and the characteristics of different types of ambient noise and the characteristics of the power spectrum of the target sound signal are compared.According to the above noise data analysis results,an environment noise filtering scheme combing both hardware and software is designed,which includes a set of methodologies on how to make selections on different devices that pick up sound signals,selections on different sound insulation devices,as well as a novel noise filtering algorithm that is based on adaptive noise cancellation principle.According to the basis of GIS mechanical fault diagnosis,two algorithms that extract GIS vibration or sound signal features are proposed,which are,namely,a improved approach to extracting power frequency harmonic structures,as well as a multi-resolution Gaussian filter bank time-frequency analysis method.The former can extract power frequency harmonic structure more accurately compared with the traditional method,and thus effectively avoid the impact of spectrum leakage;while the latter exploits the characteristics of the ear hearing model and wavelet transform,and can perform fine time-frequency analysis on GIS acoustic and vibration signals to provide the time-frequency spectrum of the signal features.Several suspected GIS electrical components were found to be in abnormal operation status in substations from Xi’an and its surrounding cities during the routine inspections by the State Grid.Through the analysis of the vibration and sound data collected from the field test,it is confirmed that there are some mechanical defects such as the poor contact of the grounding switches and looseness of the shield.Moreover,the effectiveness of the proposed system in the monitoring and fault diagnosis of GIS in the actual substation is verified.