GIS Mechanical Failure Research Based On Vibration Data And Image Data

With the vigorous development of our country's economy,the scale of the power system is also getting larger and larger.Building a safe and reliable power transportation system is related to all aspects of people's production and life.Gas Insulated Switchgear(GIS)is one of the core equipment of the power system.This equipment uses SF6 gas instead of air at atmospheric pressure as the insulation medium.It is a metal folly enclosed switchgear.This equipment plays the role of protection and control circuit,and other equipment of the substation is controlled and protected by it,so it is an important guarantee for the normal operation of the power system.According to statistics,mechanical failures in GIS account for about 39%of all failure types,and are the most common type of failure.When the equipment fails,the substation needs power outages and maintenance,which has greatly affected people's production and life.Therefore,It is necessary to carry out relevant tests and research on mechanical failure of GIS equipment.The GIS mechanical defect model is built in the laboratory,and explores the use of sound source localization equipment,vibration detection equipment,and X-ray imaging equipment through multiple experiments.On the basis of studying the vibration mechanism of GIS equipment,the acoustic data and vibration data are continuously analyzed,and how to find the point of equipment failure is mastered.Then use the X-ray imaging system to capture the perspective image of the fault point.Considering the complexity of the equipment components,the identification of different components should be realized first,and the fault diagnosis of different components should be realized secondly.After analyzing and comparing the X-ray images,the invariant moment features and texture features capable of distinguishing the two types of images of the circuit breaker adsorbent cover and the tube weld are extracted,and then these features are sent to the neural network for use.The component recognition model was trained,and finally the fault diagnosis was performed on the welded joint of the tube by using image processing techniques such as local threshold method.The final result confirms that compared with the convolutional neural network model,the component recognition model proposed in this thesis has higher generality,can obtain better recognition results in the case of small samples,and can be left over when welding can be detected at the welding place of the pipe bus And spot welding marks on the original X-ray fluoroscopy image.According to the judgment principle that the existence of welding spots is regarded as a failure,the final result does not affect the qualitative diagnosis of the welding place of the tube mother.So as to meet the requirements of auxiliary staff for fault diagnosis.Using vibration and image data to automatically diagnose equipment failures not only saves time and personnel resource costs,but also ensures the accuracy of the diagnosis,avoids subjective interference by the staff,and provides equipment maintenance methods in a timely manner.This article draws on the prior knowledge of professional researchers,using image processing,neural networks,and pattern recognition technology to realize the identification of the circuit breaker adsorbent cover and the welding of the pipe bus,and mark the welding spots on the welding of the pipe bus.To help staff in troubleshooting.