Research On Fault Pattern Recognition Technology For The Mining HV Vacuum Distribution Equipment

The paper in this thesis is an important part of the major project “Development of the Condition monitoring and Life Cycle Management for Mine Major Production Equipment Based on Internet of Things”(No :20131101029) sponsored by the Shanxi Science and Technology major projects. It is aimed at the problem that the high failure rate and the imperfect fault monitoring and diagnosis system of HV mobile station distribution equipment in the coal mine.High voltage vacuum distribution equipment is responsible for the main tasks of power distribution in underground coal mine. It is one of the core equipment of mine power supply system. Whether it can work reliably will directly affect the power supply safety and production safety of coal mine enterprises. At the same time, with the increasing of the voltage level and load capacity of the power supply system in the mine, the use of high voltage vacuum distribution equipment has also been an unprecedented increase. However, due to the harsh environment of coal mine, large load change, and the lack of the application of condition monitoring and fault pattern recognition technology, mine high-voltage vacuum distribution equipment during operation due to their own fault caused by coal mine safety accidents occurred frequently. Therefore, in order to accurately diagnose the current state of high voltage vacuum distribution equipment, put forward the reliable equipment operation guidance opinions and suggestions to the user before the high voltage vacuum distribution equipment fault, and to ensure the safety of mine power supply and production safety, the research on the fault pattern recognition technology of mine high voltage vacuum distribution equipment is of great practical significance.Aiming at the existing problems of mine high-voltage vacuum distribution equipment, this paper researched the fault pattern recognition technology, the main research contents are as follows:Through field investigation and consulting a lot of literature, combing the structure of high-voltage vacuum distribution equipment for mining and common fault, the fault of high voltage switch proportion. State characteristics are analyzed and summarized the fault may cause and fault monitoring information, and the overall strategy of system failure mode identification is established.According to the failure mechanism, focus on the analysis of the status of the monitoring objects, including the tripping and closing coil current status characteristics and storage capacitance voltage characteristics, action state characteristics of circuit breaker, vibration characteristics, vacuum arc extinguishing chamber vacuum characteristics. The method of obtaining the state information of the control circuit, the electrical signal, the mechanical state signal and the vacuum degree signal of the arc quenching chamber were further studied.The fault data monitoring and fault pattern recognition experiment platform was build, including the selection of sensors for monitoring, and the configuration of the upper machine and lower machine. The host computer based on LABVIEW platform, and the computer acquisition and communication program, the fault feature extraction classification program and data storage management was written. Through the serial debug, the data acquisition was realized, the transmission is accurate and reliable, which can effectively extract the fault information.Studing of support vector machine(SVM) of the basic theory, a method of fault pattern recognition was put forward based on support vector machine and incremental learning. Combining with the actual industry, the establishment method of fault feature data sample was summarized, and the realization scheme of fault classification and recognition was proposed.The control loop electrical faults and mechanical faults were simulated. Including the power supply voltage, the abnormal of the circuit resistance, the increase of auxiliary switch failure, spindle clamping and separating spring screw loose. The support vector machine classification model and incremental learning model were build, the electrical fault pattern recognition model and mechanical fault pattern recognition model was established. The verification shows that the accuracy of each model was more than 90%, and the purpose of the fault pattern recognition was achieved.A simulation test scheme for the vacuum degree degradation in the arc quenching chamber was proposed. The change of the potential of the shielding cover and the change of the partial discharge in the arc quenching chamber were studied. The experimental results have important theoretical significance and practical value for the recognition of the vacuum degree of arc quenching chamber.