Study on Data Mining for Grounding Fault Line Selection in 6kV Ineffectively Grounded System of Coal Mine

A great amount of fault wave has been recorded by the devices for detecting phase-to-ground faults in ineffectively grounded systems. However, a better method hasn´t found for effectively taking advantage of these data to improve the result of fault line selection. Data mining techniques can be used for fault line selection in ineffectively grounded system to gain knowledge from the existing data and to improve the technique of fault line selection. This paper briefly describes the principles, methods and implementation of data mining techniques, classifies the fault samples of ineffectively grounded systems by using clustering analysis method, employs different fault line selection methods according to the types of faults, and consequently provides a set of criteria for modeling of typical ineffectively grounded systems and verifying the validity of real-time fault line selections. The validity of the methods has been convinced by the calculation using the data obtained from the real performance of a substation in coal mine. It has been shown to be promising to employ the data mining techniques in ineffectively grounded systems fault detection. This paper provides very good methods for resolving the difficulties with onsite tests, enhancing the techniques of fault line selection and establishing the fault detection management systems.