A Method For Faulty Phase Selection In HV Transmission Lines Based On Data Mining Technology

Fault phase selection is a critical link in relay protection of HV-transmission lines, and fast and accurate selection is of great significance to the safety of the lines and stability of the system. In the HV-transmission line systems in our country, the fault phase selection elements play an important role in assuring the regular function of the protection equipments. Especially those new type protections that gain wide attention recently such as traveling wave protection and transient protection are more dependent on reliable fault phase selection . Phase selection elements are important parts of the line protection equipments, and they serve the combined reclosing in conventional protections, to select the single fault phase when a single fault occurs and then trip the circuit breaker , otherwise cut the three phase circuit breakers.The paper researches and analyzes the present state of the phase selection methods, and gives deep analysis, comparisons and discussions to the basic algorithm and its application condition of each method, focusing on the application of Shannon entropy and SVM in the transmission line fault selection. The association between fault characteristics and EMD entropy and EMD singular entropy are illustrated .Through finite EMD entropy and EMD singular entropy with definite physics meaning, the fault characteristics of transient signals are described completely and clearly, and fast and exact decisions of the fault type and fault phase are easily realized. The application of EMD in phase selection overcomes disadvantages of other methods , such as inaccurate decision of the fault phase and disability to recognize whether a grounded fault occurs or not and mistakes in phase selection.Combined with analysis of fault signals, the paper applies EMD, Singular Value analysis, SVM and Shannon entropy to the extraction and analysis of transient signals. The method is fast and accurate in phase selection, and is immune to the effects of the noises ,the transitional resistance ,the fault locations and the initial degree of the fault. Large number of simulation results indicate that the application of the proposed data mining technology in HV transmission line fault phase selection is feasible and effective.