Association Research Between Lightning Activity And Topographical Factors Based On Classification And Clustering

Lightning disturbance has great effect on safety operation, thus lightning protection of power grid and especially transmission lines is very important. The works of mastering lightning activity characteristics and research of basic parameters can help effective opera-tion of lightning protection of power system. As a result, association analysis between lightning activities and topographical factors is helpful for construction of transmission lines, site selection of towers and assurance of lightning protection level, for avoiding lightning disturbance in advance and for improving safety and stability.In the past, the characteristics of lightning activities were studied through statistical methods, such as calculation of correlation coefficient and establishment of regression equ-ation. Nevertheless, those methods didn't consider randomicity of lightning activities with numerous influencing factors, so that traditional research methods are hard to obtain effec-tive and correct results. Therefore, a kind of new thought for research of lightning activity characteristics is put forward in this paper. Classification and Clustering are used to analyze historical monitoring lightning data in Anhui province in order to study the association be-tween lightning activities and topographical factors and deeply mine the characteristics of lightning activities.In this paper, a sampled database with 5612 columns and 52 rows is built as the re-search base of the decision tree model. The database include flash density attribute and to-pographical attributes. The flash density attribute is one of the fundamental parameters of corridors of transmission lines. And the topographical attributes are absolute altitude dif-ferences between the tower and specify points which is 100-500 meters away from the tower in the directions of East, West, South, North, North-East, South- East, North-West and South-West. Normality test and correlation analysis are carried out based on all the attributes, 11 of that with high correlation with flash density attribute selected. Then the C4.5 decision tree model is built with the attribute of flash density classes as a target attribute and 11 topographical attributes as candidate attributes. If-then rules are obtained from the decision tree, which reflect the characteristics of lightning activities in small areas defined with longitude and latitude. The prediction of flash density classes is of high accu- racy using the decision tree.In order to study the correlation between a thunderstorm and special topography, 3 times of thunderstorm occurring in the waters in Anhui are selected to carry out clustering analysis. Time step is set in the clustering model to avoid the effect of time for lightning activities. The centers of corresponding clusters obtained from the model are displayed on Google Earth with different colors according to the number of the flashes in every cluster, which reflect the intensive extent of lightning activities. The results show that the flashes of a thunderstorm tend to occur in waters and mountainous region.Data mining techniques are applied to association research between lightning activity and topographical factors in this paper. Quantitative description way is used to represent the characteristics of lightning activities. Data mining techniques prove to be efficient and adaptable for study of Lightning activity characteristics.