A Designation Of Short Term Lightning Warning System Based On Cluster Analysis And Grey Model

Lightning is a natural phenomenon formed by multi factors. It is influenced by the thickness of the clouds, air humidity, terrain, vegetation and other factors in the process of its formation, occurrence and development. In addition, it is almost instantaneous from the occurrence of lightning to its end. As a result, it is too hard to observe the formation of lightning to obtain useful information when it happens. Therefore, the research of lightning has long been a difficult topic. Especially, it is difficult to accurate prediction on where the lightning will happen at, when it will happen and the intensity of it. Because of its huge and destructive power, lightning is also a kind of nature disaster. There are many people died in lightning in our country each year, what’s more, a huge number of loss are caused by lightning, and the numbers of this two will be much bigger in worldwide. So it is necessary and important to explore the secret of lightning, and the research also has important natural and realistic significance. With the continuous development and progress in science and technology, China’s relevant departments have mastered a lot of historical monitoring data of lightning activity. The regularities in lightning activity which are hidden behind the data can be mined and made use of in the guidance of national product through make full use of these data, such as the distribution of lightning in space and time.Two classical cluster analysis methods and the grey model are chosen in this paper after a series of experiments, test and comparison. A combination algorithm based on the two cluster analysis methods has been proposed to make the prediction of the range in which the lightning may happen, in the method, Firstly, extract the lightning sample points, then cluster the sample points into different categories through the maximum minimum distance algorithm; in addition, optimize each clustering category through K-means algorithm; Secondly, regard the clustering centers as samples and cluster them through K-means algorithm to get a predictive category. Besides, an improved grey disaster prediction model is proposed to make forecast of time at which heavy lightning may happens, in this model, first, get the time at when the latest historical lightning happened; then make a process of this data through the knowledge of grey wave model; thus, the processed data can be used to establish a grey disaster model to predict the possible time of the lightning.A brief introduction of related knowledge has been made in this paper, including the background of data mining, cluster analysis and grey system. And the following is a detail discuss on how the two cluster analysis algorithm and grey disaster model functions on real data has been given.Finally, a summary of the paper is given in which shows the limitations and shortages of the algorithm. At the same time, the plan to re design the models is also put forward on condition that more useful information can be used, such as meteorological information.