Research And Application Of Rough Set Theory In Cumulonimbus Forecast

Small scale weather is a grave threat to civil aviation flight safety. Small scale weather prediction methods still primarily uses traditional meteorology methods, mainly by the subjective factors of forecasters to predict weather occurrence, development and changes. The traditional method of prediction has been unable to meet the safety of the flight requested by the accurate, quantitative, and timely prediction requirements. As a new soft computing method, rough set theory connects the knowledge with sorts and supports a mathematics tool which accords with the human's for the classification problem of inadequacy data. The knowledge discovery technology based on rough set has become the hot topic of the decision science. For cumulonimbus this article makes use of data discretization, attribute reduction, rule extraction posed by rough set theory to analyze weather data. Furthermore, we establish a new cumulonimbus weather forecast model based on rough set and artificial immune classifier.First, this article has studied carefully the problem of data discretization based on rough set theory. By study and analysis we proposed an improved method of data discretization based on Cramer's V, the method enhances effectively the class- attributes related degree and makes the interval distribution more rational after data discretization. By experiment we proved the effectivity of the method.Second, this article has deeply studied rough set theory, such as attribute reduction, rule extraction and so on. By analyzing and processing weather data about cumulonimbus we established a meteorological database, completed the acquisition of meteorological data, the selection of meteorological factors, the establishment of a rough set decision table based on meteorological data, And then by combining artificial immune classifier, we established a new cumulonimbus weather forecast model based on rough set. Abundant experiments show the model achieve basically our research goal, but we still need enhance the accuracy of forecasting and acquire more effective meteorological factors.