The Application And Research Of Time-series Data Mining In Meteorology Field

Time series data mining is an important component in data mining field, it is one of the hot spots of computer technology. Especially, research on multi-dimensional time series has become a frontier topic. At present, the time series data mining in the financial, telecommunications, medical, power and life sciences and other fields has been widely applied, and has been a great impetus to the development of various fields.Based on the latest research results of time series data mining in domestic and abroad, with the background of analyzing and applying the ground-based automatic weather station data, this thesis researched and proposed a multi-dimensional time series data mining model for the meteorological data. In this model the dimensions redundant reduction algorithm is used for reduced the redundant dimensions and no-correlation dimensions, reduces the complexity of data mining; The extremum slope piecewise linear fitting method is used to implement multi-dimensional meteorological time series segmentation, data compression , feature value extraction and sequence analysis of the trend of change, and further reduces the sequence data; A k-means cluster algorithm is used for make the symbols of sequence, final rule extraction is used for get useful rules in experiments.In the experiment a novel rainfall weather model has been mined through multi-dimensional time series data mining model. The relationship of rain-weather phenomena and meteorological factors have been explained very good .If the rainfall model is been further improvements, it can be used on the forecast of weather. The results of experiment show this model has a great practicability.