Study of Long-term Runoff Forecast Model Based on Association Rules Mining

摘要

Association rule is an important method of data mining techniques. There are a mass of hydrological and forecasting data in the region of long-term runoff forecast. How to fully analyze and mine those data via various intelligent algorithms to formulate accordingly hydrological forecast for precise forecast is of crucial importance. Considering the characteristics of hydrological forecast, association rules mining method is applied to the long-term runoff forecast. The hydrological and meteorological data from 1956 to 2005 are selected to constitute the runoff forecast database of Jiangqiao hydrologic station at Nenjiang River. So as to find the strong association rules which accord with the min-support and min-confidence, we discretize the values of the attributes by the standards. In the practical example of Jiangqiao station, three strong association rules are mined and these rules reveal the effects of the north pacific sea surface temperature (SST) on the flood season runoff at Jiangqiao hydrologic station. The results of test show that the qualified rate of the model comes to 80%, and the model is highly effective for the flood prediction of Jiangqiao station in flood season. Furthermore, association rules mining may be one of effective tools for the long-term hydrological forecast.