In this paper, we present a hybrid AI(artificial intelligence) approach to the implementation of predicting the tom over point of time series. The hybrid AI system integrates the rule-based fuzzy system and the genetic algorithm to accurately predict the turn over point of time series. Firstly, the clustering- based method is applied to determine the optimal number of fuzzy rules and establish an initial fuzzy model. Secondly, the genetic algorithm is exploited to tune the parameters of rule base and membership functions.