基于完备模糊规则集的混沌时间序列预测

基于数据挖掘思想，使用置信度度量和改进的梯度下降法，提出一种新的构造完备．模糊规则集的方法来建模和预测混沌时间序列。所提方法通过确定最优输出模糊子集的质心和模糊规则的置信度度量，能够推理数据未覆盖区域的空缺规则，并构造一个完备的模糊规则集，进而解决了混沌时间序列的可预测问题。仿真结果表明新方法是有效和准确的。它能很好地辨识系统的特征，并且提供了一种混沌时间序列预测的新方法。

Prediction of Chaotic Time Series Based on the Complete Fuzzy Rule Sets

Based on the idea of data mining, a new approach, which can construct a complete fuzzy rule sets, was developed to model andpredict chaotic time series by means of confidence measure and improved gradient descent method. By determining the centroids of optimal output fuzzy sets and the confidences of fuzzy rules, the proposed approach could extrapolate missing rules over the regions that data do not cover, and construct a complete fuzzy rule sets, and then resolve the predictable issue of chaotic time series. The simulation results show the effectiveness and accuracy of the proposed approach, which can identify the systems characteristics quite well and provide a new way to make the prediction of tile chaotic time series.