Research of Time Series Finding Algorithm Based on Artificial Neural Network
Yu Zhang · 2009
In this paper, a new method for time-series pattern finding is proposed. The time-series curve is modeled as sequence of straight lines by piecewise linear approximation and the correspondent Arctg(slopes) of the lines are obtained as samples. Classification accuracy of time-series pattern finding algorithm is improved by using artificial neural network (ANN) and hierarchical processing technology. After we analyze experiment result with way of comparing, it prove that classification accuracy of the new time-series pattern finding algorithm is improved.