关键词: 谐波小波分解, 钢铁价格预测, 混沌分析, 神经网络

Abstract: Using harmonic wavelet, chaos analysis and neural network theory, a method is presented to model and forecast steel price. Using harmonic wavelet theory, the steel price time serial is decomposed into different frequency ingredient which can significantly represent potential information of original time serial, and the further analysis of each frequency ingredient indicates that China steel price exists a chaos feature. Then, by using chaos theory and neural network, the forecasting models are established to forecast the every frequency ingredient respectively. By these means, the model can be improved to learn various objective function and more precious prediction can be obtained. The experiments show that the presented method can effectively improve the prediction accuracy.

Key words: harmonic wavelet, steel price prediction, chaos analysis, neural network