论文部分内容阅读
提出了基于Chebyshev数值逼近的时间序列直接多步预测算法。该算法具有模型简单、所需的观测样本容量小、易于在线计算及预测精度较高的特点 ,特别适合于有高实时性要求的场合进行实时预测。解决了基于传统ARMA模型建模繁琐 ,模型阶次对预测精度影响大 ,以及神经网络模型收敛速度慢 ,难于满足实时性要求的问题。仿真及实验结果表明了该算法的可行性和有效性。
A time series direct multi-step prediction algorithm based on Chebyshev numerical approximation is proposed. The proposed algorithm has the advantages of simple model, small required sample size, easy on-line calculation and high prediction accuracy. It is especially suitable for real-time prediction with high real-time requirements. It solves the problems that the traditional ARMA model is cumbersome, the model order has a great influence on the prediction accuracy, and the neural network model converges slowly and is difficult to meet the real-time requirements. Simulation and experimental results show that the algorithm is feasible and effective.