时间序列模型的进化识别方法及其应用

Evolutionary identification approach for time series modelling and its application

针对常规时间序列分析存在的模型结构和参数选择的困难,引入进化算法的全局优化思想,设计了一种模型结构和参数分别进化、共同识别方案,实现对非线性动力学模型结构和参数的全局最优识别.实例分析表明该方法具有较高的精度和推广预测能力.

As to the difficulties in selecting model structures and model parameters in traditional time series analysis,the genetic evolutionary algorithm was introduced and a new evolutionary method was designed to perform global identification of the nonlinear dynamic model where the model structure and model parameters are evolved separately.Application results proved the high precision and generalization capacity of the new method.