论文部分内容阅读
本论文基于样本数据的模型辨识方法,通过对时间序列的可叠加性特点分析,并采用典型时间序列模型组合形式构造时间序列模型。针对时间序列中可能存在的离群点,在模型中引入影响函数来表示。论文中采用遗传算法来确定模型的结构,然后使用粒子群算法来确定每个模型中的参数值。通过对几个实际问题的仿真分析可以得出结论:无论是对已知模型结构的辨识还是对未知模型结构的辨识问题,这种方法都是可行的。
Based on the model identification method of sample data, this paper analyzes the stackability characteristics of time series and constructs the time series model by using the combination of typical time series models. For the possible outlier in time series, we introduce the influence function in the model. The paper uses genetic algorithm to determine the structure of the model, and then uses the particle swarm algorithm to determine the parameter values in each model. Through the simulation analysis of several practical problems, we can draw a conclusion: This method is feasible whether it is the identification of the known model structure or the identification of the unknown model structure.