基于一步迭代策略的非线性系统模型辨识

One Step Iterative Strategy Based Model Identification for Nonlinear System

针对复杂非线性系统难以建模的特点,将系统的数学函数进行一阶泰勒级数展开,创新性地采用一步迭代算法对系统的线性部分进行辨识,并用严格的数学理论研究样本的选取对模型精度的影响,在此基础上采用BP网络辨识泰勒余项和噪声.然后,将所提出的算法应用于一类典型的间歇系统中,仿真结果表明提出的算法具有比较高的建模精度,为复杂系统的建模提供了一条新途径.

In view of the difficulty of modelling for complex nonlinear system,a novel one step iterative identification algorithm is proposed.Based on taylor series expansion,the algorithm is used for the linear part of the system.The effect of sample on the precision of the model is analyzed by utilizing rigorous mathematical theory,and BP network is used to identify the taylor remainder and noisy.Then,to verify the efficiency of the proposed algorithm,it is applied to a classical benchmark batch process.The simulation results show that the algorithm proposed has a good performance and provides a new way for the modelling of complex nonlinear system.