复杂过程的多模型建模方法研究

针对复杂工业过程的非线性、工况范围广的特点,本文提出了一种新的多模型建模方法.首先对系统按照工况划分准则进行双层K均值聚类,在此基础上采用LS-SVM算法建立各局部模型,并用粒子群算法对多模型权值进行辨识.此建模方法收敛速度快,对辨识过程有良好的全局适应性.并以典型非线性主汽温系统作为辨识对象,采用上述方法建立其系统的多模型,仿真结果验证了该方法的有效性.

Multi-model modeling approach for complex process

Aiming at the characteristics of nonlinearity and wide operating condition range of complex industrial processes,a new multi-model modeling approach is presented in this paper.Firstly,the system is processed with double level K means clustering according to the classification criteria of operating conditions;then,LS-SVM is adopted to establish the local models,and particle swarm algorithm is used to identify the weights of multi-models.This modeling method has fast convergence speed and good global adaptabi...