摘要
针对多变量工业过程中存在的耦合性、建模困难以及不确定性的控制难点,论文研究了基于解耦设计的多变量隐式广义预测控制。该方法利用加权最小二乘递推方法根据输入输出数据直接辨识系统参数,这样就降低了控制器设计的复杂性;采用分散设计的方式来降低各通道间的关联关系。论文利用基于最小二乘方法,根据输入输出数据来建立被控对象的数学模型,并在此基础上对控制方法进行了仿真分析与试验研究,结果表明该方法具有良好的控制效果。
To address the control problems of multivariable industrial system with coupling and nonlinearities, we present our IGPC method for dealing with them. Sections 1 and 2 of the full paper explain our IGPC method, which we believe is better than existing ones. The core of sections 1 and 2 consists of： （ 1 ） according to input and output data, the parameters of optimal control law can be identified through use of least squares approximation ; （2） using decentralized objective function, we apply the decoupling IGPC to control a multivariable system; （3） recursive weighted least squares （RWLS） method is used to establish the mathematic model of system. Simulation and exper- imental results, presented respectively in Fig. 1 and Fig. 2, and their analysis show preliminarily that our IGPC method is indeed better than previous ones.
出处
《西北工业大学学报》
EI
CAS
CSCD
北大核心
2012年第6期936-940,共5页
Journal of Northwestern Polytechnical University
关键词
多变量系统
解耦设计
隐式广义预测控制
系统辨识
温度控制
computer simulation, design, experiments, identification （control systems）, least squares approxima-tions, mathematical models, multivariable systems, nonlinear systems, optimization, parameter ex-traction, temperature control
decoupling, implicit generalized predictive control（IGPC）
