Research On Multi-hierarchical Model Predictive Control Of Nonlinear System

Some practical industrial systems have strong nonlinear and dynamic characteristics, and they are difficult to be controlled. Multiple model predictive control algorithm has become an effective method to solve control problems of nonlinear systems. However, when the working condition of the multivariable nonlinear system is under a large range change, traditional multi-model methods cann't effectively describe the dynamic characteristics of systems which can affect dynamic performances. According to this problem, a methodology called multi-hierarchical model predictive control is obtained in this thesis. The main research results can be listed as follows,1) For the modeling problems of complex nonlinear systems, multi-hierarchical model structure is proposed. The structure is composed of several layers with multi-models, and the global operation spaces are the same while the sub-model has different operation spaces in each layer. Experimental results show that the multi-hierarchical model can solve nonlinear system modeling problems better than the traditional multi-model structure.2) In consideration of the MIMO nonlinear system's control problem with large variation of working condition, multi-hierarchical model predictive control method is proposed, which combines the multi-hierarchical model structure with GPC algorithm. Experimental results show that the novel predictive control method based on multi-hierarchical model can effectively solve the control problems of complex nonlinear systems than the traditional multi-model predictive control algorithm.3) For switching problem in the multi-hierarchical model strategy, a switching method based the output error and its change ratio is proposed on analysis of the characteristics of the different layer models and their impact on control systems. Experimental results show that the method has well control effect.4) On analysis of the influence of different channels with different nonlinearity on the layer switching process, a model switching method which is based on output change ratio of weak channels is proposed. And on research of the influence of model switching methods between the layers on the dynamic performances of the system, two other simplified model switching methods are proposed. Experimental results show the effectiveness of these methods.At last, the thesis is concluded with a summary and the problems which need further research are also discussed.