A Multi-Model Modeling Approach to MIMO Nonlinear Systems

For real industrial processes in which systems are multi-input multi-output (MIMO),nonlinear and large operating range,a new multi-model modeling approach is presented in this paper.Firstly,a set of scheduling variables is partitioned into c subsets by Satisfactory Fuzzy c-mean Clustering algorithm,then the multi-model system is identified by the identification algorithm based on weighted performance function.The resulted model can obtain good trade-off in terms of global fitting and local interpretation.For each local model,its valid area is got as well.The result of application with the modeling approach to typical pH process illustrates the performance of the proposed algorithm.