Identification of Nonlinear Multi-Input Multi-Output Systems Based on Maximum Likelihood Principle

Identification of multi-input multi-output (MIMO) industrial system with block structure is helpful to analyze the dynamic characteristics of the system, and beneficial to design the controller based on the model and the operation monitoring. Based on the identification strategy, the model of MIMO Hammerstein structure has a unique advantage to fit the dynamic input and output data of the process. A parameterized identification algorithm without increasing the number of parameters to be identified is proposed based on the maximum likelihood principle (MLC). Compared with the conventional parameterized identification method, the proposed identification algorithm can effectively separate the coupled nonlinear static parameters from the system dynamic parameters, and reduce the compute calculation amount of the identification algorithm and the variance of parameter estimation error. Besides, an adaptive forgetting factor is also constructed to improve the convergence rate and stability. Finally, the superiority and advantage of the proposed algorithm is verified by a chemical process.