Design of a Data-oriented 2DOF Nonlinear Controller

This study considers about the 2 degree-offreedom (2DOF) nonlinear controller. The proportionalintegral-derivative (PID) controllers have been widely used in many process systems. There are broadly-divided two design methods. One is a model-based controller design that tunes PID parameters based on a system model. The other is a dataoriented controller design that tunes these parameters by using obtained experimental data. Model-based controller design schemes are effective for designing a robust controller. On the other hand, data-oriented control design scheme can design a controller with a high regard for tracking property. In this research, a hybrid data-oriented 2DOF nonlinear controller is proposed. According to the scheme, a robust closed-loop control system is first constructed by prior information of a system. Then, a set of experimental data is obtained by the closedloop system, and a constructed feedforward cerebellar model articulation controller based proportional-derivative (CMACPD) compensator is learned by using the data in an offline manner. The method can guarantee a robustness of the closedloop system, and the CMAC-PD compensator can absorb nonlinearity of a system. The effectiveness of the proposed control system is evaluated by a simulation example.