Intelligent MIMO ORFBLS-Based Setpoint Tracking Control with Its Application to Temperature Control of an Industrial Extrusion Barrel

This paper presents a novel intelligent control method using an output recurrent fuzzy broad learning system (ORFBLS) for robust setpoint tracking control of nonlinear digital multi-input multi-output (MIMO) time-delay dynamic systems and one real industrial extrusion barrel, in order to effectively adapt to changing setpoints and exogenous disturbances. The weighting parameters of the used ORFBLS controller are iteratively updated using the deepest gradient descent algorithm to recursively minimize the quadratic form of tracking errors, and its closed-loop stability is well analyzed by establishing a sufficient inequality condition of a learning rate. The effectiveness, superiority, and applicability of the proposed controller are well demonstrated by conducting three comparative simulations and experimental results on a real extrusion barrel in a plastic injection molding machine. These results indicate that the proposed MIMO ORFBLS control method works well with a better robust setpoint tracking performance and a better disturbance rejection.