TSK Model-based Energy Efficient Control with Application to Temperature in Laboratory Fruits Dryer

Abstract

Most industrial plants are characterized by nonlinearity, model uncertainty, inertia and a heavy energy load. Energy efficiency becomes nowadays an important goal of their control. The present research presents a methodology for Takagi-Sugeno-Kang (TSK) modelling of complex nonlinear plants and design on their basis of fuzzy logic controllers using the principle of parallel distributed compensation (PDC). The methodology is illustrated for the energy efficient control of the temperature in a laboratory convective dryer. The TSK plant model is built on expert determined linearisation zones and types of dynamic models for each zone which parameters via genetic algorithms (GA) minimise the error between the plant responses to different step inputs from experiments and from model simulation. The model validation uses data from the online real time temperature PID control via an industrial programmable logic controller. Then a PDC is designed with local linear PI controllers. Their parameters via GA minimise the system dynamic error and control variance which are measures for high energy efficiency. The system simulations prove that the PDC outperforms other energy efficient controllers designed for a linear plant model or for an open loop pulse-relaxation control computed analytically from a-priory measurements.