Optimal PI and PID Temperature Controls for a Dehydration Process

Abstract

In this work, the drying air temperature for an experimental platform of sliced tomatoes dehydration is optimally regulated by using the very well-known integral control approach which could be equivalent to the PI control when the output system is the whole state. In a first stage, the chosen controller is a PI because the plant step response does not present oscillations with low damping. However, obtained experimental results, applying this controller, present some oscillations in the closed-loop transient response. So, in a second stage, a discrete PID controller is applied and its gains are computed by using a numerical optimization algorithm based on the hill climbing method, which considers a quadratic performance index; the proposed procedure assumes an input delayed discrete model of the plant; therefore, the stability is verified by the closed-loop poles location. The reason to apply an optimal control strategy is motivated by the energy saving and the preserving of a desired closed-loop performance of the process. In order to contrast the advantages to use an optimal controller, the obtained experimental results are compared with a PID controller tuned with the improved Ziegler–Nichols rules, which is optimized only considering a criterion depending on the error (the energy consumption it is not involved). Moreover, the performance given by optimal PI control is also compared with an optimal state feedback control. The synthesized controllers (computed gains) are programmed in an industrial programmable logic controller Siemens S7-1200, and they are applied to a sliced tomato dehydration plant. According to the obtained results, when the optimally tuned PID controller is applied, it produces energy saving rates, more uniform drying conditions for the product and a better closed-loop plant performance (shorter settling time and nonovershoot, which are desired features in this type of process), with respect to other tested controllers. The presented experimental results by using the PI and PID optimized controllers allow to verify the feasibility of our proposal.