Controle PID multivariável descentralizado: sintonia e aplicação prática

Abstract

This dissertation develops PI controllers (Proportional-Integral) and PID (Proportional-Integral-Dervativo) projects for SISO (Simple Input Simple Output) and MIMO (Multiple Input Multiple Output). The objective of this study is to combine the techniques of SISO and SISO PID controllers based on specification of gain and phase margins to synthesize a new design method of multivariable MIMO PID controllers with self-tuning or automatic tuning. Special emphasis is given to projects MIMO PID controllers using the relay method as frequency identification of the loops to be controlled. Two MIMO design methods are discussed in this work. These methods are applied in simulation systems and two practical applications: a process formed by neonatal incubator loops and a process of double attached tanks. These processes have strong interaction between control loops. The first method of multivariable design is a sequential design which the controllers are designed systematically and considering the interaction between loops for each step. This method is efficient and simple and has advantages such as: i) decentralized structure and sequential SISO design of control loops, ii) stability and robustness is guaranteed every step of the project; iii) the method is self-adjusting, therefore process knowledge is not required. The second method of multivariable controller design refers to a generalization of the Åström and Wittemark method, also known as critical point method for MIMO systems. The concept of critical surface and local characteristics for the cases of MIMO processes must be well established to design PID controllers with this approach. Many theoretical and practical aspects still need to be investigated in this method. Practical and theoretical aspects of the DRF method (Decentralized Relay Feedback) using nonlinearities like relay MIMO process are addressed through new solutions for multiple limit cycles. The present study showed that simulated and experimental applications seem to point to the advantages of decentralized PID controller design that utilizes the critical point method. The choice of the best method of project always depends on several aspects such as the robustness, stability and complexity, context and exhaust which are not the purpose of this study.