Controle PID gaussiano com otimização dos parâmetros das funções gaussianas usando algoritmo genético e PSO

Abstract

This work proposes the use of a Gaussian adaptive PID control technique (GAPID) in order to increase the performance of the traditional PID control applied to a Buck converter. The Gaussian function used to define adaptive gains has characteristics such as; it is a smooth function with smooth derivatives, it has well defined lower and upper bounded and it has the adjustable concavity. Because it is a smooth function, it helps avoid problems related to abrupt gains transition, commonly found in other adaptive methods. However, there is no algebraic methodology to obtain the adaptive gains, since originally the GAPID parameter set consists of eight elements. Therefore, was used techniques such as optimization through bio-inspired metaheuristics, performance evaluation metrics, and change in the method to obtaining the settling-time, in order to increase the performance of this controller (GAPID) and obtain the adaptive gains. The use of the eight elements in the optimization generated optimized but very specialized solutions, causing the controller not to behave well when the operating conditions change. In this way, a link between the nonlinear parameters of the gaussian curves and the linear parameters of the PID controller was proposed, which demonstrated to generate solutions almost as good as with free and less specialized parameters, with a more homogeneous behavior in relation to changes in the operating point of the controller and bringing as a main advantage the use of the same traditional PID design requirements, which would facilitate the migration of PID controller to GAPID within most industries. The results obtained in both the simulation and the prototype were similar. This is due to careful modeling and rigor in design procedures, implemented in the same way in the model and the prototype.