Comparative study of tuning techniques for fractional PID controllers for FOPDT systems

• Autor(a) principal: Aguiar, Rodrigo Teixeira [UNESP]
• Data de Publicação: 2022
• Outros Autores: Lima, Bruno Silva de, Maia, Antônio Augusto Torres
• Tipo de documento: Artigo
• Idioma: eng
• Título da fonte: Repositório Institucional da UNESP
• Texto Completo: http://dx.doi.org/10.1007/s40430-022-03491-1; http://hdl.handle.net/11449/239946
• topic: Control systems; Fractional Calculus; Fractional PID controller; Fractional PID tuning methods

Traditionally, PID (proportional–integral–derivative) controllers are used in the process control field to keep the system output at a desirable value. Fractional-order PID controllers, however, tend to perform better than the traditional integer-order version due to characteristics like its nonlinear nature and the two extra tuning parameters that allow a better adjustment in the controller action. Similar to the integer controllers, there are different methods to tune fractional-order PID controllers. In this work, it is presented a comparative study of fractional PID controller tuning techniques applied to first-order plus dead-time plants (FOPDT). The equations are thoroughly detailed, and the algorithm used in the simulations is presented. Five tuning methods for fractional PIDs were found in the available literature and adjusted for being applicable to the problem under study. Two of these methods are analytical and the other three heuristics (genetic algorithm, differential evolution, and Nelder–Mead algorithm). The best methods found were the methods based on the use of genetic algorithm and Nelder–Mead simplex algorithm which presented values of the ITAE criterion at least two times lower when compared to the other methods. Also, robustness and disturbance rejection for each controller considered were analyzed, presenting adequate responses. Moreover, the step response curves represented overshoot values lower than 10% and a response time below 10 seconds, values considered satisfactory for the studied system.