Globally Convergent Fractional Order PID Tuning for AVR Systems Using Sequentially Niching Metaheuristics

Fractional order PID (FOPID) tuning has surfaced as a substitute controller for PIDs offering better regulatory system operation with an additional increase in estimation parameters. Over the years, the three main representative metaheuristics global optimisation techniques (Genetic algorithm, differential evolution and particle swarm optimisation) have been interchangeably used in optimal fractional order PID tuning problems. In order to further improve the gain estimation capabilities of the state-of-the-art metaheuristic candidates for FOPID tuning operations, the current research proposes the use of sequential niching search frameworks while estimating gain parameters. The main benefit of the approach is that while the classical metaheuristic approach leads to a very explorative unimodal search that often stalls to a potential improved solution, sequentially niching frameworks are evolutionary in nature, ban previously discovered good solutions and continuously change search directions indefinitely until eventually stopped, thus increasing the likelihood of obtaining better estimates over time. This search approach has been used to improve the performance index results of an Automatic voltage regulator (AVR) system yielding improved control performances than when using traditional metaheuristic frameworks.