Dynamic modelling of a single-link flexible manipulator system: A particle swarm optimisation approach

This paper presents investigations into modelling of a single-link flexible manipulator system using the particle swarm optimisation (PSO) algorithm. PSO is a population-based search algorithm and is initialised with a population of random solutions, called particles. Basic PSO with best model can hardly provide suitable solutions in the case of real-world multimodal problems. In order to improve diversity in the population set and hence to improve the global searching capability a local version of PSO with time varying inertia and acceleration coefficients is proposed and used in this work. The effectiveness of the algorithm in modelling is validated and verified in terms of tracking, stability and the ability of the derived model in capturing a system's dynamics. The effect of swarm size on the convergence of the proposed PSO algorithm is also analysed. Time domain and frequency domain results of derived models clearly show the potential of the modelling technique and the proposed algorithm in solving such control problems.