PID controller optimization for a rotational inverted pendulum using genetic algorithm

Abstract

Rotational inverted pendulum (RIP) is widely used as a benchmark system in evaluating various control strategies, as it is an ideal test bed that represents under-actuated plants. Although a proportional-integral-derivative (PID) controller is popular and relatively simple in structure, its use in balancing the RIP is usually not recommended due to the difficulties in finding the suitable PID controller parameter values for the RIP system, which has two unstable open-loop poles. Therefore, this paper presents an investigation on a PID tuning problem for stabilizing the RIP using a variant of multi-objective Genetic Algorithm (GA), called the Global Criterion Genetic Algorithm (GCGA). Simulation work has shown that GCGA is capable in tuning the PID controller gains to balance the RIP. Compared to the standard single objective GA, the solutions found in GCGA have better control performances. The optimization results have then been applied to the real RIP in order to validate the results in the real-time environment.