Multi-criteria Genetic Algorithm Optimization Approach for Balancing Bicopter Control

Abstract Efforts to find the best controller for stabilizing a flying vehicle have been widely developed in recent years. This work aims to join the development track by exploring the use of genetic algorithm (GA) to find the most optimum controller parameter considering settling time and overshoot criteria. Roll attitude control of a balancing bicopter was chosen as the applied plant to confirm the designed method. The controller structure applied in this research is an ideal proportional-integral (PI) controller. Two equations considering the highlighted criteria for the objective function of the GA were developed with one of them involving integral absolute error (IAE) (MCIAE-GA), and the other one does not (MC-GA). Two GAs with both equations were compared to ziegler-nichols and the one with a standalone IAE (IAE-GA). The test results confirm that the proposed MC-GA and MCIAE-GA outperform the comparative controllers at their best performance by keeping the settling time as small as possible without overshooting. Despite the latter showing a slightly faster settling time, the former has better repeatability given a smaller standard deviation for Kp and Ki parameters with a mere 0.0820 and 0.0980, respectively.