Bio-inspired Algorithms for Modeling and Control of Underwater Flexible Single-Link Manipulator

Abstract

This research focuses on bio-inspired modeling and control system of an underwater flexible manipulator system (UFM). The dynamic behavior of the UFM was first modeled using system identification (SI) methods utilizing bio-inspired algorithms. The input–output data used for identification were acquired directly from a laboratory-sized UFM experimental rig developed earlier by the previous researcher. The models were developed using cuckoo search algorithm (CSA) and flower pollination algorithm (FPA) using parametric ARX model structured. For the controllers of the UFM, proportional-integral-derivative (PID) controllers were tuned using conventional heuristic and intelligent FPA methods. These algorithms were utilized to obtain the optimal values of controller parameters for trajectory tracking control of rigid-body motion of the UFM system. The PID controller is tuned offline based on the best identified SI model. The performance of these control schemes was analyzed via real-time PC-based control and observed in terms of trajectory tracking and error. The overall result of UFM described in this research revealed the superiority of the PID controllers tuned using bio-inspired flower pollination algorithm (FPA). It was found that the percentage of improvement achieved experimentally by the PID controller tuned by FPA indicates superiority compared to PID tuned heuristically with 45.6% improvement on overshoot and 66% improvement of MSE for negative pulse and 100% improvement on overshoot for positive pulse.