Evolutionary algorithms for active vibration control of flexible manipulator

Abstract

Flexible manipulator systems offer numerous advantages over their rigid counterparts including light weight, faster system response, among others. However, unwanted vibration will occur when flexible manipulator is subjected to disturbances. If the advantages of flexible manipulator are not to be sacrificed, an accurate model and efficient control system must be developed. This thesis presents the development of a Proportional-Integral-Derivative (PID) controller tuning method using evolutionary algorithms (EA) for a single-link flexible manipulator system. Initially, a single link flexible manipulator rig, constrained to move in horizontal direction, was designed and fabricated. The input and output experimental data of the hub angle and endpoint acceleration of the flexible manipulator were acquired. The dynamics of the system was later modeled using a system identification (SI) method utilizing EA with linear auto regressive with exogenous (ARX) model structure. Two novel EAs, Genetic Algorithm with Parameter Exchanger (GAPE) and Particle Swarm Optimization with Explorer (PSOE) have been developed in this study by modifying the original Genetic Algorithm (GA) and Particle Swarm Optimization (PSO) algorithms. These novel algorithms were introduced for the identification of the flexible manipulator system. Their effectiveness was then evaluated in comparison to the original GA and PSO. Results indicated that the identification of the flexible manipulator system using PSOE is better compared to other methods. Next, PID controllers were tuned using EA for the input tracking and the endpoint vibration suppression of the flexible manipulator structure. For rigid motion control of hub angle, an auto-tuned PID controller was implemented. While for vibration suppression of the endpoint, several PID controllers were tuned using GA, GAPE, PSO and PSOE. The results have shown that the conventional auto-tuned PID was effective enough for the input tracking of the rigid motion. However, for end-point vibration suppression, the result showed the superiority of PID-PSOE in comparison to PID-GA, PID-GAPE and PID-PSO. The performance of the best simulated controller was validated experimentally later. Through experimental validation, it was found that the PID-PSOE was capable to suppress the vibration of the single-link flexible manipulator with highest attenuation of 31.3 dB at the first mode of the vibration. The outcomes of this research revealed the effectiveness of the PID controller tuned using PSOE for the endpoint vibration suppression of the flexible manipulator amongst other evolutionary methods.