Evolutionary algorithms for self-tuning Active Vibration Control of flexible beam

Abstract

This paper presents the development of self tuning Active Vibration Control (AVC) strategy for flexible beam structure. An experimental procedure was conducted on a flexible beam structure with clamped-free boundary condition. The beam was forced to vibrate using an external force and a set of input-output vibration data was acquired. Using the input-output data, the flexible beam model was developed using Least Squares (LS) algorithm that incorporated the Auto Regressive (ARX) model structure. The AVC controllers developed are proportional-derivative (PD) and proportionalintegral-derivative (PID). The parameters of PD and PID controllers were tuned using iterative learning algorithm (ILA) and evolutionary Particle Swarm Optimization (PSO) techniques. Mean squared errors (MSE) were used to compare PSO tuned PD (PD-PSO), PSO tuned PID (PID-PSO) and PID with ILA (PID-ILA) controllers. It was found that the PID-ILA controller tuned using ILA had performed better than PID-PSO but PD-PSO is the best among the three controllers.