Evolutionary swarm algorithm for modelling and control of horizontal flexible plate structures

Abstract

Numerous advantages offered by the horizontal flexible structure have attracted increasing industrial applications in many engineering fields particularly in the airport baggage conveyor system, micro hand surgery and semiconductor manufacturing industry. Nevertheless, the horizontal flexible structure is often subjected to disturbance forces as vibration is easily induced in the system. The vibration reduces the performance of the system, thus leading to the structure failure when excessive stress and noise prevail. Following this, it is crucial to minimize unwanted vibration so that the effectiveness and the lifetime of the structure can be preserved. In this thesis, an intelligent proportional-integral-derivative (PID) controller has been developed for vibration suppression of a horizontal flexible plate structure. Initially, a flexible plate experimental rig was designed and fabricated with all clamped edges boundary conditions at horizontal position. Then, the data acquisition and instrumentation systems were integrated into the experimental rig. Several experimental procedures were conducted to acquire the input-output vibration data of the system. Next, the dynamics of the system was modeled using linear auto regressive with exogenous, which is optimized with three types of evolutionary swarm algorithm, namely, the particle swarm optimization (PSO), artificial bee colony (ABC) and bat algorithm (BAT) model structure. Their effectiveness was then validated using mean squared error, correlation tests and pole zero diagram stability. Results showed that the PSO algorithm has superior performance compared to the other algorithms in modeling the system by achieving lowest mean squared error of 6103947.4 , correlation of up to 95 % confidence level and good stability. Next, five types of PID based controllers were chosen to suppress the unwanted vibration, namely, PID-Ziegler Nichols (ZN), PID-PSO, PID-ABC, Fuzzy-PID and PID-Iterative Learning Algorithm (ILA). The robustness of the controllers was validated by exerting different types of disturbances on the system. Amongst all controllers, the simulation results showed that PID tuned by ABC outperformed other controllers with 47.60 dB of attenuation level at the first mode (the dominant mode) of vibration, which is equivalent to 45.99 % of reduction in vibration amplitude. By implementing the controllers experimentally, the superiority of PID-ABC based controller was further verified by achieving an attenuation of 23.83 dB at the first mode of vibration and 21.62 % of reduction in vibration amplitude. This research proved that the PID controller tuned by ABC is superior compared to other tuning algorithms for vibration suppression of the horizontal flexible plate structure.