An intelligent framework for modelling and active vibration control of flexible structures

Abstract

This thesis presents investigations into the development of an intelligent framework for modelling and active vibration control (AVC) of flexible structures. Dynamic characterisations of one-dimensional flexible beam and two-dimensional flexible plate structures are presented and simulation algorithms characterising the behaviour of each structure is developed using finite difference methods. The deflection of the structures in several modes, obtained from the simulation, has been validated by comparing these with previously reported theoretical work. Parametric and non-parametric modelling of such systems is investigated. Parametric approaches include linear parametric modelling of the system using recursive least squares (RLS) and genetic algorithms (GAS); and non-parametric approaches include multi-layered perceptron neural networks (MLP-NNs) and adaptive neuro-fuzzy inference systems (ANFIS) are employed. A comparative assessment of the techniques used is presented and discussed in terms of accuracy, efficiency and performance in estimating the modes of vibration of the system. Single-input single-output AVC systems using RLS, GAS, MLP-NN and ANFIS are developed and implemented within a flexible beam and flexible plate simulation environments to yield optimum cancellation of broadband vibration at an observation point on the structure. Two controller design formulations are proposed. The first controller design is formulated so as to allow on-line modeling, controller design and implementation and thus, yield a self-tuning control algorithm. Performance of the AVC algorithm is assessed based on parametric design techniques, using RLS and GAS, and non-parametric design techniques, using MLP-NN and ANFIS in the suppression of vibration of the flexible structures. The second controller design strategy is based on a cost function optimization using GAS. This approach bypasses modelling of the plant and results in direct estimation of the controller characteristics. Performance of this controller formulation is assessed in flexible beam and plate structures. The work is further extended to developing and integrating the idea of active control of flexible structures into an interactive learning environment. The environment is implemented in such a way that it allows the user to simulate and visualise behaviour of flexible structures with given physical characteristics, to test and validate controller designs, and furthermore, to execute such processes repeatedly in a friendly and easy manner. The simulation algorithm and interactive environment thus developed and validated form suitable test and verification platforms for the development of AVC strategies for flexible structures as well as for learning and research purposes.