INTELLIGENT MODELLING OF GRADIENT FLEXIBLE PLATE STRUCTURE UTILISING HYBRID EVOLUTIONARY ALGORITHM

Muhammad Hasbollah, Hassan (2023) INTELLIGENT MODELLING OF GRADIENT FLEXIBLE PLATE STRUCTURE UTILISING HYBRID EVOLUTIONARY ALGORITHM. Masters thesis, Universiti Malaysia Sarawak.

PDF Intelligent Modelling of Gradient Flexible Plate Structure Utilising Hybrid Evolutionary Algorithm.pdf Download (6MB)

Abstract

The gradient flexible plate structure has been widely used in engineering industries. However, the gradient flexible plate is susceptible to vibrational disturbances and affecting its durability and performance over time. Hence, the unwanted vibration needs to be controlled and can be accomplished by developing an accurate model. Despite that, the accurate model is hard to be obtained especially in estimating the model parameters. Thus, the research presents the development of dynamic modelling for gradient flexible plate structure (GFPS). A slanted GFPS with orientation angle of 30° and all edges clamped was developed and fabricated to represent the actual dynamics of the system. Then, data acquisition and instrumentation system were integrated to the rig to collect the input-output vibration data. The research utilised parametric system identification based on autoregressive with exogenous input (ARX) model structure. First, evolutionary algorithms, namely particle swarm optimisation (PSO) and grey wolf optimisation (GWO) were used in developing GFPS dynamic model and their performances were compared. It was discovered that GWO model outperformed PSO model. However, the computational time of GWO is slower compared to PSO. Thus, a hybrid of grey wolf and particle swarm optimisation (GWO-PSO) were proposed to further improve the system modelling. It was found out that the hybrid GWO-PSO model outperformed PSO and GWO models by achieving the lowest mean squared error, correlation up to 95 % confidence level, and good stability. The obtained GWO-PSO models which is model order 2 and model order 4 were verified by using proportional-integral-derivative (PID) based controller. Their performances were measured in terms of model robustness based on vibration suppression. The final result confirmed that model order 2 of GWO-PSO is the optimum model to represent GFPS system modelling with 71.08% vibration attenuation.

Actions (For repository members only: login required)

View Item