PhD Programme

 No Registration & Status
 1 20102011/1 -20132014/2
Main Supervisor
Lim Kian Sheng

Particle Swarm Optimisation based on Domination and Multi Leader Concept for Multi Objectives Problems

Multi Objective Optimisation (MOO) problem involves simultaneous minimization or maximization of many objective functions. Various MOO algorithms have been introduced to solve the MOO problem. One of these algorithms is Vector Evaluated Particle Swarm Optimization (VEPSO) algorithm. In VEPSO, each objective’s function is optimised by a swarm of particles under guidance of the best solution, known as leader from another swarm. However, this leader is updated only when the newly generated solutions has better fitness at the objective function optimised by that swarm, and ignore the other objective functions. In this study, an improved VEPSO algorithm, namely VEPSO incorporated non-dominated solution (VEPSOnds), is introduced by the use of non-dominated solution as a leader to improve the performance of VEPSO. Since the mechanism of VEPSO algorithm is based on only one leader from another swarm, particles will converge quickly towards the leader without consider if there are a lots of non-dominated solutions around it. This will result in premature convergence and unable to explore for better solutions. Subsequently, the VEPSOnds algorithm is further modified with multi leaders, namely VEPSO with multi leaders (VEPSOml), such that particles are guided towards multiple non-dominated solutions and further improve the performance of VEPSO. Besides, the concept of using multi leaders is implemented in another PSO-based MOO algorithm to improve its performance. This new PSO-based MOO algorithm is called Multi Leader Particle Swarm Optimisation (MLPSO). All improved algorithms are subjected to a series of numerical experiments based on fourteen test problems from Zitzler, Deb, and Thiele’s (ZDT) problems and Walking Fish Group’s (WFG) problems. The experimental results show that substantial improvements are achieved by the improved algorithms as compared to their predecessor algorithm. The VEPSOnds shows significant difference as compared to other MOO algorithms while VEPSOml and MLPSO do not show significant difference. In terms of ranking, the MLPSO has shown better in ranking than some MOO algorithms.

Main Supervisor

Mustapha Mohammed


A two wheeled inverted pendulum (TWIP) mobile robot is an under-actuated mechanical system, which is inherently open-loop unstable with highly nonlinear dynamics. Thus, an effective control is essential for its proper operation. Nonlinear systems such as TWIP mobile robot can be modeled and controlled using Takagi-Sugeno (T-S) fuzzy model-based design. This thesis is aimed at developing a new fuzzy model based controller for the balancing and velocity tracking control of a TWIP mobile robot, which had been achieved through four steps. Firstly, the nonlinear dynamical equations of motion of the TWIP mobile robot were derived using Kane’s method, based on which the T-S fuzzy model of the TWIP mobile robot was developed using local approximation techniques. Secondly, new LMI-based stabilization conditions for a class of continuous-time T-S fuzzy model-based control system were formulated. The stabilization conditions were derived using a fuzzy Lyapunov function and a non-parallel distributed compensation (non-PDC) control law. Next, the stabilization conditions were modified to include constraints on the closed-loop poles location, and then a fuzzy state feedback controller for the continuous-time T-S fuzzy model-based control system which guarantees global stability for a desired transient performance was designed. Finally, based on the developed T-S fuzzy model of the TWIP mobile robot and the designed fuzzy state feedback controller, a velocity tracking controller was proposed for balancing and velocity tracking control of the TWIP mobile robot. The performance of the proposed controller was investigated via simulation, and the results were compared to those from linear and nonlinear controllers. The results showed that the proposed controller outperform the other controllers in terms of robustness and transient performance, and require less controlled input signal compared to other controllers.

Main Supervisor
Amir Abdullahi Bature





A Two Wheeled Inverted Pendulum (TWIP) mobile robot is a mechanical system with three degrees of freedom but has only two wheels driven by DC motors to control all the degrees of motion. This makes it an under-actuated system. It is also highly nonlinear, besides being an open-loop unstable system, thus making the automated control of the robot quite challenging. This thesis aims at designing a robust Composite Nonlinear Feedback (CNF) control for balancing and position tracking of the TWIP mobile robot, which has been achieved through four steps. Firstly, the dynamic equations describing the nonlinear robot were derived by using Euler-Lagrange co-ordinate equation method, and the nonlinear model is then linearized. Secondly, the linear model is further refined by using grey box identification technique in order to improve the approximation of the linear model to the nonlinear dynamic model. Next, based on the model, a new approach of CNF design is used for balancing and position control of the robot. The linear part of the CNF was designed using Linear Matrix Inequality (LMI) pole placement technique to ensure global stability for a desired transient response. Whereas the nonlinear part of the CNF controller’s output feedback of the tilt angle is biased with optimized weight to give the robot tilt angle concern not only in the linear part, but in the nonlinear part of the CNF as well. Finally, Integral Sliding Mode (ISM) control technique is used to make the proposed scheme robust to input disturbance and parameters uncertainties. The performance of the proposed controller was then investigated via simulation, and the results were compared to those controlled by a linear controller, nonlinear controller and conventional CNF controller. The results show that the proposed controller shows better performance in terms of balancing with similar transient response with other controllers. 

20112012/1 -20152016/1 (completed)

Mohd Fadzli Haniff

Power-Comfort Optimized Scheduling of Air Conditioning System

 520112012/2  (Active) 

Nor Muzakkir NorAyob@Nordin

Development of Ultrasonic Tomography System for Two-Phase Flow Measurement in a Steel Column

Master By Research Programme
 NO Registration & Status
 120082009/1- 2
(Completed 29 Jan10)
Noor Khafifah bt Khalid DNA sequence design based on Binary Vector Evaluated Particle Swarm Optimization
 2 20092010/2 (completed)
Husnaini bt Azmy Target selection using single electroencephalogram electrode based on mental task
 320112012/2 (completed)
Mohd Fahajumi Jumaah Optical Tomography using lens for Concentration Profile Measurement.