Vol. 14 No. 3 (1999)

An optimization process could be performed through a priori specification of a set of conditions to be satisfed by the optimum design. By manipulating the optimality criteria, simple redesign equations can be derived. This research looks at mathematical criteria obtained from diferentiation of Lagrangian with respect to the design variabels. In deriving the optimality criterion and developing the algorithm, fill use is made of the knowledge of the behaviour of the constraint imposed on the structure. Scale factor is applied to a design which meets the optimality criteria. "A computer…

During its operation, Slide-Water-Gate structure, receive an excessive vibration due to resonance effect caused by water current. To avoid this undesirable effect, in this paper we propose the application of structural optimization method to modify the initial structure. The goal of this optimization is to maximize the first natural frequencies so that the resonance effect can be minimized and the weight of structure is remained constant. The optimization is conducted by using the ANSYS 5.4 finite element program package.

This paper presents a robust decentralized adaptive controller for robot manipulators. The proposed controller does not require special knowledge of the robot dynamics model. Each joint of the manipulator is controlled by using an adaptive PD feedback controller and an adaptive gain acceleration feedforward. An auxiliary input is introduced to improve the convergence rate and to improve the robustness to torque disturbances and to a class of unmodelled dynamics. Lyapunov-based Model Reference Adaptive Control is used to establish the global stability of the tracking errors. Experimental…

This paper presents an empirical result on the application of neural network controller in a precision straight motion system. As a controller, neural networks learn in two phase. In the first learning phase, neural network was doing inverse system identification. It is called general learning. In the second learning phase, neural network was applied as an adaptif controller which doing online learning. It is ca//ed special learning. Through these two learning phases, neural network shows a good performance for wide range operation with a minimum operator/human intervention.