Browsing by Author "Yu, M. T."

Now showing 1 - 3 of 3

Active-set sequential quadratic programming method with compact neighbourhood algorithm for the multi-polygon mass production cutting-stock problem with rotatable polygons
(2009-09-01)
Yu, M. T.
;
Lin, T. Y.
;
Hung, C.
;
機械工程學系
;
Department of Mechanical Engineering
The cutting-stock problem, which considers how to arrange the component profiles on the material without overlaps, can increase the utility rate of the sheet stock, and is thus a standard constrained optimisation problem. In some applications the components should be placed with specific orientations, but in others the components may be placed with any orientation. This study presents an overlap index and it is much more suitable for the active-set SQP method which can reduce the time spend for constraint consideration. Using this method, various object orientations can be considered easily and the number of object on the sheet stock can be improved by up to eight percent. (C) 2009 Elsevier B.V. All rights reserved.
Novel optomechanical alignment method using a numerical optimization methodology
(2007-06-01)
Lin, Tsung Y.
;
Yu, M. T.
;
Liu, W. T.
;
機械工程學系
;
Department of Mechanical Engineering
Alignment of optomechanical components is important when designing and manufacturing optical systems. This study used an optical fiber as a case study to develop an alignment method. The core diameter of a single-mode fiber is about 9 mu m, and any slight misalignment or deformation of the optical mechanism will cause significant optical losses in connections. Previous studies have shown that the currently used alignment methods are not efficient, and the precise position for the connection is not easy to locate. This study proposes a two-stage method to overcome these problems. In the first stage, the Nelder-Mead simplex method is used to move quickly to the optimum solution. In the second stage, a numerical optimization method is used to improve the accuracy. This study compares different numerical optimization method that can be used to find the ideal connection position. It can be concluded that the most stable method for the search direction is the steepest-descent method, because the light intensity distribution is similar to a Gaussian one, and the most efficient method for the step-size determination is polynomial interpolation. Therefore, the second stage uses the steepest-descent method with polynomial interpolation. (c) 2007 Society of Photo-Optical Instrumentation Engineers.
Sequential quadratic programming method with a global search strategy on the cutting-stock problem with rotatable polygons
(2012-06-01)
Yu, M. T.
;
Lin, T. Y.
;
Hung, C.
;
機械工程學系
;
Department of Mechanical Engineering
The cutting-stock problem, which considers how to arrange the component profiles on the material without overlaps, can increase the utility rate of the sheet stock. It is thus a standard constrained optimization problem. In some applications the components should be placed with specific orientations, but in others the components may be placed with any orientation. In general, the methods used to solve the cutting-stock problem usually have global search strategies to improve the solution, such as the Genetic Algorithm and the Simulated Annealing Algorithm. Unfortunately, many parameters, such as the temperature and the cooling rate of the Simulated Annealing method and the mutation rate of the Genetic Algorithm, have to be set and different settings of these parameters will strongly affect the result. This study formulates the cutting-stock problem as an optimization problem and solves it by the SQP method. The proposed method will make it easy to consider different orientations of components. This study also presents a global search strategy for which the parameter setting is easy.