Please use this identifier to cite or link to this item:
|Title:||STUDIES ON NONLINEAR OPTIMIZERS|
INTERIOR PENALTY FUNCTION METHOD
SEQUENTIAL EXPERIMENTAL DESIGN
|Abstract:||In the areal' of chemical engineering, two aspects are very common. Firstly, expressing the behaviour of a system in a mathematical form ( that is, model ) and secondly 'the system performance is. optimized for its known variables. For both 'the cases, one need to have optimizer which could minimize or maximize the function for these two aspects. The object of the present study was to provide the FORTRAN code of different optimizers (mainly non-linear optimizers ) to facilitate the department's software facilities. Three optimizers were selected namely, Complex method, Interior penalty function method and non-linear least ! square Marquardt lethod. Two objective functions in the area of chemical reaction ‘ngineering field namely, parameter estimation and sequential experimental design,were utilized to check their functioning and compare performance. Marquardt method showed superiority over Complex method for parameter estimation problem. Also, Marquardt method showed that it can utilize the worst starting point. However, Complex method could be used for design of experiment in comparison to Interior penalty function method. Also, Complex method showed a tendency to reach in vicinity of optimum solution in for fewer function evaluations. This behavior could be utilized to initiate the optimization and could be ovtakeri the superior optimizer to achieve the optimum solution,|
|Appears in Collections:||MASTERS' DISSERTATIONS (Chemical Eng)|
Items in DSpace are protected by copyright, with all rights reserved, unless otherwise indicated.