Optimization of Technological Process by Experimental Design

Abstract

In the design of technological processes, i.e. in determining of the process parameters, selection of optimal parameters is important factor, regardless of the optimization criterion. If the first order experimental plans are used for finding the response function of measured variable (criterion function), problem is reduced to the determination of boundary values of individual parameters, which correspond to the extreme value of criterion function. Application of the second order experimental plans, however, makes certain difficulties, since, in this case, criterion functions are nonlinear (they usually consist of second order elements and interaction elements, besides linear elements), namely the saddle surface. Considering of many parameters (three, four and more), and many optimization criteria, creates even greater problem. The paper deals with the optimization procedure for such functions by analyzing different cases regarding the number of parameters and complexity of the model. The algorithm is tested on suitable practical examples. The article is accompanied with suitable graphs and diagrams.