Global search algorithm for the problem of optimization cascade of extractors
2013
Pevneva, A., National Mineral Resources Univ., Saint Petersburg (Russian Federation)
The article proposes the theoretical scheme of extreme computer simulation to determine the global maximum of the objective function. This is represented in the form of a step-by-step. As the strategy of building of experiment a random search algorithm with directional cone is selected. Modification of this algorithm is based on the principle of sifting ''unpromising '' trajectory search. Result of this modification can significantly reduce the amount of calculations. In addition, estimates are given for the empirical parameters of random search and some advice on their choice. As well as a comparative table of search results is displayed for different values of parameters. The basis of computer simulation is mathematical model of the cascade of extractors with recycle. It is the system of algebraic equations for the concentrations of substances extracted and balance equations. The criterion adopted to optimize present value of the plant. This value is linearly dependent on the cost of reagents and concentrations of extracted substances. After determining the approximate global optimization is performed by local refinement of this method (used known simplex method). The results of this computational problem clearly demonstrated the need for a global approach in the design of extreme experiment.
显示更多 [+] 显示较少 [-]