Researching On The Parameters' Optimization In The Reduction Process Step Of Smelt Magnesium By Silicon-thermo-reduction

The magnesium and magnesium alloy as a result of own merit, obtained the widespread application in each domain, has brought the obvious technology economic efficiency. Builds up smelt magnesium by silicon-thermo-reduction reduction process step parameter to carry on the optimized design, may enhance the production efficiency, reduces the production cost, enhances the production benefit. Different from the traditional mathematics optimization, took an actual production question, relations very complex is unable to establish the explicit mathematical relation model, this had decided it uses the conventional method to realize with difficulty. This article in builds up smelt magnesium by silicon-thermo-reduction reduction process step in the foundation, builds up smelt magnesium by silicon-thermo-reduction reduction process step parameter to optimize this complex actual combination optimization question, by leaves the magnesium rate to take finally optimizes the goal, established based on the artificial neural networks and the genetic algorithms to build up smelt magnesium by silicon-thermo-reduction reduction process step parameter the optimized system, thus might obtain the most superior craft parameter combination.Analyzed smelt magnesium by silicon-thermo-reduction the basic principle, elaborated smelt magnesium by silicon-thermo-reduction thermodynamics principle, its return to reduction process step mechanism and macroscopic dynamics. Explained smelt magnesium by silicon-thermo-reduction the condition as well as these conditions which must achieve in the return to original state process to the return to original state process influences. Has determined the return to reduction process step optimized parameter.Artificial neural network extracts the domain knowledge from a large number of discrete experimental data after learning and training. The knowledge is put into the network joint weight and corresponding mathematic model could be set up. Accordingly the artificial neural network serves as the kernel algorithm.Genetic algorithm is an effective method of resolving the optimization problems. Its computational model simulates the heredity selection and natural elimination through selection or contest of Darwinian's biologic evolution. It remarkably characterizes in the simple and general usage, sounding robustness, qualified to parallel processing, high efficiency, etc. That's why to choose genetic algorithm to do the optimization.