Application Research On Multi-objectives Optimization Based On The Genetic Algorithm

Optimization problems are always been the subjects that people pay more attention to, especially in modern time that science and technology have a rapid development. In the application process of the engineering technique, people often need to study the optimization problems that have more than one objective which in given constraint condition, that is multi-objectives optimization problem. Because of those objectives are restricted and excluded by each other, so they can't reach the optimization results respectively. Genetic algorithm is the optimization method based on the biology evolution theory, because of the complexity in multi-objectives optimization problem and the advantage of global optimized researching in genetic algorithm, more and more researchers apply genetic algorithm in this field, and have got some results.Non-Dominated Sorting Genetic Algorithm (NSGA) and Elitist Non-Dominated Sorting Genetic Algorithm (NSGA-II) can get the optimization results which distributed symmetrically; these two methods have strong advantage in multi-objectives optimization field. In this paper, the basic theory of NSGA-II is studied, and more importantly, NSGA-II is resorted to solve some practical problems of multi-objective optimization. The main contents are as follows:(1) To overcome the normal problems of BP, such as it is easy to be trapped in local minima and its convergence speed is slow when error back propagation network training. A method is put forward to bring the concept of multi-objective optimization in back propagation network, and regard the output errors of BP network as the multi-objectives to minimize in parallel. Use NSGA-II to optimize the initial weights and threshold values of BP. The modeling result of a single input-double outputs system indicated the feasibility of NSGA-II combined with BP network. The simulation result indicated that the combination method would overcome the limitation of BP training and the times for back propagation network training could be decreased a lot.(2) To overcome the disadvantage that it's usually to determine the parameters of support vector machine by empirical experiment. An algorithm is put forward to bring the concept of multi-objective optimization in optimizing parameter of support vector machine, and regard the three parameters of support vector machine (width parameterσ, insensitive parameterε, penalty parameter C ) as the decision variables, the optimal objectives in reality application are used as the multi-objectives to optimize by NSGA-II. The useful parameters of support vector machine can be obtained by this method. Through the optimization process with goal constraint, the randomicity and empirical in parameter selection of support vector machine is overcome. The procedure of using this method in building the model of penicillin ferment process is given. The simulation result indicated that the method is validity. The model of penicillin ferment process has a good prediction result.