Prediction And Control Of Resources Optimization Based On Fuzzy Genetic Algorithm

As we known, the classic genetic algorithm does well on the optimization problem which is deterministic. However, there is still no in-depth study for the uncertain variables(such as fuzzy variables) optimization. Aiming at the problem of the optimal allocation of resources, we found that the deterministic result was hard to make it realized for many uncertain factors. In this paper, we focus on how to define and design the fuzzy genetic algorithm to solve the fuzzy optimization problem: through fuzzy individual definition, fuzzy seletion, crossover, mutation operators, fuzzy fitness fuction and fuzzy optimum, we expend the application of fuzzy optimization problem which is a hot issue nowadays and the key point of this paper.Since the classical genetic algorithm can not solve the fuzzy optimization problems,he traditional genetic algorithm is put forward to expendfor solving fuzzy optimization problems on the basis of the study. In view of the five genetic elements, the definition is as follows: first, we use the membership degree to make the input fuzzy; then, use the fuzzy neural network as the objective function calculation and make the result fuzzy, the crossover, mutation operator is introduced in the process of making the membership function to make the process fuzzy; at last output the fuzzy individual and its fuzzy fitness value.Compared to some of the so-called “fuzzy genetic algorithm” that they use fuzzy control rules to make the crossover and mutation probability change adaptively, or just make the individual fuzzy simply with the same parts of the classic, in this paper, according to the five genetic elements, we have the strict definition for the fuzzy individual, fuzzy fitness fuction, fuzzy operators and fuzzy output which is a true fuzzy genetic algorithm. And it is proved its validity in theory to be applied to the fuzzy optimization problem.In this paper, the data from our project team is used to predict and control the best allocation of resources: prediction is to predict the performance by different configuration which is realized by neural network; and the control is to seek the optimal configuration as the control standard of reference in the given performance(such as excellence). When we get the final result from the fuzzy genetic algorithm, it is compared to the accurate result so as to test the effectiveness of the fuzzy genetic algorithm. We realize the fuzzy genetic algorithm through Mat lab with experimenting and testing the best solution by 100 samples which are generated randomly. Finally, we get a satisfied result.