Study On DNA-Based Hybrid Genetic Algorithm

In the mass production today, there are a lot of NP problems, function optimization, which is one of the most typical problems, has a wide range of applications in the industrial production, engineering construction, and other fields, there is of great significance if find a way to solve the function optimization problem fast and accurate. The local optimal solution is the optimal solution of function in a certain field, the complex of function optimization is, there are not only global optimal solution, but also many local optimal solutions similar to the global optimal solution. It may get one of the local optimal solution and could not do further search to the solution space if use either the traditional mathematics based on gradient algorithm or the direct search algorithm to solve the function optimization problem.Genetic algorithm is a kind of the evolutionary algorithms which simulate the evolution process, based on population, the search of genetic algorithm has parallel characteristic, which not only gives genetic algorithm very strong global search ability, but also keep the diversity of population. Genetic algorithm has birth defects, however, the algorithm also has very strong robustness, and it can serve as an algorithm framework which can add other optimization algorithm or their ideas to make up for the defects. By constructing fitness function as the standard of evolution, and encoding the to build the relationship between individual and decision variables, genetic algorithm can search adaptively without depending on the problem characteristics, it has very strong application space.In this paper, it gives an improved genetic algorithm, which use the unified architecture of genetic algorithm, to which join the DNA code and ideas, constructs the hybrid genetic algorithm, which is used to solve unconstrained function optimization problems and constrained function optimization problem. With the analysis of the two types of problems to build the model, in view of their respective characteristics, adopt different genetic operators and evolution strategy, ensure high efficiency and high precision of the algorithm. This paper mainly studies the following aspects:1) As for unconstrained function optimization problems, in this chapter, on the basis of genetic algorithm, it introduces DNA coding and DNA replication technology to construct the hybrid genetic algorithm based on DNA, which is called DNA-BHGA. On the basis of the DNA code, the selection operator and the crossover operator are improved to enhance the global search ability of genetic algorithm, and the selective mutation operator is also improved to make up for the local search ability of genetic algorithm, with the adaptive niche algorithm, the diversity of population is guaranteed, all of those are to be sure that the algorithm has strong commonality. To test and verify the efficiency of algorithm, eight examples of general unconstrained optimization function are selected, and the result of the test shows DNA-BHGA is superior to the algorithm of the literature.2) As for constrained function optimization problems, with the analysis of the problem, it is supposed to be solved in the two steps, the constraint processing and optimization processing. For constraint handling method, by analyzing and comparing, the penalty function method is selected to deal with the constraints of the problem. For optimization processing, on the basis of DNA-BHGA which is designed for the solution of unconstrained function optimization, the genetic operators are changed based on the characteristics of the constrained optimization, to ensure the population diversity, one part of the population retaining and others reborn strategy is designed to instead of the niche algorithm. The test results of11international general constrained optimization algorithm example show that DNA-BHGA is superior to literature’s algorithms.