DNA Genetic Algorithms And Applications In Chemical Processes

Being stochastic global optimization technique, Genetic algorithms (GAs) simulate the natural biological evolution mechanism and can provide a problem solving complex system of general framework without depending on the specific problem area. Furthermore, duto to GAs has a good global search ability, it has been widely employed in many science research and engineering applications. However, GA still has some defects, such as weak local search ability, high similarity of the population individual in later evolution, easy premature convergence and so on.DNA genetic algorithms (DNA-GAs) develops based on genetic algorithm and inspired by biological DNA and DNA computing, which enormously enrichs genetic operations and provides a new way for the further development of the genetic algorithms. This thesis has relatively deep research to the DNA-GAs, and the main contents are as follows.(1) The DNA genetic algorithm with parameters interval adaptive strategy is proposed. In this algorithm, three kinds of crossover operators are designed to increase the population diversity. Moreover, in order to enhance the local search ability and overcome the traditional genetic algorithm shortcoming that initial solution value ranges are set on experience, the parameters interval adaptive strategy is applied. This algorithm is adopted to solve the test function optimization problems and the parameter estimation of the heavy oil cracking modeling problem, and the experiment results verify the effectiveness of the proposed algorithm.(2) Inspired by the evolutionary strategy and biological DNA, a hybrid DNA genetic algorithm based on (μ,λ) evolutionary strategy is proposed. The algorithm uses the DNA base coding method. Inspired by the characteristics of DNA molecular manipulation, new crossover operators and self-adaption mutation operator are designed. Moveover, the selection strategy in population evolutionary process is improved based on the simulated annealing. In order to improve the algorithm global and local search ability, the population update operation based on (μ,λ) evolutionary strategy is present and the the parameters interval adaptive strategy is applied. The algorithm is adopted to find optimal solutions of test functions and dynamic system parameter estimation problems, and the results all illustrate the satisfactory effectiveness of the proposed algorithm.. (3) Gasoline-blending process is an important link in refinery production which directly affects the enterprise profit margin and relates to environmental protection and energy saving. This problem is a nonlinear optimization problem with constraints. A DNA genetic algorithm with special ring operations is proposed and a new penalty function method is used to solve the constraints problem. The algorithm is applied to find optimization of test functions and the gasoline-blending scheduling problem, and the results verified the proposed algorithm effectiveness.(4) A DNA genetic algorithm with population taboo eliminate strategy is proposed. The algorithm comes from the idea of mean fitness value to evaluate premature degree and uses the value of premature degree to adaptively improve implementation probability of the genetic operation including the corossover operation and mutation operation. The algorithm is used to solve the optimization problem of the neuro-sliding mode controller. The experiment results demonstrats that the optimal neuro-sliding mode controller with the proposed algorithm has better control performance and robustness.