DNA Genetic Algorithm And Its Application In Parameter Estimation Of Fuel Cell Model

Genetic algorithm(GA) has strong search capability and it is simple, it can solve complex systems with a universal framework. It is a typical representative of the intelligent optimization algorithms. However, some disadvantages of GA such as the low search efficiency, the tendency to premature convergence and so on.DNA genetic algorithm(DNA-GA) is based on GA and DNA computing. It adopts DNA coding method and outperforms tradition GA with the help of DNA operators, whose technique can enhance the diversity of the population, update and improve the algorithm. Based on the previous algorithm research, the study is combined with the advantage of others algorithms to improve the DNA genetic algorithm, and the proposed algorithm is applied to the parameter estimation of fuel cell model. The main contents of the papers are as follows:1. Deep research to the DNA-GA is conducted based on the previous research. Comparisons of the effects of algorithm evolutionary rate and convergence rate of the crossover between different numbers of individuals of have been done. Three new crossover operators between several individuals are designed based on the DNA coding method to ensure the diversity of the population. Numerical experiments on four typical test functions are carried out to show the efficiency and effectiveness of the proposed algorithms. From the comparison of the results, the optimal algorithm is obtained when the number of cross individuals is three.2. To simplify the algorithm and improve the efficiency of the algorithm, we propose selective crossover and mutation operation DNA-GA. The algorithm only performs crossover or mutation operating in accordance with a certain probability and Crossover operates among a number of different types of individuals, mutation probability of mutation changes with genetic algebra. Four test function’s optimization results demonstrate the effectiveness of the algorithm.3. Inspired by DNA genetic algorithm’s multi individual crossover operation, we proposed an improved multi-group DNA-GA. Each sub-population inherited to the next generation independently according to different emphases, when populations evolve to certain algebra, we can exchange information between populations, and use population convergence strategy to fusion subdivided populations, therefore the algorithm’s global search ability and local search capabilities are improved. Test functions’ results show the superiority of the proposed multi-group DNA-GA.4. The accurate model is of grout significance for the simulation and design of fuel cell system. The selective crossover and mutation operation DNA-GA and the improved multi-group DNA-GA are applied to the parameter estimation of PEMFC. Analyze the performance of the proposed algorithm in different authentication policies based on the relationship between the given PEMFC output voltage and load current. By contrast with the existing four intelligent algorithms, the feasibility and availability of the two proposed algorithms in the optimizing of PEMFC model parameter estimation are proved. The improved multi-group DNA-GA is applied to the parameter estimation of SOFC model, Analyze and compare the voltage- load current characteristics under different temperature conditions, the results are compared with the existing algorithms and the effectiveness of the proposed algorithm are derived.