Online optimization of an engine controller by means of a genetic algorithm using history of search

In the present paper, online optimization of an engine controller by means of genetic algorithms (GA) is discussed. In optimization of real complex systems through experiments and computer simulation using random variables, optimization methods must cope with uncertainty of objective function and limitation of possible number of evaluation. Sano et al. (2000) proposed a GA utilizing history of search (GA with memory-based fitness evaluation: MFEGA) so as to reduce the number of fitness evaluation for such applications of GA. In the proposed method, the value of fitness function at a novel search point is estimated not only by the sampled fitness value at that point, but also by utilizing the fitness values of individuals stored in the history of search. In the present paper, this method is applied to online optimization of an engine controller for vehicles. Computer experiments using an engine simulator show that the proposed method outperforms conventional GAs both in convergence speed and accuracy of solution under fluctuation of fitness evaluation.