ADAPTING TO NOISE
暂无分享,去创建一个
Abstract An Expert System for Optimising Combustion in Multiple Burner Furnaces and Boiler Plants and the simulation on which the system was developed are briefly described. The performance of the Expert System on the 108 burner furnace of a continuous annealing line for rolled steel is critically examined. A learning system which generates, tests and improves rules for optimising combustion in multiple burner installations, based on a genetic algorithm is fully described. The results of experiments to compare the performance of the two systems in dealing with “noisy” data are then fully reported. Different simulations of multiple burner furnaces were set up and the signals from them distorted using various noise levels. It was found that, although the performance of the Expert System may be superior when there is no noise, it deteriorates considerably and proportionally as the level of noise increases. The performance of the Genetic Learning System, on the other hand, is unvarying, no matter what the level of noise, and is not significantly different from that of the Expert System at the low and medium noise levels used, although it may be better at the high level. These results form the basis for the development of a classifier system for combustion control in multiple burner installations.
[1] Kenneth de Jong,et al. Adaptive System Design: A Genetic Approach , 1980, IEEE Trans. Syst. Man Cybern..
[2] John H. Holland,et al. COGNITIVE SYSTEMS BASED ON ADAPTIVE ALGORITHMS1 , 1978 .
[3] E. J. Petit,et al. Analysis of the performance of a genetic algorithm-based system for message classification in noisy environments , 1987 .
[4] John H. Holland,et al. Escaping brittleness: the possibilities of general-purpose learning algorithms applied to parallel rule-based systems , 1995 .