Search of optimum gas mixture ratio as gas insulating medium by genetic algorithm

摘要

Genetic algorithm is applied to find the optimum gas mixture ratio as gas insulating medium substituting pure SF/sub 6/. Genetic algorithm is very useful to find the optimum solution from vast searching possibilities. Testing each gas mixture by experiment requires a long time. The present method is very efficient to preselect the candidates of gas mixtures before more thorough but time-consuming investigation via experiment is carried out. The gas mixture ratio is coded as a series of bits simulating a genetic sequence of a life form. Two-term Boltzmann equation is used to calculate the effective ionization coefficient of each gas mixture that is used to evaluate the degree of adaptation of each individual representing one set of mixture ratios. Two types of degree of adaptation are used to evaluate each individual, the effective ionization coefficient at the critical ratio of the electric field to the gas density of SF/sub 6/ of 359.3/spl times/10/sup 21/Vm/sup 2/, and the global warming potential. Based on the degree of adaptation, better individuals can be selected as parents of the next generation, leaving their genes to future generations. After some generations, the group of individuals converges into the optimum with the best degree of adaptation.