多变异遗传算法优化生物脱硫反应动力学模型参数

Multimutation Genetic Algorithm to Optimize Kinetic Model Parameters of Biodesulfurization

由于常规遗传算法(SGA)的全局寻优效率不高,用于复杂的生物脱硫反应动力学模型参数优化时效果欠佳,为此设计了一种新的多变异遗传算法(MGA)以提高全局寻优效率。MGA的改进措施包括散射变异、微扰变异和单纯形变异各算子的设计,多变异操作实施方案的制定,选择操作和交叉操作方式的选择和改进等。Shaffer's F6函数和10维Alpine函数测试表明,与SGA相比,MGA的全局寻优效率大大提高。将MGA应用于红球菌DS-3脱除二苯并噻吩(DBT)的动力学模型参数优化,建立了更为准确的反应动力学模型。

Simple genetic algorithm （SGA） was a stochastic global optimization algorithm, however, due to its poor performance in local optimization and poor results to optimize kinetic model parameters of biodolesulfurization, a new multi-mutation genetic algorithm （MGA） was designed to improve the global optimization performance and local optimization. The improvement of the new method included designing scattering mutation operator, small perturbation operator and simplex method search mutation operator, establishing the course of performing multi-mutation operation, and choosing or improving available selection operation and crossover operation. Shaffer＇s F6 function and ten-dimensional Alpine function were applied to test MGA. The results demonstrated that the global optimization performance of MGA was superior to that of SGA. Furthermore, MGA was applied to optimize the kinetic model parameters of biodesulfurization of dibenzothiophen by Rhodococcus Sp DS-3, and a more accurate kinetic model was established.