改进遗传算法及其在SO_2氧化反应动力学参数优化中的应用

An improved genetic algorithm with its application in SO_2 oxidation reaction kinetics parameters' optimization

针对化学反应动力学机理模型复杂、非线性等特点,本文提出了1种基于黄金分割的改进遗传算法。该算法主要将整个遗传进程分成3个阶段,前2段配备不同的改善种群多样性的选择交叉算子,最后采用改进的"最佳保存策略"加速收敛,这样能够有效避免经典遗传算法中出现的计算效率较低,收敛迟缓,容易早熟等问题。本文改进的算法依托MATLAB软件平台对二元多峰Schaffer函数进行了仿真性能测试,最后将其成功地应用到了SO2氧化反应动力学参数优化案例中,并与文献中的方法进行对比,结果表明,其在收敛精度(接近5×10^(-5))与收敛效率方面效果更好,而且这种算法的应用不依赖于案例的具体领域,可用于解决化工系统中类似的参数估计问题。

To deal with the problems of parameters for complex and nonlinear chemical reaction kinetics mechanism models,this paperpresents an improved genetic algorithm based on the golden section.The algorithm mainly separates the genetic process into threephases,different selective and crossed operators for the first two stages aiming at maintaining the whole population diversity and an improvedelitist strategy at last accelerating the convergence,which,may effectively avoid the existing problems of the classical genetic algorithm, like lowcomputation efficiency,slow convergence and easy to be premature and so on.In addition, we make a simulation performance test to thefunction of Schaffer depending on the platform of MATLAB, f＇mally,we found the better convergence accuracy （nearly 5x10^5） andconvergence efficiency effect after comparing with the methods in literatures when applied to the case of parameters optimization of SO2oxidation reaction kinetics,furthermore,this algorithm＇s application does not depend on the specific areas of the case,which makes it possible tosolve the similar problem of parameter estimation.