CIESC Journal ›› 2003, Vol. 54 ›› Issue (3): 327-332.
Previous Articles Next Articles
MENG Fanxu, YANG Bolun, YAO Ruiqing ; TAO Xianhu
Online:
Published:
孟凡旭;杨伯伦;姚瑞清;陶贤湖
Abstract: Hybrid evolutionary algorithms were developed. In this new method, the traditional genetic algorithm was modified by using adaptive multi-annealing crossover and mutation strategies instead of simple strategy. The multi-evolutionary mode was also adapted to improve search efficiency. The modified genetic algorithm avoided the problem of local optimum and showed the higher estimating precision and better convergence than the traditional genetic algorithm and Runge-Kutta method. This new hybrid evolutionary algorithm was successfully used to estimate the kinetics parameters for the synthesis of methyl tert-amyl ether from methanol and tert-amyl alcohol on strong acid cation exchange resin A-15. The calculated results using these parameters from the modified genetic algorithm agreed well with experimental data.
Key words: 遗传算法, 模拟退火算法, 进化算法, 反应动力学, 甲基叔戊基醚
遗传算法,
摘要: 将遗传算法与模拟退火算法相结合,以自适应多重退火交叉策略和自适应多重退火变异策略分别代替传统遗传算法中的单一交叉策略和单一变异策略,在此基础上进一步与进化策略相结合,开发出了一种多进化模式的混合进化算法. 用以上算法对以强酸性阳离子交换树脂A-15为催化剂用甲醇和叔戊醇为原料合成甲基叔戊基醚反应动力学方程中的速度常数、水的阻害系数等参数进行了求解计算,并与传统算法进行了比较. 结果表明:本研究所提出的混合进化算法收敛速度快,估算精度高,有效抑制了早熟现象的发生,所得到的有关动力学参数可靠,模型计算结果与实验结果吻合良好.
关键词: 遗传算法, 模拟退火算法, 进化算法, 反应动力学, 甲基叔戊基醚
MENG Fanxu, YANG Bolun, YAO Ruiqing , TAO Xianhu. KINETICS RESEARCH IN SYNTHESIS OF METHYL tert-AMYL ETHER USING HYBRID EVOLUTIONARY ALGORITHMS[J]. CIESC Journal, 2003, 54(3): 327-332.
孟凡旭, 杨伯伦, 姚瑞清, 陶贤湖. 混合进化算法研究甲基叔戊基醚合成反应动力学 [J]. 化工学报, 2003, 54(3): 327-332.
0 / / Recommend
Add to citation manager EndNote|Ris|BibTeX
URL: https://hgxb.cip.com.cn/EN/
https://hgxb.cip.com.cn/EN/Y2003/V54/I3/327
Soft sensor modeling based on differential evolution-particle swarm optimization based hybrid optimization algorithm
Effect of metal chlorides on glucose decomposition kinetics in high temperature liquid water
DUAN Donghong;SUN Yanping
Kinetic performance of BH-4 oxidation on Cu anode
Soft sensor modeling based on DE-LSSVM