摘要
针对催化裂化复杂反应体系 ,提出了一种在NLJ算法基础上改进的随机搜索优化算法NLJ +,该方法引入了优化的中间结果和目标函数的变化来修正搜索范围 ,从而减少了搜索的随机性。该方法应用于催化裂化十集总动力学模型轻循环油反应网络动力学参数估计 ,降低了搜索算法对初值和实验误差的要求 ,保证了收敛稳定性。
A modified direct search procedure, NLJ+, based on NLJ method utilizing pseudo random numbers over a region was presented to predict parameters of the FCC complex reaction system. The optimization procedure could reduce the size of search region according to the variations of the target function and the variations of the predicted parameters. The new method not only lowered the requirements to initial values, but also insured the convergence stability in using it to predict the parameters of the rate constants of light cycle oil in FCC 10 lumping kinetic model.
出处
《石油炼制与化工》
CAS
CSCD
北大核心
2004年第2期56-59,共4页
Petroleum Processing and Petrochemicals
