论文部分内容阅读
模拟退火是一种全局优化算法,具有跨越局部最优点的机制。最小一乘是一种较常用的最小二乘更为稳健的优化准则,更适用于可能偏离正态分布的实际数据集。本文探讨了用最小一乘为准则并利用模拟退火方法同时测定多组分体系的可能性。应用于2～3组分药物体系分析,获得满意的结果。本文还探讨了改变步长提高模拟退火算法优化精度的方法。
Simulated annealing is a global optimization algorithm with a mechanism to cross the local optimum. Least squares is a more robust least squares optimization criterion that is more suitable for actual data sets that may deviate from the normal distribution. This article explores the possibility of simultaneous determination of multicomponent systems using the least squares method and simulated annealing. Applied to 2 to 3 components of the drug system analysis, to obtain satisfactory results. This article also discusses the method of changing the step size to improve the precision of the simulated annealing algorithm.