摘要
一般的萤火虫算法只能用于连续的实数空间,本文提出了改进的离散萤火虫算法,并结合轮盘赌算法避免算法陷入局部最优。提出的离散萤火虫算法用于近红外光谱的波长筛选,并应用于谷物的四种性质和汽油的辛烷值预测。由实验结果可知,提出的萤火虫算法能筛选到高信息变量,且收敛速度快。
Most Firefly algorithms（FA） are applied in continuous and real-number space. In this paper, the modified firefly algorithm（FA） is proposed and the roulette wheel algorithm is added in FA to avoid falling into local optimum. The proposed FA is used to select variables in NIR spectroscopy and to predict moisture, oil, protein and starch values for corn samples and gasoline octane value. It has been demonstrated that the modified FA can select informative variables and converges quickly towards the optimal solution.
作者
王志莹
石伟民
申琦
WANG Zhiying;SHI Weimin;SHEN Qi(College of Chemistry and Molecular Engineering,Zhengzhou University,Zhengzhou,450001,He＇nan,China)
出处
《计算机与应用化学》
CAS
北大核心
2018年第5期358-365,共8页
Computers and Applied Chemistry
基金
国家自然科学基金资助项目(21575131)
关键词
萤火虫算法
近红外光谱
变量筛选
偏最小二乘
Firefly algorithm
Near infrared spectroscopy
Variable selection
partial least-squares
