基于改进蜂群算法的果酒总黄酮软测量模型  

Soft-sensor Model on Total Flavonoids in Wine Based on Improved Artificial Bee Colony Algorithm

作　　者：陈国超[1] 成新文[1] 

机构地区：[1]四川理工学院计算机学院,四川自贡643000

出　　处：《计算机测量与控制》2015年第3期723-726,共4页Computer Measurement &Control

基　　金：四川省重点实验室开放基金项目(NJ2011–09)

摘　　要：针对RBF神经网络和支持向量机对果酒总黄酮软测量过程中,存在速度慢和精度低的缺点,提出了基于改进人工蜂群算法的果酒总黄酮软测量模型;该模型利用混沌变量的遍历性和随机性特点,对标准人工蜂群算法进行改进,在种群初始化阶段引入混沌机制,确保个体分布的均匀性,并用信息素和灵敏度模型代替轮盘赌选择策略,使模型避免了过早收敛和提前停滞;仿真实验结果表明,该模型提高了果酒总黄酮软测量的精度,具有收敛速度快,抗噪性较强的特点,便于实现果酒总黄酮的在线测量。For RBF neural networks and support vector machines is weak at speed and precision in the soft--sensor process of total fla- vonoids in the wine, the soft--sensor model on total flavonoids in wine which based on improved artificial bee colony algorithm was proposed. The model which utilizes the ergodic and randomness characteristics of chaotic variable, improved the standard artificial bee colony algorithm, introduced chaos mechanism in population initialization phase to ensure distribution uniformity of the individual, replaced roulette selection model with pheromones and sensitivity model to avoid its premature convergence and stagnation in advance. The simulation results show that the model which has fast convergence and strong anti-- noise characteristics improved the soft-- sensor accuracy of total flavonoids in wine and facilitate the realization of its online test.

关 键 词：人工蜂群算法 最小二乘支持向量机 总黄酮 软测量 

分 类 号：TP39[自动化与计算机技术—计算机应用技术]