基于IGBCA与SVM的质量预测  

Quality prediction model based on IGBCA and SVM

作　　者：李先飞 高琦[1,2] 高菲 LI Xianfei;GAO Qi;GAO Fei(School of Mechanical Engineering,Shandong University,Jinan 250061,CHN;Key Laboratory of High Efficiency and Clean Mechanical Manufacture,Shandong University,Jinan 250061,CHN)

机构地区：[1]山东大学机械工程学院,山东济南250061 [2]山东大学高效清洁机械制造教育部重点实验室,山东济南250061

出　　处：《制造技术与机床》2019年第12期128-132,共5页Manufacturing Technology & Machine Tool

基　　金：“十三五”装备预研领域基金项目(61409230102)

摘　　要：为了提高制造过程的质量预测精度和效率,提出了一种基于改进遗传蜂群算法(IGBCA)与支持向量机(SVM)的质量预测模型。首先,针对实际生产中,影响质量变动的因素具有多变量、复杂非线性、时变性等特点,选择支持向量机进行预测;然后针对现有的支持向量机参数选择方法存在精度不高、收敛速度慢的问题,提出了一种改进遗传蜂群算法对支持向量机的参数进行搜索寻优。经过实例对比分析得出,利用改进的遗传蜂群算法得到的支持向量机质量预测模型比利用遗传算法(GA)、人工蜂群算法(ABCA)得到的预测模型精度更高、收敛时间更短。In order to improve the accuracy and efficiency of the quality prediction in manufacturing process,a quality prediction model based on the improved genetic bee colony algorithm(IGBCA)and support vector machine(SVM)is proposed.Firstly,considering that the factors affecting the quality in the actual production are multivariable,nonlinear and time-varying,the support vector machine is selected to make the prediction.Then,aiming at the problems such as low accuracy and slow convergence rate in the existing parameter selection methods,the improved genetic bee colony algorithm is proposed to search the preferable parameters of SVM.After the analysis of the comparison test,we find the quality prediction model of SVM obtained through the improved genetic honeybee algorithm has higher accuracy and shorter convergence time than that obtained through genetic algorithm(GB)and artificial bee colony algorithm(ABCA).

关 键 词：质量预测 改进遗传蜂群算法 支持向量机 

分 类 号：TP391[自动化与计算机技术—计算机应用技术]