摘要
以钢铁产品为例,在分析多工序多阶段产品质量预测控制特点的基础上,建立了多控制点递阶SVM预测控制模型,在模型的求解过程中,提出了基于粗集理论和主成分分析法的数据预处理与模型简化,并利用带约束的PSO算法分别优化了SVM的核超参数和相关影响因素的决策范围,实现了多阶段产品质量预测和相关过程参数的全局优化,为生产过程的质量改进提供了科学的决策依据。
Take steel product for an example, based on the analysis of quality predictive control characteristics of multi - phased production, a multi - control point hierarchical SVM predictive control model is established. During the solving, a data pre -processing method based on rough set theory and PCA method is proposed. A constrained PSO algorithm is used for optimizing the hyper kernel parameters of SVM and process parameters. The prediction of the muhi - phased products and the global optimization of its related process parameters are realized. Thus, some scientific evidences for production process quality improvement is offered with the method.
出处
《大连民族学院学报》
CAS
2013年第1期37-41,共5页
Journal of Dalian Nationalities University
基金
中央高校基本科研业务费专项资金资助项目(DC12010201)
中央高校博士启动资金资助项目(20106204)
关键词
预测控制
过程参数优化
支持向量机
递阶模型
微粒群算法
predictive control
process parameter optimization
support vector machine （ SVM ）
hierarchical model
particle swarm optimization （PS0）
