摘要
纱线的生产是一个多环节的复杂工业过程,其质量控制大多需要依赖领域专家的个人经验,为此提出一种基于支持向量机的纱线质量预测模型.探讨了模型选择与验证问题,并利用“网格搜索”法对模型参数进行了优化.试验结果表明,在小样本和“噪音”数据环境下,支持向量机模型仍能保持一定的预测精度,与人工神经网络模型相比,更适应于真实纺纱生产过程.
Yarn production is a multiple stage complex industrail process, and its quality control is heavly depended upon the domain expert＇s experience. An SVM model for predicting yarn properties is presented, and the model parameters are optimized with ＂grid-research＂ method. Experimental results show that under the real data sets and small population circumstances, SVM models are capable of maintaining the stability of predictive accuracy, and more suitable for noisy spinning process.
出处
《控制与决策》
EI
CSCD
北大核心
2007年第6期693-696,共4页
Control and Decision
基金
国家自然科学基金项目(70371040)
国家经贸委技术创新项目(02LJ-14-05-01)
关键词
支持向量机
统计学习
预测模型
人工神经网络
纺纱生产
Support vector machines
Statistical learning
Predictive model
Artificial neural networks
Spinning process
