摘要
在基于数据驱动理论的软测量建模过程中,样本的数量会对所建模型的精度产生影响.针对训练样本数量较少的情况,提出了一种利用欧氏距离和角度原则进行数据扩充的方法.该方法通过分析数据的分布特征来确定样本扩充的区间范围,利用扩充数据完善建模对象在各阶段的信息,并进一步重构建模数据集,从而提升了所建模型的预测精度.通过2个工业过程的仿真研究,验证了该方法具有良好的泛化性能和建模精度.
In the data-driven based soft sensor modeling procedure, the number of data samples has an apparent affect on the model accuracy. In the case of a small number of training samples, a method of data expansion combining Euclidean distance and angle principle is proposed. This method can determine the range of the sample expansion by analyzing the distribution characteristics of data, and the process information of modeling plant in each stage is improved by the extended data-set. By reconstructing the modeling data sets, the prediction performance of the model is improved. The simulation results of different industrial processes have indicated that the proposed method has good prediction accuracy and generalization performance in the case of less number of samples.
作者
毕略
熊伟丽
BI Lve;XIONG Wei-li(School of Internet of Things Engineering, Jiangnan University, Wuxi 214122, China;Key Laboratory of Advanced Process Control for Light Industry of Ministry of Education, Jiangnan University, Wuxi 214122, China)
出处
《控制工程》
CSCD
北大核心
2019年第7期1431-1436,共6页
Control Engineering of China
基金
国家自然科学基金项目(21206053,21276111)
江苏省“六大人才高峰”计划资助(2013-DZXX-043)
江苏高校优势学科建设工程资助项目(PAPD)
关键词
数据分布特征
样本扩充
相似度准则
软测量
Data distribution feature
sample expansion
similarity criterion
soft sensor
