摘要
为提高多模过程故障检测的准确率,提出改进引力搜索算法-核主元分析邻域建模的故障检测方法。首先应用及时学习算法在参考数据集中找到待检数据的相关数据,再将相关数据和待检数据作为核主元分析检测模型的输入进行故障检测。核主元分析模型中的参数对故障检测性能有较大影响,提出改进引力搜索算法对模型中参数进行优化,提高检测性能。将所提方法应用于青霉素多模过程进行实验验证,仿真结果表明所提方法在多模过程故障检测中用时短、准确率高。
In order to improve multimode process fault detection low accuracy,an ensemble method called improved gravitational search algorithm-kernel principal component analysis（ IGSA-KPCA） neighborhood modeling is proposed. Firstly,the related data is found in reference data sets by using just in time learning（ JITL） approach,then the related data is set and current data are used as inputs of the KPCA model. KPCA model parameters have great influence on fault detection performance and improved GSA is put forw ard to optimize the KPCA model parameters,which improves fault detection performance. Finally,the proposed method is applied to penicillin multimode process and the simulation results show that IGSA-KPCA neighborhood modeling method is better than traditional method for multimode process fault detection with fast and high accuracy.
出处
《沈阳理工大学学报》
CAS
2016年第1期22-26,共5页
Journal of Shenyang Ligong University
基金
辽宁省教育厅科学研究基金资助项目(L2014083)
