基于局部模型的多阶段在线产品质量预测

系统仿真学报 ›› 2016, Vol. 28 ›› Issue (4): 966-971.

基于局部模型的多阶段在线产品质量预测

李元¹, 燕亚运¹, 唐晓初²

1.沈阳化工大学信息工程学院,辽宁沈阳 110142;
2.沈阳航空航天大学自动化学院,辽宁沈阳 110136

收稿日期:2014-12-13 修回日期:2015-02-15 出版日期:2016-04-08 发布日期:2020-07-02
作者简介:李元(1964-),女,辽宁沈阳,教授,博士,研究方向为基于数据驱动复杂过程故障诊断;燕亚运(1990-),男,江苏泰州,硕士生,研究方向为质量预测与控制。
基金资助:
国家自然科学基金重大项目(61490701),国家自然科学基金(60774070, 61174119)

Online Product Quality Prediction for Multi-phase Based on Local Model

Li Yuan¹, Yan Yayun¹, Tang Xiaochu²

1. College of Information Engineering, Shenyang University of Chemical Technology, Shenyang 110142, China;
2. Shenyang Aerospace University, Automation College, Shenyang 110136, China

Received:2014-12-13 Revised:2015-02-15 Online:2016-04-08 Published:2020-07-02

摘要/Abstract

摘要： 针对间歇过程产品质量离线预测精度不高的问题,提出了一种基于局部模型的多阶段在线预测产品质量方法。利用间歇过程的周期重复性,采用重复因子将间歇过程划分为稳定阶段和过渡阶段。对稳定阶段采用相同相位时间片建立最小二乘支持向量机(LSSVM)模型,过渡阶段采用扩散距离(Diffusion Distance)选取最优子集建立LSSVM模型,使得当前样本的稳定阶段和过渡阶段的属性能分别和历史数据集的稳定阶段和过渡阶段的属性相似,预测出不易测量的产品质量。通过在Pensim仿真平台青霉素发酵过程中的应用表明,与整体离线预测相比,基于局部模型的多阶段在线预测方法有更好的预测性能。

关键词: 重复因子, 多阶段, 扩散距离, 最小二乘支持向量机, 局部预测模型

Abstract: For offline quality prediction accuracy for batch process, an online prediction method for multi-phase product quality was proposed based on the local model. According to the repetitive cycle of batch process, batch process could be divided into stable phase and transitional phase using the repeatability factor. The least squares support vector machine (LSSVM) model was established in stable phase using the time slice of same phase position, and the LSSVM model was established in transitional phase using optimal subset based on diffusion distance, which made the natural properties of current stable phase and transitional phase be similar to the natural properties of historical stable phase and transitional phase respectively, and the product quality which is difficult to be measured could be obtained. The application to penicillin fermentation process generated in Pensim simulation platform shows that the method based on multi-phase online prediction has better predictive performance than overall offline prediction.

Key words: repeatability factor, multi-phase, diffusion distance, least squares support vector machine, local prediction model

中图分类号:

TP29

李元, 燕亚运, 唐晓初. 基于局部模型的多阶段在线产品质量预测[J]. 系统仿真学报, 2016, 28(4): 966-971.

Li Yuan, Yan Yayun, Tang Xiaochu. Online Product Quality Prediction for Multi-phase Based on Local Model[J]. Journal of System Simulation, 2016, 28(4): 966-971.

参考文献

[1] 周东华, 李钢, 李元. 数据驱动的工业过程故障诊断技术 [M]. 北京: 科学出版社, 2011.
[2] P Nomikos, J F MacGregor. Monitoring batch processes using multiway principal component analysis[J]. AIChE Journal (S0001-1541), 1994, 40(8): 1361-1375.
[3] P Nomikos, J F MacGregor. Multi-way partial least square in monitoring batch processes[J]. Chemometrics and Intelligent Laboratory Systems (S0169-7439), 1995, 30(1): 97-108.
[4] Y Yao, F R Gao.A survey on multistage/multiphase statistic al modeling methods for batch processes[J]. Annual Reviews in Control (S1367-5788), 2009, 33(2): 172-183.
[5] S P Reinikainen, A Hoskuldsson.Multivariate statistical analysis of a multi-step industrial process[J]. Analytica Chimica Acta (S0003-2670), 2007, 595(1-2): 248-256.
[6] X T Doan, R Srinivasan.Online monitoring of multi-phase batch processing phase-based multivariate statistical process control[J]. Computers and Chemical Engineering (S0098-1354), 2008, 32(1): 230-243.
[7] N Y Lu, F R Gao.Stage-based process analysis and quality prediction for batch processes[J]. Industrial and Engineering Chemistry Research (S0888-5885), 2005, 44(10): 3547-3555.
[8] Y Yao, F R Gao.Phase and transition based batch process modeling and online monitoring[J]. Journal of Process Control (S0959-1524), 2009, 19(5): 816-826.
[9] Philippe Lauret, Eric Fock, Rija N.Randrianarivony. Bayesian neural network approach to short time load forecasting[J]. Energy Conversion and Management (S0196-8904), 2008, 49(5): 1156-1166.
[10] 张捷, 薄煜明. 基于神经网络预测的网络控制系统故障检测[J]. 南京理工大学学报, 2010, 34(1): 19-23.
[11] WW Hu, YC Huang, CP Li.Improved algorithm of muscle fatigue detection using linear regression analysis[J]. Electronics Letters (S0013-5194), 2013, 49(2): 89-91.
[12] 张宇山. 多元线性回归分析的实例研究[J]. 科技信息, 2009, 9: 54-56.
[13] Cai Yudong, Liu XiaoJun, Xu Xuebiao. Support vector machines for predicting HIV protease cleavage sites in protein[J]. Journal of Computational Chemistry (S1096-987X), 2002, 23(2): 267-274.
[14] 姜万录, 吴胜强. 基于SVM和证据理论的多数据融合故障诊断方法[J]. 仪器仪表学报, 2010, 31(8): 1738-1743.
[15] Cao Shugang, Liu Yanbao.A forecasting and forewarning model for methane hazard in working face of coal mine based on LS-SVM[J]. Journal of China University of Mining & Technology (S1006-1266), 2008, 18(2): 172-176.
[16] 赵清艳, 张超. 基于GA-LSSVR的铣削加工变形预测[J]. 组合机床与自动化加工技术, 2011, 12: 57-60.
[17] Huang Yixiang, Xuan F Zha.Discriminant diffusion maps analysis: A robust manifold learner for dimensionality reduction and its applications in machine condition monitoring and fault diagnosis[J]. Mechanical Systems and Signal Processing (S0888-3270), 2013, 34(1-2): 277-297.
[18] 胡刚, 刘侍刚. 基于扩散距离的SIFT特征匹配算法[J]. 计算机系统应用, 2012, 21(9): 92-96.
[19] Zhiqiang Ge, Luping Zhao.Utilizing transition information in online quality prediction of multiphase batch processes[J]. Journal of Process Control (S0959-1524), 2012, 22(3): 599-611.
[20] Zhiqiang Ge, Zhihua Song.Online montoring of nonlinear multiple mode processes based on adaptive localmodel approach[J]. Control Engineering Practice (S0967-0661), 2008, 16(12): 1427-1437.
[21] 刘毅, 王海清. Pensim仿真平台在青霉素发酵过程的应用研究[J]. 系统仿真学报, 2006, 18(12): 3524-3527.