改进主元分析方法及数据重构在工业系统中的故障诊断研究
|
摘要
为了更加准确地对复杂工业生产系统进行故障判断,使生产系统更加稳定地运行,采用了改进的主元分析(Principal component analysis,PCA)方法及数据重构对工业过程进行故障诊断研究。采集工业系统正常和故障状态时的数据,将传统的PCA算法中平方预测误差(Squared prediction error,SPE)统计量分成两个,分别为主元显著关联的检测残差变量(Principal-component-related variable residual,PVR)和一般变量残差(Common variable residual,CVR)对系统进行故障判断。为了使系统在检测出故障之后尽量减少故障数据对系统的影响,又进一步应用数据重构方法,将故障数据重构成正常数据,并采用有效度指标进行验证。在故障发生的过程中对故障部分进行检修和排除,把生产系统受到故障的影响降到最低。改进的PCA方法和数据重构方法运用田纳西—伊斯曼过程的数据验证,使故障的检测结果更加准确,保证了生产系统的正常运行行。
Not only in order to determine the fault more accurately in the industrial system,but also in order to make the production system operation more stable,the improved principal component analysis method and data reconstruction method is used in the industrial process. The data of the normal and fault state of industrial system are collected,the SPE statistics of the traditional principal component analysis is divided into principal-component-related variable residual(PVR)and common variable residua(CVR),which are used to diagnose the system. In order to minimize the impact of the failure data on the system after detecting the failure,the data reconstruction method is further applied. The failure data are reconstituted into normal data,and the validity index is used to verify. When the fault happenes,the fault is repaired and excluded,and the failure impact on the production system is minimized. In order to verify the diagnosis method,the method is applied to the data of the Tennessee-Eastman system,the detection result of the fault is more precise,and the normal production system is ensured to work.
Not only in order to determine the fault more accurately in the industrial system,but also in order to make the production system operation more stable,the improved principal component analysis method and data reconstruction method is used in the industrial process. The data of the normal and fault state of industrial system are collected,the SPE statistics of the traditional principal component analysis is divided into principal-component-related variable residual(PVR)and common variable residua(CVR),which are used to diagnose the system. In order to minimize the impact of the failure data on the system after detecting the failure,the data reconstruction method is further applied. The failure data are reconstituted into normal data,and the validity index is used to verify. When the fault happenes,the fault is repaired and excluded,and the failure impact on the production system is minimized. In order to verify the diagnosis method,the method is applied to the data of the Tennessee-Eastman system,the detection result of the fault is more precise,and the normal production system is ensured to work.
引文
[1] Wang Yifei,Ma Xiandong,Joyce M J. Reducing sensor complexity for monitoring wind turbine performance using principal component analysis[J]. Renewable Energy,2016,97:444-456.
[2] Gao Xin,Hou Jian. An improved SVM integrated GS-PCA fault diagnosis approach of Tennessee Eastman process[J]. Neurocomputing,2016,174(22):906-911.
[3] Xu Xianzhen,Xie Lei,Wang Shuqing. Multimode process monitoring with PCA mixture model[J]. Computers & Electrical Engineering,2014,40(7):2101-2112.
[4] Zhou Bo,Ye Hao,Zhang Haifeng,et al. Process monitoring of iron-making process in a blast furnace with PCA-based methods[J]. Control Engineering Practice,2016:47:1-14.
[5] Pan Yijun,Yang Chunjie,An Ruqiao,et al. Fault detection with improved principal component pursuit method[J]. Chemometrics and Intelligent Laboratory Systems,2016:157:111-119.
[6] 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,2011:23-27,58-59.
[7] 肖应旺,徐保国. 改进PCA在发酵过程监测与故障诊断中的应用[J]. 控制与决策,2005,20(5):571-574.Xiao Yingwang,Xu Baoguo. Application of improved PCA to fermentation process monitoring and fault diagnosis[J]. Control and Decision,2005,20(5):571-574.
[8] 张新荣,熊伟丽,徐保国. 基于Q统计量分量的故障检测算法研究[J]. 计算机与应用化学,2008.Zhang Xinrong,Xiong Weili,Xu Baoguo. A study of approach on fault detection based on Q statistic separation[J]. Computers and Applied Chemistry,2008,25(12):1537-1542.
[9] 王海清,宋执环,李平. 改进PCA及其在过程监测与故障诊断中的应用[J]. 化工学报,2001,52(6):471-475.Wang Haiqing,Song Zhihuan,Li Ping. Improved PCA with application to process monitoring and fault diagnosis[J]. Journal of Chemical Industry and Engineering,2001,52(6):471-475.
[10] 赵忠盖,刘飞,徐保国. 基于改进混合概率主元分析模型的过程监控[J]. 控制与决策,2006,21(7):745-749.Zhao Zhonggai,Liu Fei,Xu Baoguo. Process monitoring based on improved mixture probabilistic principal component analysis model[J]. Control and Decision,2006,21(7):745-749.
[11] 徐涛,王祁. PCA在火箭发动机试车台传感器故障诊断中的应用[J]. 南京理工大学学报,2006,30(6):669-672.Xu Tao,Wang Qi. Application of PCA in sensor fault diagnosis of rocket engine ground testing bed[J]. Journal of Nanjing University of Science and Technology,2006,30(6):669-672.
[12] 成成,黄道. IPCA在故障检测与分离中的应用[J]. 华东理工大学学报,2000:502-506.Cheng Cheng,Huang Dao. Fault detection and diagnosis by integrated principal component analysis[J]. Journal of East China University of Science and Technology,2000:502-506.
[2] Gao Xin,Hou Jian. An improved SVM integrated GS-PCA fault diagnosis approach of Tennessee Eastman process[J]. Neurocomputing,2016,174(22):906-911.
[3] Xu Xianzhen,Xie Lei,Wang Shuqing. Multimode process monitoring with PCA mixture model[J]. Computers & Electrical Engineering,2014,40(7):2101-2112.
[4] Zhou Bo,Ye Hao,Zhang Haifeng,et al. Process monitoring of iron-making process in a blast furnace with PCA-based methods[J]. Control Engineering Practice,2016:47:1-14.
[5] Pan Yijun,Yang Chunjie,An Ruqiao,et al. Fault detection with improved principal component pursuit method[J]. Chemometrics and Intelligent Laboratory Systems,2016:157:111-119.
[6] 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,2011:23-27,58-59.
[7] 肖应旺,徐保国. 改进PCA在发酵过程监测与故障诊断中的应用[J]. 控制与决策,2005,20(5):571-574.Xiao Yingwang,Xu Baoguo. Application of improved PCA to fermentation process monitoring and fault diagnosis[J]. Control and Decision,2005,20(5):571-574.
[8] 张新荣,熊伟丽,徐保国. 基于Q统计量分量的故障检测算法研究[J]. 计算机与应用化学,2008.Zhang Xinrong,Xiong Weili,Xu Baoguo. A study of approach on fault detection based on Q statistic separation[J]. Computers and Applied Chemistry,2008,25(12):1537-1542.
[9] 王海清,宋执环,李平. 改进PCA及其在过程监测与故障诊断中的应用[J]. 化工学报,2001,52(6):471-475.Wang Haiqing,Song Zhihuan,Li Ping. Improved PCA with application to process monitoring and fault diagnosis[J]. Journal of Chemical Industry and Engineering,2001,52(6):471-475.
[10] 赵忠盖,刘飞,徐保国. 基于改进混合概率主元分析模型的过程监控[J]. 控制与决策,2006,21(7):745-749.Zhao Zhonggai,Liu Fei,Xu Baoguo. Process monitoring based on improved mixture probabilistic principal component analysis model[J]. Control and Decision,2006,21(7):745-749.
[11] 徐涛,王祁. PCA在火箭发动机试车台传感器故障诊断中的应用[J]. 南京理工大学学报,2006,30(6):669-672.Xu Tao,Wang Qi. Application of PCA in sensor fault diagnosis of rocket engine ground testing bed[J]. Journal of Nanjing University of Science and Technology,2006,30(6):669-672.
[12] 成成,黄道. IPCA在故障检测与分离中的应用[J]. 华东理工大学学报,2000:502-506.Cheng Cheng,Huang Dao. Fault detection and diagnosis by integrated principal component analysis[J]. Journal of East China University of Science and Technology,2000:502-506.