粮情监控系统中传感器故障诊断和数据恢复  

Fault Diagnosis and Data Reconstruction for Sensors in Monitoring System for Condition of Stored-Grain

作　　者：王海涛[1] 刘倩[2] 陈桂香[1] 王军[1] 王海霞[1] 陈雁[1] 

机构地区：[1]河南工业大学土木建筑学院,郑州450001 [2]华北水利水电大学软件学院,郑州450000

出　　处：《中国粮油学报》2013年第11期86-90,96,共6页Journal of the Chinese Cereals and Oils Association

基　　金："十二五"国家科技支撑计划(2011BAD03B01);博士基金(2012BS064)

摘　　要：粮情监控系统对粮食的安全储藏十分重要,然而,粮情监控系统中的传感器经过长期使用后往往会出现故障。针对粮情监控系统中的传感器故障,本研究给出了一种基于主成分分析法的传感器故障诊断和数据恢复方法。利用主成分分析法对正常运行条件下的测量数据进行建模,可将测量空间分为主成分子空间和残差子空间。利用平方预测误差检测传感器故障,利用传感器有效性指数辨识故障和寻找故障源,通过多次迭代和逐步逼近主成分子空间的方法实现数据恢复。最后,利用实际粮情监控系统的测量数据验证了传感器故障诊断和数据恢复方法的正确性和有效性。Monitoring system for condition of stored - grain is very important to the safety of grain storage. However, sensors in monitoring system for condition of stored - grain tend to more faults due to the complexity of the sys- tem and poor work environment. A fault diagnosis and data reconstruction strategy under the using of principal component analysis（ PCA）has been presented in this paper for sensors in monitoring system condition of stored -grain. The measured data under operation condition was used to build principal component analysis models. The PCA model was utilized to partition the measurement space into the principal component subspace （PCS）and the residual subspace （RS）. Square prediction error（SPE） statistic was utilized to detect sensor faults. Sensor validity index（SVI） was eraployed to identify and locate faulty sensors. Faulty data was recovered by sliding the faulty data to PCS via iteration. Finally, the strategy proposed was validated using data from a real monitoring system for condition of stored - grain. The validation results showed the PCA -based sensor fault diagnosis and data reconstruction strategy is accurate and effective.

关 键 词：粮情监控系统 传感器故障 故障诊断 主成分分析 数据恢复 

分 类 号：TP273[自动化与计算机技术—检测技术与自动化装置]