Even though there has been a recent interest in the use of principal component analysis (PCA) for sensor fault detection and identification, few identification schemes for faulty sensors have considered the possibility of an abnormal operating condition of the plant. This article presents the use of PCA for sensor fault identification via reconstmction. The principal component model captures measurement correlations and reconstructs each variable by using iterative substitution and optimization. The transient behavior of a number of sensor faults in various types of residuals is analyzed. A sensor validity index (SW) is proposed to determine the status of each sensor. On-line implementation of the SVI is examined for different types of sensor faults. The way the index is filtered represents an important tuning parameter for sensor fault identification. An example using boiler process data demonstrates attractive features of the SW.

Fault detection and process monitoring using principal-component analysis (PCA) and partial least squares were studied intensively and applied to industrial processes. The fundamental issues of detectability, reconstructability, and isolatability for multidimensional faults are studied. PCA is used to define an orthogonal partition of the measurement space into two orthogonal subspaces, a principal-component IntroductionProcess equipment deteriorates with time and use, and requires frequent maintenance. In some processes, equipment degradation affects the quality and yield. However, in other cases, the plant production is not affected by the equipment degradation until a major failure occurs. This type of drastic failure of process equipment is usually very costly because the production would be interrupted for a period of time, and other equipment or product stocks could be affected by such a failure. For this reason, in general, it is important to detect sensor faults as well as process faults that indicate significant deterioration of equipment.Process disturbances are another source of operational problems in a plant (Ku et al., 1995). Because not all the variables that affect the process are under the control of the plant operations, the changes in one of these variables may take the process operation out of its optimum. Moreover, severe process interaction increases the possibility that a control action, which tends to correct a disturbance in a particular process unit, will affect other units of the plant. This interaction also shows the necessity of monitoring the process in an integrated manner to avoid incorrect fault diagnostics. Therefore, for appropriate monitoring of a process, multipleCorrespondence concerning this article should be addressed to S. J. Qin.sensors located at different places in the plant should be considered to detect a process fault.The detection of changes in the process operation of a plant requires a monitoring technique that quantitatively represents the major relations among the process variables. Violation of such relations would indicate a potential malfunction in the plant. Principal-component analysis (PCA) (Jackson, 1980;Wold et al., 1987) is a reliable and simple technique for capturing variable correlation. Several articles have recently illustrated the use of PCA in process monitoring and statistical process control (Kresta et al., 1991; De Veaw, et al., 1995), fault diagnosis Raich and Cinar, 1994), and sensor validation (Tong and Crowe, 1995; Dunia et al., 1996a,b,c). The applications vary from batch processes to continuous processes (Dunia and Qin, 1998). Miller et al. (1993) proposed a contribution plot approach based on PCA models to help identify variables strongly affected by the underlying fault.Even though many authors have reported the success of using PCA for process monitoring, there are some fundamental issues related to statistical process monitoring that have not been developed in the PCA framework. For example, based on the confidence region used...

No abstract

No abstract

No abstract

Inferential sensors, or soft sensors, refer to a modeling approach to estimating hard-to-measure process variables from other easy-to-measure, on-line sensors. Since many sensors are used as input variables to estimate the output, the probability that one of the sensors fails increases significantly. In this paper, we propose a self-validating inferential sensor approach based on principal component analysis (PCA). The input sensors are validated using a fault identification and reconstruction approach proposed in Dunia et al. AIChE J. 1996, 42, 2797-2812. A principal component model is built for the input sensors for sensor validation, and the validated principal components are used to predict output variables using linear regression or neural networks. If a sensor fails, the sensor is identified and reconstructed with the best estimate from its correlation to other sensors. The principal components are also reconstructed accordingly for prediction. The number of principal components used in sensor validation and prediction are chosen differently based on different criteria. The typical input correlation or collinearity is utilized for sensor validation and removed in predicting the output to avoid ill-conditioning. The selfvalidating soft sensor approach is applied to air emission monitoring, where continuous monitoring of the air pollutants is required for environmental regulations.

No abstract

No abstract