Abstract— Due to aging and environmental factors, system components may either fail or not function as expected, which causes unprecedented changes in the quality of the system. A timely detection of the onset of a fault in a component is crucial to a quality monitoring of a process if costly failures are to be avoided. However, finding the source of the failure is not trivial in systems with a large number of components and complex component relationships. In this paper, an efficient scheme to detect adverse changes in system reliability and find the failed component is proposed in order to have an effective process quality monitoring. The monitoring scheme has been made effective by implementing first the techniques of fixed- parameter Shewhart and Hoteling’s T 2 control chart, and then the adaptive versions of Shewhart Chart and T 2 control chart for counter- checking the precision of quality reports. Once detected, the fault isolation scheme uses a Bayesian decision strategy based on the maximum correlation between the residual and one of a number of hypothesized residual estimates to generate a fault report. By doing so, the critical information about the presence or absence of a fault, and its isolation, is gained in a timely manner, thus making the quality monitoring system an effective tool for a variety of maintenance programs, specially of the preventive type. The proposed scheme is evaluated extensively on simulated examples, and on a physical fluid system exemplified by a benchmarked laboratory-scale two-tank system to detect and isolate faults including sensor, actuator and leakage ones. Keywords— Quality Monitoring, Fault detection, adaptive shewhart chart, adaptive T 2 Chart, benchmarked laboratory-scaled two tank system. I. INTRODUCTION he problem of monitoring the quality of a plant has always been recognized as being of primary importance and has been the subject of intense research for quality engineers. Diagnosis of unprecedented changes in system Abdur Rahim is with the Faculty of Business Administration, University of New Brunswick, Fredericton, NB, Canada, E3B 5A3, e-mail: rahim@unb.ca Haris M. Khalid is with the Systems Engineering Department, King Fahd University of Petroleum and Minerals, P. O. Box 8283, Dhahran 31261, Saudi Arabia, e-mail: mhariskh@kfupm.edu.sa. Muhammad Akram is with the Systems Engineering Department, King Fahd University of Petroleum and Minerals, P. O. Box 116, Dhahran 31261, Saudi Arabia, e-mail: akram@kfupm.edu.sa . Lahouari Cheded is with the Systems Engineering Department, King Fahd University of Petroleum and Minerals, P. O. Box 116, Dhahran 31261, Saudi Arabia, e-mail: cheded@kfupm.edu.sa. Amar Khoukhi is with the Systems Engineering Department, King Fahd University of Petroleum and Minerals, P. O. Box 116, Dhahran 31261, Saudi Arabia, e-mail: amar@kfupm.edu.sa . Rajamani Doraiswami is with the Department of Electrical & Computer Engineering, University of New Brunswick, Fredericton, Canada, P. O. Box 4400, e-mail: dorai@unb.ca reliability and detection of the source of the change is essential for removing faulty components, replacing them with better ones, restructuring system architecture, and thus improving the overall system reliability. However modern complex systems create challenges for systems engineers to understand and trouble-shoot possible system problems. Therefore due to the large system size, the use of efficient monitoring and fault diagnosis methods become unavoidable for complex systems [1]. Measurements are needed to monitor process efficiency and equipment condition. Data from a faulty component is composed of a time series of measurements of all the state variables describing this component. For example, in the case of a simulated leak, the leak flow is initially set to zero for the first element of the series, and its value is gradually increased with time. A single excursion of a component out of its limits, is not enough to detect safely the presence of a fault in the process. An error is only flagged when a component remains out of bounds during several consecutive steps. Monitoring is a continuous real-time task of determining the conditions in a physical system. It consists of recording information, recognizing changes and detecting abnormalities in the system’s behavior. The faults to be monitored that are considered here include sensor, actuator and leakage faults, and can be classified broadly as either parametric faults or additive ones. An additive fault manifests itself as an additive exogenous signal in the measured data, while a parametric one induces a variation in the system parameters. II. RELATED WORKS Related studies conducted in this area consist of three categories: quality monitoring models, statistical process control tools and both univariate and multivariate approaches to monitoring. A. Quality Monitoring Various hydraulic models have been proposed to detect leaks in water distribution systems. Minimizing the difference between measured and calculated pressure and flow gives the solution to an inverse problem [2]. Ligget and Chen [3] extended this method to transient flow. These approaches can detect network leakage at nodal points only and require large amounts of data. Liou and Tian [4] developed a time marching algorithm to detect small and moderate size leaks under both steady and transient flow An Adaptive SPC approach to Quality Monitoring and Fault Diagnosis: A Comparative Study M. A Rahim, Haris M. Khalid, M. Akram, Lahouari Cheded, Amar Khoukhi and R. Doraiswami T DISCLAIMER: This is the authors' version of an article published in early access of International Conference on Computer, Electrical and Systems Engineering (ICEME) as CONFERENCE PAPER. Changes were made to this version by the publisher prior to publication.