Chunyan (Cheryl) Qu

Education:

Ph.D., Texas A&M University, College Station, TX (2009)
M.Eng., National University of Singapore, Singapore (2003)
B.Eng., Zhejiang University, Hangzhou, China (2000)
Chunyan (Cheryl) Qu

Research Interests

Process monitoring & parameter estimation

One important aspect of process safety is detection of abnormal operating conditions. A common approach to this problem is that important states and parameters in a process are monitored and compared against their upper and lower bounds. However, some of these states and most of the parameters cannot be directly measured and instead have to be computed from plant data. This raises the question to what degree the results are affected by the procedure used for computing the values of unmeasured states and parameters. Therefore, a detailed comparison has been made between several state estimation methodologies such as linearized Kalman filter (LKF), extended Kalman filter (EKF) and moving horizon estimation (MHE) with a specific emphasis on unscented Kalman Filter (UKF) as it is a relatively new technique in this part of research. Additionally, as an extension of this work, an approach for computation of arrival cost for MHE via UKF is proposed and its performance is investigated as compared to the one via the commonly used EKF.

Fault Detection and Identification

Early and accurate fault detection and diagnosis is an essential component of operating modern chemical plants as the level of instrumentation in chemical plants increases. It plays an essential role in reducing downtime and costs, increasing safety and product quality and minimizing the impact on the environment. While alarm management is one form of fault detection, the information contained in the HAZOP (Hazard and Operability Studies) is often very qualitative in nature and the exact threshold for initiating alarms are determined from past experience with the plant. Additionally, alarm management is usually performed by setting threshold for individual variables, thereby neglecting the effect on variables on one another. As a result of this, it often happens that several alarms are initiated at the same time which complicates the response to the abnormal situation. These points will be addressed in this part of research by investigating a fault diagnosis system which will be able to determine the type and location of the fault (sensor fault, process fault, actuator fault) as well as the magnitude in the presence of measurement noise and uncertainty in the model of the plant. Subsequently appropriate verification of HAZOP procedures and alarm thresholds could be determined.