Mohammad Hajizadeh             Email:



• Ph. D.: Mechanical Engineering, Jan. 2009-present, University of Alberta, Edmonton, Canada

• M.Sc.: Automation and Instrumentation Engineering, 2006, Petroleum University of Technology (PUT), Tehran, Iran

• B.Sc.: Mechanical Engineering, 2001, Petroleum University of Technology (PUT), Ahwaz, Iran



• M. Hajizadeh, S. Kasiri and K. Salahshoor, "Online Prediction of Nonlinear Processes using an adaptive correction approach", 8th IASTED Conference on Control and Applications (CA), Montreal, Quebec, Canada, 2006.

• S. Kasiri, M. Hajizadeh and K. Salahshoor, "Adaptive Radial Basis Function (RBF) Neural Networks for Online Identification", IEEE International Conference on Control Applications, Munich, Germany, 2006.

• S. Kasiri, M. Hajizadeh and K. Salahshoor, "A Comparative Study of Online Radial Basis Function (RBF) Neural Networks Identification Approach", 8th IASTED Conference on Control and Applications (CA), Montreal, Quebec, Canada, 2006.


Project Title:  Applying Kalman filtering and particle filtering techniques for detecting and modeling anomaly and probabilistic damage events in time-varying systems

An anomaly or fault is a sudden and unpredictable change in the system performance or characteristic property (feature) from the acceptable or usual condition and it may initiate a failure or a malfunction in the system. Fault and anomaly detection refers to the problem of finding this pattern in data from the system. An undetected fault may lead to poor quality off-spec products, resulting in poor plant economic performance or accidents and injuries, which could be prevented by detection and identification of faults at an early stage. Therefore, fast and accurate fault detection and identification is important for efficient, economic, and safe system operation.

Anomaly detection in general is a challenging problem that is not easy to solve and lots of anomaly detection methods are exist. In fact, most existing anomaly detection methods solve a particular type of the problem. A model-based analysis approach using statistical classification techniques may be able to identify anomalous events that contribute to process degradation and damage accumulation in a nonlinear time-varying system. Feature vectors related to such events would give insights into what process and reliability factors are important for continuous improvement of operating and maintenance practices. In this project I want to use Kalman filter as a model-based technique in combination with some statistical techniques for fault detection and diagnosis in nonlinear time-varying system.

Progress to Date: I did a comprehensive literature survey on the existing fault detection methods and their pros and cons.  Some of these methods were used on benchmark problems to study their performance and reliability, and on an oilsands equipment problem using actual operating and maintenance data.


Expected completion date: 2013.


 Return to Mike Lipsett’s Research Page