The Development and Implementation of a Classification System for Power Quality Phenomena

Jerry Collins, Prof. W. G. Hurley

This project details the development and implementation of a computerised system for the classification of Power Quality (PQ) phenomena. The system uses a hybrid Expert System / Neural Network approach, allowing a reduction in the rules required by the Expert System as well as a reduction in the time needed to train artificial neural networks. Various Expert Systems and Pattern Recognition (PR) Systems approaches where investigated with the view to developing a PQ classification system. Advantages and disadvantages of each system (as well as the hybrid system) are examined. The hybrid system is implemented using the computer languages Prolog and C. Technical issues regarding the implementation of the hybrid system are also discussed. Digital Signal Processing (DSP) techniques used to extract salient features from data collected during PQ phenomena are described. In particular spectral estimates and digital filtering are examined. Competing DSP techniques are compared by example.

Start date: 1st of September 1992. This project has been completed.

Project Publications

Conference
J. Collins, W. G. Hurley, "Application of Expert Systems and Neural Networks to the Diagnosis of Power Quality Problems", Third International Conference on Power Quality, PQA'94, Amsterdam, A-2.03, October 1994.
J. Collins, W. G. Hurley, T. P. McHale, P. J. Nolan, "Classification of Power Quality Problems Using Neural Networks and Expert System Approaches", 28th Universities Power Engineering Conference, UPEC'93, Staffordshire University, vol. 2, pp. 506-509, September 1993.

• Disclaimer
• Privacy
• Copyright
• Accessibility