A Novel Approach to Corona Monitoring

Intelligent Automation, Incorporated (IAI) with the support from its subcontractor, Prof. W. Lee at University of Texas at Arlington propose an integrated approach to detect PC/CD. It consists of three major steps: preprocessing to remove background andsensor noise, Principal Component Analysis (PCA) for feature extraction and dimensionality reduction, and neural net classifier for fault detection and isolation. In the preprocessing step, our main objective is to eliminate as much as possible theexternal and sensor noise. This step is quite critical as different sensors may have different noise characteristics. We will customize noise filtering algorithms for each sensor. An adaptive algorithm is proposed to deal with the time varying backgroundnoise. In the feature extraction step, we apply PCA to extract the important feature out of the sensor signals. Another advantage of PCA is that the signal dimension has been significantly reduced. Consequently, the subsequent computational load will bereduced tremendously. Overall training and testing of the neural network classifier will be greatly improved. In the classification step, we propose to use a standard multiplayer perceptron neural net to do the classification. The learning capability of NNallows the monitoring system to learn from historical data. Hence performance will be improved as more operational knowledge has been accumulated. Our integrated software and hardware for corona monitoring will be directly used in power systems for theAir Force. IAI is also under contract with Boeing for development of Health Monitoring algorithms for the Future Combat System. Moreover, IAI also has good relationships with Ford. The Phase 2 work will provide an excellent demonstration for our partnersso that they will directly transition our technologies to their product lines.