Using disturbance records to automate the diagnosis of faults and operational procedures in power generators

Nowadays, it is a common practice in power generation utilities to monitor the generation units using digital fault recorders. As the disturbance records are generally analysed and stored at a central office or control centre, it has become difficult for engineers to analyse all this data. Some of the main steps in developing automated diagnosis tools to help in this task are the segmentation and feature extraction of the recorded signals and decision making. This study presents a methodology to extract meaningful information from each segment of a disturbance signal. In the approach described in this study, the segmentation is performed by an extended complex Kalman filter. The main features extracted from each segment are symmetrical components at fundamental frequency of voltage and current signals. Feature extraction uses root-mean-square values to obtain the symmetrical components of the three phase quantities. This methodology focuses on offline analysis of fault recorder data of power generators and it is developed not only to fault analysis, but also to verify normal operational procedures, from which result most of the disturbance records. This study also describes an expert system that can be used to automatically classify each record into known categories, focusing the engineer's attention to the most relevant occurrences.