Natural gas pipeline valve leakage rate estimation via factor and cluster analysis of acoustic emissions

This paper estimates the leakage rate of a valve in a natural gas pipeline via factor and cluster analysis of acoustic emission signals. Factor analysis was used to reduce the amount of redundant information in the highly dimensional features and extract the optimal features for the cluster analysis. Three types of clustering algorithm-fuzzy C means, k-means and k-medoids-were used to classify leakage rates. Performance was evaluated in terms of overall accuracy, computational time, iterations, Jaccard coefficients and Cohen's kappa. A model based on factor analysis and k-medoids clustering was found to be exceedingly effective for recognizing internal valve leakage rates. This method proved to be superior to the k-means and fuzzy C means clustering methods, and has potential value in real-world applications.