Abstract
Presented is a novel method for disturbance classification using a higher-order statistics-based technique for extracting a reduced and representative event signature vector and a neural network for classification. The signature vectors provide enough separability among classification regions, resulting in a classification rate of 100% for the validation events set.