This version is available at https://strathprints.strath.ac.uk/41327/ Strathprints is designed to allow users to access the research output of the University of Strathclyde. Unless otherwise explicitly stated on the manuscript, Copyright © and Moral Rights for the papers on this site are retained by the individual authors and/or other copyright owners. Please check the manuscript for details of any other licences that may have been applied. You may not engage in further distribution of the material for any profitmaking activities or any commercial gain. You may freely distribute both the url (https://strathprints.strath.ac.uk/) and the content of this paper for research or private study, educational, or not-for-profit purposes without prior permission or charge.Any correspondence concerning this service should be sent to the Strathprints administrator: strathprints@strath.ac.ukThe Strathprints institutional repository (https://strathprints.strath.ac.uk) is a digital archive of University of Strathclyde research outputs. It has been developed to disseminate open access research outputs, expose data about those outputs, and enable the management and persistent access to Strathclyde's intellectual output. A number of different approaches to PD defect diagnosis have been developed to extract defect-specific information from PD pulse data in both the time and frequency domains. Frequency based PD pulse analysis has previously been demonstrated to offer a low-power approach to PD defect identification, where a mixture of passive and active analog electronics can be used to generate diagnostic features in a low-power device suited to wireless sensor network operation.This paper examines approaches to implementing diagnostic methods for frequency-based PD pulse diagnosis targeted at compositional frequency spectrum features in a computationally efficient manner. Dirichlet and Gaussian distributions are used to demonstrate the complex probabilistic form of fault class decision surfaces, which motivates the proposed application of the log ratio transform to frequency composition data.The results demonstrate that PD defects can be differentiated using these frequency-based methods and that employing the log ratio transform to the compositional frequency content data yields increases in classification accuracy without necessarily resorting to more complex classifiers.