Presentation Profile
Applications of Machine Learning for Improvements in Reliability, Detection and Quantification Capabilities for Methane Leak Detection Systems
Currently Scheduled: 10/12/2021 - 9:25 AM - 9:55 AM
Room: Iris Room
Main Author
Andrew Schaub - Southwest Research Institute
- Heath Spidle - Southwest Research Institute
Abstract:
Southwest Research Institute, through a partnership with the Department of Energy’s National Energy Technology Laboratory have been developing the Smart Leak Detection-Methane (SLED/M) system for autonomous real-time methane leak detection. Application of machine learning techniques on midwave infrared optical gas imagery data resulted in improved reliability, detection and quantification capabilities relative to similar approaches. In the initial phase, real-time capable predictive models were developed for pipeline monitoring of remote or difficult-to-access sites. In the more recent phase of research, an optical gas imager was augmented with the capability to quantify flow-rate from a safe distance and evaluated for field conditions.
