This paper discusses recent work focused on the development of an innovative, multi-platform, machine learning-based technology that is capable of reliably and autonomously detecting small hazardous liquid pipeline leaks in near real-time. The technology is aimed at providing reliable detection of small leaks (
The technology is suitable for both mobile platforms (manned and unmanned aircraft, all-terrain vehicles, etc.) and stationary platforms, such as fixed installations at pump stations and block valve sites. The focus of the development was on the detection of liquid hydrocarbon leaks, but the technology also shows promise for detecting gas leaks.
Based on sensor input data, the system uses machine learning techniques to reliably detect “fingerprints” of small hazardous liquid leaks. The combination of the different types of sensors raises the possibility for detecting spilled product from an existing leak event, even if the leak is not actively progressing. Furthermore, the incorporation of different types and/or combination of sensors is also a possibility, making this technology very extensible.
Leak characterization was performed by imaging a variety of different types of hazardous liquid constitutions (e.g. crude oil, refined products, crude oil mixed with a variety of common refined products, etc.) in several different environmental conditions (e.g., lighting, temperature, etc.), and on various surfaces (e.g., grass, pavement, gravel, etc.).
Techniques were developed to extract a variety of features across the several spectral bands to identify unique attributes of different types of hazardous liquid constitutions in different environmental conditions, as well as non-leak events. The characterization of non-leak events is crucial in significantly reducing false alarm rates. Classifiers are then trained to autonomously detect small leaks and reject false alarms, followed by system performance testing. Trial results of this work are discussed in this paper.