Conductor Recognition Using Semantic Segmentation

Worldwide electricity distribution companies are moving towards risk-driven network replacement and maintenance strategies, and part of the risk assessment is to evaluate or estimate the asset condition. Unison has an ageing overhead conductor fleet consisting of over 4,200 km of high voltage lines, there are approximately 500 km of 7/0.064 Plain Hard-drawn Copper (PHC) 11 kV conductor in the fleet. The conductor has low tensile strength and an aged installation profile. Research that Unison conducted in 2018 identified the conductor’s visual appearance as a key indicator for condition assessment – hence Unison is currently developing a system that can use conductor images obtained by drones, to automatically evaluate the appearance and condition of the conductor.
This presentation will show the process followed to use machine learning and semantic segmentation to recognise/identify the overhead conductor region of interest in images with natural backgrounds. Such segmentation of the conductor is a required first step before evaluating the conductor appearance and condition, which also uses automated machine learning techniques

Speakers

Dr Peter Brady

Dr Peter Brady is a Senior Application Engineer at the MathWorks where he leverages maths to accelerate our customers projects. He works across the engineering spectrum with a focus on maths, statistics, optimisation, machine and deep learning as well as cloud scale out. Prior to joining MathWorks Peter worked for several civil and defence contractors delivering projects in surface water, hydraulics and hydrology as well as fluid dynamics and submarine cavitation inception. His prior research work includes free surface hydraulic simulation, thermodynamics and nanofluidics.
Peter has a PhD in Mechanical Engineering and a Bachelors in Civil Engineering both from UTS and is a Chartered Practicing Engineer in the fields of Mechanical Engineering and Leadership and Management (CPEng NER) as well as a Certified Professional with the Australian Computer Society. He is the current Chair of Sydney Division Water Engineering Panel and has chaired a number of Engineers Australia conferences.

Alex Castellanos

Alex Castellanos is working as Digital Engineering Technical Lead at Unison Networks Ltd. He graduated in electrical engineering/power systems (2001) from Mexico’s National Institute of Technology (Orizaba, Veracruz); he obtained a PhD degree (2015) from the University of Canterbury (Christchurch, New Zealand). He has over 10 years of experience in the power and energy industry including energy use, power generation, and power transformers monitoring and control. His current area of interest is data analytics and condition monitoring for the management of electricity distribution network assets.