Monday, May 11, 2020
2:00 pm: Automating Wind Turbine Inspection through Deep Analytics for Sulzer&Schmid Laboratories
Dr. Jonathan Masci, Director of Deep Learning, NNAISENSE
2:30 pm: ChildrEN SafEty Retrieval (CENSER) ...
Amarjot Singh, Founder and CEO, Skylark Labs
To remain cost-competitive with alternative forms of energy in the subsidy-free era, the wind industry must find new efficiencies from data-driven cost-saving measures such as transitioning away from a reactive maintenance strategy to a more proactive one which exploits advanced analytics to identify damage. In this talk, Jonathan will discuss their work with Sulzer&Schmid Laboratories, a Swiss company that provides turbine inspection services to wind farms, to automate part of their damage detection operations using deep learning-based predictive analytics. Sulzer&Schmid uses a two stage process where high-resolution images of the rotor blade surface, taken by autonomous drones, are first pre-annotated by humans to identity areas of interest which focus the attention and dramatically increase the efficiency of blade experts who, in the second stage, make the final determination as to whether damage is present. Their goal in this work was to replace the pre-annotation stage with a deep learning predictive model trained using previously labeled images. Jonathan will discuss the challenges related to training large supervised models with incomplete and noisy labels, and demonstrate the solution in action, showing how it not only reduces cost and increases throughput, but also actually improves upon the performance of the human annotators in terms of both accuracy and utility. If time permits, he will end by discussing how data from multiple inspections (i.e. a sequence of inspections per turbine) could be used for predictive maintenance by learning a deep model of blade surface deterioration dynamics.