Open Tech Calendar

Augmented Engineering Intelligence for Industrial Equipments in the Energy Sector

Predicting failure is a very well known and difficult problem. There are good reasons for that - known failure modes are scarce, let alone unknown failure modes (dreaded “black swans”), capable of causing the most damage. Also, equipments generate immense streams of multidimensional data, representing highly connected systems, thus anomalous patterns can be easily missed by humans alone. The mostly unlabelled data makes it almost impossible to create an acceptable supervised ML model. Instead, we present an unsupervised deep learning approach, inspired and adapted by the works of Google on smart buildings, to create a set of Early Warning Notifications for industrial equipments.

Bio: Pedro Santos (www.linkedin.com) is the Lead Data Scientist at Digital & Data (Total Energies UK). He is specialised in turning ideas into actionable business insights with machine learning. In 2021, the team which he is part of won the Oil & Gas UK Business Innovation Large Enterprise Award.