14:30 - 15:00
Thursday, 19 September 2019
T2.9 Machine Learning for Automated Model Building in Oil and Gas
A supermajor oil and gas company partnered with SparkCognition to obtain predictive AI capable of detecting abnormalities in their gas system for offshore rigs. Specifically, the goal was to provide a single, scalable solution that could monitor all assets, despite the differing behaviors and life cycles of machines within the gas system.
An unsupervised approach was used to identify normal versus anomalous behavior, and to track anomalies to downtime events. A model was built that could correctly predict the cluster of any given data point in real time, which was retrained each month. The model was able to successfully identify anomalies and track them to production downtime events, giving advanced warning to engineers of what, specifically, they needed to investigate, and when, and providing leading indications for root cause analysis.
This proof of concept was so successful that the oil and gas company has decided to go forward with a full deployment of the same technology. As valuable as predictive maintenance is to oil and gas operators, it can be difficult to implement. This use case demonstrates that AI can successfully be employed to overcome this challenge.