Advances in AI and machine learning have been tremendous in recent years and impact our lives in many different and significant ways. We’ve seen self-driving cars become a reality, we are able to communicate with our homes using Alexa, and we are able to draft responses to our emails using AI virtual assistants.
Naturally, we expect similar advancements within heavy industries. Predictive Maintenance has been hailed as the Holy Grail for industries including oil and gas and maritime, where equipment failures are costly and safety is a priority. The reality is disappointing, and that is that we have fallen short.
We’re constantly told that the technology is there – look at AlphaGo, DeepMind or Libratus consistently outperforming humans at more complex tasks. But we are talking about building models to describe the behaviour of equipment under stochastic processes – like wind generation or oil production – with tremendously complex underlying physical behaviour. These are not easily encapsulated into a set of fixed rules, and would rely on many examples and boundaries to be successful.
Statistically (and luckily!), equipment fails very rarely, leaving us with a difficult predicament – how can we successfully build a model to identify specific failure modes on equipment, let alone predict it?
In this talk, we will discuss a methodology for building hybrid engineering and machine learning models that may be applied cross-equipment and industry.