We caught up with Jonathan Hodson, Machine Learning Engineer at ONYX InSight, to discuss key challenges for the wind industry such as data access and inspection data, and how ONYX InSight supporting the industry to solve these challenges.
Hi Jonathan! To kick things off, could you give us an overview of your experience and career to date?
After graduating from the University of Nottingham with a degree in Physics, I started work as a tool designer and data scientist at a wireline logging firm, which lowers sensors into oil wells on a line to measure formation properties. I was involved with signal processing and electronic tool design. After a couple of years I joined the graduate scheme at Romax Technology.
There, I worked on electric drivetrains, R&D on thermal modelling – but also gained a solid background in gearbox design.
How did you come to work for ONYX InSight?
After working in oil and gas, I wanted to pursue a career in a greener sector, focusing my efforts on sustainable forms of energy production. This was coupled with a desire to build on my data analysis skills from University, so the opportunity to combine both of these propositions was particularly attractive.
I’d worked at ONYX InSight as part of my Romax graduate scheme, and it seemed like a natural fusion of my previous experience with gearboxes and data science. Once I started working with machine learning algorithms at ONYX Insight, everything clicked into place!
Could you give us a quick run-down on what you’re currently focusing on at work?
We are constantly building new iterations of our data tools so that we can continue to deliver the most accurate results. We’ve been working on SCADA health checks, for example, and it’s a collaborative process. We get inputs from our mechanical engineers to ensure that our models are grounded in the latest concrete engineering experience from wind farms across the world. The end result? Data analysis backed with real-world engineering.
What are the most exciting developments in wind energy right now?
Significant strides are being made in data collection and sharing. ONYX InSight have been working with the industry to improve data access, which is still a challenge – but things are better than before.
Owners and operators are realising that the real value of data is unlocked through sharing with data analytics partners that can extract actionable insights – which ultimately saves them money.
What is the main challenge for the wind industry right now?
One of the key areas for improvement throughout the sector is maintenance and inspection data. Currently, this valuable information on turbine health is untapped, as it is often held in a format which makes analysis difficult, such as written notes. This means the industry is missing out on significant savings as data showing key trends is locked away in filing cabinets.
Applications like ONYX InSight’s fieldPRO support technicians on the ground by digitising and automatically organising inspection data, opening the door to analysis through machine learning. As part of a wider suite of datasets, inspection data can give a clear picture of developments in turbine health to enable owners and operators to gain control of their maintenance.
We’re also excited to roll out our new AI Hub platform next year. Among other functionalities, this will use a case management system that consolidates maintenance reports and scheduling. For example, a triggered alarm will be entered as a ‘case’ in the system – and then followed up with an inspection. The inspection log and photos are then digitised and uploaded to the case file, drawing critical data into one place.
What are the key trends in data analytics which could impact the wind energy sector in the next 5 years?
Computational power is becoming cheaper all the time, which means we can handle increasing amounts of data. In the next 5 years, these costs will reduce further, which could open avenues to analyse even higher resolutions of data, enabling the most accurate characterisations of turbine behaviour yet and improving diagnostics.
To take advantage of this and avoid missing out on this valuable information the industry must adopt more granular data collection and reporting standards. We currently have access to 10-minute SCADA data, but with one-second data, we can deliver even more accurate and powerful insights into turbine health!
What will a typical wind farm look like in 2030?
For a start, there would be significantly fewer people there! Reducing engineer time-on-turbine is a key opportunity for the wind industry, and this is done through building in sensors to capture key data on turbine health. Wireless sensors and the internet of things will open up a new wave of predictive maintenance and data analytics as data-rich assets become the norm.
One of the biggest trends we’re seeing now is retrofitting online oil sensors. Combining oil and vibration data produces the most accurate insights to enable more effective predictive maintenance, and as owners and operators start to benefit from this, turbines will inevitability be designed with built in capabilities to gather this vital data-stream.
Who is your engineering hero?
There’s plenty of famous historical engineers to choose from here, but the answer is closer to home. My father, an engineer himself, fostered my love of engineering from an early age, encouraging my natural passion for solving real world problems – which continues to this day at ONYX InSight.