AI to deliver simplified predictive maintenance for Coal Authority

In what is believed to be a world first machine learning deployment data analytics for 75 pumps spread across 50 locations will be automated to enable multi-site predictive and preventive maintenance.

IT provides SolutionsPT, together with Schneider Electric, are partnering with the UK’s Coal Authority in the project.

Andrew Ballard, Vice President Process Automation at Schneider Electric commented, ‘When a coal mine ceases its operations, the essential maintenance and dewatering systems that prevent flooding stop working, too. Flooding poses a threat for the structural security of the mine, so it is important that preventative measures are taken.’

‘For this project, we focused on bringing together the IIoT data from the extensive distributed architecture into a single model, and as a result, enable the Coal Authority to continuously reduce risk and improve efficiency.’

Mazhar Hussain Business Unit Manager – Infrastructure, SolutionsPT, explained, ‘[The system] uses predictive models to generate alerts without input from operators. This is unsupervised machine learning in which the software can draw its own inferences and compare deviations from normal operating parameters to highlight potential issues before they cause a failure. This is made possible by the system constantly analysing real-time data and comparing it to the expected modelling.’

The system also ‘learns’ by inviting the specific user to give the information it provides a ‘thumbs up’ or a ‘thumbs down’ as to the usefulness of the information it offers. With different job roles using the data in different ways, the system is continuously learning which information to share with whom, and when.