Intelligent algorithms help plan machine maintenance
Industrial machines and vehicles and the environment they operate in can be continuously monitored by a range of intelligent sensors. The result is huge amounts of data concerning usage history, operating conditions, location, movement and other physical properties. These systems form part of a larger network of diverse collaborative systems like renewable energy parks and vehicle fleets. Unplanned machine downtimes due to unexpected equipment failures can prove costly for the company and reduce profitability. There is a need to accurately predict component failures as well as wear and tear. The Horizon 2020 MANTIS(opens in new window) project has designed and developed a proactive maintenance service platform reference architecture based on CPS. This platform can be used for predicting and preventing imminent failure and scheduling proactive maintenance for industrial machinery, transforming raw data into knowledge while minimising bandwidth. The CPS are controlled by computer-based algorithms, which are closely integrated with the internet and its users. Better visualisation helps explain data The consortium collated data using CPS and sent it to the cloud. “We have learnt how to select the optimal sensors that can provide the wear-related information of a critical asset, and the type of intelligent functions that are available for providing an increased level of knowledge and have validated them,” says project coordinator Dr Urko Zurutuza. “We designed, developed, and validated new sensors for the manufacturing community and have investigated how best to communicate with them in challenging environments,” he adds. Machine learning algorithms learn the behaviour of the machines to identify potential problems before they begin. “We have proposed, created, described and validated artificial intelligence algorithms in the form of algorithm portfolios, so industry can similarly use or apply them, based on our experience, and the type of data they are faced with,” explains Dr Zurutuza. Knowing the future condition of machines and using advanced visualisation techniques allowed maintenance engineers and operators to better understand the information the algorithms provided. Dr Zurutuza comments: “The aim is to identify any problems before they begin and combine online and historical data and present it to maintenance engineers and technicians using adaptable views, when too much data would be confusing using conventional visualisation techniques.” Saving time and money Researchers carried out pilot studies to test and validate the innovative functionalities of the proactive maintenance service platform architecture and its future use in the industrial sector. Dr Zurutuza states: “As well as the architecture, intelligent sensors, and algorithms, we also studied how it was implemented in several industries, such as photovoltaic and wind turbine parks, rail networks, or manufacturing equipment.” MANTIS will benefit companies by limiting the downtime of industrial assets and transport during maintenance, thereby reducing costs and increasing competitiveness, growth and sustainability. “Seeing how the project has won over industrial partners and helped them progress towards a completely new approach to maintenance services has been extremely rewarding,” points out Dr Zurutuza.
