An Arrowhead and Mimosa Based IoT Framework with an Industrial Predictive Maintenance Application

Manufacturing is undergoing an immense change triggered with widespread sensorisation, volumes of data being generated, and advanced machine learning technologies. Problems once solvable via simpler approaches considering more monolithic paradigms have evolved to become larger systems (Cyber Physical Systems; CPS) and Systems of Systems. The scaling, manageability, security, data handling requirements of such systems, as well as the industry’s common goal to reusability have led to several outcomes at the broader European level, Arrowhead and Mimosa being two of those so far. In this study, we consider an Industry 4.0 "Predictive Maintenance" problem. Instead of a rushing with straight data analysis approach as defined under CRISP-DM, we first delve into creating a more widely consumable and reusable set of building blocks by implementing an Arrowhead and Mimosa framework, which together form the route to the machine learning steps that finally lead to the solution.