IOT data-driven experimental process optimisation for kevlar fiberglass components for aeronautic

This paper describes the work carried out during the PROOF experiment (IOT data-driven experimental PROcess Optimization for kevlar Fiberglass components for aeronautic), winner of the second open call of the MIDIH EU project. The main objectives of the experiment are the integration of smart sensing devices with the Energy@Work IoT gateways and the development of cloudified innovative data-driven methodologies and data analytics tools to support process optimization in the production of hybrid composite material parts for the aeronautical sector. Collection of real-time production-data from multiple sensors with several industrial protocols and data transfer to the MIDIH project platform has been performed adopting the IoT gateway developed by Energy@Work, following MIDIH reference architecture for advanced data processing and visualization (e.g., Fiware Orion Context Broker, Apache Flink and Fiware Knowage) by using MQTT protocol. Then, historical and new acquired data has been analysed using advanced clustering techniques and trends, with the purpose to allow a novel CPS-based predictive system on the production process. Machine-Learning algorithms and visualisations (GUI based on Fiware Knowage) in real operating conditions have been used to validate the performance and assess the outcome. Finally, thanks to the implementation of specific optimization rules, able to process data gathered from the sensor network, a framework for distributed processing engine has been exploited by (i) generating tips for energy efficiency and process optimization and (ii) providing different type of alarms based on expected consumptions, resulting in concrete support to production managers for the improvement of the whole production value chain.