Energy Prediction and Energy Management in Kinetic Energy-Harvesting Wireless Sensors Network for Industry 4.0

Real-time control and monitoring are some of the main goals Industry 4.0. To meet these requirements, sensors needed at every step production process. Wireless (WS) better suited due to their flexibility but limited in energy. In this work, kinetic energy harvesting using piezoelectric technologies considered ensure autonomy a Sensor Network (WSN). First, unlike most existing works, paper focuses on WSN rather than single WS since entirety industrial processes requires several WSs. The solution proposed here is based deep learning harvestable power signals each sensor deployed system. Specifically, vibration measurements were performed 12 locations an ore crushing mill mine. From there, mechanical–electrical conversion model considering system’s dynamics was set up evaluate profile WSs can harvest. Considering that has many peaks different operating states engine, we first Predictor Harvestable Power from Vibrations (PHPV). Using large database, compared state-of-the-art predictor, Energy vibrations (PHEV) allows for significantly reducing Root Mean Square Error (RMSE). More specifically, lowest reduction achieved RSME ranged 9.4 μW (with PHEV) 5.9 PHPV). A decrease RMSE ranging 18.45 4 obtained another measurement point. Since harvest rates differ one location another, Hierarchical Energy-Balancing Protocol (HEBP) maximize number capable transmitting information about state, thus avoiding interruption network coverage. HEBP, it envisaged WSs, besides data, will supply other nodes with deficit allow them communicate location. For minimum packet size 1100 bits, all ensured, only 66% previous protocols.