Piezoelectric Energy Harvesting Prediction and Efficient Management for Industrial Wireless Sensor

Abstract

The vibrations, due to their abundance in most industrial processes, constitute an attractive solution for the power supply of Industrial Wireless Sensor (IWS). However, the amount of energy that can be harvested presents numerous fluctuations due to the engines’ different operating modes (overload, full load, or even operation without charge). Most designs do not incorporate this fluctuation in the definition of the specifications of the autonomous IWS. This paper then presents a design method to ensure the node’s energy autonomy while maximizing its Quality of Service (QoS). To precisely define the specifications of the IWS, vibration measurements were carried out at its location for one month. The recorded data was used to propose a new Predictor of the Harvestable Energy from Vibrations (PHEV). A comparative evaluation of the proposed PHEV performances with a state-of-the-art predictor is carried out. The results obtained show that the PHEV makes it possible to minimize the Root Mean Square Error (RMSE) from 28.63 mW to 19.52 mW. A model of energy dissipation in IWS, considering the Internet of Things’ requirements, was established. The model is based on Long-Range (LoRa)/Long-Range Communication Wide Area Network (LoRaWan). The amount of data transmitted is then maximized according to the expected energy harvest rate by setting up a Maximization Data Size Protocol (MDSP). The proposed method makes it possible to ensure an acceptable QoS without resorting to reconfigurable circuits, which are sometimes bulky for miniature devices such as the IWS.