Vibration-based sound power level estimation: A method to reduce the number of sensors

The acoustic description of an industrial machine is crucial as it can provide information about its health status and assess the effectiveness of design modifications. Experimental acoustical measurements exist, but they may be impractical in noisy environments and limited by spatial requirements. Vibration-based approaches have been explored to overcome these limitations. However, the experimental procedure requires a large number of sensors. The aim of this work is to introduce a novel method which, starting from the vibration-based procedure, allows to reduce the number of sensors required to accurately estimate the sound power. An analysis of the number and the position of the accelerometers has been conducted to find the positions that better fit the resulting vibration-based sound power. An experimental campaign on a system consisting of a worm gear gearbox driven by an electrical motor has been used as a case study to verify the accuracy of the procedure. Acoustical based measurements were conducted and used as reference. Thus, starting from the data of twenty-seven accelerometers, the improved procedure was performed. The results will be plotted considering two and four accelerometers as input data and four operational conditions. The resulting vibration-based sound power levels demonstrate a good agreement in both octave bands and overall values. Furthermore, the trend of the radiation factor was investigated, aiming to determine the influence of parameters such as resistant torque and rotational velocity.