The geometric mean is a superior frequency response averaging method for human body vibration

Abstract

The frequency response data of human body vibration are often used for standardisation, design of transport vehicles and occupational health and safety measures. This article shows that the commonly used methods of averaging frequency response spectra, such as arithmetic averaging in the complex or magnitude domain and median averaging, are not as suitable as the less commonly used geometric averaging in the complex domain. This is because it is necessary to minimise the deviation of the measured values about the mean value and to minimise the bias from the true mean value due to noise, distortion and nonlinearity. Practitioner summary: For averaging frequency response spectra, it is necessary to minimise the bias from the true mean value. This research shows that the commonly used averaging methods, such as arithmetic averaging in the complex or magnitude domain and the median, are not as suitable as geometric averaging in the complex domain. Abbreviations: H1 Estimator: frequency response function estimation method using the cross-spectrum of the output with the input divided by the auto-spectrum of the input; ISO: International Organization for Standardization; NHK: Nippon Hatsujo Kabushiki Kaisha; PCB: PCB Group ("PCB" is abbreviation for "PicoCoulomB"); RMIT: Royal Melbourne Institute of Technology; r.m.s.: root mean square