Abstract
This paper introduces a potential seating comfort measurement system based on integrating temperature and relative humidity information. Placement of sensors was consistent with our previously published data with temperature and humidity sensors placed under each thigh (2 of each sensor) and the coccyx (1 of each sensor). Before the in situ trials were initiated, the temperature and humidity sensors were assessed for variance in output. The sensors performed consistently with standard deviations (SDs) of: temperature sensor ± 0.2°C and humidity sensor ± 0.3 %. In order to suppress unwanted noise, an EMD-based filter was applied to smooth raw data sets. The filtering performance of this EMD-based filter was compared with Moving Average filter, Local Regression filter and Savitzky-Golay filter using the goodness of statistics and residual charts as judgment criteria. Comparison results show the EMD-based filter has the lowest RMSE (root-mean-square-error) and the highest R-square values as well as smallest vibrations around zero line in temperature and relative humidity residual charts. This proves the EMD-based filter is the best choice in noise suppression against the other three classic filters.