Research on proper sampling parameters determination approach of digital filters in small steady-state heat flux measurement

Abstract

Noise fluctuation often occurs in small steady-state heat flux measurement and it significantly disturbs the measurement precision. When using the digital filters denoising the raw heat flux signals on-line, the sampling parameters, i.e. the sampling size and sampling period have an important influence on the filtering effect. In this paper, the aim is to explore an appropriate sampling parameters determination approach. Firstly, a typical small steady-state heat flux measurement apparatus was built with a refrigerator cabinet based on the reverse heating loss method. With the help of error analysis and via particle image velocimetry technique, it is inferred that the noise mainly originates from the error of the acquisition system and the flow field unsteady-state on the given measurement condition. Then, three cases are performed to explore the determination approach of appropriate sampling parameters for six digital filters with the assist of running coefficients of variation of filtering effect assessment metrics. The numerical results indicated that the sampling independent sizes for rolling average, median, moving average, Savitzky-Golay, Gaussian and Kalman filters are 9780, 3210, 3926, 3676, 3179 and 8089, respectively. The start-point has a slight impact on the sampling independent size. The raw heat flux signals with large sampling period span can capture more noise information of the time series signal. Considering the filtering effect and sampling efficiency, the moving average filter with the sampling size of 3926 and the sampling period of 3.308 s is suitable for the steady-state heat flux denoising work of refrigerator cabinet.