Questions Smoothing Time Series Data on the Levels of Pollutant Emissions from Industrial Plants

• Published in: 2019 International Multi-Conference on Industrial Engineering and Modern Technologies (FarEastCon) pp. 1 - 4
• Main Author: Belov, A.A.
• Format: Conference Proceeding
• Language: English
• Published: IEEE 01-10-2019
• Subjects: data smoothing; detail coefficients; Distortion; pollutant emissions; Signal to noise ratio; Smoothing methods; threshold processing; time series; Time series analysis; Wavelet analysis; wavelet transform; Wavelet transforms
• DOI: 10.1109/FarEastCon.2019.8934855

The paper presents an algorithm for smoothing the time series of concentrations of gaseous pollutants contained in industrial emissions. In this case, smoothing is performed using the method of threshold processing of the detail coefficients of the wavelet decomposition a time series concentrations at different levels of the wavelet processing. An approach to finding the most efficient and optimal in terms of computational costs and accuracy in restoring the data of the threshold level, based on a preliminary assessment of the statistical characteristics of the high-frequency random components of the time series is presented. In the course of the research conducted on the effectiveness of the smoothing algorithm, various criteria were used to select threshold levels for processing detailed wavelet coefficients, such as the criterion for adaptive unbiased risk assessment, a modified heuristic risk assessment criterion, a minimax criterion, a criterion for a multilevel penalty cut-off factor. Conducted research showed that time series of monitoring systems for pollutant emissions, the optimal criterion for threshold treatment is the adaptive penalty threshold method.