Automatic Outlier Detection in Automated Water Quality Measurement Stations
Author: Atefeh Saberi
Publisher:
Published: 2015
Total Pages: 96
ISBN-13:
DOWNLOAD EBOOKWater quality monitoring stations are used to measure water quality at high frequency. For effective data management, the quality of the data must be evaluated. In a previously developed univariate method both outliers and faults were detected in the data measured by these stations by using exponential smoothing models that give one-step ahead forecasts and their confidence intervals. In the present study, the outlier detection step of the univariate method is improved by identifying an auto-regressive moving average model for a moving window of data and forecasting one-step ahead. The turbidity data measured at the inlet of a municipal treatment plant in Denmark is used as case study to compare the performance of the use of the two models. The results show that the forecasts made by the new model are more accurate. Also, inclusion of the new forecasting model in the univariate method shows satisfactory performance for detecting outliers and faults in the case study data.