An Outliers Detection Method of Time Series Data for Soft Sensor Modeling

Aiming at the particularity of data outliers of soft sensor modeling in complex industrial processes, a new outliers detection method for time series is proposed. The new method combines the traditional density-based clustering algorithm (DBSCAN) with soft sensor modeling process. The soft sensor modeling errors are used as the guidance of outliers detection process and replace the traditional manual intervention in the clustering process. Meanwhile the outlier detection is completed as well as the soft sensor modeling is established. The experiment shows that the new outliers detection method has good performance.