Fuzzy Treatment Method for Outlier Detection in Process Data

Abstract

A novel fuzzy-logic based data treatment framework is proposed for the detection of outliers in process data. The proposed method incorporates outlier detection parameters into a fuzzy strategy. The method utilizes Hampel identifier for screening and fuzzy c-means cluster analysis for further evaluation. The Hampel identifier and fuzzy c-means clustering membership values are used as inputs. The outlierness of a data point is computed as a result of a 2-input/1-output fuzzy inference system. The overall fuzzy treatment framework is a generalized approach and can be modified to suit the application. The fuzzy treatment method was applied to benchmark penicillin production process data containing artificial data points with suspected outliers. The proposed method was able to detect the outliers in the process data with some irregularities. The results are presented along with a discussion on the advantages of this method as a flexible treatment of process data.