Anomaly detection using control flow pattern and fuzzy regression in port container handling

Abstract

Deviations in port container handling can be detected by many factors. One of them is anomalies in the process model. Several studies have proposed anomaly detection methods. However, these methods do not accommodate verbal judgments of experts. These methods treat instances with low deviation as containing anomalies while in reality not all instances with low deviation contain anomalies. Considering this, a method was developed for detecting anomalies in port container handling using fuzzy regression in order to accommodate verbal expert judgments on the rate of anomaly (ROA). First, a control flow pattern is built to form an anomaly pattern that will be used for detecting wrong patterns. Then, five anomaly attributes were declared, i.e. skip sequence, wrong throughput time (max), wrong throughput time (min), wrong patterns and wrong decisions. In the experiment, the rate of anomaly was found using three methods, namely fuzzy regression (FR), support vector regression (SVR) with radial basis function (RBF) kernel, and multiple linear regression (MLR). The results showed that fuzzy regression was better at detecting anomalies than multiple linear regression and support vector regression. The experimental validation showed that fuzzy regression combined with control flow pattern was able to reduce false positives and false negatives. The sensitivity, specificity and accuracy of the proposed method were 96%, 97% and 99%, respectively.