Autocorrelation Integrated Gaussian Based Anomaly Detection using Sensory Data in Industrial Manufacturing

In industrial processes, early detection of anomalies is crucial for reducing process failures, meeting the quality assurance (QA) requirements, and lowering raw material wastage. Therefore, anomaly detection algorithms should identify an anomaly in a timely manner, and hence, allows immediate corrective actions to be applied. In this context, this paper proposes a low-complexity algorithm for detecting anomalies in industrial steelmaking furnaces operation. The algorithm utilizes the vibration measurements collected from several built-in sensors to compute the temporal correlation using the autocorrelation function (ACF). Furthermore, the proposed model parameters are tuned by solving multi-objective optimization using a genetic algorithm (GA). The proposed algorithm is tested using a practical dataset provided by an industrial steelmaking plant. The obtained results show that the proposed algorithm outperforms the support vector machine (SVM) and random forest (RF) algorithms in most key performance measures with the advantage of a substantial decrease in training and execution times.