Track: Modeling and Simulation

Abstract

Due to the rapid advancement of Industry 4.0, the increasing complexity of industrial applications leads to the expanding dimensionality of time series data. To maintain the performance, avoid economic losses, and ensure safety during the industrial processes, anomaly detection draws great attention. In view of advantages in dimensionality reduction and feature retention, autoencoder (AE) technology is widely applied for anomaly detection monitoring. In this work, considering both high dimensionality and dynamic relations between elements in the hidden layer, an improved autoencoder with dynamic hidden layer (DHL-AE) is proposed and applied for anomaly detection monitoring. Two case studies including Tennessee Eastman process and Wind data are used to show the effectiveness of the proposed algorithm. The results demonstrate that compared with classical AE approaches that are most commonly used, DHL-AE exhibits the best overall performance in anomaly detection monitoring.