Skip to main content Skip to main navigation


Deep learning-based clustering of processes and their visual exploration: An industry 4.0 use case for small, medium‐sized enterprises

Nijat Mehdiyev; Lea Götz; Johannes Lahann; Peter Fettke
In: Expert Systems (EXSY), Vol. 39, Pages 1-24, Wiley, 2022.


This paper proposes a multi-stage approach consisting of deep learning-based imageclassification, process trace clustering, and visual/statistical knowledge discovery ofprocess data. The proposed decision augmentation solution aims to facilitate the pro-duction planners in estimating the process-specific production parameters such asactivity duration, idle time, or machine utilization. This study focuses on‘one-of-a-kind production’(OKP). Planning in OKP is especially challenging due to the increas-ing individualization of customer requirements. Furthermore, the uniqueness of prod-ucts adds complexity to data and information structuring. To tackle this issue, wefirst train deep convolutional neural networks (CNN) with image data of productionparts obtained from computer-aided design (CAD) systems to extract meaningful fea-tures. After cross-validation, uncertainty, and robustness assessment of the adopteddeep learning approach, we use the data representation from the penultimate layeras input for clustering production parts. The goodness of clustering results is evalu-ated using a series of internal clustering validation indices. Finally, process event logdata provided by manufacturing execution systems (MES) is mapped to each produc-tion part, allowing us to conduct statistical and visual knowledge discovery of processparameters for each cluster. The relevance of our proposed approach has been vali-dated by studying a real-world use case in a small, medium-sized enterprise (SME)operating in the fixture and jig manufacturing industry.