A deep learning unsupervised approach for fault diagnosis of household appliances

Fault detection and fault diagnosis are crucial subsystems to be integrated within the control architecture of modern industrial processes to ensure high quality standards. In this paper we present a two-stage unsupervised approach for fault detection and diagnosis in household appliances. In particular a suitable testing procedure has been implemented on a real industrial production line in order to extract the most meaningful features that allow to efficiently classify different types of fault by consecutively exploiting deep autoencoder neural network and k-means or hierarchical clustering techniques.