Data-driven quality monitoring is highly demanded in practice since it enables relieving manual quality inspection of the product quality. Conventional data-driven quality monitoring is constrained by its offline characteristic thus being unable to handle streaming nature of sensory data and nonstationary environments of machine operations. Recently, there have been pioneering works of online quality monitoring taking advantage of online learning concepts in the literature, but it is still far from realization of minimum operator intervention in the quality monitoring because it calls for full supervision in labelling data samples. This paper proposes Parsimonious Network++ (ParsNet++) as an online semisupervised learning approach being able to handle extreme label scarcity in the quality monitoring task. That is, it is capable of coping with varieties of semisupervised learning conditions including random access of ground truth and infinitely delayed access of ground truth. ParsNet++ features the one-pass learning approach to deal with streaming data while characterizing elastic structure to overcome rapidly changing data distributions. That is, it is capable of initiating its learning structure from scratch with the absence of a predefined network structure where its hidden nodes can be added and discarded on the fly in respect to drifting data distributions. Furthermore, it is equipped by a feature extraction layer in terms of 1D convolutional layer extracting natural features of multivariate time-series data samples of sensors and coping well with the many-to-one label relationship, a common problem of practical quality monitoring. Rigorous numerical evaluation has been carried out using the injection molding machine and the industrial transfer molding machine from our own projects. ParsNet++ delivers highly competitive performance even compared to fully supervised competitors.

No abstract

No abstract

A cross domain multistream classification is a challenging problem calling for fast domain adaptations to handle different but related streams in never-ending and rapidly changing environments. Notwithstanding that existing multistream classifiers assume no labelled samples in the target stream, they still incur expensive labelling cost since they require fully labelled samples of the source stream. This paper aims to attack the problem of extreme label shortage in the cross domain multistream classification problems where only very few labelled samples of the source stream are provided before process runs. Our solution, namely Learning Streaming Process from Partial Ground Truth (LEOPARD), is built upon a flexible deep clustering network where its hidden nodes, layers and clusters are added and removed dynamically in respect to varying data distributions. A deep clustering strategy is underpinned by a simultaneous feature learning and clustering technique leading to clustering-friendly latent spaces. A domain adaptation strategy relies on the adversarial domain adaptation technique where a feature extractor is trained to fool a domain classifier classifying source and target streams. Our numerical study demonstrates the efficacy of LEOPARD where it delivers improved performances compared to prominent algorithms in 15 of 24 cases. Source codes of LEOPARD are shared in https://github.com/wengweng001/ LEOPARD.git to enable further study.