Knowledge extraction from deep convolutional neural networks applied to cyclo-stationary time-series classification
Modelling complex processes from raw time series increases the necessity to build DeepLearning (DL) architectures that can manage this type of data structure. However, as DLmodels become deeper, larger and more diverse datasets are necessary and knowledgeextraction will become more difficult. In an...
| Autores: | , , , , |
|---|---|
| Tipo de recurso: | artículo |
| Estado: | Versión enviada para evaluación y publicación |
| Fecha de publicación: | 2020 |
| País: | España |
| Institución: | Universidad de Sevilla (US) |
| Repositorio: | idUS. Depósito de Investigación de la Universidad de Sevilla |
| OAI Identifier: | oai:idus.us.es:11441/107225 |
| Acceso en línea: | https://hdl.handle.net/11441/107225 https://doi.org/10.1016/j.ins.2020.03.039 |
| Access Level: | acceso abierto |
| Palabra clave: | Deep learning Convolutional Neural Networks (CNN) Cyclo-stationary time-series analysis Knowledge extraction Fault diagnosis |
| Sumario: | Modelling complex processes from raw time series increases the necessity to build DeepLearning (DL) architectures that can manage this type of data structure. However, as DLmodels become deeper, larger and more diverse datasets are necessary and knowledgeextraction will become more difficult. In an attempt to sidestep these issues, in this pa- per a methodology based on two main steps is presented, the first being to increase sizeand diversity of time-series datasets for training, and the second to retrieve knowledgefrom the obtained model. This methodology is compared with other approaches reportedin the literature and is tested under two configuration setups of Condition-Based Mainte- nance problems: fault diagnosis of bearing, and fault severity assessment of a helical gear- box, obtaining not only a performance improvement in comparison, but also in retrievingknowledge about how the signals are being classified. |
|---|