Development of Data-based Air Leakage and Location Detection Technology in Pneumatic Facilities
Embedded Files
Funding
ETRI
Role
Participating Researcher
(Sep 2023 - Dec 2023)
Task
-A deep learning approach for early air leakage detection in IoT-connected air compressors, further addressing imbalanced data and suitable for deployment on edge servers (under Sejong University)
Overview
A dropout-enabled deep neural network specially designed for air leakage detectionÂ
Strategically leverages Monte Carlo Dropout, an extension of the dropout regularization method known for its effectiveness in enhancing generalization and mitigating overfitting in neural networks
An unsupervised-enhanced data sampling method called UEDSM to address class imbalance
This method fully integrates three main components: principal component analysis, k-means clustering, and cluster similarity scoring
Page updated
Google Sites
Report abuse