Abstract
In this paper, we propose a novel fault detection method for multivariate industrial processes. The method is based on AutoEncoder. AutoEncoder is a single-hidden-layer neural network that can learn low-dimensional nonlinear representations for high-dimensional data. In the proposed fault detection method, offline normal data are used to train an AutoEncoder, which is then used for online fault detection. The proposed method is compared with conventional methods on Tennessee Eastman process. The experimental results show that the proposed method is able to outperform other methods.
Full Text
Published Version
Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have
Schedule a call