The failure prognosis is crucial for industrial equipment in prognostics and health management field. The vibration signal is the commonly used data for failure prognosis. The conventional prognostic approaches have limitations to handle the features extracted from the vibration signal because of the large data quantity, complex feature relations, and limited degeneration mechanisms. In this paper, a deep learning-based approach is proposed to predict the failure of the complex equipment. To supply plenty of features, three different domain features are extracted from vibration signals. Next, these features are preprocessed and reconstructed by arctangent normalization and data stream, respectively. Finally, a deep neural network, namely, multistream deep recurrent neural network (MS-DRNN) is built to fuse these features for failure target. The presented deep neural network is hybrid, involving recurrent layer, fusion layer, fully connected layer, and linear layer. To benchmark the proposed approach, several prognosis approaches are evaluated with the testing data from six large bearing datasets. Simulation results demonstrate that the prediction performance of the MS-DRNN-based approach is effective and reliable.
scite is a Brooklyn-based organization that helps researchers better discover and understand research articles through Smart Citations–citations that display the context of the citation and describe whether the article provides supporting or contrasting evidence. scite is used by students and researchers from around the world and is funded in part by the National Science Foundation and the National Institute on Drug Abuse of the National Institutes of Health.