Search

Fault prediction technology provides a way to reduce the loss caused by equipment failure. Currently, many efforts of failure prediction pay more attention to the fault trend and the remaining useful life. This paper devotes to predicting a specific failure mode of device in the future. We propose a two-step prediction method with two sub-models to predict the fault mode. The first model, named regression model, combines Attention-based-LSTM (Long Short-Term Memory) and Random Forest (RF) to predict time series trends, where wavelet packet analysis is adopted to extract elementary features of time series and form multivariable time series. Based on the status data forecasted by the first model, the second model, named classification model, combines Attention-based-LSTM and Extra-Tree (ET) to classify fault mode. The classified fault will occur in the device at some point in the future. In the regression model, Attention-based-LSTM performs deep level feature extraction of vibration sensor signals, and the result is fed to the last layer (i.e., an RF layer), which takes responsibility for predicting the future sequence. The future sequence is then provided to the classification model, where Attention-based-LSTM is used to extract implicit features, and ET classifier is used to accept these features and figure out the failure phenomenon. Our approach for predicting failure modes includes the severity and type of fault, which are rarely investigated synthetically. The experimental results of bearing data show that the sub-models can achieve high prediction accuracy and our method can predict the future failure mode with precision.

This work proposes an efficient framework for prediction of filled elastomer damping properties based on imaged microstructures. The efficiency of this method stems from a hierarchical multiscale modeling scheme, in which the constitutive response of subcell regions, smaller than a representative volume element (RVE), are determined using micromechanics; the resulting constitutive parameters then act as inputs to finite element simulations of the RVE, from which damping properties are extracted. It is shown that the micromechanics models of Halpin–Tsai and Mori–Tanaka are insufficient for modeling subcells with many filler clusters, and thus these models are augmented by an additional interaction term, based on stress concentration factors. The multiscale framework is compared to direct numerical simulations in two dimensions and extended to predictions for three dimensional systems, which include the response of matrix–filler interphase properties. The proposed multiscale framework shows a significant improvement in computational speed over direct numerical simulations using the finite element method, and thus allows for detailed parametric studies of microstructural properties to aid in the design of filled elastomeric systems.

In order to find all occurrences of a twig pattern in XML documents, a considerable number of twig pattern matching algorithms have been proposed. Previous algorithms mainly focus on the conjunctive twig queries whose sibling edges are only connected by AND connectives. However, meaningful twig queries typically contain arbitrarily specified compound predicates including AND, OR and NOT in practical applications. Moreover, as far as we know, none of these twig matching algorithms have examined the processing of twig queries which contain all the compound predicates over fuzzy XML data. In this paper, we present the first study on evaluating twig queries with AND, OR and NOT connectives in fuzzy XML. We propose a novel holistic twig matching algorithm called LTwig for answering these complex queries. Our algorithm guarantees that the answers can be obtained by scanning the relevant data of the data streams associated with the nodes appearing in the twig pattern only once. A comprehensive set of experiments is finally carried out to demonstrate the effectiveness and efficiency of our proposed approach.