No abstract

Impact load is the load that machines frequently experienced in engineering applications. Its time-history reconstruction and localization are crucial for structural health monitoring and reliability analysis. However, when identifying random impact loads, conventional inversion methods usually do not perform well because of complex formula derivation, infeasibility of nonlinear structure, and ill-posed problem. Deep learning methods have great ability of feature learning and nonlinear representation as well as comprehensive regularization mechanism. Therefore, a new feature learning-based method is proposed to conduct impact load reconstruction and localization. The proposed method mainly includes two parts. The first part is designed to reconstruct impact load, named convolutional-recurrent encoder–decoder neural network (ED-CRNN). The other part is constructed to localize impact load, called deep convolutional-recurrent neural network (DCRNN). The ED-CRNN utilizes the one-dimensional (1-D) convolutional encoder–decoder to obtain low-dimension feature representations of input signals. Two long short-term memory (LSTM) layers and a bidirectional LSTM (BiLSTM) layer are uniformly distributed in this network to learn the relationship between input features and the output load in time steps. The DCRNN is constructed mainly by two 1-D convolutional neural network (CNN) layers and two BiLSTM layers to learn high-hidden-level spatial as well as temporal features. The fully connected layers are placed at the end to localize an impact load. The effectiveness of the proposed method was demonstrated by two numerical studies and two experiments. The results reveal that the proposed method has the ability to accurately and quickly reconstruct and localize the impact load of complex assembled structure. Furthermore, the performance of the DCRNN is related to the number of sensors and the architecture of the network. Meanwhile, the strategy of alternating layout is proposed to reduce the number of training locations.

No abstract

Intelligent diagnosis is an important manner for mechanical fault diagnosis in the era of industrial big data, and deep network has received extensive attention in this field because of automatically learning features and classifying entered samples. As a classic deep learning model, Convolutional Neural Network has been applied in mechanical intelligent fault diagnosis. However, the limitation is that entered samples must be balanced to achieve satisfactory recognition rate. During the operation of machinery, the normal samples are abundant and the fault samples are rare. Therefore, the recognition rate of the minority category is minor when processing the imbalanced data with Convolutional Neural Network. To solve the above problem, an intelligent classification method for imbalanced mechanical data based on Deep Cost Adaptive Convolutional Network is proposed. According to this model, first, it learns intrinsic state characteristics in mechanical raw signals through multiple convolution and pooling operations. Second, it maps these characteristics to mechanical health condition by fully connected layers. Finally, the cost adaptive loss function adaptively assigns different misclassification costs for all categories and keeps updating them in training process to effectively classify the imbalanced mechanical data. The proposed method is verified by bearing data and milling cutter data with different imbalanced ratio, and compared with other methods. The experimental results show that the proposed method is robust and is able to effectively classify the imbalanced mechanical data.

No abstract