Convolutional neural network has been widely used in fault diagnosis of mechanical devices. In particular, a 2-D convolutional neural network requires manual selection of multiscale transformation to transform vibration signal into the 2-D structure. Although a 1-D convolution neural network can directly use the vibration signal for convolution processing, it cannot make full use of the nonlinear information in the 1-D space. In order to make full use of the advantages of 1-D and 2-D convolutional neural networks, in this article, we develop a one-dimension in tandem with 2-D joint convolutional neural network (1D–2D JCNN) for rotating machinery fault diagnosis. More specifically, 1-D convolution is employed to adaptively obtain the multiscale feature vectors of the vibration signal, and these feature vectors are constructed into 2-D maps, and then these 2-D vectors are used as the input of the 2-D convolutions neural network. Take the cross-entropy loss function as the loss function and use the error back propagation algorithm to optimize the filter parameters of the 1D–2D JCNN model to obtain the final fault diagnosis model. Using the motor bearing dataset and the worm gearbox dataset, the experimental results show the excellent classification performance of bearings and gears under different working conditions. The average diagnostic accuracy of ten runs on Case Western Reserve University bearing dataset is 99.92%, and the variance is 3.96e-6. The average diagnostic accuracy of ten runs on the worm gear dataset is 99.82%, and the variance is 7.82e-6. Compared with the traditional fault diagnosis model and the latest convolution neural network method, the 1D–2D JCNN shows better diagnosis performance.

Multifractal detrended fluctuation analysis (MF-DFA) has been used for vibration-based fault diagnosis because it is able to uncover multifractality buried in nonlinear and nonstationary vibration signals and thus offers an opportunity to explore a new set of multifractal features for fault diagnosis. However, the choice of detrending polynomial orders is one of the major concerns in MF-DFA because improper polynomials can cause the underfitted or overfitted scale-dependent trend of the signals. To address this issue, adaptive MF-DFA (AMF-DFA) is developed in this paper. More specifically, the developed AMF-DFA uncovers multifractality of the signals by adaptively extracting a variable number of scale-dependent fluctuations present in the signals and automatically eliminating irrelevant trend components to the fundamental structure of the signals based on correlation analysis. Accordingly, the developed AMF-DFA does not require a priori knowledge (i.e., detrending polynomial order). The effectiveness of the developed AMF-DFA was verified for fault diagnosis applications.

As vital renewable energy devices, wind turbines suffer from gearbox failures due to harsh speed increasing operations. Therefore, gearbox fault diagnosis is crucial for wind turbine maintenance with reducing economic costs. However, obtaining faulty data is rather challenging, especially at the early fault stage. For this reason, a sparse isolation encoding forest (SIEF) is proposed aiming at both anomaly detection and novel fault discrimination for wind turbine gearboxes. In the present SIEF method, a sparse autoencoder (SAE) is first trained with only normal data to obtain an optimized and robust weight structure. Newly acquired data corresponding to faulty or healthy conditions are sent to this encoder for feature extraction by encoding to its low-dimensional space. All the data in low-dimensional space are fed to an isolation forest (IF) for anomaly detection and novel fault discrimination. In the addressed SIEF approach, only normal data are required to train the model for fault detection and further discrimination. It is consistent with the actual operations of the wind turbines, with much less dependence on the fault data for model training. The proposed method was evaluated by fault diagnosis tests on wind turbine gearboxes. Results show good performances of the proposal compared with peers.

Induction motors, the key equipment for rotating machinery, are prone to compound faults, such as a broken rotor bars and bearing defects. It is difficult to extract fault features and identify faults from a single signal because multiple fault features overlap and interfere with each other in a compound fault. Since current signals and vibration signals have different sensitivities to broken rotor and bearing faults, a multi-channel deep convolutional neural network (MC-DCNN) fault diagnosis model based on multi-source signals is proposed in this paper, which integrates the original signals of vibration and current of the motor. Dynamic attenuation learning rate and SELU activation function were used to improve the network hyperparameters of MC-DCNN. The dynamic attenuated learning rate can improve the stability of model training and avoid model collapse effectively. The SELU activation function can avoid the problems of gradient disappearance and gradient explosion during model iteration due to its function configuration, thereby avoiding the model falling into local optima. Experiments showed that the proposed model can effectively solve the problem of motor compound fault identification, and three comparative experiments verified that the improved method can improve the stability of model training and the accuracy of fault identification.

Machinery remaining useful life (RUL) prediction plays a pivotal role in modern industrial maintenance. Traditional methods entail the manual selection of useful features, which requires prior knowledge and lack adaptability to diverse cases. Moreover, as features may have different relevance to the degradation process at various stages, the prognostic performance will be limited by the utilization of fixed features throughout the full life-time. Additionally, most deep learning methods lack the perception of global information of features, which is critical to RUL prediction. To tackle these issues, an adaptive feature utilization method with separate gating mechanism and global temporal convolutional network (SGGTCN) is proposed in this paper. First, a separate gating mechanism is proposed to adaptively model temporal information within each feature individually through a series of designed separate gated residual modules. Second, an adaptive feature utilization method is proposed to evaluate and dynamically weight feature importance. Third, a global temporal convolutional network is proposed to model and fuse global temporal information for comprehensive sequential modeling. The effectiveness and superiority of the proposed method are validated by two prognostic case studies of turbofan engines and bearings.