Rotating machinery plays a critical role in many significant fields. However, the unpredictable machinery faults may lead to the severe damage and losses. Hence, it is of great value to explore the precise approaches for fault diagnosis. With the development of the intelligent fault diagnosis methods based on deep learning, convolutional neural network (CNN) has aroused the attention of researchers in machinery fault diagnosis. In the light of the reduction of difficulty in feature learning and the improvement of final diagnosis accuracy, data preprocessing is necessary and crucial in CNN-based fault diagnosis methods. This review focuses on CNN-based fault diagnosis approaches in rotating machinery. Firstly, data preprocessing methods are overviewed. Then, we emphatically analyze and discuss several main techniques applied in CNN-based intelligent diagnosis, principally including the fast Fourier transform, wavelet transform, data augmentation, S-transform, and cyclic spectral analysis. Finally, the potential challenges and research objects are prospected on data preprocessing in intelligent fault diagnosis of rotary machinery.

/pdf/data-preprocessing-techniques-in-convolutional-neural-11gl3o9h4l.pdf

Data Preprocessing Techniques in Convolutional Neural Network Based on Fault Diagnosis Towards Rotating Machinery

In engineering practice, mechanical equipment is usually in polytropic working conditions, where the data distribution of training set and test set is inconsistent, resulting in insufficient generalization ability of the intelligent diagnosis model. Simultaneously, different tasks often need to be modeled separately. Domain adaptation, as one of the research contents of transfer learning, has certain advantages in solving the problem of inconsistent feature distribution. This article designs and establishes a domain adaptation framework based on multiscale mixed domain feature (DA-MMDF) for cross-domain intelligent fault diagnosis of rolling bearings under polytropic working conditions. The proposed method first uses the MMDF extractor to obtain features from the collected data, which constructs a complete feature space through variational mode decomposition (VMD) and mixed domain feature extraction to fully mine the state information and intrinsic attributes of the vibration signal. Second, the dimensionality reduction and optimization of features are achieved through extreme gradient promotion, and meaningful and sensitive features are selected according to the importance of features to eliminate redundant information. The optimized important features are combined with the manifold embedded distribution alignment method to realize the distribution alignment of data in different fields and cross-domain diagnosis. In order to verify the effectiveness of the proposed approach, the rolling bearing data sets gathered from the laboratories are employed and analyzed. The analysis result confirms that DA-MMDF is able to achieve effective transfer diagnosis between polytropic working conditions. Compared with traditional intelligent fault diagnosis methods and DA methods, the method proposed in this article achieved the state-of-the-art performances.

An Intelligent Fault Diagnosis Method Based on Domain Adaptation and Its Application for Bearings Under Polytropic Working Conditions

Accuracy-improved bearing fault diagnosis method based on AVMD theory and AWPSO-ELM model

Adaptive energy-constrained variational mode decomposition based on spectrum segmentation and its application in fault detection of rolling bearing

Intelligent fault diagnosis of diesel engine via adaptive VMD-Rihaczek distribution and graph regularized bi-directional NMF

Fault diagnosis for rolling bearing based on VMD-FRFT

The invention disclose a vehicle-mounted wheel rail displacement image detection method and detection device based on double laser sources, and belongs to the technical field of railway vehicle profile measurement devices and measurement methods. The detection method comprises the following steps that a first laser source and a second laser source are fixed on the central position P of a wheel set, a laser point A emitted form the first laser source is located at a first steel rail surface, a laser point B emitted from the second laser source is also located at the first steel rail surface, and the connecting line between the two laser points of A and B and the edge of the first steel rail surface are perpendicular; a camera is installed on a vehicle body right above the region of the twolaser points through a support, a principal optic axis of the camera is perpendicular to the steel rail surface, and during the train running, images of the steel rail surface where the laser point Aand the laser point B are located are collected by the camera. According to the method, non-contact measurement is used to skillfully solve the detection of complex contact parameters of the wheel rail, the method is simple, the computation speed is high, and the feasibility is high.

Vehicle-mounted wheel-rail displacement image detection method and detection device based on double laser sources

The invention discloses a rolling bearing fault diagnosis method based on VMD-FRFT (Variational Mode Decomposition-Fractional Fourier Transform), and relates to the technical field of the fault diagnosis method of a bearing. The method comprises the following steps that: extracting the vibration signal of a rolling bearing, and processing the vibration signal to obtain a multi-component LMF (Linear Frequency Modulation) vibration signal; carrying out VMD-FRFT on the multi-component LMF vibration signal to obtain the modal component of the vibration signal; according to the obtained modal component of the vibration signal, reconstructing the vibration signal; and carrying out wavelet transform on the reconstructed vibration signal to obtain an instantaneous frequency conversion curve, and according to the instantaneous frequency conversion curve, obtaining the fault information of the rolling bearing. The method which is put forward is successfully applied to rolling bearing inner ringfault data analysis, instantaneous fault characteristic frequency can be accurately extracted, and noise robustness is high.

Rolling bearing fault diagnosis method based on VMD-FRFT (Variational Mode Decomposition-Fractional Fourier Transform)

The invention discloses a twist image detection method based on laser source positioning and relates to the technical field of devices and methods to monitor railway track lines. The twist image detection method comprises the steps of mounting two laser scanning sensors on two sides in front of a track detection cart, and allowing laser beams to irradiate specified positions of a rail flange of awheel pair in the front, wherein laser beam outlets are perpendicular to the same positions of a rail flange at the inner side of the track; acquiring and returning positional data by a computer, splicing the multiple sets of data, processing the data, fitting a rail flange curve function, and calculating a vertical displacement in the presence of twist; acquiring differences between the corresponding positions via the two sets of data, marking whether points are convex or concave points, calculating through data packet time and specified velocity to judge whether the points are in the basal length range of dynamic twist measurement. Non-contact laser measurement has high measuring precision and high speed, and the method is more feasible for long-line railways.

Twist image detection method based on laser source positioning

The invention discloses a self-adaptive signal decomposition method, and relates to the technical field of signal processing methods and is used for noise reduction and decomposition of multi-component complex signals. The method comprises the following steps of converting an input signal into a frequency domain, initializing each variable in the frequency domain; carrying out circulated updatingon each component, and the center frequency, bandwidth parameters and Lagrangian coefficient of each component according to an updating formula, increasing k in unit step size, until the sum of all the components obtained by updating and the mutual entropy of the input signals is smaller than a threshold value, stopping the inner circulation; repeating the above steps until an error limit requirement is met; and finally realizing automatic determination of the number K of the components and the bandwidth parameters, wherein the obtained components are all effective components, and the sum of all the components is an input signal. According to the method, no priori knowledge is needed to set parameters in advance, and the method has complete self-adaptability The method is high in noise robustness, and high in decomposition efficiency, and finally effective single-component signals are obtained.

Kang De

Papers

Fault diagnosis for rolling bearing based on VMD-FRFT

Vehicle-mounted wheel-rail displacement image detection method and detection device based on double laser sources

Rolling bearing fault diagnosis method based on VMD-FRFT (Variational Mode Decomposition-Fractional Fourier Transform)

Twist image detection method based on laser source positioning

Self-adaptive signal decomposition method