Search

An artificial-intelligence (AI)-based method for fault diagnosis is a strong candidate for industrial applications in the health management of rolling bearings. However, traditional fault diagnosis methods fail to improve the detection accuracy because they only extract a single feature and have limitations in feature representation. In addition, advanced object detection frameworks such as region-based convolutional neural networks have not yet been applied in fault diagnosis. To this end, a fault diagnosis model using a Time-Frequency Region-Based Convolutional Neural Network (TF-RCNN) is proposed in this paper. This method was mainly adopted to extract multiple regions that can characterize fault features from the Time-Frequency Representation (TFR). Specifically, an attention module was introduced so the model could focus on representative features. The existing classification strategy was also enhanced to perform multiple types of fault classification. Finally, an end-to-end rolling bearing fault diagnosis framework based on the TF-RCNN was developed with the aforementioned improvements. The effectiveness of this method was proven experimentally on artificial faults and real faults. The superiority of the proposed method is demonstrated using a comparison with the typical object detection method and an advanced fault diagnosis method. [ABSTRACT FROM AUTHOR]

With the continuous development of artificial intelligence, data-driven fault diagnosis methods are gradually attracting widespread attention. However, in practical industrial applications, noise in the working environment is inevitable. This leads to the fact that the performance of traditional intelligent diagnosis methods is hardly sufficient to satisfy the requirements. In this paper, a developed intelligent diagnosis framework is proposed to overcome this deficiency. The main contributions of this paper are as follows: Firstly, a fault diagnosis model is established using EfficientNet, which achieves optimal diagnosis performance with limited computing resources. Secondly, an attention mechanism is introduced into the basic model for accurately establishing the relationship between fault features and fault modes, while improving the diagnosis accuracy in complex noise environments. Finally, to explain the proposed method, the weights and features of the model are visualized, and further attempts are made to analyze the reasons for the high performance of the model. The comprehensive experiment results reveal the superiority of the proposed method in terms of accuracy and stability in comparison with other benchmark diagnosis approaches. The diagnostic accuracy under actual working conditions is 86.24%. [ABSTRACT FROM AUTHOR]

The rolling bearing, though a widely used component in rotating machinery, is also failure-prone. There are some difficulties in extracting valuable fault features from the vibration signal generated by the rolling bearing operation. Based on the interval coded deep belief network, an improved method for fault diagnosis is introduced to deal with the existing problems. Interval code is an encoding method to characterize fault features by using bearing vibration amplitude distribution. This method can also reduce the complexity of the original data and integrate input dimension. The deep belief network can remove the dependence from manual feature extraction and quickly extract the fault features from coding results. This proposed method is employed in the fault diagnosis experiments of rolling bearing to evaluate the performance. From the obtained experiment results, it can be concluded that the interval coded deep belief network boosts performance, which can eliminate the limitation of the input data dimension in the traditional deep learning model. [ABSTRACT FROM AUTHOR]

• A fault signal detection method based on object detection is proposed. • Using the fault feature regions (FFRs) to detect the bearing fault impact features. • Realizing the compound fault diagnosis of bearing with single fault data. • Showing the superior performance of accuracy and efficiency of fault diagnosis. Bearing faults of rotating machinery are common compound faults, and diverse fault categories are coupled, which makes it challenging to achieve state monitoring. For this purpose, a fault diagnosis method based on fault feature region (FFR) detection for bearings is proposed. The key point is an FFR proposal network for several regions with fault features from training samples. This process only uses the single fault data. These obtained regions are used as the training dataset of a module based on a deep belief network. In the application, if the output probabilities satisfy the discrimination terms, then an untrained compound fault state is output. Furthermore, the bearing compound fault dataset is employed to assess the diagnosis performance. The results reveal that the diagnosis accuracy of compound faults and the overall accuracy exceeds 80%. This fascinating discovery proves the superiority of the proposed approach to achieving compound fault diagnosis for rolling bearings. [ABSTRACT FROM AUTHOR]

• A Bi-directional deep belief network is developed for bearings fault diagnosis. • Quantum genetic algorithm is utilized to get the optimal model. • The effectiveness of the proposed method is verified on a printing press platform. Intelligent fault diagnosis model based on machine learning algorithm has extensive application, while the unsatisfied training data quality leads to the lower accuracy of intelligent fault diagnosis and inferior generalization performance. To address this problem, a method called Bi-directional deep belief network (Bi-DBN) is developed for fault diagnosis of rolling bearings. The forward training part of Bi-DBN can learn the fault features from the original vibration signals first, and then the reverse generation part synthesizes samples according to the weight sharing by the forward training part. The noise time-shift layer is innovatively introduced to reduce the similarity between the synthesized sample and the original sample. Finally, Quantum genetic algorithm (QGA) is applied to optimize the key parameters of Bi-DBN to improve the feature learning efficiency. The experimental results of rolling bearing signals indicate that the developed method significantly limits the effect of training data quality on the diagnosis accuracy. [ABSTRACT FROM AUTHOR]