Vibration signals always contain noise and irregularities, which makes spectrum analysis difficult to extract high-level features. Recently, graph theory has been applied improve the performance of feature extraction. By converting raw data into graphs, hidden structural topological information can be obtained. In this article, a spatial-temporal graph-based extraction, called SuperGraph, for rotating machinery fault diagnosis is proposed. Specifically, theory-based used construct graph....

Lithology identification based on conventional well-logging data is of great importance for geologic features characterization and reservoir quality evaluation in the exploration production development petroleum reservoirs. However, there are some limitations traditional lithology process: (1) It very time consuming to build a model so that it cannot realize real-time during well drilling, (2) must be modeled by experienced geologists, which consumes lot manpower material resources, (3)...

Due to its ability learn the relationship among nodes from graph data, convolution network (GCN) has received extensive attention. In machine fault diagnosis field, it needs construct input graphs reflecting features and relationships of monitoring signals. Thus, quality affects diagnostic performance. But still two limitations: 1) constructed usually redundant edges, consuming excessive computational costs; 2) cannot reflect between noisy signals well. order overcome them, a dynamic...

Graph data-driven machine fault diagnosis methods make success using sufficient data recently. However, in the actual industry, there are rare failure historical data, leading to insufficient graph representation ability and reducing performance. In this article, a generalized contrastive learning (GCL) framework for few-shot is proposed. First, spectrum features of vibration data-based samples used calculate Euclidean distance matrix constructing K-nearest neighborhood (KNNG), where <italic...

Colon polyps have a greater chance of developing into colon cancer, and colonoscopy is one the most commonly used methods to detect polyps. However, effectiveness depends largely on technical level physicians, there are fewer experienced physicians. Besides, traditional artificial intelligence cannot obtain unified model with good results for all patients. To solve these problems, polyp detection segmentation method based mask regions convolutional neural network (MRCNN) precise region...

Transfer learning-based fault diagnosis methods borrow source-domain knowledge to achieve task for the unlabeled target domain. However, existing research articles mainly lie in feature mapping and model transfer, ignoring relationship between cross-domain samples. Once connections samples with same label can be constructed, propagation will easier even if there is a distribution discrepancy. In this article, transfer graph-driven rotating machinery considering construction proposed....

Permeability estimation plays an important role in reservoir evaluation, hydrocarbon development, etc. Traditional methods have problems of time consuming and high cost. At present, the application machine learning are more extensive, however, some models developed for permeability fewer samples, requiring prior knowledge, parameters need to be calculated indirectly. To this end, based on a certain scale dataset, hybrid method embedded feature selection light gradient boosting (EFS-LightGBM)...

Modern detectors in remote-sensing images follow the pipeline that feature maps extracted from ConvNets are shared between classification and regression tasks. However, there exist obvious conflicting demands multiorientation object detection of (RSOD) is insensitive to orientations, while quite sensitive. In addition, previous works cannot promise reliability rotation-invariant or rotation-equivariant features with only qualitative intuitive analysis. To address these issues, we propose an...

Abstract Rotating machinery is a primary element of mechanical equipment, and thus fault diagnosis its key components very important to improve the reliability safety modern industrial systems. The point diagnose faults these extract effectively hidden information. However, actual vibration signals rotating have nonlinear non-stationary characteristics, so traditional signal decomposition methods are unable frequency accurately, leading spectrum overlap decomposed sub-signals. Therefore,...

Bridge piers on river channels can cause obstacles for flood flow by reducing the cross-sectional area and inducing local eddy currents high velocities, which may destroy hydraulic structures. A two-dimensional numerical model was used to investigate effects of bridge hazards in Jialing River, China. For modeling, Mike 21 FM used, is an unstructured mesh finite volume that solves shallow water equations. The validated with collected historical traces, sensitivity analyses identified Manning...

Graph deep learning-based prognostic methods have been successfully applied in bearing remaining useful life (RUL) prediction, as graph represents spatial and temporal dependencies of signals. However, data-driven prediction using single-sensor data are still insufficiently studied. And the construction is not interpretable, where physical meaning edges unclear. To overcome these limitations, a node-level PathGraph-based RUL method proposed, Chebyshev convolutional network (ChebCGN) with...

By learning effective information from unlabeled nodes, node-level graph data-driven diagnosis methods perform better than graph-level methods. However, features of indirectly involved in feature learning, are not fully utilized. To overcome aforementioned limitations, a semisupervised machine fault fusing unsupervised contrastive (GCL) is proposed. A new GCL framework, where positive and negative graphs generated by calculating Pearson correlation coefficient, fused into the transformer...

The hydropower generator unit (HGU) is a vital piece of equipment for frequency and peaking modulation in the power grid. Its vibration signal contains wealth information status characteristics. Therefore, it important to predict tendency HGUs using collected real-time data, achieve predictive maintenance as well. In previous studies, most prediction methods have only focused on enhancing stability or accuracy. However, insufficient consider one criterion (stability accuracy) prediction....

Abstract Deep learning-based rotating machinery remaining useful life (RUL) prediction approaches rarely consider spatial dependencies and global temporal correlation of monitoring signals simultaneously. Superiorly, graph convolutional networks (GCNs) learn relationship information among nodes, considering the signals. It is beneficial for constructing high-quality graphs to improve performance in single-sensor scenarios. In this paper, an RUL approach based on a dynamic spatial–temporal...

Unsupervised domain adaptation (UDA) has been widely exploited for machinery fault diagnosis (MFD). However, existing UDA approaches always require direct access and sharing to the labeled source domain, raising privacy concerns. In this paper, a practical challenging scenario, source-free (SFUDA), is considered privacy-preserving MFD. SFUDA, only pre-trained model provided unlabeled target data inaccessible during adaptation. A novel SFUDA approach, namely cluster (SF-CA), proposed, which...