Rotating machinery usually operates under variable-speed conditions, and how to effectively handle nonstationary signal in fault diagnosis is a critical task. The time-frequency analysis (TFA) method is widely used in the processing of nonstationary signal. To improve the time-frequency resolution and clearly identify instantaneous frequency (IF) characteristics, the adaptive generalized S-synchroextracting transform (AGSSET), which is a novel TFA method proposed in this paper. Firstly, a new transform named adaptive generalized S-transform (AGST) is put forward by optimizing the window function of generalized S transform. In this paper, an adaptive window function optimization method based on the frequency spectrum of the vibration signal is introduced, and the energy concentration measure is used to determine the window function’s parameters in AGST. Simultaneously, the synchrony extraction idea is incorporated into the AGST, then the AGSSET is derived. To address more complex IF characteristics, the synchronous extraction operator (SEO) is reconstructed. In the simulation experiment, the GMLC signal model is selected to represent nonstationary signal and to verify the effectiveness of the proposed method. In addition, bearing fault data is also used for fault diagnosis experiments. The results of both numerical simulation and experimental analysis indicate that AGSSET performs well in identifying the time-varying IF characteristic in nonstationary signals. It can also efficiently detect faults with high accuracy and strong stability.

Star sensors are easily disturbed by stray light when working in the space environment: the quality of star image can be seriously reduced and the difficulty of star image preprocessing can be greatly increased. In this paper we propose a novel method for eliminating stray light interference based on the grayscale difference of star images. Firstly, since the grayscale difference between the edge of a star spot and the surrounding background pixels was very large, a grayscale forward difference was performed on the star image row by row. Based on the equivalent difference star image obtained, we could propose an edge pixel criterion for the star spot. Secondly, Stray light interference was eliminated from the star image through a connected component analysis based on the edge pixels criterion for the star spot. The relatively complete star spot areas were marked on the marker map. The simulation results demonstrate that the star spot extraction rate achieved through the proposed method was much higher than that obtained through traditional star image preprocessing methods. As long as the star spot is not submerged by the stray light, the proposed method is robust enough to ensure a successful extraction of the star spot.

In intelligent fault diagnosis for rolling bearings, the variable operating conditions of rotating machinery lead to inconsistent distributions between training and testing data, inevitably decreasing fault diagnosis accuracy. To overcome this issue, we propose a bearing fault diagnosis method based on transfer learning with a dual-flow manifold residual network and improved capsule network (DMRCN). Firstly, we introduce a dual-flow manifold residual network structure consisting of continuous wide-kernel convolution, continuous narrow-kernel convolution, and manifold learning spatial information description blocks. This architecture is utilized to extract low-level features from the data. Then, we present a novel architecture of the capsule network with self-attention output. We incorporate a self-attention mechanism into the dynamic routing algorithm to transfer richer high-level features from the low-level capsule layer to the higher-level capsule layer. During training, we propose a loss function known as global-local distribution discrepancy loss to enhance the model’s generalization ability and accelerate model convergence. Finally, we validate the performance of DMRCN in cross-condition transfer fault diagnosis on both the Case Western Reserve University public dataset and a laboratory-built bearing experimental rig dataset. Experimental results demonstrate that DMRCN outperforms other commonly used deep transfer learning methods regarding fault diagnosis across different operating conditions.

In response to the challenges posed by complex and variable operating conditions of rolling bearings and the limited availability of labeled data, both of which hinder the effective extraction of key fault features and reduce diagnostic accuracy, this study introduces a model that combines a spatial attention (SA) mechanism with a multi-scale depthwise separable convolution module. The proposed approach first employs the Gramian angular difference field (GADF) to convert raw signals. This conversion maps the temporal characteristics of the signal into an image format that intrinsically preserves both temporal dynamics and phase relationships. Subsequently, the model architecture incorporates a spatial attention mechanism and a multi-scale depthwise separable convolutional module. Guided by the attention mechanism to concentrate on discriminative feature regions and to suppress noise, the convolutional component efficiently extracts hierarchical features in parallel through the multi-scale receptive fields. Furthermore, the trained model serves as a pre-trained network and is transferred to novel variable-condition environments to enhance diagnostic accuracy in few-shot scenarios. The effectiveness of the proposed model was evaluated using bearing datasets and field-collected industrial data. Experimental results confirm that the proposed model offers outstanding fault recognition performance and generalization capability across diverse working conditions, small-sample scenarios, and real industrial environments.

As a low-carbon and clean energy source, Coalbed methane (CBM) is of great significance in reducing greenhouse gas emissions, optimizing the energy structure, safeguarding mine safety, and promoting the transformation to a green economy to achieve sustainable development. Coalbed methane (CBM) in Xinjiang’s steeply dipping coal seams is abundant but difficult to predict due to complex geology and distinct gas flow behaviors, making traditional methods ineffective. This study proposes GCN-BiGRU, a parallel dual-module model integrating seepage mechanics, reservoir engineering, geological structures, and production history. The GCN module models wells as nodes, using geological attributes and spatial distances to capture inter-well interference; the BiGRU module extracts temporal dependencies from production sequences. An adaptive fusion mechanism dynamically combines spatiotemporal features for robust prediction. Validated on Baiyanghe block data, the model achieved MAE 59.04, RMSE 94.25, and improved accuracy from 64.47% to 92.8% as training wells increased from 20 to 84. It also showed strong transferability to independent sub-regions, enabling real-time prediction and scenario analysis for CBM development and reservoir management.