The scale of targets in remote sensing images varies greatly and diversity. It has many small targets which distribute densely, and high complexity of image background. The number of network model parameters and the computation amount of the object detection algorithms based on deep learning is huge. It is difficult to apply them on the platform with fixed performance and limited computing resources. A lightweight remote sensing object detection model is proposed in this paper, which called Attention and Multi-Scale Feature Fusion Lightweight-YOLO (AMFLW-YOLO). The deep separable convolution, inverted residual, and linear bottleneck structure are employed to replace the standard convolution layer to reduce the model parameters in the backbone network of the model. The Coordinate Attention (CA) mechanism is introduced into the feature fusion network to capture the direction-aware and location-aware information across channels at the same time, which improves the accuracy of the network. The Bidirectional Feature Pyramid Network (BiFPN) structure is employed to strengthen feature extraction. The learnable weights are introduced to learn the importance of different input features. The multi-scale feature fusion is applied to improve the detection effect. Experimental results show that the algorithm achieves satisfactory performance in terms of efficiency and accuracy and has advantages in detection accuracy and model lightweight.

Rockburst is a common geological disaster in mines, tunnels, deep underground engineering, and during excavation, mining, and construction. Rockburst frequently occurs as the depth of burial increases, and its early warning technology is in urgent need of further development. At present, the most effective monitoring and analysis method of rockburst is microseismic technology, which detects a large number of rock micro-fracture signals through geophones. The identification of microseismic monitoring data is an essential part of microseismic data processing. It is necessary to identify effective microseismic signals from considerable monitoring data for subsequent early warning. Aiming at the identification of rock micro-fracture signals, this thesis proposes a microseismic data identification method based on the Deep Convolution Neural Network Inception (DCNN-Inception) algorithm. The algorithm uses an existing Convolutional Neural Network (CNN) model, adding Inception structure in the middle of the model to form a DCNN-Inception model. A data set was established depending on the actual measured data of Baihetan Hydropower Station, and CNN and DCNN-Inception were employed to identify effective microseismic signals. The results demonstrate that the DCNN-Inception algorithm is better than CNN in recognition accuracy and can effectively identify effective microseismic signals. It provides an essential foundation for the identification of microseismic abnormal signals of rock microfracture and the early warning of rock rupture precursors and is of practical significance for the study of rockburst warning technology.

Rock burst early warning technology is currently applied mainly in microseismic monitoring. Rock burst signals indicate the micro-fracture phenomena of a rock and can transmit earthquake waves through the rock before they are finally received by a detector. A characteristic decomposition of rock micro-fracture signals was conducted by the singular value Empirical Mode Decomposition (EMD) algorithm to effectively decompose the characteristic signals of a rock micro-fracture from mixed microseismic signals, with a low signal to noise ratio to ensure prediction precision. When comparing the proposed method with wavelet decomposition and EMD, it was found that the local characteristics of the signals were retained effectively. The proposed algorithm was verified by applying it in a laboratory simulation and to the decomposition of microseismic signals from a hydro-power station. It was concluded that the improved algorithm had a better decomposition precision than wavelet decomposition and EMD decomposition and could effectively separate the characteristic signals of micro-earthquakes. This could provide a significant basis for the identification of the abnormal microseismic signals of rock micro-fractures as well as a pre-warning of rock fractures. It is therefore of practical significance to study rock fracture early warning technology.

Determining the precise arrival time of earthquake recordings in noisy settings is crucial for effective earthquake data processing and analysis. This study introduces a novel U-Net3+-based network specially designed for accurate first arrival picking in noisy data. To reduce information loss in seismic waveform extraction, enhancements were made to the U-Net3+ decoder, featuring a residual structure for better information propagation, a convolutional layer replacing the pooling layer to preserve high-frequency seismic waveform details, and a convolutional block attention module (CBAM) integrated into the deeper layers to prioritize critical information at the onset of the seismic wave. We evaluated our model against U-Net, U-Net2+, U-Net3+, and Phase-Net using noisy earthquake recordings, and carried out comprehensive ablation studies on each module. The model’s generalizability was further validated using the “DiTing” dataset. The proposed network model demonstrated superior accuracy and robustness against noise, with picking rates exceeding 92.42%, 90.82%, and 87.81% when signal-to-noise ratios (SNRs) dropped below 20, 10, and 5 dB, respectively. This work contributes a promising approach to developing a more noise-resistant, low-cost seismic phase picking model, and offers potential significant gains for seismic monitoring techniques.

Cracks inside urban underground comprehensive pipe galleries are small and their characteristics are not obvious. Due to low lighting and large shadow areas, the differentiation between the cracks and background in an image is low. Most current semantic segmentation methods focus on overall segmentation and have a large perceptual range. However, for urban underground comprehensive pipe gallery crack segmentation tasks, it is difficult to pay attention to the detailed features of local edges to obtain accurate segmentation results. A Global Attention Segmentation Network (GA-SegNet) is proposed in this paper. The GA-SegNet is designed to perform semantic segmentation by incorporating global attention mechanisms. In order to perform precise pixel classification in the image, a residual separable convolution attention model is employed in an encoder to extract features at multiple scales. A global attention upsample model (GAM) is utilized in a decoder to enhance the connection between shallow-level features and deep abstract features, which could increase the attention of the network towards small cracks. By employing a balanced loss function, the contribution of crack pixels is increased while reducing the focus on background pixels in the overall loss. This approach aims to improve the segmentation accuracy of cracks. The comparative experimental results with other classic models show that the GA SegNet model proposed in this study has better segmentation performance and multiple evaluation indicators, and has advantages in segmentation accuracy and efficiency.