7 research outputs found
New exploration model and engineering application of urban karst groundwater channel based on the 3D electrical method
No full textThe detection of karst groundwater channels is of great significance for the development of underground cities and the prevention and control of geological disasters. Aiming at the characteristics of high spatial variability of karst geological groundwater channels, a new exploration model based on the three-dimensional (3D) electrical method is proposed. Combined with an urban exploration project in Hunan Province, China, the exploration efficiency and effect of the new mode of one-shaped layout arrangement exploration and conventional three-dimensional electric method exploration are compared and analyzed. The main process of this model is as follows: Firstly, determine the X and Y directions of electrophysical exploration according to the direction of the geological anomalies in the exploration area. Then, according to the exploration length and measuring point spacing in X and Y directions, the electrical grid parameters are determined. Finally, the coordinates of all power supply grids and the actual relative coordinates of the exploration area are calculated, and the power supply points are arranged in a pattern of one. The feasibility of the above model has been verified through the exploration project, and the results show that the exploration efficiency of this exploration model has been significantly improved compared with the conventional three-dimensional electric method, and the effect is excellent and highly applicable.The comparison of data shows that the number of power supplies and data acquisition of conventional 3D electrical method are 121 and 14520 respectively, while the number of new mode of one-shaped layout is only 17 and 2046 respectively, with a reduction of 85.9%. The research results can provide some reference for the exploration scheme design of urban karst groundwater channels
Resilience-based post disaster recovery optimization for infrastructure system via Deep Reinforcement Learning
No full textA global feature fusion and adaptive optimization method to enhance detection accuracy and computational efficiency based on YOLOv8
No full textWeed detection is vital for agricultural productivity but faces challenges like target scale diversity and leaf shading-induced asymmetry. To address these challenges, this paper proposes a global feature fusion adaptive optimization method based on YOLOv8 to enhance detection accuracy and computational efficiency. (Global Feature Fusion-YOLOv8, GFF-YOLOv8). First, to enhance the accuracy of small object detection, we propose the C2f-FADC (C2f-Frequency-Adaptive Dilated Convolution) to replace the traditional C2f method, thereby improving the backbone network of YOLOv8. Next, to improve information exchange between different dimensions within the network, we propose the Global Fusion Diffusion Pyramid Networks (GFDPN) to replace the Neck structure in YOLOv8. This is achieved through adaptive feature selection and global fusion diffusion methods. Finally, to improve the model's ability to learn features, we introduce an Adaptive Task-Aligned Dynamic Detection Head (ATDDH), which modifies the traditional detection head to enhance the model's robustness and accuracy in weed detection. Experiments on the DeepWeeds dataset show that GFF-YOLOv8 achieved a mAP@50 of 76.98 %, outperforming other YOLO-based weed detection models
Harnessing hybrid digital twinning for decision-support in smart infrastructures
No full textDigital Twinning (DT) has become a main instrument for Industry 4.0 and the digital transformation of manufacturing and industrial processes. In this statement paper, we elaborate on the potential of DT as a valuable tool in support of the management of intelligent infrastructures throughout all stages of their life cycle. We highlight the associated needs, opportunities, and challenges and discuss the needs from both the research and applied perspectives. We elucidate the transformative impact of digital twin applications for strategic decision-making, discussing its potential for situation awareness, as well as enhancement of system resilience, with a particular focus on applications that necessitate efficient, and often real-time, or near real-time, diagnostic and prognostic processes. In doing so, we elaborate on the separate classes of DT, ranging from simple images of a system, all the way to interactive replicas that are continually updated to reflect a monitored system at hand. We root our approach in the adoption of hybrid modeling as a seminal tool for facilitating twinning applications. Hybrid modeling refers to the synergistic use of data with models that carry engineering or empirical intuition on the system behavior. We postulate that modern infrastructures can be viewed as cyber-physical systems comprising, on the one hand, an array of heterogeneous data of diversified granularity and, on the other, a model (analytical, numerical, or other) that carries information on the system behavior. We therefore propose hybrid digital twins (HDT) as the main enabler of smart and resilient infrastructures
Gray matter atrophy patterns within the cerebellum-neostriatum-cortical network in SCA3
No full textObjectiveTo investigate the spatial patterns and the probable sequences of gray matter atrophy in spinocerebellar ataxia type 3 (SCA3).MethodsA total of 47 patients with SCA3 and 49 age-and sex-matched healthy controls participated in the study. High-resolution T1-weighted MRI were examined in all participants. We used the causal network of structural covariance (CasCN) to identify the sequence of gray matter atrophy patterns. This was achieved by applying Granger causality analysis to a gray matter atrophy staging scheme performed by voxel-based morphometry from the network level.ResultsParticipants in the premanifest stage of the disease showed the presence of focal gray matter atrophy in the vermis. As the disease duration increased, there was progressive gray matter atrophy in the cerebellar, neostriatum, frontal lobe, and parietal lobe. The patients with SCA3 also showed proximal and distal cortical atrophy sequences exerting from the vermis to the regions mainly located in the cerebellum-neostriatum-cortical network.ConclusionOur results, although preliminary in nature, indicate that the gray matter atrophy in SCA3 lies and extends to involve more regions according to distinct anatomical patterns, mainly in the cerebellum-neostriatum-cortical network. These findings advance our understanding on the natural history of structural damage in SCA3, while confirming known clinical features. This could provide unique insight into the ordered sequential process of regional brain atrophy that targets a particular network
