The structural responsiveness of urban energy systems has emerged as a central challenge in the governance of shrinking cities. Urban shrinkage entails more than a redistribution of resources—it reflects deep tensions embedded in population spatial configuration, functional redundancy, and institutional inertia. To investigate the evolutionary trajectory and driving mechanisms of urban energy consumption (UEC) under the context of urban shrinkage, this study focuses on China, a country undergoing rapid internal regional transformation. Drawing on panel data from 281 cities between 2008 and 2021, the study integrates two-way fixed effects (TWFE) models, mediation analysis, and spatial econometric models to ensure the scientific rigor and robustness of the quantitative analysis. Contrary to the intuitive assumption that declining population leads to reduced energy loads, the results reveal a non-linear and asymmetric trajectory wherein per capita energy consumption increases alongside continued demographic decline. Mechanism decomposition further shows that declines in population density and the share of secondary industry suppress UEC through spatial dispersal and the retreat of energy-intensive sectors, respectively. In contrast, fiscal contraction and institutional path dependency collectively elevate the share of traditional energy consumption, reinforcing the structural inertia of UEC. This study illuminates the non-linear dynamics of energy system evolution under urban shrinkage and argues for a shift away from linear and target-driven governance paradigms toward governance frameworks that emphasize structural adaptation, distributive equity, and systemic resilience—thereby offering a theoretical and empirical foundation for multi-objective sustainable urban transitions.
LiDAR-captured 3D point clouds are widely used in self-driving cars and smart cities. Point-based semantic segmentation methods allow for more efficient use of the rich geometric information contained in 3D point clouds, so it has gradually replaced other methods as the mainstream deep learning method in 3D point cloud semantic segmentation. However, existing methods suffer from limited receptive fields and feature misalignment due to hierarchical downsampling. To address these challenges, we propose PSNet, a novel patch-based self-attention network that significantly expands the receptive field while ensuring feature alignment through a patch-aggregation paradigm. PSNet combines patch-based self-attention feature extraction with common point feature aggregation (CPFA) to implicitly model large-scale spatial relationships. The framework first divides the point cloud into overlapping patches to extract local features via multi-head self-attention, then aggregates features of common points across patches to capture long-range context. Extensive experiments on Toronto-3D and Complex Scene Point Cloud (CSPC) datasets validate PSNet’s state-of-the-art performance, achieving overall accuracies (OAs) of 98.4% and 97.2%, respectively, with significant improvements in challenging categories (e.g., +32.1% IoU for fences). Experimental results on the S3DIS dataset show that PSNet attains competitive mIoU accuracy (71.2%) while maintaining lower inference latency (7.03 s). The PSNet architecture achieves a larger receptive field coverage, which represents a significant advantage over existing methods. This work not only reveals the mechanism of patch-based self-attention for receptive field enhancement but also provides insights into attention-based 3D geometric learning and semantic segmentation architectures. Furthermore, it provides substantial references for applications in autonomous vehicle navigation and smart city infrastructure management.