Accurate solar power forecasting is a key technology for efficient operation of photovoltaic (PV) power plants and safe grid dispatch. Under the “dual carbon” goals and the increasing share of renewable energy connected to the grid, ultra-short-term power forecasting is important for improving dispatch decisions and supporting system operation. To address the ultra-short-term forecasting task at two PV sites, this study develops an end-to-end framework that integrates machine learning, Bayesian optimisation, and SHAP-based interpretability. First, correlation analysis was performed on the datasets from the two sites to provide a foundation for subsequent model development. Next, seven forecasting models, including CatBoost, NGBoost, Random Forest (RF), AdaBoost, ARIMA, CNN-LSTM, and LSTM, were developed and uniformly optimised using Bayesian optimisation. Under a unified framework of data partitioning, optimisation budget, and evaluation metrics, the predictive performance of all models at the two sites was systematically assessed. The results show that the optimal model varies across sites: at Site 1, LSTM delivered the best performance, with test-set R², MSE, RMSE, and MAE values of 0.972, 17.610, 4.196, and 2.267, respectively; at Site 2, CatBoost achieved the best results, with corresponding values of 0.994, 0.385, 0.621, and 0.249, respectively. These findings highlight pronounced site-specific differences in model performance, indicating that different modeling approaches exhibit distinct adaptability under varying data characteristics and operational conditions. Further error analysis and SHAP interpretation indicate that solar irradiation and key meteorological variables are the main drivers of power output, and their effects are nonlinear, confirming the model’s ability to capture complex nonlinear relationships in PV power forecasting. Finally, a graphical user interface (GUI) tool was developed to support site selection, real-time forecasting, and parameter input, providing a practical and convenient solution for PV plant operation and grid dispatch.

For power-grid applications such as transmission corridor inspection, substation asset inspection, and post-disaster emergency repair, reliable UAV self-localization under GNSS-degraded or GNSS-denied conditions is critical to ensuring operational safety and accurate defect geotagging. Due to substantial discrepancies in viewpoint, scale, and geometric structure between oblique UAV images and nadir satellite images, conventional RGB-based cross-view retrieval methods often suffer from unstable alignment and insufficient geometric modeling, particularly in scenarios with repetitive textures and partial overlap. To address these challenges, we propose a cross-view visual geo-localization model that integrates RGBD multimodal inputs with multi-scale attention enhancement. Specifically, MiDaS is used to estimate relative depth from UAV imagery, which is concatenated with RGB to form a four-channel input, while satellite images are padded with an additional zero channel to maintain dimensional consistency. A shared-weight ViTAdapter is adopted to learn joint semantic–geometric representations, and a lightweight Efficient Multi-scale Attention (EMA) module is adopted on spatial feature maps to strengthen multi-scale spatial consistency. In addition, an IoU-weighted InfoNCE loss is employed to accommodate partial matching during training, thereby improving the robustness of feature alignment. Experiments on the GTA-UAV dataset under the cross-area protocol show stable performance across both retrieval and localization metrics. Specifically, Recall@1, Recall@5, and Recall@10 reach 18.12%, 38.83%, and 49.47%, respectively; AP is 28.01 and SDM@3 is 0.53; meanwhile, the top-1 geodesic distance error Dis@1 is 1052.73 m. These results indicate that explicit geometric priors combined with multi-scale spatial enhancement can effectively improve cross-view feature alignment, leading to enhanced robustness and accuracy for localization in challenging power inspection scenarios.

Accurate wind power forecasting is essential for efficient wind farm operation and reliable grid dispatch. This study proposes a site-adaptive forecasting framework that integrates machine learning, Tree-structured Parzen Estimator (TPE)-based Bayesian hyperparameter optimization, and SHapley Additive exPlanations (SHAP) for interpretability. Using real-world meteorological and power generation data from two wind farms, we first perform joint-distribution feature analysis to characterize statistical relationships between key inputs and power output, supporting model development and interpretation. TPE optimization is then applied to six benchmark models (CatBoost, Extra Trees, GBM, LightGBM, TabNet, and XGBoost). The optimized Extra Trees model achieves the best performance at Site 1 (R² = 0.965, RMSE = 3.872 kW, MAE = 2.333 kW), whereas the optimized XGBoost model performs best at Site 2 (R² = 0.921, RMSE = 3.049 kW, MAE = 1.382 kW), demonstrating the effectiveness of TPE tuning and the strong predictive capability of tree-ensemble learners. SHAP analysis further reveals heterogeneous drivers across sites: Site 1 benefits from synergistic wind-speed contributions across multiple heights, while Site 2 is primarily governed by hub-height wind speed. Overall, the proposed framework achieves both high accuracy and robust interpretability for multi-site wind power forecasting.

Urban underground spaces are increasingly developing toward large-scale and multifunctional configurations, imposing higher requirements on the large span and ultimate load-bearing capacity of protective doors for critical infrastructure. To address the low load-bearing efficiency and susceptibility to local instability of conventional large protective doors, this study proposes a novel double-leaf, double-arched, bidirectionally supported protective door incorporating a rear grillage and horizontal coupling beams, based on a spatially coordinated load-bearing system. First, the dynamic response of a rib-stiffened double-leaf arched protective door, designated Type I, was investigated using the finite element code LS-DYNA. To address its closely spaced higher-order modal frequencies and insufficient stiffness, a preliminary optimization was conducted by adding a rear grillage, resulting in the Type II configuration. The stress concentration and localized connection damage caused by stiffness incompatibility in Type II were then examined, and a final optimized design incorporating horizontal coupling beams was proposed (Type III). The results indicate that the coordinated restraint provided by the rear grillage and horizontal coupling beams effectively mitigates the reduction in the fundamental frequency and the local stress concentrations observed in the Type II structure. The fundamental frequency of the Type III structure is 7.9% higher than that of the Type I structure. Under severe blast loading, the Type III structure exhibits a substantially improved load-transfer path. Compared with the Type I and Type II structures, respectively, its maximum displacement is reduced by 7.30% and 14.09%, and its peak equivalent plastic strain by 19.15% and 22.84%. Consequently, the Type III structure exhibits improved global deformation compatibility and effectively suppresses local instability. This study elucidates the spatially coordinated load-bearing mechanism of the double-leaf, double-arched, bidirectionally supported protective door under severe dynamic loading and represents a promising structural candidate for the innovative design and optimization of large protective doors for critical underground infrastructure.

With the development of various emerging structures, concrete-filled steel tubular (CFST) columns have become critical load-bearing components in key infrastructures such as subways and underground utility tunnels. Accurately predicting their ultimate bearing capacity (N_u) is essential for guaranteeing structural safety. To address the limitations of traditional empirical formulas and code-based calculation approaches, this paper proposes a prediction model for ultimate bearing capacity based on the CatBoost algorithm optimized by Random Search. Furthermore, the marginal contribution of each key feature to the prediction results is measured through interpretability analysis. First, a database containing 438 CFST column ultimate bearing capacity test cases was established, with key parameters such as geometric dimensions and material properties as input variables. Second, the predictive performance of six machine learning algorithms—CatBoost, LightGBM, Random Forest (RF), Gradient Boosting (GB), K-Nearest Neighbors (KNN), and XGBoost—was compared. A five-fold cross-validation integrated with a Random Search strategy was employed for joint hyperparameter optimization. The results show that the optimized CatBoost model significantly outperforms other algorithms and conventional design codes, achieving a coefficient of determination (R²) as high as 0.99 and a root mean square error (RMSE) of 174.29 kN. Furthermore, the SHAP (Shapley Additive exPlanations) method was used to perform global and local interpretability analyses of the prediction model. This not only quantified the individual contribution and interaction effects of each feature parameter on the bearing capacity but also revealed that geometric parameters are the primary influencing factor. This finding confirms a high degree of consistency between the prediction mechanism of the data-driven model and classical mechanical theories, effectively validating the model’s reliability. This study provides an efficient and reliable tool for the optimal design and rapid evaluation of CFST columns and establishes a new data-driven paradigm for the design and reinforcement of key components in underground structures.