Power-system safety is fundamental to protecting lives and ensuring reliable grid operation. Yet, hierarchical text classification (HTC) methods struggle with domain-dense accident narratives that require cross-sentence reasoning, often yielding limited fine-grained recognition, inconsistent label paths, and weak evidence traceability. We propose EG-HPT (Evidence-Grounded Hierarchy-Aware Prompt Tuning), which augments hierarchical prompt tuning with Global Pointer-based nested-entity recognition and a sentence–entity heterogeneous graph to aggregate cross-sentence cues; label-aware attention selects Top-k evidence nodes and a weighted InfoNCE objective aligns label and evidence representations, while a hierarchical separation loss and an ancestor-completeness constraint regularize the taxonomy. On a HFACS-based power-accident corpus, EG-HPT consistently outperforms strong baselines in Micro-F1, Macro-F1, and path-constrained Micro-F1 (C-Micro-F1), with ablations confirming the contributions of entity evidence and graph aggregation. These results indicate a deployable, interpretable solution for automated risk factor analysis, enabling auditable evidence chains and supporting multi-granularity accident intelligence in safety-critical operations.

Modeling accident investigation reports is crucial for elucidating accident causation mechanisms, analyzing risk evolution processes, and formulating effective accident prevention strategies. However, such reports are typically long, hierarchically structured, and information-dense, posing unique challenges for existing language models. To address these domain-specific characteristics, this study proposes SAFE-Transformer, a Structure-Aware and Format-Enhanced Transformer designed for long-document modeling in the emergency safety context. SAFE-Transformer adopts a dual-stream encoding architecture to separately model symbolic section features and heading text, integrates hierarchical depth and format types into positional encodings, and introduces a dynamic gating unit to adaptively fuse headings with paragraph semantics. We evaluate the model on a multi-label accident intelligence classification task using a real-world corpus of 1632 official reports from high-risk industries. Results demonstrate that SAFE-Transformer effectively captures hierarchical semantic structure and outperforms strong long-text baselines. Further analysis reveals an inverted U-shaped performance trend across varying report lengths and highlights the role of attention sparsity and label distribution in long-text modeling. This work offers a practical solution for structurally complex safety documents and provides methodological insights for downstream applications in safety supervision and risk analysis.

Urban ground collapse (UGC) threatens dense coastal cities because failures can occur abruptly beneath roads and buried lifelines. We developed a dynamic and interpretable machine-learning framework for UGC susceptibility assessment in Shenzhen, China. We integrated 1687 events from 2017 to 2024 with 20 predictors covering terrain, rainfall, drainage infrastructure and urban disturbance. Event-date rainfall was assigned to collapse samples, and background controls received year-constrained pseudo-event dates after spatial exclusion. Three tree-ensemble models were evaluated under random 75/25 testing, leave-one-district-out spatial validation and temporal hold-out testing. The selected LightGBM 1:5 model achieved ROC-AUC = 0.934, AP = 0.790 and BA = 0.860 in random testing, with more conservative ROC-AUC values of 0.905 and 0.884 under spatial and temporal validation. High-rainfall mapping expanded high and very high susceptibility zones from 4.60% to 15.26% and increased event capture from 19.32% to 58.92%. Grouped SHAP indicated comparable pipeline and rainfall contributions (26.01% and 25.13%). The leading predictors were pipe burial depth, road density and 30-day rainfall. PDP and spatial SHAP diagnostics highlighted high model responses where deep pipes coincided with high antecedent rainfall or dense roads in mature urban cores. This framework supports the rainfall-conditioned, mechanism-informed susceptibility diagnosis for early warning and mitigation.