A Diagnostic Method for Port Safety Regulations Based on AI Large Model

Abstract

This invention relates to the fields of port safety management and artificial intelligence technology, specifically disclosing a method for diagnosing missing or incomplete port safety regulations based on an AI large-scale model. The method includes: constructing a safety regulations knowledge graph by hierarchically decomposing a standard regulations library; dividing the system into scenario units and identifying inherent risk points and priority indices; constructing a dual-channel comparative diagnostic model of regulations and scenarios; completing feature encoding to obtain feature vectors of the same dimension on both the regulations and scenarios sides; analyzing the similarity of feature vectors through a cross-attention mechanism; calling knowledge subgraphs of the knowledge graph to determine the type of difference and establish rectification priorities. This method solves the problems of rigid compliance benchmarks, detachment from practical application, difficulty in multimodal data fusion, ambiguous difference determination, and lack of clear rectification priorities in traditional port safety regulations diagnosis. It achieves dynamic adaptation of compliance benchmarks, diagnosis aligned with port practices, cross-modal data fusion and comparison, and clear priorities for both diagnosis and rectification, thus improving diagnostic accuracy and efficiency.

Images