A railway operation risk intelligent prediction method based on bidirectional semantic constraint

Abstract

The application discloses a kind of railway operation risk intelligent prediction methods based on bidirectional semantic constraint, comprising: according to historical accident report, construct railway operation risk knowledge graph;Based on railway operation risk knowledge graph, the semantic embedding representation of risk factor and risk relationship is optimized;The semantic embedding representation of risk factor and risk relationship after optimization is stored as semantic embedding representation knowledge base;Based on semantic embedding representation knowledge base, the candidate tail risk event corresponding to target risk event is matched;Based on scoring function, the scoring value of all candidate tail risk events is calculated, all scoring values are sorted, and candidate risk event list is output in order.The present application can automatically and intelligently deduce the complete risk chain possibly caused by the given initial event, and output in probability order.This greatly improves the predictability of complex and rare risk events, and changes risk control from "passive response" to "active early warning".

Images