Research on Accident Risk Early Warning System Based on Railway Safety Management Data under Cloud Edge Collaborative Architecture

At present, the early warning and prediction of railway accidents are mostly carried out from the perspective of technical conditions, that is, the unsafe state of equipments, such as the degree of rail wear, the operation state of locomotive parts, etc. These analyses ignored the management factors behind the accident, such as the impact of people's unsafe behaviors on the accident. This research constructs an accident risk early warning system based on safety inspection data and establishes a corresponding relationship between massive safety inspection data and accident risk factors. In this study, the analysis model and mining algorithm are based on Heinrich's rule and the time decay factor of historical events, and the cloud edge collaboration method is adopted to give consideration to model training and application. Finally, the results of risk assessment are visualized through different forms of seaweed ball models, which is helpful for safety managers to master the risk situation. The above early warning system has been applied in some railway bureaus and achieved good application results.