Abstract:

In order to effectively check the compliance of coal mine building design, this paper used word similarity calculation (Word2Vec) model and database technology to organize and standardize a large number of regulations data from the perspective of accident causation. Firstly, the Word2Vec model was used to train word vectors on 225 accident reports from 2009 to 2021 caused by non-compliance of coal mine building design, so as to find out the risk elements of coal mine building design which were most associated with the occurrence of various accidents. Then, the relevant existing coal mine construction regulations were collected and screened from the acquired building design risk factors, a standardized expression framework was constructed for the regulations by analyzing specific regulations provisions. Finally, the regulations were organized into a structured query language (SQL) database to build a standardized regulation database. Using the Word2Vec model, the construction risk elements associated with the occurrence of accidents in accident analysis reports can be uncovered based on the deep-level characteristics of words. The construction risk elements can be used as the central word for the initial search and filtering of specification information, thus building the standardized library of coal mine construction regulations, which breaks the barriers between regulations and standards.

Key words: accident causation, coal mine building, regulation database, building regulation, Word2Vec model