Causal relationship analysis and risk assessment of contributing factors in coal mine ordinary accidents: A case study (2002–2023)

Abstract

Systematically analyzing accident causes and identifying causal pathways is essential for eliminating accident triggers and effectively managing risks. This study conducted an in-depth analysis of 386 coal mine ordinary accident cases from 2002 to 2023, covering six major types of accidents: object striking, vehicle injury, mechanical injury, high falling, roof falling, and other injuries. The accident causation model-24Model, grounded theory, association rule mining, and Bayesian networks were integrated in this study, enabling a thorough exploration of internal causal relationships and the assessment of risk profiles. The main findings are as follows: Accident causes are categorized into four primary dimensions—safety culture, safety management system, safety capability, and unsafe acts/conditions—from which 75 factors are identified. Employing the Apriori algorithm, 78 association rules are extracted, revealing the causal logic between six types of accidents and their respective contributing factors. Based on these findings, a Bayesian network is constructed, and through conditional probability analysis and local sensitivity analysis, six probability-sensitivity risk matrices are developed, enabling the classification of factors into different risk levels for each accident scenario. A tiered prevention strategy is designed to address these risk levels. An accident prevention system is proposed, guided by safety culture, integrated into the management system, enhanced individual safety capabilities, and strengthened behavior control processes. Additionally, a three-line of defense risk management mechanism is introduced, providing a systematic, data-driven framework for safety management and accident prevention.