Abstract: To provide the decision support and effective rescue plan for prevention and mitigation of coal mine accidents, an architecture of coal mine intelligent emergency plan automatic generation system is proposed in this paper. First of all, by integrating emergency plan and accident cases, we utilize text analysis to carry out structured and semi-structured storage. Based on the analysis of accident cases, we detect patterns for accident time and spots and then classify the accident cases into several categories to find accident characteristics for further study on accident treatment. Besides, real-time data from coal mine is analyzed and visualized by math models, which facilitates the real-time situation sensing and rescue work monitoring. Finally, according to history emergency plans, an intelligent emergency plan template is designed. Based on the expert knowledge base, accident treatment methods and real-time data analysis results, rule-based reasoning is used to generate intelligent emergency plan automatically.