Abstract: With the rapid development of cities, the construction of urban underground comprehensive pipe galleries has been rapidly carried out. Due to its main use to carry urban resources such as electricity, gas, and heat, it has many risk factors and high fire hazards. In response to the above issues, a fire warning and hazard identification system based on the improved Apriori algorithm is proposed in the article. The STM32 microcontroller is used as the main controller and various sensors are used to monitor data inside the pipe gallery in this system. LabView software is adopted as the upper computer to provide real-time data monitoring screens and alarm information display for the user end. Through experimental data verification, the proposed improved algorithm can save about 60% of time compared to the original algorithm, and its accuracy can be maintained at over 90%. For different types of fires that require different data mining association rules, the underground pipe gallery electrical cable fire is taken as an example. The fire data obtained by establishing the fire mode by using PyroSim software is used to mine association rules using the algorithm proposed in this article. Finally, three characteristics of early cable fires are obtained, which are the warning basis for cable fires.

Key words: comprehensive pipe gallery, fire warning, association rules, Apriori algorithm, PyroSim