The electrical fire monitoring system will automatically alarm to tell people where the residual current is abnormal before the fire occurs, which greatly reduces the occurrence of fire. Use data mining technology to find useful data from a large amount of data recorded by the electrical fire monitoring system, so as to reduce the occurrence of electrical fires. The purpose of this paper is to study the electrical fire monitoring system with different intelligent algorithms, and obtain the residual current data of different materials and materials with different cross-sectional areas in each time period. And then the electrical fire monitoring system feeds back whether the identification is successful or not through data mining technology. Experiments showed that the recognition rates of electrical fire monitoring systems with different cross-sections of the same material are roughly the same, and the recognition rates of electrical fire monitoring systems with the same cross-section of different materials are also roughly the same, and their recognition rates are roughly above 90%. The electrical fire monitoring system deserves further research to find a system with a higher recognition rate.