Research on Fire Alarm System Problems Based on Comprehensive Evaluation Method and Machine Learning

Machine learning applied to fire alarm systems is an increasingly common optimization problem. In this paper, a method based on comprehensive evaluation and machine learning is proposed. Firstly, relying on the literature and data sets, 10 indicators such as the number of fires, the failure rate, the accuracy rate, etc. are extracted to comprehensively evaluate the performance of each type of alarm component, and the entropy weight method, the information weight method and the TOPSIS method are used to derive the performance level rating of each component, and the fire alarm intelligent research and judgment system is constructed. Then, a BP neural network was used to build a training model and test the alarm data set for a certain region, and finally predicted the real number of alarms of each test device with an accuracy of 95.4%. These results can help the government plan to select more reliable fire detector types and improve the fire alarm accuracy to further optimize the fire alarm system.