Fire alarm model based on AHP and improved BP neural network

Establishing a reliable fire warning system is of great significance to national property and life safety. Based on data processing, this paper uses analytic hierarchy process to set different weights for each technical index, so as to evaluate and analyze the classified and summarized data. Secondly, Select the component name, fire frequency, failure rate of different components, reliability and other parameters to establish a grey correlation model, so as to obtain the correlation analysis between the occurrence of fire and other indicators. At the same time, adopt the l-BFGD algorithm improved neural network and Softmax classifier model for fitting analysis. Take the practical application background of a certain group as an example. Evaluate the authenticity of the alarm signals sent by different components and get the probability that each alarm signal is a real fire.