City Fire Risk Assessment Model Based on the Adaptive Genetic Algorithm and BP Network

Abstract

Based on the risk evaluation index system of city fire, a comprehensive evaluation model with the adaptive genetic algorithm and BP neural network (AGA-BP) is established in the article. In former process of the hybrid algorithm, the adaptive genetic algorithm is applied to adjust weights and thresholds of the three-layer BP neural network and train the BP neural network for locating the global optimum, and the error back propagation algorithm is used to search in neighborhoods of the approximate optimal solution in the later process. The program written in VB6.0 is used to learn some samples of city fire risk according to the AGA-BP algorithm and the general BP algorithm. The results show that the learning precision of AGA-BP algorithm is more correctly than that of the general BP algorithm. The training speed and convergence rate of the former is significantly improved because of the combination of AGA and BP algorithm. It is helpful to realize automated evaluation for city fire risk.