Abstract: On the problems of BP neural network in the process of fitting, such as low detection precision, easy to fall into local optimum, an optimized BP neural network model based on genetic algorithm(GA) and simulated annealing algorithm(SA) was developed. The model can significantly improve the recognition accuracy, while avoid the network over fitting phenomenon, and make the forecast results jump out of local optimal so as to achieve the global optimal. Firstly, the hidden layer structure was improved by GA, and then the connection weight was improved by SA. Finally, the optimized GA-SA-BP model was used for information fusion of fire experimental data to realize fire detection. Experimental results show that compared with the single BP neural network, the BP neural network improved by GA and SA can effectively improve the fitting ability of the network, and improve the accuracy of fire detection to 98.91%.

Key words: fire detection; simulated annealing algorithm; genetic algorithm; BP neural network