Fire Detecting Technology of Information Fusion Using Support Vector Machines

In this paper, fire detection technology based on multi-sensors information fusion is presented for the complexity of fire process and the multiplicity of fire environment. And support vector machines are used to establish fire detecting model of information fusion, for Support vector machines show excellent performance in generalization and optimization. The improved genetic algorithm by Auto-adaptive crossover probability and the mutation probability is used to optimize the nuclear function parameters of support vector machines. The training and simulation results show that the combination of support vector machine and improved genetic algorithms has good ability to optimize model. It has good generalization ability and higher calculation speed comparing with other approaches. The superiority and the feasibility are proved by through the simulation.