Fire Warning Based On Multi-sensor Information Fusion System Research

A fire is a burning phenomenon that is without of control in time and space,which is a serious threat to human life and property.The randomness and complexity of the fire make the fire warning technology using the traditional single parameter can not meet the people 's requirement of the accuracy of fire early alarm.In view of this problem,this paper presents a fire early alarm system based on multi-sensor information fusion.Firstly,the mechanism of fire generation and the development law are studied.Based on the analysis of the characteristics of the fire,the CO concentration,temperature and smoke concentration are selected as the fire characteristic parameters.The system uses the intelligent fusion algorithm to fuse the CO concentration,temperature and smoke concentration to realize the identification of fire,smoldering and fire.Firstly,the mechanism and the development law of fire generation are studied.Based on the analysis of the characteristics of initial fire,the CO concentration,temperature and smoke concentration are selected as the fire characteristic parameters and can be analysed of fusion by the system through the intelligent fusion algorithm to realize the identification of no fire,smoldering fire and open fire.Secondly,because of the complexity of living environment,the early fire alarm system which only depends on one means of communication mode can not meet all the circumstances,a multi-communication mode fire early alarm system is designed.It is involved that the design of the whole structure of the fire early warning system,the software design,the hardware design and the software design on PC.The hardware and software design of the fire eary warning system include the regional fire detector's Zigbee node,Wi-Fi node,and fixed node,the fire early warning controller,as well as the hardware and software design of alarm device of fire emergency.Thirdly,the fire test environment is designed,a fire test platform is set up,the wood smoldering fire and wood open fire test are completed.Finally,BP neural network,support vector machine(SVM)and extreme learning machine(ELM)are introduced.The fire alarm algorithm based on BP neural network,one-to-one support vector machine extreme learning machine are designed.Compared and analyzed among the three kinds of fire warning algorithms by simulation using the experimental data,it draws conclusions that the algorithm based on extreme learning machine has the advantages of high accuracy and generalization ability.But the limit learning machine is poorly stable,and the kernel extreme learning machine(KELM)is proposed for this shortcoming.However,there are some shortcomings of parameter sensitivity in the kernel extreme learning machine.In this design,the improved particle swarm optimization algorithm is used to optimize the kernel extreme learning machine.The algorithm of fire early warning based on kernel extreme learning machine optimized inertial particle swarm is proposed.The simulation results show the advantages of the algorithm.