Research Of Multi Source Information Fire Detection Algorithm Based On Wireless Network

Intelligentization and security put forward higher requirements for contemporary fire alarm systems.How to improve the defects of traditional fire detection and alarm systems and how to optimize the safety level of fire protection systems have become the focus of current research.On account of this reason,the thesis expands research on multi-source information fire detection methods based on wireless networks.The topic starts with two aspects: Firstly,due to the characteristics of high transmission rate,simple wiring,and easy retrofitting of wireless networks,this topic has designed a fire detection system based on wireless sensor networks according to related specifications.This system can accurately collect fire information so that it can predict the fire behavior.Secondly,from the perspective of fire information prediction,in order to overcome the shortcomings of traditional fire alarm system singulation prediction and pre-judgment,etc.,the multi-source information fusion technology and artificial intelligence algorithm are used to establish a fire prediction model.The focus of this study is how to make it better to apply it to the fire prediction model.For the characteristics of fire information data,the fire prediction model was designed using three methods: fuzzy reasoning,artificial neural network and extreme learning machine.The three fire characteristic information of CO concentration,smoke concentration,and temperature were used as the network input with no fire probability,smoldering fire probability and open flame probability used as network outputs.MATLAB software is used for network design and simulation verification respectively.Finally,the comparison with the national standard fire test data is carried out.The simulation results show that the adaptive fuzzy neural inference,RBF neural network model,GRNN neural network model,PNN neural network model and ELM algorithm can all predict fire with their own advantages.