Gombustible Gas Explosion Based On Sensor Data Fusion Monitoring And Evaluation System

Hazard characteristics of combustible gas mainly reflect in combustibility,explosibility,diffusivity,toxicity,corrosivity,asphyxiation and other aspects.Combustion and explosion will not only bring economic loss,but also affect the process of drilling and exploiting.Therefore,it is vital to have quick and accurate detection towards combustible gas content,as well as burning and explosion state prediction.Artificial neural network is a kind of information processing system based on human brain nerve.Founded on mathematical model,it processes information data by simulating the net structure and action mode of human brain nerve.Neural network is widely applied to various fields like AI and prediction because of it's superior intelligence.BP neural network which is featured in strong learning capability and great non-linear function is the most classic mode of neural network.Besides,RBM doesn't rely on initialization and can find the statistical law of data.Therefore,this article adopts DBN mode which combines both BP neural network and RBM neural network.The explosion security state evaluation is acquired by fusing the sensor data which is collected by flaming and explosion detection system via DBN.By doing this,the misjudgments due to data deviation collected by sensor can be avoided,accuracy rate of combustible gas explosion state judgments can be improved and combustion state's data collect work can be done quickly.At the same time,this article adopts CAN Bus,and combines the data communication method of ZigBee to transfer multi-sensor data.Based on multi-sensor date fuse,it develops a combustible gas' combustible sate detection and evaluation system.It fulfills the collection and transmission of combustible gas data,and the judgments towards combustion state.Besides,the efficiency of this system is also verified by experiment.