Development Of Signal Processing System For Domestic Fire Monitoring And Alarm

With the advancement of industrialization and modernization of society,fire has become one of the disasters that pose the greatest threat to the safety of people’s lives and property.Fire warning systems and fire extinguishing equipment can effectively detect and extinguish fires in time or fight for rescue time in the early stage of fires.2018 China issued a national standard specifying the installation requirements for fire detection equipment in high-rise buildings and large places,but domestic fire monitoring and alarm system has not yet been introduced to improve the specification standards.This thesis will address the development of a signal processing system for domestic fire monitoring and alarm,and the main research contents are as follows.A numerical simulation model of indoor fires is established,and the household structure data from real residential houses are modeled using PyroSim.The numerical simulation of indoor fire signals is studied and characterized by the design of parameters such as reaction type,furniture surface combustion material parameters,fire source and sensor placement.A pre-processing method is designed to aggregate and expand the dimensionality of 3D independent sensor data,so that the 3-dimensional time series are converted into a picture-like matrix format.Improving the ConvNeXt network model structure by modifying three aspects of the model: the stage size,the number of stage channels,and the design of the inverse bottleneck structure of the block,and proposing a neural network model structure suitable for home fire monitoring and alarm systems.This model reduces the number of parameters and computation volume to 1.5% of the original model at a slight cost of 0.63% decrease in accuracy compared to the ConvNeXt-T model.The prototype of the signal processing system for home fire monitoring and alarm was developed,and the system structure with separate acquisition and monitoring and alarm was designed.The software and hardware of the acquisition and monitoring and alarm system are designed and selected,and finally experiments are designed to verify the performance of the prototype.