Studies On A Neuro-Fuzzy Network Applied In Fire Detection System

Humen being have made a great progress because using fire,made science advance.The people have created those more civilization thanks to fire.The fire is good enough to fill men's needs,but once fire is out of control,it not only deprives humen's lifes and estates,but also destroy entironment and nature source which support all persons in the world. Therefore,it is a forever task for humen to prevent fire breaking out and reduce losing owning to fire.For several thousands years,humen being have been researching how to prevent fire to happen and break out.To this day until 1980s,humen have developed a new science named fire science.lt is the main point of this subject to research fire how to take place,how to prevent fire and some base theories about preventing fire.Those thoreies include building a fire,fire extending, smoke spreading.Morever,interaction of envriment and system.A new method of prventing fire is by using computer to calculate, analyze and prevent fire successful in the end.the arose of this science make humen enter a new epoch.A neuro-fuzzy network applied in fire detection system is proposed in this paper .Input sigals are first fuzzified and then processed in a neural network,which is a forward BP network with three layers.BP. BP algorithm is useed to train network parameters.Thereafter,fire is detected by means of fuzzy logic.Experiments show that this network can detected different standard test fire accurately and have strong anti-interference capability.Together with this technology,a hardware flat base on PC 104 computer is established.And software of this system is given and a problem about how to improve speed of calculation is discussed...