模糊神经网络在火灾探测中的应用

Neuro-Fuzzy Network Applied in Fire Detection

本文提出将串联型模糊神经网络应用于火灾探测系统。首先将输入信号模糊化后送入神经网络进行处理，其模糊系统用指派法定义隶属度函数，神经网络采用三层前馈ＢＰ网络结构，用ＢＰ算法进行网络参数的训练。其后由模糊逻辑判决火灾或非火灾。实验结果表明该网络能够准确探测各种标准试验火并有很强的抗干扰能力。

:A neuro-fuzzy network applied in fire detection system is proposed in this paper. Input signals are first fuzzified and then processed in a neural network,which is a forward BP network with three layers.BP algorithm is used to train netwrok parameters.Thereafter,fire is detected by means of fuzzy logic. Experiments show that this network can detect different standard test fires accurately and have strong anti-interference capability.