模糊神经网络在火灾探测中的应用研究

【作者】 高珊珊；

【导师】 李洪春； 曾昭华；

【作者基本信息】 大连理工大学 ， 控制理论与控制工程， 2004， 硕士

【摘要】 本文根据火灾探测的特点，针对于火灾探测复杂的、非线性结构的对象，提出将模糊神经网络应用于火灾探测中，以降低火灾探测的误报率、提高准确性的思想。本文首先阐述了模糊系统和神经网络在火灾探测中的应用依据，并对模糊控制、神经网络以及模糊神经网络的发展和原理等进行综述。本文采用模糊神经网络用于火灾探测中不仅能将模糊系统与神经网络的仿人思维的功能与处理非线性结构的共同特点发挥出来，而且还能各取所长，共生互补。应用多层前馈网络构造模糊变量隶属函数和模糊推理控制模型，使神经网络不再表现为黑箱式映射，其所有节点和参数都具有模糊系统等价意义。将模糊规则与隶属度函数用神经网络表现出来，利用神经网络的自学习特性实现隶属度函数和模糊规则的自动提取，可优化调整隶属函数，并且模糊系统也弥补了神经网络运算速度较慢的缺点，因此将其用于火灾探测会具有较低的误报率、较高的可靠性和较强的环境适应能力。同时本文提出模糊神经网络用于火灾探测中的模型结构，详细介绍了模糊神经网络的设计过程与算法。并对模糊神经网络进行训练，得到较为满意的结果，证明了将模糊神经网络应用于火灾探测的思想是符合实际要求的。更多还原

【Abstract】 According to the characteristics of fire detection, the thesis mainly puts forward a sort of thought of applying fuzzy neural network to fire detection in order to reduce the false alarm rate and improve the veracity. It first expatiates on the basis of applying fuzzy neural network in fire detection . It also summarizes the development and the principle of fuzzy control system, neural network and fuzzy neural network. The fuzzy system and the neural network have the ability of apery thought. They can also deal with non-linearity configuration. The application of fuzzy neural network in fire detection can exert the characteristic in common of the fuzzy system and the neural network. When the fuzzy system is combined with neural network, fuzzy rules and subjection function can be exhibited by the neural network and be picked up through self-study trait of neural network. Moreover, Fuzzy system can also make up the shortcoming of slow operational rate of the neural network. So fuzzy neural network in fire detection will have lower misinformation rate , higher security and stronger environmental adaptive capability. The thesis also introduces the construction of fuzzy neural network, expounds the design course and the arithmetic. Besides , it trains the fuzzy neural network and gains the satisfying results. This work shows that the thought is feasible.更多还原