基于多波段的火灾探测系统神经网络算法研究-通信与信息系统专业论文.docxVIP

华 华 中 科 技 大 学 硕 士 学 位 论 文 II II Abstract Fire is a disaster which occurs frequently and results in huge loss. Nowadays, it is an important subject in the field of safety protection. However, traditional method of fire detection is to employ a single fire detector such as heat detector, smoke detector and so on, it has a high rate of false alarm. As a result, such detection method cannot meet the demands of modern safety protection. With the fast development of sensor technology, microprocessor technology and intelligent algorithm, it is possible to research intelligent fire detector based on multi-sensor. So it is possible to cut down the time of early fire-alarm, increase the accuracy of fire judging, and actively protect all kinds of industrial and high-rise civil buildings. In this thesis, firstly we introduce the basic theoretical knowledge about current development situation of fire detection at home and abroad, principle of fire occurrence and fire detection method,etc.Then we expound the function and constitute of system, and introduce microprocessor，named MSP430F149 and main peripheral modules.Meanwhile, we make a detailed analysis of system software designing. After this, we introduce a typical algorithm named cross correlation, which is simple and has a high performance. At last, we introduce an intelligent algorithm named neural network. It has the property of self-adaption, learning capacity, fault-tolerance and parallel processing, so it is widely used in the field of information processing, and opens up a new development approach for fire signal processing. We detailedly expound the structure model and algorithm principle of neural network algorithm, and make use of MATLAB neural network toolbox to make emulation. After analyzing the result, we make a conclusion that there is a high accuracy if neural network is used in fire detection. So we put neural network to use in fire detection. By this, the rate of fire false-alarm can be decreased, and the ac