数据融合技术在基于物联网的火灾探测系统中的应用研究

发布时间：2018-03-15 00:17

本文选题：火灾探测　切入点：数据融合技术　出处：《吉林大学》2015年硕士论文　论文类型：学位论文

【摘要】：如今，火灾已经成为最常见和最普遍的灾害之一，它不仅威胁公众安全，也影响社会发展，而火灾探测技术对于预防火灾发生以及降低灾后损失十分有效。传统的消防系统功能相对简单，用于火灾判断的特征参数单一，容易出现误报和漏报的情况。将数据融合技术应用在火灾判断中，可以获得更多被测目标和环境的信息，可以进一步提高消防系统的智能性和准确性。早在1973年，美国就已经开始有了对数据融合技术的研究，1988年将其列为重点研发技术，，1991年已经将该技术应用于军事战争。目前多传感器数据融合技术已经由分散研究转变成一个独立的研究领域，并且在信息电子学领域、计算机科学领域以及自动化领域都有着重要的影响。比较常见的数据融合方法有贝叶斯方法、D-S证据推理、人工神经网络以及模糊理论算法等。数据融合具有三层结构，针对这一特性，本文选择在不同的层次使用合适的算法。这样的做法可以利用算法各自的优势，弥补彼此的不足，进一步提高融合结果的准确性和可靠性，从而减少火灾判断的误报率。本文先是构建一个基于ZigBee技术的小型火灾特征数据采集系统，分别从硬件和软件设计两个方面进行详细的阐述，分析Z-stack协议栈的概念与应用方法。分别对该采集系统进行了相关的功能和性能上的测试，实验结果表明，该系统可以有效地采集温度、烟雾浓度和CO浓度三种火灾特征数据。然后提出将神经网络和模糊理论相结合的火灾探测算法应用在火灾探测系统中。详细地从理论上分析了这两种算法的原理以及优缺点，利用MATLAB仿真软件中的神经网络工具箱以及模糊推理工具箱，对两种算法分别进行了仿真和分析，得出将两种算法结合适用于火灾探测的结论。最后在VS2010开发平台上开发后台数据处理软件，分别实现串口数据传输、火灾探测算法以及对数据库访问三个功能。为系统提供了一个较好的人机交互平台，开发者可以观测到实时采集出来的火灾特征数据值，并可以看到系统对火灾判断后的结果，也可以访问数据库查看历史数据。
[Abstract]:Today, fire has become one of the most common and common disasters, which not only threaten the safety of the public, but also affect social development. Fire detection technology is very effective to prevent fire and reduce the loss after disaster. The function of traditional fire protection system is relatively simple, and the characteristic parameters used for fire judgment are single. The application of data fusion technology in fire judgment can obtain more information about the target and environment, and further improve the intelligence and accuracy of fire protection system. As early as 1973, The United States has already begun to study data fusion technology, which was listed as a key research and development technology in 1988 and applied in military warfare in 1991. At present, multi-sensor data fusion technology has changed from decentralized research to a new one. A separate field of research, And in the field of information electronics, computer science and automation have an important impact. The common data fusion methods are Bayesian method D-S evidential reasoning, artificial neural network and fuzzy theory algorithm. Data fusion has a three-layer structure. In this paper, we choose to use appropriate algorithms at different levels, which can make use of the respective advantages of the algorithms, make up for each other's shortcomings, and further improve the accuracy and reliability of the fusion results, thus reducing the false alarm rate of fire judgment. In this paper, a small fire feature data acquisition system based on ZigBee technology is constructed, which is described in detail from two aspects: hardware and software design. The concept and application method of Z-stack protocol stack are analyzed, and the related function and performance of the acquisition system are tested. The experimental results show that the system can effectively collect fire data of temperature, smoke concentration and CO concentration. Then a fire detection algorithm combining neural network and fuzzy theory is proposed. The principle, advantages and disadvantages of the two algorithms are analyzed in detail. By using the neural network toolbox and fuzzy reasoning toolbox of MATLAB simulation software, the two algorithms are simulated and analyzed respectively, and the conclusion that the two algorithms are suitable for fire detection is obtained. Finally, the background data processing software is developed on the VS2010 development platform, which realizes three functions respectively: serial port data transmission, fire detection algorithm and database access. It provides a better human-computer interaction platform for the system. The developer can observe the fire characteristic data collected in real time, see the result of the system judging the fire, and visit the database to view the historical data.
【学位授予单位】：吉林大学
【学位级别】：硕士
【学位授予年份】：2015
【分类号】：TP274;TN92;TP391.44

【相似文献】