基于小波分析及数据融合的电气火灾预报系统及应用研究

发布时间：2018-07-12 18:42

本文选题：电气火灾 + 故障电弧　；参考：《燕山大学》2013年博士论文

【摘要】：火推动了人类社会的文明进步，而火灾却给人类带来了巨大的危害。随着现代社会经济的快速发展以及工业的不断繁荣，各种电气化产品的层出不穷，给火灾的发生提供了更大的可能性。多年来电气火灾的数量一直呈现居高不下的局面，而且损失惨重的重特大火灾往往也由电气火灾造成。传统的火灾预报由于探测技术、信号处理方法和理论研究的局限性，在电气火灾监测过程中时常会出现误报及漏报。论文深入分析了电气火灾形成机理，在分析出电弧（电火花）和高温为电气火灾火源的根本形式的基础上，通过大量实验，深入研究了不同负载形式下的交流故障电弧燃烧时的电弧电压、电流波形特性后发现，交流电弧在燃烧过程中有潜在着的“零休现象”。故障电弧的“零休现象”特性，给故障电弧的检测拓宽了思路。提出了利用故障电弧检测与分析监测及预报电气火灾的方法。运用小波函数对故障电弧电流信号进行了小波奇异性分析。构造了正交二次样条小波为小波函数，利用多孔算法的二进小波变换实现了快速小波变换算法。故障电弧周期零休现象这一特征信息用小波分析时表现为周期性的奇异点，因此提出了周期性奇异点检测故障电弧的新算法，并分析了该故障电弧检测算法的可行性和有效性。在检测故障电弧发生的基础上，对电气火灾早期现场的主要特征信号进行了多参数实时监测，运用多信息融合技术完成了对所探测的电气火灾特征信息的融合，实现了电气火灾的准确辨识。设计了基于故障电弧的信息融合的三层模型，并运用我国标准火数据以及典型干扰数据进行了实验仿真，仿真结果表明，该融合模型能够很好地完成电气火灾的快速准确预报，有效地避免了电气火灾的误报和漏报。采用集散控制方法，完成了基于故障电弧和多信息融合的电气火灾预报系统的系统设计。整个系统分为上下位机，下位机又分为主机和从机。下位机主要完成信号的采集、预处理以及传输，其中的主机可完成一定的信号处理与判断；上位机主要完成各种信号处理算法的实现、存储以及监控系统画面的实现。所研发的“电气火灾预报系统”经过了反复试验、调试并在多家应用单位进行了推广使用，较好地实现了电气火灾的预防。论文在电气火灾预报方面进行了一定的研究工作，取得了一定的进展，但是，电气火灾仍然有许多值得研究的热点，例如，在故障电弧进一步与电气火灾其他参量的融合方面、电气火灾融合模型结构的优化方面、采用新型探测技术和探测器扩展现有系统的能力方面、其它领域的新技术（如激光图像粒径分群、激光前向/后向散射的应用）引发电气火灾探测技术的新途径方面以及电气火灾监测技术在与自动化、现代通讯技术、智能大厦技术的进一步结合使得电气火灾探测系统更趋于自动化、开放性和模块化等方面还会有更进一步的发展。
[Abstract]:Fire has promoted the civilization and progress of human society, and fire has brought great harm to mankind. With the rapid development of modern social economy and the continuous prosperity of industry, the emergence of various electrified products has provided greater possibility for the occurrence of fire. The number of caller gas fires has been in high level for many years. And the heavy and heavy fires are often caused by electrical fires. The traditional fire prediction, due to the limitations of detection techniques, signal processing methods and theoretical research, often Misreports and Misreports in the process of electrical fire monitoring.
In this paper, the formation mechanism of electric fire is deeply analyzed. On the basis of the basic form of arc (electric spark) and high temperature as the fire source of electric fire, through a lot of experiments, the arc voltage of alternating current fault arc burning under different load forms is deeply studied. After the current wave shape characteristics, it is found that the AC arc has the potential in the combustion process. The "zero rest" phenomenon, the "zero rest" characteristic of the fault arc, has widened the idea of the detection of the fault arc. The method of using the fault arc detection and analysis to monitor and predict the electric fire is put forward.
Wavelet function is used to analyze the singularity of the fault arc current signal by using the wavelet function. The orthogonal two spline wavelet is used as the wavelet function, and the fast wavelet transform algorithm is realized by the two progressive wavelet transform of the porous algorithm. The characteristic information of the fault arc period zero rest is a periodic singular point in the small wave analysis. This paper proposes a new algorithm for detecting arc fault by cyclic singularity detection, and analyzes the feasibility and effectiveness of the algorithm.
On the basis of detecting the occurrence of the fault arc, the main characteristic signals in the early stage of the electric fire are monitored in real time, and the information fusion of the electric fire detection is completed by the multi information fusion technology, and the accurate identification of the electric fire is realized. The three layer model of information fusion based on the fault arc is designed. The experimental simulation is carried out with the standard fire data and typical interference data in China. The simulation results show that the fusion model can complete the rapid and accurate prediction of electrical fire, and effectively avoid the false alarm and false alarm of electrical fire.
The system design of electric fire prediction system based on the fault arc and multi information fusion is completed by the method of distributed control. The whole system is divided into the upper and lower computer, the lower machine is divided into the host and the slave. The lower machine mainly completes the acquisition, preprocessing and transmission of the signal, and the host can complete certain signal processing and judgment; The bit machine mainly completes the realization of various signal processing algorithms, stores and monitors the realization of the system picture. The "electric fire prediction system" has been tested repeatedly, debugged and popularized in many application units, and the electric fire prevention is well realized.
In this paper, some research work has been carried out in the field of electric fire prediction, and some progress has been made. However, there are still a lot of hot spots in the electric fire. For example, in the aspect of the fusion of the fault arc and other parameters of the electric fire, the optimization of the structure of the electric fire fusion model, the new detection technology and detection are adopted. The ability to expand existing systems, new technologies in other fields (such as laser image particle size distribution, application of laser forward / backward scattering) lead to a new approach to electrical fire detection technology and electrical fire monitoring technology in conjunction with automation, modern communication technology, Intelligent Building technology to make electrical fire detection The system will be more automated, and there will be further development in terms of openness and modularity.
【学位授予单位】：燕山大学
【学位级别】：博士
【学位授予年份】：2013
【分类号】：TM76;X934

【参考文献】