KNN算法在矿井水源识别中的应用

发布时间：2018-03-24 01:20

本文选题：矿井水源　切入点：KNN　出处：《安徽理工大学》2017年硕士论文

【摘要】：在煤矿井下,发生的水害水灾是矿井安全工作中的重点防治对象。突水是水灾主要的体现,一旦发生,则会造成严重的人身和经济损失。所以,防治水害的工作是非常重要的。在水害防治工作中,对于矿井水源的识别工作也是必不可少的,对于传统的识别方法,如水化学方法,其耗时长、效率低等缺点都没能很好地解决。针对这些情况,本文提出了利用KNN算法结合LIF技术在矿井水源识别的应用。首先分析煤矿井下水源的由来,详细介绍其产生的原因与现阶段矿井水源所处的地下层,分析对矿井安全的危害。然后对矿井水源的水样提取做出了要求和介绍,对于矿井下水样的提取工作,是非常困难的,而且所提取的水样需要进行实验前处理,达到实验所需的要求。再对实验所用的实验设备进行了介绍,实验的设备是自主研制的矿用设备,目前处于实验室阶段。利用该设备,对所采集到的矿井水源进行光谱数据的采集,设置好设备参数,保证采集过程在暗室进行,之后将采集的光谱原始数据存储在上位机中,待用。在光谱数据处理之前,需要对其进行光谱预处理,本文采用多种预处理方式,起到对比的作用,在其中选取最佳的光谱预处理方法。本文还介绍了 KNN算法以及一些改进的KNN算法,对于改进的算法进行了原理分析。并在实验中进行多种改进的KNN算法同时对光谱数据进行处理分类,在改变K值的基础上,对多种改进KNN算法的准确度进行分析,选取最佳的KNN算法。实验所用到的软件有MATLAB和SPSS,对数据处理有很大的功能,操作起来也非常简单。最后,对来自淮南某一矿区所采集的矿井水源进行了实际的分类实验,利用改进的KNN算法对光谱数据进行分类,所分类的准确度非常可观,再次证明了 KNN算法在矿井水源识别中的应用是非常可行的,而且具有很高的使用价值。对于KNN算法在矿井水源中的应用,本文所提出的这种识别分类方法是第一次应用。对于其仿真结果和实际的实验分析结果来说,都说明了,KNN算法在矿井水源识别的应用中是非常值得研究的。也充分展示了,LIF技术在此领域的特殊之处,能够快速的建立模型对未知的水样进行识别分类。这对于今后的煤矿产业安全工作,起到了里程碑性的进步。
[Abstract]:In the coal mine underground, the water disaster flood is the key prevention object in the mine safety work. Water inrush is the main embodiment of the flood. Once it occurs, it will cause serious personal and economic losses. The prevention and control of water hazards is very important. In the prevention and control of water hazards, it is also necessary for the identification of mine water sources, and for traditional identification methods, such as hydrochemical methods, it takes a long time. The shortcomings of low efficiency have not been solved well. In view of these conditions, this paper puts forward the application of KNN algorithm combined with LIF technology in mine water source identification. Firstly, the origin of underground water source in coal mine is analyzed. This paper introduces the causes of mine water source and the underground layer of mine water source at this stage, analyzes the harm to mine safety, and then makes a request and introduction to the water sample extraction of mine water source, which is very difficult to extract mine water sample. Moreover, the extracted water samples need to be treated before the experiment to meet the requirements of the experiment. Then the experimental equipment used in the experiment is introduced. The experimental equipment is a self-developed mine equipment, which is currently in the laboratory stage. To collect the spectral data of mine water source, set the parameters of the equipment to ensure that the collection process is carried out in the dark room, and then store the original spectral data in the upper computer to be used. Before the spectral data processing, In this paper, we choose the best spectral pretreatment method, and we also introduce the KNN algorithm and some improved KNN algorithm. The principle of the improved algorithm is analyzed, and several improved KNN algorithms are used to process and classify the spectral data in the experiment. On the basis of changing the K value, the accuracy of the improved KNN algorithm is analyzed. The best KNN algorithm is selected. The software used in the experiment is MATLAB and SPSS, which has great function in data processing and is very simple to operate. Finally, the actual classification experiment of mine water collected from a mining area in Huainan is carried out. Using the improved KNN algorithm to classify the spectral data, the accuracy of the classification is very considerable. It is proved that the application of the KNN algorithm in mine water source recognition is very feasible. For the application of KNN algorithm in mine water source, the method proposed in this paper is the first application. It shows that the application of KNN algorithm in mine water source identification is very worthy of study, and fully demonstrates the special features of LIF technology in this field. It can quickly establish the model to identify and classify the unknown water samples, which is a milestone progress for the future safety work of coal mine industry.
【学位授予单位】：安徽理工大学
【学位级别】：硕士
【学位授予年份】：2017
【分类号】：TD745.2

【相似文献】