基于SVM的刮板输送机故障诊断

发布时间：2019-05-18 17:53

【摘要】：目前,随着采煤机械化程度的不断提高,煤矿刮板输送机也向着大型化、自动化方向发展。SGZ1250/2565刮板输送机具有破碎、转载和运送功能,同时与采煤机和液压支架等矿用设备组成一个串联的输送系统。然而,刮板输送机故障频繁发生,导致整个采煤系统瘫痪,对煤矿造成巨大的经济损失与安全隐患。为确保煤矿大型设备的安全可靠运行,人们在矿用设备的故障诊断研究中投入了更大的人力和物力,煤矿设备的故障诊断技术也向着智能化方向努力。支持向量机(Support Vector Machine)是针对的小样本数据进行分类的算法,在机电设备故障数据难以获得,且数据稀少的情况下比较适用。上世纪九十年代,Vapnik和其合作者研究出SVM至今,SVM已广泛被应用在各类设备的故障诊断中。其原理是对设备的非线性故障数据,通过核函数映射到高维空间,使得故障数据在低维空间线性不可分时,在高维空间中变得线性可分。因为支持向量机是一种基于两分类问题的分类算法,怎样进行多分类,如何选取核函数,以及分类时惩罚参数和核函数参数该如何选取等问题,一直是支持向量机研究的热点。本文主要以煤矿刮板输送机SGZ1250/2565为研究对象,针对该输送机的多类故障,研究了支持向量机的分类算法理论,建立了二分类及多分类模型以及将其应用在输送机的故障诊断中,实验证明支持向量机多分类算法能够准确的对刮板输送机进行多故障诊断。在以往支持向量机研究中,核函数以及参数是通过设计者自己在实验时设定,依赖人工经验,没有科学依据,也造成实验的准确率下降,针对这一问题,本文做了适当的改进,首先采用改进网格搜索交叉验证法得到最优参数,再用此参数来建立多类故障模型,实验证明,改进网格搜索得到的参数提高了刮板输送机的故障诊断准确度。
[Abstract]:At present, with the continuous improvement of coal mining mechanization, the coal mine scraper conveyor is also developing towards the direction of large scale and automation. SGZ 1250 / 2565 scraper conveyor has the functions of crushing, reloading and transporting. At the same time, it forms a series transportation system with mining equipment such as shearer and hydraulic support. However, the failure of scraper conveyor occurs frequently, which leads to the paralysis of the whole coal mining system and causes huge economic losses and hidden dangers to coal mines. In order to ensure the safe and reliable operation of large coal mine equipment, people have invested more manpower and material resources in the fault diagnosis research of mine equipment, and the fault diagnosis technology of coal mine equipment is also working towards the direction of intelligence. Support vector machine (Support Vector Machine) is an algorithm for classification of small sample data, which is suitable for mechanical and electrical equipment fault data when it is difficult to obtain and the data is scarce. In the 1990s, Vapnik and its partners developed SVM, SVM has been widely used in fault diagnosis of all kinds of equipment. The principle is that for the nonlinear fault data of the equipment, the kernel function is mapped to the high-dimensional space, so that the fault data becomes linearly divisible in the high-dimensional space when the fault data is linearly inseparable in the low-dimensional space. Because support vector machine is a classification algorithm based on two classification problems, how to carry out multi-classification, how to select kernel functions, and how to select penalty parameters and kernel function parameters in classification have always been the focus of support vector machine research. In this paper, the SGZ1250/2565 of coal mine scraper conveyor is taken as the research object, and the classification algorithm theory of support vector machine is studied in view of many kinds of faults of the conveyor. The double classification and multi-classification models are established and applied to the fault diagnosis of conveyor. The experimental results show that the support vector machine multi-classification algorithm can accurately diagnose the multi-fault of scraper conveyor. In the previous research of support vector machine, the kernel function and parameters are set by the designer himself in the experiment, rely on artificial experience, there is no scientific basis, but also cause the accuracy of the experiment to decline. In order to solve this problem, this paper has made appropriate improvements. Firstly, the optimal parameters are obtained by using the improved grid search cross verification method, and then many kinds of fault models are established by using this parameter. The experimental results show that the improved grid search parameters improve the fault diagnosis accuracy of scraper conveyor.
【学位授予单位】：西安科技大学
【学位级别】：硕士
【学位授予年份】：2015
【分类号】：TD634.2

【相似文献】