基于双谱图像的机械故障嵌入式诊断系统研究

发布时间：2018-01-09 20:36

本文关键词：基于双谱图像的机械故障嵌入式诊断系统研究　出处：《郑州大学》2017年硕士论文　论文类型：学位论文

【摘要】：作为机械系统传动组件最重要的齿轮,它的工作状况关系到整个机械设备的运行状态。设备一旦发生故障,将带来十分严重的危害,造成巨大的经济损失。因此,人们对机械设备故障诊断高度重视并研究。与通常的故障识别的途径不同,本文利用图像识别技术,通过分析齿轮故障信号双谱灰度图的纹理特征,搭建具有满足对功耗、功能、体积、成本、可靠性严格要求性能的嵌入式平台,并在该平台上编写Android应用程序通过使用BP神经网络分类算法识别出齿轮的故障类型。主要研究内容如下:(1)研究信号的滤波、降噪等处理技术,得到齿轮振动信号的双谱灰度图。(2)研究基于灰度矩的图像特征提取,以作为分类的输入特征向量。(3)研究BP神经网络模式识别算法对第一类灰度矩进行分类,从而识别出齿轮故障类型。(4)了解嵌入式系统基础知识,掌握嵌入式系统环境的搭建和移植。(5)设计各模块原理图,编写硬件驱动程序并进行调试。(6)开发基于Android的齿轮箱故障诊断系统软件。
[Abstract]:As the most important gear of the transmission component of mechanical system, its working condition is related to the running state of the whole mechanical equipment. Once the equipment breaks down, it will bring very serious harm. Therefore, people attach great importance to the fault diagnosis of mechanical equipment and study it. Different from the usual way of fault identification, this paper uses image recognition technology. By analyzing the texture features of the bispectral grayscale image of gear fault signal, an embedded platform with strict requirements for power consumption, function, volume, cost and reliability is built. The Android application program is written on this platform and the fault type of gear is identified by using BP neural network classification algorithm. The main research contents are as follows: 1) study signal filtering. The image feature extraction based on gray moment is studied by using noise reduction and other processing techniques to get the bispectral grayscale image of gear vibration signal. Based on the input feature vector of classification, the BP neural network pattern recognition algorithm is used to classify the first grayscale moment, and the gear fault type is identified. 4) to understand the basic knowledge of embedded system. Master the construction of embedded system environment and transplant. 5) Design the schematic diagram of each module, write the hardware driver and debug. 6) develop the software of gearbox fault diagnosis system based on Android.
【学位授予单位】：郑州大学
【学位级别】：硕士
【学位授予年份】：2017
【分类号】：TH17

【相似文献】