基于神经网络的水晶研磨机数控系统的设计与实现

发布时间：2018-04-29 00:42

本文选题：水晶研磨 + 嵌入式　；参考：《杭州电子科技大学》2017年硕士论文

【摘要】：水晶作为一种伴随着人民生活水平提高而需求量大增的物品,对现代水晶加工行业也提出了新的要求。现有的水晶研磨机数控系统或多或少都存在一些问题,这些问题导致机器的效率没法进一步提高,水晶成品的质量也存在着一些不稳定的地方。同时随着现有的一些技术的发展,传统的水晶研磨机数控系统也需要进一步的改进。本文首先分析了水晶加工技术的国内外现状和未来的发展趋势,并详细介绍了水晶研磨机数控系统的总体设计思路和水晶研磨的原理,使用STM32F103ZET6微处理器作为控制核心,连接各个功能模块实现对研磨机的控制,详细描述了数控系统的硬件设计和软件设计。针对水晶在研磨过程中由于磨盘的损耗导致研磨出现不稳定的情况,使用神经网络建立模型优化对水晶研磨的控制。本文分析了在水晶研磨过程中的受力情况,并讨论几个影响水晶成品质量的因素,针对磨盘因为作业而磨损导致水晶表面研磨不稳定的问题,提出使用神经网络优化研磨时间以改善研磨状况,降低工人操作的要求。接着介绍了神经网络的发展和应用广泛的BP神经网络的原理,并讨论了几种BP算法的改进。在对研磨过程中的几个关键参数进行分析之后建立BP神经网络模型,采集水晶研磨机的磨盘电机转矩、接触面压力、打磨次数作为神经网络的输入,研磨时间作为输出建立单隐含层的BP神经网络,在对数据进行误差剔除和归一化处理之后对网络进行训练,使用MATLAB训练后的模型能够很好地拟合目标值。数控系统的各个模块之间有非常多的总线连接,工作的现场环境非常复杂,传统的数字键盘输入的方式不是很便利,本文使用Wi-Fi网络,通过上位机进行参数设置和数据传输,增加了设备操作的便利性,并对设备的各个模块进行了测试。本文在设计水晶研磨机数控系统的同时运用神经网络优化研磨过程,智能控制水晶的研磨时间,优化后的系统在实际工作中运行稳定,结果表明工作取得了一定的成果,为后续系统的进一步改善打下了基础。
[Abstract]:As a kind of product with the improvement of people's living standard, crystal has put forward new requirements to the modern crystal processing industry. There are some problems in the existing CNC system of crystal grinder, these problems lead to the machine efficiency can not be further improved, and the quality of crystal products also has some instability. At the same time, with the development of some existing technologies, the traditional CNC system of crystal grinder also needs to be further improved. This paper first analyzes the present situation and future development trend of crystal processing technology at home and abroad, and introduces in detail the general design idea of NC system of crystal grinder and the principle of crystal grinding. The STM32F103ZET6 microprocessor is used as the control core. The hardware design and software design of NC system are described in detail. In view of the instability of crystal grinding due to the wear of the disc during the grinding process, the neural network model is used to optimize the control of crystal grinding. In this paper, the stress in the process of crystal grinding is analyzed, and several factors affecting the quality of crystal products are discussed. The neural network is used to optimize the grinding time to improve the grinding condition and to reduce the requirement of workers' operation. Then it introduces the development of neural network and the principle of BP neural network, and discusses the improvement of several BP algorithms. After analyzing several key parameters in the grinding process, the BP neural network model is established, and the torque of the disc motor, the contact pressure and the grinding times of the crystal grinder are collected as the input of the neural network. The BP neural network with single hidden layer is established as the output of grinding time. After error elimination and normalization of data, the network is trained. The model after MATLAB training can fit the target value well. There are many bus connections between the modules of the NC system, the working environment is very complex, the traditional way of digital keyboard input is not very convenient. In this paper, we use the Wi-Fi network to set parameters and transmit data through the upper computer. The convenience of equipment operation is increased, and each module of the equipment is tested. In this paper, the NC system of crystal grinder is designed, and the grinding process is optimized by using neural network, the grinding time of crystal is controlled intelligently, and the optimized system runs stably in practice. The results show that certain achievements have been made in the work. It lays the foundation for the further improvement of the follow-up system.
【学位授予单位】：杭州电子科技大学
【学位级别】：硕士
【学位授予年份】：2017
【分类号】：TS933;TP183

【参考文献】