基于人工神经网络的水压机活动横梁位置控制系统研究

【作者】 陈晖；

【导师】 谭建平；

【作者基本信息】 中南大学 ， 机械电子工程， 2006， 硕士

【摘要】 我国航空航天事业的不断发展以及国防现代化的需要，对我国大吨位锻造水压机的控制水平提出了更高的要求。我国现有的300MN模锻水压机是目前亚洲最大的水压机，是加工各种航空航天锻件以及各类军工产品的关键设备。然而，水压机现有的操作控制系统已经明显老化，不能适应现代的高精度、智能化的控制要求，对水压机的改造势在必行。在对水压机的改造研究中，水压机活动横梁位置控制系统的控制精度是衡量水压机技术先进性的重要指标之一，直接影响着水压机锻件产品的加工质量及水压机自动化生产，具有重要的研究意义。本文研究了基于人工神经网络的水压机活动横梁位置控制系统，着重对系统的基本思想、系统构建、神经网络模型建立以及神经网络超程值预测算法在上位机中的实现进行了分析和研究，其主要研究内容如下：1．研究了水压机活动横梁位置控制系统的方案，分析了水压机活动横梁位置控制系统需要解决的问题—横梁超程，分析了现有的解决方法及其局限性，提出了基于人工神经网络的水压机活动横梁超程值预测。2．研究了水压机活动横梁位置控制系统的构建，介绍了重要元件的选型。3．研究了如何根据系统实际，建立适合于本系统的神经网络模型。分析了BP神经网络常规学习训练算法的缺陷，针对其缺陷，研究了如何避免常规算法的缺陷。4．研究了如何在上位机中通过软件编程实现了水压机工作活动横梁超程值的人工神经网络预测算法。更多还原

【Abstract】 With the development of the spaceflight project and the requirement of the modern national defence, it call for higher precision and larger tonnage for the hydraulic press machine. 300MN hydraulic press machine is the biggest hydraulic press machine in both our country and Asia, It is a key machine to machining kinds of spaceflight equipment and military establishment. But the existing operating and controlling system already aging, it can not suits modern control, So rebuild the operating and controlling system of 300MN hydraulic press machine is necessary. While the rebuilding of the operating and controlling system of 300MN hydraulic press machine, The stop position accuracy of the control system of 300MN hydraulic press is the main index of its advanced performance. It affects the quality of forging products and the process automation directly. So it is very important.This paper analysis the stop position control system of the hydraulic press using artificial neural networks, emphasize on the basic principal of the system, the structure of the system and how to build the structure of the neural networks model, how to realize the program in the upper computer. The main research contents as following:1. Study on the stop position control system of 300MN hydraulic press, analysis the problem Which the stop position control system should solve. Analysis the traditional methods and the limits of these methods. Bring forward a new way to compensate the overstork using artificial neural networks.2. Study on the structure of the stop position control system of 300MN hydraulic press, and give a introduction of the election of key components of the system.3. Study on how to build the structure of the artificial neural networks model according to the system. Analysis the limits of the traditional study arithmetic of BP neural networks, and study on how to find a mended study arithmetic for the BP neural networks4. Study on how to realize the program in the upper computer system.更多还原