【作者】 曾永龙；

【导师】 曾良才；

【作者基本信息】 武汉科技大学 ， 机械电子工程， 2008， 硕士

【摘要】 设备监测和故障诊断一直以来都是工业领域研究的热点,一方面人们对设备的安全、稳定、可靠性的要求逐步提高;另一方面网络技术、信号分析与处理技术广泛成功应用,使得基于网络的远程监测和故障诊断模式越来越受到国内外的重视。正是在这一背景下,本文结合炼铁厂出铁场除尘风机监测和故障诊断课题做了相关的研究,文中着重研究了风机监测和故障诊断的技术实现,并对关键技术进行了分析,对需要解决的问题作了探讨。(1)设计风机状态监测与故障诊断系统的框架结构。系统主要由在线监测子系统和智能诊断子系统组成,并对数据采集系统、在线监测系统、信号分析和故障诊断系统等各部分进行了分析和设计。(2)开发基于单片机的数据采集系统。设计了数据采集方案,运用Proteus ISIS建立仿真系统,并写入KeilC平台下对应的编译程序对系统仿真,最后开发了基于AT89S52的数据采集系统。系统按照指令采集各个传感器上的状态值然后通过CAN总线传给服务器,服务器对数据进行处理并存储,用户可以通过网络远程监测风机的实时数据。(3)研究故障诊断数据融合处理的方法。首先要根据系统的要求以及信源的特点探讨一种以ART-2网络聚类分析为核心的数据融合算法,运用ART-2网络数据融合实现分类的机理实现同类数据融合,然后再根据己有的多源信息和系统的融合知识采用一定的学习方法,对所建立的神经网络系统进行离线学习,确定网络的连接权值和连接结构拓扑,最后把得到的网络用于冶金风机故障诊断系统中。系统采用BP网络实现异类数据融合并实现对风机系统设备运行状态的实时监测和故障诊断,验证了该算法的有效性和可行性。更多还原

【Abstract】 Equipment monitoring and faulty diagnosis have always been the research hotspots in industrial field. On the one hand, there are rising requirements of the safety, stability and reliability for equipments. On the other hand, the network-based remote monitoring and fault diagnosis model has been getting more and more attention at home and abroad, along with the widely successful application of network technology as well as the signal analyzing and processing technology. Our research was launched just under this background. It was combined with the monitoring and fault diagnosis study on dedusting fan of the cast house at ironmaking plant. In this paper, we mainly studied the realization of the technology of fan monitoring and fault diagnosis, analyzed the key techniques and made a relatively deep discussion about the problems to be resolved.The frame structure of the fan remote monitoring and diagnosis system was designed. The system was mainly composed of the on-line monitoring subsystem and the intelligent diagnosis subsystem, which were able to analyze and design the data acquisition system, on-line monitoring system, signal analysis and fault diagnosis system and so on.To develop the data acquisition system based on a single chip microcomputer. After the design of the data acquisition scheme and the establishment of the simulation system using Proteus ISIS, we wrote the corresponding compiler of the KeilC platform in order to simulate the system. Lastly, we developed the AT89S52-based data acquisition system which could collect the state values of each sensor according to the instruction and then transmit them to server through the CAN bus. The server was to process and store data, therefore, users could remotely monitor the real-time data through network.Research on faulty diagnosis by the method of data fusion disposal. First of all, it was essential to select an appropriate neural network model in accordance with the requirements of the system and the characteristics of the source, and to study a data fusion algorithm with ART-2 clustering analysis as the core. As the implementation of classifying mechanism and similar data fusion technology with the application of ART-2 network data fusion, some learning method was adopted into the off-line learning towards the established neural network system on the basis of the present multi-source information and the systematic fusion knowledge. Through the off-line learning, the connection weight and connecting structure topology of the network were determined. The lastly obtained network was used in the faulty diagnosis system of the metallurgical fan. The system utilized the BP network to realize the fusion of different kinds of data and to accomplish the real-time monitoring and fault diagnosis of the running state for the fan system equipment. The algorithm was proved to be effective and feasible.更多还原