旋转机械故障知识的知识化表达模型建模问题研究

项目名称： 旋转机械故障知识的知识化表达模型建模问题研究

项目编号： No.50875118

项目类型： 面上项目

立项/批准年度： 2009

项目学科： 金属学与金属工艺

项目作者： 赵荣珍

作者单位： 兰州理工大学

项目金额： 34万元

中文摘要： 针对旋转机械智能故障诊断技术的知识获取瓶颈，从知识发现的系统工程思想出发，对旋转机械故障知识发现得以工程应用的故障状态量化特征提取、故障分类与辨识、故障知识运算模型建模问题进行了研究。一个双跨转子系统被作为研究对象。定位于使用粗糙集理论作为知识发现工具开展了研究工作。研究问题包括：1）实验转子系统多功能集成信息系统的开发；2）转子系统振动信号消噪；3）故障状态的量化信息熵特征描述问题；4）故障数据集的分类与故障辨识方法研究；5）故障特征数据库架构环境下基于决策树算法的知识规则提取方法研究；6）一个故障知识发现的概念模型与数据处理模型建模及模型优化方法。其中，在6)中的模型建立，涉及故障知识挖掘的粒度计算、知识发现数据运算的基础理论模型建立与模型工程应用方法与途径的探讨。研究表明，将机械运行信息转化成为具有开发利用价值的故障知识资源，已成为能够推动机械信息技术向数据驱动的科学方向发展而引发出的新型工程科学问题。该问题的尽快解决对发展先进的机械信息技术具有奠基作用。 研究发现，工业数据已成为关系国家制造业科学发展的重要战略资源。故建议加大对此类研究项目的资助力度。

中文关键词： 旋转机械；故障诊断；知识挖掘；知识表达模型；粒度计算。

英文摘要： The project was located in the knowledge acquisition bottleneck of intelligent faults diagnosis technology of rotating machinery. Starting from the ideas of system engineering and knowledge discovery in database, this project investigated the quantitative features extraction about faults conditions, and classification & patterns recognition of the faults, as well as to build the concepts model to calculate the faults knowledge. The way in engineering application of faults knowledge discovery was explored. A rotor system with two spans was designed as the research's object. And the researches were developmented around the rough sets theory to use as a tool of knowledge discovery. The several tasks are completed as follows. First, a set of multi-functions integration information system with to acquisition the vibration signals of rotor system is developed successfully. Second, The methods of signals's de-noising is studied. Thirdly, the information entropy to description the faults conditions with quantitative values are considered. Fourth, both the classification of faults dataset and method of pattern recognition are investigated sequentially. In the fifth, to extract out the knowledge rules by the decision-tree algorithm embedded in the SQL-Server 2005 paltform are tried to. In the last part, both a concepts model of faults knowledge discovery and its data processing are investegated out. Among them the model consistes of the granular computing of fault knowledge mining and the establishment of basic theory model to knowledge discovery as well as the application of modeling method. The results shows that the running information of rotating machinery can be transformed into a valuable data and knowledge resources. They have the developing and utilizating value to solve the bottleneck of knowledge acquisition. So the related problems on the thing should be turn into a class of new type basic science research contexts. It can put forward that one of developing trands in the information technology of mechanical equipment is data driven direction. To solve the decision-making problems by machnie intelligence the data driven will paly an important role to achieve an advanced level in the information technology of mechanical equipment. The industrial data resource has become a kind of important strategic resources involved a country science development. So investigations to preserve the industrial data with scientific way should be paid attention to as soon as possible.

英文关键词： Rotating machinery; Fault diagnosis; Knowledge mining; Knowledge expression model; Granular computing.