中文摘要：

在分析旋转机械振动特点和Rough Set理论的基础上，针对传统的频谱分析方法对质量不平衡、动静碰摩、支座松动等3种典型故障识别效率低的缺点，提出了一个基于Rough Set的振动故障诊断模型．该模型根据故障和能量的映射关系，分别在时域、频域、时-频域中定义4种信息熵作为条件属性，推导了3种典型振动的决策规则，实现了对振动信号中不一致信息的处理．通过汽轮发电机组振动实验对上述方法进行了验证．结果表明，该模型能够很好地识别这3种典型故障．

英文摘要：

In view of the inefficiency of traditional spectrum analysis methods to identify faults due to mass unbalance, rubbing between stationary and moving components, or slackening of pedestal＇s fastening, after having analyzed vibration characteristic of rotary machinery and the Rough Set theory, a model for vibration faults diagnosis, based on Rough set, is brought forward. According to the mapping relationship between fault and energy, 4 kinds of information entropies are respectively defined in the time domain, frequency domain and the time-frequency domain, to serve as conditional attributes. A judging criterion for the three typical vibration faults is then deduced, thus realizing a way of processing disaccording informations concerning vibration signals. The method described above has been validated by vibration tests of a steam turbo-generator unit. Results show that the method can very well identify the 3 typical faults.