Case-based reasoning introduced into RCM analysis process simplifies the tasks of RCM analysis and shortens time. However, redundant features may not only increase the case memory, but also make the case retrieval algorithm more complicated. Additionally, traditional methods of weight allocation increase the human subjective influence on the accuracy of case retrieval. This paper applies fuzzy rough set algorithm in feature reduction and weight allocation which is used for case retrieval of similar equipment in RCM analysis case-based reasoning. This method effectively avoids information loss caused by discretizing continuous feature value in cases. Finally, a case study is implemented to steam feed pump performance features.