In order to solve the problem of redundancy, complexity and low accuracy of feature information in the open circuit fault of an inverter, a rough set greedy algorithm is proposed to discretize data, reduce attributes and extract an effective rule table. This method can realize the discretization of data without changing the indiscernible relationship of the system, extract the condition attributes with high importance from the data sample set for attribute reduction, and finally extract the effective rule table to realize the open circuit fault diagnosis of the inverter. This method solves the problem of data redundancy and complexity in open circuit fault diagnosis of the inverter, greatly reduces the feature dimension of fault data, and uses an effective rule table to diagnose the fault quickly and accurately, saving a lot of time. This work is supported by National Key Research and Development Program of China (No. 2016YFC0600906).