As a key component, the rolling bearing plays a decisive role in the safe operation of the whole wind turbine. To solve the problem of fault diagnosis in rolling bearings, a fault diagnosis method based on optimized directed acyclic graph combing with large margin distribution machine (O-DAG-LDM) is proposed. Combining the advantages of DAG multi-class scalable features with the generalization performance of LDM two-classifier, a DAG structure extended LDM multiple classifier method for rolling bearing fault diagnosis is constructed. In the framework of the DAG-LDM method, a node optimization algorithm is used to optimize the DAG nodes to reduce the cumulative error caused by random permutation, and improve the accuracy of LDM fault classification. The experiment shows that the proposed O-DAG-LDM method for fault diagnosis has higher accuracy and better capability of anti-noise immunity in comparison with other mainstream intelligent diagnosis methods.