Firstly, in view of SVM parameters optimization problem, combination to the advantage of ACROA, a new classification model, called ACROA_SVM was presented herein. Furthermore, the effectiveness and superiority of the ACROA_SVM model was identified via benchmark datasets, which was downed from the sit web of UCI. Lastly, combination to local mean decomposition and energy moment feature extraction, ACROA_SVM was served as approach of pattern recognition to identify rotating machinery fault types. The experimental results show ACROA_SVM method has higher precision, better generalization ability of fault diagnosis, and less time consumption, higher efficiency of fault diagnosis, which is conducive to realize online intelligent fault diagnosis.
