To efficiently and quickly deal with noisy data,bearing fault diagnosis was modeled as a time series classification task.The singular spectrum analysis was used to extend the one-dimensional time series to the multi-dimensional time series and to realize noise reduction.The convex optimization algorithm was used to learn the FCMs model from the noise data quickly and effectively,and the time series were transformed into C×C FCMs weight matrix for representation.Neural network was used to classify the feature matrix.The CWRU bearing dataset was used to validate the proposed method.The results show that compared with the traditional method,the performance of the proposed method is better in bearing fault diagnosis.