The evolving clustering method(ECM) is a kind of efficient online clustering method, adapting the clustering results with new incoming data samples automatically. However, the algorithm not only depends on the predefined maximum distance threshold but also be sensitive to the data order, which are easily misleading if the parameter is not appropriate or the data order is different. For the above problem，an evolving clustering method based on the self- adaptive learning ECM(SALECM) is proposed. Under the situation that no priori knowledge is accessible, it can adjust clusters adaptively without any predefined parameter. Experiment results show that the algorithm can improve online clustering adaptively compared with the ECM and relieve the influence of the input sequence to the ECM to some degree.