In order to solve the problem of performance degradation

poor robustness

easy to generate overlearning and zero classification interval for support vector data description（ SVDD） in training and learning process

a double surround hypersphere datadomain description（ DSHDD） model is proposed. The method designs a big bounding sphere on the basis of small bounding ball

and the data is divided into three parts by the double bounding sphere: the normal region

the anomalous region

and the reject region. The experimental results show the applicability and effectiveness of the proposed method.