Anomaly detection algorithm based on the Support Vector Data Description (SVDD) usually brings low detection rates due to background examples being contaminated by anomalous data. To deal with the problem, a new method based on SVDD with negative examples is proposed. By leading into the origin and a few anomalies as negative examples, the capability of description to anomaly and background is improved. In order to control the influence of false examples, the distance of each sample deviating from the mean of background example is mapped into the sample as weighting coefficient. The results show that the proposed method could obtain higher detection rate under low false rate than the algorithm based on SVDD. The effectiveness of the proposed method is validated by experimental results obtained from real data.