Aiming at the problems of single feature extraction and low detection accuracy of the existing abnormal power consumption behavior detection methods, an improved ant lion optimization algorithm and improved support vector machine are proposed to detect the abnormal power consumption behavior of power users. The decision tree is used to optimize the support vector machine into a multi-level classifier, and the ant lion optimization algorithm is improved to optimize the parameters of the support vector machine to improve the training speed. A variety of abnormal power consumption behaviors are analyzed through experiments to verify the superiority of the proposed method. The results show that the proposed method has higher detection accuracy and lower training time than traditional anomaly data detection methods.