Abstract: In the research of abnormal energy consumption detection of industrial extruder, because the data characteristics are not comprehensive and appropriate, the detection accuracy is not high. Therefore, a method based on GMM-LDA clustering feature learning and particle swarm optimization support vector machine (PSO-SVM) is proposed. GMM-LAD clustering feature learning algorithm was used to cluster some data sets to obtain the optimal features of normal data and abnormal data. The update pattern was used to dynamically generate the nearest normal pattern library and abnormal pattern library to improve the adaptability of the database, so that the proposed method could adapt to the dynamic changes of the network environment and label the data set. The parameters of SVM were optimized based on PSO to obtain the optimal model. Experimental results show that the proposed detection model not only avoids the dependence of manual classification in supervised training samples, but also has higher detection accuracy and lower false alarm rate compared with single algorithm or other algorithms.