Abstract: This paper studies the method of identifying anomalies in electric energy metering data in the electricity consumption information collection system. The abnormal data in the electric energy measurement data is collected through the electricity consumption information collection system,and the improved particle swarm algorithm is selected to optimize the support vector machine kernel function parameters to construct the power quality disturbance model,and the classification of the electric energy measurement abnormal data collected is implemented. The LOF algorithm calculates the anomaly factor,and then the flyaway anomaly intelligent analysis method is used to determine the disturbance model to judge whether the indication value of the electric energy meter is abnormal,thus the abnormal identification process of the electric energy measurement data is completed. The experimental results show that the accuracy of this method for classifying abnormal data is as high as 98.50%,and its detection time is only 1.121 s,better than the comparison method,therefore it can better prevent judgment errors, and effectively improve the quality and efficiency of abnormality judgment of electric energy metering data.