Electric Anomaly Monitoring Model Based on Improved SVM

Aiming at the abnormal power consumption monitoring of smart grid, this paper puts forward a power abnormal monitoring model based on improved support vector machine (SVM), which can meet the needs of real-time monitoring while improving the detection accuracy and response speed. In this paper, the key electrical parameters, including voltage, current, power and frequency, are collected from smart meters and environmental sensors, and then the collected data are preprocessed to remove noise, deal with abnormal values and standardize the data. Then, in data preprocessing, the penalty parameter c of SVM model is optimized by radial basis function kernel. The experimental results show that the average detection delay of the improved SVM monitoring analysis is 287.36ms, and the highest detection coverage rate of the monitoring analysis is 99.5%, which reflects its efficiency and comprehensiveness in anomaly monitoring.