A method for identifying outliers of power grid load data fluctuations based on IPSO-Stacking integrated learning

The traditional power load anomaly recognition method has limitations, such as low efficiency, strong subjectivity, and many human factors, and it is difficult to meet the accuracy requirements of modern power systems. Therefore, an integrated learning method based on IPSO-stacking for power grid load data fluctuation outliers was designed. The feedback mechanism of the Elman function was used to modify the DNN training network, and improved particle swarm optimization was used to optimize the weight and threshold of the DNN model. An integrated learning framework is introduced, and the abnormal load data domain is innovatively constructed for the load curve of abnormal electricity usage patterns to identify the fluctuating outliers in the grid load data. The experimental results show that the load identification effect of the design method is good, the coverage performance is stable above 90%, and the accuracy of the calculation score is higher, which can meet the accuracy requirements of modern power systems for the recognition of outliers.