Harmonic Current Probabilistic Modeling Based on K-Fold Cross-Validation Kernel Density Estimation

In the construction of the new energy system, the new power system's "double high" characteristics bring harmonic pollution to the power grid. Meanwhile, due to the different habits of power users and new energy generation affected by some uncertainties such as weather, randomness needs to be taken into account in the harmonic assessment, and the probabilistic model is an effective method to characterize the harmonic distribution. In this paper, a kernel density estimation model with bandwidth selection by the k-fold cross-validation method is proposed for estimating the probability density of harmonic currents. The efficacy of the proposed model is assessed through a conventional kernel density estimation method with a rule-of-thumb bandwidth selection. All models are tested using the coefficient of determination, mean absolute error, and root mean square error. The results show the proposed models perform better in assessing the probability density of harmonic currents for residential and commercial loads.