A Method for Modeling the Output Power of Photovoltaic Power Supplies Based on Non-parametric Kernel Density Estimation

Abstract

To address the shortcomings of the traditional PV power output model, which requires the assumption of parameter distribution and cannot fully consider the influence of various random factors, this paper adopts the k-nearest neighbor estimation method and the kernel density estimation method based on the non-parametric estimation theory to describe the PV power output respectively, and proposes an improved optimal bandwidth calculation model without reference to the overall distribution for the kernel density bandwidth selection problem, and selects the actual measured data of a large PV power supply in a certain region for simulation and comparison analysis to verify the correctness, validity, and adaptability of the proposed kernel density estimation probability model to the random characteristics of different PV power supplies.