A Multi-model Interval Prediction Method for Photovoltaic Power

Abstract

The present invention provides a seasonal multi-model interval prediction method for photovoltaic power generation based on extreme learning machine and kernel density estimation. First, the output power, power deviation and power change rate of photovoltaic power station are analyzed. Obvious seasonal distribution characteristics; the deterministic prediction model of photovoltaic output in different seasons was established through the extreme learning machine neural network; secondly, the error distribution of the deterministic prediction was fitted by the non-parametric kernel density estimation method, and then a certain Confidence level for PV power prediction intervals. The invention can describe the possible fluctuation range of the photovoltaic power under different confidence levels, provides a way to evaluate the reliability of the prediction interval, and provides support for the risk assessment and system reliability analysis of the photovoltaic power station.

Description

technical field [0001] The invention relates to a photovoltaic power generation seasonal multi-model interval prediction method based on extreme learning machine and kernel density estimation, which belongs to the field of photovoltaic power generation prediction. Background technique [0002] Photovoltaic power generation is developing rapidly and has become one of the most promising energy sources. As of the end of 2017, the global installed capacity of photovoltaic power generation increased by 102GW, and the cumulative installed capacity reached 405GW. Photovoltaic power generation is affected by uncontrollable meteorological and environmental factors, and its output has large fluctuations and randomness. The grid-connected large-scale photovoltaic power generation system brings great impact on the stability of the grid, the quality of power grid power, and the operation and scheduling of the power system. influences. Accurate photovoltaic power forecasting can effecti...

AI Technical Summary

Problems solved by technology

[0003] At present, the research in the field of photovoltaic power forecasting mainly focuses on deterministic forecasting, but the deterministic forecasting method cannot effectively describe the distribution of forecasting errors, and it is impossible to make decisions about the reserves that the system needs to reserve based on pure photovoltaic deterministic forecasting in scheduling operation. Engineering applications are constrained

[0004] In addition, photovoltaic output has obvious periodicity and nonlinearity, and its distribution characteristics are different under different weather conditions in different seasons, resulting in obvious differences in the mapping relationship that needs to be fitted by the photovoltaic power generation prediction model under different weather conditions, and a single model is used to realize Accurate prediction of power generation under various weather conditions is very difficult

Images