Photovoltaic Power Forecasting Based on SVM Optimized by Improved ABC

Abstract

In order to improve the accuracy of photovoltaic power output prediction, a photovoltaic power prediction method based on similar days and improved artificial bee colony support vector machine is proposed. Firstly, through calculating the Euclidean distance of history day and measured day meteorological factors to determine similar days. Secondly, select historical data of photovoltaic power output, temperature, humidity and daily radiation on the slope of similar days and temperature, humidity and daily radiation on the slope of test date as input variables of support vector machine. And we adopt the improved artificial bees colony to optimize kernel function parameters and the penalty factor of support vector machine. Finally get the output in each period of photovoltaic power prediction. The experimental results showed that the proposed method can effectively improve the prediction accuracy of photovoltaic power.