Short-term PV Power Prediction Based on K-means Clustering and Extreme Learning Machine Combination Algorithm
Considering that the prediction accuracy of PV power strongly depends on the influence of factors such as weather modes and climatic conditions,a short-term PV power prediction method based on an extreme learning machine combination algorithm was proposed.Firstly,the weather typing based on K-means clustering algorithm was divided into 12 different groups of weather categories of sunny,cloudy and rainy weather under four seasons.Secondly,the PV power prediction model was constructed based on three algorithms:extreme learning machine(ELM),genetic algorithm im-proved extreme learning machine(GA-ELM),and bird flock algorithm improved extreme learning machine(BSA-ELM)for weather typing results.Finally,the proposed model was validated with a PV plant data.The prediction results show that the BSA-ELM has the highest prediction accuracy,and the prediction accuracy of 12 kinds of weather reaches about 90% ,and the weather type with the highest prediction accuracy in each season is sunny,and the accuracy of cloudy weather is higher than that of rainy weather,which can provide effective data support for the safe and stable operation of the new power system containing a high proportion of grid-connected PV.
万方数据
维普
