Short-term forecasting of wind turbine power generation based on Genetic Neural Network

Abstract

Measurement Model is the main approach for short-term power generation prediction of a wind turbine generator system (WTGS), which utilizes the relationship between power generation and wind speed. This paper introduces genetic neural networks (NN) technique for wind speed and power generation prediction of a wind turbine generator system. Firstly, the following 3 hours wind speed was predicted by means of Neural Network with measured wind speed data of latest 24 hours, and then the wind power generation was forecasted based on the standard power curve of the WTGS. In order to test the predict precision different neural networks (NN), this paper also compares three NN models: standard BP, Momentum BP and Genetic Algorithm. The results show that Genetic Neural Network is a more effective and accurate method to predict wind speed and wind turbine power generation.