Wind power generation system and its wind alignment regulation based on feedforward control

This study aimed to improve wind resource utilization efficiency and overcome the effects of wind fluctuation on wind power generation systems (WPGSs). A novel WPGS and a method of regulating its wind alignment (WA) based on feedforward control were developed. First, the changes in wind direction and speed were obtained through a convolutional neural network–bidirectional long short-term memory (LSTM) network modeled in an end-to-end manner. Second, an improved multi-degree-of-freedom permanent-magnet synchronous wind generator (WG), which neglected the yaw system, was designed. Furthermore, the generator was optimized by adopting different deflection stator slot widths, thereby increasing both the response speed and WA precision. Finally, a novel wind regulation control strategy combining an LSTM network and model predictive control was formulated, which can be used to control the proposed WG efficiently and accurately in real time. Furthermore, parameters such as the start time of WA were optimized. This study demonstrated the effective use of wind-forecasting results for the WA regulation of a WG. The proposed system could achieve approximately 6 %–30 % higher power and 3 %–29 % higher efficiency than those of traditional wind power systems.