Research On The Intelligent Control Methods Of Doubly-fed Wind Energyconversion Systems

As a renewable clean energy which has the fastest development and large-scale development prospects,wind power is very significant to relieve energy shortage, and has gradually become a hot study all over theworld. However, the randomness of the wind will lead to problems such as low efficiency of wind energy,poor stability of wind energy conversion system (WECS), so it has theoretical significance and practicalengineering value to study the high performance control methods for WECS to improve energy conversionefficiency, output power quality and system reliability problems.According to the characteristics of the WECS based on doubly-fed induction generator (DFIG), in thisdissertation, combining with the latest developments of neural network control technology, stable operationof WECS and maximum wind energy dynamic capture are systematically investigated.The main contentsare as follows.1.The detailed review about the study of WECS at home and abroad is represented. The structure ofthe doubly-fed WECS and the operating principle are analysed in the dissertation, and double dynamicmathematical models are set up based on the multiple-time-scale features of wind speed.2.The neural network PID intelligent controller for WECS is designed, the designed method ofcontroller is explained, and stability of the controller is discussed.Combining fuzzy control theory andneural network PID controller,and fuzzy neural network PID controller for WECS is designed to realizetorque control of the wind generator and compensation of the pitch angle, and stability of the system isdiscussed.3.The data-driven controller for doubly-fed WECS is designed based on the data-driven thought. Inorder to improve the system control precision and capture maximum wind energy, the neural networkcompensator is designed to dynamic compensate the output of the system.4.Compensation control method of multi-variable collective pitch with dynamic feedforward neuralnetwork for doubly-fed WECS is studied, using weight pruning technique to dynamic adjust the weight inhidden layer of the neural network compensator so as to ensure stability of system output and restrainfluctuation of the wind generator speed.Based on neural network inverse control,model-free adaptivecontrol of multi-variable independent pitch for WECS is studied to reduce the influences of unbalancedload.The proposed methods are verified through the Matlab simulation and part hardware-in-loopsimulation experiment. The research will promote the control of wind power systems and provide a newthought to its rapid development in China.