Maximum power point tracking control of the permanent magnet synchronous generator based wind turbine

摘要

Wind power generation is a promising renewable energy source. The reduced cost of electricity supplied from wind power plants may be attributed to good control strategies such as maximum power point tracking. The control algorithm for maximum power generation is analysed in this thesis. The control algorithm is proposed by regulating the d-q axis voltages of electrical machines in order to control machine torque and rotational speed that allows wind turbines to always extract maximum power from the wind energy source. A conventional way to control the electrical machine is by using vector control together with PI controllers to regulate voltages. This control method is mature and robust enough for electrical machine control. However, vector control may have difficulties in handling system interconnected nonlinearity and time varying wind power input variables. To improve the control strategy and provide controllers with a wider range of applicability, feedback linearization and nonlinear adaptive control algorithms are investigated. Feedback linearization control cancels out all the nonlinearities of q-axis items to expand operational range and develop interaction between the d-axis and q-axis dynamics for machine torque. For nonlinear adaptive control, the original nonlinear multi-input multi-output system is divided into inter-related subsystems and the system nonlinear items and uncertainties are estimated in order to cancel out the existing nonlinearities. Wind power generation maximum power point tracking is accomplished by using conventional vector control, feedback linearization control and nonlinear adaptive control. Practically, due to the small range of control capability, the gain-scheduled conventional control strategy requires a set of control parameters in order to match the different input wind speed. And a mapping technique which relies on the wind speed and current sensors is essential for this control strategy. The feedback linearization control strategy proposed in the report gives global trajectory tracking, so only one set of controller parameter is able to handle all the different wind speed inputs. However, the feedback linearization control still requires some of the machine operational parameters such as rotor speed, stator winding current, etc. Therefore, the nonlinear adaptive control strategy is proposed which uses the estimated machine operational parameters instead of actual parameters. This would further improve the controller capability and robustness. The simulation in this thesis have shown that the proposed nonlinear control strategies are also able to conduct wind turbine maximum power point tracking compare to conventional gain-scheduled control strategy. In the real case, if the proposed nonlinear control strategies can be successfully implemented for wind turbine, it will reduce the number of sensors and the corresponding devices used and thus reduce the cost and enhance the wind turbine robustness. A magnetic equivalent circuit model of the permanent magnet synchronous machine is developed to analyse the electrical machine performance consider magnetic saturation. This model is usually used for electrical machine design and optimization purpose. It has a significant advantage in computational speed compared to another popular tool, finite element method. The magnetic equivalent circuit model may be used to calculate electrical machine properties such as electromotive force and flux linkages for machine control. The flux worked out by using this model is compared with finite element method analysis and the result shows that this model is five times faster in calculations and gives the percentage error less than ten. Currently, due to the uncertainties of magnetic saturated machines, the electrical machine controller only handles the linear region of machine power speed curve. If the proposed model has the calculation speed fast enough to give real time machine operational parameters, the uncertain parameters can be obtained even when the machine encounters magnetic saturation. It has to be emphasized that the nonlinearities in the magnetic equivalent circuit model is due to the magnetic material, while the nonlinearities in machine controller are due to the summation or product of multiple state variables, they are essentially different.