Research On The Optimal Operation Of Doubly-fed Wind Turbines

Wind power for the most mature technology and the most efficient clean energy innowadays has been attended and applied by the countries. For the output power of thedoubly-fed wind generator affected by the wind speed changing, that has caused theintermittent and volatility of the wind turbine output power, and this situation poses a hugechallenge on the stable operation of power system. Optimization control method fordoubly-fed wind turbine at constant voltage operation mode and equivalent modeling basedon the flow field characteristics for doubly-fed wind farm have been mainly researched in thispaper.For the shortcoming that the PI controller parameters can’t dynamic tuning in constantvoltage control system for doubly-fed wind turbines, a PI controller parameters dynamictuning strategy based on the BP neural network and ant colony optimization algorithm ispresented. Three layers neural network structure is adopted in dynamic tuning algorithmbased on BP neural network. The optimal weights can be got from learning samples by thenerve cells of the neural network, so the parameters of PI controller can be dynamicallyadjusted during the network’s input changing. The PI controller parameters dynamic tuningbased on the ant colony optimization algorithm uses the two parameters in PI controller as thecolony of ants, uses the controlled absolute error integral function that between terminalvoltage actual value and its reference value for doubly-fed wind turbines as the optimizationobjective, and dynamically adjusts the PI controller parameters in the control process. Whenthe wind speed changes and the grid voltage drops, the dynamic and non-dynamic parametertuning methods are compared and analyzed.At present, the entire wind farm is regarded as a large-capacity wind turbine formodeling wind farm, the influence of flow field characteristics in the wind farm has beenignored. For the shortcoming of this modeling approach, an equivalent modeling method ofdoubly-fed wind farm based on the flow field characteristics has been proposed. A calculationmodel of flow field in the wind farm has been built and the influence of terrain and wake forflow characteristics was taken into consideration. The wind speed was calculated at each windturbine installed point based on the flow field calculation model, and that wind speed is asinput wind speed for each wind turbine, the dynamic characteristics of the wind farmproduction process is studied. The equivalent modeling based on that method for thedoubly-fed wind farm which has15wind turbines has been built, and the comparative studies have been done with the model using anemometer tower wind.Comparative analysis of the contents researched in this paper indicate that a PI controllerparameters dynamic tuning strategy based on the BP neural network and ant colonyoptimization algorithm have improved the stability of gird-connected voltage for doubly-fedwind turbines and reduced the response time, and equivalent modeling based on the flow fieldcharacteristics for doubly-fed wind farm can reflects the dynamic characteristics of the windfarm production process more accurately.