The Optimal Load Dispatch Strategy Based On Controllable Power Units In Wind Farm

In recent years, wind power has developed rapidly. As the proportion of wind power capacity in the new energy power system is increasing, wind farms should improve the operation and control level in order to ensure the safety. For the less volatility and less impact from randomness, wind farms should increase their output control level to meet the demand of power grid. Around this goal, paper did some research on short-term wind speed forecast, adjustable power of wind turbine unit, optimal load dispatch strategy in wind farm.The thesis includes the following aspects:This paper established short-term wind speed forecast model based on AdaBoost-BP neural network and least squares support vector machine (LSSVM), then did the simulation with field data and proved the effectiveness of the short-term wind speed forecast. As characteristics and laws are the same in wind speed signal of the same frequency part, we choose the wavelet packet transform as the data preprocessing method. In basis of wavelet transform, wavelet packet transform provides more analysis of high frequency signal with higher resolution. Paper established short-term wind speed forecast model with optimal wavelet packet transform and LSSVM. Through the test with field data, results show higher prediction precision by using optimal wavelet packet transform.For the different purpose of short-term wind speed forecast, paper gave two types of wind speed forecast:multistep forecast and information granulation forecast. Then paper analyzed the contrast of two application modes through the simulation. Scheduling instruction to wind farm from power grid are usually serial numbers. Wind speed multistep forecast can show the variation trend in the future for grid continuous scheduling. Information granulation forecast can eliminate data redundancy and get accurate wind speed level and distribution for analyzing output of different wind farms or different wind power units in the future.It is proposed the adjustable power strategy based on wind power model. Traditional wind unit operation uses the most power strategy and keeps the pitch angle minimum and optimal tip speed ratio; keep rated power by adjusting the pitch angle. Paper established the wind turbine power model, analyzed wind units characteristic, proposed adjust wind turbine power strategy in the full range. Within wind speed, strategy chooses generator torque for small range adjustment and pitch angle and generator torque together for large range adjustment.We calculated the unit feature vector, established wind farm feature matrix. As the number of units in wind farm, we analyzed the clustering of all units, then set up several level clusters to reduce the dimension of unit commitment problem. Last, in the condition of wind farm power limit from power grid, we solved the unit commitment problem by considering units relative loss.