Probabilistic Decision Methods Of Voltage And Reactive Power Control For Wind Power Integrated Power System

The large-scale wind power centralized integration affects the performance of Automatic Voltage Control(AVC)in power system,and presents challenges in traditional AVC which is formulated as a deterministic static optimization.To guarantee the security and economic,it is a key issue in dealing with the stochastic,intermittent behavior of wind power injection,and achieving the spatio-temporal coordination of different kinds of control devices.This thesis proposes probabilistic decision methods of voltage and reactive power control for wind power integrated power system,which includes the pilot-bus selection,network partitioning algorithm,and stochastic reactive power optimization.The major research is as follows:Considering the stochastic behavior of load and wind power injection,an optimal pilot-bus selection method considering source-load uncertainty is proposed.Load uncertainty is characterized by forecasting load value of typical period with forecasting error following normal distribution,and wind power uncertainty is determined by the wind speed model following Weibull distribution.The objective is to minimize the square sum expectation of load bus voltage deviation under source-load stochastic disturbance,while the voltage of pilot-bus is constant.The bus voltage deviation is calculated by the sensitivity considering the performance of primary voltage control.Immune genetic algorithm combined scene method is utilized to solve this problem.The source-load power is characterized by three scenarios,and load power scenario is corresponding to the peak,middle and valley state.Case studies based on the IEEE 39-bus system and comparative analysis of existing study are provided to illustrate the effectiveness of the proposed method.Only one load level is considered in traditional pilot-bus selection method and network partitioning algorithm,and the interaction of each other is ignored.Considering the load uncertainty,a coordinative optimization of pilot-bus selection and network partitioning algorithm is proposed.Load uncertainty is characterized by forecasting load value of typical period with forecasting error following normal distribution.There are three objectives as follows:(1)maximize the controllability of reactive source to pilot-bus,which is characterized by the short circuit impedance of the zonal reactive power;(2)maximize the controllability of pilot buses to load buses,which is characterized by voltage control sensitivity considering the performance of secondary voltage control;(3)minimize the effect of pilot buses to load buses of other zones.Static reactive power balance of each control zone under stochastic load power disturbance is considered as the constraint.Multi-objective immune genetic algorithm combined scene method is utilized to solve this problem.Case studies based on the IEEE 39-bus and IEEE 118-bus system are provided to illustrate the effectiveness of the proposed method.The existing research on stochastic static reactive power optimization does not differentiate the regulating characteristics of discrete/continuous devices,which is difficult to satisfy the control decision requirement of AVC.Therefore,stochastic static optimal reactive power optimization considering requirement of AVC for discrete devices is proposed.Source-load uncertainty is characterized by forecasting value with forecasting error following normal distribution.The adjusting cost of discrete devices and the expected energy loss in optimization period are minimized while considering the voltage amplitude security constraints under source-load stochastic disturbance.In the model,the discrete control variables represent deterministic decision variables which are constant under random fluctuation of source-load power,and the continuous control variables represent random decision variables which follow the random fluctuation of source-load power.A hybrid intelligent algorithm combined point estimate method is utilized to solve this problem.The hybrid intelligent algorithm is based on parallel immune genetic algorithm and predictor corrector primal dual interior point method.Two case studies on IEEE 14-bus and IEEE 118-bus system with real wind farm and system load data are provided to illustrate the effectiveness of the proposed method.The existing stochastic dynamic optimal reactive power optimization considering the adjusting cost of discrete devices,separating the coupling between various segmentations,solves this problem by converting to 24-independent-segmentation stochastic static optimal reactive power optimization,which may lead to frequent switching of discrete devices.Therefore,stochastic dynamic optimal reactive power optimization combined source-load curve segmentation is proposed.Fisher optimal segmentation is used to segment the load curve and wind power injection curve with the objective of minimizing the power fluctuation in each segment.Source-load power uncertainty is characterized by forecasting value of 288 time sections with forecasting error following normal distribution.Considering the bus voltage amplitude security constraints under stochastic source-load disturbance,the objective is to minimize the total adjustment costs in each period.A hybrid intelligent algorithm combined point estimate method is utilized to solve this problem.The hybrid intelligent algorithm is based on parallel immune genetic algorithm and predictor corrector primal dual interior point method.Two case studies on IEEE 14-bus and IEEE 118-bus system with real wind farm and system load data are provided to illustrate the effectiveness of the proposed method.