Operation Pattern Based Probabilistic Load Flow Analysis Of Power System With Wind Farms

The rapid growth of wind power penetration and wind power’s characteristicsof strong randomness and being difficult to regulate are bringing more challenges tothe safe operation of power system. For the promotion of wind power developmentin China, one of the key problems is how to realize the on-line monitoring of theoperation risk of power system with wind farms and then the on-line optimization ofpower system operation mode. To solve this key problem, one of the methods is toimprove the prediction precision of wind speed and wind power, and then buildreasonable probability distribution models which are applicable to on-lineprobabilistic load flow, both on the load side and on the wind power generation side.Firstly, on the basis of previous research, this paper proposed a supershort-term forecast method of wind speed and wind power based on similar windseries to improve the forecasting accuracy. A searching algorithm of similar windseries was given, and the characteristic of daily periodic variation of wind speed wasused to eliminate the interference of irrelevant wind series. Using mean absolutepercentage error (MAPE) as the evaluation index, the presented method wascompared with time series forecasting method and persistence model forecastingmethod. Simulation results showed that the presented method was more accurate.Then, this paper discussed how to realize wind power forecast on the basis of windspeed forecast. The proposed forecasting method could provide a technical tool forthe study of the operation pattern and the operation pattern based joint probabilitydistribution on the wind power generation side in subsequent chapters.Secondly, this paper analyzed the unreasonableness of taking the injectionpower of different nodes as independent random variables in the process ofprobabilistic load flow calculation. Then this paper considered the correlation of theload of different nodes and that of the wind power of neighboring wind power plantsindirectly from the perspective of operation patterns, and built joint probabilitydistribution models of the load of different nodes and the power injection ofneighboring wind power plants under different operation patterns. The total powerinjection and the ratios of the power injection of each area or each node to the totalpower injection, both on the load side and on the wind power generation side, wereobtained by Monte Carlo simulation based on joint probability distribution modelsunder relative operation patterns, and then probabilistic load flow calculation wasconducted. A test in IEEE14-bus system showed that the presented model couldsimultaneously consider the probabilistic characteristics of the total power injectionand the power injection of each area or each node, as well as their correlation, which made the results of probabilistic load flow calculation more reliable.At last, this paper applied the proposed joint probability distribution of powerinjection and probabilistic load flow method, which were based on operation pattern,to the on-line regulation of power system operation mode. Sampling wasimplemented on the basis of the proposed joint probability distribution, and all ofthe typical operating states and their probabilities were gained throughcomparatively less times of power flow calculation, and then the CDF curves andprobability indices of all the node voltage amplitudes and branch power flows wereobtained. Simulation results demonstrated that the presented method could greatlyreduce the number of power system operating states to be considered and thecalculating amount, which presented a practical technical approach to applyingprobabilistic load flow to the on-line evaluation of power system operation risk andthe on-line optimization of power system operation mode.This research work is supported by the National High Technology Research andDevelopment of China (863Program)(NO.2011AA05A105) and State GridCorporation of China (SGCC)(NO. SGCC-MPLG-023-2012).