Research On Steady-state Wind Farm Modeling And Application

With the excessive consumption of fossil energy and the deterioration ofenvironment, people’s demand for clean energy is growing. Wind energy hasbeen widely concerned because it is clean, renewable and abundant. The windenergy technology is increasingly mature and the scale of wind farms isexpanding, so the impact of wind power’s randomness and uncontrollability onthe safety and realibility of power system is more and more serious. Therefore, itis imperative that an exact steady-state model of wind farms established toevaluate its random output and accurately grasp the influence of grid-connectedwind farms on the security and economics of power system.The various factors influencing the steady-state output of wind farms areanalyzed and two wind farm steady-state modeling methods are proposed in thispaper, one is steady-state wind farm modeling considering wake effects andterrain and the other one is steady-state wind farm modeling consideringoperating datas. The two methods are applied into the steade-state outputevaluation of wind farms and wind power forecasting.The main research is as follows:(1) A wind farm modeling method of considering wake effects and terrain is proposed. First, the influences of wake effects and terrain on the wind speedof wind turbines are analysized and the wind speed models considering wakeeffects under flat terrain, considering and not considering wake effects undermountain terrain are built. Then, the steady-state model of Dongmafang windfarm is established based on the wind speed and output models of the windturbine. Finally, the impacts of wake effects and terrain on the output of thewind farm are studied by simulations which verify the accuracy of the modelingmethod considering wake effects and terrain.(2) Considering influences of wake effects and terrain on the output of awind farm, a steady-state output evaluation system of a wind farm is designed toanalysize changes of the steady-state output when wind speeds and winddirections vary for three wind farm models under flat terrain, mountain terrainand flat+mountain terrain. And the best wind direction and ranges of windspeeds in a certain wind direction are found out based on the system, whichprovide theoretical basis for the micro-siting of a wind farm.(3) Considering it is difficult to accurately model the Dongmafang windfarm for its complex terrain, a wind farm modeling method of consideringoperating datas is proposed. First, piecewise operating characteristics of thewind turbine are analysized. Then, wind turbines in Dongmafang wind farm areclassified by two classification methods, the maximum tree method andimproved maximum tree method, the wind speed-active power curves andfunctions of the wind turbine clusters are respectively obtained by means of statistic and piecewise linear fitting principles, and a three-turbine equivalentsteady-state model of the wind farm is established. Finally, an improvedsteady-state model is established to compare with the three-turbine equivalentsteady-state model, and the results show that the three-turbine equivalentsteady-state model is more accurate.(4) According to the classified results of Dongmafang wind turbines,average wind speeds of the three wind turbine clusters and wind farm areforcasted by BP neural network, and the forecasting results are separatelyapplied into the three-turbine equivalent steady-state and improved model toforecast the wind power. And the forecasting results show that the accuracy ofthe forecasting model is effectively improved when applying three-turbineequivalent steady-state model into the wind power forecasting model, whichprovides a effective reference for optimizing grid dispatching and reasonablyarranging the overhaul and maintenance of wind turbines.