The WRF Model Forecast Wind Speed Correction Under Complex Terrain Based On Kalman Filter

As one of the clean energy, wind energy is being widely developed all over the world. Wind power generation is an important means of wind energy utilization, hence accurate wind power forecasting will effectively improve the wind power prediction level, thereby increasing the effective wind farm output and providing a powerful guarantee for the stable operation of power grid. This research analyzes five typical spatial-temporal distribution characteristics of wind energy resources in five typical and complex terrains via using observation data from the anemometer tower. Based upon this foundation, Kalman filter is employed to correct the wind field simulated with WRF mesoscale in the aforementioned five terrains. Conclusions of this research are as follows:(1) Among the five typical and complex terrains, the maximum wind speed is in Inner Mongolia at each height. Wind speed in Xinjiang is the second fastest and Jiangxi the slowest. However with the influence of altitude, the wind power density in Xinjiang is the biggest(with an altitude of263m). The power density of Inner Mongolia (with an altitude of1520m) is less than Xinjiang, and again the density in Jiangxi is the least. Obviously the wind direction under5typical complex terrains is greatly influenced by different terrains, seasons and atmospheric circulation.(2) Via correcting for the WRF model forecast wind speed based on Kalman Filter under different terrains, it is found that the corrected wind speed is much closer to the observed data. In addition, root mean square error, relative root mean square error, mean absolute error and relative mean absolute error decreased clearly. The correction effect is remarkable.(3) When the simulation result of WRF model has bigger errors and the prediction result is poorer, the Kalman filter correction is but more effective and the number of corrected errors decreased more dramatically. This is mainly due to the fact that the Kalman filter is excellently capable of tracking.(4) By comparing the WRF model simulation results under MRF and MYJ, which are two different boundary layer schemes, the result shows that the calculation results under both schemes are significantly similar in both Xinjiang Alashankou and Shandong Weihai. However in both Wulatefeng of Inner Mongolia and Poyang Lake of Jiangxi, the calculation results under MRF prevails MYJ. This indicates that wind field simulation under different complex terrains accordingly requires different parameterization schemes of boundary layers for the WRF mode.(5) For the simulation of WRF model under different terrains, climate characteristics, atmospheric circulation and local weather situation will produce different effects. Seasonal alteration may also influence the effect of simulation result. However, Kalman filter uses the state-space model of linear random processes which consist of state equation and measurement equation to describe. With the rule of the minimum variance estimation, Kalman filter uses a recursive method to make an optimal estimation of the state variables in order to filter out the noises arisen from different climate characteristics, atmospheric circulation and weather situation. Meanwhile Kalman filter joins extremum regression, therefore produces a closer corrected result to the observed data, resulting in a better correction.