Wind Farm Clutter Mitigation For Weather Radar Based On Clutter Characteristics

With the rapid development of the wind power industry in China,the construction of wind farms may have serious interference on the performance of existing weather radars and other equipment.As a special big moving obstacle,wind turbine will cause weather radar to produce detection blind zone.The rotation of the blades and the strong reflectivity of the mast will affect the estimation accuracy of weather target parameters.Traditional ground clutter suppression techniques cannot completely eliminate such wind farm clutter with rotational characteristics.Therefore,the study of wind farm clutter suppression technology for weather radar is of great significance to improve the accuracy of weather forecasting and reduce the requirements of wind farm site selection.In this paper,the signal models of weather target and wind farm clutter for weather radar are firstly established.On this basis,the cluster dispersion characteristics of weather targets and the cyclostationary characteristics of wind farm clutter are studied,which provides a theoretical guidance for the subsequent wind farm clutter suppression techniques.Secondly,based on the different motion characteristics of weather targets and wind farm clutter,the clutter dictionary is constructed according to the three parameters of wind turbine speed,initial angle and observation azimuth.The inverse solution of Bessel's geodetic problem is used to estimate the relative distance and azimuth of weather radar and wind turbine,and only the clutter pollution area data is extracted.The matching pursuit technology is used to realize the clutter suppression of the wind turbine in the scanning mode for weather radar.Finally,because the fixed clutter dictionary matrix in the matching tracking algorithm leads to degradation of clutter suppression performance for the radar echo with random interference,the K-SVD algorithm in image processing can be applied to wind farm clutter suppression.The wind turbine clutter data is only selected as the training sample to build the dictionary matrix in the proposed method.The singular value decomposition technique is used to continuously update the dictionary matrix and the corresponding coefficient matrix until convergence,and the clutter suppressed.The experimental results of simulation data and measured data show that the proposed method can effectively suppress wind farm clutter.At the same time,considering that the dictionary training and updating in this algorithm consumes a lot of computing resources,the prior information such as the orientation angle of the wind turbine and the blade rotation speed is used to reduce the training sample size.Therefore,the update speed and computational efficiency of the dictionary are significantly improved.On this basis,it is compared with the matching pursuit algorithm,and their advantages,disadvantages and applicable conditions are given.