Recognition of unknown meteorological objects in radar probing

The present paper gives a solution to the problem of recognition of unknown objects in radar probing of the meteorological phenomena. A non-traditional algorithm of selection and recognition of unknown signals is proposed. The algorithm is considered on the base of autoregressive model description of the intensity fluctuations of the reflected signals for the meteorological objects. Estimations of the autoregressive coefficients in the models are obtained at the training stage using the classification reflected signal samples for the given objects. An data of experimental study of the object recognition efficiency are presented using real reflected signals from a weather radar.