Abstract:

Traditional turbulence detection methods need to use empirical formulas and parameterized models, and the correctness of the formulas and models greatly influence the accuracy of the detection. The turbulence target detection algorithm based on back propagation (BP) neural network multi-level classification method does not need to use empirical formulas and parameterized models, with the utilization of neural network classification function, it can effectively establish the relationship between radar echoes and the intensity of turbulence only through the study of large amounts of data. The simulation results show that the proposed method has a good accuracy in the classification of four turbulence intensity grades, namely, negligible, mild, moderate, strong. And the accuracy will be greatly improved in the classification of two turbulence intensity levels, namely to determine whether there is turbulence or not. The theory and practice results show that the proposed method can effectively detect the turbulence target.