熵权特征融合的声雷达近地面风速反演算法OA

Sonic Detection and Ranging serves as a critical technology for near-surface wind field monitoring,holding significant value in wind energy development and meteorological exploration.However,its practical application faces severe challenges from Doppler spectral peak drift and complex electromagnetic noise interfer-ence,which constrain wind speed retrieval accuracy.To address these issues,this study proposes an entropy-weighted feature fusion algorithm.To overcome spectral peak drift,a Doppler window dynamic migration model based on the power-law distribution of turbulent dynamics is constructed.Targeting the critical bottleneck of insufficient robustness and accuracy in spectral peak discrimination under complex noise environ-ments,this study introduces an entropy-weighted feature fusion method grounded in information entropy theory.This method constructs a feature space incorporating key discriminative indicators such as signal-to-noise ratio and Doppler frequency shift difference,while dynamically assigning weights to each feature using information entropy theory,thereby effectively mitigating the subjectivity and environmental adaptability limitations inherent in traditional feature weight assignment methods.Furthermore,to ensure the physical rationality of symmetric beam results,a symmetry constraint is introduced for optimization.Experimental validation demonstrates that the entropy-weighted feature fusion algorithm significantly enhances the spectral peak identification capability and wind speed inversion accuracy of sonic wind radar in complex interference scenarios,providing an effective solution for high-reliability near-surface wind field monitoring.