Abstract: Bird strikes constitute the primary threat to aviation safety. These incidents predominantly occur during aircraft takeoff and landing phases. Within the airport apron environment， the strong radar echoes generated by large targets such as civil aviation aircraft can overwhelm the faint echoes from bird targets. Consequently， detecting bird targets against airport apron clutter is critically important. The micro-Doppler signatures generated by the wing-flapping motion of bird targets contain critical physical information that serves as a valuable basis for the identification and classification of bird targets. However， under strong airport apron clutter conditions， these components cannot easily be directly extracted. Therefore， the micro-Doppler components arising from wing-flapping echoes within the received radar signals should be separated. To address the challenge of separating micro-Doppler components of bird targets under the background of airport apron clutter， this study proposes a separation method for bird wing-flapping echoes using basis pursuit denoising （BPDN） optimized by an improved sparrow search algorithm （ISSA）. This method first improves the sparrow search algorithm （SSA） by integrating the Circle chaotic mapping， osprey optimization algorithm （OOA）， and Cauchy variation strategy. The Circle chaotic sequence is used to initialize the sparrow population， enhancing the diversity of populations. To enhance the global optimization capability， the OOA is employed to modify the explorer update formula. The Cauchy variation is used to perturb the follower positions， enhancing the algorithm’s ability to escape local optima. Subsequently， to mitigate the degradation of micro-Doppler signal separation performance caused by manual parameter selection in BPDN， the ISSA is employed to optimize the regularization parameter and the augmented Lagrangian parameter applied in BPDN. This optimization strategy mitigates the influence of manually configured parameters on algorithm performance， thereby enabling the determination of critical parameters. Finally， based on the optimal parameter combination， the radar echo signals from bird targets are reconstructed using the different sparse characteristics of multi-component signals in various transform domains， achieving the separation of airport apron clutter components， bird body components， and micro-Doppler components from bird echoes. Experimental results from both simulation and measured data demonstrate that the ISSA exhibits higher convergence speed and accuracy than conventional optimization algorithms （e.g.， particle swarm optimization）. The BPDN with optimized parameters effectively separates the micro-Doppler components of bird echoes， providing a prerequisite and theoretical foundation for the subsequent bird target parameter estimation.