RCB object detection method based on Eckart weighted diagonal loading

In the context of passive sonar target detection for Unmanned Underwater Vehicles (UUVs), conventional beamforming and Minimum Variance Distortionless Response (MVDR) methods are commonly employed. However, due to the limited number of snapshots in real environments and potential errors such as element spacing inaccuracies and channel amplitude-phase errors, both the sampled covariance matrix and the nominal steering vector of the desired signal exhibit considerable errors compared to their ideal values. The computed beamforming weighting vector significantly differs from the theoretically optimal vector, leading to poor performance of the beamformer. Additionally, weak targets in underwater passive signal processing often exhibit relatively prominent spectral features only at certain frequency components within the processed frequency band. The commonly used energy accumulation-based post-processing method tends to smooth out weak targets, ultimately resulting in ineffective target detection. To address this issue, this study proposes an improved robust beamforming algorithm that utilizes the Eckart optimal filter to extract the desired spectrum from the beamformed signal spectrum, thereby determining the beam spectrum weighting values. The enhanced Robust Capon Beamformer (RCB) algorithm enhances the robustness of the beamformer against errors in the steering vector and covariance matrix. Simulation analysis and real-world data are utilized to validate the superiority of the proposed algorithm over conventional beamforming methods.