A Novel EV-ELM-Based Recognition Method for Ship and Corner Reflector Array on HRRP Using Multivariate Statistical Features

In complex electromagnetic environments, radar detection capabilities can be significantly degraded. Corner reflectors, as a critical type of interference source in maritime ship monitoring, often form arrays to mimic the radar echoes of real ships, making them difficult to distinguish from ships in high-resolution range profile (HRRP). To address this issue, most existing methods suffer from problems such as high demand for multi-angle samples, long required accumulation time, and complex network computations. This paper proposes a learning and classification method that combines statistical features with an ensemble voting ELM network. By leveraging multivariate statistical features to suppress the peaks of nonship targets and integrating multiple ELM networks to improve model discrimination accuracy, the proposed method demonstrates advantages such as reduced demand for angle samples, faster data collection, and simpler, more computationally efficient networks. Additionally, it exhibits superior performance in recognizing centroid interference compared to conventional techniques.