Radar Signal Classification Based on Quantum‐Classical Hybrid Convolutional Neural Network

ABSTRACT Due to the increasing complexity of radar signals and the limitations of traditional machine learning techniques in handling noise and high‐dimensional data, this paper proposes a quantum‐classical hybrid convolutional neural network (QCHCNN) model for radar signal classification. In our approach, radar signals are first transformed into time‐frequency amplitude spectrum images, effectively capturing their essential features. These images are then processed through a quantum convolution layer, leveraging the advantages of quantum computing to enhance feature extraction. The output is subsequently fed into a classical convolutional neural network, which further refines the data and produces accurate classification results. Our experiments show that QCHCNN outperforms in radar signal multi‐classification, demonstrating stability across datasets, noise resistance, and robustness to quantum noise. It improves accuracy by 5.65% over classical convolutional neural network, reaching 96.07%. Finally, this study highlights the potential of quantum‐classical hybrid methods to advance radar signal multi‐classification and machine learning applications.