Radar Emitter Signal Classification (RESC) plays a critical role in electronic warfare and communication systems, particularly for accurately identifying signals under challenging conditions, including low Signal-to-Noise Ratio (SNR). This study addresses these challenges by proposing an Advanced Particle Swarm Optimization - Maximum Overlap Discrete Wavelet Packet Transform (APSO-MODWPT) framework. MODWPT offers enhanced frequency resolution and alignment suitable for radar signals; however, parameter optimization remains complex. This issue is addressed using Particle Swarm Optimization (PSO) guided by Kullback-Leibler divergence. Experimental results across BPSK, QPSK, 8-PSK, and 16-QAM signals demonstrate significant SNR improvements. Visual analysis via Continuous Wavelet Transform further validates the framework’s effectiveness, significantly reducing noise while preserving echo signal fidelity. This highlights the approach’s effectiveness in accurately extracting vital signal features and filtering out unwanted noise, enhancing signal analysis quality and reliability in radar research. The proposed method promises improved signal fidelity and more accurate radar insights.