The gray wolf algorithm is a type of optimization algorithm that mimics the behavior of gray wolves in nature. This heuristic optimization algorithm is widely used in radar classification and other complex military industrial applications due to its effectiveness. This study focuses on the application of the gray wolf algorithm in modern electronic battlefield radar for radiation source detection. An innovative enhancement scheme is proposed by optimizing the index probability weight of the gray wolf algorithm to create the enhanced gray wolf optimization (EGWO) algorithm. This enhancement aims to improve the convergence speed and global search capabilities of the gray wolf population optimization problem. The EGWO algorithm is compared with six other classic swarm intelligence algorithms, including particle swarm optimization (PSO), through MATLAB simulation tests on nine standard benchmark functions. Results show that the EGWO algorithm achieves superior global solutions and faster convergence speeds. Additionally, the EGWO algorithm is applied to a real four-classification problem of radiation sources. The results indicate that the EGWO-enhanced model outperforms the original model, a convolutional neural network (CNN)-based improved model, and three models enhanced by PSO. The EGWO model demonstrates lower classification loss and superior performance in terms of F1 score, precision, recall, and accuracy. These findings highlight the efficient search performance of the EGWO algorithm in addressing high-dimensional complex optimization problems. This suggests potential future applications of the EGWO algorithm in radar classification, intelligent optimization of bio-fermentation processes, and other complex scenarios, particularly within the military industry domain.