High-entropy alloys break the design limitations of traditional metal material component systems， greatly expand the application space of metal material component lineage and performance design and engineering， and the generative model design method provides a new technical means for the design and performance prediction of high-entropy alloy systems. In this paper， a structured generative adversarial model composed of Maximum Mean Difference Variational Autoencoder（MMD-VAE） and Wasserstein Generative Adversarial Network with Gradient Penalty （WGAN-GP） is established， and three new energetic high-entropy alloy systems are generated by learning the performance parameters of three types of NbTaW energetic high-entropy alloys， and the energy density of the material system under the constraint of copper-like density is taken as the core index， and their energy density characteristics are predicted and analyzed. The results show that the accuracy of the structured generative adversarial model for the design and performance prediction of energetic high-entropy alloy systems is significantly better than that of the single MMD-VAE model algorithm， with an overall coefficient of determination of 0.7326 and a root mean square error of 0.0540， the accuracy of the generated data is balanced. This paper provides an efficient and reliable model method for the design and performance prediction of energetic high-entropy alloys of the new system.