To address the disconnection between aesthetic requirements and parameter configuration in ceramic ornament design, this study proposes a bidirectional coupling model acting as an algorithmic bridge connecting aesthetic artistic expectations and ceramic firing physical constraints, integrating aesthetics and parameters. The model uses the Analytic Hierarchy Process to quantify indicator weights, employs a Generative Adversarial Network for ornament generation, and integrates Particle Swarm Optimization for global parameter optimization, forming a closed-loop framework. Experimental results show that the model achieves a style feature matching rate of 92.53 ± 0.97%, a decorative detail restoration degree of 97.41 ± 0.75% (N=5 independent replicates, 95% confidence interval, p<0.05), and a process parameter compliance rate of 91.47 ± 1.18% in the simple style of Song Dynasty ornaments. Compared with other models, it outperforms the Variational Autoencoder–Analytic Hierarchy Process–Particle Swarm Optimization model (87.42±1.35% style matching), the Generative Adversarial Network with elitist strategy and non-dominated sorting genetic algorithm Ⅱ (82.43±1.58% detail restoration), and the Generative Adversarial Network–Particle Swarm Optimization model (73.34±1.81% process compliance). The model achieves over 90% accuracy in recognizing multiple ceramic styles. Note that the present model is only validated against Song-Ming-Qing Chinese ceramic styles and its generalization to other ceramic cultural systems remains limited. These results demonstrate its effectiveness in coupling aesthetic requirements with parameter configuration and supporting digital inheritance and innovative ceramic ornament design.
