• Proposes a novel intelligent design framework integrating CDCGAN with SIMP for thermal conductivity topology optimization.
• Achieves approximately 98% faster computational speed than standard SIMP and 55.5% faster than conventional deep-learning methods.
• Demonstrates significant thermal performance improvements: 50.1% reduction in average temperature and 28.2% reduction in highest temperature compared to theoretical designs.
• Addresses key challenges of continuous condition domain adaptation and design structure stability in intelligent thermal design.