This paper presents DT-AOF, a decision-theoretic adaptive optimization framework designed to address dynamic optimization problems under uncertainty. The proposed method integrates three key components: explicit uncertainty modeling to capture stochastic system variations, an adaptive weighting mechanism to balance multiple competing objectives over time, and a physics-informed model to enforce structural consistency and improve solution stability. By jointly optimizing these components within a unified framework, DT-AOF achieves robust and efficient decision-making in non-stationary environments. Extensive experiments and ablation studies demonstrate that DT-AOF consistently outperforms deterministic baselines and partially constrained variants in terms of overall cost reduction and convergence behavior. The results further reveal that removing any individual component leads to noticeable performance degradation, highlighting the necessity of their synergistic integration. These findings indicate that DT-AOF provides a principled and effective solution for complex adaptive optimization tasks with inherent uncertainty.