Topology optimization(TO)has become a core computational paradigm for structural design by defining optimality through physics-based objectives and constraints.However,practical engineering design often involves incom...Topology optimization(TO)has become a core computational paradigm for structural design by defining optimality through physics-based objectives and constraints.However,practical engineering design often involves incomplete and imperfect physical modeling due to multi-physics coupling,manufacturing uncertainty,and computational constraints,leaving critical design factors insufficiently captured in purely physics-driven formulations.In parallel,data-driven and generative methods have enabled rapid topology generation and intent-aware design exploration,yet often weaken explicit optimality guarantees.This review argues that these seemingly divergent developments can be organized under a unified information-physics perspective.We term this emerging field Topology Optimization Informatics(TOI):optimal structural design is obtained through the joint modeling and optimization of physical laws and design-relevant information.We first summarize the integration of artificial intelligence(AI)and TO into two major paradigms:AI-based one-shot TO,which learns mappings or distributions of near-optimal designs from data and prioritizes fast generation and diversity,and AI-enhanced iterative TO,which embeds learning-based modules into the classical solver-in-the-loop pipeline while keeping the underlying governing equations unchanged.Finally,we show that traditionally separate tasks—design control,computational acceleration,and fidelity enhancement—can be interpreted as different manifestations of information-physics co-modeling within a single optimization framework,thereby clarifying their connections and design implications and outlining opportunities for semantic-and data-enabled next-generation structural design.展开更多
基金funded by the Guangdong Basic and Applied Basic Research Foundation(2024A1515011786 and 2025A1515010672).
文摘Topology optimization(TO)has become a core computational paradigm for structural design by defining optimality through physics-based objectives and constraints.However,practical engineering design often involves incomplete and imperfect physical modeling due to multi-physics coupling,manufacturing uncertainty,and computational constraints,leaving critical design factors insufficiently captured in purely physics-driven formulations.In parallel,data-driven and generative methods have enabled rapid topology generation and intent-aware design exploration,yet often weaken explicit optimality guarantees.This review argues that these seemingly divergent developments can be organized under a unified information-physics perspective.We term this emerging field Topology Optimization Informatics(TOI):optimal structural design is obtained through the joint modeling and optimization of physical laws and design-relevant information.We first summarize the integration of artificial intelligence(AI)and TO into two major paradigms:AI-based one-shot TO,which learns mappings or distributions of near-optimal designs from data and prioritizes fast generation and diversity,and AI-enhanced iterative TO,which embeds learning-based modules into the classical solver-in-the-loop pipeline while keeping the underlying governing equations unchanged.Finally,we show that traditionally separate tasks—design control,computational acceleration,and fidelity enhancement—can be interpreted as different manifestations of information-physics co-modeling within a single optimization framework,thereby clarifying their connections and design implications and outlining opportunities for semantic-and data-enabled next-generation structural design.
