From bio-inspired concepts to intelligent design: evolution and development trends of bio-inspired mechanical structural design methods

As the geometric complexity, design dimensionality, and functional requirements of bio-inspired structures continue to increase, conventional approaches based on morphological imitation, parametric modeling, and iterative finite element optimization face growing limitations in structural representation, design-space exploration, and computational efficiency. These challenges are driving bio-inspired structural design from experience-driven approaches toward data- and image-driven intelligent design. Unlike existing reviews organized mainly by structural type or biological prototype, this review follows the evolution of design paradigms and the complete technical workflow, from biological inspiration and digital representation to structure–property database construction, forward prediction, inverse generation, and closed-loop design. Conventional strategies, including morphological reconstruction, mechanism extraction, hierarchical, gradient, topology, and optimization-based design, are first summarized with their advantages and limitations. Methods for digital characterization, automated modeling, high-throughput finite element analysis, and database construction are then reviewed, emphasizing the transition from explicit parameters to image, voxel, and graph representations. The applications of machine learning, deep learning, and graph neural networks to scalar, full-process, and full-field response prediction are compared, followed by recent advances in inverse design using surrogate models and generative models such as GANs, VAEs, and diffusion models. Finally, key challenges and future directions are discussed. This review establishes a unified framework linking biological inspiration, digital representation, performance prediction, intelligent generation, and experimental validation for next-generation intelligent bio-inspired structural design.