Fundamental Advances in Inverse Mechanics Towards Self-Aware and Intrinsically Adaptable Structural Systems
Abstract
The objective of this project was to advance computational methods for solving inverse problems related to smart morphable structures that can evaluate their current environment and then adapt accordingly to optimize performance with minimal energy expense. A computational framework was established for the optimal design of morphing structures comprised of active smart materials through locally controllable actuation and activation. The computational optimal design framework for smart material morphing structures was then extended to substantially improve the computational efficiency of the overall solution procedure by replacing the non-gradient-based optimization with gradient-based optimization. In addition, this work sought to explore the use of reduced-order modeling for accurate and efficient inverse problem solution algorithms, and investigated inverse problem solution efficiency in a general sense. The reduced-order modeling component was primary focused on nondestructive evaluation type inverse problems (e.g., inverse characterization). A computational approach was established to create physics-based reduced-order models for representing the behavior of a boundary value problem (e.g., solid mechanics, heat transfer, etc.) in a generally applicable way, with minimal computational cost, but with accuracy commensurate with traditional finite element analysis methods. Then, an entirely new variation on the use of proper-orthogonal decomposition (POD) bases for inverse material characterization problems through a Gappy POD reconstruction procedure combined with direct inversion was hypothesized, developed, and tested. Lastly, the use of multi-objective optimization, rather than the typically utilized single-objective format, for non-gradient-based optimization-based inverse problem solution strategies was investigated and evaluated.
Document Details
- Document Type
- Technical Report
- Publication Date
- Nov 30, 2014
- Accession Number
- AD1010972
Entities
People
- John C. Brigham
Organizations
- University of Pittsburgh