Surrogate-Assisted Inverse Design of the Power-Law Index in Axially Functionally Graded Fluid-Conveying Pipes for Target Modal and Deflection Performance.
Saved in:
| Title: | Surrogate-Assisted Inverse Design of the Power-Law Index in Axially Functionally Graded Fluid-Conveying Pipes for Target Modal and Deflection Performance. |
|---|---|
| Authors: | Gao, Lun1,2 (AUTHOR), Gu, Jijun2,3 (AUTHOR) gu@cup.edu.cn, Guo, Tianjin1,2,3 (AUTHOR), Zhao, Shanshan1,2 (AUTHOR), Li, Junjie1,2 (AUTHOR) |
| Source: | Materials (1996-1944). Sep2026, Vol. 19 Issue 18, p3936. 20p. |
| Subjects: | Surrogate-based optimization, Power law (Mathematics), Particle swarm optimization, Frequencies of oscillating systems, Performance-based design, Fluid flow |
| Abstract: | A surrogate-assisted inverse design framework is presented for selecting the power-law index of clamped–clamped axially functionally graded (AFG) Timoshenko pipes conveying fluid. The GITT model, 336-sample database, and trained MLP forward surrogate were developed in our previous study; the present contribution begins with the formulation and solution of the inverse problem. Millisecond-speed MLP inference is embedded in grid, particle swarm optimization (PSO), and genetic algorithm (GA) searches for prescribed modal-frequency and deflection targets. Single-variable, two-variable, weighted-sum, constrained, and Pareto formulations are examined. Continuous candidates from the single-variable cases are subjected to GITT-database interpolation verification, which is explicitly distinguished from a new independent GITT calculation. The feasible single- and dual-modal examples produce small database-interpolated target residuals, whereas an intentionally unattainable triplet case retains an approximately 16% fundamental-frequency residual and demonstrates the need for feasibility screening. The two-variable maps reveal non-unique parameter couplings and are treated as exploratory surrogate results unless the complete candidate coincides with, or is independently assessed against, the available database. A five-network ensemble assesses repeatability with respect to training-data partitioning only, while a sensitivity analysis connects the selected indices to the high-sensitivity gradation range. In the synchronized Case A benchmark, the online MLP grid and PSO searches require approximately 0.044 and 0.210 s, respectively, excluding the inherited offline GITT-database construction cost. [ABSTRACT FROM AUTHOR] |
| Copyright of Materials (1996-1944) is the property of MDPI and its content may not be copied or emailed to multiple sites without the copyright holder's express written permission. Additionally, content may not be used with any artificial intelligence tools or machine learning technologies. However, users may print, download, or email articles for individual use. This abstract may be abridged. No warranty is given about the accuracy of the copy. Users should refer to the original published version of the material for the full abstract. (Copyright applies to all Abstracts.) | |
| Database: | Engineering Source |
|
Full text is not displayed to guests.
Login for full access.
|
|
| Abstract: | A surrogate-assisted inverse design framework is presented for selecting the power-law index of clamped–clamped axially functionally graded (AFG) Timoshenko pipes conveying fluid. The GITT model, 336-sample database, and trained MLP forward surrogate were developed in our previous study; the present contribution begins with the formulation and solution of the inverse problem. Millisecond-speed MLP inference is embedded in grid, particle swarm optimization (PSO), and genetic algorithm (GA) searches for prescribed modal-frequency and deflection targets. Single-variable, two-variable, weighted-sum, constrained, and Pareto formulations are examined. Continuous candidates from the single-variable cases are subjected to GITT-database interpolation verification, which is explicitly distinguished from a new independent GITT calculation. The feasible single- and dual-modal examples produce small database-interpolated target residuals, whereas an intentionally unattainable triplet case retains an approximately 16% fundamental-frequency residual and demonstrates the need for feasibility screening. The two-variable maps reveal non-unique parameter couplings and are treated as exploratory surrogate results unless the complete candidate coincides with, or is independently assessed against, the available database. A five-network ensemble assesses repeatability with respect to training-data partitioning only, while a sensitivity analysis connects the selected indices to the high-sensitivity gradation range. In the synchronized Case A benchmark, the online MLP grid and PSO searches require approximately 0.044 and 0.210 s, respectively, excluding the inherited offline GITT-database construction cost. [ABSTRACT FROM AUTHOR] |
|---|---|
| ISSN: | 19961944 |
| DOI: | 10.3390/ma19183936 |
