Data-driven pressure estimation and optimal sensor selection for noisy turbine flow with blocked clustering strategy

Physics of Fluids, Journal Year: 2024, Volume and Issue: 36(12)

Published: Dec. 1, 2024

The design and control of turbomachinery require a wealth spatiotemporal data. Thus, the low-cost robust estimation global aerodynamics from extremely limited data noisy measurements is an important problem. This paper describes data-driven approach to estimate full-field pressure distribution turbine cascade flow in combination with sparse-distributed sensor measurements. For offline library building online reconstructing, reduced-order model based on standard proper orthogonal decomposition (POD) least squares approximation, sparse representation overcomplete dictionary L1 norm minimization are leveraged. To enhance reconstruction accuracy robustness varied selections, novel blocked K-means clustering strategy developed reconstruct field through superposition multiple local clusters. statistical results indicate that outperforms gappy POD high-noise measurement environments due its superior noise effective feature selection. By applying proposed strategy, significantly improved. mean square error can be reduced by 9.86% for at 90% span blade. Sparse produces excellent enhancement when intensity exceeds 0.3. Overall, framework this exhibits outstanding advantages robustness.

Language: Английский

Fractal-constrained deep learning for super-resolution of turbulence with zero or few label data

Computer Physics Communications, Journal Year: 2025, Volume and Issue: 312, P. 109548 - 109548

Published: March 24, 2025

Language: Английский