An iterative data-driven turbulence modeling framework based on Reynolds stress representation

Data-driven turbulence modeling studies have reached such a stage that the basic framework is settled, but several essential issues remain strongly affect performance. Two problems are studied in current research: (1) processing of Reynolds stress tensor and (2) coupling method between machine learning model flow solver. For issue, we perform theoretical derivation to extend relevant arguments stress. Then, representation theorem employed give complete irreducible invariants integrity basis. An adaptive regularization term enhance an iterative with consistent convergence proposed then applied canonical separated flow. The results high consistency direct numerical simulation true values, which proves validity approach.