Extracting information overlap in simultaneous OH-PLIF and PIV fields with neural networks

Simultaneous measurements, such as the combination of particle image velocimetry (PIV) for velocity fields with planar laser induced fluorescence (PLIF) species fields, are widely used in experimental turbulent combustion applications analysis a plethora complex physical processes. Such analyses driven by interpretation spatial correlations between these experimenter. However, also imply some amount intrinsic redundancy; simultaneous carry overlapping information content. The goal this work lies quantitative extraction overlap field measurements. Specifically, PIV contained simultaneously measured OH-PLIF domain swirl-stabilized combustor is sought. This task accomplished using machine learning techniques based on artificial neural networks designed to optimize PLIF-to-PIV mappings. It was found that most content could be retrieved when considering linear combinations neighborhoods signal spanning roughly two integral lengthscales (half considered domain), and PLIF interactions residing smaller, local regions (less than half domain) captured no information. Further, visualizing coherent structures within network parameters, role multi-scale related retrieval from became more apparent. Overall, study reveals useful pathway (in form extraction) develop diagnostic tools capture same resources minimizing redundancy.