Gappy Proper Orthogonal Decomposition for Flow Data Reconstruction

The proper orthogonal decomposition(POD) is a data-driven feature information extraction technology of flow, which can give the mode of the flow field by the amount of energy, and can obtain an approximate description of the high order data by lower order mode superposition. This method combined with a set of linear equations will form the Gappy POD method, which can achieve reconstruction of the missing flow. This paper studied the application of the Gappy POD method in filling the missing flow field data, developing from two cases where the snapshots are known and the snapshots are not completely known. Firstly, for the case where the snapshots are known, the reconstruction ability under any parameter value within the range of snapshots parameters was studied. Secondly, for the case that the snapshots are not completely known, the influence of modal order, Gappy rate and snapshot number on the accuracy of missing flow field reconstruction were studied. The results show that the Gappy POD method can efficiently reproduce any complete flow field data within the parameter range. However, for snapshots with high Gappy rates, it is necessary to increase the number of snapshots appropriately to improve the reconstruction accuracy.