On the Proper Orthogonal Decomposition Technique for PIV Post-Processing
PIV速度场坏矢量的本征正交分解处理技术

A new technique for data post-processing of particle image velocimetry (PIV) based on proper orthogonal decomposition (POD) is introduced in this paper. This technique changes the situation of current data post-processing that separately achieves the identification of spurious vector in velocity field and its correction. Through iterative method, the unification of identification of dead pixels in velocity field and its repairment algorithm is effectively achieved. A reference velocity filed is reconstructed from the lower order mode of POD for identification of spurious vectors. Based on the reference velocity field, the identification and repairment of dead pixels are achieved in the whole field. Through introducing Gaussian distribution of random error for velocity field data of smooth PIV, a test measurement verifies the superiority of the proposed POD method. Results show that for spurious vector identification, the new method presents better performance than that of normalized median test, including the identification of spurious vector in a large area; while for interpolation algorithm of spurious vector repairement, it presents higher calculation efficiency than that of traditional Gappy POD and higher accuracy than that of common vector field interpolation methods. Especially, this method can provide good prediction for original flow field in a large connected and area, where the data are missing.