Genetic Algorithms in Search, Optimization and Machine Learning
Genetic Algorithms in Search, Optimization and Machine Learning
ICIC'10 Proceedings of the Advanced intelligent computing theories and applications, and 6th international conference on Intelligent computing
Hi-index | 0.00 |
This paper presents a new particle pairing algorithm using "Genetic Algorithms" for DPIV (Digital Particle Image Velocimetry), which are searching algorithms for obtaining an optimal solution based on the mechanism of evolution. The particle pairing between two tracer images with a constant time interval is needed to obtain a velocity vector field. Since the algorithm adopts a fitness function which totally evaluates the similarity between respective small particle patterns in the two tracer images over the field, it promises to give a more correct velocity vector distribution than the conventional PTV (Particle Tracking Velocimetry) which identifies each particle based on its local information. In addition, a particle pattern matching for the similarity is performed after correcting fluid rotation. It therefore is robust against a high particle density and an increase in the time interval. The algorithm is applied to the PIV standard images distributed through the Internet (http://www.vsj.or.jp/piv). It gives a correct velocity vector distribution as a result even if a pair of the successive images has a large time interval.