Automatization of strain measurement method using picture processing and finite element method

In order to measure the strain distribution under critical conditions, we had developed a strain measurement system based on a grid method using computer picture processing techniques. This method, however, does not operate efficiently when the gray scale distribution is uneven, so that a complementary interactive operation is needed. But an interactive operation is not suitable for the feed-back operation systems or when dealing with a large amount of data. In this study, we have developed an automatic picture processing method to detect grid points (marks) in uneven picture data by utilizing the laplacian operator which responds to the sharp change of gray levels. Strain distribution has been calculated from the displacement distribution approximated by using finite elements and the least square method. But this method has also some problems, so it has been replaced by a new approach based on the displacement constraint of finite element method to improve the accuracy of calculation. For an application of this approach, dynamic strain distribution under an electromagnetic field is measured and compared with the static strain distribution to investigate the dynamic loading effect. (author)