| Summary: | 碩士 === 國防大學中正理工學院 === 兵器系統工程研究所 === 93 === At present, automatic alignment technologies for optical fibers can be divided into the passive and active ones. In the passive alignment, optical connectors are usually used to make coupling. Because of higher connection losses, the performance and efficiency using optical connectors are not very well. On the contrary, the active alignment can offer an accurate way and have lower losses during the connection. Also in the manufacturing process of optical components, active alignment plays an important role to increase the accuracy.
This research first uses the “Blind Search Method” to examine the distribution of the light source, and set as the boundary conditions for the following optimization process. In the beginning of the optimization, the “Nelder-Mead Simplex Method” is used for course alignment instead of using expensive CCD cameras. It also ensures that local minimum traps can be avoided. Then, the “Steepest Descent Method” is applied to get a more accurate alignment. In the “Steepest Descent Method”, the “One Dimensional Search” is used to get the optimum value of the step size instead of setting as a constant value. It ensures that this algorithm can converge fast in limited steps. In summary, combining these two methods can provide a fast and accurate way for the active alignment.
|
|---|
