Genetic Simulated Annealing Algorithm In Computer-aided Alignment Of Optical System

With the speedy development of the remote sensing engineering and the technology of lithophotography, the optical system becomes more complex and the imaging quality's request becomes higher. Such system can be designed perfectly with optical CAD software like Code V, Zemax and so on and the optical elements, especially the aspheric surface can be manufactured well. The key of the higher imaging quality is assuring the accuracy of optical system assembling and alignment. Because of the complexity of the optical system, it's more difficult to satisfy the requirement of imaging quality using the conventional alignment method which is non-visual, non-quantitative, random and long-period. The computer-aided alignment is a new way. The misalignments can be given quantitatively using this method. The alignment process can be directed and the optimal imaging quality will be gotten.The core of computer-aided alignment is to solve the misalignments. The computer-aided alignment has been studied in the 1980s and the least-squares method was used to solve the misalignments. However, it generates the accurate results only in a limited range of linearity round the ideal optical system. This method is suitable for the fine-alignment step.The genetic simulated annealing algorithm is proposed to solve the misalignments in this paper. This algorithm is the combination of genetic algorithm and simulated annealing algorithm, which has the advantages of strong global search and good local convergence.The program of misalignment solving using the genetic simulated annealing algorithm has been completed in VC. The numerical simulation is conducted for an off-axis three-mirror system with Zemax and the local testing is accomplished for Cassegrain system and three-mirror system. The good results are obtained and the accuracy and stability for misalignments solving is demonstrated. Moreover, the main parameters of the genetic simulated annealing algorithm such as populative scale, operator which will influence the misalignment solving are studied.