Автоматизований Розрахунок Ширококутних Об’єктивів
Year: 2016
Type: article
Abstract: One of tasks of modern computational optics is the problem of complete automatization of an optical system (OS) design procedure according to different criteria. The widespread local optimization techniques have a significant drawback: initial OSs with good quality are typically required. To solve the problem of finding the lowest value of the evaluation (merit) function in given space of allowed parameters, modern global optimization (GO) methods can be applied. For purpose of automated lens design, different authors have proposed to use such GO techniques as simulation annealing, genetic algorithms, etc. In this paper, the possibilities of one adaptive method (namely, differential evolution Cauchy, proposed in 2013) are numerically studied. An algorithm of the method is characterized by presence of an internal mechanism for self-adapting the coefficients of a classical differential evolution algorithm. Besides, it incorporates the Cauchy distribution to generate new "random" values of above-mentioned coefficients. The algorithm has been implemented in own computer program, developed for automated lens design. An experimental verification of the method’s effectiveness has been done during the design of several OS systems, similar to those, presented recently in US patents.
Authors: Сокуренко, Вячеслав Михайлович, Буйлов, Іван Сергійович
Language: Ukrainian
Subfield: Biomedical Engineering
Field: Engineering
Domain: Physical Sciences
Open Access status: green
OpenAlex ID: W7112540066