基于太赫兹反射时域光谱技术, 分别利用迭代算法和遗传算法提取葡萄糖多晶太赫兹频段的两个光学参数, 即折射率和吸收系数, 并将遗传算法与迭代算法进行对比分析。 研究结果表明: 当将利用弱吸收近似条件所获得的光学参数作为迭代算法的初始值时, 采用迭代算法可以提高计算效率。 然而, 迭代算法对初始值敏感, 当初始值偏离实际值较大时, 迭代算法不能保证获得较高精度的光学参数； 遗传算法通过参数编码、 初始种群、 遗传操纵、 参数控制和条件约束等设计, 保证了算法的收敛性和种群多样性。 而且, 遗传算法的最优解并不依赖于初始种群。 与迭代算法相比, 采用遗传算法所获得的物质的光学参数拥有更高的精度。 因此, 在基于THz光谱技术提取物质的光学参数时, 本文建议采用智能优化的算法, 这样可以获得高精确度的光学参数。

Based on the terahertz time-domain reflection spectroscopy, the optical parameters of anhydrous D-glucose polycrystalline, i.e. the refractive index and the absorption coefficient were extracted by using iterative and genetic algorithm, respectively. After comparing and analyzing the two algorithms we had drawn the following conclusions: first, the calculation efficiency of iterative algorithm was improved using the solution of weak absorption approximation as initial values. However, the iterative algorithm was sensitive to the initial values. When the big difference between the initial values and real values existed, the accuracy of optical constants would be affected; Secondly, the genetic algorithm was not insensitive to the initial populations. It ensured the convergence of the algorithm and the population diversity through the design of parameter coding, initial population, genetic manipulation, parameter control and constraint condition. Last, compared with the iterative algorithm, the optical parameters obtained by the genetic algorithm had higher accuracy. Therefore, we suggest that the optical parameters of materials with higher accuracy based on the THz spectroscopy can be obtained by using an intelligent optimization algorithm.