近红外光谱(NIRS)是一种间接分析技术,其应用需建立相应的校正模型。为了提高模型的解释能力、预测准确度和建模效率,需要对NIRS 进行波长选择,优选最小化冗余信息。选用蚁群优化(ACO)、遗传优化(GA)、粒子群优化(PSO)、随机青蛙(RF)和模拟退火(SA)5 种智能优化算法对烟叶总氮和烟碱近红外光谱数据进行特征波长选择,结合偏最小二乘(PLS)算法,构建了多个烟叶总氮和烟碱的校正模型,(Near infrared spectroscopy (NIRS) is an indirect analytical technique, and its application needs to establish corresponding correction models. In order to improve the explanatory power of the model Power, prediction accuracy and modeling efficiency require wavelength selection of NIRS to optimize and minimize redundant information. Five intelligent optimization algorithms for smoke control: group optimization (ACO), genetic optimization (GA), particle swarm optimization (PSO), random frog (RF) and simulated annealing (SA)Leaf total nitrogen and nicotine near infrared spectroscopy data were selected as characteristic wavelengths. Combined with partial least squares (PLS) algorithm, several tobacco leaf total nitrogen and nicotine were constructed.)