应用遗传算法结合连续投影算法近红外光谱检测土壤有机质研究。 采集浙江省文城地区农田土壤样品近红外光谱数据, 土壤样品数为394个。 为简化模型, 采用遗传算法结合连续投影算法挑选出18个特征波长建模, 应用偏最小二乘回归建立有机质预测模型, 建模集的决定系数为081, 均方根预测误差为022, 剩余预测偏差为231, 预测集的决定系数为083, 均方根预测误差为020, 剩余预测偏差为245。 研究发现, 遗传算法结合连续投影算法在简化模型同时, 模型的预测评价指标同采用全谱波长建模并没有明显降低。 因此, 遗传算法结合连续投影算法挑选的特征波长可以应用于近红外光谱检测土壤有机质含量。

Visible near infrared spectroscopy combined with genetic algorithm and successive projection algorithm was investigated to detect soil organic matter (OM). A total of 394 soil samples were collected from Wencheng, Zhejiang province. In order to simplify calibration model, a total of 18 characteristic wavelengths were selected with usinggenetic algorithm and successive projections algorithm. These characteristic wavelengths were subjected to partial least squares regression (PLSR) with leave-one-out cross validation to establish calibration model of soil organic matter (OM) with coefficient of determination (R2) of 081, 083, RMSEP of 022, 020 and residual prediction deviation (RPD) of 231, 245 for the calibration set and prediction set respectively. The results showed that using genetic algorithm and successive projections algorithm can simplify the model greatly while the assessing indexes of model such as R2, RMSEP and RPD were not reduced greatly compared with indexes of model using full spectra data to develop calibration model. Therefore, genetic algorithm combined with successive projections algorithm can be used to simply the model to predict soil organic matter.