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

发布时间：2018-11-10 08:43

【摘要】：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。
[Abstract]:Application of genetic algorithm and continuous projection algorithm to detect soil organic matter by near infrared spectroscopy. The near infrared spectrum data of farmland soil samples in Wencheng area Zhejiang Province were collected and the number of soil samples was 394. In order to simplify the model, 18 characteristic wavelengths were selected by genetic algorithm and continuous projection algorithm. The organic matter prediction model was established by partial least square regression. The determination coefficient of the modeling set was 0.81and the root mean square prediction error was 0.22. The residual prediction deviation is 2.31, the decision coefficient of prediction set is 0.83, the root mean square prediction error is 0.20, and the residual prediction deviation is 2.45. It is found that the genetic algorithm combined with continuous projection algorithm can simplify the model, and the prediction and evaluation indexes of the model are not significantly reduced with the full-spectrum wavelength modeling. Therefore, the characteristic wavelengths selected by genetic algorithm and continuous projection algorithm can be used to detect soil organic matter by near infrared spectroscopy.
【作者单位】：浙江大学生物系统工程与食品科学学院;华东交通大学电气工程与自动化学院;
【基金】：国家自然科学基金项目(61134011) 江西省科技支持项目(20161BAB202060,20161BBF60060,20151BAB207009,20142BDH80021)资助
【分类号】：S153.6;O657.33

【相似文献】