基于支持向量机的微咸水灌溉下土壤盐分预测

土壤水盐运移过程是农田土壤水盐调控的重要基础,是一个极其复杂的物理化学过程。本文在室内微咸水灌溉试验的基础上,引入支持向量机(SVM)模型应用于咸淡水交替灌溉后对土壤溶液EC、pH值的预测研究,并对其预测效果进行了分析。研究表明,支持向量机回归模型能够有效地模拟预测咸淡水交替灌溉下土壤EC和pH值变化规律,模型平均相对误差均小于10%,具有较高的预测精度。本方法为土壤水盐运移研究提供了一条新的思路和途径,具有较大的实用价值。

Application of support vector machine method to prediction of soil salinity

The soil water and salt migration process is one of the most important foundations for water salt regulation in farmland. It is also an extremely complicated physical and chemical process. Based on the ex-periments on saline and fresh water alternate irrigation in laboratory, this study introduced the model of supporting vector machine （SVM） was introduced in to predict soil electrical conductivity （EC） and pH af-ter saline and fresh water alternate irrigation. The results show that support vector machine （SVM） models can predict soil EC and pH values effectively under saline and fresh water alternate irrigation, the average relative error is less than 10%, and the higher forecasting accuracy can be acquired by using SVM model. Therefore,the SVM model is a very useful tool for soil water and salt migration study.