基于支持向量机优化粒子群算法的作物生育期ET_0预测

Prediction of ET_0 Based on Particle Swarm Optimization and Support Vector Regression

ET0是计算作物需水量、进行农田灌溉管理及区域水资源优化配置的重要依据。为了提高ET0的预测精度,将粒子群(particle swarm optimization,PSO)算法引入到ET0预测中,并用支持向量回归机(support vector machine,SVM)优化参数。PSO-SVM将最高气温、最低气温、相对湿度、平均风速与日照时数输入到SVM中学习,将SVM参数作为PSO中的粒子,把ET0值作为PSO的目标函数,然后通过粒子之间相互协作得到SVM最优参数,对ET0进行预测,并采用PM模型计算值验证。该文以新疆喀什地区为例,通过采用粒子群耦合支持向量机(PSO-SVM)算法训练得到模型,并用10组数据进行预测;最后引用BP神经网络算法和PSO-SVM算法进行了对比,其结果表明,PSO-SVM算法预测准确率较高,预测值与实测值间相关系数达0.682,平均相对误差为3.19%。

ET0 is an important basis for computing crop water requirement ,optimal allocation of regional water resources. In order to improve the prediction accuracy of ET0,the particle swarm optimization algorithm was introduced to the prediction ET 0,and using the support vector regression to optimize the parameters. PSO-SVM took the maximum temperature ,minimum temperature ,relative humidity ,wind and sunshine duration into SVM learning,the SVM parameter as the particles in the PSO,the ET0 value as the target function of PSO,and then through the mutual cooperation of SVM particles obtained optimal parameters to forecast ,ET0 and PM model was used to calculate the value of verification.Taking Kashi area as an example , this paper got model by algorithm training with PSO-SVM;and finally,compared the results with PSO-SVM and BP neural network by predicting 10 groups samples.The results showed that PSO-SVM ,which showed that the higher correlation coefficient between the predicted and measured values （0.682）and lower average relative error rate（3.19%）,which were better than BP neural network.