支持向量机回归在线建模及应用

Support vector machines regression on-line modellingand its application

支持向量机 (SVM)回归理论与神经网络等非线性回归理论相比具有许多独特的优点。讨论了建模中 SVM核函数、损失函数的选取和容量控制等问题 ,并用实验加以验证。将 SVM回归动态建模理论应用于非线性、时变、大时延温室环境温度变化的建模和预测 ,模型简单 ,预测效果好。

The support vector machines theory is shown to have excellent performance compared with other non-linear regression, such as neural networks. The problems how to select the kernel function, lossfunctionandcontrolG*2capacity,and so on, are discussed with simulation demonstration. The dynamic SVM regression modelling is applied to the process of greenhouse temperature change which is non-linear, time-varying, dead-time. The model is simplified and the result of prediction is fine.