基于支持向量机的结霜过程特性参数预测模型

为解决结霜过程中有明显的非线性和时变性特征及测试数据中受噪声干扰较大、特性参数预测效果较差的难题,引入了基于结构风险最小化的支持向量机方法,建立了以热力参数集、时间、空间等为特征向量的预测模型.应用实验数据对模型进行了验证和评估,并与基于最小二乘法的预测模型进行了对比分析.结果表明,基于支持向量机的预测模型具有良好预测性能、非线性逼近能力和抗噪声干扰能力.

A Frost Growth Characteristics Predication Model Based on Support Vector Machine

Frost growth was a strong nonlinear and time-varying process,and the experimental data was(usually) interfered by noise;so it was hardly to predict its characteristics precisely.In order to solve this problem,a learning method support vector machine(SVM) which is based on the structure risk minimization principle was introduced;then a set of thermal parameters,time and space coordinates were selected;finally the predication models of frost growth characteristics were presented based on those characteristic vectors.The predicated results were verified by the experimental data,which shows a good agreement.Comparing SVM predication model with LS predication model shows that the SVM model has better ability to deal with nonlinear and noise interference problems.