中文摘要：

针对电厂过热汽温控制中存在大滞后和强非线性的特点,采用最小二乘支持向量机方法建立过热汽温系统模型并给出基于贝叶斯证据框架的LS-SVM的参数选择方法。在第一推断准则选择模型参数,第二推断准则选择模型超参数,第三推断准则选择模型的核参数。仿真结果表明该模型具有灵活的结构,较快的计算速度以及很好的泛化能力。

英文摘要：

Aiming at large time-varying and strongly nonlinear characteristics in controlling of super-heater temperature in plant,the method of LS-SVM is used to model of super-heater temperature and a parameter selecting method on Bayesian evidence framework is proposed for LS-SVM.On the first level of inference,model parameters were selected and on the second level of inference the hyper-parameters were selected.The kernel parameter were selected on the third level of inference.The result of the simulation shows that this model has flexible structure,rapid calculation speed and good generalization ability.