精确在线支持向量回归及其在热工过程在线建模中的应用

热工过程往往具有非线性和不确定性,传统描述热工过程动态数学模型的方法难以建立非线性模型,从而难于精确实施热工过程优化控制。文章提出了一种基于精确在线支持向量回归算法的热工过程自校正辨识方法,并与基于最小资源分配网络的非线性模型进行比较分析。仿真研究结果验证了建模方法的有效性,且所得模型精度高,可直接应用于基于模型的控制算法。

Accurate On-line Support Vector Regression and Its Application in Thermal Process Modeling

Thermal processes generally contain nonlinearity and randomicity, it is difficult to build the nonlinear mod- els by the traditional method, and so the whole optimal control for thermal process is impossible. This paper proposes an auto-tuning identification method of thermal process based on accurate on-line support vector regression （AOS- VR） , and compares with the method based on minimal resource allocation network（ MRAN）. Simulation study results proves the validity of this method,which is distinguished by a higher precision and this method can directly apply to model based control algorithm.