【摘要】 对于复杂多变的热工对象,难以建立较为准确的动态模型,这是影响动态控制品质的主要因素。对此,提出了运用空间统计学中的克里格估计方法来建立复杂多变的热工对象的自适应动态模型。该方法是一种基于样本数据的非参数回归建模方法,具有多维插值的形式,不需历史数据和模型训练,只需添加实时数据即可建立对象的自适应动态模型。通过对Mackey-Glass混沌时间序列的预测仿真和实际电站锅炉NOx排放的动态预测,表明了该方法在复杂热工对象动态建模中的有效性。更多还原

【Abstract】 For the complicated and variable thermodynamic objects,it is hard to establish comparatively accurate dynamic models,this is a main factor affecting the quality of dynamic control.For this,a Kriging estimation method used in the spacial statistics has been put forward to establish the adaptive dynamic model for complicated and variable thermodynamic objects.This method is a model-establishing method of non-parametric regression based on samples’ data,boasting muti-dimentional interplating form,needing no historic data and training model.An adaptive dynamic model of the object can be established only by adding the real-tiem data.Through prediction and emulation of Mackey-Glass chaos time series and dynamic prediction of NOx emission from actual utility boilers,it has been proven that the said method has effectiveness in establishing dynamic model for the complicated thermodynamic objects.更多还原