论文部分内容阅读
本文提出了一种基于独立成分分析的模型对象失配程度评估方法。该方法可以应用于对模型预测控制器的控制模型质量进行评估。该方法可以处理两种不同情况的假设:第一种假设情况下设定值可以发生改变,而在第二种假设条件下只有常规运行数据可以利用。对后一种假设条件,模型的实际匹配程度应该和一个基准匹配程度进行比较。因为有比较好的灵活性,本文所提的方法可以很方便的应用于工业过程中对模型对象失配程度进行评估。此外,还对工业应用对模型预测控制性能评估的理解以及该研究领域中的局限和存在的问题进行了介绍。
This paper presents a method for assessing the degree of mismatch of model objects based on independent component analysis. This method can be applied to evaluate the quality of the control model of the model predictive controller. The method can handle two different assumptions: the first one assumes that the setpoint can be changed, and only the second one assumes that normal operating data is available. For the latter assumption, the actual degree of matching of the model should be compared to a benchmark match. Because of its good flexibility, the method proposed in this paper can be easily applied to evaluate the degree of mismatch of model objects in industrial process. In addition, the understanding of the industrial application on the evaluation of the model predictive control performance as well as the limitations and existing problems in the research field are also introduced.