原油冶炼中主精馏塔装置的渐近辨识及分析

Multi-variate Identification of Crude Oil Refining Processes Using ASYM Method

原油冶炼过程的精馏塔装置属于典型的强耦合多变量化工对象,对其进行控制需要准确、合理的数学模型,辨识模型的精确程度与控制效果和经济效益直接相关。出于安全和效率考虑,对原油精馏过程的测试需要在闭环条件下进行,渐近辨识方法在测试过程中能够同时改变多路操作变量并根据频域方法给出模型。本文基于渐近辨识理论,采用先升阶再降阶的解决方案,根据ARMAX模型描述精馏塔装置,取得符合要求的辨识模型。由于新观测数据能够降低过程建模中的不确定性,以输出误差为指标考察了新的测量数据对改善辨识模型精度的影响,分析测试数据信息量与有效估计之间的联系,量化分析新观测数据的引入对降低模型参数不确定性的影响。

The main distillation column of a crude unit at a refinery is a strong-coupled multivariate system. The precision of identifiable models has close relationship with performance of control and ecomonic benefit. Considering the safety and efficiency, test should be performed under closed-loop situations. ASYM method has the potential to change multiple manipulated variables at the same time, providing identified models based on frequency domain methods. This paper uses ARMAX model to describe the main distillation column by ASYM method. As efficient measured data reduce the uncertainty in the process of parameter estimation,this paper quantitatively analyzes accuracy improvement brought by new measured data, using output error as performance index.