一种新的双线性数据协调方法

A New Method for Bilinear Data Reconciliation

在流程工业数据校正中,若涉及多组分物料平衡或能量平衡时,问题则转换为一类特殊的非线性问题,即双线性数据协调。今针对双线性数据协调传统方法的不足,给出了一种新的方法。首先提出了一种消除不可观测变量的方法,通过消除不可观测变量及部分非冗余变量将协调问题降维,并将问题分解为两个子问题;然后针对分解后的子问题,利用微粒群优化算法(PSO)求解。与传统方法相比,该方法在确保高协调精度的基础上具有较好的协调运算效率,并能处理过程中含有不可观测变量的情况。今对一个实例进行了仿真,仿真结果表明该方法的有效性。

For data reconciliation of the process concerning the multi-component mass or energy balances, it will be a particular nonlinear problem, i.e. bilinear data reconciliation. To overcome the defects of the traditional methods for bilinear data reconciliation, a new method was proposed. Firstly, a new unobservable variable elimination approach was developed to decompose and regularize the data reconciliation problem. Then, the problem was solved by particle swarm optimization algorithm （PSO） after elimination of the constraint equations. Data from simulation of a bilinear steady case were reconciled and the performance of the proposed strategy was compared with that of traditional methods. The comparison results show that the new method may give more accurate estimates, have good computing efficiency and can work well even if some of the process variables are unobservable.