基于广义似然比法的化工非线性动态过程过失误差侦破

Gross Errors Detection for Nonlinear Dynamic Chemical Process Based on Generalized Likelihood Ratios

广义似然比法(GLR)是一种有效适用于线性稳态化工过程的过失误差侦破方法。通过将动态化工数据协调模型中的微分约束和代数约束转化为矩阵形式和非线性约束线性化方法,成功将GLR应用到连续搅拌釜(CSTR)非线性动态系统中,同时计算了GLR在该系统中的过失误差侦破性能。统计结果表明,GLR的过失误差侦破率与过失误差大小和窗口长度有关:侦破率随过失误差增大而增大,随窗口长度增大而增大。

Generalized likelihood ratios（GLR） is an effective gross errors detection method for linear steady data reconciliation.In the paper,the differential constraints and algebraic constraints of dynamic data reconciliation model were transformed into the form of matrix,and the nonlinear constraints were linearized.Based on the two methods,GLR was successfully applied to a continuous stirred tank reactor（CSTR） system.The performance of gross errors detection of GLR in the nonlinear dynamic system was also calculated.Statistic results show that gross error detection rate relates to the size of gross error and the length of moving window.With the increase of gross error,the detection rate is improved;with the increase of length of moving window,the detection rate is also improved.