整体检验法与鲁棒法结合的数据协调新方法

A New Data Reconciliation Method Based on Global Test and Robust Method Combination

数据协调技术能为工业过程物料平衡、生产统计、生产计划优化等提供准确、一致的数据,而仪表故障等引起的显著误差会污染数据协调的结果。今提出一种整体检验法与鲁棒法结合的数据协调新方法。该方法首先在整体检验法中基于节点残差信息找出无显著误差的"好"流股和候选显著误差流股;然后给出了基于污染正态分布模型的鲁棒数据协调权重矩阵,其中"好"流股的权重为对应方差的倒数,而候选显著误差流股的权重与数据协调值有关;最后根据权重矩阵进行鲁棒数据协调迭代求解。对蒸汽测量网络的仿真结果表明了该方法能够有效降低显著误差对数据协调结果的影响。

Data reconciliation method can provide accurate and consistent data for mass balance,statistics and optimization of production in industrial process.However,the gross errors of process measurements caused by failure of the measurement instruments can contaminate the data reconciliation results.A new method based on combination of global test and robust method was proposed.Firstly,the global test uses nodal residuals to find good streams without gross errors and suspicious streams with gross errors.Then,based on the contaminated Gaussian distribution model,the weight matrix of robust data reconciliation was given.The weight of good stream is the inverse of the according variance and the weight of the suspicious stream depends on the reconciled results.Finally,the weight matrix was used to obtain the iteration reconciled results.The simulation results of a steam metering system verify that the proposed method can decrease the bad effects caused by the gross errors.