基于TBM的工业过程显著误差检测方法研究

基于可传递信度模型(TBM),提出了综合多种传统显著误差检测算法的新方法.新方法基于多种传统显著误差检测算法,提出了显著误差辨识框架子集上可信度分配策略,并进一步在此基础上利用可传递信度模型对各子集上的可信度分配进行合成,从而得到对各个测量数据的诊断决策.新方法中使用了迭代补偿的策略.实例考核结果证明了新方法的有效性.

Study on Gross Errors Detection Based on the Transferable Belief Model in Process Industries

A new method was proposed to detect gross errors by combining several traditional algorithms based on the transferable belief model(TBM).The method explained a strategy to assign the belief of different sets of a collection of measurements(the frame of discernment)based on the results of different traditional algorithms.The TBM was used as a reasoning model.Serial compensation strategy was also applied in the proposed method.Simulation results show the new method possesses high performance to identify gross errors.