Reconciliation of outliers in CO2-alkanolamine-H2O datasets by robust neural network winsorization

摘要: It is normal to find at least a few measured values in CO2-alkanolamine-H2O datasets that deviate greatly from the majority of published data, as data come different sources. These values, termed outliers, are major source conflict modeling, simulation and process development studies. Therefore, removal outliers mandatory. However, available statistical techniques known lose information boundaries system exhibit substantial deviation holistic trend. Hence, an adaptive approach combining artificial neural networks robust winsorization presented for identification reconciliation system. The proposed flexibly transforms nonlinear distribution predicts corrected (winsorized values), thus maintaining extremes results have been graphically analyzed show good conformance treated with retention winsorized values. method improves shortcomings previous approaches can be potentially extended other experimental chemical systems.