HYBRID MLR(MULTIPLE LINEAR REGRESSION)-ANN(ARTIFICIAL NEURAL NETWORK) MODEL FOR PREDICTING HEAT CAPACITY OF IDEAL GAS OF PURE ORGANIC COMPOUNDS

摘要

PURPOSE: A hybrid MRL-ANN model for predicting heat capacity of ideal gas of pure compounds is provided to improve prediction performance.;CONSTITUTION: A molecular descriptor value for heat capacity of ideal gas of hydrocarbon series compounds is prepared. A QSPR(Quantitative Structure-Property Relationship) model for heat capacity of ideal gas is constructed. The prediction performance is tested by experimental data.;COPYRIGHT KIPO 2012