NEURAL-NETWORK MODELING OF HEAT TRANSFER OF BENZENE IN AN ELECTRIC FIELD

摘要

A neural-network model allowing for extraction of new knowledge from experimental data is developed on the basis of studies using modern computer technology. Prediction of an output parameter (relative change in coefficient of thermal conductivity) with a relative error of 4% on a network previously trained by means of a knowledge base is demonstrated. Some features and laws of the heat transfer of benzene in an electric field are established.