Abstract: In order to explore a new method to predict the acentric factor of narrow fractions from direct coal liquefaction oil( DCLO),artificial neural network and group bond contribution coupled model( ANN-GBC) were established. The coupled model used 45 group-bonds and atmospheric boiling point( Tb) of DCLO as input parameters,the relevance between acentric factor and molecular structure of 15 coal liquefaction narrow fractions was investigated.By calculating the acentric factors of 20 model compounds,the ANN-GBC model presents good simulation calculation function,and the average relative error between the calculated value and the theoretical value is less than2. 5%.These comparative data show that acentric factor increases with the increasing of the distillation temperature.The predicted value of ANN-GBC model is higher than that from Watanasiri and NEDOL.In terms of <380 ℃ fractions,ω is less than 1,and the deviation is relatively small,nevertheless,the deviation of > 380 ℃ fractions is larger. The > 420 ℃ fraction can be qualitatively and quantitatively analyzed,because only 20% substances is derived from coal liquefaction narrow fractions. In addition,the actual differences in specific substances could induce a larger deviation,the deviation is very large.