摘要：Aimed at biggish prediction error of the overall heat transfer coefficient caused by running parameters fluctuations, analyze the correlations between various elements including outlet oil temperature, oil physical properties, flow rate, environmental temperature, wax deposition rate as well as running time after pigging and the overall heat transfer coefficient. And four dimensionless number determining the laws of overall heat transfer coefficient was put forward. Simplified model came in the use of collinearity diagnosis theory and dimensionless experimental analysis, followed by determining the optimal functional form of prediction model and physical meaning for each dimensionless numbers, then establish a new prediction model. Prediction of the overall heat transfer coefficient for several pipelines came after applying Ten-fold Cross Validation to solve the model. The results showed that, compared with back calculation, the error of overall heat transfer coefficient reduces by 3.62%, and average relative error of end-point temperature lowers by 2.88%. The model is of good steadiness and generalization, and better characterization on heat transfer mechanism of overall heat transfer coefficient, to some degree, solving the biggish prediction error led by running parameters fluctuations..
