1. School of Resource Exploration Engineering, Kim Chaek University of Technology, Pyongyang 999093, D. P. R. Korea;
2. College of Construction Engineering, Jilin University, Changchun 130026, China;
3. School of Information Science and Technology, Kim Chaek University of Technology, Pyongyang 999093, D. P. R. Korea

Foam is used widely in underbalanced drilling for oil and gas exploration to improve well perfor-mance. Accurate prediction of the cutting transport and pressure loss in the foam drilling is an important way to prevent stuck pipe, lost circulation and to increase the rate of penetration(ROP).In foam drilling, the cuttings transport quality may be defined in terms of cuttings consistency and downhole pressure loss, which are controlled by many factors. Therefore, it is very difficult to establish the mathematical equation that reflects nonlinear relationship among various factors. The field and experimental measurements of these parameters are time consuming and costly. In this study, the authors suggest a cuttings transport mathematical modeling using BPN (back propagation network), RBFN (radial basis function network) and GRNN (general regression neural network) based on various experiment data of cuttings transport of previous researchers and compared the result with experiment data. Results of this study show that the GRNN has a correlation coefficient of 0.999 62 and an average error of 0.15 in training datasets, and a correlation coefficient of 0.998 81 and an average error of 0.612 in testing datasets, which has higher accuracy and faster training velocity than the BP network or RBFN network. GRNN can be used in many mathematical problems for accurate estimation of cuttings consistency and downhole pressure loss instead of field and experimental measurements for hydraulic design in foam drilling operation.