Terrestrial heat flow has a high reference value for the evaluation of geothermal resources since it can directly indicate the Earth's internal heat on the surface.However,no high-quality and high-resolution terrestrial heat flow measurements have been conducted in the Songliao Basin due to costly and difficult conventional heat flow measurements based on the drilling technology.Machine learning,as a technology for data analysis,can identify patterns in data and utilize these patterns to automatically calculate unknown data.This study calculated the regional terrestrial heat flow using the machine learning method.Based on the measured data of global terrestrial heat flow and the geological structure data,both the Kriging regression algorithm and the machine learning algorithm were used to calculate the terrestrial heat flow in a known heat flow distribution area,as well as the root mean square error and the correlation coefficient.The machine learning algorithm yielded results with a smaller error and a higher correlation.Then,the terrestrial heat flow in the Songliao Basin was calculated using the machine learning method.As revealed by the calculation results,the terrestrial heat flow is the highest (more than 80 mWm^-2) in the Songliao basin and gradually decreases outward in a circular pattern centered on the area between Daqing and Songyuan.The results are highly consistent with the measured results of the regional geothermal gradient,providing a reference for further analysis of the distribution patterns of geothermal resources in the Songliao Basin.Finally,the sensitivity of geological characteristics was analyzed using the Sobol method to quantify the effects of various parameters.This study verifies that the machine learning method has a high research and application value in the calculation of terrestrial heat flow.