With the development of science and technology, the super high accurate time comparison techniques with several ten picoseconds or higher accuracy are required in many advanced and basic fields. The atomic clock system in the space station has better performance than that on the ground, but the traditional common-view time comparison method cannot be applied to the space station because there are some limitations. At first, the space station common-view time comparison principle aiming at several ten picoseconds accuracy is analyzed, and the sources of delay larger than 1 picosecond are considered. According to the space station common-view time comparison principle, the visibility of the space station is simulated based on several main geographical cities in China. The analysis results show that the time interval is short for ground station to observe the space station, and the common-view time interval is shorter. A more serious problem is shown that some areas cannot receive the signal send by the space station simultaneously, so the traditional common-view time comparison method is invalid when the ground stations are in these areas. Then the effect of space station orbit error is studied in theory and simulation based on the traditional method. The research result shows that the orbit error cannot be cancelled effectively by the traditional method, and the remnant orbit error is on the order of about several hundred picoseconds. These remnant orbit errors have a direct influence on the time comparison. A new asynchronous common-view time comparison method is proposed, and its principle and advantages are introduced. The geometric expression that describes the position relationship between the space station and two ground stations is proposed to find the observation time when the orbit errors can cancel completely. And the high stability of the space atomic clock and the ground atomic clock are advantaged to model and extrapolate the space-ground clock bias. The geometric position relationship and the modeling and extrapolating of the space-ground clock bias are combined together to solve the problems of time comparison accuracy and common-view blind area, because the optimized method does not require that two ground stations observe the space station simultaneously. Finally, the simulation experiments are done to validate the new method. The experimental result shows that the asynchronous common-view time comparison method is valid to realize the time comparison with the accuracy of several ten picoseconds. And it also shows that the new method is helpful in solving the problem of blind area that exists in traditional space station common-view time comparison method.