Abstract: A combination method which is applicable to real-time cycle slip detection is selected based on the summary and comparison of various existing improved TurboEdit algorithms, and the method is optimized and improved by adding in the threshold models aiming at its defects. The specific optimizations are as follows: Adding the threshold model which is adaptive to the change of observation sequence RMS to the MW combination cycle slip detection based on the moving average filter, which reduces misjudgments and false negatives in the cycle slip detection effectively; Adding weighted threshold model which is changing with sampling rate and satellite elevation angle to the GF Combination cycle slip detection based on the GF difference between adjacent epochs, which improves the reliability of the cycle slip detection significantly when the sampling rate is low. A large amount of data are tested and the results show that, the combination method is effective and feasible. Compared with other improved methods, the method not only improves the detection precision to some extent, but also has great improvement in avoiding misjudgments and false negatives in cycle slip detection.