Score Test Method of Real-Time Cycle Slip Detection Considering the Influence of Gross Error

The appearance of gross error in un-differenced observations will cause the false detection of cycle slip, but the existing real-time cycle slip detection algorithm has little research on it, this paper introduces the score test model in hypothesis test, and on this basis, constructs a score test method that both gross error and cycle slip are processed at the same time. Considering the characteristics of less information available and irreversible processing in real-time cycle slip detection, the score test model is optimized by using the prediction residual instead of the fitting residual of the original model. In the detection process, select the appropriate window width, test GF (geometry free) data series by using the idea of sliding and score test, mark the detected abnormal points, and use the abnormal situation of the next point as the judgment index of gross error and cycle slip. The experimental results show that this method can not only effectively separate the gross error and cycle slips in the un-differenced observations, but also identify the small, continuous and special cycle slips in the data.