A refined triple-frequency cycle slip detection and repair algorithm for GPS/BDS undifferenced observables under high ionospheric disturbances is proposed. In this method, three linearindependent optimal observables combinations for GPS/BDS are selected. The residual ionospheric delay estimated from a "calculation-prediction mechanism", namely flexibly determine whether to calculate delay by observables themselves or to predict delay by a feedforward neural network (FNN), is used to compensate for the detection values. Additionally, we devise an adaptive detection threshold based on actual noise level to detect the cycle slip, and adopt the modified least-square decorrelation adjustment (MLAMBDA) to fix integer cycle slip. The performance of the proposed algorithm was tested with observables at 30 s sampling rate in a 2-day geomagnetic storm period. Results showed that the proposed algorithm can detect and repair all kinds of cycle slips as small as one cycle in the case of high ionospheric disturbances. No false repairs are generated despite the occurrence of very few misjudgments., Nijia Qian, Jingxiang Gao, Zengke Li, Fangchao Li and Chen Pan., and Obsahuje bibliografii