关键词: 多分辨奇异值分解（MRSVD）, 敏感分量, 单频周跳, 探测与修复, 北斗, 最小二乘支持向量机（LS-SVM）

Abstract: It is critical for repairing cycle slips by accurately detecting epochs of small cycle slips, while it is difficult for high accuracy Chinese BeiDou Navigation Satellite System（BDS） to precisely repair cycle slips. Thus, a sensitive component prediction model based on Multi Resolution Singular Value Decomposition（MRSVD） is proposed to detect and repair cycle slips. In using this method, Hankle matrices are constructed for detected signals, which are analyzed via MRSVD to obtain a group of component signals. Subsequently, the component signals are evaluated and analyzed from the perspective of sensitive factors to select components with characteristics of cycle slips and thereby compose sensitive feature vectors to highlight information about cycle slips of signals, in order to precisely detect singular points of signals through maximum modulus and accurately detect small cycle slips. At last, sensitive feature vectors are used as training samples to construct a predictive LS-SVM（Least Square Support Vector Machine） model for repairing cycle slips. Experimental results suggest that the method proposed in this paper can accurately detect and repair small cycle slips appearing in carrier phases, which prove that this method is feasible and effective.

Key words: Multi Resolution Singular Value Decomposition（MRSVD）, sensitive vector, single frequency cycle slip, detection and repair, BeiDou, Least Square Support Vector Machine（LS-SVM）