中文摘要：

RTK模糊度固定通常采用序贯四舍五入方法，该方法只固定能可靠固定的部分模糊度，因此控制四舍五入的纳伪概率是实现RTK可靠性的关键。从含多个备选假设的假设检验理论出发，研究纳伪概率可控的四舍五入方法，根据风险水平和实数解的质量自适应地确定四舍五入取整区域，从而有效地控制整数固定的纳伪概率。并结合GNSS应用中两种特殊情况，发展两种简单实用的应用策略。采用基线长94．6km的CORS站GPS双频数据进行网络RTK模糊度固定试验，分析本文纳伪概率可控的四舍五入方法的效果。结果表明，纳伪概率可控的四舍五入方法的固定率与传统四舍五入方法的固定率相当，且较传统方法更能有效地控制纳伪概率，提高整数解的可靠性。

英文摘要：

RTK ambiguity resolution is usually preferable to sequential rounding method to partially fix the ambiguities that can be fixed with high reliability. Hence, it is crucial to control the error probability of rounding. The issue of reliable rounding based on the theory of multiple hypothesis testing will be addressed, where the rounding interval is adaptively determined in terms of the user-defined critical level to tightly control the error probability of ambiguity fixing. Furthermore, regarding to the two special scenarios in real GNSS applications, two easy-to-use rounding strategies are developed. The dual-frequency GPS data collected from two CORS stations of 94.6 km baseline is carried out as a network RTK ambiguity resolution instance to evaluate the performance of the proposed new rounding strategies. The results show that, comparing with the ordinary rounding strategies, the new rounding strategies can indeed control error probability of ambiguity fixing whilst keeping the moderate fix-rate