Aiming at the problem that the ambiguity of carrier-phase is usually more difficult to fix and the positioning result is less effective when using low-cost receiver for real-time kinematic (RTK) positioning, a combined adaptive partial ambiguity resolution (CA-PAR) algorithm is proposed in this paper. The algorithm adaptively adjusts the boundary value of the bootstrapped (BS) success rate based on the decorrelated float state matrix and its covariance matrix, and iteratively uses the criterion of satellite selection to remove the satellites with large measurement noise and residual error, so as to select the optimal subset of the ambiguity parameters. Based on the low-cost receiver and single-frequency GPS/BDS combination model, the experimental results show that fixed rate of ambiguity resolution of the new algorithm has a significant improvement compared with the existing partial ambiguity resolution (PAR) algorithm, static and dynamic positioning can reach 92.5% and 56.7%, respectively. And the horizontal and vertical accuracy are increased by 35%~45% and 80%~ 85% respectively based on static positioning of RTK.