Abstract: Experimental research and theoretical analysis of MPMD matching algorithm based on the GRPM under- ground positioningHigh-precision positioning service is important for the underground production management,the in- telligent escape and the emergency search. A novel method named GRPM was proposed in this paper,which is the combination of the radio frequency identification and the geomagnetic matching. GRPM is the positioning method with a high precision,autonomous,and suitable for the mine Internet of Things. The MPMD algorithm based on GRPM is a matching model for the optimal estimation of the joint distance of two-dimensional magnetic parameters,which can solve the problem of low accuracy of geomagnetic matching because the spatial distribution of geomagnetic anomaly field is not obvious,slowly changing or similar. Two horizontal roadways named H-11 and H-21 in simulated mine were selected in the study. The suitability of geomagnetic positioning in the H-21 roadway is good and in the H-11 roadway is poor,which were obtained through the statistical analysis of the mean,standard deviation,roughness and information entropy of their geomagnetic data. The geomagnetic matching studies with MSD,MAD,and MPMD algorithms in the H- 11 and H-21 roadways were carried out to compare the positioning accuracy and anti-noise performance of the three al- gorithms under the conditions of geomagnetic variation and noise disturbance. The study results show that the matching probability of the three matching algorithms all reached more than 95% in the roadway H-21 with abundant geomag- netic features,the matching accuracy of MPMD is higher than that of MSD and MAD,its matching error keep within a grid. MPMD algorithm is better than MSD and MAD algorithm in dealing with the problem of fuzzy matching value in H-11 roadway with poor geomagnetic features,and its matching results show few false locations. The algorithm perform- ance was tested in the H-21 roadway,there is a small amount of false location in the MSD and MAD matching under the 200 nT measurement noise,and the matching result of MPMD is nearly error-free. There are a lot of virtual locali- zation in the matching result of MSD and MAD under the 400 nT or 600 nT measurement noise,and MPMD has a few false locations,and the matching probability still reaches 89% . MPMD matching algorithm is an effective and high-pre- cision geomagnetic matching method,which has obvious advantages in accuracy and robustness. However,its matching process is more computational and time-consuming than other algorithms, which needs further optimization in the search strategy.