Abstract: The on-site microseismic monitoring results in large data， which shows the damage and evolution of rock mass because of excavation. On the basis of big data technology， the quantitative relation between the microseismic data and the rock mass damage is sought through the excavation of the effective data， and the dynamic evaluation of the stability of the rock mass and the early warning of the potential instability disaster are carried out at the same time. A local Spark platform which can interact with the cloud is built in this paper， and it is used as microseismic monitoring data analysis system. Based on the local anomaly detection model and fuzzy statistical programming classification model written on Spark platform， the microseismic data of Shirengou iron mine are analyzed from the views of time series and spatial location， respectively. The stability of the rock mass in Shirengou iron ore is evaluated by synthesizing the results of the two models. The results show that the evaluation algorithm of rock mass stability based on Spark and cloud server， can detect that the large faults between 19# and 21# in Shirengou Iron Mine to develop into deep rupture， and the degradation of rock mass properties caused by the complete penetration of the boundary top column near 16#. The above research work is a beneficial attempt of cloud computing and big data platform application in mines， which can provide a reference for mine safety production.

Key words: Source parameters, Local outlier detection, Fuzzy statistics, Stability evaluation, Early warning