Abstract: The large-scale renewable energy penetration to the power system leads to the reduction of the system inertia. It is of great significance to use the synchronous phasor measurement unit (PMU) to estimate the system inertia level online. Considering the spatial distribution characteristics of the power system frequency dynamics, it is feasible to carry out the online inertia estimation in regions. The frequency response regioning is not only related to the power grid topologies and parameters, but also to the system operation points, disturbance types and locations. So, itis difficult for a single fixed regioning scheme to ensure that the regioning results meet various operating conditions of the power system. At the same time, the real-time regioning scenarios also puts forward higher requirements for the rapidity of the inertia parameter identifications. This paper analyzes the principle of the on-line estimation of the regional inertia of the power system, and expounds the requirements of the real-time regioning of the frequency dynamic response. Then, a dynamic real-time frequency partition method is proposed, which uses the dynamic time warping index and the k-medoids algorithm to cluster the measured frequency response curves of nodes in the whole network online. The inertia parameter identification based on the Kalman filter is given to ensure that the inertia parameter calculation meets the real-time requirements. The influence of various errors on the system regioning and inertia parameter identification results is further analyzed. Finally, the effectiveness, the real-time inertia tracking capability and the adaptability to different operation scenarios of the proposed method are verified in the IEEE 39 bus system. The online estimation of the regional inertia of the power system proposed in this paper uses the measured frequency response of the PMU to update the regioning of the power system in real time so as to ensure the consistency of the regional frequency response under various operating points and disturbances, which has the potential of field application.