The large-scale integration of wind turbine generators has led to a significant decrease in power system inertia, posing threats to system frequency security and stability. To address the challenges of optimal frequency measurement point selection and high estimation error in regional inertia identification of high wind power penetration systems, a regional inertia identification method based on a combined model is proposed. First, the K-shape clustering algorithm based on the shape-based distance (SBD) index is adopted to cluster frequency response curves, and the optimal path of frequency dynamic responses within each region is determined. Second, the least squares support vector machine (LSSVM) and the autoregressive moving average with exogenous input (ARMAX) model are combined and used to identify regional inertia levels under different wind power penetration rates. The results are compared with those obtained using the traditional ARMAX-based inertia identification method. Finally, the effectiveness of the proposed method is verified through simulations using the modified IEEE 10-generator 39-bus system. The results show that the proposed method effectively improves the identification accuracy of both regional and system-level inertia.