Accurate inverter impedance information is essential for analyzing and solving issues related to the stability and power quality of new energy grid integration. To address the high cost and safety risks of injected impedance measurement, where disturbances are intentionally injected into the system, as well as the low accuracy of non-injected impedance estimation, a coordinated injected and non-injected inverter impedance estimation method is proposed. First, an inverter parameter estimation model is constructed, using historical impedance measurement data as input and minimizing the deviation between the estimated and measured impedance values as the objective. Considering constraints from the filtering circuitry and control loops, optimal inverter parameters are obtained. Second, a dynamic impedance clustering method is proposed, which uses the solved inverter parameters as features to classify and aggregate impedance measurement data, while the clustering results characterize inverter operating sub-conditions. Then, an inverter impedance estimation method is proposed to construct the feature matrix of sub-conditions based on historical impedance measurements, and real-time power data acquired from non-injected measurements are matched with elements of the feature matrix to estimate inverter impedance in real time. Finally, simulation results verify that the proposed impedance estimation method has strong anti-interference ability, achieves higher accuracy than either injected-only or non-injected-only methods, imposes less impact on the system, and holds practical engineering value.