The parameters of the induction machine per-phase equivalent circuit (T-model) have been shown not to be uniquely identifiable from online external measurements of steady-state speed, stator currents and voltages, due to T-model redundancy. This paper presents a new model-based online parameter identification technique which overcomes the non-identifiability problem associated with the T-model, eliminating the possibility of obtaining incorrect parameter sets that satisfy the model solution. The proposed method uses a combination of the steady-state induction machine equivalent circuit (T-model) and the inverse Γ-model, which has been shown to be uniquely identifiable from external measurements, in conjunction with an optimization algorithm to estimate the induction machine T-model equivalent circuit parameters. The structural identifiability of the proposed combined technique is analytically confirmed using a transfer function analysis. The performance of the proposed parameter identification method is experimentally demonstrated using a 400 V, 1.1 kW, inverter-fed, three-phase induction motor.