Abstract: This paper designed a composite framework observer combining a general sliding mode observer based on a hybrid nonsingular terminal sliding surface and a Luenberger observer for parameter identification of surface-mounted permanent magnet synchronous motor (PMSM). Firstly, a composite observer framework, consisting of an Extended Sliding Mode Observer (ESMO) and a Luenberger observer, was constructed to enable real-time mechanical parameter identification of the PMSM. Secondly, based on an extended state equation formed by the mechanical parameter error signals used in the ESMO design, an algorithm for mechanical parameter identification was developed. This algorithm endowed the observer with filtering capabilities, thereby avoiding the adverse effects caused by introducing additional filters.After obtaining two mechanical parameters using the ESMO, the load torque was identified. However, the identified parameters still exhibited reliability uncertainty and coupling issues among different parameters. To address this, a new Luenberger observer was introduced into the composite observer framework to achieve decoupling and independent identification of parameters. Finally, simulation experiments were conducted on the MATLAB/Simulink platform to verify the accuracy and feasibility of the proposed method.