Electro-mechanical actuators (EMA) are highly practical in modern multi-electric aircraft and their reliability significantly impacts the safe operation of such aircraft. Therefore, research on the safety and reliability of EMA is crucial. Currently, traditional fault diagnosis methods rely on mathematical models and data-driven approaches, exhibiting some latency in fault prediction. The emerging digital twin technology offers a promising direction to address this challenge. This paper proposes a method to address the real-time tracking issue of key parameters in the digital twin model of EMA. Specifically, a parameter updating estimation method based on the back electromotive force of permanent magnet synchronous motors is introduced. This method utilizes observers to estimate the back electromotive force of permanent magnet synchronous motors and employs gradient descent to fit curves for updating key parameters. The method effectively corrects the parameter of the digital twin model through Simulink simulations.