Model Parametrization-Based Genetic Algorithms Using Velocity Signal and Steady State of the Dynamic Response of a Motor.

The study of dynamic models and their parameterization remains a relevant topic in research. Motors and their models have been extensively analyzed, studied, and parameterized using various techniques due to their broad applicability in motorering and industrial settings. However, most methods for obtaining model parameters require at least two averaged signals from the motor, such as torque, current, speed, position, or acceleration. In this work, we propose the parameterization of a motor's dynamic model using only the speed signal and the steady-state values of the variables. Through evolutionary computation, the mechanical and electrical equations of the motor are reconstructed based on this signal. This approach offers a significant advantage, as it enables parameter estimation without requiring the instrumentation needed for full current signal measurement or, alternatively, torque measurement. To achieve this, the transfer function representing the motor's speed is utilized. The function reconstruction is performed with a Root Mean Square Error (RMSE) of less than 1% for both the speed and current signals. Since the original current signal is not required for this estimation, this work presents an innovative approach to estimating a system of dynamic equations using only a single measured variable and the dynamic relationships of its step-input response.