Recursive Least Squares Algorithm for Online Parameter Identification of DC Motor: Theory and Practice

In the automation era, DC motors are widely utilized across various industries due to their adaptability and efficiency. In the downside of reality, accurately modelling and controlling their behavior under diverse operational conditions remains challenging. Hence, this study proposes a comprehensive approach integrating Recursive Least Squares (RLS) for parameter estimation within ARX models, combined with experiment data to enhance model accuracy. Moreover, State Feedback Control is applied to optimize DC motor operation. Specifically, the model coefficients are predicted over time, alongside recording the error between the actual output and predicted output within a 30 -second timeframe. The results indicate that the error is controlled and converges to 0 within 0.5-second (approximately 10 cycles). As a result, in the online model updates, system error increases and deviates from zero over time due to model drift, likely caused by environmental factors or changes in the motor's temperature. In contrast, the error remains stabilized around zero with online model updates, demonstrating improved control effectiveness through continuous model adaptation.