Wiener Model Structure Estimation of DC Motor Through Online Neural Network System Identification
Mohamed Najib Ribuan, Dirman Hanafi, Ayad Mahmood Kwad, Hisyam Abdurman, Sepannur Bandri, Sabir Meftah · 2024
It is essential to have an informative mathematical model of the system to properly analyze its behavior and design a good controller for the system. The system identification method is the best technique for deriving a mathematical model of a system. This paper uses online system identification to derive the mathematical model of a bidirectional DC motor. The neural network is applied as an estimator to identify the system. To adopt the properties of systems in the real world, where all systems have nonlinear properties, in this paper, the candidate model for DC motors in system identification is assumed to have a nonlinear Weiner model structure and constructed into neural networks. Two types of Wiener model structures were developed: Parallel Weiner Neural Network (PWNN) and Series Parallel Weiner Neural Network (SPWNN). Based on experimental results, the PWNN model provides a goodness of fit of 93.70%, and on the other hand, the SPWNN model only provides a goodness of fit of 89.48%. Therefore, it can be said that the PWNN model is the most suitable model structure for a bidirectional DC motor.