Improved speed and load torque estimations with adaptive fading extended Kalman filter

Background Extended Kalman filter (EKF) is one of the most preferred observers for state and parameter estimation of induction motor. To achieve optimal estimations, EKFs require a stochastic system with complete dynamic or measurement equation. However, those equations are partially known in practice and may vary depending on operating conditions, leading to a degradation in the estimation performance of conventional EKFs (CEKFs). Aim To overcome this drawback, this paper proposes an adaptive fading EKF (AFEKF) observer that can compensate for the effect of the incomplete dynamic equation for the estimations of stator currents, rotor fluxes, rotor mechanical speed, and load torque. Materials & Methods To show the superiority of AFEKF, its estimation performance is compared to that of CEKF in both simulations and real-time experiments. Both observers have been implemented through the S-Function block in Matlab/Simulink so that the same observer blocks can be used in both simulation and experimental studies. For real-time implementations, a DS1104 controller board is used. In addition, the computational burdens of both CEKF and AFEKF are compared with real-time experiments. Results and Discussion The simulation and experimental studies indicate that the forgetting factor in AFKEF clearly improves the estimation performance of CEKF, especially in transient states. It also prevents the observer from diverging. Considering its advantages, the additional computational load that causes an increase in the computational load of about 4% can be ignored. Conclusion The proposed AFEKF observer significantly improves the estimation performance and compensates for the effect of dynamic model inaccuracies. Its superiority has been validated by simulation and experimental studies. Finally, an observer with a better estimation performance has been proposed with a slight increase in computational load.