On-line parameters estimation of low scale SPSG using discrete Kalman filters

Abstract

Parametric identification techniques are applied in year's space from in appliances based on electrical machines. Many of these techniques are verging on each other in this field. In fact, certain techniques are more adapted to a recorded time parametric identification and which are so-called 'off-line methods'. The off-line methods are principally based on standstill tests. The other techniques are more suitable to a real time estimation of the parameters and are known as 'on-line methods'. The on-line methods concern mainly the case where the machine is functioning under load conditions but both methods (off-line and on-line) may use an optimization algorithm to minimize the error between the real and the estimated parameters. In This article the study was carried out on a salient-pole synchronous generator (SPSG) of 0.3 kW, at first, experimental parametric identification perform using off-line tests is presented; subsequently different estimators were applied to on-line parametric identification; the discrete Kalman filter (DKF) is the estimator applied in this work; it can be used in its traditional form (DTKF) for linear systems or in its extended form (DEKF) when the system is nonlinear. Another attractive application of the DKF is when it is biased (DBEKF). The consideration of the bias makes it possible to reduce the mean squared error (MSE) between measured and estimated values of the system state variable accordingly the normalized MSE (NMSE) can be minimized. Likewise, standard deviation (STD) between real and estimated values of the parameter can be limited in the tolerable percentage. All this study is discussed and the different DKFs are implemented in Matlab/Simulink code in order to demonstrate the effectiveness of DBEKF estimator compared to other filter. All estimators can be used in on-line estimation in both steady and transient states for low scale generator.