The modified gain extended Kalman filter and parameter identification in linear systems (Q579211)

The modified gain extended Kalman filter (MGEKF) is applied to the parameter identification of linear systems. First, a globally convergent observer, called the modified gain extended Kalman observer (MGEKO), is developed for a special system composed of both nonlinear dynamics and nonlinear measurements \(x_{i+1}=f_ i(x_ i)\), \(z_ i=h_ i(x_ i)\), in an effort to generalize earlier results of the authors. The gain algorithm of the MGEKO is the same as that of the Kalman filter. By using Lyapunov's second method, the resulting algorithm is shown to be exponentially convergent. Next, a stability analysis of the MGEKF is studied in the probabilistic Hilbert space \(L_ 2\) by introducing an exponentially bounded nominal filter called the intermediate MGEKF. The authors obtain a generalization of earlier results. In the last section two examples of the application of the MGEKF to the parameter identification problem are presented.