|Jan9-12, 02:08 AM||#1|
Extended Kalman Filter(EKF) concept (SIMULINK)
I am currently designing a Extended Kalman Filter, estimating temperature in a permanent magnetic synchronize motor, in the Matlab Simulink. Attached pdf is the paper i am referring for my covariance matrix and state vector matrices. I have built the system in Simulink but the results are undesirable. I have some questions which hope can help me in my trouble shooting.
1. The covariance matrix P should it be initialized or just designed in a loop format?
2. The EKF system in SIMULINK should it be digitalised or in analog state?
I have also attached the overview of my EKF design would appreciate any advice and recommendations to improve my EKF system.
Greatly Appreciate any help. Cheers
|Jan13-12, 09:08 AM||#2|
I haven't looked at your attachment yet, but here are comments on your two questions:
1. If you're initializing the state vector, you should initialize the covariance along with (i.e. what is your confidence in your initial state estimate?).
2. Can you be more descriptive? Are you using some special package/toolbox or a third-party subsystem? Not sure what you mean by "digitalised or analog state." If you're asking about integration method, you should use a discrete-time method (Kalman filter is a discrete-time filter, and doesn't use the standard state-space - you probably need to include the time step in your state transition matrix, and additional state derivatives in your state vector compared to a standard x_dot = A*x+B*u system).
Hope this helps.
EDIT2: The paper you attached describes the process of creating a discrete-time model from a continuous model (page 2, eqs. 7 and 8) and the screen shot of your simulink model seems to indicate that you've done this. Still not sure what your second questions means.
|Similar Threads for: Extended Kalman Filter(EKF) concept (SIMULINK)|
|sensor fusion using extended kalman filter||General Engineering||5|
|A question about Extended Kalman Filter||General Math||4|
|The Extended Kalman Filter||Mechanical Engineering||4|
|Understanding the Extended Kalman Filter||Astrophysics||17|