Simple Kalman Filtre with 2-D state vector

Thank you very well Geoff. I do it with your helps. Thanks

[Geoff Hayes]

bahadir - what is your question? Because it almost looks like you have posted homework and are asking (or expecting?) someone to do it for you. Please attempt the problem and then post questions concerning what you have tried (include any errors that you have observed).

You seem to be tasked with implementing a Linear Kalman Filter. You have the algorithm and all of the (noise, covariance, etc.) matrices so what is missing?

[bahadir safak]

first step My problem is that: Rt and Qt matrix in algorithm.

Second step:B=0

3th.step: what is the Et(is a 2-D random vector from a multivariate....) what is the &t(is a 2-D random vector from a multivariate....)

Answer 1

[Geoff Hayes]

bahadir - unless you have an algorithm to estimate what R(t) and Q(t) should be at time t, then I would just use the R and Q that have been defined above for all t. This should be sufficient.

Since there is no action, then I suspect B can be zero.

As for Sigma(t) (your E(t)), this is your covariance matrix. Given how it is updated at step three, I suspect that you can initialize it to R...so Sigma(0) = R.

As for your last point, I'm not sure what you mean by &t. Do you mean delta t? If so, it can probably be zero or you can randomly generate it at each iteration using the zero mean and Q for this Gaussian distribution.

[Geoff Hayes]

bahadir's answer moved here

Geoff Hayes thanks you very well.

I am a new student.

R= 2x2 matrix (size=2). İt should be 2x1 matrix because A*x(t-1)+sigma(t); A=2x2, initial X=2x1, but R=2x2. İs it true?

How can I do zero mean for this Gaussian distribution.

Thanks you; Best regads

[Geoff Hayes]

R should be 2x2 since the measurement input is 2x1. From your equation, A*x(t-1)+sigma(t); there isn't any reference to R. See normrnd which describes how you can generate a normal (gaussian) random variable.

[Geoff Hayes]

bahadir's answer moved here

Thank you

ı will use R=[2 1;1 2] and Q=[1 0.2;0.2 1] at all iteration. is it true

Could you write simple kalman matlab code with my parameters only one iteration.

Best Regard.

[Geoff Hayes]

Yes, I could write a simple Kalman filter in Matlab using your parameters for only one iteration. But then that would be doing your homework for you... ;)

[bahadir safak]

You are best person.