If I am following the process of identifying State-Space Models with Separate Process and Measurement Noise Description outlined in the link below:
The matrices A and C are known, there is no input into the system so the B and D matrices are going to be all zeros, and the system output data is known. How do I solve for the partially known covariance matrices R1, R2, and R12 if I also have to make an initial guess for the parameterization vector, theta?
Best Answer