Hidden Markov model implementation

Hello,

I am trying to implement a Hidden Markov Model for predicting the state of health of battery.

The observation symbols is an array of real numbers and hidden states is an array of integers from 1 to 5.

The data I have is a 2*n matrix consisting of hidden states for n observation symbols.

When I used hmmtrain function in Matlab, the parameters are not learnt as expected.

I set the initial values for emissions, transition and initial state matrix randomly.

I couldn't figure out, what has to be done! Do you have some advice on what might be the problem is?