Neural net approach for constrained state-space realization

Using the neural network approach for determination of the constrained state-space realization from Markov parameters of the transfer function is proposed. The neural network approach is suggested for determining realization A, B, and C in such a manner that there are some constraints on some of the elements of A, B, and C. Such constraint cases cannot be achieved using conventional algorithms. A single-layer neural network and heuristic random optimization algorithm are used for constrained state-space realization