Non-feasible gradient projection recurrent neural network for equality constrained optimization

A recurrent neural network for equality constrained optimization problems is proposed, which makes use of a cost gradient projection onto the tangent space of the constraints. The proposed neural network constructs a generically non-feasible trajectory, satisfying the constraints only as t → ∞. Generalized convergence results are given which do not assume convexity of the optimization problems to be solved. Convergence in the usual sense is obtained for convex optimization problems. A circuit realization of the proposed architecture is given to indicate practical implementability of our neural network. Numerical results indicate that the proposed method is efficient and accurate.