A Neural Network Methodology of Quadratic Optimization with Quadratic Equality Constraints.

This paper presents a feedback recurrent neural network for solving the quadratic programming with quadratic equality constraint (QPQEC) problems based on project theory and energy function. In the theoretical aspect, we prove that the proposed neural network has one unique continuous solution trajectory and the equilibrium point of neural network is stable and convergent when the initial point is given. Employing the idea of successive approximation and convergence theorem from [6], the optimal solution of QPQEC problem can be obtained. The simulation result also shows that the proposed feedback recurrent neural network is feasible and efficient. [ABSTRACT FROM AUTHOR]