Fast contingency screening through optimizing Hopfield neural networks

Contingency ranking attempts to estimate the impact of various contingencies without actually solving the power network. Existing methods of contingency ranking methods suffer from masking effects in approximate methods and slow execution in more accurate ranking methods. In an effort to improve this situation, this paper proposes a fast contingency screening scheme using an Extended Hopfield Neural Network (EHN). This scheme is based on a novel formulation of the contingency ranking problem as a combinatorial optimization problem for efficient solution by the EHN. Because of the recent advances in problem mapping methods, this formulation results in accurate model codification free from trial and error and elimination of local minima problems encountered by previous Hopfield network schemes. The application results of the proposed EHN scheme to three sample systems are presented. The ranking performance is compared with other methods