Fast and accurate solution for the SCUC problem in large-scale power systems using adapted binary programming and enhanced dual neural network

• A novel hybrid method based on decomposition of SCUC into QP and BP problems is proposed.
• An adapted binary programming and an enhanced dual neural network model are applied.
• The proposed EDNN is exactly convergent to the global optimal solution of QP.
• An AC power flow procedure is developed for including contingency/security issues.
• It is suited for large-scale systems, providing both accurate and fast solutions.

This paper presents a novel hybrid method for solving the security constrained unit commitment (SCUC) problem. The proposed formulation requires much less computation time in comparison with other methods while assuring the accuracy of the results. Furthermore, the framework provided here allows including an accurate description of warmth-dependent startup costs, valve point effects, multiple fuel costs, forbidden zones of operation, and AC load flow bounds. To solve the nonconvex problem, an adapted binary programming method and enhanced dual neural network model are utilized as optimization tools, and a procedure for AC power flow modeling is developed for including contingency/security issues, as new contributions to earlier studies. Unlike classical SCUC methods, the proposed method allows to simultaneously solve the unit commitment problem and comply with the network limits. In addition to conventional test systems, a real-world large-scale power system with 493 units has been used to fully validate the effectiveness of the novel hybrid method proposed.