Two-phase optimization approach to unit commitment problems

A two-phase optimization method(LR-DE) was presented for power system unit commitment(UC),a high dimensional,non-convex,nonlinear problem.First the problem was decoupled by Lagrangian Relaxation algorithm,the optimization of multi-machine was changed into double-counting of single optimization to simply the model,using dynamic programming method and sub-gradient method to derive the dual solution Lagrange multiplier;Secondly,the space of updating Lagrange multipliers was determined by optimal dual solution,searched by Differential Evolution Algorithm(DE) with all constraints considered,the duality gap will be narrowed continually and the optimal unit commitment will be obtained.Analysis of examples shows that the algorithm can get better solutions,has the comprehensive ability to search,which is very prospective for large-scale unit commitment problem.