Research To Optimal Unit Commitment Problem Of Power System Based On Tabu-Dynamic Programming

Unit Commitment problem is an important part of the electric power enterprise short-term generation dispatching schedule. It is of great significance for realizing energy-saving power generation dispatching and promoting the construction and development of thermal power generation to work out a safe, economic scheduling abort units Commitment and active power distribution within the time required by engineering practice,In allusion to the shortage that the dimension disaster problem of traditional dynamic programming and the poor searching efficiency and dependencies on the initial state of tabu search, explore the complementarity of the two. In order to utilize the advantages that dynamic programming is convenient for modeling to solve the unit commitment problem directly and dimension reduction by division of stages and the tabu search that is less likely to plunging local optimum, try to build a kind of hybrid algorithm combined with dynamic programming and tabu search, simplify the dynamic programming and design the corresponding taboo search rules to avoid repeating visit infeasible paths and woven so as to achieve better optimization performance.In order to narrow the search scope of undetermined variables of units’ states, the paper presents a method of identifying undetermined units’ states. It determines some units’ states before optimization by analyzing the characteristics of the load changes and unit commitment scheme by priorities during peak and dip periods, and thus reduces the scale of discrete variables.Aiming at the problems of the loss of states caused by the simplification strategy, two methods is raised to explore new states, the neighborhood search based on adjusting infeasible solutions and unit replacement by classification, which are beneficial to expanding the search scope and keeping the diversity of state space.The numerical results of test systems of10-unit24-hour and its development, the IEEE-118nodes with36units24hours system show that the tabu-dynamic programming can solve the unit commitment problem with mixed0-1variables effectively and the performance of the hybrid algorithm is superior than the single algorithm; The computing time of the proposed algorithm changes relatively gently along with the growth of the scale of system. It shows the potential of solving large-scale unit commitment problem.