Optimization of Unit Commitment based on improved Max-Min Ant System algorithm

Unit Commitment (UC) is one of the key issues which involves both binary programming and real number programming. In this paper, in order to get the minimum coal cost of each period, the Dynamic Programming (DP) is used to obtain the optimal load distribution among qualified UCs, whose output can meet the period load order. The points with minimum coal cost in every period are adopted to form an initial heuristic strategy to discard the unreasonable operating candidate points. After that, Max-Min Ant System (MMAS) is chosen to deal with binary part. By adding the number of operating units into both the heuristic information expression and the pheromone updating expression, times of units´ start-up decrease so that the starting cost reduces too. After once MMAS iteration, a kind of creativity local search method helps ants find the best solution in the neighborhood of themselves´ MMAS solutions. At last, a 10-unit UC problem is used to test both improved and original MMAS. The Simulation shows that improved MMAS guides ants to search the space more effectively and obtain a better result.