A New Method To Solve Unit Commitment Problems In Power Systems Based On Generalized Benders Decomposition

With the growing depletion of the traditional fossil energy, human’s needs for resource are growing every passing day. The question how to reduce energy consumption and pollutant emissions are the common goals for all the people. Under this background, an in-depth practical work has been carried out to solve the unit optimal commitment (Short for UC) problem which influence the operation and dispatch of power system, which has important theoretic significance and practical value.Based on the optimal theory, according to the generalized Benders decomposition (GBD) which is pop research methods of mathematic programming in recent years, this thesis focuses on the studies for the Unit Commitment (UC), which means a detailed and in-depth research has been carried out.GBD is a very effective algorithm to solve mixed integer programming which is obtained through nonlinear duality theory. According to the different type of variables, GBD decomposes the problem into a master problem (MP) and a subproblem (SP), MP and SP are solved in turn, the SP utilizes the solution of the MP to form the Benders cuts and returns the cuts to the MP with the purpose of updating the feasible region of MP, until all the conditions of constraints are met so that the optimum solution is obtained eventually.The UC is a mixed integer nonlinear programming problem in which the number of discrete variable amounted to50%of the total number of the variables, the characteristics of the time variation in startup cost, and the characteristics of the time period coupled in unit ramp rate constraints, all of these add some kind of difficulties to solve the problem. In this thesis, a new method based on GBD was presented to solve the UC problem, named as GBD-UC. In this proposed method, the UC problem is decomposed into a master problem and a sub problem which possess clear physical meaning, it has modified the object function and constraints both on MP and SP in order to speed up the convergence of algorithm and improve the results. The proper on/off states of the generating units are found by solving the master problem. In the subproblem, the generating output of the accepted units in the master problem is acquired. The proposed cuts have explicit meaning which contacts the two problems, and improve the speed of convergence of the algorithm to a feasible solution efficiently. The proposed method uses decomposition algorithm to reduce the computational scale and complexity, and takes advantage of the commercial optimum software to solve the discrete variables, the continuous variables make the best of the advantages of the interior pointed method, for example with good astringency and high precision, thus utilizing the sophisticated mathematics optimization method is used to accelerate the speed of calculation. The simulation results for10-200units systems, TEST-6system, IEEE-30system for24intervals demonstrate the effectiveness and correctness of the proposed method, algorithm has good performances such as fast convergence and computationally efficient, and is suitable for the application of large-scale problem, and shows a prospective prospect.