| Thermal power unit dispatch problem is a typical optimization problem. It aims at allocating the electricity load demand between the generating units while considering the limitation of generating units and the security need of the systerm. The main goal is to minimize the operation cost of generators. Solving the problem is of great importance to the economic-safe running of electric power enterprise, energy saving, emission reduction and sustainable development. The thesis firstly introduces some related knowledge of unit optimal dispatch problem, which include how to obtain the coal consumption curve, modeling and analyzing the process of some common algorithms to solve the unit optimal dispatch problem.Differential Evolution(DE) algorithm is an excellent group intelligence optimization algorithm. It has been widely used to solve various optimization problems, it can get satisfactory results when solving complex, large-scale problems. Since the algorithm has some shortcomings as it is easy to be trapped in local minima, besides, search stagnation phenomenon may also happen sometimes. The thesis makes some improvements on the basis of standard DE algorithm, mainly in mutation operation, variable crossover factors and random disturbance. Then the thesis uses the modified DE to solve ten benchmark functions in low and high dimension respectively, results are compared with the results of other classical modified DE algorithms. The comparison results show the effectiveness of the modified algorithm and results, it has much better effect in solving high dimension problem. Then by solving the Sphere function in different parameter settings with the modified algorithm, the thesis analyzes the affect of different parameter settings to the modified algorithm.The thesis uses the modified DE to solve three typical unit dispatch problems, they are thirteen units and forty units with valve-point effect and ten units with valve-point effect and multi-fuels. The thesis uses a new equation constraints handling mechanism when dealing with the power balance constraints. By simulation experiments, it can be concluded that the modified DE has much better effect in solving thermal power unit dispatch problem than other comparison algorithms. |
