Research On The Optimization Model And Method Of Hour-level Monthly Generation And Purchase Scheduling For Power Systems

Power generation scheduling is an important part of the grid economic optimization dispatch. Thereinto monthly generation and purchase scheduling can macroscopically reflect the economy and security of power system in a long period, so the reasonable planning of monthly generation and purchase scheduling are of great significance to power system economic operation. In the background of the opening regional layered electricity market, the grid company can seek suitable external electricity selling units, which is possible to make security problems and peak shaving problem for the provincial power grid. Therefore, it’s a challenge to coordinate economy and safety of power system while considering the wide-area optimal allocation. However, the traditional methods have been unable to meet the above requirements. So, the fine simulation on mid-and-long term electricity production is an important approach to the problem above. In addition, the optimization model of high dimension is very difficult to resolve. Based on this, this thesis innovatively studies on the optimization model and method of hour-level monthly generation and purchase scheduling, the specific contents are as follows:① On the issues of network security and system peak shaving, the optimization model of hour-level monthly generation and purchase scheduling considering external grid is established. On the basis of generally model, the proposed optimization model combines with the different electricity purchasing mode between internal and external grid, while considering the hour-level power balance. The units’ climb rate constraint of start-stop and movements, the hour-level network security constraint and the upper and lower spinning reserve capacity constraint are taken into consideration in this model. Accordingly the purchase scheduling is decomposed to the hour-level generation scheduling for the feasibility of monthly generation and purchase scheduling.② For the proposed optimization model with characteristics of large-scale and mixed variable is very difficult to resolve, this thesis analyses the size and coupling of the model, and constructs the sub-problems. On this basis, combined the advantages of Lagrange relaxation method, which is easy to decompose the models, and the immune genetic algorithm, which is easy to deal with all kinds of variables and constraints, the hybrid intelligent algorithm containing decomposition-coordination optimization strategy is proposed. In the first layer, all variables are in computing, as the immune genetic algorithm variables. And then the antibody population, containing the feasible solution of original problem, is obtained by heuristic-adjusted strategies. In the second layer, the decomposition-coordination algorithm based on Lagrange relaxation method is used to solve the monthly internal generation scheduling sub-problem by time decoupling method, in the context of identifying the unit commitment and external purchased power in the first layer. Then return the optimization results to the immune genetic algorithm. The proposed method improves the antibody diversity and global convergence by immune genetic algorithm, and the efficiency of local search by Lagrange relaxation method. Furthermore the proposed method uses time decomposition-coordination to reduce model solving difficulty. So the algorithm computation time and precision are improved.③ The proposed optimization model and the hybrid intelligence algorithm with decomposition-coordination optimization strategy of hour-level monthly generation and purchase scheduling are simulated and analyzed on IEEE57 node system, IEEE118 node system and a provincial power grid. The practicability of the model and the effectiveness of the algorithm are demonstrated by the simulation results.