Optimal Planning For Substation Locating And Sizing

With the rapid development of expansion and reform for urban power systems in China, urban power system planning has become an urgent work. In planning the urban power systems, it is an important stage between load forecast and network planning to acquire both the locations and sizes for substations. Its result will affect many aspects directly such as power line routing, network structure, power network investment, operation economy level and power supply reliability.Focusing on the low quality of the solution derived from existing methods, a more efficient optimal planning method for substation locating and sizing on the engineering background is put forward in this thesis, which divides the problem into two sub-problems, i.e. the sources locating one and the combinational optimization one.So far, the common solutions to the problem of sources locating are Alternative Location-Allocation (ALA) method and Genetic Algorithm (GA). The method of ALA has the advantages of simple principle and fast computation; but as a local optimization algorithm, it cannot guarantee the quality of the solution that varies with the choice of the initial location. Although GA provides a way to reach the global optimal solution, however, it suffers from slow convergence—as many futile steps due to the poor local searching ability. Besides, in solving the sources locating problem, crossover operator of ordinary GA will make the children lose some fine characteristics of their parents. Thus the efficiency of GA is rather poor as it is quite uncertain for two excellent individuals to produce a child with an equivalent good quality. To avoid such difficulties, the thesis firstly designs a specialized Partheno-Genetic Algorithm (PGA) that is suitable to tackle the sources locating problem. Secondly, the thesis combines the complementary PGA and ALA to improve the computation efficiency. Hybrid PGA based on ALA method and approximate hybrid PGA based on ALA method are proposed to deal with the sources locating problems of small-scale and large-scale respectively.As the sources locating sub-problem in substation planning is of large-scale engineering attribute, the proposed approximated PGA is effective in such circumstances. In addition, simulation examples show that the combinational optimization sub-problem is non-convex and non-concave, thus the Branch and Bound (B&B) technique is adopted in this paper to obtain or approximate the global optimal solution. To improve the efficiency of the B&B technique, some heuristic principles are proposed to cut those branches that may generate global optimization solution with low probability.Finally, the software package developed based on this method is applied to the high voltage substation planning process for an actual city. The results verify that the proposed algorithm possesses significant engineering practicability with fast speed...