Research On Model And Algorithm For Urban Distribution Network Planning With Geographic Information

With the development of city, the cables are used frequently in distribution network planning (DNP) of some areas with high request of load sight, such as the city centre, shopping centre, etc. Compared with DNP of using overhead lines, the usage of cables will bring a series of new problems including the choice of laying modes and the setting of cable branch boxes. Since the distribution lines are built along the streets, the street segmenting points should be added on the map, which may make the electric corridors along the streets uncertained and the electric branch nodes on the feeder random. Therefore, the further research of DNP of using cables with consideration of street geographic information is of important theoretical and practical significance. Two new models and Genetic Algorithm (GA) for them are proposed. The research mainly includes following parts:â‘ For DNP of new urban power supply area with wholly cables, a new model for multi-source planning of distribution network is proposed with the detailed consideration of street geographic information. The proposed model takes the integrated investment as objective function. As regards the network investment, the civil construction and electrical costs of the electric corridors are calculated separately. In the process of configuring the network, sources and electric corridors along the streets can be selected randomly for any load, and any new electric branch nodes on the feeder can also be produced if necessary. In contrast with conventional models, the proposed model is more practical in calculating investment, determining electric corridors and producing new electric branch nodes.â‘¡The model for single-source planning of distribution network is developed considering location and costs of cable branch boxes. The costs of outlet switch cabinets in switching station and cable branch boxes are added to the investment. In the process of configuring or modifying the network, the quantity and location of cable branch boxes, the loads supplied by cable switch boxes, electric corridors along the streets and electric connection can be selected randomly. In contrast with existing models,the new model is more practical and more efficient.â‘¢GA for the new models is proposed. In GA, a multi-information matrix encoding strategy is adopted, which is composed of node, branch, line, and source matrix, in order that the geographic and electric information of network can be easily modified, inquired and counted, the constraints can be effectivly dealt with and the fitness of individuals can be correctly determined. Taking advantage of the feature of the new models, edge recombination crossover strategy and different mutation strategies are adopted to give consideration to diversity and optimality of individuals, and help to realize feasibility of individuals. The mutation strategies include shortest path substitution, load node transfer, loads of cable branch boxes transfer, quantity and location of cable switch cabinet variation and electric corridors along the streets transfer.â‘£Two actual urban medium-voltage distribution network planning systems are simulated and the validity of the proposed models and algorithms are illustrated.