Distributed generation can reduce power losses and on-peak operating costs, improve voltage profiles, defer or eliminate for system upgrades and mitigate environment pollution for distribution network. However, that needs reasonable planning for the distribution networkAt present, genetic algorithm is the most used algorithm for distribution network planning including distributed generation. The problem of distribution network planning including distributed generation is a multi-objective problem. The problem objectives are combined into a single function and then one single solution is produced in a single run by genetic algorithm. The solution is not guarented to be Pareto optimal. In addition, the capacity of DG is represented by binary encode. That limits the choices of capacity of DG since the capacity of DG can be varied from a large range.The mirco-grid and types of distributed generation are discussed firstly in this paper. After analized distribution network planning and the impacts of distributed generation to distribution network, the mathematic model for distribution network planning including distributed generation is established. And a simple and effective power flow algorithm for distribution network, which is based on forward and backward substitution method, is established as well. The algorithm needs no complex codes of nodes and branches by its new branch selection method.According to the model of distribution network planning, two improved algorithms have been proposed based on NSGA-II, which is real-number encoding but with unstable convergency and easily premature. One is non-dominated sorting chaos genetic algorithm by adding chaos search into NSGA-II. The other is non-dominated sorting differential evolution algorithm by substituting genetic operation by differential operation. The appropriate approach is selected after applying them in one distribution network.
|