The Research Of Load Transfer In Distribution Network

With the development of distribution automation technology and the increasmentof onsite implementation, fast load transfer technology has attracted widespreadattention of power companies and researchers. Load transfer for distribution networkis to recover the load out of power without any violation as fast as possible after thefailures in network have been isolated. This target is reached by switching operationsand shedding load. As it can significantly reduce the losses caused by the failures andimprove system reliability, load transfer has become one of the most importantfunctions in distribution automation system.However, current research can not yet meet the requirements of the practicalapplication fully. The main methods for load transfer in other literatures can bedivided into two parts: heuristic search method and random optimization method. Theadvantage of heuristic search method is that the calculation speed is fast. Though thismethod is very fast, it is only a feasible solution, not an optimal one. The stochasticoptimization method has a better optimizing ability, but it costs much more time incalculation. And for larger systems, sometimes no solution is available. The problemof load transfer was studied and some results have been achieved in this thesis. Themain work and results are as follows:Firstly, a new adaptive immune algorithm with infeasible solution recovery andtopology radial property fast determination method based on basic-tree to solve theproblem of load transfer when a fault occurs in the distribution system is proposed.Considering all the characteristics of load transfer, a complete mathematical model isproposed. Due to the infeasible solution recovery technology, this method has anadvantage over stochastic optimization methods in calculation time. Furthermore, toaccelerate the calculation speed, the fast determination method of the topology radialproperty based on basic-tree is further introduced to judge the topological radialconstraint limits in distribution system for each new solution during the iterationprocess; it reduces the calculation time for the evolution optimization processcompared with the traverse method.Secondly, a new method that combines the heuristic search method with PSOmethod is proposed. In the optimization process, a new coding method is adopted. The randomly generated individuals can then satisfy the topological constraints, thatis, it is no need to judge the topological radial constraint limit. Compared with thetraditional heuristic method, our method considers optimization process. And it costsless time in contrast to the traditional stochastic algorithm. So it can provide a betterreference measure for operator.At last, to overcome the difficulty in processing the discrete variables and smallconvergence region of the deterministic methods, a new method, complementaryconstraint based on smooth Newton method, has been proposed for load transfer indistribution system with discrete variables. The mathematical model withcomplementary constraint for load transfer has been built up. This model is solved byNewton’s method and the calculation speed is very fast.