Research On Reactive Power Optimization Strategy For Distribution Network With Wind Power And Small Hydro Group

Due to the development of economy and the increasing load types in distribution networks, higher power quality has been required by more users. As is known, the power quality in distribution network is associated with reactive power that causes the fluctuation of voltage, so it is very important to share reactive power more economically and reasonably. Moreover, the utilization of clean renewable energy seems an ideal solution for reducing pollution and carbon emission. For instance, wind power and small hydro power are penetrating increasingly into distribution networks. However, higher penetration of renewable distributed generation(DG) in distribution network also raises serious problems at the same time, such as poor quality of voltage. Therefore, it has great significance to investigate reactive power optimization in distribution networks in this thesis, when wind turbines(WTs) and small hydropower systems(SHSs) are involved.In a distribution network with WTs and SHSs(especially the runoff hydro power station), it is more difficult to share reactive power optimally, since load demand fluctuates at random and the outputs of DGs using intermittent renewable energies are uncertain. Furthermore, the number of operation of reactive power compensation equipments in a period of time is restricted, which makes real-time optimization harder. Inspired by the idea of static state on different periods, a strategy based on adaptive time division has been proposed, which offers an effective way to share reactive power optimally in the distribution network with the fluctuation of power effectively. In this method, the time of whole optimization cycle is divided into several periods of time adaptively, but how long a period lasts relies on the real-time fluctuations of DGs outputs and load demand, i.e., the value of a DG output or load on the same period is approximate. Further, a method with decision predisposition for time division is presented, when variation of environmental conditions and demand are considered. In this method, both the fluctuation degree and the value of decision predisposition factors have influence on the results of time division. Simulation results show that adaptive time division strategy can not only deal with the fluctuation of power, but also provide an effective way to realize dynamic adjustment for reactive power.At present, multi-objective particle swarm optimization(MOPSO) has been widely applied to solve the problem of reactive power optimization. However, there still exist some disadvantages in present MOPSO. For instance, information sharing among particles is limited, the diversity of solutions is hard to be achieved and premature convergence often occurs. Hence, multi-objective particle swarm optimization based on information sharing strategy(MOPSO-IS) is proposed in this thesis. In information sharing strategy, velocity updating equation is improved based on the information of elite particles, and information sharing among the particles is promoted effectively. In order to improve the convergence accuracy and uniform distribution of solutions, some strategies, such as chaotic mutation and external archiving, are introduced into MOPSO-IS. Moreover, the convergence and the time complexity of MOPSO-IS are analyzed and discussed. Also, MOPSO-IS has been evaluated on ZDT test functions. Compared with similar algorithms, the results show that the convergence accuracy and diversity of Pareto optimal solutions are improved. More importantly, particles can escape from local optimum points, which is an effective way to avoid the aggregation of the population. Besides, the new parameters introduced by chaotic mutation have little influence on the performance of MOPSO-IS. MOPSO-IS provides a more efficient and excellent algorithm to solve multi-objective optimization problems of reactive power.As mentioned above, there are WTs and SHSs in the distribution network, so they are regarded as continuous reactive power sources and play important roles in the optimization. To evaluate the performance of methods and strategies proposed in this thesis, simulations have been carried out on the IEEE 33-node system. The results demonstrate that a set of optimal solutions with good distribution has been obtained, and network losses and voltage deviation are reduced significantly, so that voltages always stay in a normal range. Besides, our algorithm gives priority to DGs rather than reactive power compensation equipments to share reactive power, when proper punishment is introduced, and the compensation amount of reactive power compensation equipments is decreased while network losses and voltage deviation are being reduced. Additionally, the application of adaptive time division strategy reflects the fluctuation of DGs outputs and load demand, and as constraints guarantees the reactive power compensation equipments are not switched frequently.