| Distributed generators(DG)have been widely used in power systems as an effective complement of traditional centralized power generation.Microgrid,acting as an effective scheme for integrating distributed generators,has many advantages,such as environmental friendliness,flexibility and rapid investment return.Microgrids and large-scale power systems can support each other to reduce energy consumption,and improve power supply reliability and electric power quality.Nevertheless,integration of large amount of DGs and energy storages to microgrids and the usage of advanced telecommunication and control techniques make microgrid different from conventional distribution systems in network structure,operation control,and fault isolation strategy.In addition,intermittency and randomness of DG outputs,sequential characteristics of energy storage SOC(state of charge)also lead to problems in planning and quantitative performance evaluation of microgrids.It is,therefore,of great theoretical and practical value to carry out the research on reliability and planning of microgrid.Researches on wind speed modeling,microgrid probabilistic production simulation method,reliability evaluation method and energy storage optimization are investigated in this thesis,under the support in part by the National High Technology Research and Development Program of China(863 Program Grant 2011AA05A107)“Research on the key technology of microgrid with distributed generator”.Wind speed sequence modeling is the basis for output analysis of the wind turbine generator and reliability evaluation of power systems including wind power.A time series wind speed model based on nonhomogeneous Markov chain is established to consider the sequential variation characteristics,which overcome the insufficiency of conventional models in daily and seasonal variation characteristics.Fisher's optimal partition method is used to divide the historical wind speed series into segments according to the degree of seasonal variation influence,and the seasonal influence component is extracted by the defined seasonal index.Then,potential change period of the wind speed is verified and extracted by Fourier transform.A time-related variable is used to establish a nonhomogeneous Markov chain wind speed model.The procedure of synthetic wind speed sequence with daily variation characteristics and seasonal effects are proposed.Historical wind speeds at four wind sites are used for modeling.Results show that the probability distribution fitting statistic R~2 at four wind sites is higher than0.99988,and the model can be used to accurately describe the daily and seasonal variation characteristics of wind speeds,which verify the accuracy and effectiveness of the proposed model.A sequential probabilistic production simulation method for microgrids is proposed to overcome the shortcoming of traditional probabilistic production simulation methods,which have the difficulty of considering the sequential characteristics of load demands,outputs of renewable energy,and energy storage.By analyzing the influence mechanism of system operation scheduling strategies on the loading order of generating units and the charge and discharge conditions of energy storages,a single-time probabilistic production simulation based on available capacity distribution is updated by combining the effect of charging equivalent load and discharging equivalent generating unit on the loading process of generating units.A time sequence energy storage state of charge updating model is proposed,and a sequential probabilistic production simulation method for microgrid is established.The proposed method is applied to an actual microgrid in Zhejiang Province.Compared with the sequential Monte Carlo simulation,errors in LOLE,EUE and expected system total power generation are 1.81%,1.10%and 0.03%,respectively,which indicates the proposed method can be used to effectively consider the microgrid sequential operation characteristics.Meanwhile,calculation time is only 18.3%of that for sequential Monte Carlo simulation.A microgrid reliability evaluation algorithm based on time-sequence probability simulation for islands is proposed.The algorithm considers the sequential characteristics of microgrids to overcome the shortage of traditional analytical evaluation algorithms in ignoring the sequential characteristics,and maintains their merit of fast calculation speed.A system network is simplified by a network block theory.Based on the block adjacency matrix,power failure types and island operating areas are determined by searching the controlled switches when a fault occurs.According to the charging and discharging power constraints of energy storage,a constrained k-means clustering method is proposed to establish a net-load multi-state model for an island.Combined with the distribution of storage SOC at the initial state in the islanded area,a continuous power supply analysis method for islanded areas based on time sequence probabilistic simulation is established.Then a microgrid reliability evaluation algorithm considering sequential characteristics is proposed.Case study shows that the results of the proposed analytical algorithm are close to those of sequential Monte Carlo simulation methods.The maximum error of the reliability index is 1.22%,whereas the calculation time is only 9.15%of the MC method.The operation scheduling strategies,energy storage capacities,and locations of the energy storage have significant effects on the system reliability,which should be considered in reliability evaluation.An alternative way to match power shortage in microgrid effectively is to purchase reserve from the external grid,besides configuring energy storages.A two-stage optimization model is proposed for optimizing capacities and locations of energy storages considering reliability constraints.The objective is to achieve the coordinated optimization of energy storages configuration and reserve purchase so that the total system cost is reduced.At the inner optimal stage,a minimum reserve capacity purchased in each time period is determined by a step-by-step searching method,with the constraints of system reliability indices.At the outer stage,a genetic algorithm based integer coding is employed to optimize the energy storage size and location.Numerical result shows that the proposed coordinated optimization can be used to effectively reduce the capacities of energy storages,so that the system cost can be reduced. |
PDF Full Text Request