Research On The Probabilistic Modeling Of Distributed Generation And Its Influences On Power Systems

Because of the decline of earth resources, the attention to environment issues and market reforms in the power industry throughout the world,the competitive mechanism was introduced to the power industry to achieve production, supply costs reductions and realize resource optimize configure. Distributed Generation (DG) has become an increasingly prevalent issue in electrical power industry, especially the renewable energy resources, for example, wind power and solar power generation. The increasing proportion of DG in power system has deeply affected the power network structure, reduced users'dependence on large power plants and the electricity grid system. At same time it has produced direct influences to the structure, investment, operation and stability of the power system. But for the random feature of wind and solar energy, its influences on power system also had the characteristics of random and uncertainties. Analyses on the random feature of this primary energy and its influences have important theoretical and actual value, and become a very important problem in smart grid.Based on cost analysis of DG, a planning algorithm using power market theory was proposed in this paper, the trust region interior point algorithm was used for solving the optimization problem. Due to the uncertainty of renewable energy,an accurate prediction of them is difficult to be implemented. In the third chapter, the paper described how to establish Box-Jenkins model for solar radiation based on system identification method, This method can be used to predict and simulation 5~15 minutes time interval of the surface solar radiation, hourly solar radiation and daily solar radiation. The prediction results and predicted errors were analyzed in detail, the residual analysis and zero and pole points test of all models were passed. Based on the established model, a new model for real-time prediction of hourly solar radiation was established by comparison and analysis on several methods, such as Kalman filtering based algorithm, normalized gradient algorithm and recursive least squares algorithm with forgetting factor.The inclination and azimuth Angle have important influence on output power of photovoltaic cell array, so the greatest possible optimal angle must be selected. Deciding optimal tilted angle for a stand-alone photovoltaic array,the uniformity and maximum of solar radiation on the surface should be considered comprehensively according to local latitude, solar radiation and yearly load characteristics. In the fourth chapter, the relationships between the daily optimal angle of extraterrestrial inclined surface and the day of year, the daily optimal angle of earth inclined surface and the daily optimal angle of extraterrestrial inclined surface, the daily optimal angle of earth inclined surface and the day of year were fitted by using the actual radiation data in the paper, and the daily or arbitrarily intervals of time optimal angle of earth inclined surface toward the equator can be calculated by using the average daily solar radiation. After that, a multi-step prediction model and its real-time online forecasting model were developed, the solar radiation on the photovoltaic (PV) surface and the temperature of PV cell array were used as the input data, and hourly output power of PV was used as the output data. The results indicated that the method had a certain practicality and the online forecasting model had better effects.The power output from wind farm is directly related to wind speed distribution, also display the characteristics of strong random. State space model PSS1 was used for short-term wind speed forecasting in chapter 5, the subspace method and iterative prediction-error minimization method was chosen to determine model parameters, the recursive least squares algorithm with forgetting factor was chosen to determine online prediction model parameters. After that, a Box-Jenkins model was developed; continuously history wind speed data were selected as inputs variants and power output of wind farm as output. It was shown by practical prediction that the model had higher precision during short-term predication with high application value.Chapter 6 showed that distribution of surface total radiation and direct solar radiation satisfied Beta distribution, but scattered radiation only satisfied mixed Gaussian bimodal distribution. Therefore, on the basis of probability model of surface total radiation, the probability distribution of PV power system was established by means of probability calculation formula. Two-parameter Weibull distribution was applied to generate distribution of wind speed, Linear least-squares method curve fitting, nonlinear least square curve fitting using the multiplicative errors model and the additive errors model, maximal likelihood method curve fitting were proposed to fit the Weibull curve. By comparing the calculated results getting from different models, it can be seen that maximal likelihood method was better in fitting the wind speed distribution. The probability distribution of wind farm was established by means of probability calculation formula. Then an algorithm of probability optimal power flows was established in chapter 7 based on point estimate method and Cornish-Fisher expansion, it can be used in the power system that wind and PV power were included. The analysis and computation results of total transfer capability and static voltage stability of IEEE test system shows that the proposed method maintains a high degree of accuracy and keeps low the computational burden when compared to the Monte Carlo method. At last, the paper analyzed the influence of wind and PV power with different distribution forms increase gradually in installed capacity connect to different node on total transfer capability and static voltage stability of power system based on the point estimate method. The results indicated that wind and PV power within a certain penetration range in power system made valuable contributions towards active power of static voltage stability critical point, and when there was only one renewable energy generation form, the probability distribution of active power of static voltage stability critical point had the approximate shape of the renewable energy. But to the same types of renewable energy generation, the influences on probability distribution of active power of static voltage stability critical point were different in different meteorological condition or different connection locations, especially the voltage of the node connected with the renewable energy generation. The simulation results show that the proposed method can provide effective decision-making supporting information for dispatchers, these basic rules also can afford more information for power system planning.