Study Of Energy Management System For Wind/PV/Battery Micro-grid

With the development of the society, energy crisis, environmental problems and the drawbacks of traditional power system are emerging recently. Wind energy, solar energy and other new energy has become an important way to solve these problems. But, most new energy is intermittent and randomness under different circumstance. When connected to the power grid distributedly, such new energy may cause some problems like voltage fluctuation, frequency fluctuation or may change the trend of the distribution of the grid power. Micro-grid is a new structure to combine Distributed Generations, the storage unit, load, control device and protecting equipment. It is a good solution to connecting to the power grid for Distributed Generations. The energy management system for micro-grid is the key part to make the micro grid safe and reliable.The thesis introduced the consistence of the energy management system. To the Distributed Generations, the paper described different control methods and the corresponding Block diagram. The advantages and disadvantages of the different control method for the micro-grid system are compared. What's more, the thesis analysed the operation strategies of the micro-grid under different operating conditions, and provided the basis for the economic operation of micro-grid.The accurate prediction for the output of the wind turbine, the photovoltaic and the demand power for the load will optimize the operation of the micro-grid, and improve the efficiency of the energy scheduling. Support vector machine is good at solving small sample, nonlinear issues. Support vector regression is an expansion of Support vector machine, and is developed rapidly in forecasting area. The thesis collected the history data to predict the power of the wind turbine, the photovoltaic and the load in short-term through the toolbox(LIBSVM) in MATLAB based on support vector regression method. After testing many times, we can find that it is more accurate to use historical output data in front of 5 hours to predict output data in 1 hour behind. Then update the data and continue to predict the rolling 1 hour output data. Based on the output data from 0 a.m. to 5 a.m. of the wind turbine and the load, 1 hour rolling output data from 5 a.m. to 24 p.m. is predicted. Based on the output data from 6 a.m. to 11 a.m. of the photovoltaic, 1 hour rolling output data from 11 a.m. to 19 p.m. is predicted. The results show that the method is effective, the predict value is close to the true value.A practical energy management platform is designed in the thesis to apply to an actual micro-grid. The platform is consisted of three parts, the interface, energy management panel and the communication system. An effective communication system is proposed. The platform acts well in the working micro-grid.