Study On Short-term Power Forecasting Of Distributed Generation System

With the gradual depletion of fossil fuels and the deterioration of environmental pollution, the efficient use of energy has attracted a lot of attention around the world. Distributed generation(DG) system can improve the utilization of energy. As essential forms of DG systems, wind power systems and photovoltaic power systems are flexible, reliable and environmentally friendly. However, wind and photovoltaic power systems have strong randomness, which has a negative effect on power quality, energy management and grid scheduling.This thesis is based on the status quo of distributed generation system, and makes an in-depth study on short-term power forecasting of DG system. Firstly, phase space reconstruction technique is used to verify the chaotic properties of wind power, which is greatly intermittent and unpredictable, and a short-term forecasting model of wind power is proposed based on support vector machine(SVM). Several environmental factors influencing the output of PV system are analyzed, and historical data is classified according to generalized weather types. Then, a short-term power forecasting model of photovoltaic system is established based on GA and SVM. The results of practical examples show these two models are able to track short-term trends of DG outputs, together with high prediction accuracy. In order to describe the interval, indicators such as PICP, PINAW and PIACE are introduced. After that, a BPNN-based short-dated output interval prediction model of DG system is proposed. Also, an improved PSO algorithm is used to optimize parameters of the model, which can help to make up the shortcomings of easy to fall into local optima and slow to convergence. The final example shows that the model can obtain an effective short-term power interval of DG systems under a certain confidence probability.