Research On Energy Management And Expansion Opmization Of Smart Grid Under Uncertainties

Recently, with more conren on global environment, achieving the sustainable development is the most challenge for many countries around the world. Especially in China, the coal-fire dominated electricity industry, heveay indsurty, and numerous vehicles leads to terrible hazy weather, which is easy to cause respiratory diseases, heart disease, cancer and birth defects, and has great threat to human’s health. Power industry is one of the main sources of emissions of sulfur dioxide, nitrogen oxides, and particulate dust pollution. The emissions of sulfur dioxide and nitrogen oxides from thermal power plant count for 40% of country’s total emission. Facing the fast economic development, lack of resources, and environmental constract, smart grid with the advantages flexible, clean, security, economic and friendly, has brought the new era of electric power system. Although smart grid can solve many of the contemporary problems, they give rise to new control and optimization problems. Compared with traditional electricity system, the ulitization of renewable energy is the greatest change at the generation side. Renewable energy generation (mainly wind and solar), highly dependent on climate change, is intermittment and stochastic. Distributed and large-scale renewable energy access to grid brings challenge to the safety and reliability of power system. Thus, imporving the accuracy of renewable energy generation is a veray important task, and could help to decreas the uncertainties in smart grid’s energy manamgenet. Except the uncertainties in renewable generation, with the diversity and liberation of electricity market, there are more complicate uncertain factors in the manamgent and programming decision, for example, the dynamic change of final demand, the fluctuant price of energy market, and the future policy from the government. Based on the issues mentioned above, the main contents of this paper include:(1) Develop the short-term renewable energy generation forecasting models. For wind power, a hybrid method based on empirical mode decomposition (EMD) and echo state network (ESN) is proposed. The original series of wind power output are decomposited to several subseries with different frequency. According to each subseries’character, different echo state network are trained to forecast. The final forecasting results are obtained by summary. For photovoltaic power output, according to the idea of knowledge mining with environmental simulation, knowledge database including the main factors of PV generation is built. Self-organization mapping network is adopted to cluster the history samples. The samples in the same cluster with forecasting day are selected to train neural network to gain better performance. Besides, to avoid the overfitting problem in neural network traing, Bayesian theory is also employed to adjust the weights of network.(2) Study the energy management and unit size optimization of residential regional distributed energy generation with CCHP system. Considering the uncertainties such as residential energy demand, market price, technology parameters, the proposed model tries to find the optimal unit size and operation strategy of various distributed energy generation in system with the goal of minimal system cost and the constraint of resource, techonology and environment. By combining the interval two-stage programming model and robust optimization theory, we can improve the robustness of the model and reduce the system operation risk.(3) Study the energy management optimization of grid-connected regional smart grid with renewable energy generation. Except the uncertainties mentioned above (e.g. demand change, energy price, market fluctuation, government regulation), considering the operation risk caused by renewable energy forecasting, an interval two-stage stochastic programming model based on fuzzy chance constraint programming is presented. FCCP could evaluate the shortage risk by renewable energy output, and we can choose proper energy management strategies under different confidence and emission reduction goal.(4) From the long-term energy management perspective, study the risk-aversion expansion optimization of regional smart grid with renewable energy generation. Since the construction of power system needs a larege amount investment and is irreversible, the future planning of regional grid should be carefully decided. The proposed risk-aversion programming framework based on CVaR and down-side risk theory could optimize the smart grid with renewable energy sources for maximum economical and ecological benefits under uncertainties. The optimal results could provide the stratigies of unit expansioin, generation schedule, carbon capture technology investment and CO2 emission trading arrangement, according to the risk preference of decision-makers.The renewable energy generation output forcasting models and the smart grid energy management and plan programming models under uncertaines proposed in this paper could provide efficient reference for decision-maker facing comlex electricity power system. They could impove the economy of power system management, and ensure the safety and relaiblity of power system operation, which is significant to achieving the sustainable development of power industry.