Research On Several Problems In Decision-making Of Microgrid Planning And Operation

Along with the serious issue of energy and environmental protection, most countries around world propose and develop their own smart grid. The microgrid, considered as an effective implementation of connecting electric power grid with distributed generation (DG), plays an important role of Smart Grid. In microgrid, renewable energy is the main component of microgrid. The renewable energy will bring a lot of issues regarding optimizations and decision-making of microgrid owing to its output uncertainty relating to power load forecasting. Therefore, in order to make microgrid perform better, it is extremely urgent to develop quantitative analysis about optimization and decision-making of microgrid.Therefore, some plan and movement decision-making problems of microgrid including power load forecasting, optimization and dynamic division of microgrid, and resilience of power systems are studied in this dissertation. It is divided by the following parts:(1) The current researches in relative fields are surveyed by reviewing a large number of domestic and international monographs, magazines, journals and conference proceedings. Firstly, more than100year development history of the world electric power industry is summarized. Secondly, the applied prospect of microgrid is briefly analysed. Thirdly, relative issues of power load forecasting, optimizations and division of microgrid are surveyed. And finally, the main thought, improvements and applications of the principal component analysis (PCA), partial least squares (PLS), orthogonal signal correction (OSC) used in load forecasting are introduced briefly.(2) The forecasting model is built by PLS through withdrawing strong explanatory components, and the issues of weak model explanation and dropped forecasting ability caused by original least squares (OLS) are solved. In the PLS forecasting model, the independent variable sample includes many orthogonal components with the forecast variable, which not only model is fit excessively and also the forecasting precision is reduced. For this reason, OSC is usd to remove components which are not related to the forecast variable in independent variable. It means the PLS model can be built by few main components. The annual electricity consumption forecasting model based on the DPCA-BP neural network is proposed. It uses dynamic PCA to solve the problem of data succession and establishes the forecast model by the extracting main components considered as neural network input.In order to verify explanatory ability and forecasting performance of the forecasting model, recent years’electricity consumption and the influencing factors in Liaoning Province are collected and analysed. The results indicate that both of improved forecasting model correspondingly simplify the model structure and have better predictive ability.(3) The influence that microgrid brings into the flow and system loss of traditional distribution network are analysed in detail, regarding turning on position and capacity of the micro power. The mathematical model of microgrid optimization is established by minimizing systems losses for goal. The genetic algorithm of which chromosome length may be changed is developed to solve the model. In order to verify the feasibility and efficiency of model and algorithm, IEEE69node’s distribution system is model. The results indicate that the proposed models and the algorithms can dispose the mierogrid effectively.(4) The microgrid division is the guarantee of automomous movement of the microgrid after the principal network breakdown. Therefore, the principle of microgrid division is proposed and mathematical model of dynamic microgrid division is established. The model is calculated by "search and verify"algorithm. It is simulated analysed to take the IEEE69node’s distribution system. The results indicate that the proposed algorithm reduces the division step of microgrid effectively, forms the autonomous pattern of microgrid, and reduces power cut area.(5) The appraisal crterrion of electric power system resilience is proposed. The influences on system resilience due to turning on position and quantity of micro power are analszed. Then, a quantificational evaluation approach for electric power network resilience is designed. The essence is to take the ratio of distributed reliable power supply over the load demand as the resilirence measure of a load node. The network resilience is the weighted sum of all load node resilience. Based on the resilience evaluation, the computational method of distribution network structure optimization is proposed. The approaches are applied to analyze and evaluate the resilience of7bus distribution network. It comes to the result that system resilience has the best performance when micro power turns on short position of power supply, which indicates the power restoration ability of the actual electrical system.