Translated Abstract
The world’s energy development is facing the increasingly grim situation of resources shortages and environmental constraints. Smart grid, which aims at efficient, clean, safe and sustainable energy supply, becomes a technical solution to this problem. With the integration of distributed renewable energy, the diversification of energy demand, supply and the deep integration of advanced sensors, measurement and communication technologies in the smart grid, the transmission of energy and information in which is influenced by several kinds of uncertainties. To this end, how to improve the balance of power supply and demand, the rational use of renewable energy and ensure safe and stable operation of power grid have become critical issues to the development of smart grid. Smart Grid is a typical Cyber-physical System (CPS), in which, Energy Management System (EMS) has shouldered the mission of power grid operation optimization, renewable energy access, grid and user interaction, and drawn much attention from both academy and industry circles.
In this paper, the related works of the EMS optimization strategy is elaborated, the critical issues of optimization strategy to be solved urgently is summarized in detail and then the EMS optimization strategy of Smart Grid with considering uncertainties in both energy flow and information flow are studied. From the view of energy flow, with considering the uncertainty of energy supply side and consumption side, the statistical analysis and prediction of load demand are conducted; the double auction mechanisms under the premise of uncertainty of demand and transaction time are proposed to address the energy transaction issues among microgrids; the optimal electric vehicle scale penetration strategy under the premise of uncertainty of renewable energy is investigated to address the issue of electric vehicle penetration in microgrids. From the view of information flow, the integrity attacks against EMS state estimation is studied to address the uncertainty in information transmission. The main research results of this paper are as follows:
(1) To address the uncertainty raised by consumer demand, statistical analysis and load forecasting method are carried out based on a large number of real world household smart meter data sets. Firstly, based on two kinds of non-parametric methods, Shapiro-Wilk and Quantile-Quantile, statistical distribution test of load data are conducted. Secondly, the short-term load forecasting methods based on two kinds of machine learning approaches are proposed. The experimental results demonstrate that the user load can be approximated as a Gaussian distribution over a certain period of time. Compared with the traditional prediction method, the short-term load forecasting error of SVM and LSSVM based approaches can be controlled lower than 7.12% and 14.56%, the regression coefficient of SVM based approach is close to 0.76. The results show that the proposed approaches are capable of improving the load forecasting precision.
(2) To address the issue of demand side bidding and energy trading, in the grid-connected multi-microgrid system, double auction mechanisms are proposed to ensure energy transactions among local microgrids, with considering the uncertainties raised by microgird demands and transaction time. The energy transaction problem is formalized as the social welfare maximization problem with taking energy surplus and insufficient microgrids as sellers and buyers, and iteration based offline double auction mechanism and valid price based online double auction mechanism are proposed to ensure the efficient and fair energy trading. Theoretical analysis demonstrates that offline double auction mechanism can fulfill the users demands and achieve maximum social welfare, online double auction mechanism can achieve the economic properties of strategy-proof (incentive compatibility), individual rationality and budget balance. Simulation results show that the offline and online mechanisms can reduce the operating costs of microgrid by 6% and 10%, respectively. The offline double auction mechanism is able to achieve optimal social welfare, while the online double auction mechanism is capable of achieving higher customer satisfaction ratio and computational efficiency.
(3) To address the uncertainty raised by renewable energy on the energy supply side, optimal strategy of electric vehicle (EV) penetration in archipelago microgrid system is proposed. Firstly, to minimize the operating costs and environmental emissions, the optimal EV penetration problem is formalized as a stochastic programming problem with considering the uncertainties of renewable energy. Then, the uncertainties associated with renewable energy are captured by Monte-Carlo method, a realistic scenario with considering the peak load limit of microgrid is proposed to schedule the scale of EVs with satisfying the load demand and constraints. Simulation results demonstrate that the proposed approach utilizes the reasonable scheduling of the EV scale to improve the energy efficiency of the grid and reduce the environmental emissions effectively.
(4) To address the uncertainty in information flow, the information integrity attack against EMS state estimation is studied. From the attacker’s point of view, the least effort information integrity attack model is constructed, which aims at tampering with the minimum number of sensor measurements information and causing incorrect estimate of power grid operation state. The above problem is considered as a typical NP-hard problem, and it is difficult to obtain the optimal solution in polynomial time. To fully explore the gird Jacobi matrix information, a novel optimal attack vector solution algorithm based on Hermit standard is proposed, on the basis of proving the existence of the optimal solution. It is theoretically proved that the algorithm can be converged with lower time complexity. The simulation results also demonstrate that compared with the existing algorithms, the proposed algorithm is capable of solving optimal attack vector with only 0.27s on IEEE-118 test system. Moreover, the time complexity of solution is exponentially decreasing with deployment of smart sensors. The time overhead drastically decreases to the 10−2, when the fraction of smart sensors achieves to 0.25. The research of the attack strategy can provide the decision basis for the defense mechanism.
Corresponding authors email
