Optimal Operation Of Microgrids Under Generation And Demand Uncertainties

In the past decades,rapid depletion of fossil fuels and environmental pollution concerns as well as growing demand for electricity have resulted in increased advances in renewable energy generation technologies.The constant growth in renewable energy technologies has in turn paved the way for increased growth and application of distributed generation(DG).To realize the potentials of DGs,more recently,a new paradigm known as microgrid has emerged.Microgrids are a low or medium voltage small scale power networks containing distributed energy resources(DER)and local loads.Coordinated operation of distributed energy sources in the form of microgrids allows decentralization of electric power supply and hence has the potential to increase system reliability and power quality.The penetration of renewable energy into electric power systems has increased with the growth of microgrids.In order to achieve efficient and reliable operation of microgrids,economic scheduling of distributed generation units,energy storage systems and controllable loads within the system should be properly planned.The intermittent nature of the renewable energy resources like wind and photovoltaic(PV)resources poses a major challenge to microgrid operators.Electricity demand also exhibits highly stochastic characteristics.As the load and generation balance is a critical requirement in the power network,the uncertainties in renewable energy resources and load demand add challenges to power regulation.These challenges necessitate accurate generation and demand forecasting tools for planning efficient operation of microgrid systems and to ensure reliability of supply.Based on demand and generation prediction results and other operational information,optimal energy management strategies are employed to ensure economic and stable operation of the microgrid under both grid-connected and islanded modes.In this thesis,we propose hybrid modelling approaches for generation and load demand forecasting in microgrids.Adaptive network fuzzy inference systems(ANFIS)based models optimized by a combination of particle swarm optimization(PSO)and genetic algorithms(GA)are proposed for wind and PV power generation forecasting.In the modelling techniques proposed in this thesis,an integrated optimization algorithm combining PSO and GA iteratively optimizes parameters of an ANFIS model by exchanging variables of the best solution in each solution step between PSO and GA algorithms running in parallel.The proposed modelling approach benefits from the simplicity and effectiveness of the particle swarm optimization algorithm and the strong global searching capability of the genetic algorithm to iteratively optimize the relatively complex ANFIS structure.The integrated algorithm is also used to train a neural network model for wind power forecasting.In PV power forecast modelling,a feature selection strategy based on binary genetic algorithm is proposed that determines a suitable set of predictive variables to improve the efficiency and accuracy of the forecaster.To select the most important variables,the feature selection strategy uses binary GA to evaluate the mean squared error(MSE)of Gaussian process regression(GPR)models corresponding to different feature subsets.In addition,a hybrid load forecasting approach integrating empirical mode decomposition(EMD),particle swarm optimization and ANFIS is proposed.The proposed method first employs EMD to decompose the complicated load data series of the microgrid into a set of several intrinsic mode functions(IMFs)and a residue,and PSO algorithm is then used to optimize an ANFIS model for each IMF component and the residue.The final short term electric load forecast value is obtained by summing up the prediction results from each component model.Finally,a day-ahead optimal energy management strategy taking into account uncertainties in demand and generation is developed.The objective of the optimal energy management strategy is to minimize operational costs and enhance efficiency of energy utilization of the next day by optimally scheduling DGs and an energy storage system.The optimal resource scheduling model is formulated using mixed integer linear programming(MILP)approach.Conclusions are finally drawn from the results obtained in the study and future research directions are discussed.