Robust Optimal Scheduling Of Microgrids Considering Uncertainties

With the increasing environment pollution and serious fossil energy depletion,the exploitation and utilization of clean energy has become a concern of global energy development.Renewable energy(RE)generation represented by photovoltaic(PV)and wind power is favored for its low pollution,high economy and flexible operation.Microgrid(MG),as a kind of small-scale autonomous system gathering distributed generation,energy storage and load,has been an effective technology and an important way to utilize and absorb RE in the fields of power and energy systems.RE generation is affected by natural conditions with strong randomness and intermittence,which brings great challenges to the stable operation of MGs.To tackle the impact of uncertainty on MG scheduling,it is inevitable that MG scheduling changes from deterministic problem to uncertainty optimization.As robust optimization(RO)shows advantages in dealing with uncertainty problems,it has attracted extensive attention in MG scheduling.However,considering the flexibility of MG operation and the decentralized clustering of units,there is still a lack of robust scheduling architecture for MGs from centralized decision-making to distributed optimization.Therefore,based on the engineering background of widespread uncertainties and distributed scheduling framework,this paper investigates the robust scheduling problem for MGs under uncertainties from the aspects of theory analysis,system modeling and method implementation.Centralized robust decision-making methods,as well as distributed scheduling approaches,are explored for the flexible and reliable operation of MGs and the decoupling of multi-entities.This work forms the theoretical framework and implementation methods for robust MG scheduling,and also provides technical support for practical engineering.The main contents of this paper are as follows:1)Multi-interval uncertainty based robust optimal operation of MGs.Firstly,the deterministic scheduling model for MG is established without considering any uncertainty factors.To characterize the spatiotemporal characteristics of source-load uncertainties,a multi-interval uncertainty set is further constructed using probability distribution functions,and then the two-stage robust scheduling model is formulated considering the startup and shutdown of RE units.Finally,nested column-and-constraint generation(C&CG)algorithm solves the robust model to obtain the scheduling plans in the worst scenario.2)Historical-correlation-driven robust optimal operation of MGs.The historical data of PV stations is analyzed to demonstrate the temporal and spatial correlations of RE generation.The correlation information is extracted by fitting the data points of similar days,and the historical-correlation-driven uncertainty set and robust scheduling model are construct accordingly.A gradient-descent algorithm with equilibrium constraints is developed to solve the nonlinear RO problem caused by correlated uncertainties,and to accelerate the convergence of the optimization for binary recourse variables.Centralized robust decision-making method provides theoretical support for the follow-up research on distributed coordination scheduling of multi-MGs.3)Distributed robust coordination scheduling of multi-MGs with peer-to-peer cooperation.Based on robust decision-making theory,distributed robust coordination scheduling of multi-MGs considering peer-to-peer cooperation is studied.Firstly,according to the synchronously distributed scheduling framework,an improved analytical target cascading(ATC)method is proposed to ensure the consistency of shared tie-line plans and realize the decoupling of multi-entity scheduling,then the synchronously distributed scheduling model is established.Secondly,the optimization of virtual center is assigned to each entity,thus formulating the distributed scheduling model with peer-to-peer cooperation.Finally,a C&CG algorithm with looped alternating optimization procedure(AOP)is proposed for the nonconvex distributed RO problem,which ensures the convergence and efficiency of modeling solution.4)Distributed robust coordination scheduling of multi-MGs based on multi-level Stackelberg game.The multi-level Stackelberg game based distributed robust coordination scheduling approach is studied under the hierarchical scheduling architecture with supplier and users.To maximize the operating benefit of each entity,a trilevel Stackelberg game scheduling model is established,and the existence and uniqueness of the equilibrium point are proved by multistep backward induction method.Considering the information interaction between different entities,a two-stage distributed iterative algorithm is proposed for the multi-level game scheduling model.Mathematical programming with equilibrium constraints(MPEC)and bisection approach improve the computational efficiency of distributed optimization,and the oscillation in iterative optimization is also settled effectively.5)Distributed robust coordination scheduling of multi-MGs considering equilibrium of individual-collective profits.To realize an equilibrium between individual and collective profits in multi-MG scheduling under uncertainties,a bilayer distributed RO model is developed,where sub-MGs conduct two-stage robust scheduling in the upper-layer,and the coordinating center implements multi-objective Nash bargaining scheduling in the lower-layer.An iterative descent algorithm is further proposed for the distributed scheduling model.Therein nested C&CG algorithm and two-stage complementarity approach solve the upper-and lower-layer problems iteratively.The optimal objectives decrease with iterations,which ensures the convergence and optimality of distributed optimization.