Research On Collaborative Optimal Operation Method Of Integrated Energy System Considering Uncertainty

Striving a low-carbon,economic,and sustainable energy development paradigm is a crucial endeavor for humanity to confront the energy supply shortages,environmental pollution,and greenhouse effect.The integrated energy system(IES)challenges the conventional model of disjointed planning and operation of individual energy systems by facilitating coordinated planning,optimized operation,interactive response,and mutual complementarity of diverse and heterogeneous energy sources,providing a significant way for China’s energy upgrading and transformation,and the realization of the strategic goal of "carbon peak and carbon neutrality ".However,the stochastic and volatility of uncertainties bring challenges to the safe and efficient operation of IES.Meanwhile,the current subsystem of regional-level IES belongs to separate operating entities,and traditional centralized optimization may raise issues such as privacy breaches and communication difficulties.Furthermore,the interactions between the supply and demand sides of park-level IES are becoming increasingly frequent,leading to a shift from the conventional vertical management structure towards a hierarchical interactive competitive structure.To address the above issues,this paper takes regional-level and park-level IES as the main research object,and focuses on the IES collaborative optimization problem considering uncertainty according to the progressive order of electric-heat,electric gas-heat and electric-gasheat-cooling.In terms of uncertainty processing,a progressive system of uncertainty handling methods is proposed,which includes static robust optimization,dynamic adaptive robust optimization,worst-expectation robust optimization,and Wasserstein distributionally robust optimization.In the collaborative optimization of IES,distributed algorithms are applied to enhance the convergence of non-convex and multi-block models,and a collaborative optimization method based on the Stackelberg game is proposed.Finally,the theory and method of IES multi-energy collaborative optimization under an uncertain environment is established.The main work is summarized as follows:(1)In the aspect of regional-level electricity-heat IES collaborative optimization,a distributed cooperative operation method with hybrid stochastic/robust optimization is proposed.Firstly,the hybrid stochastic/robust optimization is combined to handle the multiple uncertainties of electricity-heat IES,in which the stochastic optimization is concentrated on the uncertainties and spatial-temporal correlativity of load and wind power,while the robust optimization is used to deal with the market electricity price uncertainty.Secondly,nonlinear constraints are proposed to eliminate the integer variables representing the pipe flow direction of the heat network,and an improved quantity regulation without the need for iterative calculation is constructed.Thirdly,the decomposition and coordination mechanism of the power and heat network is established for the multi-agent characteristics of electric-heat IES.The Bregman Alternating Direction Method of Multipliers(BADMM)is used for the distributed iterative solution.Finally,the case study demonstrates that the proposed hybrid stochastic/robust optimization effectively handles the uncertainties of electric-heat IES.The improved quantity regulation can reduce the model complexity while avoiding the iterative calculation of the hydraulic and thermal conditions for the heat network.Compared with the conventional Alternating Direction Method of Multipliers(ADMM),the adopted BADMM algorithm enhances the convergence of the non-convex distributed model with a flexible penalty function,and finally achieves the distributed cooperative operation of the electricity-heat IES under uncertainty.(2)In the aspect of regional-level electricity-gas-heat IES collaborative optimization operation,a dynamic adaptive robust optimization method with a multimicrogrid is proposed.Firstly,the uncertainty set with integrated spatiotemporal correlation and symmetry is constructed to address the drawback of large conservativeness for the traditional box uncertainty set.Secondly,a min-max-min adaptive robust optimization model for multi-microgrid is developed considering the uncertainty of distributed generation and electric vehicles.For the adaptive robust optimization model with integer variables in the third level,it is difficult to apply the KKT conditions or the strong duality theorem directly.Therefore,the Column and Constraint Generation(C&CG)algorithm and the alternating optimization procedure(AOP)are combined to solve the problem.Thirdly,since the traditional ADMM algorithm cannot guarantee convergence in three blocks,the PCB-ADMM algorithm with added prediction and correction parts is used to solve the distributed model of the three-block electricity-gas-heat IES.Finally,the case study shows that the proposed uncertainty sets are more accurate and realistic than the traditional box uncertainty sets,and significantly reduce the conservativeness of robust optimization.Compared with the traditional Nested C&CG,the C&CG-AOP algorithm does not need to increase the size of the subproblem,so it improves the computational efficiency.Moreover,the PCB-ADMM algorithm can effectively deal with the issue that the direct generalization form of traditional ADMM is not necessarily convergent,and realize the multi-agent autonomous decision-making of multi-block IES.(3)In the aspect of robust optimization operation of park-level IES,the worstexpectation robust optimization model is proposed.Firstly,to reduce the conservativeness of traditional min-max-min adaptive robust optimization,a min-maxmax-min worst expectation robust optimization model for park-level IES represented by Energy Hub(EH)is established,which considers the uncertainty variables from multiple classes and the probabilistic uncertainty of those classes.Secondly,to address the problem of nonlinear coupling of the objective functions in the subproblem,the C&CG and an alternating iteration strategy(AIS)are combined to accelerate the model solution.Thirdly,to reduce the conservativeness caused by the traditional uncertainty set always taking values at the boundary or outermost pole where the actual probability of occurrence is very small,an adaptive multi-interval convex hull uncertainty set(MCHUS)that can take values at the interior is proposed.Finally,the case study shows that the proposed worst expectation robust optimization model leverages the ideas of stochastic optimization,which can be regarded as a generalized form of traditional adaptive robust and stochastic-robust models.The C&CG-AIS algorithm solves faster and finds more robust solutions compared to the traditional C&CG algorithm.The proposed MCHUS can eliminate the low-probability scenarios in practice and has less conservativeness compared with traditional box uncertainty sets and convex hull uncertainty sets.(4)In the aspect of source-load cooperative operation of park-level IES,an optimization framework combining Wasserstein distributionally robust chanceconstrained(DRCC)optimization with the Stackelberg game is proposed.Firstly,the carbon emission flow(CEF)model for energy storage suitable for the change in the carbon intensity of the upstream node is proposed,and the integration of the CEF model and the ladder carbon trading model for EH low-carbon optimization is constructed.Secondly,the Wasserstein distributionally robust optimization is applied to deal with the uncertainty of EH in the horizontal dimension,and the DRCC is used to deal with some soft constraints of EH.Thirdly,the Stackelberg game model of upper-level EH and lower-level users is constructed in the vertical dimension.To address the issue that the EH model integrating Wasserstein DRCC and Stackelberg game is difficult to solve directly,various reformulation methods are used to transform the original semi-infinite model into a model that can be solved directly by a commercial solver.Finally,the case study shows that the proposed CEF model of energy storage model can effectively deal with the change in the carbon intensity of the upstream node.The framework combining Wasserstein DRCC and Stackelberg game not only leverages the advantages of stochastic and robust optimization,but also facilitates the interaction between EH and users,which guides the source-load cooperation of EH in an uncertain environment.