Dynamic Dispatch Method Of Integrated Energy System Considering Uncertainty

To achieve the goal of building a "clean,low-carbon,safe and efficient" energy system,it is necessary to further increase the consumption of non-fossil energy and improve energy efficiency.The integrated energy system(IES),which includes renewable energy and provides multiple forms of energy,is an important solution to these problems.In addition,compared with the traditional power system,the IES has the characteristics of various equipment types and operating conditions,coupling between energy sources and large differences in dynamic characteristics,and high randomness of renewable energy generation and energy demand,which makes its economic dispatch particularly complex.Based on the above background,this paper studies the dynamic-dispatch strategy based on the mixed logic dynamic model and the stochastic-dispatch strategy based on the scenario method according to the structural characteristics and management requirementsof the IES.The main research contents include:(1)In order to obtain the basic dispatch model for the follow-up research,the state space model of the main equipment in the typical IES is established.An inertial element is added to the micro-gas turbine to express its thermal dynamic characteristics.In addition,the 0-1variables are used to express the logic relationship of the start-stop or working status of various equipment.Finally,the probability distribution model of prediction error in IES is established,and its accuracy is proved by statistical indicators.(2)Aiming at the problem that conventional MPC can not solve integer optimization variables directly,the method of establishing MLD is proposed.The derivation of the transformation from logic relation to mixed integer inequality is given,and the nonlinear of objective function caused by full condition model is piecewise linearized.Then the dynamicdispatch problem is transformed into a MILP based on MPC framework,and compare with the static dispatch through simulation experiments.(3)The Two-stage stochastic optimization based on scenario method is adopted to deal with the randomness.In the scenario generation,a method based on inverse sampling is used to sample random variables and generate a large number of scenarios by using the multivariate standard normal distribution considering the correlation of variables.Then,the original scene set is deleted by the SBR method to reserve the most representative scenario for stochasticdispatch.In addition,in order to balance the computational complexity and accuracy,the ways to determine the best number of generation and reduction scenarios are presented.Finally,simulation experiments are given to verify the necessity of the proposed stochastic-dispatch strategy.