基于动态运行场景预测的多能微电网实时能量优化调控方法OA

The multi-energy microgrid(MEMG)has a high degree of integration among various types of energy systems during the planning,construction and operation process,realizing synergistic planning and coordinated operation among multiple heterogeneous energy forms of production,consumption and storage units.Through the complementarity among various forms of energy,MEMG shows great potential in enhancing the comprehensive utilization efficiency of energy,reducing the cost of energy usage and promoting the accommodation of renewable distributed generation(RDG),which is of great significance in promoting energy transition and realizing sustainable development.However,the multi-dimensional uncertainties induced by the dynamic changes in multi-energy loads and the intermittent and stochastic nature of RDG pose great challenges to the optimal dispatch and reliable operation of MEMG,which need to be urgently addressed in the process of MEMG energy dispatch.To this end,a real-time energy optimal dispatch method for MEMG based on dynamic operation scenario forecasting was proposed in this paper.The method characterizes MEMG operation uncertainties by dynamically forecasting a set of operation scenarios at each time slot,thus effectively addressing the negative impact of multi-dimensional uncertainties on MEMG operation. Firstly,a Wasserstein generative adversarial network(WGAN)suitable for the characteristics of MEMG operation scenarios was constructed in the method to mine and characterize their intrinsic statistical distribution in an unsupervised manner.Secondly,a constrained optimization problem for scenario forecasting was formulated based on known information(observations and point predictions of uncertain variables)and combined with the well-trained WGAN.It achieves efficient,high-quality and time-to-time forecasting of MEMG operation scenarios by optimizing the input vectors of the generator to effectively capture the multi-dimensional uncertainties of MEMG for a coming period.Finally,a MEMG pre-scheduling model was developed in the stochastic model predictive control framework based on the forecasted scenarios to accurately obtain the MEMG pre-scheduling commands.A real-time power compensation model was also constructed to compensate for the unbalanced electric and thermal power in the most economical manner,thereby ensuring the real-time power balance of MEMG. Comprehensive numerical simulations fully validate the effectiveness of the proposed MEMG operation scenario forecasting and real-time energy optimal dispatch method.The developed scenario forecasting method can effectively forecast the MEMG operation scenarios in different forecast time ranges and can effectively capture the edge distribution of the actual scenarios,realizing the accurate characterization of the uncertainties for MEMG operation in a coming period.Meanwhile,the real-time energy optimal dispatch method constructed based on the forecasted scenarios effectively mitigates the negative impact of multi-dimensional uncertainties on MEMG operation,and shows more significant economic benefits than the traditional model predictive control method. In the future,the energy forms integrated into MEMG will be more diversified,the forms of MEMG will be more complex and diverse,and their control will be more intelligent.Therefore,future research will focus on how to use digital twins and other advanced tools to improve the operation state sensing level and fine dispatch performance of MEMG with close coupling among energy forms such as hydrogen,biomass,heat,gas,electricity,etc.