Optimal Dispatch Methodology For Integrated Energy System Considering Photovoltaic Uncertainty

To promote carbon peaking and carbon neutral,the integrated energy system has become the main trend of energy system development due to its advantages of increasing the penetration of renewable energy.The output power of renewable energy is highly random,fluctuating and intermittent,resulting in inaccuracies and uncertainties in the power prediction process,which brings challenges for the coupling transformation and cooperative operation optimization of integrated energy system.In view of the above background,this thesis intends to investigate the optimal dispatch methodology of integrated energy systems under uncertainty conditions.The research is of great significance to improve the accuracy of power prediction of renewable energy systems,enhance the dispatch efficiency of integrated energy system and ensure the security of operation of the integrated energy system.The main contributions and novelties are summarized as follows:(1)A K-nearest model(KNM)evidence regression algorithm for photovoltaic power prediction is proposed in this thesis aiming at the uncertainty issue of PV system power generation prediction.The KNM method is a multi-model fusion forecasting method based on the framework of evidence theory,which achieves the representation of cognitive uncertainty characterization in the PV power forecasting process using evidence discounting and reduces the forecasting error through the fusion among multiple models.The KNM method uses neural network algorithms to build sub-models for each neighbor,the uncertainty of each model is represented in terms of the data dispersion among the sample space.The sub-models are combined according to evidential theory,and the parameters of sub-model as well as the structural parameters are unified for identification.The simulation results indicate that the proposed KNM evidence regression algorithm has high accuracy in deterministic prediction of photovoltaic power generation and high reliability in the uncertain prediction.(2)A conductor harmony search(CHS)algorithm is proposed for the optimal dispatch of integrated energy systems considering the coupled multiple energy flows and complementary spatio-temporal characteristics.The CHS algorithm employs a conductor state memory(CSM)and time series conjoint constraint model to achieve decoupling of time series constraints.The CHS algorithm divides the multi-energy flow and time-coupled variables into harmonic variables and solo variables in accordance with time series and equipment order.For the time series coupling constraints between solo variables,a conductor melody memory(CMM)and the corresponding CSM are constructed.Also,a time series conjoint constraint model for each period and the periods with direct correlation is built.The CMM is optimized sequentially in accordance with the CSM and the time series conjoint constraint model.The results demonstrate that the proposed CHS algorithm reduces the computation time considerably and achieves better solutions with higher accuracy and efficiency when dealing with the optimal dispatch problems of integrated energy systems.(3)A hierarchical and multi-time scale feedback optimal dispatch approach is proposed for the dispatch of integrated energy systems under multi-time scale uncertainties.The proposed method introduces the feedback of safety and economic indicators of the short time scale operation period into the long time scale dispatching process,forming a closed-loop optimization of the long-short time scale dispatching process.Also,a Monte Carlo-k nearest neighbor method is proposed for two-stage sampling for long-short time scale variables,and a stochastic robust optimization(SRO)model is built under two-stage correlated random variables.The feedback dispatching scheme is optimized according to the sampling of uncertain variables at multiple time scales and two-stage SRO model.The case studies proved that the dispatching results based on the feedback models are more conservative,which result in slight reduction in economic efficiency,but the dispatching results are significantly improved in feasibility during system operation,which leads to more backward compatibility multi-time scale dispatching solution.