Power Generation Forecasting,Load F Orecasting And Coordinated Optimization Of Multi-energy Systems

Under the challenges of national energy security and global climate change,developing renewable energy has become a major strategic move for global energy transition and addressing climate change.However,due to the intermittency and volatility of renewable wind and solar energy,the real-time matching of renewable power generation and power demand faces great difficulties,and there is serious abandonment phenomena of wind and solar energy in their development and utilization process.The 14th Five-Year Plan for Electric Power Development of Chinese government clearly pointed out that the multi-energy complementarity is an important means to promote the utilization of wind and solar power.The inherent variability and uncontrollability of wind and solar power generation have a great impact on the operation of multi-energy complementary systems.At the same time,the utilization rate of wind and solar power is also an important problem.Therefore,focusing on the distributed and centralized multi-energy complementary systems coupled with renewable energy,with the goal of improving the consumption level of renewable energy and the economy of system operation,this paper studied the high-accuracy predictions of electricity load and renewable energy generation,and based on the prediction,the paper studied the collaborative operation optimization method of multi-energies,so as to provide theoretical reference for the management of future energy systems and promote the construction of a clean energy supply system.The main contents are as follows:(1)Accurate load forecasting is the base for the operation optimization of the multi-energy system,and is an indispensable part of energy management in the multi-energy systems.In view of the problems of low accuracy and high computation complexity in the forecasting based on traditional combination of data decomposition technology and machine learning model,this paper proposed a new model that combines empirical mode decomposition and multi-layer perceptron(EMD-MLP)to forecast the one-day-ahead load.EMD was introduced to decompose the original load signal into multiple intrinsic mode functions(IMFs),and the extreme point division method was adopted to reconstruct the IMFs into two new components,i.e.,high frequency component(HFC)and low frequency component(LFC).The HFC and LFC were predicted separately,and their predicted values were superimposed as the forecasts of the load.The simulation was conducted using real load data measured in the Australian electricity market.Results show that the proposed EMD-MLP model outperforms the baseline models,i.e.,the persistence model,the single MLP model,and the traditional EMD combination model in the extrapolation performance.At the same time,this method greatly reduces the computational task compared with traditional combination forecasting method.(2)High-precision prediction of photovoltaic(PV)power generation can provide foundation for the coordinated optimization of multi-energy systems,and is the basis for improving the renewable energy utilization and system operation economy.This paper proposed an ensemble forecasting model to make the day-ahead forecasts of PV power generation.This method combined the structure and the data diversity enhancement techniques that were used separately in the traditional ensemble prediction models.Based on the prediction form of ’multiple datasets-multiple learners’,a variety of candidate base models were generated,and the ones with better performance were selected for integration to produce the final prediction result of the PV power.The measured data from the 15kW PV power station in Ashland,USA was used to carry out the simulation.Results indicate that under three different weather conditions,the generalization performance of the ensemble model is significantly superior to that of each single model,which verifies the effectiveness of the developed ensemble model.(3)Assisted by the accurate prediction models of electricity load and PV power generation,a dual-layer day-ahead and real-time optimization method was proposed for the coordinated optimization of the distributed multi-energy microgrid that contains distributed photovoltaic(PV),battery(BS)and micro gas turbine(GT).In the day-ahead layer,the day-ahead optimization operation strategy was proposed aiming at maximizing the daily operation economy of the microgrid.Based on the day-ahead forecast of electricity load and PV power generation,the day-ahead plan of the power allocations among the PV,BS,GT,and the main grid was formulated.At the real-time layer,minimizing the operating cost of each scheduling period(15-minutes)was used as the optimization goal,and the real-time optimization strategy was established.Based on the ultra-short-term forecast of electricity demand and PV power generation made 15 minutes ahead,the real-time power allocation plan was formed.Compared with the only day-ahead optimization method,the proposed dual-layer method can further correct the day-ahead plan through real-time optimization,and thereby effectively eliminates the impact of the deviations of day-ahead forecasts on the operation economy of the microgrid.(4)Based on the reliable day-ahead electricity load and PV power generation prediction results,a phased optimization scheme was proposed for the collaborative operation of centralized multi-energy system containing PV,pumped storage(PS),and thermal power units.The scheme decoupled the operation optimization of the PV and pumped storage power station(PV-PSPS)from the thermal power units.In the 1st stage,the PV-PSPS were regarded as a joint system.Based on the day-ahead forecasted electricity load and PV power,and the realistic operation mechanism of the PSPS,the optimal dispatching strategy aiming at the maximum joint power generation of the PV-PSPS system in a scheduling day was established,and the plans of the power allocation of PV,the output of the PSPS were formed.In the 2nd stage,on the base of the joint power output of PV-PSPS,the thermal power generation plan was scheduled.Taking the minimum coal consumption of thermal power units as the objective function,the equal incremental principle method was adopted to optimize the output distribution among multiple thermal power units.Compared with the operation scenario without the integration of pumped storage,in the optimized operation scenario of the multienergy system containing PV,pumped storage and thermal power units,the utilization rate of PV power generation is significantly improved under three typical weather conditions,and the proportion of thermal power and coal consumption are significantly reduced,which proves the effectiveness of the proposed phased optimization scheme.The research results of this paper can provide reference for the optimal operation of China’s multi-energy complementary demonstration project,and have important value for building a clean,low-carbon,safe and efficient energy system and achieving the goal of "double carbon".