Deep Learning-based Scheduling Of Integrated Energy System Under Short-term Load Forecasting

With the continuous implementation of the energy Internet strategy of our country,the development and utilization of the integrated energy system have been gradually promoted,so that power-to-gas technology(P2G),combined heat and power technology(CHP)develop continuously.The coupling degree between electric power system and natural gas system is further improved,and it provides a new way for the consumption of renewable energy such as wind power.In order to increase the proportion of new energy consumption,reduce carbon emissions,and improve the reliability of system power supply,it is very important to carry out short-term load prediction of comprehensive energy system and low-carbon optimal scheduling of comprehensive energy system.The main research contents of this paper are as follows:1)starting from the basic theory of short-term power load forecasting,first introduced the features of short-term power load forecasting,this paper expounds the power load characteristics,and further analyzes the micro power grid wind power output characteristics,micro gas turbine output model,analyzes the topology of the energy storage system,this paper introduces the P2 G technology,In addition,the energy storage system and P2 G mathematical model were constructed,which laid a theoretical foundation for the main work content after that.2)A short-term power load forecasting method with bidirectional gated cyclic network structure is proposed.Based on the traditional GRU neural network,the method makes a second cycle of the GRU neural network,so that the neural network model can learn the deep factors affecting the power load data from the historical and future power load data.The correlation analysis of various influencing factors of power load is made,and the multi-input and multi-output forecasting model is established,which can improve the forecasting speed of the whole model and output complete load data.In addition,a variety of short-term load prediction methods are used to make a final comparison with the model constructed in this paper to verify that the proposed(Bidirectional Gated Recurrent Unit(BIGRU)short-term load prediction model has higher accuracy.3)In order to improve the short-term dispatching level of the comprehensive energy system and solve the problems of poor economy and excessive carbon emissions of the comprehensive energy system,a method of operation optimization of the comprehensive energy system considering the carbon trading mechanism is proposed by giving full consideration to the role of the carbon trading market.In order to explore the benefits of the P2 G two-stage operation process and the potential advantages of less consideration of the adjustable thermoelectric ratio in the optimal operation with CHP.In the traditional integrated energy system,the adjustable thermoelectric ratio model,the refined P2 G operation model,and the carbon trading mechanism model are established.In order to comprehensively consider the influence of the above three models on the scheduling of the integrated energy system,the lowest sum value of the energy purchase cost,carbon trading cost and wind abandoning cost was set as the objective function,and the optimization scheduling model of the integrated energy system was established considering the constraints of the operation of various equipment in the system.In order to solve the economic dispatching problem of integrated energy system more effectively,the original mixed integer nonlinear problem is linearized into mixed integer linear problem.The base load of the dispatching day was predicted by BIGRU load forecasting method,and eight different dispatching scenarios were set in the example analysis.The influences of step carbon trading,adjustable thermoelectric ratio and refined P2 G on the optimal dispatching results of the integrated energy system were respectively compared and verified.The effectiveness of these three technologies in integrated energy system optimization and the reliability of their collaborative operation are verified.