Research On Active Distribution Network Scheduling With Distributed Energy

With the increase of the proportion of new energy in active distribution network,the unbalanced relationship between new energy output and load demand becomes increasingly prominent,and various problems caused by new energy consumption gradually emerge.On the one hand,the scheduling strategy of active distribution network can be optimized by adding dispatchable loads into the system,thus improving the scheduling flexibility of active distribution network.On the other hand,the accuracy of new energy output forecast has an important impact on the rationality of distribution network dispatching.Improving the accuracy of new energy output forecast can greatly reduce the difficulty of dispatching.Based on this background,the new energy output prediction method considering the sequence decomposition algorithm is studied,and the optimal scheduling model of active distribution network considering schedulable load is studied from two time scales of day-ahead and intra-day.Specific research contents are as follows:First of all,the new energy output prediction model is constructed to forecast the new energy output of day-ahead and intra-day.Based on the measured data of a new energy power plant in Xinjiang Province,this paper analyzed the characteristics of new energy output,and established a new energy output prediction model based on sequence decomposition and selection.The model was divided into three parts: decomposition,selection and prediction.In the decomposition part,the phase space reconstruction and empirical mode decomposition algorithm were used to decompose the related influencing factors into several sequences.In the selection part,the concept of individual was introduced into the genetic algorithm to optimize the sequence according to the requirement of the problem.LSTM neural network and BP neural network were used in the prediction part.The three parts of decomposition,selection and prediction were combined to complete the day-ahead and intra-day prediction of new energy output,and improve the day-ahead and intra-day prediction accuracy of new energy output.Then,the day-ahead scheduling optimization model of active distribution network was constructed.Based on IEEE 33 distribution network model,dispatchable loads such as electric vehicles and energy storage devices were added,and reactive power optimization devices such as capacitor banks,static reactive power compensators and on-load regulating transformers were added.The output of new energy used the day-ahead forecast results.The power consumption boundary of EV cluster was obtained by Monte Carlo sampling,and two scheduling strategies for EV were given.The comprehensive operation cost including power purchase cost,network loss cost and new energy cost was taken as the optimization target.The second-order cone relaxation technique was used to transform the optimal scheduling model into the second-order cone mixed integer model.Based on the Yalmip/ Cplex solver,the output of each equipment in the distribution network in the next 24 hours was optimized.By comparing the the new energy consumption,power purchase and network loss of the active distribution network under the two strategies,the influence of the participation of electric vehicles on the active distribution network was analyzed.Finally,the intra-day rolling optimal scheduling model of active distribution network was constructed.Day-ahead scheduling results of energy storage equipment and electric vehicles were used as constraints,and intra-day rolling forecasting results were used for new energy output constraints.Voltage offset,system network loss and new energy absorption capacity were taken as optimization objectives,which were transformed into comprehensive optimization objectives by linear weighting method.Based on the Yalmip/ Cplex solver,the second-order cone-mixed integer model was used to schedule the operation of the distribution network in the next 4 hours,and the optimal scheduling results for the whole day were obtained by rolling every 1 hour.The scheduling results under the two strategies were analyzed,and the prediction errors of new energy output during day-ahead and intra-day scheduling were compared.The experimental results show that the day-ahead and intra-day scheduling model in this paper can effectively reduce the network loss,voltage offset and power purchase cost,while maintaining a high consumption rate of new energy.