Research on Short-term Power Load Forecasting in Distribution Station Area and Adjustable Load Participating in Demand-side Cluster Control

In order to effectively realize the demand-side cluster control of the adjustable load, the forecast of the daily load curve in the distribution station area needs to be accurately predicted. So, the time series of electricity load and historical maximum temperature, minimum temperature, average temperature and relative humidity are coupled as influencing factors when power load forecasting in this paper. Then, for the typical single-peak scenarios of the load curve, the BP neural network algorithm is used to calculate the load forecasting curve under the two conditions of whether or not to consider meteorological factors, and the regression curve of the load forecasting results in this two cases is further drawn. Through comparative analysis, it can be seen that when multiple meteorological factors are considered, the coefficient of determination R2 between the load curve forecast results and actual curves can reach over 99%, in other words, the load forecast results are accurate and effective. Furthermore, the cluster topology and control method of regional adjustable load is designed for power system peak shaving. Finally, the obtained load forecasting curve is effectively applied to the demand-side response cluster dispatch, and a simulation study of cluster adjustable loads participation in power system auxiliary peak shaving is carried out.