Research On Trend Prediction Method Of Electricity Sales

Power market is an integral part of power grid planning.It includes the sale of electricity,electricity price,wholesale,commission loss and other content,including the sale of electricity is the basis of power grid planning work.In a certain period of time,the size of the sale power determines the management level of the power supply enterprise and the development of the national economy.Therefore,the prediction of the trend of the sale of electricity is the prerequisite for the marketing decision of the power supply enterprise.There are many ways to forecast electricity sales,each of which has its own scope and data set.Different regions of the electricity sales by the season,the proportion of the three major industries and other factors,often showing a different trend.Based on the modeling of Jiamusi area,the paper analyzes and forecasts the different forecasting models of Jiamusi area,and combines the characteristics of different single forecasting models to select the combined optimization model which is suitable for the sales data of Jiamusi area.(1)The ARIMA model and the grey GM(1,1)model are respectively carried out for residual and particle swarm optimization,based on factors such as seasonality,boundary value and uncertainty of the quantity of electricity sold in different industries.According to the different optimization model,the model is predicted and analyzed.(2)The optimized ARIMA model can better predict the change trend of electricity but there is large error in the prediction value and the actual value,the optimized GM(1,1)to solve the boundary value problem but error exists in the forecast trend lead to two models to predict the value of a da is less than ideal.Therefore,the two models are further improved by means of simple average,weighted average and variable weight.(3)In order to meet the requirements of practical work,through the experiment based on PSO-ARIMA-GM combinatorial optimization model prediction accuracy is higher,the key of the model is based on the dynamic variable weight method to determine the ARIMA model and grey GM(1,1)model of weight coefficient,PSO-ARIMA-GM combined forecasting optimization model to solve the single model on the part of the forecast accuracy is missing,and improve the overall prediction accuracy.Finally,the weight coefficient of PSO-ARIMA-GM combinatorial optimization model was determined by dynamic variableweight method,and PSO-ARIMA-GM combinatorial optimization model was used to predict the sales volume of jiamusi region.