Research On Forecasting Method Of Electricity Price In Electricity Market

Electricity price is the fulcrum of the electricity market,electricity price has the characteristics of volKatility and uncertainty,they have injected vitality into the electricity market and have also increased the difficulty of forecasting electricity prices.With the steady advancement of domestic electricity marketization,electricity price forecasting is particularly important for power companies and power demand customers.Accurate forecasting of electricity prices is conducive to achieving a win-win situation for both parties in the market.Power demand customers can formulate reasonable power plans based on predicted electricity prices,it can help reduce the electricity bills and avoid use electricity during peak hours,at the same time,it can reduce the cost of power supply for power suppliers.In recent years,governments of various countries have vigorously promoted the construction of electricity market,attracting more and more scholars to participate in the research of electricity price forecasting,which improves the prediction accuracy of electricity prices to some extent,however,electricity price is affected by many factors,it is difficult to predict it,the current electricity price prediction accuracy is far from meeting the requirements of power companies and power demand customers.Therefore,it is necessary to study the electricity price forecast.Electricity prices have two characteristics:(1)nonlinearity and volatility(2)periodicity and mean recovery.For the features,the models are presented separately.By analyzing the electricity price of PJM electricity market,this paper finds that the price of electricity has periodicity and volatility,which can improve the prediction accuracy of electricity price.However,the interference data of electricity price will seriously affect the accuracy of electricity price prediction.Therefore,this paper proposes an combined algorithm EMD algorithm and wavelet threshold denoising algorithm model for pre-processing of electricity price data,this model can better preserve the original characteristics of electricity price while denoising.Electricity price has the characteristics of nonlinearity and volatility.The neural network algorithm has the ability to approximate nonlinear functions,so this paper proposes theGDBA-LSTM electricity price prediction model.The accuracy and effectiveness of the bat algorithm are far better than the traditional optimization algorithm,but original bat algorithm has the disadvantage of slow convergence at a later stage and easy to fall into local extremum,this paper adds gradient descent optimization as a noise term to increase the probability of the bat algorithm jumping out of the extremum.The improved bat algorithm can speed up the convergence and improve the prediction accuracy.Electricity price has the characteristics of periodicity and mean recovery,which is a typical time series.therefore,this paper proposes a GBE-ARIMA electricity price prediction model.The ARIMA model is used to predict the linear part of the electricity price,and the Elman neural network that is optimized by bat algorithm to predict the nonlinear residual of the electricity price.The experimental results show that the prediction accuracy of the above model is higher than that of the single ARIMA model.A single forecasting model always has its own advantages and disadvantages.In addition,the single model adapts to the data in a narrower range,which greatly increases the forecasting risk in practical applications.Therefore,this paper proposes a combined model that calculates the weight of each model through the comprehensive error indicator.This paper combine GDBA-LSTM and GBE-ARIMA by the above combined model,the prediction accuracy is significantly improved,and the reliability of the prediction model is also increased in principle.The prediction model proposed in this paper is more accurate.Electricity price can be better predicted,apart from the violent fluctuation caused by accidental factors.Precise forecasting of electricity prices can help power market participants make more informed decisions in a competitive and uncertain environment.This model is also applicable to other industries with similar characteristics,such as the forecast of the number of express parcels.