Short-Term Load Forecasting Method Based on ARIMA and LSTM

Shuo Chen, Rongheng Lin, Wei Zeng

202232 citationsDOI

Abstract

This paper explores the classic prediction algorithms such as Autoregressive Integrated Moving Average model (ARIMA) and Long Short-Term Memory Neural Network (LSTM), and combines the advantages of both, and proposes a short-term forecast method ARIMA-LSTM fusion model. This model computes the final predicted value by applying a linear correctio to the LSTM model error. Using ARIMA and LSTM as a comparison algorithm, train and predict the next day's load with ARIMA-LSTM. The root mean square error (RMSE), mean absolute percentage error (MAPE), and the worst relative error (WRE) were used to evaluate the performance of the proposed algorithm. After testing, the RMSE of the proposed model is 0.433, while ARIMA and LSTM are 0.461 and 0.445 respectively. ARIMA-LSTM also has better results and performs well on different types of power datasets.

Topics & Concepts

Autoregressive integrated moving averageMean squared errorMean absolute percentage errorTerm (time)Computer scienceArtificial neural networkApproximation errorBox–JenkinsAutoregressive modelTime seriesAlgorithmStatisticsArtificial intelligenceMathematicsMachine learningPhysicsQuantum mechanicsEnergy Load and Power ForecastingGrey System Theory ApplicationsGeoscience and Mining Technology