Litcius/Paper detail

Predicting Building Energy Consumption by Time Series Model Based on Machine Learning and Empirical Mode Decomposition

Dandan Liu, Qiangqiang Yang, Fang Yang

202015 citationsDOI

Abstract

In this paper, the prediction method for building energy consumption was discussed. The building energy consumption data was regarded as time series, which was usually nonlinear and non-stationary. The traditional time series analysis model has lower prediction accuracy for nonlinear and non-stationary time series. So the joint algorithm using support vector regression (SVR) and empirical mode decomposition (EMD) was applied. EMD method decomposed the non-stationary and nonlinear energy consumption time series into several Intrinsic Mode Functions (IMFs). And support vector regression (SVR) was used to predict the decomposed time series. The sum of every predicted subsequence was final forecasting result. The experimental results showed the prediction accuracy of the combination model is better than SVR forecasting model.

Topics & Concepts

Hilbert–Huang transformSupport vector machineSeries (stratigraphy)Time seriesNonlinear systemMode (computer interface)Computer scienceEnergy consumptionRegression analysisNonlinear regressionEnergy (signal processing)Artificial intelligenceAlgorithmMathematicsMachine learningStatisticsEngineeringPaleontologyBiologyElectrical engineeringPhysicsQuantum mechanicsOperating systemEnergy Load and Power ForecastingGrey System Theory ApplicationsAdvanced Algorithms and Applications
Predicting Building Energy Consumption by Time Series Model Based on Machine Learning and Empirical Mode Decomposition | Litcius