A comparison of prediction and forecasting artificial intelligence models to estimate the future energy demand in a district heating system

Runge, Jason; Saloux, Étienne

doi:10.1016/j.energy.2023.126661

Cited by 38 publications

(8 citation statements)

References 25 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…LSTM model produced the best prediction performance. Besides, (Runge & Saloux, 2023) compared machine learning and deep learning techniques to predict heating demand over 6 h and 24 h ahead in Canada. The findings demonstrated that the LSTM and XGBoost produced good performance.…”

Section: Ai Methodsmentioning

confidence: 99%

See 1 more Smart Citation

Applications of Artificial Intelligence for Demand Forecasting

Nguyen

2023

OSCM: An Int. Journal

View full text Add to dashboard Cite

Artificial Intelligence (AI) has significantly contributed to the growth of various sectors and industries. One of the most significant contributions of AI applications is in forecasting. Consumer demand fluctuates faster than ever as a result of economic growth, advances in technology, and higher customer expectations, making forecasting future demand more difficult. Demand forecasting is a vital operation of supply chain management that aids in the best matching of supply and demand. Thus, improving demand forecasting accuracy is critical for companies and supply chains. Thanks to AI, businesses can accurately predict customer behavior. The study aims to provide a comprehensive review of how AI has been applied to forecast demand over the last decade. This research collects articles published between 2013 and 2023. According to the findings, AI is increasingly being used for demand forecasting in recent years. Energy and water demand forecasts receive the most attention. Long Short-Term Memory has gained prominence because of its advantages. Besides, this study will highlight the challenges of the adoption of AI. One of these challenges is selecting different reliable and suitable forecasting inputs for each AI method. This review will help supply chain managers and analysts select and implement suitable forecasting methods. Furthermore, this study will suggest some future research directions.

show abstract

Section: Ai Methodsmentioning

confidence: 99%

“…In addition, (Kavya et al, 2023) stated that energy forecasting influences a country's economic patterns and the profits of energy companies and related sectors. Furthermore, predicting future energy demand aids in optimizing and managing energy use, thus reducing greenhouse gas emissions (Kavya et al, 2023;Runge & Saloux, 2023).…”

Section: Areamentioning

confidence: 99%

Applications of Artificial Intelligence for Demand Forecasting

Nguyen

2023

OSCM: An Int. Journal

View full text Add to dashboard Cite

show abstract

“…It comes out that the quality of the forecasts is affected by the limitations imposed by the models' architectures (Rehman et al, 2022). Also, generating accurate forecasts with econometric models (especially autoregressive models) requires a minimum of thirty observations, whereas machine learning models requires need at least 10 5 data points to perform properly (Amber et al, 2018;Runge and Saloux, 2023;Shi et al, 2023). Grey models (GM), unlike the previous techniques, can make very accurate predictions with only four observations (Li and Zhang, 2021).…”

Section: Convolution Model To Electricity Demandmentioning

confidence: 99%

“…Also, generating accurate forecasts with econometric models (especially autoregressive models) requires a minimum of thirty observations, whereas machine learning models requires need at least 10 5 data points to perform properly (Amber et al. , 2018; Runge and Saloux, 2023; Shi et al. , 2023).…”

Section: Introductionmentioning

confidence: 99%

See 1 more Smart Citation

An optimal wavelet transform grey multivariate convolution model to forecast electricity demand: a novel approach

Sapnken,

Hamaidi,

Hamed

et al. 2023

View full text Add to dashboard Cite

PurposeFor some years now, Cameroon has seen a significant increase in its electricity demand, and this need is bound to grow within the next few years owing to the current economic growth and the ambitious projects underway. Therefore, one of the state's priorities is the mastery of electricity demand. In order to get there, it would be helpful to have reliable forecasting tools. This study proposes a novel version of the discrete grey multivariate convolution model (ODGMC(1,N)).Design/methodology/approachSpecifically, a linear corrective term is added to its structure, parameterisation is done in a way that is consistent to the modelling procedure and the cumulated forecasting function of ODGMC(1,N) is obtained through an iterative technique.FindingsResults show that ODGMC(1,N) is more stable and can extract the relationships between the system's input variables. To demonstrate and validate the superiority of ODGMC(1,N), a practical example drawn from the projection of electricity demand in Cameroon till 2030 is used. The findings reveal that the proposed model has a higher prediction precision, with 1.74% mean absolute percentage error and 132.16 root mean square error.Originality/valueThese interesting results are due to (1) the stability of ODGMC(1,N) resulting from a good adequacy between parameters estimation and their implementation, (2) the addition of a term that takes into account the linear impact of time t on the model's performance and (3) the removal of irrelevant information from input data by wavelet transform filtration. Thus, the suggested ODGMC is a robust predictive and monitoring tool for tracking the evolution of electricity needs.

show abstract