Estimates of hydroelectric energy generation in BRICS-T countries using a new hybrid model

ABSTRACT

Realistic energy production and consumption projections are needed by national governments that are developing policies to reduce greenhouse gas emissions and ensure energy supply security. In this study, four types of artificial neural network (ANN) models – including a back propagation (BP) model, ANN-BP, and three Rao algorithm models, namely ANN-Rao_1, ANN_Rao2 and ANN-Rao_3 were developed to determine hydroelectric power (HEP) generation projections for Brazil, Russia, India, China, South Africa, and Turkey (BRICS-T) countries. Gross domestic product (GDP), population, import, and export data were inputted as independent variables into the models. Data from the period of 1990‒2014 were used for ANN model development. Data from the period of 2015‒2020 were used as reference data to test the models’ predictive power. According to the error values calculated for the training and test sets, the ANN-Rao_3 model predicted HEP generation values for BRICS-T countries more accurately than the other models. The results obtained from this study predict that the HEP generation value of the BRICS-T countries will increase by 28.1% by 2040. According to the RMSE values obtained for the test datasets, compared to the BP algorithm, the Rao_3 algorithm increased ANN performance by 34.1% for Brazil, 24.1% for Russia, 3% for India, 13% for South Africa, and 12.1% for Turkey. The total HEP generation value of the BRICS-T countries in 2040 was projected to be 2815.69 TWh. Current BRICS-T countries’ HEP investments are not sufficient to achieve their renewable energy targets. In light of this research, relevant policy implications may be derived.

KEYWORDS:

• BRICS-T countries
• Hydropower generation
• Rao algorithms
• Neural networks

Highlights

• This study is associated with predicting hydropower generation in BRICS-T countries.

• GDP, population, import, and export data were used as predictor variables.

• ANN-Rao predicted the hydropower generation more accurately than classical ANN.

• Using the ANN-Rao model, the hydropower generation was forecasted until 2040.

Nomenclature

ANN	=	Artificial neural network
BP	=	Back propagation
HEP	=	Hydroelectric power
BRICS-T	=	Brazil, Russia, India, China, South Africa, and Turkey
GDP	=	Gross domestic product
IEA	=	International Energy Agency
G7	=	Group of seven
G20	=	Group of 20
OECD	=	Organisation for Economic Co-operation and Development
GM	=	Gray model
ARIMA	=	Auto-regressive integrated moving average
NGBM	=	Nonlinear gray Bernoulli model
ONGBM	=	Optimized nonlinear gray Bernoulli model
CO2	=	Carbon dioxide
DGGM	=	Data grouping-based gray model
SARIMA	=	Seasonal auto-regressive integrated moving average
GA	=	Genetic algorithm
ABC	=	Artificial bee colony
TLBO	=	Teaching-learning based optimization
GWO	=	Grey wolf optimizer
MSE	=	Mean square error
RE	=	relative error
RMSE	=	root mean square error
MAE	=	mean absolute error

Disclosure statement

No potential conflict of interest was reported by the author(s).

Data availability statement

The datasets used and analyzed during the current study can be provided on reasonable request.

Supplemental data

Supplemental data for this article can be accessed online at https://doi.org/10.1080/15567249.2024.2310094.