Adaptive neuro-fuzzy inference system-based prediction model for Malaysia’s overall energy consumption.

• Main Authors: Syamsul Bahri, Nurlaila, Mohd. Ali, Nur Syazwani, Jamaluddin, Khairulnadzmi, Hamzah, Khaidzir, Zainal, Jasman, Sazali, Muhammad Arif, Sarkawi, Muhammad Syahir, Basri, Nor Afifah, Mohd. Sies, Mohsin, Md. Rashid, Nahrul Khair Alang
• Format: Article
• Language: English
• Published: SANDKRS sdn bhd. 2023
• Subjects: TP Chemical technology
• Online Access: http://eprints.utm.my/108577/1/NurlailaSyamsulBahri2023_AdaptiveNeuroFuzzyInferenceSystemBased.pdf; http://eprints.utm.my/108577/; http://dx.doi.org/10.20967/jcscm.2023.02.005

Malaysia is one of the developing countries in South- east Asia that showed a rising in energy consumption every year. In this paper, three predictive models on total energy consumption are constructed using the Adaptive Neuro-Fuzzy Inference System (ANFIS). Three major steps are proposed to determine the predic- tive models using ANFIS, which are data extraction, construction of ANFIS and comparison of predictive and actual predictive models. In data extraction, yearly energy consumption, growth of popula- tions and GDP are determined. Next, the construction of ANFIS involved the normalization of data and MATLAB as a simulation to stimulate the predictive model. A comparison between predictive and actual predictive models is included to justify the correctness of the model. To construct the most appropriate prediction model, three models based on two input-partitioning methods—grid parti- tioning with two layers of the Gaussian membership function and subtractive clustering with radii of 0.6 and 0.7—have been chosen and compared. Three statistical methods, including the correlation coefficient, mean absolute error (MAE), and root means square er- ror (RSME), were used to assess the ANFIS model’s performance. The findings indicated that the RMSE values are 0.0601, 0.1591 and 0.0860, respectively, whereas the MAE values are 0.0560, 0.1480 and 0.4386. Additionally, Model 1, which represents the subtractive clustering of 0.6 radii, has a correlation coefficient that is close to 1, making it the most appropriate model for this study’s prediction of energy consumption through the year 2029. The ability to estimate future energy use is crucial for ensuring that there is always enough energy available to meet demand and promote sustainability.