Shaping a greener future: Energy auditing and conservation in Maldives utilizing augmented physics informed neural network-Addax Optimization approach

View at publisher Summarise with VIRAT
Authors: Sivakumar Thankaraj Ambujam, Salavutheen Noortheen, M. Mary Linda, Zahid Jameel
Cited by: 0

Summary

No summary yet. VIRAT can read this paper and write a plain-language summary of its aim, methods, key findings and limitations.

Summarise with VIRAT Free to try with your Science Catalogs account.

Abstract

Energy is essential for the world to continue evolving, growing, and existing. There are significant environmental challenges as a result of the rising energy dema5nd. Energy consumption and economic development are directly related, hence effective energy management techniques are required. Energy auditing and conservation in buildings of the Maldives focus on assessing energy usage and implementing strategies to reduce consumption, thereby enhancing sustainability in the face of growing demand and environmental concerns. However, challenges such as low prediction accuracy in energy consumption forecasts, limited access to advanced technologies, and the high initial investment required for energy-efficient solutions can hinder widespread adoption. Additionally, the limited availability of reliable data, difficulties in implementing cost-effective solutions, and a lack of technical expertise and infrastructure in the region can further impede the effective deployment of energy-saving technologies. To overcome these drawbacks, this manuscript proposes energy auditing and conservation in buildings of Maldives using hybrid approach. The data are collected from the households of Male and Hulhumale city dataset. Afterward, the data are fed to pre-processing. In pre-processing segment removes the missing values and normalizes the data utilizing Maximum Correntropy Quaternion Kalman Filter (MCQKF). The pre-processed output was fed to Augmented Physics Informed Neural Network (APINN) for prediction, which predicts energy consumption. The weight parameter of APINN is optimized using Addax Optimization (AO). The proposed method APINN-AO is implemented in MATLAB and compared with different existing techniques such as Fuzzy Logic Controller (FLC), Back-Propagation Neural Network (BPNN), Convolutional Neural Network (CNN), Encoder-Decoder Recurrent Neural Network (EDRNN), and Genetic Algorithm-Neural Network (GA-NN). The proposed APINN-AO technique achieves a Root Mean Squared Error (RMSE) of below 1.0, Mean Squared Error (MSE) of 5.1, and Coefficient of Determination (R 2 ) value of 0.76, significantly outperforming existing methods across these metrics. This highlights the technique’s effectiveness in minimizing prediction errors and enhancing the accuracy of energy consumption forecasting, making it a robust solution for energy auditing and conservation in buildings of the Maldives.

Fields of Study & Subjects

Building Energy and Comfort Optimization Energy Load and Power Forecasting Smart Grid Energy Management

Publication details

Journal / source
CRC Press eBooks
Pages
238–251
Publication date
25 March 2026
Language
EN
Document type
Book chapter