Deep learning based carbon emissions forecasting and renewable energy's impact quantification

Abstract

Global warming is one of the most challenging issues of the current era. Revolutions in industrial, Information and Communication Technology (ICT) sectors significantly contribute in increasing global warming. Green House Gases (GHG) emissions from industrial, transportation, power and other sectors cause environmental pollution, which results in climate degradation. Environmental experts are well aware of the disastrous consequences of excessive global warming; therefore, several decarbonization strategies are developed and practiced in the recent past. A widely practiced decarbonization strategy is replacing fossil fuels by Renewable Energy Sources (RES) in the power systems. Electricity consumers are also encouraged to shift their consumption load to low carbon emissions' time periods. For accomplishing this task, an estimation of future carbon emissions is required. In this paper, power system's carbon emissions are predicted accurately with the help of a novel and an efficient forecasting model. The proposed model comprises of Spearman Correlation Analysis (SCA) and Improved Shallow Denoising Autoencoder (ISDAE) based feature engineering. Forecasting is performed through an Improved Particle Swarm Optimization (IPSO) based Deep Neural Network (DNN) forecaster. In addition, a comprehensive quantification analysis is also presented in this paper. The impacts of RES integration level on the electricity price, consumption cost and Greenhouse Gases (GHG) emissions are quantified descriptively and graphically. Performance of the proposed forecasting model is evaluated by Normalized Root Mean Square Error (NRMSE), Mean Absolute Error (MAE) and Mean Square Error (MSE). Simulation results prove that the proposed model outperforms Support Vector Machine (SVM) and Multiple Linear Regression (MLR) based carbon emission forecasting models in terms of forecasting accuracy.

Enhanced prediction of stock markets using a novel deep learning model PLSTM-TAL in urbanized smart cities

“…For example, PCA, RF and embedded methods like LASSO and Ridge regression can be used for feature extraction along with CAE and the best performing method can be selected for the proposed deep model. Important features can also be selected based on Spearman Correlation Analysis (SCA) as in [52] .…”

Inteligencia Artificial - Chatgpt: un estudio bibliométrico y su aplicación en el caso de Gases de Efecto Invernadero

“…Por su parte Akhshik et al (2022), manifiestan que al probar varios algoritmos de IA y matrices de entrada se puede predecir rápidamente las emisiones de GEI para predicciones de ahorro de emisiones basadas en análisis de ciclo de vida. Mujeeb & Javaid (2023), usaron métodos basados en IA para estimar las emisiones nacionales de GEI hasta el año 2040, en Arabia Saudita. El hallazgo ayuda a los tomadores de decisiones a comprender los diversos factores socioeconómicos en la definición de los inventarios nacionales de GEI.…”

Ensemble Framework for Multistep Carbon Emission Prediction: An Improved Bi-LSTM Model Based on the Bayesian Optimization Algorithm and Two-Stage Decomposition

“…According to the literature, first, most studies have focused on using artificial intelligence (AI) models for annual carbon dioxide (CO₂) emission forecasting, while relatively few have explored daily CO₂ emission prediction (Huang et al, 2022;Jena et al, 2021;Jin, 2021;Mason et al, 2018). Second, the methods for optimizing AI model parameters are often quite singular; hence, the search for a more robust and effective optimization algorithm becomes particularly crucial (Aryai and Goldsworthy, 2023;Mujeeb and Javaid, 2023;Sun et al, 2017;Yu et al, 2023;Zhu et al, 2022). Third, data decomposition-reconstruction techniques have not yet been widely applied in this field.…”