Calculating the energy consumption of electrocoagulation using a generalized structure group method of data handling integrated with a genetic algorithm and singular value decomposition

In this study, a hybrid data mining method for predicting energy consumption is proposed, namely the group method of data handling integrated with a genetic algorithm and singular value decomposition (GMDH-GA/SVD). As the randomness of renewable sources influences prediction methods, prediction mode...

Full description

Saved in:
Bibliographic Details
Published in:Clean technologies and environmental policy 2019-03, Vol.21 (2), p.379-393
Main Authors: Bonakdari, Hossein, Ebtehaj, Isa, Gharabaghi, Bahram, Vafaeifard, Mohsen, Akhbari, Azam
Format: Article
Language:English
Subjects:
Algorithms
Data mining
Data processing
Decomposition
Earth and Environmental Science
Electrocoagulation
Energy consumption
Energy development
Environment
Environmental Economics
Environmental Engineering/Biotechnology
Environmental policy
Genetic algorithms
Group method of data handling
Industrial and Production Engineering
Industrial Chemistry/Chemical Engineering
Multiple regression models
Original Paper
Parameter sensitivity
Prediction models
Process parameters
Randomness
Regression analysis
Sensitivity analysis
Singular value decomposition
Sustainable Development
Uncertainty analysis
Citations: Items that this one cites
Items that cite this one
Online Access:Get full text
Tags: Add Tag
No Tags, Be the first to tag this record!
Description
Summary:In this study, a hybrid data mining method for predicting energy consumption is proposed, namely the group method of data handling integrated with a genetic algorithm and singular value decomposition (GMDH-GA/SVD). As the randomness of renewable sources influences prediction methods, prediction model improvements are necessary for further development. Thus, GMDH-GA/SVD is introduced to model energy consumption as the primary criterion for process evaluation in finding the optimum condition to achieve the least energy consumption process. The parameters include the initial pH, the initial dye concentration, the applied voltage, the initial electrolyte concentration and the treatment time. The uncertainty analysis is applied to survey the quantitative performance of the new proposed model compared to existing popular reduced quadratic multiple regression models and two recently published models in the form of a Taylor diagram, indicating the proposed model is the most accurate. Moreover, partial derivative sensitivity analysis was done on the key parameters in the new model to provide insight into the calibration process of the new model. Graphical abstract
ISSN:1618-954X
1618-9558
DOI:10.1007/s10098-018-1642-z