Application of artificial neural network and multiple linear regression in modeling nutrient recovery in vermicompost under different conditions

Ahmad Hosseinzadeh; Mansour Baziar; Hossein Alidadi; John L Zhou; Ali Altaee; Ali Asghar Najafpoor; Salman Jafarpour

doi:10.1016/j.biortech.2020.122926

Application of artificial neural network and multiple linear regression in modeling nutrient recovery in vermicompost under different conditions

Bioresour Technol. 2020 May:303:122926. doi: 10.1016/j.biortech.2020.122926. Epub 2020 Jan 29.

Authors

Ahmad Hosseinzadeh¹, Mansour Baziar², Hossein Alidadi³, John L Zhou⁴, Ali Altaee¹, Ali Asghar Najafpoor³, Salman Jafarpour³

Affiliations

¹ Centre for Green Technology, School of Civil and Environmental Engineering, University of Technology Sydney, NSW 2007, Australia.
² Ferdows School of Paramedical and Health, Birjand University of Medical Sciences, Birjand, Iran.
³ Social Determinants of Health Research Center, Department of Environmental Health Engineering, School of Health, Mashhad University of Medical Sciences, Mashhad, Iran.
⁴ Centre for Green Technology, School of Civil and Environmental Engineering, University of Technology Sydney, NSW 2007, Australia. Electronic address: Junliang.zhou@uts.edu.au.

PMID: 32035386
DOI: 10.1016/j.biortech.2020.122926

Abstract

Vermicomposting is one of the best technologies for nutrient recovery from solid waste. This study aims to assess the efficiency of Artificial Neural Network (ANN) and Multiple Linear Regression (MLR) models in predicting nutrient recovery from solid waste under different vermicompost treatments. Seven chemical and biological indices were studied as input variables to predict total nitrogen (TN) and total phosphorus (TP) recovery. The developed ANN and MLR models were compared by statistical analysis including R-squared (R²), Adjusted-R², Root Mean Square Error and Absolute Average Deviation. The results showed that vermicomposting increased TN and TP proportions in final products by 1.5 and 16 times. The ANN models provided better prediction for TN and TP with R² of 0.9983 and 0.9991 respectively, compared with MLR models with R² of 0.834 and 0.729. TN and C/N ratio were key factors for TP and TN prediction by ANN with percentages of 17.76 and 18.33.

Keywords: Modeling; Municipal solid waste; Nitrogen; Nutrient recovery; Phosphorus; Vermicompost.

MeSH terms

Linear Models
Multivariate Analysis
Neural Networks, Computer*
Nutrients
Phosphorus*

Substances

Phosphorus