Evaluation of Ash and Coal Response to Hybrid Polymeric Nanoparticles in Flotation Process: Data Analysis Using Self-Learning Neural Network

Abstract

An artificial neural network was used to investigate the potential of using a novel polymer aid to produce clean coal concentrates in a fine clayey coal flotation process. Fine coal particles in the size range of +38-75 µm with 25.12 wt % of ash were floated in the presence of Al(OH)3-polyacrylamide nanoparticles. Five parameters; polymer dosage, pH, impeller speed, dispersant dosage and conditioning time were used as inputs in the simulation studies. Two network types (feedforward BP and cascade-forward BP) with three training algorithms (LM, BFG and GDX) and various numbers of neurons were designed and used to validate the experimentally observed qualitative and quantitative trends. The performance of each architecture design was evaluated by the correlation coefficient (R) and the mean square error (MSE). The two outputs that were used to evaluate the response of coal particles to the polymer used were the combustible recovery and the froth ash content. The cascade-forward network and Levenberg-Marquardt algorithm were selected as the optimal networks. The optimal ANN model showed a reasonable agreement in predicting the experimental data with correlation coefficient of 0.994 and 0.997 for training and testing datasets, respectively.

Department(s)

Mining Engineering

Keywords and Phrases

Aluminum compounds; Coal; Flotation; Mean square error; Nanoparticles; Neural networks; Polymers; Architecture designs; Ash removal; Clean coal; Combustible recovery; Correlation coefficient; Mean Square Error (MSE); Polymeric nanoparticles; Self-learning neural networks; Metal recovery; Al(OH)3-PAM; Artificial neural network

International Standard Serial Number (ISSN)

1939-2699; 1939-2702

Document Type

Article - Journal

Document Version

Citation

File Type

text

Language(s)

English

Rights

© 2017 Routledge, All rights reserved.

Publication Date

01 May 2019

Share

 
COinS