Prediction of grinding behavior of low-grade coal based on its moisture loss by neural networks

ALTINER M., KUVVETLİ Y.

ENERGY SOURCES PART A-RECOVERY UTILIZATION AND ENVIRONMENTAL EFFECTS, cilt.39, sa.12, ss.1250-1257, 2017 (SCI-Expanded) identifier identifier

  • Yayın Türü: Makale / Tam Makale
  • Cilt numarası: 39 Sayı: 12
  • Basım Tarihi: 2017
  • Doi Numarası: 10.1080/15567036.2017.1320692
  • Dergi Adı: ENERGY SOURCES PART A-RECOVERY UTILIZATION AND ENVIRONMENTAL EFFECTS
  • Derginin Tarandığı İndeksler: Science Citation Index Expanded (SCI-EXPANDED), Scopus
  • Sayfa Sayıları: ss.1250-1257
  • Anahtar Kelimeler: Artificial neural network approach, Bayesian regularization learning technique, coal grindability, impact strength index, Levenberg-Marquardt learning technique, moisture, IMPACT STRENGTH INDEX, XIMENG LIGNITE, GRINDABILITY
  • Çukurova Üniversitesi Adresli: Evet

Özet

In this article, it was aimed to determine the influence of moisture amount on the grindability of Afsin-Elbistan low-grade coal using impact strength index (ISI) and hardgrove grindability index (HGI) tests. For this purpose, the sample was dried at a temperature range of 60 degrees C-150 degrees C for various times (80-240 min). A drying rate was further determined for each experiment. ISI and HGI values of each sample varied between 25.56-90 and 25-120, respectively. The obtained results show that a decrease of moisture in each sample led to an increase its grindability. In addition, artificial neural network (ANN) approach with two different learning techniques (Levenberg-Marquardt "LM" and Bayesian regularization "BR") was carried out to predict the HGI of Afsin-Elbistan coal. Three input parameters (moisture amount, ISI, and drying rate) obtained from the experiments were used for predicting HGI. LM learning algorithm gave a more satisfactory prediction (R-2 = 0.92, overall) compared to another technique.