Modeling the Throughput of Horizontal Shaft Impact Crushers Using Regression Analyses, Artificial Neural Networks and Multivariate Adaptive Regression Spline

Köken E.

Afyon Kocatepe Üniversitesi Fen ve Mühendislik Bilimleri Dergisi, cilt.22, sa.5, ss.1193-1203, 2022 (Hakemli Dergi)

  • Yayın Türü: Makale / Tam Makale
  • Cilt numarası: 22 Sayı: 5
  • Basım Tarihi: 2022
  • Doi Numarası: 10.35414/akufemubid.1116702
  • Dergi Adı: Afyon Kocatepe Üniversitesi Fen ve Mühendislik Bilimleri Dergisi
  • Derginin Tarandığı İndeksler: TR DİZİN (ULAKBİM)
  • Sayfa Sayıları: ss.1193-1203
  • Abdullah Gül Üniversitesi Adresli: Evet

Özet

In this study, the throughput (Q) of horizontal shaft impact (HSI) crushers was investigated using regression analyses, artificial neural networks (ANN) and multivariate adaptive regression spline (MARS). For this purpose, 32 different HSI-type crushers, which operated in the secondary crushing processes of various rock quarries in Turkey, were considered. Various quantitative data (i.e., rotor width (Rw), rotor diameter (Rd), rotor speed (Vr), characterized feed size (d80), operating energy (Oe), and Los Angeles abrasion value (LAAV) of the crushed stone) were collected from each crushing-screening plant. Linear and nonlinear regression analyses were first conducted using the above-mentioned collected data. Then, different ANN and MARS analyses were carried out to estimate the Q of these crushers. As a result, strong predictive models were developed to estimate the Q of HSI-type crushers. The correlation of determination (R2) of the proposed models (M6‒M10) ranged from 0.91 to 0.98, indicating the relative success of the established models. Therefore, the proposed models can reliably be used to estimate the Q of investigated HSI-type crushers. Nevertheless, the number of case studies should be increased to investigate other factors affecting the Q of HSI-type crushers.