Prediction of Abrasive Belt Wear Height for Screw Rotor Belt Grinding Based on BP Neural Network with Improved Skyhawk Algorithm

• Published in: International journal of precision engineering and manufacturing Vol. 26; no. 2; pp. 399 - 414
• Main Authors: Pan, Fei, Sun, Xingwei, Yang, Heran, Liu, Yin, Chen, Sirui, Zhao, Hongxun
• Format: Journal Article
• Language: English
• Published: Seoul Korean Society for Precision Engineering 01.02.2025; Springer Nature B.V
• Subjects: Abrasive belts; Abrasive wear; Abrasive wheels; Algorithms; Artificial intelligence; Artificial neural networks; Back propagation networks; Belt grinding; Cylinders; Engineering; Farm machinery; Grinding; Industrial and Production Engineering; Machine learning; Materials Science; Neural networks; Prediction models; Process parameters; Regular Paper; Rotors; Tensioning; Aquila optimizer; Neural network prediction; Sand belt wear; Grinding; Screw rotor
• ISSN: 2234-7593; 2005-4602
• DOI: 10.1007/s12541-024-01110-8

The influence of process parameters on the abrasive belt wear height in abrasive belt grinding screw rotors is studied in this paper. The independently developed special grinding device is used for the experiment. The improved Aquila Optimizer (IAO) algorithm is used to optimize the BP neural network, the experimental parameters and abrasive wear height data are input into the IAO-BP neural network model for training, then establish the prediction model of the average wear height of abrasive belt particles. The prediction samples and comparison samples are obtained by multi factor grinding experiments. The prediction accuracy is compared with ANN and GA-BP neural networks. The results show that the accuracy of the prediction model is better than that of ANN and GA-BP neural networks. The single factor prediction results of abrasive belt wear height show that the wear height of abrasive belt increases with the increase of driving wheel cylinder pressure and decreases with the increase of tension cylinder pressure. The wear height increases first and then decreases with the increase of the linear speed of the abrasive belt, and increases with the increase of the axial feed speed of the abrasive belt. The improved AO algorithm to optimize BP neural network prediction model can provide a theoretical basis for selecting process parameters of screw rotors in grinding belt. Abrasion of abrasive belt can be effectively alleviated by selecting higher linear speed and feed speed during grinding, appropriately reducing the pressure of positive cylinder and increasing the pressure of tensioning cylinder.