Surface Quality Evolution Model and Consistency Control Method of Large Shaft Multi-Pass Grinding

Abstract

Large shaft usually achieves high surface quality through multi-pass grinding in practice. Common surface quality indexes include surface roughness and glossiness, which are not only required numerically, but also require high consistency of distribution along the whole shaft. In multi-pass grinding, these two indexes are affected by the process parameters and the surface quality of the previous grinding pass, which leads to the difficulty of modeling. In addition, due to the uneven distribution of actual grinding depth, the surface quality along the whole shaft is usually inconsistent, resulting in the need for multiple spark-out grinding passes to ensure consistency. In this study, the surface quality evolution models for surface roughness and glossiness based on Elman neural network are developed, which build regressions between process parameters, surface quality indexes of the previous grinding pass, and surface quality indexes of the current grinding pass. Moreover, a consistency control method of surface quality is proposed by adjusting the actual grinding depth within the dimensional accuracy tolerance range at the rough grinding stage. Experimental results show that the surface roughness and glossiness prediction errors of the surface quality evolution models are only 5.5% and 5.1%. The consistency control method guarantees the consistency of surface quality, reduces the grinding passes, and increases the grinding efficiency.

Exploring New Parameters to Advance Surface Roughness Prediction in Grinding Processes for the Enhancement of Automated Machining

“…In [19], the authors introduced a technique employing the radial basis function (RBF) to tackle the issue of unevenly distributed abrasive particles in belt grinding procedures. Likewise, ANNs have been employed for surface roughness prediction in studies [20][21][22][23][24][25][26][27][28], yet cooling methods [29,30] and dressing parameters [31] have not been treated as distinct variables in the prediction algorithms. Additionally, it is important for the dataset size to be sufficiently large, which has not been consistently observed in the literature.…”