Improving the accuracy of TIF in bonnet polishing based on Gaussian process regression

A robotic polishing parameter optimization method considering time-varying wear

“…However, in the actual processing, the change of wear is dynamic and time-varying, it is difficult to obtain some information such as the surface topography of the polishing tool in real time to evaluate the degree of wear. Besides, these aforementioned theoretical models all regard the Preston coefficient k p as an independent constant with the velocity distribution and pressure distribution, which is inconsistent with the actual experiments [10] [11]. Therefore, although the theoretical model can accurately predict the material removal rate in a short period by comprehensively considering the influencing factors and certain experimental data, the prediction accuracy of the theoretical model will gradually decrease with the change of time and various variables.…”

“…Polishing time is calculated by dividing the distance by the feed rate. referring to the literature [11], the coupling relationship between polishing parameters has long-term smooth features and short-term periodic features, which can be characterized by setting a composite kernel function with SE(square exponential) kernel and PER(periodic) kernel.…”

“…In order to achieve better polishing quality, it was also necessary to determine a appropriate rotation speed. In order to verify the superiority of the material removal regression model incorporating prior knowledge proposed in this paper, the proposed model is compared with the GPR model proposed by [11], the support vector regression(SVR) model proposed by [12] and the theoretical model proposed by [20]. The test samples were also 40 data points randomly selected from other historical experimental groups.…”

“…Even so the article achieves the removal of uneven surfaces, it ignores the problem of tool wear. Feng et al [11] provided a method to improve the accuracy of the classic model in bonnet polishing applying Gaussian process regression. The article mainly focuses on the coupling relationship between Preston coefficient and polishing depth and rotation speed.…”

A material removal model for nonconstant-contact flexible grinding

“…To build the MR model, contact calculation is necessary. The most widely used methods of contact calculation include Winkler elastic foundation [11], Hertz contact theory [12][13][14] and the finite element analysis (FEA) [15][16][17]. Some simplifications are made in the above methods, including rigid workpieces, steady and constant contact and so on.…”