A data-driven machining errors recovery method for complex surfaces with limited measurement points

Accurately characterizing the machining errors of manufacturing workpieces requires dense measurement information, which is inefficient when the data are acquired with the trigger contact probes. To address this problem, this paper presents a Gaussian Processes (GP) based super resolution (SR) network. Specifically, a multi-scale SR network with dual attention is built to achieve the recovered results with more powerful feature expression. A specific kernel based GP is introduced to transform the scattered and noisy data into grid denoised features. And fractal Brown motion (fBm) is applied to synthesize simulated machining errors. The generalized neural model exploits the data to enable the network in learning to enrich features of sparse data, which dramatically reduces the sampling time and increases measurement efficiency. The effectiveness of the method is verified through a series of comparison study and this method can achieve higher performance with the same sampling points and less computing time.