Predictive surface roughness model for robotic polishing considering initial surface quality

The polishing process is complex and influenced by various parameters, making the construction of predictive models for polishing quality a significant area of research. Existing models primarily focus on four parameters: contact force, rotational speed, feed rate, and sandpaper grit, while neglecting the impact of initial surface quality, resulting in limited accuracy and applicability. This paper proposes a method for constructing a surface roughness prediction model that considers initial surface quality, which consists of two parts: First, through experimental polishing tests on workpieces with various initial surface qualities, it was shown that the initial surface quality has a significant effect on the final polishing result; second, the initial surface quality is classified into three grades based on roughness values, and a prediction model for post-polishing surface roughness is constructed by integrating the initial surface quality and the four process parameters using response surface methodology. Finally, a series of polishing experiments with different parameter combinations obtained model prediction errors ranging from 3.40 to 11.44% (average 7.48%), verifying the practicality and generality of the proposed prediction model.