Research on Optimization of Chemical Mechanical Polishing of Hard and Brittle Materials Based on Adaptive Neural Network

In order to explore the relationship between the surface quality of hard and brittle materials and the parameters of chemical mechanical polishing process, and then realize the purpose of improving the surface quality and processing speed of hard and brittle materials, this paper proposes a modeling method based on adaptive neural network to construct the surface roughness model and material removal rate model, respectively, and then conducts training and testing on them. The test results show that the maximum error of the hard and brittle material roughness model is 12.7%, the MAPE value is 5.22%, and the R2 value is 0.9984. The maximum error of the hard and brittle material removal rate model is 14.1%, the MAPE value is 5.31%, and the R2 value is 0.9896. Both models have good generalization ability and regression ability. Furthermore, both models have high prediction accuracy and can recognize the data in the test set more accurately.