A novel method for workpiece deformation prediction by amending initial residual stress based on SVR-GA

High-precision manufactured thin-walled pure copper components are widely adopted in precision physics experiments, which require workpieces with extremely high machining accuracy. Double-sided lapping is an ultra-precision machining method for obtaining high-precision surfaces. However, during double-sided lapping, the residual stress of the components tends to cause deformation, which affects the machining accuracy of the workpiece. Therefore, a model to predict workpiece deformation derived from residual stress in actual manufacturing should be established. To improve the accuracy of the prediction model, a novel method for predicting workpiece deformation by amending the initial residual stress slightly based on the support vector regression (SVR) and genetic algorithm (GA) is proposed. Firstly, a finite element method model is established for double-sided lapping to understand the deformation process. Subsequently, the SVR model is utilized to construct the relationship between residual stress and deformation. Next, the GA is used to determine the best residual stress adjustment value based on the actual deformation of the workpiece. Finally, the method is validated via double-sided lapping experiments.