Thermal error modeling of Mini/Micro LED mass transfer equipment based on support vector machine regression

Abstract

This study presents a machine learning-based approach employing Support Vector Regression (SVR) to predict and compensate for thermal errors affecting the precision of Mini/Micro chip transfer in mass transfer equipment. The correlation between the temperature variations in the linear motor drive system and thermally induced positioning errors is analyzed through experimental measurements. Key temperature measurement points are identified using K-means clustering and gray relational analysis, with the optimal number of clusters determined via the elbow method to minimize collinearity interference. In order to verify the validity and robustness of the proposed model, thermal characteristic experiments based on constant feed velocity and random velocity were carried out. The results show that SVR model has higher accuracy than MLR model and BP model. The findings indicate the potential of this tool to enhance the precision of positioning in high-precision manufacturing applications.