- Research Article
- 10.1088/2631-8695/ae2835
Thermal error modeling of linear axis on surface mount technology machine based on zebra optimization algorithm optimized least squares support vector machine
- Dec 16, 2025
- Engineering Research Express
- Yang Li + 6 more +6
Abstract As a high-precision equipment for solid-phase crystallization, surface mount technology (SMT) machine, whose accuracy has an important impact on the packaging of semiconductor chips. Due to the ultra-high speed reciprocating characteristics of the SMT machine, thermal error (TE) is the main influence factor. To reduce influence of TE on machining accuracy of SMT machine, a TE modeling method based on zebra optimization algorithm optimized least squares support vector machine (ZOA-LSSVM) is proposed. To start with, Canopy clustering algorithm is used to perform "coarse" clustering on the collected temperature variables to give the number of clusters and the initial cluster center. K-means clustering algorithm is used to cluster temperatures based on results provided by Canopy. Using entropy, an indicator of cluster validity, selecting the temperature point with the highest entropy value from each group as the temperature-sensitivity point(TSP), which can reduce the correlation interference between temperature measurement points. Secondly, with selected TSP as the input, the least squares support vector machine (LSSVM) prediction model is constructed, where zebra optimization algorithm (ZOA) is introduced to optimize penalty parameter γ and kernel function σ, and ZOA-LSSVM TE prediction model is established. Finally, taking X-axis of SMT machine as the object, the measurement and verification experiments for the machine tool's linear axis were designed.
Read more