The Application of AI for the Spindle Dynamic Balancing and Thermal Displacement in Machine Tools

• Main Authors: CHUANG, REN-HAO, 莊人豪
• Other Authors: WANG, CHI-CHANG
• Format: Others
• Language: zh-TW
• Published: 2019
• Online Access: http://ndltd.ncl.edu.tw/handle/49z5zv

碩士 === 逢甲大學 === 機械與電腦輔助工程學系 === 107 === In this paper, the mechanical balance of artificial machine tool is used to predict and identify the dynamic balance of the machine tool spindle and predict the thermal displacement generated by the inner spindle of the simulated spindle. First of all, the dynamic balance first performs vibration measurement and explores its characteristic frequency through spectrum conversion, including the difference in the number of screws installed. Then through the measuring device and establish a database at different speeds and the number of screws, the impact on the accuracy can be explored through different pre-processing and learning networks. The final accuracy rate can reach 90% The second proposed to predict the thermal displacement of the inner spindle of the spindle to make up for the insufficiency of the traditional linear fitting. Using the characteristics of the neuron, the training error is much smaller than the linear fitting, and the least square of the thermal displacement under the fixed heat source and the variable heat source is predicted. The error is reduced to 2.16um and 1.96um respectively, which also breaks through the myth of the previous measurement point position. The closer to the measurement point of the heat source, the more accurate the training model is, and the temperature rise slope of the temperature point is used to model the neural network. As an important parameter, it helps to judge the correspondence between temperature and thermal displacement.