| Summary: | 碩士 === 國立中興大學 === 機械工程學系所 === 107 === The “machine tool” is responsible for high precision, which is always a pursuing goal for manufacturing. The heat from geometrical error of the machine tool and during the cutting session is an important element affecting machining accuracy. The thermal error of spindle for the machine tool is an influential sources for machining accuracy. In recent years, thermal error compensation is highly valued and is adapted to elevate machining accuracy for a machine tool. The advantages for thermal error prediction contains rapid prediction, high precision prediction, and etc.
In this study continues previous research data for Spindle of Gantry Type Five Axis machine tool and spindle Z axis displacement to predict neural network and T-S fuzzy-neural network. The structure of neural network prediction adapts back propagation neural network to develop the model of thermal error of spindle Z axis and uses Takagi-Sugeno fuzzy-neural network to create the model of thermal error of the spindle Z axis. Then using the same temperature and the error of spindle Z axis to compare with the model of Multivariate Linear Regression developed from the previous study and then write the program simulation using Matlab.
The result of simulation shows that the prediction model of neural network decreases the thermal error of spindle Z axis from 130 to±5um up to 96% accuracy, while the T-S fuzzy-neural network decreases from 130 to ±5um up to 97% accuracy.
|
|---|
