Thermal Error Dynamic Modeling And Application For Machine Tool Based On Bayesian Network

The thermal error compensation of CNC machine tools is a common problem of preciseprocessing, and its difficulty lies in modeling and real-time compensation. This dissertationproposes a method of machine tool thermal error dynamic modeling during processing, basedon Bayesian network theory. Proposed an economical solution for real-time thermalerror compensation could compensate thermal error of machine tool at a lower economiccost, achieve domestic CNC machining to improve the accuracy.In this paper, considering the correlation of various factors, we analysis and optimizethe correlation of those factors according to experiment data, and found a dynamic modelof thermal error compensation of CNC machine tool based on Bayesian Network theory.Moreover, because of the self-learning feature of Bayesian network, the model can becontinuously optimized by updating dynamic coefficient, and reflect the changes ofprocessing condition. Finally, the feasibility and validation of this model are provedthrough the experiment we did, and proposed a microcontroller-based machinetool thermal error compensation control module. It’s really useful in the real-timethermal errors compensation in CNC machine tools.The main contents of this dissertation are shown follows：(1) According to the characteristics of the Bayesian network, analyzed the processfor dynamic thermal error modeling based on Bayesian networks. And then describethe thermal error modeling process based on BN specifically, deduced the forecastingformulas for thermal deformation, the model can be continuously optimized by updatingdynamic coefficient, and reflect the changes of processing condition.(2) Collected sample data for the thermal error compensation model. Developeda microcontroller-based temperature acquisition module, design experimental scheme,analyzed the sample data obtained for modeling.(3) Optimized temperature variables, the BN thermal machine error model isestablished with experimental data. The feasibility and advantages are verified by contrastof the LS thermal error model...