Research On Thermal Error Compensation Method Of Heavy Duty Machine Tools Based On Distributed Multi-measurement Sensors

With the development of industrial modernization,China has entered the era of Industry 4.0.As a high-end precision manufacturing equipment in industrial construction,heavy duty machine tools require higher and higher accuracy.Most research shows that thermal error is the main factor of machining accuracy for heavy duty machine tools and how to reduce the thermal error has become an urgent problem for the improvement of machining accuracy.Due to heavy duty machine tools are large volume,long stroke and heavy load,effectively monitoring for temperature field of heavy-duty machine tools becomes the first big problem.In addition,based on few temperature sensors,the traditional modeling methods of thermal error are hard to achieve high accuracy because the wide range of heat sources of heavy duty machine tools.Another problem is how to smoothly compensate during the actual machining process after modeling.In this paper,the heavy duty gantry drilling machine ZK5540 A is the main research object.Based on the distributed monitoring data of temperature field for ZK5540 A,an ahead predictive modeling method using combined models and a smooth compensation method for thermal error are proposed.The main research work of this paper show as follows:(1)Research on temperature field monitoring of heavy duty machine tools based on distributed multi-measuring sensors.Based on the analysis of the structural characteristics and heat source of the heavy duty gantry drilling machine ZK5540 A,distributed FBG sensors network is used to monitor the temperature field of key parts for ZK5540 A.Besides,experiments are designed to verify the effectiveness of distributed multi-measuring sensors and analyze the temperature field characteristics of ZK5540 A in factory.(2)Research on ahead predictive modeling method of thermal error using combined models.Based on the temperature field data monitoring by distributed multi-measuring sensors,an ahead predictive modeling method using combined models is proposed.This model consists of two part.The convolution neural network is used as the front part for thermal error modeling to predict current thermal error from current temperature field.The long short term memory neural network is used as post part for ahead prediction of thermal error to compute the next a few time steps thermal error from current thermal error.Different experiments of spindle speed are used to verify the effectiveness of the combined model.Besides,effective range of time steps ahead would also be discussed through experiments.(3)Research on smooth compensation method and compensation system for thermal error.By analysis and comparison of various compensation method,an appropriate method is selected.Based on the chosen method,compensation system for thermal error is designed.The PLC ladder diagram modification and sample G code for the compensation system would be introduced in detail.With the experiments of different compensation processing,the effectiveness and smoothness of ahead predictive model would be verified.