Design And Implementation Of Thermal Error Compensator For CNC Machine Tools Based On Edge Computing

CNC machine tools are the cornerstone of precision and ultra-precision machining,and are responsible for providing high-end equipment for aerospace,marine,nuclear power and other fields.With the proposal and implementation of the "Made in China 2025" project,the manufacturing has an increasingly demand for machining precision for CNC machine tools.A large number of studies has shown that the thermal error caused by the thermal deformation is the most important factor affecting the machining accuracy of CNC machine tools.Therefore,it is of great theoretical and practical significance to effectively compensate the thermal error of the machine tool and improve the machining accuracy of the machine tool.The ZK5540 A heavy-duty CNC machine tool is the research object in this paper.The embedded thermal error compensator based on edge computing is designed and implemented.And a thermal error prediction model is established by using lightweight convolutional neural network.Finally,the compensator is tested and verified on the CNC machine.The main research work of the thesis is as follows:(1)According to the architecture of the edge computing,the overall scheme of the thermal error compensator is formulated.The compensator is divided into three parts: the edge network layer,the edge application layer and the edge-cloud network layer.As for edge network layer,two communication interfaces are designed to obtain machine tool’s temperature data and thermal error data.The NCUC-IP core is applied to design the FPGA-based bus communication module to interact with the CNC system.In the edge-cloud network layer,the network communication circuit and the MQTT-XML-based communication protocol are designed to realize two-way communication between the edge and the cloud platform.The ARM processor is used as main controller and the edge application development environment is built.The real-time preprocessing method based on variable threshold is studied for the outliers in raw data.Besides,the compensation control strategy based on the G-code embedded parameter method is proposed.(2)A thermal error modeling method based on lightweight convolutional neural network is studied.Firstly,aiming at the problem that traditional thermal error models relying on selection result of key temperature measurement points,the convolutional neural network is used to predict the thermal error.Then,in order to reduce the resource consumption during the running period so that the model can be trained and updated at the edge of network to quickly adapt to different machining conditions,the lightweight convolutional framework shufflenet is used to replace the standard convolution operation to reduce the calculation amount of the model.At the same time,a pruning method with progressive threshold is proposed to reduce the parameter amount of the fully connected layer.Finally,the accuracy and operational efficiency of the proposed lightweight modeling method are verified on the compensator.(3)The compensator is deployed on the ZK5540 A heavy-duty CNC machine equipped with HNC-848 C CNC system.The edge network and edge-cloud network communication function of the device are tested for a long time.The effectiveness and real-time performance of the data preprocessing method are verified by simulation experiments.Finally,experiments based on the control variable method is designed.The thermal error compensation function of the compensator is verified in different seasons under the milling machining conditions.