Research On Real-time Detection And Compensation Method For Thermal Error Of Dry-cut Hobbing Machine

Gears are the core and key parts of industrial manufacturing.High-speed dry-cut hobbing machines are also important equipment for the production of small modulus and high-precision gears.Their processing accuracy plays a very important role in the development of China's industrial modernization process.During the processing of high-speed dry-cutting gear hobbing machine,because it uses high-pressure air instead of cutting fluid cooling to dissipate heat,and the thermal conductivity of air is poor,this inevitably leads to the large amount of heat generated during the cutting process cannot be spread to In the surrounding environment,the main components of the machine tool are affected by heat conduction and heat radiation,resulting in an uneven temperature gradient distribution,which leads to irregular thermal deformation.Thermal deformation will change the distance between the hob of the hobbing machine and the tooth blank in the x-axis direction,which affects the amount of cutting when the hob cuts the teeth and leads to thermal errors,which reduces the accuracy of gear processing.After analyzing the principles of generation and development of the thermal error of the dry-cut hobbing machine,the paper first proposed a direct detection method that can be detected in time according to the thermal deformation generated during the gear hobbing mechanism,and based on the direct detection method a set of compensation devices to compensate the detected thermal error.However,considering that the direct detection method is to install a specific detection device on the machine tool,and needs to detect the distance between the hob tool holder and the workpiece at a specific detection position,and the detection link will inevitably disrupt the production cycle and reduce production efficiency.At the same time,installing a detection device on the bed will also increase the cost of the machine tool and increase the difficulty of processing.Therefore,in view of the shortcomings of the direct detection method,we combined the neural network model,by establishing the mapping relationship between the machine tool characterization signal and the thermal deformation state during the tooth making process detected by the direct detection method,In the subsequent processing process,the machine tool characterization signals are collected in real time and processed,and then input to the trained mathematical model to calculate the corresponding thermal deformation state,and then make a compensation decision.In this regard,we took a certain type of machine as the research object,and researched the real-time monitoring and compensation method of the thermal error during the processing of machine through the research ideas of signal acquisition-feature extraction-model establishment-design compensation system.The main research contents are as follows:(1)By analyzing the thermal error generation and development principle of high-speed machine,combined with the direct detection method of thermal deformation of machine,the signal-time dimension data and the thermal deformation-time dimension data are designed during the gear manufacturing process.(2)After cleaning the machine tool characterization data,use the time domain,frequency domain and time-frequency joint domain analysis methods to process the cleaned signal data to obtain 36 feature data sets including temperature data that can characterize the original signal data.(3)Analyze the characteristics of the traditional BP neural network,combine the feature data set and the thermal deformation state to establish the BP neural network model of the thermal error of the dry-cutting hobbing machine,and the accuracy rate of the model is 82.39% through the test data set.(4)Analyze the two popular deep learning models of the current popular deep automatic encoder and convolutional neural network.The former combines the feature data set,and the latter combines the original signal feature map composed of current time domain & temperature signal to establish the corresponding dry-cut hobbing machine.The thermal error monitors the deep learning model in real time,and it is obtained by comparing the results.In this study,the accuracy of the CNN model is higher than the DAE model.(5)Combined with the CNN model of the thermal error monitoring of the dry-cutting hobbing machine,the upper computer was written in C #,and a thermal error compensation system was designed.This system can input the thermal error detection model and make compensation decision through real-time processing of the machine tool characterization signal during the gear making process,and write the compensation value into the CNC system through the FOCAS program package of FANUC CNC system and modify the processing parameters to achieve thermal error compensation.