Research On Automatic Maintenance Method Of Vertical Machining Center

The faults of vertical machining center can be divided into three categories: CNC system faults,electrical control system faults and mechanical system faults.Among them,CNC system faults and electrical control system faults can be diagnosed through the CNC system's own diagnostic functions.The diagnosis results provide maintenance suggestions for equipment maintenance personnel.However,there is no effective method for the evaluation and diagnosis of mechanical system failures with the highest failure rate.At present,when faced with this type of fault maintenance,enterprises mainly adopt the maintenance strategy of "ex-post maintenance" and "regular maintenance".These two maintenance methods have poor pertinence and low efficiency,and often have problems such as "excess maintenance" and "insufficient maintenance".In order to improve the maintenance support capability of the vertical machining center,this article takes the main drive system and feed shaft drive system of the Haas VF-2 vertical machining center as the object.This paper studies autonomous maintenance method for explicit and recessive failures,and build an autonomous maintenance decision support system.Focused on the information collection and data processing and fault diagnosis and health assessment to carry out in-depth research.The main research results are as follows:First,according to the failure type and failure mechanism of the mechanical transmission system in the vertical machining center,this paper proposes an autonomous maintenance method for explicit failures and recessive failures.Taking the main drive system and feed shaft drive system as the monitoring objects,this paper builds an autonomous maintenance decision support system that uses the offline model training end and the online evaluation and diagnosis end.Secondly,for the multi-source information in the mechanical transmission system,based on the LabVIEW platform,this paper uses vibration,temperature and current sensors to realize the collection of multi-source signals.Considering the characteristics of deep learning algorithms,this paper establishes the data processing process,which includes data denoising,data normalization and data set enhancement processing.Then,in view of the problems of traditional one-dimensional convolutional neural network algorithm in fault diagnosis and health assessment,this paper proposes a convolutional neural network algorithm with a multi-scale convolution architecture.This algorithm solves the problem of the lack of adaptability of convolutional neural networks when extracting multi-scale frequency features in mechanical transmission systems.Based on the multi-scale convolution architecture,this paper builds the fault diagnosis model and the health assessment model.Finally,taking the X-axis transmission system of the VF-2 vertical machining center as the test object,this paper conducted an autonomous maintenance experiment for explicit failures and recessive failures.Based on TensorFlow deep learning framework and multi-source data,offline training evaluation and online diagnosis evaluation experiments of the model were carried out.The experiment results verify the effectiveness of the proposed method.Overall,the autonomous maintenance method and strategy proposed in this paper solve the problems of "how to repair" and "when to repair" in the maintenance of vertical machining centers.This method changes passive maintenance to active maintenance,which effectively improves the maintenance efficiency.