Research On Fault Monitoring Of Warp Knitting Machine Based On Edge Computing

As an important textile equipment,condition monitoring and fault diagnosis of warp knitting machines is an important part of the realization of warp knitting digital workshop and intelligent operation and maintenance.It is of great significance to realize the fault detection of key mechanism components of warp knitting machine and to determine the type of warp knitting machine fault,in order to regulate the workshop production arrangement and reduce the equipment maintenance interval time.In order to accurately sense the working status of key components of warp knitting equipment,this paper proposes a framework structure for fault monitoring of warp knitting machine based on edge computing,and conducts signal monitoring of key components of warp knitting machine,such as yarn guide comb and electronic traverse system,and studies the physical signal data feature extraction method and fault detection algorithm of each working component The mapping relationship between the physical signal features and the working state of the components is established to realize the fault detection of the key mechanism of the warp knitting machine.The main contents are as follows.（1）A summary analysis of the faults occurring in each working mechanism of warp knitting machine is carried out,and the overall architecture of fault monitoring of warp knitting machine based on edge computing features and application architecture is designed based on the demand of fault monitoring of warp knitting machine.（2）For the possible faults of warp knitting machine guide comb,the vibration signals of the guide comb during the motion are collected,and the collected samples are extracted with time-domain features and wavelet packet energy features,and then the feature vector samples are normalized.Using the single-value classification feature of the Support Vector Domain Data Description（SVDD）algorithm,the SVDD model is trained using only the feature vectors of the guide comb in its normal state.The test results of the test samples show that the wavelet packet energy features effectively complement the equipment state information,and the constructed vibration signal multi-features and SVDD model accurately achieve state discrimination.（3）In the monitoring of the mechanical drive mechanism of the electronic traverse system of warp knitting machine,firstly,the modeling simulation of the working mechanism of the electronic traverse system was carried out based on the MATLAB Simulink environment,and the waveform characteristics of the servo motor output axis torque after changing with the transmission stiffness and damping were analyzed and observed.Then the servo motor output shaft torque signal is collected based on the experimental platform,including three different states of no abnormal transmission,loose coupling and worn ball screw,and the feature vectors of the torque signal are extracted and the diagnostic model is trained by using support vector machine（SVM）.The test experiment results verify that the fault diagnosis model based on the servo motor torque signal can identify the faults of some mechanical transmission components of the electronic traverse system.（4）Based on the above warp knitting machine fault monitoring framework design and fault detection method research,this paper carries out the design of edge computing application system from hardware equipment,software framework and communication method,using data acquisition card and industrial communication protocol Modbus to realize the signal collection of key components of warp knitting machine,and applying fault detection model at the edge end to evaluate the health status of parts,the designed edge computing monitoring The designed edge computing monitoring system provides a certain theoretical and technical basis for the realization of warp knitting machine fault monitoring.