Research On Quantitative Identification Method Of Internal And External Broken Wires In Steel Wire Rope

As a key load-bearing component in engineering applications,steel wire rope works in harsh environments and complex working conditions.As the service life increases,various forms of damage are prone to occur,resulting in a reduction in its load-bearing capacity.It is necessary to conduct damage detection to the steel wire rope to eliminate potential safety hazards.Based on the principle of magnetic flux leakage detection and machine learning theory,this essay conducts quantitative research on the internal and external broken wire damage of steel wire ropes.This study includes the following contents:(1)Based on the principle of magnetic flux leakage detection,the theoretical analysis of the wire rope damage leakage magnetic field is carried out,the magnetic charge model is established,and the influence of the lift-off value and the change of the fracture width on the leakage magnetic field is analyzed.Establishing a three-dimensional model of the damaged wire rope and exciter,and using finite element simulation to analyze the change of the magnetic flux leakage generated by the number of broken wires,the width of the fracture,internal and external broken wires,the diameter of the wire rope and the distribution of damage leakage magnetic field in the circumferential direction.Analyzing the influence of the change of the air gap,magnetic pole spacing and lift-off value on the excitation effect and damage to the magnetic field leakage.Through theoretical analysis and finite element simulation,it provides a basis for the structure design of the excitation detector.(2)The steel rope safety inspection test platform is designed,which is mainly composed of the test bench base frame,motion control system,information feedback system,excitation detector and data acquisition system.In order to reduce the sensitivity of the magnetic flux leakage signal to the lift-off value,based on the analysis of the influence of the magnetic concentrating ring on the wire rope damage leakage magnetic field,the magnetic concentrating ring is installed in the excitation detector.Due to the different form of the axial component and the radial component of the magnetic flux leakage signal,the wire rope axial excitation detector with magnetic concentrating ring and the radial excitation detector based on the Hall array are designed respectively,designing the signal preprocessing circuit and the signal acquisition system based on Lab VIEW.In order to study the distribution of magnetic flux leakage in the circumferential direction of the damage location,a multi-channel signal acquisition circuit and acquisition program were designed for the radial excitation detector.Making damage specimens with steel rope diameters of 20,22,24 cm,and carrying out damage detection experiments on the steel rope safety detection test platform.Two kinds of excitation detectors are used to collect the wire rope damage magnetic flux leakage signal and analyze the change law.Verifying the correctness of the simulation model established in this essay.(3)Using a method for magnetic flux leakage signal denoising of steel wire rope based on double-density dual-tree complex wavelet transform,setting adaptive threshold,noise reduction is performed on the noise-added magnetic flux leakage signal calculated by the magnetic charge model and the steel wire rope magnetic flux leakage detection signal.the noise reduction effects of discrete wavelet,double-tree complex wavelet,double-density wavelet and the method in this essay are comparative analysis;after detrending,multiple time-frequency domain features are extracted as the feature set;the feature selection method based on the combination of inter-class distance and mutual information is proposed to screen out effective feature combinations,and use the last retained features as the optimal feature subset.(4)Based on the classification principles of BP neural network and ELM,two quantitative recognition models of wire rope damage based on BP neural network and ELM are constructed respectively,taking the optimal feature subset as the input of the two models,the damage quantitative identification of the wire ropes with 9 types of damages of single specification wire ropes and 27 types of damages of multi-specification wire ropes was performed,and the recognition results were compared and analyzed,the9 types of single specification wire rope and 27 types of multi-specification wire rope were quantitatively identified,and the recognition results were comparative analysis,the results show the two models have a higher damage recognition rate for single specification steel wire ropes,but a low damage recognition rate for multi-specification wire ropes,the reason for this result was analyzed.(5)In order to solve the problem of low damage recognition rate of multispecification wire rope,the classification principle of convolutional neural network is analyzed.Introducing five methods for generating two-dimensional RGB images from one-dimensional signals,and generating a total of five types of magnetic flux leakage feature map for steel wire rope,including wavelet feature maps,VMD feature maps,synchronous compressed wavelet transform feature maps,Wiener-Ville distribution feature maps,and sensor fusion feature maps.A transfer learning model based on Goog Le Net was constructed,and the model was used to quantitatively identify for steel wire rope damage.A quantitative recognition model of broken wire damage in steel wire rope based on two-dimensional convolutional neural network is proposed,and the optimal parameters of the convolutional layer,learning rate and batch size are selected for the network.Using different feature maps to train the convolutional neural network,obtain the wire rope damage recognition model,and analyzing its damage recognition effect.The effects of using migration learning model based on Goog Le Net,and the convolutional neural network proposed in this essay to identify wire rope internal and external broken wire damage were comparative analysis.The results show that t the convolutional neural network proposed in this essay has the highest recognition rate for wire rope internal and external broken wire damage,and the damage recognition rate for multi-specification wire rope is 98.4%.