Research On Wear Detection Method Of TBM Disc Cutter Based On Pulse Eddy Current

When a full-section rock tunnel boring machine（TBM）is tunneling,the disc cutter（hereinafter referred to as cutter）mounted on the cutter plate is in direct contact with the rock and is subjected to huge torque and thrust,and the load conditions of TBM tunneling are complex.In recent years,with the promotion of TBMs to the"three highs"（high rock pressure,high rock strength and high quartz content）and other extreme formations,downtime due to cutter wear is bound to become more frequent.Therefore,the research of TBM operation condition monitoring technology,including cutter wear detection,has become an urgent problem in the field of TBM.Compared with the harmonic excitation eddy current detection technology,the pulse eddy current detection technology（PECT）has a wide spectrum,high current and other excitation characteristics,so the far-field detection distance is farther,and the impurities such as underground minerals penetration ability is stronger.At present,the relevant pulse eddy current detection technology has not been applied to the field of tool ring wear monitoring.For this reason,this thesis firstly proposes a PECT-based tool ring wear nondestructive detection method;subsequently,the structural parameters of the probe coil are optimized by combining finite element simulation and physical test;meanwhile,the ring wear characteristic signal is extracted,and the influence law of the probe lifting distance,metal medium,ring rotational speed and ring relative permeability on the ring wear characteristic signal is studied.In addition,further based on the machine learning method of BP neural network and SVM support vector machine,a multi-coil ring wear detection model is proposed to classify the common tool ring failure types.The main research work and conclusions are as follows.（1）Based on Maxwell electromagnetic theory,a probe-optimized finite element model for pulse eddy current detection of We wear tool coils was established in COMSOL environment,and parametric scan analysis was performed for the inner and outer radii and height of the excitation coil,respectively.The simulation results show that increasing the excitation coil height h₁ or decreasing the outer coil radius r₁,although it can help to improve the resolution of the excitation coil,diminishes the In order to take into account the requirements of both parties,it is recommended that r₁and r₂ be given as 35 and 18 mm,respectively.（2）On the basis of the obtained optimized parameters of the probe coil,a finite element detection model of the pulse eddy current of the worn tool ring with the radial wear amount d varying in a gradient is further established,and the influence law of the tool ring wear amount and the probe lifting distance on the output signal is analyzed,and the research results show that the voltage signals corresponding to different d are obviously distinguished in the early stage,so the integral value is taken as the characteristic quantity,and the function relationship between the tool ring wear amount and the characteristic quantity is established.The functional relationship equation between the amount of tool ring wear and the characteristic quantity is established.Further parametric scanning of the lifting distance reveals that the radial wear amount d can be up to 35mm when the lifting distance is in the range of 0～29mm.（3）Based on the aforementioned simulation analysis,a 19-in tool ring pulse eddy current inspection test bench was built and the pulse eddy current inspection test was carried out with different lifting distances of the probe,and the research results showed that the farthest lifting distance of the probe was 61mm,which was basically consistent with the simulation conclusion;the farthest distance was improved by about 70%compared with the eddy current inspection technology,which was much larger than the maximum radial wear of the tool ring in engineering.detection requirements.（4）combined with simulation and test means,further analysis of the metal medium,geological conditions and knife ring rotational speed on the pulse eddy current detection signal impact law,the results show that:the geological conditions of the saturated silt on the signal of the relatively largest impact,accounting for about air conditions characteristic amount;metal medium on the signal;knife ring rotational speed on the signal of 1%;overall,the pulse eddy current detection technology can In general,the pulse eddy current detection technology can effectively resist the interference of metal medium,geological conditions and tool ring rotation speed on the signal.（5）In order to meet the demand for the classification and detection of tool ring failure,a finite element model for tool ring wear detection with multiple response coils is further established,which is characterized by:in addition to the response coils arranged in the radial direction of the tool ring,there are two response coils symmetrically arranged on both sides of the tool edge;in addition,by changing the depth of partial wear e,the depth of chord wear d and the width of fracture w,the three types of failure of partial wear,chord wear and fracture are simulated.In addition,by varying the bias grinding depth e,chord grinding depth d and fracture width w,the different failure degrees of the three types of failure coils,namely,bias grinding,chord grinding and fracture,are obtained.Through simulation and analysis,the voltage signal curves corresponding to the different failure levels of the three types of failures are obtained,and the characteristic values of coil voltage,peak voltage and peak time are extracted.Further,a neural network and support vector machine based tool ring wear type classification detection method was proposed,and 70%of the data obtained from the simulation was used as the training set and the rest as the validation set to perform the sample training and validation detection,and its research results showed that the above two classification methods can complete the classification task and have high separation detection accuracy（more than 99%）.