Research On Dynamic Characteristics And Fault Diagnosis Of Shield Main Bearing With Roller Local Fault

Tunnel Boring Machine is widely used in capital construction projects like subway construction and highway tunnel due to the advantages of high efficiency,economic benefits and operational safety.As the “joint” of main drive system of cutter head,the main bearing in shield main is the most critical component in cutter head drive system.Its operation status and service life are related to the shield construction process.The main bearing is suffering compound loads from cutter head,and there are impurities intrusion and poor lubrication during the operation,which lead to some damages of the main bearing.The damage investigation of the main bearing that carried out in the early stage shows that the damage forms of the main bearing are diverse and the mechanism is complex.Once the main bearing is damaged during service,it will cause the shield equipment to deviate from the normal state and cause system failure,resulting in significant economic losses and negative impacts.Therefore,exploring the dynamic characteristics of the shield main bearing and assessing the state of the main bearing play a crucial role in safe operation of the shield machine.However,the main bearing of the shield machine has a complex loads and large structure size,which brings great challenges to the research on its dynamic characteristics and fault diagnosis.Therefore,based on the principle of scaled test,this thesis selects the three-row roller slewing bearing with the same structure and the same function as the shield main bearing as the research object,establishes the finite element model of the main thrust bearing,and carries out the researches on dynamic characteristics of the rolling element scratch fault by simulations.The dynamic characteristics of the main bearing under different working conditions are explored,and the vibration tests of the main bearing under different working conditions is carried out through the built shield main bearing damage test bench.Furthermore,a deep learning model that can effectively identify the failure types of main bearing rolling element is proposed,which provides theoretical support and practical guidance for the design,operation and maintenance of shield main bearings.The main research contents and conclusions are as follows:(1)Based on the finite element theory,a dynamic model of the shield main bearing with rolling element health/scratch damage was constructed,and the accuracy of the model was verified by experimental tests.The contact stress characteristics of the main thrust bearing under partial rolling element failure were analyzed,and the influence of load and rotational speed on the stress of rolling element was discussed.The results show that the failure aggravation of the rolling element will lead to the increase of the contact stress between the roller and the raceway.Under the action of "edge effect",when the healthy roller contacts the outer ring,the contact line presents a stress distribution of "large at both ends and small in the middle".The stress on the contact line of the faulty roller shows the distribution characteristics of "large at outer edge and small in inner edge".Within a certain range,the axial force has a greater influence on the contact stress between the roller and the outer ring than the rotational speed.(2)Based on the built shield main bearing damage test bench,the main bearing vibration test of rolling element with different degree of failure was carried out.The main bearing vibration signal database under different working conditions was constructed combined with the simulation data.The influence of service conditions and the degree of rolling element failure on the vibration characteristics of the main bearing was revealed.The results show that the vibration signal of shield main bearing presents periodic shock characteristics under the condition of rolling element failure.The variation trend of the time-domain characteristic indicators obtained from the experiment is as the same as the simulation,and the time-domain correlation features are sensitive to the degree of rolling element failure.However,the change of the main bearing’s service conditions has no significant effect on the time-domain characteristics of the vibration signal.The main frequency of the vibration signal envelope spectrum can reflect the characteristic frequency of rolling element faults well,but its main frequency is not sensitive to the degree of bearing failure.(3)In order to overcome the problem in extracting the fault features of the main bearing rollers,an end-to-end multi-scale one-dimensional convolutional neural network fault diagnosis algorithm for the main bearing is proposed based on the vibration signal characteristics of the shield main bearing.And the classification and identification of main bearing roller fault vibration signal samples under different working conditions was carried.The results show that the proposed MS-1DCNN can accurately identify rolling element with different failure degree,the accuracy rate is above 97.5% under different working conditions,and MS-1DCNN is suitable for main bearing roller fault identification.Compared with other commonly used methods,MS-1DCNN fault diagnosis algorithm has better stability and fault identification accuracy.Based on the proposed MS-1DCNN fault diagnosis algorithm and combined with the main thrust bearing dynamic simulation data and test data,an offline main bearing data management and analysis system was developed by using MATLAB APP Designer.