Design And Implementation Of Inspection Robot For Rollers Of TBM Continuous Belt

Continuous belt conveyor is the key equipment to realize the efficient,stable and safe ballast discharge of the full section rock tunnel boring machine(TBM).The running state and fault situation are related to the safety of TBM tunnel construction,among which the fire hazard caused by the idler sticking fault is particularly serious.In order to ensure the smooth development of TBM tunnel construction,it is necessary for the staff to patrol and test the continuous belt conveyor regularly with handheld detection equipment.However,TBM continuous belt conveyor is hundreds of meters short,tens of thousands of meters long,and the tunnel construction environment is harsh and complex,the staff need to shift day and night inspection,the detection effect is poor and low efficiency.Therefore,it is imperative to develop a continuous belt conveyor inspection robot to complete the inspection work instead of manual work,which is of great significance to improve the quality and efficiency of inspection,ensure the normal operation of continuous belt conveyor and the construction progress of TBM.In order to realize the fault diagnosis of idlers along the continuous belt conveyor,this thesis studies and designs a continuous belt conveyor inspection robot.By analyzing the on-site working environment and the work content of inspection personnel,the design requirements of the inspection robot were summarized.The suspension track type was selected as the transmission scheme,and a mechanical structure design and software and hardware control system scheme suitable for the track type inspection robot were designed based on dynamic analysis and various types of equipment required.The walking,positioning system,power supply Information collection and fault diagnosis systems have been developed.At the same time,in order to enable the inspection robot to accurately detect the fault idler,the detection and diagnosis of the fault idler is carried out by using the significance target detection theory based on RGB-T in combination with the collected visible light(RGB)images and thermal infrared(T)images.In this thesis,the contextguided stack refinement network CSRNet is used to detect the significance target(temperature anomaly)based on RGB-T deep fusion.The detection accuracy is much higher than the traditional algorithm,and the fault area can be detected quickly and accurately.Finally,according to the design scheme,the preliminary prototype of the inspection robot was manufactured,and the functions of the robot systems were tested.At the present stage,the inspection robot prototype basically met the established design requirements.