Research On The Design And Control System Of The Pipeline Robot For Monitoring The State Of Vertical Transmitting Wellbore

The operation space between the silo is narrow,which increases the difficulty of pipeline maintenance and the risk of equipment safety operation.In order to grasp more objectively the equipment status of the onshore simulated silo,and ensure the safety of the ejection experiment,Based on the support of "data analysis platform of integrated monitoring station" and "key technology of intelligent controland fault diagnosis of vertical launching cylinder cover system of underwater missile" jointly funded by the Ministry of education of equipment pre research,A robot control system for monitoring the condition of vertical launch wellbore was developed in this paper.It is mainly used to complete the tasks of status monitoring,video return and real-time diagnosis in a narrow space.Off-line detection was performed on the returned video to make further statistics on crack damage inside the pipe.First,The mechanical structure of the pipeline robot was designed,and the working principle,mechanical characteristics and driving performance of the mechanical unit for the pipeline robot were analyzed.The model of the pipeline robot was established,the movement analysis and simulation of the pipeline robot were carried out,the key parts of the robot were analyzed by finite element method,the reliability was preliminarily verified,and the motion simulation of the robot prototype was carried out to verify the effectiveness of the mechanical structure of the robot.Secondly,According to the functional requirements of the control system,we determine the overall scheme of the system,the system work principle,According to the function of the control system hardware design,build the industrial control motherboard module,The function is to load the upper computer system of the pipeline robot system,and send control instructions,display and store the returned video data;To realize the charging and discharging function of the display control platform by build the power management module;Determine the specific connection mode of the system hardware interface;The design of integrated controller's power circuit,starting mode interface circuit,indicator lamp circuit and drive motor speed regulation circuit are carried out.Combined with the control system hardware,the video module and motor control module of the control system software are designed by C language as the main programming language,including the initialization of video acquisition and the specific process of video acquisition,as well as the communication process,transmission protocol and interface configuration of the integrated controller.With C# as the main programming language,the upper computer interface is designed,and the composition of the upper computer interface and the system operation process are introduced.Then,in order to improve the pipeline robot control system's ability to identify cracks and damage and ease YOLO v3 on tiny crack in the pipe damage detection effect is not ideal,We made two improvements to the algorithm: combines GIOU loss into YOLO v3 target detector and improve the prediction accuracy of bounding box by GIOU bounding box evaluation method;Aiming at the problem of too much loss of superficial feature information caused by too many times of feature map reduction,an auxiliary network was added on the shallow layer network to directly transmit the superficial feature information to the corresponding YOLO detection layer,which enhanced the ability of YOLO detection layer to identify surface features such as crack texture,feature and color.Then the pipeline crack data set was made,the network structure was built,the network model was trained,and the accuracy rate,error rate and missed rate were introduced to evaluate the performance of the network model.Finally,An experimental study was conducted after the completion of the system construction,testing the walking performance of the pipeline robot prototype in the vertical tapering pipeline.The movement control experiment was conducted to verify the rationality of the mechanical structure of the pipeline robot and the practicability of the control system.An experimental study on pipeline crack identification was carried out.Identifying the video information returned by the handheld display control platform,the network model's recognition effect on the pipeline crack data set before and after the algorithm improvement was compared to verify the effectiveness for pipeline crack.The experimental results show that the system can meet the basic functional.