| In rock tunneling,especially in hard rock tunneling,the drilling and blasting method is the main construction form.Technicians mark the drilling points on the drilling face according to the perforation drawings,and complete the drilling through manual hand-held pneumatic drill or traditional hydraulic drilling trolley.This will lead to poor drilling positioning effect and potential safety risks.In this thesis,centering on the positioning requirements of the end of drill pipe in the working process of the drilling rig,the target identification method and single visual distance algorithm are used to measure the position of the working drill pipe of the drilling rig.Combined with the computer Flask-Web technology,the remote monitoring management system is designed to realize the remote monitoring of the real-time position of drill pipe of the drilling rig.The details are as follows:Based on the detailed analysis of the monitoring requirements of the drilling rig,the overall architecture of the drilling rig end positioning system and remote monitoring management system is designed,the layout scheme of the machine vision system is determined,and the camera selection and calibration are completed,considering the complex environment and lighting conditions in the actual production process.The multi-factor image acquisition and data set amplification of the self-made target were completed.An improved target recognition algorithm based on convolutional neural network is studied and verified by experiments.Based on the single-stage target detection algorithm YOLOv5 s,and aiming at the single target recognition requirement,the five aspects of attention mechanism,activation function,data enhancement strategy,loss function and initial anchor frame selection scheme are improved.By combining and comparing the above improved strategies,the results show that the improved target recognition algorithm has higher detection effect,and the time delay of the algorithm meets the requirements.A method of locating the end of drill pipe of rock drilling trolley based on monocular camera is proposed.Firstly,an improved target detection algorithm was used to detect the target fixed under the propulsion beam,and the two-dimensional coordinates of the target in its plane were determined.Secondly,the spatial distance between the target plane and the camera is measured by the single visual distance principle.Finally,the spatial Angle of the drill pipe is measured by a dip sensor mounted on the propulsion beam.Combined with the data of the above three parts,the spatial position of the end point of drill pipe is calculated.The results show that the positioning error and delay meet the system requirements.The fore and aft ends of the drilling pipe end positioning monitoring system of the drilling rig are designed and completed.Firstly,the functional and non-functional requirements of remote monitoring of rock cutting trolley are analyzed.Secondly,the front and back ends of the monitoring system are built using B/S architecture mode.Finally,According to the requirements of the system,functional modules such as login and registration function,calculation and positioning of the end of drill pipe,real-time hole position monitoring and vehicle PLC communication were built.In order to verify the feasibility of the whole scheme and the effectiveness of each module,the system test and performance analysis are carried out.First of all,the functional modules of the system are tested,including login registration function,data access and export function,target detection algorithm function,accuracy verification of single visual distance,detection of algorithm identification at various angles during drill pipe operation,and overall experiment of system integration.Secondly,the nonfunctionality of the system is verified,including the compatibility,security,stability and real-time performance of the monitoring and management system.The experimental results show that the remote monitoring system for the end positioning of drill pipe of drilling rig is feasible and effective.This thesis has 83 figures,21 tables and 103 references. |