Data Processing And Software System Development Of Large Oil Tank Inspection Robot

Vertical oil tank is the main equipment for oil storage in oil depot.Due to construction technology,its own weight,environmental changes and liquid pressure,it is easy to deform,which leads to inaccurate volume measurement and even safety accidents.At present,oil tank deformation detection has been listed as one of important contents of annual inspection of oil tank,and it is very important to solve the problem of accurate volume measurement and data processing of large oil tank.At present,there are many technologies and methods for deformation detection and accurate measurement of vertical oil tank,and they have developed towards automation.The data obtained by automatic measurement need to be further processed automatically,and the parameters are displayed intuitively.This research work needs to be further solved.Aiming at the data processing problem of large oil tank inspection,this thesis takes the point cloud data obtained by the large vertical oil tank intelligent inspection robot studied by our group as the research object,and develops a data processing software system of large oil tank intelligent inspection robot to realize the automation of large vertical oil tank inspection.(1)Software design and development.According to the needs of users,a software is developed which can preprocess the acquired point cloud data,calculate and analyze the geometric parameters and deformation of oil tank,display and save the results.On the platform of Visual Studio 2017,combined with point cloud library,C++ programming language is used to write programs of functional modules such as point cloud data preprocessing,data calculation and analysis,result saving and display software functions.Qt is used as a graphical user interface library to design the software user interface,which realizes the functions of setting processing parameters,displaying point clouds and displaying results.(2)Point cloud data preprocessing.Based on the point cloud data collected by the largescale vertical intelligent inspection robot for oil tank,the point cloud data of oil tank are removed by voxel grid sampling method,and the point cloud data that keeps the characteristics of oil tank are obtained.Improved random sampling consistensus segmentation algorithm and statistical filtering denoising algorithm are used to remove large-scale noise,and moving least square method is used to remove small-scale noise,so as to remove the noise that affects the calculation accuracy and get the point cloud data representing the tank wall and bottom.(3)Calculation of geometric parameters of oil tank.The point cloud data of the tank bottom is fitted to the plane of the tank bottom by principal component analysis,and the position of the point cloud of the tank is adjusted according to the normal vector of the plane,so that the point cloud of the tank bottom is parallel to the XOY plane and the opening is in the positive direction of the Z axis.Poisson surface reconstruction algorithm is used to reconstruct the point cloud data of tank bottom into triangular mesh surface,and the tank bottom displacement is obtained by combining digital terrain model technology.The point cloud on the tank wall is sliced,and the outline of the scattered points on each slice is reconstructed by convex hull algorithm.The outline area of the scattered points is taken as the area of the ring plate at the corresponding height,and the area of the outline of each slice is accumulated to obtain the tank volume;Fitting the scattered points on the slice with the least square method,and taking the radius,center and roundness of the circle as the radius,center of mass and roundness of the ring plate at the corresponding height,fitting all the center of mass of the ring plate to obtain the center of mass line,and then obtaining the overall inclination;Correct the errors caused by tank bottom displacement,liquid pressure,temperature and accessories in the tank to the volume meter.(4)Software testing.The correctness of each module of the software is verified by standard case base testing method.By establishing a standard case base,the point cloud data of the standard model is input into the module to be verified,and the results obtained are compared with the data of the standard model.The accuracy of the software is verified by comparison test.The same object is measured by the strapping method and the three-dimensional laser scanner method,and then the point cloud data obtained by the scanner are input into the software to get the calculation results.The accuracy of the software is verified by comparing the calculation results of the software with the s results of the strapping method.