Research On Intelligent Analysis Method For Welding Quality Based On X-ray Imaging

In the petroleum engineering, especially in the long-distance transport pipeline construction and maintenance projects, welding and nondestructive testing is the key process, and the timeliness and accuracy of welding is important guarantee for the quality of petroleum engineering. The long-distance transport pipeline has the characteristics of high complex operating environment, long distance transmission, low information transmission reliability, difficult identification of welding micro defects, and is more complex and technically difficult compared with the conventional weld quality inspection and analysis problems. The inspection and analysis method for welding quality based on X-ray imaging is currently the most extensive range used method for welding quality inspection at home and abroad, and the problems of weld imaging based on X-ray, digital image transmission, weld defect nondestructive detection and recognition, and weld quality analysis and evaluation are all challenges for theoretical research and engineering technical personnel.This dissertation starts with the analysis of the long-distance transport pipeline domain features such as weld image collection under complex environment, large spatio-temporal span transmission, high energy consumption and low reliability of welding image, and propose a welding quality inspection and intelligent analysis method with more practical value. Using the knowledge technological means of petroleum engineering, wireless sensor network, image recognition and other subjects, this work studys the issues of weld quality evaluation method, X-ray welding image acquisition and long-distance transmission optimization problem, image reduction and micro weldpoint visual intelligent recognition theoretical problems, and designs and develops a set of X-ray based weld quality detection system using the automation technology of petroleum engineering and software engineering, which is applied in the long-distance transport pipeline weld quality analysis field.First, aiming at how to implement the content and method of long-distance transport pipeline weld quality evaluation based on X-ray, this dissertation analyzes the characteristics of the operation environment, the transmission distance, image transmission reliability and weld defect recognition difficulty of the long-distance transport pipeline.By compared the traditional film photography technology, computed radiography, and digital radioscopy, this work discusses the method which is applicable for long-distance transport pipeline weld quality detection. This paper presents the weld quality evaluation method based on X-ray digital imaging, including the welding quality evaluation index based on X-ray and the imaging standard and standard system which contains the standard classification, standard grade, standardization and its characteristics, and standardization of object. This provides a theoretical basis for quality evaluation of digital imaging based on X-ray, and puts forward the route of X-ray digital imaging technology from the angle of engineering realization. This paper discusses the visual detection method based on X-ray for long-distance transport pipeline welding quality detection, and analysis the principle of X ray imaging technology for long-distance transport pipeline; and discusses the application of X-ray digital imaging method in the weld quality of long-distance transport pipeline detection and technical difficulties of existence through the analysis of two kinds of X-ray digital imaging method of typical comparative, which are X-ray computed radiography detection method and X-raydetection digital imaging method.Next, in order to realize the remote transmission reliability of welded digital image data, especially in wireless sensor networks to reduce the energy consumption of data transmission and enhance the data transmission rate of accuracy, this paper presents the model of wireless sensor networks based on the regression model, and designd the corresponding image data transmission optimization method. This work describes the problems and key technology related of weld image data acquisition and transmission in the process; the simple linear regression model and multivariate linear regression model are established, to build the regression model of welding digital image transmission relies on wireless sensor network from different complexities; in order to solve the problem and the proposed mathematical model, this paper proposes a method for optimizing the transmission of image data based on distributed regression algorithm, and the formal definition of the algorithm is given for the core description, and a detailed description is proposed for the process design and parameter setting of the distributed regression algorithm. In order to verify the correctness and efficiency of the proposed algorithm, this paper evaluates the energy loss and the prediction precision by simulation. The simulation experiment results show that the algorithm is very suitable for compression of multiple monitoring data, to achieve the goals of data low energy consumption and high accuracy transmission of weld digital image, which has important meaning for long-distance transport pipeline weld image transmission accuracy and reliability, and directly influences the weld image reduction accuracy and the quality of the welding visual recognition efficiency. Meanwhile, the proposed wireless sensor network linear regression algorithm of data collection has certain significance for the exploration of application and development of wireless sensor network based theory.Subsequently, this dissertation carries out the work aiming at the intelligent recognition of weld quality analysis weld point intelligent recognition, which is the coree problem of this domain, especially in digital image recognition problems with close small weld point. The paper proposes and analyzes the weld defects of digital image aided identification problem, and emphasizes the weld points X-ray image recognition and intelligent recognition of weld defect image. This work presents respectively the weld point digital image intelligent recognition method and micro weld points X-ray vision recognition optimization method. The former analysises from the digital image pretreatment method and weld point image feature, and the latter focus on the X-ray image of two value processing method and edgedetection method of micro weld point, to present a visual identification method of Sobel operator based on X-ray microweld point. Through a series of experiments and contrast analysis, this verifies that the proposed algorithm has advantages of the stability, availability and robustness compared with traditional algorithms, especially in high precision welding tiny weld point recognition process, this algorithm has higher accuracy than the traditional algorithm, to reduce the recognition error effectively and guarantee the quality of the welding object.Finally, on the fundamental research of the problems of long-distance transport pipeline based on X-ray welding image quality evaluation methods, remote data transmission optimization method, and welding micro weld point visual intelligent identification method, this paper designs and implements of a X-ray digital radiography weld quality detection system, to verify the effect of application of X-ray imaging method of long-distance transport pipeline welds quality detection. In order to verify the X-ray imaging method based long-distance transport pipeline weld quality detection correctness and effectiveness, this paper collected large amount of the long-distance transport pipeline data from a oil refining transportation enterprises, and simulated the theoretical results in this paper using the preceding chapters to verify the weld quality of long-distance transport pipeline detection effect. The actual data are collected in the simulated operation in X-ray digitalimaging detection system, and analysises the system application effect in X-ray welding digital imaging, digital image network transmission, micro weld point intelligent recognition and so on, so as to verify the work for improving the analysis efficiency of weld quality, reduce the production cost of the actual application value.