Design And Implementation Of Defect Inspection System Based On Radiography

With the computer technology and industrial automation, non-destructive testing techniques have more room for development. Currently present on the market, based on the X-ray casting defect detection system has a low degree of automation, the operation is complicated and does not meet the needs of efficiency and other issues. Moreover, since the radiographic characteristics and complexity of the runtime environment, the image produced by the noise, affecting the defect detection. In this paper, to solve the above problems, we develop an inspection system for casting flaws based on radiography, and launch a series of work for the design and implementation of the system. This work is as follows:Overall system design. Through research of the NDT development, we first analyze the system requirements, then compelete the design of system hardware, including mechanical devices and mechanical PLC control system; then turn to the design of PC software, including the adoption of the framework, design patterns and functional design, function modules include image acquisition, image processing, model management, high-pressure-ray control, mechanical motion control and network communications.The implementation of PC control software. Based on the WPF framework of Windows operating system to complete software implementation and each part of the software interface design; based on high-resolution digital flat panel detector to implement image acquisition and transmission; based on MySQL database to implement data management for casting model; based on Siemens PLC and serial protocol to implement.the control of voltage and current for X-Ray; based on Delta PLC and Modbus protocol to achieve mechanical motion control; based on TCP / IP to achieve data synchronization and statistics between multiple devices.The analysis of casting defect detection algorithm. This paper presents an edge filter algorithm based on maximum cross gradient sequence: first complete defect localization by linear scanning method, and then use the improved LoG edge detection algorithm, setting a significant edge threshold and real-edge value to filter closed contour that weak edge has made up most of the part, then followed by the maximum cross gradient sequence algorithm for further noise edges filter, then we make the final selection for flaw edge with the area and length condition. With the experiment, we find the algorithm is effective and the running time can meet the real-time testing requirements.Overall system test. With the usage of test cases, the system is functional and performance testing. The test results show that the system has achieved the desired goal and meet the production needs.