The Research Of Recognize Ship Weld Flaw Defect System

Ship jointing is the key to guarantee the intensity and the quality of ship,it is also the important condition to guarantee safe shipping and working.If there is disfigurement in the welding line,that may crack the frame and induce leakage, finally beget ship sink,so checkout of welding line is much important during the process of ship's inspection.We usually use x-radiogram methods to check ship welding line,which would produce large number of films.As to assess and examine the films,if we adopt Computer Assistant Assess System,comparing to assessing by manpower which needs heavy workload,low work efficiency and more easily affected by technology diathesis and experience of the assessing people or something else,it will largely reduce workload of assessing,largely improve the work efficiency. In addition,it will make film's assessing process more impersonality,more scientific and more standardization.However,it is still difficult to distill,divide,and identify the disfigurement message exactly because of imperfects produced during the process of the welding line figure input to computer,such as overabundance yawp,illegibility edge of disfigurement,low contrast and so on.On the base of analyzing the status of welding line area's development in and out of our country,through various of methods trying to disposal the disfigurement figure,This paper give a further study on relative theory and key technology of welding line disfigurement identify system,adopting distilling technology of no-change character of square and nerve-network method.As follows are the key-point of the paper:(1) Studying the keystone of CCD imaging and collection,,designing the hardware system of negative's collection.It has been proved by experiment that the figure satisfied the requirement to assess.At the same time,functions as the figure's memory,delete,search and mimeograph have also been finished by image data-base technology.(2)Studying the arithmetic of the pretreatment,division of image and the edge's inspection,from comparison of experiments,it brought forward the arithmetic flow of processing welding line's image of ship and welding line disfigurement distilling methods based on DSA,which make a foundation of character distilling and disfigurement identify.(3)Studying the classification of welding line's usual disfigurement,distilling its no-change characteristic square parameters using method of no-change characteristic square,binding with its characteristic parameters of disfigurement shape,it offer a good supply to disfigurement identify.(4)Based on the methods of nerve-network's classification and actual situation of welding line disfigurement,an implement for nerve-network classification has been designed.It has proved that disfigurements,such as crack,air hole, dregs-nipping,non-welding on the root and non-melting,could be classified. Ultimately it access to the needs of assessing film assistant by computer.(5)According to Guide of Ship Jointing Inspection,printed by CCS on Oct.1, 2008,the paper has established ship welding line grades inspection module and ship welding line non-damage inspection data-base management system.