The X-ray automatic detection device and method of cylinder jacket defect
Technical field
The present invention relates to X ray non-destructive testing technology, more particularly to a kind of X-ray automatic detection device of cylinder jacket defect
And method.
Background technology
Important component of the cylinder jacket as automobile engine, production is of large quantities, and quality requirement is high, but in casting, machinery
In process, because the effects such as material internal stress itself, clamping force easily cause what is formed in microfissure and casting process
The latent defects such as stomata, sand holes, if these defects are appeared in the engine under HTHP condition of work, it is easy to cause
The failure of engine simultaneously causes security incident.Therefore, to the quick, efficient of high-volume cylinder jacket, Non-Destructive Testing with regard to particularly important.
Current more advanced detection means mainly has detection method of eddy and X-ray detection method.Eddy-current method is only able to detect
Cylinder-barrel surface or near surface flaw, deep zone defect is difficult to detect;X ray rule be by cylinder be placed on x-ray source with into
As between device, observing ray image of the workpiece under multiple orientation repeatedly by human eye to judge that whether qualified workpiece is, one
Detection of the aspect to double-deck wall thickness can cause the resolution ratio for obtaining image to decline, and for there is externally threaded cylinder jacket to influence more
Greatly, on the other hand differentiate that result is influenceed by the experience of worker and the degree of fatigue of eyes, is difficult to soon for defective cylinder jacket
Speed determines that the defect is located at the antetheca or rear wall of beam penetration direction.This method is costly, accuracy rate is low and time-consuming, difficult
To meet the production requirement in high volume quick and precisely detected.
Specifically, current cylinder jacket defect detection method of X-ray such as Fig. 1,4 be computer, and 2 be to detect workbench, 3
It is cylinder jacket (to be detected) for X-ray tube, 9,6 be planar array type image device.
Existing detection method is that lead room is opened manually, cylinder jacket 9 be placed on X-ray tube 3 and planar array type image device 6 it
Between, radiographic source, which is sent, to be imaged on image device after ray penetrates workpiece and is sent to computer 4, and worker is observed on computer screen
The radioscopic image with cylinder jacket internal structural information so that differentiate that whether qualified workpiece is, and generally require turning of work piece
Observation differentiates repeatedly under multiple angles.This artificial detection method drawback is as follows:
(1) eye recognition method speed is slow, efficiency is low, it is difficult to meets large batch of efficiently quick detection requirement, and identifies
As a result the unavoidable flase drop with worker's subjectivity.
(2) planar array type detector is used in order to improve detection speed, cylinder jacket must be positioned over radiographic source (X-ray tube)
Among planar array type image device, the imaging of double-deck wall thickness causes the decline of real image resolution ratio, and cylinder is cased with external screw thread
When due to the irregular juxtaposition Imaging of front and rear wall screw thread it is more serious.In addition, for the cylinder jacket of uniform wall thickness, in reality
Border can produce the sinusoidal effect relevant with cylinder jacket itself arc length when being imaged, and cause the casting actually passed through on ray beam optical path
Wall thickness intermediate thin both sides are thick, and institute is inconsistent into the partially dark gray scale in partially bright both sides among image, and difficulty is brought to identification, for identifying
It is antetheca defect or rear wall defect that the defects of going out, which is also difficult to differentiation,.It is illustrated in figure 2 a parallel X-ray beam 10 and penetrates uniform wall
Behind the thick a certain section 11 of cylinder jacket into image intensity profile figure, as seen from the figure, upward in x-axis, 1,5 point of gray value f (x)
Highest, because beam is cylindrical tangent with cylinder jacket, ray, which is not decayed, directly to be beaten on imaging screen;2nd, 4 gray values are minimum,
Because the cylinder jacket wall that beam actually passes through is most thick, ray attenuation is also most;3 points of intermediate point then the most truly reflects ray
Shu Chuizhi penetrate double-deck wall thickness into image gray value.
The content of the invention
The present invention has what the parts of tubular structure collected when carrying out ray detection to solve current cylinder jacket etc.
Image can not accurately reflect product defects and then cause the inaccurate technical problem of testing result, there is provided a kind of cylinder jacket defect
X-ray automatic detection device and method.
The X-ray automatic detection device of cylinder jacket defect of the present invention is realized using following technical scheme:One
The X-ray automatic detection device of kind cylinder jacket defect, including computer control are penetrated with graphics processing unit, detection workbench and X
Spool;Also include the Linear Array Realtime image device being engaged with X-ray tube;The Linear Array Realtime image device passes through detector support arm
Support and controlled with computer and is connected with graphics processing unit;The detection workbench can oscilaltion and 360 degree of freedom
Rotate and be used to jack up cylinder jacket the periphery for being enclosed on Linear Array Realtime image device, computer control passes through machine with graphics processing unit
The action of electric control module control detection workbench and detector support arm.
Cylinder jacket can be enclosed on the periphery of Linear Array Realtime image device by device of the present invention, pass through rotation cylinder set
The cylinder jacket radioscopic image of one week is obtained with, a plane is launched into equivalent to by the cylinder jacket of cylindrical shape, so as to obtain
The single wall imaging of cylinder jacket, so avoiding the decline of image resolution ratio present in existing detection method causes testing result
The problem of inaccurate.Detection workbench can freely rotate around rotary shaft.
The X ray automatic testing method of cylinder jacket defect of the present invention adopts the following technical scheme that realization:One
The X ray automatic testing method of kind cylinder jacket defect, comprises the following steps:
Step 1:After detection means starts, system initialization:1. the cylinder jacket model of detection is determined, it is automatic according to its wall thickness
It is determined that the x-ray tube voltage needed and tube current size;2. according to actually detected it needs to be determined that the cylinder jacket defect type of detection
With flaw size, detection means automatically determines detection workbench and does the maximum rotary step to rotate in a circumferential direction around rotary shaft accordingly
N °, n can be divided exactly by 360, N=360/n;
Step 2:Computer controls sends instruction with graphics processing unit, Electromechanical Control module by batch cylinder jacket workpiece according to
It is secondary to be sent to detection operating position;
Step 3:After workpiece arrival detection workbench is detected, Electromechanical Control module will be loaded with the detection work of cylinder jacket automatically
Make platform to jack up vertically upward so that cylinder jacket is enclosed on Linear Array Realtime image device periphery completely, and cylinder jacket is sent out close to X-ray tube
Penetrate imaging surface of the antetheca as close to Linear Array Realtime image device of end side;
Step 4:X-ray tube sends X-ray beam transillumination cylinder jacket, and Linear Array Realtime image device is obtained before its cylinder jacket
One column data of wall, is designated as X1:
X1=(x11,x21,…,xM1)T (1)
Wherein, X1For M dimensional vectors, the pixel number that Linear Array Realtime image device in the vertical direction collects is represented, symbol
Number T represents transposition;
Step 5:Detection workbench drive cylinder jacket is done with maximum n ° of rotary step corresponding to the model cylinder jacket between circumference
Formula of having a rest at the uniform velocity rotates and gathered data, and following N-1 groups M dimension datas are obtained with step 4 identical method:
Step 6:By the image that the data unified integration of the N group M dimensional vectors in formula (1) and formula (2) is M rows N row
Matrix X:
Wherein, matrix X is comparable to formed single wall panoramic x-ray image after cylinder jacket circumferentially deploying, in matrix
Element xijThe grey scale pixel value that image arranges in the i-th row jth is represented, image size is M × N;
Step 7:The single wall panoramic x-ray image of acquisition is judged;The detection to cylinder jacket is completed one by one.
Due in imaging process, the view data of cylinder jacket is between radiographic source and detector and presses close to detector
The single wall of imaging surface is vertically imaged, and this imaging method is launched into after rectangular panel by linear array equivalent to by cylinder jacket in itself
Detector scans by column imaging.For a uniform wall thickness and do not have defective qualified cylinder jacket, ignore the inconsistency of X ray
With the influence factor such as system noise, ideal situation hypograph X each column data corresponding element, in theory all should it is approximate completely
It is identical, i.e., in above-mentioned formula (1) and formula (2), it should have:
X1=X2=...=Xj=...=XN (7)
Namely the data gray value per a line in formula (3) in theory should be equal, so avoids bilayer
The sinusoidal effect caused by arc length when wall configuration is imaged.
Be illustrated in figure 4 a parallel X-ray beam 10 penetrate behind uniform wall thickness cylinder jacket a certain individual layer expansion section 12 institute into
Intensity profile Fig. 4 of image.As seen from the figure, perpendicular to the axial a certain section part of cylinder jacket, circumferentially on cylinder
It is in the same size to cover section gray value, overcomes sinusoidal effect caused by arc length.It is imaged compared to double wall, single wall imaging method is notable
The resolution ratio of image is improved, the front and rear wall screw thread in double wall imaging is substantially being overcome in the case of cylinder is cased with external screw thread
Irregular cross jamming.
In actual imaging detection, due to the influence of X ray and other extraneous factors, each columns described in formula (7)
Should approximately equal according to corresponding element:
X1≈X2≈…≈Xj≈…≈XN (8)
For the single wall panoramic x-ray image obtained, can be detected using various existing detection methods;Due to
Obtain the single wall panoramic x-ray image of cylinder jacket to be measured so that image can reflect cylinder jacket to be measured to greatest extent
Feature, so as to be laid a good foundation for the accuracy of measurement.
Further, computer is controlled with graphics processing unit by quick difference shadow method to collecting single wall panoramic x-ray
The cylinder jacket of image is detected, and realizes automatic detection;The quick difference shadow method comprises the following steps:
Step a:By the type and size of step 1 known defect, it is assumed that the minimum defect to be detected reflection is in the picture
Size be m × n, i.e., sub-district area image of the defect in cylinder jacket entire image be m rows n row, then can use m × n template
It is used for quickly detecting;
Step b:Preceding m rows data in formula (3) are taken, from the 2nd row to the pixel grey scale Value Data of Nth column, subtract it successively
Previous column pixel corresponding grey scale value is simultaneously averaged, and is preserved data and is:
D1=(d11,d12,…,d1,N-1) (4)
Step c:The 2nd row is taken in formula (3) successively to m+1 row data, the 3rd row to m+2 row data ... ..., until
Last m row data, respectively perform above-mentioned steps b, obtain and preserve data be:
Step d:Data unified integration by the M-m+1 group N-1 dimensional vectors in formula (4) and formula (5) is M-m+1 rows
The matrix D of N-1 row:
Wherein, D is image X recognition matrix, dijIn the image array of representation formula (3) from i rows to i+m-1 rows
M grey scale pixel value of j+1 row correspondingly subtracts the average of m grey scale pixel value of jth row;
Step e:Data characteristics in analytical formula (6) per a line, you can complete the detection of cylinder jacket:1. for preferable
Qualified products, all data element d in DijIt is small value, and each row of data should be very close;2. for same data line
The continuous two abnormal data member sizes of middle emergence close to it is positive and negative on the contrary, and its immediate two row data up and down it is normal, then this
Position is isolated noise point, is also considered as certified products;3. the defects of if there is not less than m × n, then in same a line defective locations
Left margin point there is positive exceptional value, there is negative sense exceptional value, continuously n data is no less than between right boundary in right margin point
Present it is irregular float but all greater than the normal data value of area free from defect, the area data outside right boundary is normal,
More than the adjacent m rows of this feature continued presence, then assert detected cylinder jacket existing defects;
Step f:Testing result is provided according to step e, Electromechanical Control module will be loaded with the table vertical of workpiece automatically
Classify after drop blanking;
Step g:Continue to detect next cylinder jacket.
The cylinder jacket detection method of X ray is currently based on, is all to obtain human eye interpretation pattern after image.Meter is utilized in others
Calculation machine detects identification field to radioscopic image, and typically all first radioscopic image is pre-processed, takes medium filtering mostly
The randomness noises such as isolated point present in original image are eliminated Deng preprocess method.The image intensity value in defective region is more than
The gray value of area free from defect, so in calculus of differences, what the left margin of defect subtracted that previous column obtains is one big just
Number, i.e., positive exceptional value, what the right margin of defect subtracted each other to obtain by latter row normal pixel is a big negative, i.e. negative sense is different
Constant value.In order to improve detection speed and ensure Detection accuracy, the invention provides a kind of quick difference shadow method to penetrate to single wall panorama X
The fast automatic judgement of line image;, can be accurate according to the detection level required by product by each step of quick difference shadow method
Really quickly identify the defects of corresponding (such as sand holes, stomata, crackle).Those skilled in the art are according to work described herein
The process of work can be easy to write out corresponding software.
Beneficial effects of the present invention:
1st, the cylinder jacket automatic detection process based on X ray is all automatically performed by computer control, and whole process includes
Feeding, cylinder jacket rotating scan imaging, radioscopic image identification, classification blanking.Imaging device can be according to the cylinder that will be detected
Set model automatically determines required radiographic source tube voltage and tube current size, type and defect chi according to the defects of actually detected
It is very little, automatically determine workbench and do the maximum step-length that rotates in a circumferential direction around rotary shaft so as to ensure that the quick of minimum valid data is obtained
Take;
2nd, imaging method is the single wall imaging method being placed in using Linear Array Realtime detector inside cylinder jacket, the picture number of cylinder jacket
According to be between radiographic source and detector and press close to detector image-forming face single wall it is vertical be imaged, this imaging method is in essence
On imaging is scanned by column by linear array detector equivalent to cylinder jacket is launched into after rectangular panel, overcome in double-walled imaging due to
Sinusoidal effect caused by arc length.It is imaged compared to double wall, single wall imaging method significantly improves the resolution ratio of image, in cylinder jacket
Have and the front and rear wall screw thread in double wall imaging is effectively prevent in the case of external screw thread be irregularly crossed as the interference of picture, and overcome
It is difficult to the problem of determination is antetheca or rear wall defect in double wall imaging;
3rd, quick difference shadow method proposed by the present invention directly detects to raw image data, eliminates the centre of image preprocessing
Link, it is effectively improved the speed of image detection and has evaded the random noise disturbances such as isolated point simultaneously, and computer is automatic
Identification efficiently avoid the uncertainty of human eye interpretation;
4th, apparatus and method of the present invention can also realize the fast automatic detecting to other tubular workpieces.
Brief description of the drawings
The existing cylinder jacket structure of the detecting device schematic diagrames of Fig. 1.
The existing detection methods of Fig. 2 obtain the schematic diagram of radioscopic image.
The structural representation of Fig. 3 devices of the present invention.
Fig. 4 is of the present invention to take precautions against the schematic diagram for obtaining single wall panoramic x-ray image.
1- computers control and graphics processing unit, 2- detection workbench, 3-X ray tubes, 4- computers, 5- Linear Array Realtimes into
As device, 6- planar array type image devices, 7- detector support arms, 8- Electromechanical Control modules, 9- cylinder jacket, 10- parallel X-rays
Beam, a certain section of 11- cylinder jacket, a certain individual layer expansion section of 12- cylinder jacket.
Embodiment
A kind of X-ray automatic detection device of cylinder jacket defect, including computer control and graphics processing unit 1, detection
Workbench 2 and X-ray tube 3;Also include the Linear Array Realtime image device 5 being engaged with X-ray tube 3;The Linear Array Realtime image device 5
Supported by detector support arm 7 and controlled with computer and is connected with graphics processing unit 1;It is described detection workbench 2 can on
Lower lifting and 360 degree of peripheries for rotating freely and Linear Array Realtime image device 5 being enclosed on for cylinder jacket 9 to be jacked up, computer control
System controls the action of detection workbench 2 and detector support arm 7 with graphics processing unit 1 by Electromechanical Control module 8.
The Electromechanical Control module 8 includes feed mechanism;Batch cylinder jacket workpiece is sent to detection work by feed mechanism successively
Make platform position;Feed mechanism can be mounted with the machines such as traffic cycle or conveyer belt or the conveyer chain of some detection stations
Structure.
A kind of X ray automatic testing method of cylinder jacket defect, comprises the following steps:
Step 1:After detection means starts, system initialization:1. the model of cylinder jacket 9 of detection is determined, according to its wall thickness certainly
The dynamic voltage of X-ray tube 3 for determining to need and tube current size;2. according to actually detected it needs to be determined that the cylinder jacket defect class of detection
Type and flaw size, detection means automatically determine detection workbench 2 and the maximum rotation step to rotate in a circumferential direction are done around rotary shaft accordingly
Long n °, n can be divided exactly by 360, N=360/n;
Step 2:Computer controls sends instruction with graphics processing unit 1, and Electromechanical Control module 8 is by batch cylinder jacket workpiece
Detection workbench 2 position is sent to successively;
Step 3:After workpiece arrival detection workbench 2 are detected, Electromechanical Control module 4 will be loaded with the inspection of cylinder jacket 9 automatically
Survey workbench 2 to jack up vertically upward so that cylinder jacket 9 is enclosed on the periphery of Linear Array Realtime image device 5 completely, and cylinder jacket 9 is close to X
Imaging surface of the antetheca of the transmitting terminal side of ray tube 3 as close to Linear Array Realtime image device 5;
Step 4:X-ray tube 3 sends X-ray beam transillumination cylinder jacket 9, and Linear Array Realtime image device 5 obtains and abuts its cylinder
A column data of 9 antethecas is covered, is designated as X1:
X1=(x11,x21,…,xM1)T (1)
Wherein, X1For M dimensional vectors, the pixel number that the in the vertical direction of Linear Array Realtime image device 5 collects, symbol are represented
Number T represents transposition;
Step 5:Detection workbench 2 drives cylinder jacket 9 to do circumference with maximum n ° of rotary step corresponding to the model cylinder jacket
Batch (-type) at the uniform velocity rotates and gathered data, and following N-1 groups M dimension datas are obtained with step 4 identical method:
Step 6:By the image that the data unified integration of the N group M dimensional vectors in formula (1) and formula (2) is M rows N row
Matrix X:
Wherein, matrix X is comparable to formed single wall panoramic x-ray image after the circumferentially deploying of cylinder jacket 9, in matrix
Element xijThe grey scale pixel value that image arranges in the i-th row jth is represented, image size is M × N;
Step 7:The single wall panoramic x-ray image of acquisition is judged;The detection to cylinder jacket 9 is completed one by one.
Computer is controlled with graphics processing unit 1 by quick difference shadow method to collecting the gas of single wall panoramic x-ray image
Cylinder sleeve 9 is detected, and realizes automatic detection;The quick difference shadow method comprises the following steps:
Step a:Mainly there are sand holes, stomata by the type and size of the known defect of preceding step 1, the defects of cylinder jacket, split
Line etc., and mainly based on sand holes.In order to briefly describe the embodiment of quick difference shadow method automatic detection in the present invention, it is assumed that required
Defect is detected as the sand holes not less than 0.5 millimeter × 0.5 millimeter, and assumes the reality of 1 0.1 millimeter of pixel correspondence mappings in image
Border size, then it can be used for quickly detecting using 5 × 5 template;
Step b:Preceding 5 row data in formula (3) are taken, from the 2nd row to the pixel grey scale Value Data of Nth column, subtract it successively
Previous column pixel corresponding grey scale value is simultaneously averaged, and is preserved data and is:
D1=(d11,d12,…,d1,N-1) (4)
Step c:Take the 2nd row to the 6th row data in formula (3) successively, the 3rd row to the 7th row data ... ..., to the last 5
Row data, respectively perform above-mentioned steps b, obtain and preserve data be:
Step d:It is M-4 rows N-1 by the data unified integration of the M-4 group N-1 dimensional vectors in formula (4) and formula (5)
The matrix D of row:
Wherein, D is image X recognition matrix, dijIn the image array of representation formula (3) from i rows to the jth of i+4 rows+
5 grey scale pixel values of 1 row correspondingly subtract the average of 5 grey scale pixel values of jth row;
Step e:Data characteristics in analytical formula (6) per a line, you can complete the detection of cylinder jacket:1. for preferable
Qualified products, all data element d in DijIt is small value, and each row of data should be very close;2. for same data line
The continuous two abnormal data member sizes of middle emergence close to it is positive and negative on the contrary, and its immediate two row data up and down it is normal, then this
Position is isolated noise point, is also considered as certified products;3. if there is the sand holes defect not less than 0.5 millimeter × 0.5 millimeter, then exist
There is positive exceptional value in left margin point with a line sand holes position, and negative sense exceptional value occurs in right margin point, between right boundary
In 5 data irregular floating is presented but all greater than the normal data value of area free from defect in continuous be no less than, in right boundary
Outside area data it is normal, more than adjacent 5 row of this feature continued presence, then assert the detected existing defects of cylinder jacket 9;
Step f:Testing result is provided according to step e, Electromechanical Control module will be loaded with the table vertical of workpiece automatically
Classify after drop blanking;
Step g:Continue to detect next cylinder jacket.
The Electromechanical Control module includes feed mechanism;Electromechanical Control module is by feed mechanism by batch cylinder jacket workpiece
Detection operating position is sent to successively;Feed mechanism can be traffic cycle or the transmission for being mounted with some detection stations
The mechanism such as band or conveyer chain.
Electromechanical Control module, detection workbench and detector support arm can be realized using existing device;Inspection
Surveying workbench can be rotated with oscilaltion and 360 degree, be realized by corresponding stepper motor and cylinder can;Detector
Support arm can be realized using screw-nut body, as long as being capable of oscilaltion can.