CN104502451A - Method for identifying flaw of steel plate - Google Patents

Abstract

The invention relates to a method for identifying a flaw of a steel plate. The method comprises the steps of processing an ultrasonic echo signal to obtain a wave curve, intercepting the wave curve to acquire a flaw wave curve, calculating the wave curve to extract peak data to obtain a flaw-type data group, further processing the flaw-type data group, and finally acquiring the flaw type of the steel plate by utilizing a flaw threshold value judging method and a morphology analysis method. According to the method for identifying the flaw of the steel plate, the flaw of the steel plate can be identified only according to the echo signal acquired by an ultrasonic probe device, the calculation speed is high, rapidness and accuracy in identification can be realized, a great amount of sample data is not needed for training, the data calculation amount is reduced, the identification speed is increased, and the flaw identification cost is reduced.

Description

A kind of steel plate defect recognition methods

Technical field

The present invention relates to one to utilize ultrasound wave to carry out steel plate defect type automatically to know method for distinguishing.

Background technology

Along with the develop rapidly of China's industrial construction, the demand of steel plate can be increasing, also more and more higher to the requirement of its inherent quality.Medium plate may form layering in process of production, crackle, diffusion-type be mingled with exceed standard, white point, the various defect such as segregation and hydrogen induced cracking.At present, domestic in middle thickness high intensity steel intralamellar part quality judging, depend on ultrasonic detection technology.

The large-scale fixed inspection system of the many employings of large-scale steel mill detects steel plate defect, is mainly applicable to steel plate detection of shaping in enormous quantities.In addition, the checkout equipment of short run, the detection of small-sized steel plate defect is applicable in addition.

Authorization Notice No. is the Chinese utility model patent of CN202693526U, CN202101975U, CN201141855Y, CN201503418U, and application number is the Chinese invention patent application of 201310750784.5 (application publication number is CN103698409A), the steel sheet detector of ultrasound wave disclosed in it has all done detailed elaboration to its detection architecture and monitoring principle.Above-mentioned several steel sheet detector can detect position and the equivalent size thereof of defect, can not the type of defect recognition.Again because dissimilar defect requires different to plate quality grading, so, the accurate judgement of the defect type safe handling for steel plate is had very important significance.The shape of the ultrasonic Flaw A ripple signal mainly collected according to ultrasonic instrument at present, the flaw detection experience relying on testing staff manually judges, inevitably can introduce error like this.

The patent No. is the Chinese invention patent " a kind of method extracting the spectral amplitude phase information of ultrasound echo signal " of ZL97109099.8 (Authorization Notice No. is CN1065961C), wherein adopts Fourier transform to obtain the frequency spectrum of ultrasonic Flaw signal.The patent No. is the Chinese invention patent " digital signal processing method of ultrasonic signal " of ZL200410011403.2 (Authorization Notice No. is CN100410925C), wherein by lifting wavelet transform, joint time frequency analysis is carried out to ultrasonic Flaw signal, extract the energy feature of flaw indication in different frequency range.These feature extracting methods are all that extraction rate is slower based on Hilbert transform and Fourier transform above.In defect inspection, conventional method is pattern-recognition, wherein has a variety of sorter.If the patent No. is ZL200710059575.0 (Authorization Notice No. is CN100567978C) Chinese invention patent " ultrasonic phased array detects oil gas pipeline girth weld defect type automatic identifying method ", wherein defect type automatic identifying method is adopted to be lifting wavelet transform combined with fractal technology, based on the automatic identifying method of supporting vector machine model, the method needs the training carrying out supporting vector machine model, the training sample number obtaining higher recognition correct rate needs is larger, and the acquisition of a large amount of training sample not easily realizes, its use does not have popularity.

Summary of the invention

Technical matters to be solved by this invention provides a kind of for above-mentioned prior art can identify steel plate defect type and the steel plate defect recognition methods that recognition speed is fast, recognition accuracy is high.

The present invention's adopted technical scheme that solves the problem is: a kind of steel plate defect recognition methods, is characterized in that: comprise the following steps:

Step one, Ultrasound Instrument start and initialization, and host computer sends controling parameters to Ultrasound Instrument;

Step 2, ultrasonic probe device start and carry out work according to the controling parameters in Ultrasound Instrument, ultrasonic probe device moves on tested steel plate, host computer obtains ultrasonic probe device to the real-time location coordinates data of steel plate check point, forms coordinate data group the W=[(x of check point ₁, y ₁), (x ₂, y ₂) ..., (x _j, y _j) ..., (x _b, y _b)], wherein j and b is positive integer, and 1≤j≤b, b is check point sum, (x _j, y _j) be the position coordinates of a jth check point, x _jrepresent the value of a jth check point on steel plate length direction, y _jrepresent the value of a jth check point on steel plate width direction;

Ultrasonic probe device launches ultrasound wave to tested steel plate, receive the ultrasonic echo signal from steel plate simultaneously, wherein, ultrasonic echo signal comprises from the start signal of surface of steel plate reflection, from the flaw indication of defective locations reflection, the end ripple signal from steel plate bottom reflection;

Step 3, Ultrasound Instrument collection store the echoed signal of each check point that ultrasonic probe device returns;

Step 4, Ultrasound Instrument process echoed signal according to the velocity of sound in steel plate and gain, and then for each check point position, the degree of depth-amplitude squiggle that within the scope of the corresponding tested steel plate of acquisition, echoed signal is formed, thus Formation Depth-amplitude squiggle data group A=[A ₁, A ₂..., A _j..., A _b], wherein j and b is positive integer, and 1≤j≤b, b is check point sum, A _jrepresent the degree of depth-amplitude squiggle of a jth check point;

Curve data in the degree of depth-amplitude squiggle data group A uploads in host computer by Ultrasound Instrument;

Step 5, host computer carry out subsequent treatment to the degree of depth-amplitude squiggle data group A;

First, in the degree of depth-amplitude squiggle data group A, obtain start signal amplitude corresponding to each check point, thus form initialize signal amplitude data group I=[I ₁, I ₂..., I _j..., I _b], in the degree of depth-amplitude squiggle data group A, obtain end wave amplitude corresponding to each check point, thus form end wave amplitude data group D=[D ₁, D ₂..., D _j..., D _b];

Calculate initialize signal amplitude mean value $\stackrel{\‾}{I}=\frac{I_1+I_2+...+I_j+...+I_b}{b};$

Calculate end wave amplitude mean value $\stackrel{\‾}{D}=\frac{D_1+D_2+...+D_j+...+D_b}{b};$

Wherein j and b is natural number, and 1≤j≤b, b is check point sum, D _jrepresent that a jth check point is at its corresponding degree of depth-amplitude squiggle A _jthe end ripple signal amplitude of upper correspondence;

Pre-service is carried out to the degree of depth-amplitude squiggle data group A, namely corresponding to each check point in the degree of depth-amplitude squiggle data group A degree of depth-amplitude squiggle intercepts, the waveform that in surface to the bottom surface depth range retaining tested steel plate, echoed signal is corresponding, thus form defective waveform curve data group B=[B ₁, B ₂..., B _j..., B _b], wherein j and b is positive integer, 1≤j≤b, B _jrepresent the defective waveform curve of a jth check point, b is check point sum;

The defective waveform curve corresponding to each check point in defective waveform curve data group B carries out derived function, thus obtains all wave crest points in corresponding check point defective waveform curve, thus builds wave crest point information data group C=[C ₁, C ₂..., C _j..., C _b]; C _j=[C _{[j] [1]}, C _{[j] [2]}], C _{[j] [1]}=(S _j0, S _j1..., S _ji..., S _ja), C _{[j] [2]}=(F _j0, F _j1..., F _ji..., F _ja);

Wherein j and b is natural number, and 1≤j≤b, b is check point sum, C _jrepresent the defective waveform curve B of a jth check point _jin all wave crest point information data groups of comprising; I and a is natural number, and 0≤i≤a, a is wave crest point sum, C _{[j] [1]}represent the depth of defect array of a jth check point, S _jifor the defective waveform curve B of a jth check point _jin depth of defect value corresponding to i-th wave crest point, C _{[j] [2]}for the defect amplitudes array of a jth check point, F _jifor the defective waveform curve B of a jth check point _jin flaw indication amplitude corresponding to i-th wave crest point;

Arrange the information data of each check point, build an information database M=[M ₁, M ₂..., M _j..., M _b], M _j=[(x _j, y _j), C _j, D _j], wherein j and b is natural number, and 1≤j≤b, b is check point sum, M _jrepresent the message data set that a jth check point is corresponding;

Step 6, treating that ultrasonic probe device is to after tested steel plate scanning, is (x for arbitrary coordinate _n, y _n) check point, wherein 1≤n≤b, n and b is positive integer, b be check point sum;

According to its corresponding depth of defect array C _{[n] [1]}in depth of defect value corresponding to each wave crest point, to check point (x _n, y _n) corresponding message data set M _nintegrate;

I.e. check point (x _n, y _n) corresponding defective waveform curve B _nin, in two adjacent successively wave crest points, when the absolute value that the depth of defect value that a rear wave crest point is corresponding deducts the difference of depth of defect value corresponding to previous wave crest point is less than degree of depth relevance threshold q, information data corresponding with previous wave crest point for information data corresponding for a rear wave crest point is classified as a sub-message data set; Otherwise, a newly-built sub-message data set;

By that analogy, thus form check point (x _n, y _n) corresponding sub-information data cluster:

P _n=[P _n1, P _n2..., P _nm..., P _nk], wherein m and k is positive integer, and 1≤m≤k, k is check point (x _n, y _n) corresponding sub-message data set sum, P _nmrepresent check point (x _n, y _n) corresponding sub-information data cluster P _nin m sub-message data set;

For sub-information data cluster P _n=[P _n1, P _n2..., P _nm..., P _nk] in each sub-message data set, obtain its depth of defect extreme value data group L _n=[L _n1, L _n2..., L _nm..., L _nk], L _nm=(MaxC _{[nm] [1]}, MinC _{[nm] [1]}), wherein m and k is positive integer, and 1≤m≤k, k is check point (x _n, y _n) corresponding sub-message data set sum, L _nmrepresent check point (x _n, y _n) corresponding sub-information data cluster P _nin m sub-message data set P _nmin depth of defect extreme value, MaxC _{[nm] [1]}represent its depth of defect maximal value, MinC _{[nm] [1]}represent its depth of defect minimum value;

Wave crest point information data corresponding for all check points is carried out integrate the new information database M'=[P of rear formation ₁, P ₂..., P _n..., P _b], wherein n and b is natural number, and 1≤n≤b, b is check point sum, P _nrepresent the n-th check point (x _n, y _n) corresponding sub-information data cluster;

Obtain the depth of defect extreme value data set L=[L that each sub-message data set in sub-information data cluster corresponding to all check points is corresponding simultaneously ₁, L ₂..., L _n..., L _b], wherein n and b is natural number, and 1≤n≤b, b is check point sum, L _nrepresent the n-th check point (x _n, y _n) corresponding depth of defect extreme value data group;

Be (x for arbitrary coordinate _n, y _n) the corresponding depth of defect extreme value data group L of check point _n=[L _n1, L _n2..., L _nm..., L _nk] in each depth of defect extreme value data, wherein m and k is positive integer, and 1≤m≤k, k is check point (x _n, y _n) corresponding sub-message data set sum; Searching and detecting point (x _n, y _n) place eight territory { (x _n, y _n-1); (x _n, y _n+1); (x _n-1, y _n-1); (x _n-1, y _n); (x _n-1, y _n+1); (x _n+1, y _n-1); (x _n+1, y _n); (x _n+1, y _n+1) in each depth of defect extreme value data in the corresponding depth of defect extreme value data group of each check point, occur simultaneously if depth of defect extreme value data exist, then sub-message data set corresponding for the check point of its correspondence merged;

So, the sub-message data set of all check points is integrated after process through search according to depth of defect extreme value data, builds new defect information database Q=[Q ₁, Q ₂..., Q _u..., Q _z], wherein u and z is positive integer, and 1≤u≤z, z is sub-defect information data set sum, Q _urepresent the sub-defect information data set of u defect;

Step 7, calculating defect information database Q=[Q ₁, Q ₂..., Q _u..., Q _z] in the corresponding defect parameters of each sub-defect information data set, thus obtain defect parameters array V=[V ₁, V ₂..., V _u..., V _z], wherein u and z is positive integer, and 1≤u≤z, z is sub-defect information data set sum, V _urepresent the defect parameters of u sub-defect information data set;

$V_u=\frac{v\×x\×y}{\underset{e=1}{\overset{v}{\mathrm{\Σ}}}(\mathrm{Max}{C}_{\left[e\right]\left[1\right]}-\mathrm{Min}{C}_{\left[e\right]\left[1\right]})};$

Wherein v represents word defective data collection Q _uin the check point sum that comprises, x represents the distance between the check point center that two on steel plate length direction are adjacent, and y to represent on steel plate width direction the distance between two adjacent check point centers, MaxC _{[e] [1]}represent defective data collection Q _uin depth of defect maximal value corresponding to e coordinate points, MinC _{[e] [1]}represent defective data collection Q _uin depth of defect minimum value corresponding to e coordinate points;

Computing information database Q=[Q simultaneously ₁, Q ₂..., Q _u..., Q _z] in the corresponding average amplitude of each sub-defect information data set, thus obtain defect average amplitude array wherein u and z is positive integer, and 1≤u≤z, z is sub-defect information data set sum, represent the average amplitude of u sub-defect information data set;

By defect parameters array V=[V ₁, V ₂..., V _u..., V _z] in each defect parameters and rarefaction defect threshold value R ₁compare, if V _u<R ₁, then defect parameters V is judged _ucorresponding sub-defective data collection Q _ucorresponding steel plate defect type is rarefaction defect; .

Weed out information database Q=[Q ₁, Q ₂..., Q _u..., Q _z] in message data set corresponding to rarefaction defect, by defect parameters corresponding for remaining each message data set and average amplitude respectively with lamination defect threshold value R ₂and amplitude thresholds f compares, if R ₂<V _u<1 and then judge defect parameters V _ucorresponding sub-defective data collection Q _ucorresponding steel plate defect type is lamination defect;

Step 8, weed out information database Q=[Q ₁, Q ₂..., Q _u..., Q _z] in the rarefaction defect defect information data set corresponding with lamination defect, remaining each sub-defect information data set is carried out pseudo color image reconstruct on two dimensional surface, form corresponding subset image, then carry out morphological analysis for subset image, thus acquisition steel plate defect type is crack defect or gas hole defect;

The nearly circularity T calculating each subset image, to judge the gas hole defect of steel plate, calculates the flexibility G of each subset image to judge the crack defect of steel plate;

Wherein T=S/ (π × d ²/ 4), wherein T represents nearly circularity, and S represents that the area that subset image comprises, d to represent in subset image the distance of 2 farthest, as nearly circularity T≤t, then think the corresponding subdata of this subset image integrate corresponding to steel plate defect as porous defect;

G=S/ (h × l), wherein I represents flexibility, S represents the area occupied by subset image, h to represent in subset image the distance of 2 farthest, l represent in subset image with the mean breadth on perpendicular direction, h direction, as flexibility G≤g, then judge the corresponding subdata of this subset image integrate corresponding to steel plate defect as crack-type defect.

Preferably, in described step 2, ultrasonic probe device drives arc movement on steel plate by walking aids tool, and described scrambler is arranged on described walking aids tool with the coordinate data of Real-time Obtaining ultrasonic probe device check point on steel plate.

Easily, described ultrasonic probe device comprises multiple ultrasonic probe be set up in parallel, and forms ultrasonic probe group, in described ultrasonic probe group the orientation of multiple ultrasonic probe and the scanning direction of described ultrasonic probe device to steel plate perpendicular;

Correspondingly, described Ultrasound Instrument is channel ultrasonic instrument;

Walking aids tool is the inspection car with two BOGEY WHEELs, and two BOGEY WHEELs install a scrambler respectively;

After inspection car takes ultrasonic probe group to opposite side from the side of steel plate, with one of them BOGEY WHEEL for point of fixity, another one BOGEY WHEEL rotates, thus completes the arc track route of ultrasonic probe device on steel plate.

In order to reduce the friction between ultrasonic probe and steel plate, ensureing that ultrasound wave enters steel plate smoothly, in the process of described ultrasonic probe device movement on tested steel plate, utilizing spray coupling mechanism to spray couplant at ultrasonic probe device and tested steel plate contact region.

Preferably, described host computer has human-computer interaction interface, by human-computer interaction interface input of control commands and systematic parameter.

Compared with prior art, the invention has the advantages that: the echoed signal that this steel plate defect recognition methods only obtains according to ultrasonic probe device can complete the identification of steel plate defect automatically, computing velocity is fast, fast recognition is accurate, and train without the need to a large amount of sample datas, reduce data calculated amount, accelerate recognition speed, thus reduce defect recognition cost.

Accompanying drawing explanation

Fig. 1 is the scanning wiring diagram of ultrasonic probe group in the embodiment of the present invention.

Fig. 2 is the process flow diagram of embodiment of the present invention light plate defect identification method.

Fig. 3 is the degree of depth-amplitude squiggle schematic diagram of the echoed signal that check point is corresponding.

Embodiment

Below in conjunction with accompanying drawing embodiment, the present invention is described in further detail.

The device used in steel plate defect recognition methods in the present embodiment comprises host computer, the Ultrasound Instrument be connected with upper machine communication, the ultrasonic probe device 1 be connected with Ultrasound Instrument communication, the walking aids tool be connected with upper machine communication, is arranged on walking aids tool and the scrambler be connected with upper machine communication, the spray coupling mechanism that is connected with upper machine communication.

Wherein, host computer has human-computer interaction interface, can by human-computer interaction interface input to the control command of ultrasonic probe device 1 and systematic parameter.

Ultrasonic probe device in the present embodiment 1 for be set up in parallel by eight, ultrasonic probe group that frequency is 5MHZ, wafer diameter is 40mm ultrasonic normal probe is formed, spacing between adjacent two ultrasonic probes is 50mm, then the spacing at two ultrasonic probe centers is 90mm, in ultrasonic probe group the orientation of ultrasonic probe and the scanning direction of ultrasonic probe device 1 pair of steel plate perpendicular.

Corresponding with ultrasonic probe group, Ultrasound Instrument is channel ultrasonic instrument, thus completes the data transmit-receive with each ultrasonic probe group in ultrasonic probe group.

As shown in Figure 1, ultrasonic probe group drives arc movement on steel plate by walking aids tool.

In the present embodiment, walking aids tool is the inspection car with two BOGEY WHEELs, two BOGEY WHEELs are installed respectively a scrambler detects position on steel plate coordinate data with each ultrasonic probe of Real-time Obtaining, according to the spacing between each ultrasonic probe, 8 groups of coordinate datas can be obtained simultaneously.After inspection car takes ultrasonic probe group to opposite side from the side of steel plate, with one of them BOGEY WHEEL for point of fixity, another one BOGEY WHEEL rotates, thus completes the arc track route of ultrasonic probe device 1 on steel plate.

The present embodiment for be 100mm to one piece of thickness, the testing process of width is 4000mm, length is 6000mm steel plate, the steel plate defect recognition methods in the present invention is described.Arranging the sampling interval of inspection car in advance process is that every 40mm gathers three ultrasonic signals.Then inspection car drives ultrasonic probe group on steel plate, carry out arc all standing detection, wherein on steel plate length direction, often group comprises 6000 ÷ (50+40)=66 check point, on steel plate width direction, often group comprises 4000 ÷ 40 × 3=300 check point, then the check point of ultrasonic probe group on steel plate adds up to 66 × 300=19800.

As shown in Figure 2, the steel plate defect recognition methods in the present embodiment, comprises the following steps:

Step one, Ultrasound Instrument start and initialization, and in Ultrasound Instrument, sent the controling parameters such as the velocity of sound in gain, steel plate, trigger rate and A ripple display parameter by human-computer interaction interface, wherein trigger rate is set to 5MHZ, and in steel plate, the velocity of sound is set to 5800m/s.Curvilinear abscissa that Ultrasound Instrument panel shows and actual grade value so can be made to be consistent, regulate A ripple scanning display delay, A curve line is made to show starting point before surface of steel plate echo, A ripple scanning indication range is set, make the scope >100mm of A curve line display depth distance, so just cover steel plate thickness scope, arranging A ripple scan display mode is positive wave.

Step 2, in Ultrasound Instrument, send startup command by human-computer interaction interface, then ultrasonic probe device 1, spray coupling mechanism and scrambler start work simultaneously, and wherein ultrasonic probe device 1 launches ultrasound wave according to the controling parameters in Ultrasound Instrument in tested steel plate.

Inspection car drive ultrasonic probe device 1 move on tested steel plate, simultaneously spray coupling mechanism spray water, using water as the couplant between ultrasonic probe and steel plate to ensure that ultrasonic probe and steel plate are coupled preferably.

Ultrasonic probe device 1 launches ultrasound wave to tested steel plate, receive the ultrasonic echo signal from steel plate simultaneously, wherein, ultrasonic echo signal comprises from the start signal of surface of steel plate reflection, from the flaw indication of defective locations reflection, the end ripple signal from steel plate bottom reflection.

Utilize scrambler to obtain the real-time location coordinates data of ultrasonic probe device 1 each Ultrasonic Detection point on tested steel plate simultaneously, and the coordinate data of each check point is uploaded in host computer, form coordinate data group the W=[(x of check point ₁, y ₁), (x ₂, y ₂) ..., (x _j, y _j) ..., (x _b, y _b)], wherein j and b is positive integer, and 1≤j≤b, b is check point sum, (x _j, y _j) be the position coordinates of a jth check point, x _jrepresent the value of a jth check point on steel plate length direction, y _jrepresent the value of a jth check point on steel plate width direction.

If extension position concrete on steel plate is 1800mm, when width position is 800mm, its corresponding horizontal ordinate is 1800 ÷ (50+40)=20, and corresponding ordinate is 800 ÷ 40 × 3=60, namely the corresponding coordinate of this position is (20,60).

Step 3, Ultrasound Instrument collection store the echoed signal of each check point that ultrasonic probe device 1 returns.

Step 4, Ultrasound Instrument are amplified echoed signal, filtering, analog/digital conversion process, and according to the velocity of sound in steel plate and gain, echoed signal is processed, and then for each check point position, the degree of depth-amplitude squiggle that within the scope of the corresponding tested steel plate of acquisition, echoed signal is formed, thus Formation Depth-amplitude squiggle data group A=[A ₁, A ₂..., A _j..., A _b], wherein j and b is positive integer, and 1≤j≤b, b is check point sum, A _jrepresent the degree of depth-amplitude squiggle of a jth check point.

Curve data in the degree of depth-amplitude squiggle data group A uploads in host computer by Ultrasound Instrument.

As shown in Figure 3, for start signal, ultrasound wave turns back to ultrasonic probe time t used at surface of steel plate after a vertical reflection ₁=10mm × 2 ÷ 5800m/s=3.45 μ s.

For end ripple signal, the ultrasound wave impinging perpendicularly on steel plate inside does not run into defect, directly arrives steel plate bottom surface, then is reflected back ultrasonic probe time t used ₃=(10mm+100mm) × 2 ÷ 5800m/s=37.93 μ s.

For flaw indication: all defect echo time is distributed between 3.45us and 37.93us, Ultrasound Instrument is converted by the velocity of sound of ultrasound wave in steel plate and obtains corresponding depth value accordingly between 10mm and 110mm.

Step 5, host computer carry out subsequent treatment to the degree of depth-amplitude squiggle data group A.

First, in the degree of depth-amplitude squiggle data group A, obtain start signal amplitude corresponding to each check point, thus form initialize signal amplitude data group I=[I ₁, I ₂..., I _j..., I _b], in the degree of depth-amplitude squiggle data group A, obtain end wave amplitude corresponding to each check point, thus form end wave amplitude data group D=[D ₁, D ₂..., D _j..., D _b];

Calculate initialize signal amplitude mean value $\stackrel{\&OverBar;}{I}=\frac{I_1+I_2+...+I_j+...+I_b}{b};$

Calculate end wave amplitude mean value $\stackrel{\&OverBar;}{D}=\frac{D_1+D_2+...+D_j+...+D_b}{b};$

Wherein j and b is natural number, and 1≤j≤b, b is check point sum, D _jrepresent that a jth check point is at its corresponding degree of depth-amplitude squiggle A _jthe end ripple signal amplitude of upper correspondence.

Pre-service is carried out to the degree of depth-amplitude squiggle data group A, namely corresponding to each check point in the degree of depth-amplitude squiggle data group A degree of depth-amplitude squiggle intercepts, the waveform that in surface to the bottom surface depth range retaining tested steel plate, echoed signal is corresponding, namely retain the shape information of depth value between 10mm and 110mm in its degree of depth-amplitude squiggle, thus form defective waveform curve data group B=[B ₁, B ₂..., B _j..., B _b], wherein j and b is positive integer, 1≤j≤b, B _jrepresent the defective waveform curve of a jth check point, b is check point sum.

The defective waveform curve corresponding to each check point in defective waveform curve data group B carries out derived function, thus obtains all wave crest points in corresponding check point defective waveform curve, thus builds wave crest point information data group C=[C ₁, C ₂..., C _j..., C _b]; C _j=[C _{[j] [1]}, C _{[j] [2]}], C _{[j] [1]}=(S _j0, S _j1..., S _ji..., S _ja), C _{[j] [2]}=(F _j0, F _j1..., F _ji..., F _ja).

Wherein j and b is natural number, and 1≤j≤b, b is check point sum, C _jrepresent the defective waveform curve B of a jth check point _jin all wave crest point information data groups of comprising; I and a is natural number, and 0≤i≤a, a is wave crest point sum, C _{[j] [1]}represent the depth of defect array of a jth check point, S _jifor the defective waveform curve B of a jth check point _jin depth of defect value corresponding to i-th wave crest point, C _{[j] [2]}for the defect amplitudes array of a jth check point, F _jifor the defective waveform curve B of a jth check point _jin flaw indication amplitude corresponding to i-th wave crest point.

Arrange the information data of each check point, build an information database M=[M ₁, M ₂..., M _j..., M _b], M _j=[(x _j, y _j), C _j, D _j], wherein j and b is natural number, and 1≤j≤b, b is check point sum, M _jrepresent the message data set that a jth check point is corresponding.

Step 6, treating that ultrasonic probe device 1 is to after tested steel plate scanning, is (x for arbitrary coordinate _n, y _n) check point, wherein 1≤n≤b, n and b is positive integer, b be check point sum.According to its corresponding depth of defect array C _{[n] [1]}in depth of defect value corresponding to each wave crest point, to check point (x _n, y _n) corresponding message data set M _nintegrate.

I.e. check point (x _n, y _n) corresponding defective waveform curve B _nin, in two adjacent successively wave crest points, when the absolute value that the depth of defect value that a rear wave crest point is corresponding deducts the difference of depth of defect value corresponding to previous wave crest point is less than degree of depth relevance threshold q, information data corresponding with previous wave crest point for information data corresponding for a rear wave crest point is classified as a sub-message data set.Otherwise, a newly-built sub-message data set.According to the statistics in this field, the degree of depth relevance threshold q=3mm in the present embodiment.

By that analogy, thus form check point (x _n, y _n) corresponding sub-information data cluster:

P _n=[P _n1, P _n2..., P _nm..., P _nk], wherein m and k is positive integer, and 1≤m≤k, k is check point (x _n, y _n) corresponding sub-message data set sum, P _nmrepresent check point (x _n, y _n) corresponding sub-information data cluster P _nin m sub-message data set.

For sub-information data cluster P _n=[P _n1, P _n2..., P _nm..., P _nk] in each sub-message data set, obtain its depth of defect extreme value data group L _n=[L _n1, L _n2..., L _nm..., L _nk], L _nm=(MaxC _{[nm] [1]}, MinC _{[nm] [1]}), wherein m and k is positive integer, and 1≤m≤k, k is check point (x _n, y _n) corresponding sub-message data set sum, L _nmrepresent check point (x _n, y _n) corresponding sub-information data cluster P _nin m sub-message data set P _nmin depth of defect extreme value, MaxC _{[nm] [1]}represent its depth of defect maximal value, MinC _{[nm] [1]}represent its depth of defect minimum value.

Wave crest point information data corresponding for all check points is carried out integrate the new information database M'=[P of rear formation ₁, P ₂..., P _n..., P _b], wherein n and b is natural number, and 1≤n≤b, b is check point sum, P _nrepresent the n-th check point (x _n, y _n) corresponding sub-information data cluster.

Obtain the depth of defect extreme value data set L=[L that each sub-message data set in sub-information data cluster corresponding to all check points is corresponding simultaneously ₁, L ₂..., L _n..., L _b], wherein n and b is natural number, and 1≤n≤b, b is check point sum, L _nrepresent the n-th check point (x _n, y _n) corresponding depth of defect extreme value data group.

Check point (x _n, y _n) the corresponding concrete computation process of sub-information data cluster is: check point (x _n, y _n) corresponding defective waveform curve B _nin information data corresponding to first wave peak dot be { (x _n, y _n); [S _n1, F _n1]; D _n, the information data that second wave crest point is corresponding is { (x _n, y _n); [S _n2, F _n2]; D _n, wherein S _n1, S _n1represent defective waveform curve B respectively _nin first wave peak dot and the depth of defect value of Second Wave peak dot, F _n2, F _n2represent defective waveform curve B respectively _nin first wave peak dot and the flaw indication amplitude of Second Wave peak dot.

By { (x _n, y _n); [S _n1, F _n1]; D _nas initial subdata collection P _n1in first element, if S _n2-S _n2during <q, then by information data { (x corresponding for second wave crest point _n, y _n); [S _n2, F _n2]; D _nbe included into subdata collection P _n1in.If S _n2-S _n2during>=q, then set up new subdata collection P _n2, correspondingly by information data { (x corresponding for second wave crest point _n, y _n); [S _n2, F _n2]; D _nbe included into subdata collection P _n2in.By that analogy, coordinate points (x is formed _n, y _n) corresponding sub-information data cluster P _n=[P _n1, P _n2..., P _nm..., P _nk].

Be (x for arbitrary coordinate _n, y _n) the corresponding depth of defect extreme value data group L of check point _n=[L _n1, L _n2..., L _nm..., L _nk] in each depth of defect extreme value data, wherein m and k is positive integer, and 1≤m≤k, k is check point (x _n, y _n) corresponding sub-message data set sum; Searching and detecting point (x _n, y _n) place eight territory { (x _n, y _n-1); (x _n, y _n+1); (x _n-1, y _n-1); (x _n-1, y _n); (x _n-1, y _n+1); (x _n+1, y _n-1); (x _n+1, y _n); (x _n+1, y _n+1) in each depth of defect extreme value data in the corresponding depth of defect extreme value data group of each check point, occur simultaneously if depth of defect extreme value data exist, then sub-message data set corresponding for the check point of its correspondence merged.

Be described as follows with example: coordinate is a sub-message data set P in the check point of (20,60) _xycorresponding depth of defect extreme value L _xy=(80,100).Searching and detecting point (20,60) place eight territory { (20,59), (20,61), (19,59), (19,60), (19,61), (21,59), (21,60), (21,61), if there is a sub-message data set P in the sub-information data cluster of check point (20,61) correspondence in all sub-message data set } _abcorresponding depth of defect extreme value L _ab=(90,110), L _xywith L _abexist and occur simultaneously (90,100), then by L _xyand L _abmerge into a new message data set Q _x.

So, the sub-message data set of all check points is integrated after process through search according to depth of defect extreme value data, builds new defect information database Q=[Q ₁, Q ₂..., Q _u..., Q _z], wherein u and z is positive integer, and 1≤u≤z, z is sub-defect information data set sum, Q _urepresent the sub-defect information data set of u defect.

Step 7, calculating defect information database Q=[Q ₁, Q ₂..., Q _u..., Q _z] in the corresponding defect parameters of each sub-defect information data set, thus obtain defect parameters array V=[V ₁, V ₂..., V _u..., V _z], wherein u and z is positive integer, and 1≤u≤z, z is sub-defect information data set sum, V _urepresent the defect parameters of u sub-defect information data set;

$V_u=\frac{v\&times;x\&times;y}{\underset{e=1}{\overset{v}{\mathrm{\&Sigma;}}}(\mathrm{Max}{C}_{\left[e\right]\left[1\right]}-\mathrm{Min}{C}_{\left[e\right]\left[1\right]})};$

Wherein v represents word defective data collection Q _uin the check point sum that comprises, x represents the distance between the check point center that two on steel plate length direction are adjacent, and y to represent on steel plate width direction the distance between two adjacent check point centers, x=90mm, y=40/3mm, MaxC in the present embodiment _{[e] [1]}represent defective data collection Q _uin depth of defect maximal value corresponding to e coordinate points, MinC _{[e] [1]}represent defective data collection Q _uin depth of defect minimum value corresponding to e coordinate points.

Computing information database Q=[Q simultaneously ₁, Q ₂..., Q _u..., Q _z] in the corresponding average amplitude of each sub-defect information data set, thus obtain defect average amplitude array wherein u and z is positive integer, and 1≤u≤z, z is sub-defect information data set sum, represent the average amplitude of u sub-defect information data set.

By defect parameters array V=[V ₁, V ₂..., V _u..., V _z] in each defect parameters and rarefaction defect threshold value R ₁compare, if V _u<R ₁, then defect parameters V is judged _ucorresponding sub-defective data collection Q _ucorresponding steel plate defect type is rarefaction defect.R in the present embodiment ₁=0.5, this R ₁value is experiment statistics value.

Weed out information database Q=[Q ₁, Q ₂..., Q _u..., Q _z] in message data set corresponding to rarefaction defect, by defect parameters corresponding for remaining each message data set and average amplitude respectively with lamination defect threshold value R ₂and amplitude thresholds f compares, if R ₂<V _u<1 and then judge defect parameters V _ucorresponding sub-defective data collection Q _ucorresponding steel plate defect type is lamination defect.R in the present embodiment ₂=0.8,

Step 8, weed out information database Q=[Q ₁, Q ₂..., Q _u..., Q _z] in the rarefaction defect defect information data set corresponding with lamination defect, remaining each sub-defect information data set is carried out pseudo color image reconstruct on two dimensional surface, form corresponding subset image, then carry out morphological analysis for subset image, thus acquisition steel plate defect type is crack defect or gas hole defect;

The nearly circularity T calculating each subset image, to judge the gas hole defect of steel plate, calculates the flexibility G of each subset image to judge the crack defect of steel plate;

Wherein T=S/ (π × d ²/ 4), wherein T represents nearly circularity, and S represents that the area that subset image comprises, d to represent in subset image the distance of 2 farthest, as nearly circularity T≤t, then think the corresponding subdata of this subset image integrate corresponding to steel plate defect as porous defect; Wherein t is nearly roundness threshold, t=0.8 in the present embodiment.

G=S/ (h × l), wherein I represents flexibility, S represents the area occupied by subset image, h to represent in subset image the distance of 2 farthest, l represent in subset image with the mean breadth on perpendicular direction, h direction, as flexibility G≤g, then judge the corresponding subdata of this subset image integrate corresponding to steel plate defect as crack-type defect; Wherein, g is flexibility threshold value, g=0.2 in the present embodiment.

Claims (5)

1. a steel plate defect recognition methods, is characterized in that: comprise the following steps:

Step one, Ultrasound Instrument start and initialization, and host computer sends controling parameters to Ultrasound Instrument;

Step 2, ultrasonic probe device (1) start and carry out work according to the controling parameters in Ultrasound Instrument, ultrasonic probe device (1) moves on tested steel plate, host computer obtains ultrasonic probe device (1) to the real-time location coordinates data of steel plate check point, forms coordinate data group the W=[(x of check point ₁, y ₁), (x ₂, y ₂) ..., (x _j, y _j) ..., (x _b, y _b)], wherein j and b is positive integer, and 1≤j≤b, b is check point sum, (x _j, y _j) be the position coordinates of a jth check point, x _jrepresent the value of a jth check point on steel plate length direction, y _jrepresent the value of a jth check point on steel plate width direction;

Ultrasonic probe device (1) launches ultrasound wave to tested steel plate, receive the ultrasonic echo signal from steel plate simultaneously, wherein, ultrasonic echo signal comprises from the start signal of surface of steel plate reflection, from the flaw indication of defective locations reflection, the end ripple signal from steel plate bottom reflection;

Step 3, Ultrasound Instrument collection store the echoed signal of each check point that ultrasonic probe device (1) returns;

Step 4, Ultrasound Instrument process echoed signal according to the velocity of sound in steel plate and gain, and then for each check point position, the degree of depth-amplitude squiggle that within the scope of the corresponding tested steel plate of acquisition, echoed signal is formed, thus Formation Depth-amplitude squiggle data group A=[A ₁, A ₂..., A _j..., A _b], wherein j and b is positive integer, and 1≤j≤b, b is check point sum, A _jrepresent the degree of depth-amplitude squiggle of a jth check point;

Curve data in the degree of depth-amplitude squiggle data group A uploads in host computer by Ultrasound Instrument;

Step 5, host computer carry out subsequent treatment to the degree of depth-amplitude squiggle data group A;

First, in the degree of depth-amplitude squiggle data group A, obtain start signal amplitude corresponding to each check point, thus form initialize signal amplitude data group I=[I ₁, I ₂..., I _j..., I _b], in the degree of depth-amplitude squiggle data group A, obtain end wave amplitude corresponding to each check point, thus form end wave amplitude data group D=[D ₁, D ₂..., D _j..., D _b];

Calculate initialize signal amplitude mean value $\stackrel{\&OverBar;}{I}=\frac{I_1+I_2+...+I_j+...+I_b}{b};$

Calculate end wave amplitude mean value $\stackrel{\&OverBar;}{D}=\frac{D_1+D_2+...+D_j+...+D_b}{b};$

Wherein j and b is natural number, and 1≤j≤b, b is check point sum, D _jrepresent that a jth check point is at its corresponding degree of depth-amplitude squiggle A _jthe end ripple signal amplitude of upper correspondence;

Pre-service is carried out to the degree of depth-amplitude squiggle data group A, namely corresponding to each check point in the degree of depth-amplitude squiggle data group A degree of depth-amplitude squiggle intercepts, the waveform that in surface to the bottom surface depth range retaining tested steel plate, echoed signal is corresponding, thus form defective waveform curve data group B=[B ₁, B ₂..., B _j..., B _b], wherein j and b is positive integer, 1≤j≤b, B _jrepresent the defective waveform curve of a jth check point, b is check point sum;

The defective waveform curve corresponding to each check point in defective waveform curve data group B carries out derived function, thus obtains all wave crest points in corresponding check point defective waveform curve, thus builds wave crest point information data group C=[C ₁, C ₂..., C _j..., C _b]; C _j=[C _{[j] [1]}, C _{[j] [2]}], C _{[j] [1]}=(S _j0, S _j1..., S _ji..., S _ja), C _{[j] [2]}=(F _j0, F _j1..., F _ji..., F _ja);

Wherein j and b is natural number, and 1≤j≤b, b is check point sum, C _jrepresent the defective waveform curve B of a jth check point _jin all wave crest point information data groups of comprising; I and a is natural number, and 0≤i≤a, a is wave crest point sum, C _{[j] [1]}represent the depth of defect array of a jth check point, S _jifor the defective waveform curve B of a jth check point _jin depth of defect value corresponding to i-th wave crest point, C _{[j] [2]}for the defect amplitudes array of a jth check point, F _jifor the defective waveform curve B of a jth check point _jin flaw indication amplitude corresponding to i-th wave crest point;

Arrange the information data of each check point, build an information database M=[M ₁, M ₂..., M _j..., M _b], M _j=[(x _j, y _j), C _j, D _j], wherein j and b is natural number, and 1≤j≤b, b is check point sum, M _jrepresent the message data set that a jth check point is corresponding;

Step 6, treating that ultrasonic probe device (1) is to after tested steel plate scanning, is (x for arbitrary coordinate _n, y _n) check point, wherein 1≤n≤b, n and b is positive integer, b be check point sum;

According to its corresponding depth of defect array C _{[n] [1]}in depth of defect value corresponding to each wave crest point, to check point (x _n, y _n) corresponding message data set M _nintegrate;

I.e. check point (x _n, y _n) corresponding defective waveform curve B _nin, in two adjacent successively wave crest points, when the absolute value that the depth of defect value that a rear wave crest point is corresponding deducts the difference of depth of defect value corresponding to previous wave crest point is less than degree of depth relevance threshold q, information data corresponding with previous wave crest point for information data corresponding for a rear wave crest point is classified as a sub-message data set; Otherwise, a newly-built sub-message data set;

By that analogy, thus form check point (x _n, y _n) corresponding sub-information data cluster:

P _n=[P _n1, P _n2..., P _nm..., P _nk], wherein m and k is positive integer, and 1≤m≤k, k is check point (x _n, y _n) corresponding sub-message data set sum, P _nmrepresent check point (x _n, y _n) corresponding sub-information data cluster P _nin m sub-message data set;

For sub-information data cluster P _n=[P _n1, P _n2..., P _nm..., P _nk] in each sub-message data set, obtain its depth of defect extreme value data group L _n=[L _n1, L _n2..., L _nm..., L _nk], L _nm=(MaxC _{[nm] [1]}, MinC _{[nm] [1]}), wherein m and k is positive integer, and 1≤m≤k, k is check point (x _n, y _n) corresponding sub-message data set sum, L _nmrepresent check point (x _n, y _n) corresponding sub-information data cluster P _nin m sub-message data set P _nmin depth of defect extreme value, MaxC _{[nm] [1]}represent its depth of defect maximal value, MinC _{[nm] [1]}represent its depth of defect minimum value;

Wave crest point information data corresponding for all check points is carried out integrate the new information database M'=[P of rear formation ₁, P ₂..., P _n..., P _b], wherein n and b is natural number, and 1≤n≤b, b is check point sum, P _nrepresent the n-th check point (x _n, y _n) corresponding sub-information data cluster;

Obtain the depth of defect extreme value data set L=[L that each sub-message data set in sub-information data cluster corresponding to all check points is corresponding simultaneously ₁, L ₂..., L _n..., L _b], wherein n and b is natural number, and 1≤n≤b, b is check point sum, L _nrepresent the n-th check point (x _n, y _n) corresponding depth of defect extreme value data group;

Be (x for arbitrary coordinate _n, y _n) the corresponding depth of defect extreme value data group L of check point _n=[L _n1, L _n2..., L _nm..., L _nk] in each depth of defect extreme value data, wherein m and k is positive integer, and 1≤m≤k, k is check point (x _n, y _n) corresponding sub-message data set sum; Searching and detecting point (x _n, y _n) place eight territory { (x _n, y _n-1); (x _n, y _n+1); (x _n-1, y _n-1); (x _n-1, y _n); (x _n-1, y _n+1); (x _n+1, y _n-1); (x _n+1, y _n); (x _n+1, y _n+1) in each depth of defect extreme value data in the corresponding depth of defect extreme value data group of each check point, occur simultaneously if depth of defect extreme value data exist, then sub-message data set corresponding for the check point of its correspondence merged;

So, the sub-message data set of all check points is integrated after process through search according to depth of defect extreme value data, builds new defect information database Q=[Q ₁, Q ₂..., Q _u..., Q _z], wherein u and z is positive integer, and 1≤u≤z, z is sub-defect information data set sum, Q _urepresent the sub-defect information data set of u defect;

Step 7, calculating defect information database Q=[Q ₁, Q ₂..., Q _u..., Q _z] in the corresponding defect parameters of each sub-defect information data set, thus obtain defect parameters array V=[V ₁, V ₂..., V _u..., V _z], wherein u and z is positive integer, and 1≤u≤z, z is sub-defect information data set sum, V _urepresent the defect parameters of u sub-defect information data set;

$V_u=\frac{v\&times;x\&times;y}{\underset{e=1}{\overset{v}{\mathrm{\&Sigma;}}}(\mathrm{Max}{C}_{\left[e\right]\left[1\right]}-\mathrm{Min}{C}_{\left[e\right]\left[1\right]})};$

Wherein v represents word defective data collection Q _uin the check point sum that comprises, x represents the distance between the check point center that two on steel plate length direction are adjacent, and y to represent on steel plate width direction the distance between two adjacent check point centers, MaxC _{[e] [1]}represent defective data collection Q _uin depth of defect maximal value corresponding to e coordinate points, MinC _{[e] [1]}represent defective data collection Q _uin depth of defect minimum value corresponding to e coordinate points;

Computing information database simultaneously in the corresponding average amplitude of each sub-defect information data set, thus obtain defect average amplitude array wherein u and z is positive integer, and 1≤u≤z, z is sub-defect information data set sum, represent the average amplitude of u sub-defect information data set;

By defect parameters array V=[V ₁, V ₂..., V _u..., V _z] in each defect parameters and rarefaction defect threshold value R ₁compare, if V _u<R ₁, then defect parameters V is judged _ucorresponding sub-defective data collection Q _ucorresponding steel plate defect type is rarefaction defect; .

Weed out information database Q=[Q ₁, Q ₂..., Q _u..., Q _z] in message data set corresponding to rarefaction defect, by defect parameters corresponding for remaining each message data set and average amplitude respectively with lamination defect threshold value R ₂and amplitude thresholds f compares, if R ₂<V _u<1 and then judge defect parameters V _ucorresponding sub-defective data collection Q _ucorresponding steel plate defect type is lamination defect;

Step 8, weed out information database Q=[Q ₁, Q ₂..., Q _u..., Q _z] in the rarefaction defect defect information data set corresponding with lamination defect, remaining each sub-defect information data set is carried out pseudo color image reconstruct on two dimensional surface, form corresponding subset image, then carry out morphological analysis for subset image, thus acquisition steel plate defect type is crack defect or gas hole defect;

The nearly circularity T calculating each subset image, to judge the gas hole defect of steel plate, calculates the flexibility G of each subset image to judge the crack defect of steel plate;

Wherein T=S/ (π × d ²/ 4), wherein T represents nearly circularity, and S represents that the area that subset image comprises, d to represent in subset image the distance of 2 farthest, as nearly circularity T≤t, then think the corresponding subdata of this subset image integrate corresponding to steel plate defect as porous defect;

G=S/ (h × l), wherein I represents flexibility, S represents the area occupied by subset image, h to represent in subset image the distance of 2 farthest, l represent in subset image with the mean breadth on perpendicular direction, h direction, as flexibility G≤g, then judge the corresponding subdata of this subset image integrate corresponding to steel plate defect as crack-type defect.

2. steel plate defect recognition methods according to claim 1, it is characterized in that: in described step 2, ultrasonic probe device (1) drives arc movement on steel plate by walking aids tool, and described scrambler is arranged on described walking aids tool with the coordinate data of Real-time Obtaining ultrasonic probe device (1) check point on steel plate.

3. steel plate defect recognition methods according to claim 2, it is characterized in that: described ultrasonic probe device (1) comprises multiple ultrasonic probe be set up in parallel, form ultrasonic probe group, in described ultrasonic probe group the orientation of multiple ultrasonic probe and the scanning direction of described ultrasonic probe device (1) to steel plate perpendicular;

Correspondingly, described Ultrasound Instrument is channel ultrasonic instrument;

Walking aids tool is the inspection car with two BOGEY WHEELs, and two BOGEY WHEELs install a scrambler respectively;

After inspection car takes ultrasonic probe group to opposite side from the side of steel plate, with one of them BOGEY WHEEL for point of fixity, another one BOGEY WHEEL rotates, thus completes the arc track route of ultrasonic probe device (1) on steel plate.

4. steel plate defect recognition methods according to claim 2, it is characterized in that: in the process of described ultrasonic probe device (1) movement on tested steel plate, utilize spray coupling mechanism to spray couplant in ultrasonic probe device (1) and tested steel plate contact region.

5. steel plate defect recognition methods according to claim 1, is characterized in that: described host computer has human-computer interaction interface, by human-computer interaction interface input of control commands and systematic parameter.