CN104897777A - Method for improving longitudinal resolution of TOFD (time of flight diffraction) detection with Burg algorithm based autoregressive spectrum extrapolation technology - Google Patents

Abstract

The invention discloses a method for improving longitudinal resolution of TOFD (time of flight diffraction) detection with a Burg algorithm based autoregressive spectrum extrapolation technology, and belongs to the technical field of ultrasonic non-destructive detection. The method adopts an ultrasonic detection system comprising a TOFD ultrasonic detection instrument, TOFD probes, a scanning device, software integrating TOFD conventional analysis functions and a computer. According to the method, TOFD scanning is performed on defects, collected time-domain aliasing signals containing upper-end and lower-end diffraction waves of the defects are subjected to autoregressive spectrum extrapolation processing; the -6dB frequency bandwidth is taken as the benchmark, defect signals in the frequency band range are selected, defect signals outside the frequency band range are estimated with the Burg algorithm, so that the frequency bandwidth is broadened, the longitudinal resolution of defect detection is improved, and the defect height quantification is realized. Compared with other methods for improving the longitudinal resolution of TOFD detection, the method does not have additional requirements for a hardware system, is not limited by exciting pulse time width of the detecting probes, and has higher engineering application value.

Description

Autoregressive spectrum extrapolation technique based on Burg algorithm improves the method that TOFD detects longitudinal frame

Technical field

The present invention relates to a kind of autoregressive spectrum extrapolation technique based on Burg algorithm and improve the method that TOFD detects longitudinal frame, it belongs to field of ultrasonic nondestructive detection.

Background technology

For the physical integrity of effective evaluation detected object, just must detect wherein all defect, and accurate location, quantitatively.At present, the ultrasonic diffraction time difference (Time of Flight Diffraction, TOFD) technology is one of the most accurate Ultrasonic Detection and quantivative approach, is that to utilize the diffracted signal propagation time difference of defect upper and lower side to realize flaw height quantitative.Compared with traditional pulse echo method, when TOFD technology detects vertical crack defect, have more advantage, the accurate quantification of flaw height can be realized.

Detecting longitudinal frame is one of key factor affecting TOFD quantitative accuracy.Detect longitudinal frame for improving TOFD further, existing method comprises parameter optimization method, image energy distribution and Spectral Analysis Method etc.TOFD detection resolution can be improved to a certain extent by parameter optimization, but the factor restrictions such as examined frequency probe, depth of defect, and longitudinal frame improves limited; In image energy distribution, the extraction effect of flaw indication affects comparatively large by pending picture quality, the often problem of existing defects Signal separator difficulty; Spectral Analysis Method limits, when defect upper and lower side propagation time difference is less than by probe frequency span time (f is frequency probe), the method is no longer applicable.

Summary of the invention

The object of this invention is to provide a kind of autoregressive spectrum extrapolation technique based on Burg algorithm and improve the method that TOFD detects longitudinal frame.The defect upper and lower side diffracted signal Aliasing Problem that in detecting for TOFD, longitudinal frame deficiency causes, utilize a kind of autoregressive spectrum extrapolation technique based on Burg algorithm, by widening flaw echoes frequency span, realize the separation of aliasing diffracted wave signal in time domain, thus improve detection signal longitudinal frame, realize flaw height quantitative.

The technical solution used in the present invention is: a kind of autoregressive spectrum extrapolation technique based on Burg algorithm improves the method that TOFD detects longitudinal frame, it is characterized in that: with-6dB the frequency span of system pulses signal for benchmark, choose the flaw indication in this frequency band range, utilize the flaw indication that Burg algorithm is estimated outside this frequency band range, thus widen frequency span, improve the longitudinal frame of defects detection, realize flaw height quantitative, the measuring process of described method is as follows:

A () is first carried out TOFD to examined workpiece and is detected D scanning, tentatively determine defective locations, select suitable frequency probe, head angle according to depth of defect information, and adjust the parameters such as center probe spacing, time window scope, detection sensitivity, pulse repetition rate and scanning increment;

B () carries out B scanning according to the TOFD detected parameters determined in step (a) to the target defect in examined workpiece, and store B scan image, by TOFD analysis software, A sweep signal y (t) at para-curve summit place in B scan image is derived;

C () utilizes the reference block identical with examined workpiece material, obtain its bottom reflection echoed signal, it can be used as system pulses signal, the i.e. reference signal h (t) of autoregressive spectrum extrapolation process, respectively Fourier transform is carried out to detection signal y (t) and reference signal h (t), utilize the frequency-domain expression Y (ω) of detection signal y (t) directly divided by the frequency-domain expression H (ω) of system pulses signal h (t), obtain the estimated signal X (ω) of defect, namely

$X\left(\mathrm{\ω}\right)\≅\frac{Y\left(\mathrm{\ω}\right)}{H\left(\mathrm{\ω}\right)}=\frac{Y\left(\mathrm{\ω}\right)H^{*}\left(\mathrm{\ω}\right)}{{\left|H\left(\mathrm{\ω}\right)\right|}^2}---\left(1\right)$

H in formula ^*(ω) be the conjugate complex number of H (ω);

D () is according to the determined frequency band window of system pulses signal spectrum maximum amplitude decline 6dB [ω _l, ω _h], obtain the data area [i corresponding with this window _l, i _h], choose flaw indication X (ω) the conduct reference model in this data area, utilize Final prediction error criterion Confirming model exponent number p, use Burg algorithm to try to achieve autoregressive spectrum extrapolation coefficient a _k;

E () utilizes forward prediction formula (2) and back forecast formula (3) to estimate i>i respectively _hand i<i _lx in scope _i(ω) estimated value,

${\tilde{X}}_i\left(\mathrm{\&omega;}\right)=-\underset{k=1}{\overset{p}{\mathrm{\&Sigma;}}}{a}_k{X}_{i-k}\left(\mathrm{\&omega;}\right),i>i_H---\left(2\right)$

${\tilde{X}}_i\left(\mathrm{\&omega;}\right)=-\underset{k=1}{\overset{p}{\mathrm{\&Sigma;}}}{a}_k^{*}{X}_{i-k}\left(\mathrm{\&omega;}\right),i<i_L---\left(3\right)$

In formula for extrapolated value, it is the conjugate complex number of autoregressive spectrum extrapolation coefficient;

F () is by frequency band window [ω _l, ω _h] data in outer data and frequency band window combine, and obtain flaw indication frequency spectrum to flaw indication frequency spectrum carry out inverse Fourier transform, obtain its time-domain signal read the mistiming Δ t of defect upper and lower side diffracted wave signal, according to formula (4), the height h of defect can be calculated,

$h=\frac{1}{2}(\sqrt{c^2{(t+\mathrm{\&Delta;t})}^2-{4S}^2}-\sqrt{c^2{t}^2-4S^2})---\left(4\right)$

In formula, c is examined workpiece longitudinal wave velocity, and S is center probe spacing half, and t is the defect upper end diffracted wave travel-time, and Δ t is defect upper and lower side diffracted wave propagation time difference.

The invention has the beneficial effects as follows: compared with detecting the method for longitudinal frame with the existing TOFD of raising, autoregressive spectrum extrapolation technique based on Burg algorithm passes through to estimate the flaw indication outside-6dB frequency band range, thus widen frequency span, overcome the restriction of the factors such as detection probe frequency span, improve TOFD and detect longitudinal frame.On this basis, the present invention can be separated aliasing diffracted wave signal in time domain, realize quantitative, the location of defect, and flaw height quantitative error is no more than 10%.Meanwhile, the present invention is workable, and to hardware system without extra demand, has good engineer applied and be worth and higher economic benefit.

Accompanying drawing explanation

Below in conjunction with accompanying drawing and example, the present invention will be further described.

Fig. 1 is TOFD ultrasonic test system connection diagram.

Fig. 2 is examined workpiece defect distribution and TOFD probe placement schematic diagram.

Fig. 3 is the B scan image of height 1.00mm crackle.

Fig. 4 is system impulse response signal, (a) time domain waveform h (t), (b) spectrum H (ω).

Fig. 5 is the spectrum extrapolating results of height 1.00mm crackle A sweep signal.

Embodiment

Autoregressive spectrum extrapolation technique based on Burg algorithm improves TOFD and detects in the method for longitudinal frame, adopt ultrasonic test system as shown in Figure 1, comprising the computing machine of TOFD ultrasound measuring instrument, integrated conventional analysis functional software, the TOFD probe, scanning equipment etc. of nominal frequency 5MHz.Measurement and the treatment step of its employing are as follows:

A () research object is carbon steel coupons, test block size 800mm × 500mm × 250mm, and wherein crackle wall thickness direction height is 1.00mm, and crackle lower end is 48.5mm apart from the detection faces degree of depth, material longitudinal wave velocity 5935m/s.Adopt TOFD ultrasound measuring instrument, select frequency probe 5MHz, wafer size 6mm, incident angle is two TOFD probes of 45 °, and arranges center probe spacing 2S=100mm (see Fig. 2).Before A sweep time window reference position is set to straight-through ripple arrival receiving transducer, final position is after flaw indication arrives receiving transducer, sample frequency 100MHz.

B a pair TOFD probe symmetry as shown in Figure 2, is positioned over above examined workpiece defect and carries out B scanning, by TOFD analysis software, B is scanned the time-domain signal derivation at para-curve summit place, be used for carrying out signal subsequent treatment by ().For the crackle of height 1.00mm, its B scanning result as shown in Figure 3.As can be seen from Figure 3, defect upper end and defect lower end echoed signal generation aliasing, can not directly read upper and lower side propagation time difference, cannot determine flaw height.

C (), in the reference block identical with examined workpiece material, utilizes single probe to obtain bottom reflection echoed signal, it can be used as system pulses signal, i.e. reference signal h (t) (as Suo Shi Fig. 4 (a)).Fourier transform is done to detection signal and system pulses signal, with the direct frequency-domain expression H (ω) divided by system pulses signal h (t) of the frequency-domain expression Y (ω) of detection signal y (t), obtain the estimated signal X (ω) of defect.

D according to system pulses signal, () determines that-6dB frequency band window is 1.7MHz-7.1MHz, its corresponding data point is 15-55, namely chooses the 15th data to the 55th point in flaw indication.According to FPE order selection criteria Confirming model exponent number p=15, Burg algorithm is utilized to calculate autoregressive spectrum extrapolation coefficient a _k(see table 1), and utilize formula (2) and formula (3) interpolation to estimate data in low frequency (f < 1.7MHz) and high frequency (f > 7.1MHz) scope.

Extrapolation coefficient composed by table 1

a 1	a 2	a 3	a 4	a 5	a 6	a 7	a 8
								1.0000	0.9857	1.0040	0.3593	0.4632	0.7899	0.3146	0.0429
a 9	a 10	a 11	a 12	a 13	a 14	a 15	a 16
								-0.5233	0.4854	-0.3848	-0.5224	-0.4577	-0.6075	-0.2690	-0.0184

E () carries out inverse Fourier transform to expanding the flaw indication obtained, obtain defect pulse signal time domain estimated value (as Fig. 5).After autoregressive spectrum extrapolation process, crackle upper and lower side aliasing diffracted signal is separated.From Fig. 3, directly read defect upper end degree of depth 47.36mm, from Fig. 5, read crackle upper and lower side echo time difference is 0.2674 μ s, and it is 48.50mm that substitution formula (4) can be calculated the defect lower end degree of depth.Therefore, flaw height quantitative result is 1.10mm, and its absolute error is 0.10mm, and relative error is 10.0%, can meet engineering demand.

Claims (1)

1. the autoregressive spectrum extrapolation technique based on Burg algorithm improves the method that TOFD detects longitudinal frame, it is characterized in that: with-6dB the frequency span of system pulses signal for benchmark, choose the flaw indication in this frequency band range, utilize the flaw indication that Burg algorithm is estimated outside this frequency band range, thus widen frequency span, improve the longitudinal frame of defects detection, realize flaw height quantitative, the measuring process of described method is as follows:

A () is first carried out TOFD to examined workpiece and is detected D scanning, tentatively determine defective locations, select suitable frequency probe, head angle according to depth of defect information, and adjust the parameters such as center probe spacing, time window scope, detection sensitivity, pulse repetition rate and scanning increment;

B () carries out B scanning according to the TOFD detected parameters determined in step (a) to the target defect in examined workpiece, and store B scan image, by TOFD analysis software, A sweep signal y (t) at para-curve summit place in B scan image is derived;

C () utilizes the reference block identical with examined workpiece material, obtain its bottom reflection echoed signal, it can be used as system pulses signal, the i.e. reference signal h (t) of autoregressive spectrum extrapolation process, respectively Fourier transform is carried out to detection signal y (t) and reference signal h (t), utilize the frequency-domain expression Y (ω) of detection signal y (t) directly divided by the frequency-domain expression H (ω) of system pulses signal h (t), obtain the estimated signal X (ω) of defect, namely

$X\left(\mathrm{\&omega;}\right)\widetilde{=}\frac{Y\left(\mathrm{\&omega;}\right)}{H\left(\mathrm{\&omega;}\right)}=\frac{Y\left(\mathrm{\&omega;}\right)H^{*}\left(\mathrm{\&omega;}\right)}{{\left|H\left(\mathrm{\&omega;}\right)\right|}^2}---\left(1\right)$

H in formula ^*(ω) be the conjugate complex number of H (ω);

D () is according to the determined frequency band window of system pulses signal spectrum maximum amplitude decline 6dB [ω _l, ω _h], obtain the data area [i corresponding with this window _l, i _h], choose flaw indication X (ω) the conduct reference model in this data area, utilize Final prediction error criterion Confirming model exponent number p, use Burg algorithm to try to achieve autoregressive spectrum extrapolation coefficient a _k;

E () utilizes forward prediction formula (2) and back forecast formula (3) to estimate i>i respectively _hand i<i _lx in scope _i(ω) estimated value,

${\tilde{X}}_i\left(\mathrm{\&omega;}\right)=-\underset{k=1}{\overset{p}{\mathrm{\&Sigma;}}}{a}_k{X}_{i-k}\left(\mathrm{\&omega;}\right),i>i_H---\left(2\right)$

${\tilde{X}}_i\left(\mathrm{\&omega;}\right)=-\underset{k=1}{\overset{p}{\mathrm{\&Sigma;}}}{a}_k^{*}{X}_{i-k}\left(\mathrm{\&omega;}\right),i<i_L---\left(3\right)$

In formula for extrapolated value, it is the conjugate complex number of autoregressive spectrum extrapolation coefficient;

F () is by frequency band window [ω _l, ω _h] data in outer data and frequency band window combine, and obtain flaw indication frequency spectrum to flaw indication frequency spectrum carry out inverse Fourier transform, obtain its time-domain signal read the mistiming Δ t of defect upper and lower side diffracted wave signal, according to formula (4), the height h of defect can be calculated,

$h=\frac{1}{2}(\sqrt{c^2{(t+\mathrm{\&Delta;t})}^2-{4S}^2}-\sqrt{c^2{t}^2-{4S}^2})---\left(4\right)$

In formula, c is examined workpiece longitudinal wave velocity, and S is center probe spacing half, and t is the defect upper end diffracted wave travel-time, and Δ t is defect upper and lower side diffracted wave propagation time difference.