CN104376881A - Nuclear power plant loose part positioning method based on Hilbert transform and data screening - Google Patents

Abstract

The invention discloses a nuclear power plant loose part positioning method based on Hilbert transform and data screening. The method comprises the steps of dividing grids on an impacted object; mounting acceleration sensors; calculating a distance different between a distance from the center point of each grid to one acceleration sensor and a distance from the center point of each grid to another acceleration sensor; acquiring impact signals received by all the acceleration sensors; eliminating low-frequency noise of the signals by adopting a moving average method, and filtering environmental background noise by a Butterworth filter; performing Hilbert transform on the impact signals subjected to the noise elimination to obtain oscillation starting point moments of the impact signals; calculating a nominal transmission speed of the center point of each grid according to a transmission distance difference from the center point of the corresponding grid to the sensors and arrival time differences of all channels; and performing data screening on the nominal transmission speeds of the center points of all the grids, and calculating a variance; and finding out a minimum value in all the variances, recording coordinates of the center point, and displaying a positioning result. The nuclear power plant loose part positioning method has the advantages of high positioning precision and high searching speed.

Description

Based on the positioning method for loosening member of nuclear power station of Hilbert transform and data screening

Technical field

The present invention relates to nuclear power station technical field, be specifically related to the positioning method for loosening member of nuclear power station based on Hilbert transform and data screening.

Technical background

There is the web members such as a large amount of screws, nut in nuclear power station, due to the continuous impact of high-velocity flow, occur burn into depreciation and loosen even fall, also have in system testing, refuel, the overhaul stage enters system metal fragment from the external world, this all can make the stability of system cloud gray model and reliability reduce, and even has influence on the safety of whole nuclear power station.Loosening element location is as the important component part of loose parts monitoring system, locate loosening element to be exactly conducive to finding loosening element fast when Shutdown, and repair accordingly, the minimizing maintenance personal that tries one's best is exposed to the time under nuclear radiation, ensure the safety of maintenance personal, the stability of nuclear power station and security are very helpful.

Existing loosening element location pertinent literature has:

G..Por, J.Kiss, I.Sorosanszky, G..Szappanos, Development of afalse alarm free advanced loose parts monitoring system (ALPS) [J], Progress in Nuclear Energy, 2003,43 (1-4): 243-251. mono-kind are based on signal SPRT (Sequence Probability Ratio Test, SPR sequential probability ratio test) time difference estimation method, first carry out albefaction with the AR model of noise to signal, the SPRT then by calculating whitened signal carrys out estimating signal step-out time.

Yong Beum Kim, Seon Jae Kim, Hae Dong Chung, Yong Won Park, Jin Ho Park, A Study on Technique to Estimate Impact Location of LoosePart Using Wigner-Ville Distribution [J], Progress in Nuclear Energy, 2003,43 (1-4): 261-266. mono-kind are based on the Study on estimation method of the loosening element impact position of Wigner-Willie distribution, propose and Wigner-Willie conversion is carried out to signal, obtain time-frequency figure, and then obtain loosening element falls position.The method accuracy is high, but calculated amount is very large.The time-frequency domain line of each signal is different, and needs manual drafting, is unfavorable for realizing robotization.

S.Figedy, G..Oksa, Modern methods of signal processing in theloose part monitoring system [J], Progress in Nuclear Energy, 2005,46 (3-4): 253-267. mono-kind, based on the time difference estimation method of Wavelet Denoising Method, pass through Wavelet Denoising Method, namely remove the impact of noise, and then estimate the time difference.The method is owing to eliminating the impact of noise, effect is estimated preferably so still have when low signal-to-noise ratio, but because the method is still using the zero crossing of signal as time of arrival (toa), do not consider the complicated communication mode of flexural wave, so still there is comparatively big error when actual location when propagation distance is far away.

Seong-Nam Jeong, Kyoung-Hang Woo, Eoun-Taeg Hwang, Won-HoChoi, A Study on the Estimation Method of Impact PositionUsing the Frequency Analysis [J], Strategic Technology.The 1stInternational Forum on, 2006:392-395. mono-kind, based on the impact position Study on estimation method of frequency analysis, proposes a kind of method for positioning loosening element based on frequency dispersion.The method accuracy is high, and stability is also higher, but early stage needs to set up fairly perfect database, and workload is large.

Summary of the invention

In order to achieve the above object, the technical scheme that the present invention takes is:

Based on the positioning method for loosening member of nuclear power station of Hilbert transform and data screening, comprise the following steps:

1) according to positioning accuracy request, grid division on thing is being knocked, then to ready-portioned grid numbering 1 ~ N;

2) be knocked by equilateral triangle layout installation 3 acceleration transducers on thing, shock signal f (t) produced during to obtain loosening element falls;

3) according to the range difference d of each grid element center point of geometry computations to each acceleration transducer between two that are knocked thing _i,j;

4) by shock signal f (t) that each acceleration transducer of data collecting card synchronous acquisition receives, impact signal s (t) when signal f (t) comprises loosening element falls and environmental background noise n (t) is clashed into;

5) adopt the method for moving average to eliminate the low-frequency noise of clashing into signal f (t), then adopt 8 rank Butterworth wave filters by environmental background noise n (t) filtering, be eliminated impact signal s (t) of noise;

6) to step 5) in process impact signal s (t) that obtains and analyze with Hilbert transform, obtain the oscillation starting points moment of impact signal s (t); Owing to cannot know the moment that impact signal s (t) falls, can only obtain the mistiming of wave traveling from impact signal s (t) of 3 acceleration transducers, difference time of arrival between every two passages is t _i,j;

7) according to the propagation distance difference d of grid element center point to each acceleration transducer _i,jand difference t time of arrival between each passage _i,jcalculate the nominal velocity of propagation v of each grid element center point _i,j;

8) the speed bound v that flexural wave is propagated in the structure is calculated _maxand v _min;

9) to the nominal velocity of propagation v of all grid element center points _i,jcarry out data screening: judge nominal velocity of propagation v _i,jspeed interval [the v whether propagated at flexural wave _max, v _min] in, if the nominal velocity of propagation v of grid element center point _i,jall at speed interval [v _max, v _min] in, then calculate the variance D (v) of the speed of this grid element center point, otherwise compose with a large value to variance D (v);

10) search for minimum value in the variance D (v) of all grid element center point, record the coordinate of its central point;

11) positioning result display.

Described step 3) in, for plane, according to the range difference of formula (1) computing grid center to each acceleration transducer:

$d_{i,j}=\sqrt{{(x-x_i)}^2+{(y-y_i)}^2}-\sqrt{{(x-x_j)}^2+{(y-y_i)}^2}---\left(1\right)$

Wherein (x, y) is the coordinate of grid element center point, (x _i, y _i) be the coordinate of acceleration transducer i, (x _j, y _j) be the coordinate of acceleration transducer j, i=1,2,3; J=1,2,3; I ≠ j;

For hemisphere face, according to the range difference of formula (2) computing grid center to each acceleration transducer:

Wherein r is the radius of ball, for the spherical coordinates of grid element center point, for the spherical coordinates of acceleration transducer i, for the spherical coordinates of acceleration transducer j, i=1,2,3; J=1,2,3; I ≠ j.

Described step 5) in, eliminate according to formula (3) and clash into signal f (t) low-frequency noise:

$f^{\′}\left(t\right)=\frac{(f\left(t\right)+f(t-1)+...f(t-n+1))}{n}---\left(3\right)$

Wherein f'(t) be the shock signal after moving average, n is the number in period of moving average, and n is set to 7.

Described step 6) in, determine impact signal s (t) the oscillation starting points moment according to the peak value of envelope, specifically comprise the following steps:

6.1) Hilbert transform of impact signal s (t) is asked according to definition:

$\hat{s}\left(t\right)={\∫}_{-\∞}^{\∞}s\left(t\right)\frac{1}{\mathrm{\π}(t-\mathrm{\τ})}\mathrm{d\τ}---\left(4\right)$

Wherein, for the hubert transformed signal of impact signal s (t);

6.2) with impact signal s (t) for real part, its hubert transformed signal for imaginary part, form a new function as formula (5):

$z\left(t\right)=s\left(t\right)+j\hat{s}\left(t\right)=\left|z\left(t\right)\right|\exp \left(\mathrm{j\θ}\left(t\right)\right)---\left(5\right)$

Wherein, be the magnitude function of new function, θ (t) is phase function, | z (t) | be then the envelope function of impact signal s (t);

6.3) to magnitude function | z (t) | peaking: the amplitude of 6 points that more each point is adjacent, if this point is amplitude maximum, then amplitude peak value being of this point, get the starting of oscillation moment that the time point t corresponding to first peak value tried to achieve is impact signal s (t), calculate the mistiming between each acceleration transducer according to formula (6):

t _i,j＝t _i-t _j(6)

Wherein, t _ifor the starting of oscillation moment of acceleration transducer i, t _jfor the starting of oscillation moment of acceleration transducer j, i=1,2,3; J=1,2,3; I ≠ j.

Described step 7) in, calculate nominal velocity of propagation v according to formula (7) _i,j:

$v_{i,j}=\frac{d_{i,j}}{t_{i,j}}---\left(7\right)$

Wherein, d _i,jfor grid element center point is to the range difference of each acceleration transducer between two, t _i,jbe that time of arrival between two passages is poor, i=1,2,3; J=1,2,3; I ≠ j.

Described step 8) in, according to (8) formulae discovery flexural wave speed bound v _maxand v _min:

$\begin{array}{c}{v}_{\max }=2{{\mathrm{\ω}}_{\max }}^{\frac{1}{2}}{\left(\frac{{\mathrm{Eh}}^3}{12\mathrm{\ρ}(1-{\mathrm{\υ}}^2)}\right)}^{\frac{1}{4}}\\ v_{\min }=2{{\mathrm{\ω}}_{\min }}^{\frac{1}{2}}{\left(\frac{{\mathrm{Eh}}^3}{12\mathrm{\ρ}(1-{\mathrm{\υ}}^2)}\right)}^{\frac{1}{4}}\end{array}---\left(8\right)$

Wherein ω _max, ω _minbe respectively the highest angular frequency of flexural wave and minimum angular frequency, E is the Young modulus of material, and h is the thickness being knocked object, and ρ is the density of material, and υ is the Poisson ratio of material.

The present invention need not demarcate bending velocity of wave propagation in advance, avoids the error that rate calibration is introduced.Velocity range can be calculated by the material parameter and structural parameters being knocked thing, also can carry out adjustment by experiment and determine.The present invention, by limiting velocity range, only retains the valid data of rum point near zone, reduces hunting zone, considerably reduce calculated amount, reduce the impact of estimation precision on result of mistiming and range difference simultaneously.It is high that the present invention has positioning precision, the advantage that search speed is fast.

Accompanying drawing explanation

Fig. 1 is method flow block diagram of the present invention.

Fig. 2 is embodiment impact signal Hilbert envelope figure.

Fig. 3 is embodiment positioning result display figure.

Fig. 4 is embodiment loosening element positioning system overall framework figure.

Fig. 5 is embodiment acceleration transducer layout.

Embodiment

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

As shown in Figure 1, based on the nuclear power station loose positioning parts method of Hilbert transform and data screening, comprise the following steps:

1) according to positioning accuracy request, grid division on thing is being knocked, then to ready-portioned grid numbering 1 ~ N;

2) be knocked by equilateral triangle layout installation 3 acceleration transducers on thing, shock signal f (t) produced during to obtain loosening element falls;

3) according to the range difference d of each grid element center point of geometry computations to each acceleration transducer between two that are knocked thing _i,j;

4) by shock signal f (t) that each acceleration transducer of data collecting card synchronous acquisition receives, impact signal s (t) when signal f (t) comprises loosening element falls and environmental background noise n (t) is clashed into;

5) adopt the method for moving average to eliminate the low-frequency noise of clashing into signal f (t), then adopt 8 rank Butterworth wave filters by environmental background noise n (t) filtering, be eliminated impact signal s (t) of noise;

6) to step 5) in process impact signal s (t) that obtains and analyze with Hilbert transform, obtain the oscillation starting points moment of impact signal s (t); Owing to cannot know the moment that impact signal s (t) falls, can only obtain the mistiming of wave traveling from impact signal s (t) of 3 acceleration transducers, difference time of arrival between every two passages is t _i,j;

7) according to the propagation distance difference d of grid element center point to each acceleration transducer _i,jand difference t time of arrival between each passage _i,jcalculate the nominal velocity of propagation v of each grid element center point _i,j;

8) the speed bound v that flexural wave is propagated in the structure is calculated _maxand v _min;

9) to the nominal velocity of propagation v of all grid element center points _i,jcarry out data screening: judge nominal velocity of propagation v _i,jspeed interval [the v whether propagated at flexural wave _max, v _min] in, if the nominal velocity of propagation v of grid element center point _i,jall at speed interval [v _max, v _min] in, then calculate the variance D (v) of the speed of this grid element center point, otherwise compose with a large value to variance D (v);

10) search for minimum value in the variance D (v) of all grid element center point, record the coordinate of its central point;

11) positioning result display.

Described step 3) in, for plane, according to the range difference of formula (1) computing grid center to each acceleration transducer:

$d_{i,j}=\sqrt{{(x-x_i)}^2+{(y-y_i)}^2}-\sqrt{{(x-x_j)}^2+{(y-y_i)}^2}---\left(1\right)$

Wherein (x, y) is the coordinate of grid element center point, (x _i, y _i) be the coordinate of acceleration transducer i, (x _j, y _j) be the coordinate of acceleration transducer j, i=1,2,3; J=1,2,3; I ≠ j;

For hemisphere face, according to the range difference of formula (2) computing grid center to each acceleration transducer:

Wherein r is the radius of ball, for the spherical coordinates of grid element center point, for the spherical coordinates of acceleration transducer i, for the spherical coordinates of acceleration transducer j, i=1,2,3; J=1,2,3; I ≠ j.

Described step 5) in, eliminate according to formula (3) and clash into signal f (t) low-frequency noise:

$f^{\′}\left(t\right)=\frac{(f\left(t\right)+f(t-1)+...f(t-n+1))}{n}---\left(3\right)$

Wherein f'(t) be the shock signal after moving average, n is the number in period of moving average, and n is set to 7.

Described step 6) in, determine impact signal s (t) the oscillation starting points moment according to the peak value of envelope, specifically comprise the following steps:

6.1) Hilbert transform of impact signal s (t) is asked according to definition:

$\hat{s}\left(t\right)={\∫}_{-\∞}^{\∞}s\left(t\right)\frac{1}{\mathrm{\π}(t-\mathrm{\τ})}\mathrm{d\τ}---\left(4\right)$

Wherein, for the hubert transformed signal of impact signal s (t);

6.2) with impact signal s (t) for real part, its hubert transformed signal for imaginary part, form a new function as formula (5):

$z\left(t\right)=s\left(t\right)+j\hat{s}\left(t\right)=\left|z\left(t\right)\right|\exp \left(\mathrm{j\θ}\left(t\right)\right)---\left(5\right)$

Wherein, be the magnitude function of new function, θ (t) is phase function, | z (t) | be then the envelope function of impact signal s (t);

6.3) to magnitude function | z (t) | peaking: the amplitude of 6 points that more each point is adjacent, if this point is amplitude maximum, then amplitude peak value being of this point, get the starting of oscillation moment that the time point t corresponding to first peak value tried to achieve is impact signal s (t), calculate the mistiming between each acceleration transducer according to formula (6):

t _i,j＝t _i-t _j(6)

Wherein, t _ifor the starting of oscillation moment of acceleration transducer i, t _jfor the starting of oscillation moment of acceleration transducer j, i=1,2,3; J=1,2,3; I ≠ j.

Described step 7) in, calculate nominal velocity of propagation v according to formula (7) _i,j:

$v_{i,j}=\frac{d_{i,j}}{t_{i,j}}---\left(7\right)$

Wherein, d _i,jfor grid element center point is to the range difference of each acceleration transducer between two, t _i,jbe that time of arrival between two passages is poor, i=1,2,3; J=1,2,3; I ≠ j.

Described step 8) in, according to (8) formulae discovery flexural wave speed bound v _maxand v _min:

$\begin{array}{c}{v}_{\max }=2{{\mathrm{\ω}}_{\max }}^{\frac{1}{2}}{\left(\frac{{\mathrm{Eh}}^3}{12\mathrm{\ρ}(1-{\mathrm{\υ}}^2)}\right)}^{\frac{1}{4}}\\ v_{\min }=2{{\mathrm{\ω}}_{\min }}^{\frac{1}{2}}{\left(\frac{{\mathrm{Eh}}^3}{12\mathrm{\ρ}(1-{\mathrm{\υ}}^2)}\right)}^{\frac{1}{4}}\end{array}---\left(8\right)$

Wherein ω _max, ω _minbe respectively the highest angular frequency of flexural wave and minimum angular frequency, E is the Young modulus of material, and h is the thickness being knocked object, and ρ is the density of material, and υ is the Poisson ratio of material.

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

To install the flat board of three acceleration transducers by equilateral triangle, the falling position adopting Scan orientation method accurately can estimate loosening element is described.If the coordinate of certain arbitrfary point Q is Q (x, y) on steel plate, acceleration transducer i coordinate is (x _i, y _i), acceleration transducer j coordinate is (x _j, y _j), then arbitrfary point Q to the range difference between acceleration transducer i and acceleration transducer j is $d_{i,j}=\sqrt{{(x-x_i)}^2+{(y-y_i)}^2}-\sqrt{{(x-x_j)}^2+{(y-y_i)}^2}$ (i＝1,2,3；j＝1,2,3；i≠j)。

Gather and clash into signal f (t), obtain impact signal s (t) after de-noising, and carry out Hilbert transform to impact signal s (t), each passage is asked for as shown in Figure 2 time of arrival.

Suppose that the mistiming that impact signal s (t) produced propagates into 3 acceleration transducers is respectively t _1,2, t _1,3, t _2,3, so nominal velocity of propagation

First data screening can be calculated flexural wave speed bound v by hitting the structural parameters of thing, material parameter and the highest low-limit frequency of impact signal _maxand v _min, then judge v _1,2, v _1,3, v _2,3whether at speed interval [v _min, v _max] in.If v _1,2, v _1,3, v _2,3all in speed interval, then calculate nominal velocity of propagation [v _1,2, v _1,3, v _2,3] variance $D\left(v\right)=\frac{1}{3}[{(v_{\mathrm{1,2}}-\stackrel{\‾}{v})}^2+{(v_{\mathrm{1,3}}-\stackrel{\‾}{v})}^2+{(v_{\mathrm{2,3}}-\stackrel{\‾}{v})}^2],$ Wherein if [v _1,2, v _1,3, v _2,3] not in speed interval, then make variance D (v)=10000.

Find minimum value in the variance D (v) of all grid element center point, and record the coordinate of its central point, and display is as Fig. 3.

Below in conjunction with test panel test, further illustrate the present invention:

1 test condition

Loose-parts monitoring system experiment porch is primarily of steel plate, industrial computer, acceleration transducer, composition such as sound prison device and signal conditioner etc.Experimental system overall framework as shown in Figure 4.

The tested object of experimental system comprises steel plate, steel ball and support thereof.Wherein, steel plate is of a size of 200 × 200 × 2cm.In order to reduce the impact of neighbourhood noise as far as possible, all added under four edges of steel plate buffer compartment from.Each buffer compartment forms from by 3 pieces of supporting steel plates and 3 blocks of rubber slabs, and by bottom, be respectively supporting steel plate, rubber slab, supporting steel plate, rubber slab, supporting steel plate, rubber slab, gross thickness is about 9.6cm.Wherein, supporting steel plate is of a size of 20 × 20 × 1.2cm, and rubber slab is of a size of 20 × 20 × 2cm.Test steel ball weight used and be respectively 176g, 1400g and 10000g.Layout 3 sensors on flat board, as shown in Figure 5, arrange by equilateral triangle, the sample frequency of each sensor is 100kHz.

2 test findings and analysis

As shown in Figure 3, in Fig. 3, position of collision is label 1, and the vector error of positioning result is (6.667cm ,-6.667cm) in positioning result display.

The present invention adopts following methods to calculate absolute error and relative error:

Absolute error: E=Δ d, wherein Δ d is the distance between loosening element falls position and positioning result of the present invention.

Relative error: wherein S is the area that 3 sensors form equilateral triangle.

Table 1 176g steel ball impact experiment outcome record table

Table 2 1400g steel ball impact experiment outcome record table

Table 3 10000g steel ball impact experiment outcome record table

As can be seen from table 1 ~ 3,3 kinds of quality steel balls fall and are respectively 6.64%, 5.93% and 8.34% in the location average relative error of all positions, the mass range span of experiment is large, representative, illustrates that the method has good locating effect to different quality loosening element.

Content described in the present embodiment is only enumerating the way of realization of inventive concept; protection scope of the present invention should not be regarded as being only limitted to the concrete form that embodiment is stated, protection scope of the present invention also and conceive the equivalent technologies means that can expect according to the present invention in those skilled in the art.

Claims (6)

1., based on the positioning method for loosening member of nuclear power station of Hilbert transform and data screening, it is characterized in that, comprise the following steps:

1) according to positioning accuracy request, grid division on thing is being knocked, then to ready-portioned grid numbering 1 ~ N;

2) be knocked by equilateral triangle layout installation 3 acceleration transducers on thing, shock signal f (t) produced during to obtain loosening element falls;

3) according to the range difference d of each grid element center point of geometry computations to each acceleration transducer between two that are knocked thing _i,j;

4) by shock signal f (t) that each acceleration transducer of data collecting card synchronous acquisition receives, impact signal s (t) when signal f (t) comprises loosening element falls and environmental background noise n (t) is clashed into;

5) adopt the method for moving average to eliminate the low-frequency noise of clashing into signal f (t), then adopt 8 rank Butterworth wave filters by environmental background noise n (t) filtering, be eliminated impact signal s (t) of noise;

6) to step 5) in process impact signal s (t) that obtains and analyze with Hilbert transform, obtain the oscillation starting points moment of impact signal s (t); Owing to cannot know the moment that impact signal s (t) falls, can only obtain the mistiming of wave traveling from impact signal s (t) of 3 acceleration transducers, difference time of arrival between every two passages is t _i,j;

7) according to the propagation distance difference d of grid element center point to each acceleration transducer _i,jand difference t time of arrival between each passage _i,jcalculate the nominal velocity of propagation v of each grid element center point _i,j;

8) the speed bound v that flexural wave is propagated in the structure is calculated _maxand v _min;

9) to the nominal velocity of propagation v of all grid element center points _i,jcarry out data screening: judge nominal velocity of propagation v _i,jspeed interval [the v whether propagated at flexural wave _max, v _min] in, if the nominal velocity of propagation v of grid element center point _i,jall at speed interval [v _max, v _min] in, then calculate the variance D (v) of the speed of this grid element center point, otherwise compose with a large value to variance D (v);

10) search for minimum value in the variance D (v) of all grid element center point, record the coordinate of its central point;

11) positioning result display.

2. localization method according to claim 1, is characterized in that: described step 3) in, for plane, according to the range difference of formula (1) computing grid center to each acceleration transducer:

$d_{i,j}=\sqrt{{(x-x_i)}^2+{(y-y_i)}^2}-\sqrt{{(x-x_j)}^2+{(y-y_j)}^2}---\left(1\right)$

Wherein (x, y) is the coordinate of grid element center point, (x _i, y _i) be the coordinate of acceleration transducer i, (x _j, y _j) be the coordinate of acceleration transducer j, i=1,2,3; J=1,2,3; I ≠ j;

For hemisphere face, according to the range difference of formula (2) computing grid center to each acceleration transducer:

Wherein r is the radius of ball, for the spherical coordinates of grid element center point, for the spherical coordinates of acceleration transducer i, for the spherical coordinates of acceleration transducer j, i=1,2,3; J=1,2,3; I ≠ j.

3. localization method according to claim 1, is characterized in that: described step 5) in, eliminate according to formula (3) and clash into signal f (t) low-frequency noise:

$f^{\′}\left(t\right)=\frac{(f\left(t\right)+f(t-1)+...f(t-n+1))}{n}---\left(3\right)$

Wherein f'(t) be the shock signal after moving average, n is the number in period of moving average, and n is set to 7.

4. localization method according to claim 1, is characterized in that: described step 6) in, determine impact signal s (t) the oscillation starting points moment according to the peak value of envelope, specifically comprise the following steps:

6.1) Hilbert transform of impact signal s (t) is asked according to definition:

$\hat{s}\left(t\right)={\∫}_{-\∞}^{\∞}s\left(t\right)\frac{1}{\mathrm{\π}(t-\mathrm{\τ})}\mathrm{d\τ}---\left(4\right)$

Wherein, for the hubert transformed signal of impact signal s (t);

6.2) with impact signal s (t) for real part, its hubert transformed signal for imaginary part, form a new function as formula (5):

$z\left(t\right)=s\left(t\right)+j\hat{s}\left(t\right)=\left|z\left(t\right)\right|\exp \left(\mathrm{j\θ}\left(t\right)\right)---\left(5\right)$

Wherein, be the magnitude function of new function, θ (t) is phase function, | z (t) | be then the envelope function of impact signal s (t);

6.3) to magnitude function | z (t) | peaking: the amplitude of 6 points that more each point is adjacent, if this point is amplitude maximum, then amplitude peak value being of this point, get the starting of oscillation moment that the time point t corresponding to first peak value tried to achieve is impact signal s (t), calculate the mistiming between each acceleration transducer according to formula (6):

t _i,j＝t _i-t _j(6)

Wherein, t _ifor the starting of oscillation moment of acceleration transducer i, t _jfor the starting of oscillation moment of acceleration transducer j, i=1,2,3; J=1,2,3; I ≠ j.

5. localization method according to claim 1, is characterized in that: described step 7) in, calculate nominal velocity of propagation v according to formula (7) _i,j:

$v_{i,j}=\frac{d_{i,j}}{t_{i,j}}---\left(7\right)$

Wherein, d _i,jfor grid element center point is to the range difference of each acceleration transducer between two, t _i,jbe that time of arrival between two passages is poor, i=1,2,3; J=1,2,3; I ≠ j.

6. localization method according to claim 1, is characterized in that: described step 8) in, according to (8) formulae discovery flexural wave speed bound v _maxand v _min:

$\begin{array}{c}{v}_{\max }={{2\mathrm{\ω}}_{\max }}^{\frac{1}{2}}{\left(\frac{{\mathrm{Eh}}^3}{12\mathrm{\ρ}(1-{\mathrm{\υ}}^2)}\right)}^{\frac{1}{4}}\\ v_{\min }={{2\mathrm{\ω}}_{\min }}^{\frac{1}{2}}{\left(\frac{{\mathrm{Eh}}^3}{12\mathrm{\ρ}(1-{\mathrm{\υ}}^2)}\right)}^{\frac{1}{4}}\end{array}---\left(8\right)$

Wherein ω _max, ω _minbe respectively the highest angular frequency of flexural wave and minimum angular frequency, E is the Young modulus of material, and h is the thickness being knocked object, and ρ is the density of material, and υ is the Poisson ratio of material.