CN103363880B - Pipeline girth weld joint identification and positioning method - Google Patents

Abstract

The invention discloses a method for identifying and positioning a pipeline girth weld, which is characterized in that an upper computer calculates a position sequence { s } of each weld_bGet the position of the weldNormalized sequence s_c}; acquiring a magnetic field wavelet sequence, an acceleration wavelet sequence and a sound wavelet sequence; down-sampling the acceleration wavelet sequence and the sound wavelet sequence, and calculating a total wavelet sequence; calculating the average sampling point number E between two welding lines and the cut-off frequency of a low-pass filter; passing the total wavelet series with a cut-off frequency of f_bsFiltering by a beta-order low-pass filter to obtain a smooth wavelet sequence, and taking the moment when the peak value appears as the moment when the welding seam appears; and arranging all the weld moments in sequence to obtain a weld positioning moment sequence, and performing normalization processing to obtain a normalized weld positioning moment sequence. The method improves the identification rate and the positioning precision of the pipeline circumferential weld, reduces the power consumption and the positioning cost, and enlarges the range in practical application.

Description

A kind of pipeline girth weld recognition positioning method

Technical field

The present invention relates to pipe signal process field, particularly a kind of pipeline girth weld recognition positioning method.

Background technology

In-pipeline detector can detect defect of pipeline and defect be positioned under the state of conduit running, to ensureing pipelineSafe operation has important function. Because the same pipeline of defect of pipeline (internal detector) position one that in-pipeline detector is measured is a pair ofAnswer, must accurately know the position of each moment in-pipeline detector. The conventional localization method of in-pipeline detector at present, as innerJourney wheel method and inertial navigation method etc., all need land mark device to carry out auxiliary positioning to eliminate cumulative errors, uses inconvenience.

Due to the surface configuration at pipeline girth weld place and the particularity of inner crystal phase structure, acoustic impedance and electromagnetic resistivity and pipeline all the otherThere is a great difference in position, can adopt in theory eddy current, the active non-destructive detecting device such as ultrasonic to identify and tack weld.But these two kinds of active methods all need exciting bank, power dissipation ratio is larger, conventionally need to tube wall special compounding, structure is alsoMore complicated, it is very inconvenient to use. In addition,, for gas or product oil conveying pipe, also can consider to use optical sensingDevice detects weld seam, but price comparison costliness, and the transparency of oil product is had to very high requirement, very impracticable.

The simple pipeline internal magnetic field that relies on removes to identify weld seam, and discrimination is limited, need to adopt new method to reduce faulty identification and omissionThe probability of identification.

Summary of the invention

The invention provides a kind of pipeline girth weld recognition positioning method, the present invention does not need exciting bank, has reduced power consumption and fixedPosition cost, has expanded range of application, described below:

A kind of pipeline girth weld recognition positioning method, said method comprising the steps of:

Three-component Magnetic Sensor, three-component accelerometer, sound transducer are fixed on to the optional position in in-pipeline detector,Obtain ducted internal magnetic field signal, acceleration signal and voice signal, and 3 kinds of signals are transferred to host computer;

Described host computer calculates the position sequence { s that each weld seam occurs_b, and obtain position while welding normalization sequence { s_c}；

Obtain magnetic field wavelet sequence { DB_j, acceleration wavelet sequence { DA_jAnd sound wavelet sequence { DV_j}；

To described acceleration wavelet sequence { DA_jAnd described sound wavelet sequence { DV_jCarry out down-sampledly, and calculate total little wave trainRow { D_k}；

Calculate the count cut-off frequency of E and low pass filter of average sample between two weld seams;

By total wavelet sequence { D_kBe f by cut-off frequency_bsThe low pass filter filtering of β rank, obtain level and smooth wavelet sequence, willThe moment that the moment that peak value occurs occurs as weld seam;

The moment that all weld seams are occurred arranges sequentially and obtains weld seam location moment sequence, and is returned as normalizedOne changes weld seam location moment sequence.

Described host computer calculates the position sequence { s that each weld seam occurs_b, and obtain position while welding normalization sequence { s_cStep toolBody is:

1) from construction information, read welding seam No and every joint duct length, obtain every pipe joint road length sequences;

2) calculate the position { s that each weld seam occurs_a, supplement s₀=0 to { s_aMust position while welding sequence { s_b|b=0,1,2,…,A}，AFor pipeline number;

3) by position while welding sequence { s_bIn each element there is position s divided by last weld seam_A, obtain position while welding normalizationSequence { s_c|c=0,1,2,…,A}。

The described magnetic field wavelet sequence { DB that obtains_jStep be specially:

Calculate every bit magnetic field total amount sequenceWherein B_jx，B_jy，B_jzFor Magnetic Sensor measureThe magnetic field three-component of every bit;

To magnetic field total amount sequence { B_jCarry out M layer continuous wavelet transform, obtain every layer of magnetic field coefficient of wavelet decomposition { CB_ij}；

WhereinBe the small echo of the magnetic field B of j position of i layer,For female small echo used, the change certainly that τ is small echoAmount;

To the magnetic field coefficient of wavelet decomposition { CB of different layers_ijDo following computing, obtain magnetic field wavelet sequence { DB_j}：

${\mathrm{DB}}_j=\sqrt{\underset{i=1}{\overset{M}{\mathrm{\Σ}}}}{C}_{\mathrm{Bij}.}^2$

Described to described acceleration wavelet sequence { DA_jAnd described sound wavelet sequence { DV_jCarry out down-sampledly, and calculate totalWavelet sequence { D_jStep be specially:

NB represents field pulses element number; NV represents sound sequence element number; NA represents the unit of acceleration total amount sequenceElement number; The down-sampled coefficient of sonic transducer and acceleration transducer is respectively:

N_S=NV/NB，Na=NA/NB

Voice signal after down-sampled and the wavelet sequence of acceleration signal are respectively:

DV_k=DV_j|j=k*Ns,k=1,2,3,...,NV/Ns

DA_k=DA_j|j=k*Na,k=1,2,3,...,NA/Na

DB_k=DB_j|j=k,k=1,2,3,...,NB；

Total wavelet sequence is:

$D_k=\sqrt{{{\mathrm{DB}}_k}^2+{{\mathrm{DA}}_k}^2+{{\mathrm{DV}}_k}^2}$

The count step of cut-off frequency of E and low pass filter of average sample between two weld seams of described calculating is specially:

$E=\frac{f\·l}{v}$

$l=\frac{1}{A}\underset{i=1}{\overset{A}{\mathrm{\Σ}}}{l}_i$

Wherein, the average length that l is pipeline; V is the average pace of internal detector; The cut-off frequency of low pass filter is:f_bs=pf/E, the sample frequency that f is Magnetic Sensor, p is setting value.

The beneficial effect of technical scheme provided by the invention is: by pipeline internal magnetic field signal, acceleration signal and voice signalCarry out data processing, select suitable small echo to carry out continuous wavelet transform, to girth joint place magnetic field breakpoint, sound breakpoint, accelerationDegree breakpoint strengthens and adopts the method butt welded seam place jump signal of LPF identify and locate, thereby improves pipelineThe discrimination of girth joint and positioning precision, reduced power consumption and location cost, expanded the scope in practical application.

Brief description of the drawings

Fig. 1 is a kind of flow chart of pipeline girth weld recognition positioning method;

Fig. 2 is magnetic signal actual measured value and wavelet sequence;

Fig. 3 is the acceleration signal wavelet sequence after down-sampled;

Fig. 4 is the voice signal wavelet sequence after down-sampled;

Fig. 5 is total wavelet sequence.

Detailed description of the invention

For making the object, technical solutions and advantages of the present invention clearer, below in conjunction with accompanying drawing to embodiment of the present invention do intoOne step ground is described in detail.

In order to reduce power consumption and location cost, expand range of application, the embodiment of the present invention provides a kind of pipeline girth weld identification fixedMethod for position, referring to Fig. 1, described below:

101: sensor is fixed on to the optional position in in-pipeline detector, obtains ducted signal, and signal is transferred toHost computer;

Being operating as in detail of this step: three-component Magnetic Sensor, three-component accelerometer, sound transducer are fixed in pipelineOptional position in detector, throws pipe by internal detector and patrols and examines, measuring channel internal magnetic field signal, acceleration signal and voice signal,Patrol and examine completely, the above-mentioned signal of internal detector record is downloaded to host computer, carry out data processing.

Wherein, the embodiment of the present invention to sensor (being three-component Magnetic Sensor, three-component accelerometer, sound transducer) andThe model of in-pipeline detector does not limit, if the device that can complete above-mentioned functions all can, for example: in-pipeline detector canFor cylindricality internal detector or spherical internal detector etc.

102: host computer calculates the position sequence { s that each weld seam occurs_b, and obtain position while welding normalization sequence { s_c}；

Being operating as in detail of this step:

1) from construction information, read welding seam No and every joint duct length, obtain every pipe joint road length sequences { l_i|i=1,2,…,A}，AFor pipeline number;

2) calculate the position that each weld seam occursSupplement s₀=0 to { s_aMust position while welding sequence{s_b| b=0,1,2 ..., A}, A is pipeline number;

3) by position while welding sequence { s_bIn each element there is position s divided by last weld seam_A, obtain position while welding normalizationSequence { s_c|c=0,1,2,…,A}。

103: calculate every bit magnetic field total amount sequence { B_jAnd carry out M layer continuous wavelet transform, every layer of magnetic field wavelet decomposition system obtainedNumber { CB_ij, to every layer of magnetic field coefficient of wavelet decomposition squared do and, obtain magnetic field wavelet sequence { DB_j}；

Being operating as in detail of this step:

Calculate every bit magnetic field total amount sequenceWherein B_jx，B_jy，B_jzFor Magnetic Sensor measureThe magnetic field three-component of every bit.

To magnetic field total amount sequence { B_jCarry out M layer continuous wavelet transform, be decomposed into different zoom coefficient and difference by field signalThe Wavelet-Weighted sum of translation coefficient, obtains every layer of magnetic field coefficient of wavelet decomposition { CB_ij, different zoom coefficient and different translation coefficientsThe small echo of the corresponding different layers of small echo and diverse location.

WhereinBe the small echo of the magnetic field B of j position of i layer,For female small echo used, the change certainly that τ is small echoAmount.

To the magnetic field coefficient of wavelet decomposition { CB of different layers_ijDo following computing, obtain magnetic field wavelet sequence { DB_j}：

${\mathrm{DB}}_j=\sqrt{\underset{i=1}{\overset{M}{\mathrm{\Σ}}}}{\mathrm{CB}}_{\mathrm{ij}}^2$

104: in like manner adopt the method in step 103 to obtain acceleration wavelet sequence { DA_jAnd sound wavelet sequence { DV_j}；

, calculate every bit acceleration total amount sequenceWherein A_jx，A_jy，A_jzFor every bitAcceleration three-component. To acceleration total amount sequence { A_jCarry out continuous wavelet transform:

WhereinFor the small echo of j position acceleration A of acceleration total flow control i layer; To the acceleration wavelet decomposition of different layersCoefficient { CA_ijDo following computing, obtain acceleration wavelet sequence { DA_j}：

${\mathrm{DA}}_j=\sqrt{\underset{i=1}{\overset{M}{\mathrm{\Σ}}}}{\mathrm{CA}}_{\mathrm{ij}}^2$

Using the voice signal getting in step 101 as sound total amount sequence { V_j, to sound total amount sequence { V_jCarry out continuously littleWave conversion:

WhereinFor the small echo of j position sound V of the total flow control i of sound layer; To the sound coefficient of wavelet decomposition { CV of different layers_ij}Do following computing, obtain sound wavelet sequence { DV_j}：

${\mathrm{DV}}_j=\sqrt{\underset{i=1}{\overset{M}{\mathrm{\Σ}}}}{\mathrm{CV}}_{\mathrm{iy}}^2$

Wherein, in the time solving magnetic field wavelet sequence,For magnetic field female small echo used, τ is the independent variable of magnetic field small echo; PhaseAnswer in the time solving acceleration wavelet sequence,For acceleration female small echo used, τ is the independent variable of acceleration small echo; Or,In the time solving sound wavelet sequence,For sound female small echo used, τ is the independent variable of sound small echo.

105: to acceleration wavelet sequence { DA_jAnd sound wavelet sequence { DV_jCarry out down-sampledly, and calculate total wavelet sequence{D_j}；

NB represents field pulses element number, i.e. { the maximum of j in Bj} sequence; NV represents sound sequence element number, i.e. { Vj}The maximum of j in sequence; NA represents the element number of acceleration total amount sequence, i.e. { the maximum of j in Aj} sequence. Sound passesThe down-sampled coefficient of sensor and acceleration transducer is respectively:

N_S=NV/NB，Na=NA/NB

Voice signal after down-sampled and the wavelet sequence of acceleration signal are respectively:

DV_k=DV_j|j=k*Ns,k=1,2,3,...,NV/Ns

DA_k=DA_j|j=k*Na,k=1,2,3,...,NA/Na

And calculate and represent for convenient, by DB_jSequence is carried out as down conversion:

DB_k=DB_j|j=k,k=1,2,3,...,NB。

Can learn that by above-mentioned calculating the value of k in three sequences is all NB,

Total wavelet sequence is:

$D_k=\sqrt{{{\mathrm{DB}}_k}^2+{{\mathrm{DA}}_k}^2+{{\mathrm{DV}}_k}^2}$

106: calculate the count cut-off of E and Butterworth Butterworth LPF of average sample between two weld seamsFrequency;

$E=\frac{f\·l}{v}$

$l=\frac{1}{A}\underset{i=1}{\overset{A}{\mathrm{\Σ}}}{l}_i$

Wherein, the average length that l is pipeline, i.e. average distance l between every adjacent two weld seams; V is the average of internal detectorPace.

The cut-off frequency of getting Butterworth low pass filter is: f_bs=pf/E, the sample frequency that wherein f is Magnetic Sensor, pDesirable 0.5, can be according to actual conditions adjustment.

107: by total wavelet sequence { D_jBe f by cut-off frequency_bsThe Butterworth low pass filter filtering of β rank, obtain relativelyLevel and smooth wavelet sequence, carries out peak value detection to it, in the moment that the moment that peak value is occurred occurs as weld seam, has identified welderingSeam;

Wherein, the value of β can get 4, can also adjust according to the situation in practical application, but in order to get preferablyFilter effect, the value of β is preferably 4 to 8, exceedes 8 filter effects not obvious.

108: the moment that all weld seams are occurred arranges sequentially and obtains weld seam location moment sequence, and obtains as normalizedTo normalization weld seam location moment sequence.

Verify the feasibility of a kind of pipeline girth weld recognition positioning method that the embodiment of the present invention provides below with concrete test,Described below:

Adopt method of the prior art can obtain the result of Fig. 2, as can be seen from the figure magnetic signal can be told a fixed numberThe weld seam of amount, but there is undetected situation. If in conjunction with falling in the acceleration signal wavelet sequence after down-sampled in Fig. 3 and Fig. 4Voice signal wavelet sequence after sampling calculates, and can obtain wavelet sequence total in Fig. 5. Can tell institute by Fig. 5Some weld seams. Than only relying on magnetic signal to detect, the method has reduced undetected situation, has improved weld seam recognition rate.

It will be appreciated by those skilled in the art that accompanying drawing is the schematic diagram of a preferred embodiment, the invention described above embodiment sequence number onlyOnly, in order to describe, do not represent the quality of embodiment.

The foregoing is only preferred embodiment of the present invention, not in order to limit the present invention, all the spirit and principles in the present invention itIn, any amendment of doing, be equal to replacement, improvement etc., within all should being included in protection scope of the present invention.

Claims (2)

1. a pipeline girth weld recognition positioning method, is characterized in that, said method comprising the steps of:

Three-component Magnetic Sensor, three-component accelerometer, sound transducer are fixed on to the optional position in in-pipeline detector,Obtain ducted internal magnetic field signal, acceleration signal and voice signal, and 3 kinds of signals are transferred to host computer;

Described host computer calculates the position sequence { s that each weld seam occurs_b, and obtain position while welding normalization sequence { s_c}；

Obtain magnetic field wavelet sequence { DB_j, acceleration wavelet sequence { DA_jAnd sound wavelet sequence { DV_j}；

To described acceleration wavelet sequence { DA_jAnd described sound wavelet sequence { DV_jCarry out down-sampledly, and calculate total little wave trainRow { D_k}；

Calculate the count cut-off frequency of E and low pass filter of average sample between two weld seams;

By total wavelet sequence { D_kBe f by cut-off frequency_bsThe low pass filter filtering of β rank, obtain level and smooth wavelet sequence, willThe moment that the moment that peak value occurs occurs as weld seam;

The moment that all weld seams are occurred arranges sequentially and obtains weld seam location moment sequence, and is returned as normalizedOne changes weld seam location moment sequence;

Wherein, described host computer calculates the position sequence { s that each weld seam occurs_b, and obtain position while welding normalization sequence { s_c?Step is specially:

1) from construction information, read welding seam No and every joint duct length, obtain every pipe joint road length sequences;

2) calculate the position { s that each weld seam occurs_a, supplement s₀=0 to { s_aMust position while welding sequence { s_b|b＝0,1,2,…,A}，AFor pipeline number;

3) by position while welding sequence { s_bIn each element there is position s divided by last weld seam_A, obtain position while welding normalizationSequence { s_c|c＝0,1,2,…,A}；

Wherein, described in, obtain magnetic field wavelet sequence { DB_jStep be specially:

Calculate every bit magnetic field total amount sequenceWherein B_jx，B_jy，B_jzFor Magnetic Sensor measureThe magnetic field three-component of every bit;

To magnetic field total amount sequence { B_jCarry out M layer continuous wavelet transform, obtain every layer of magnetic field coefficient of wavelet decomposition { CB_ij}；

WhereinBe the small echo of the magnetic field B of j position of i layer,For female small echo used, the change certainly that τ is small echoAmount;

To the magnetic field coefficient of wavelet decomposition { CB of different layers_ijDo following computing, obtain magnetic field wavelet sequence { DB_j}：

${\mathrm{DB}}_j=\sqrt{\underset{i=1}{\overset{M}{\Σ}}{C}_{Bij}^2};$

Wherein, described to described acceleration wavelet sequence { DA_jAnd described sound wavelet sequence { DV_jCarry out down-sampled, and meterCalculate total wavelet sequence { D_kStep be specially:

NB represents field pulses element number; NV represents sound sequence element number; NA represents the unit of acceleration total amount sequenceElement number; The down-sampled coefficient of sonic transducer and acceleration transducer is respectively:

N_S＝NV/NB，Na＝NA/NB

Voice signal after down-sampled and the wavelet sequence of acceleration signal are respectively:

DV_k＝DV_j|j＝k*Ns,k＝1,2,3,...,NV/Ns

DA_k＝DA_j|j＝k*Na,k＝1,2,3,...,NA/Na

DB_k＝DB_j|j＝k,k＝1,2,3,...,NB；

Total wavelet sequence is:

$D_k=\sqrt{{{\mathrm{DB}}_k}^2+{{\mathrm{DA}}_k}^2+{{\mathrm{DV}}_k}^2}.$

2. a kind of pipeline girth weld recognition positioning method according to claim 1, is characterized in that, two welderings of described calculatingThe count step of cut-off frequency of E and low pass filter of average sample between seam is specially:

$E=\frac{f\·l}{v}$

$l=\frac{1}{A}\underset{i=1}{\overset{A}{\Σ}}{l}_i$

Wherein, the average length that l is pipeline; l_iFor every pipe joint road length sequences; V is the average pace of internal detector; LowThe cut-off frequency of bandpass filter is: f_bs=pf/E, the sample frequency that f is Magnetic Sensor, p is setting value.