CN102537670B - Pipeline leakage diagnosis method - Google Patents

Abstract

The invention provides a pipeline leakage diagnosis method, which includes the following steps: mounting dynamic pressure transmitters or acoustic wave leakage monitors at head and tail stations of the pipeline respectively, so as to continuously monitor a dynamic pressure signal or an acoustic wave signal in the pipeline in real time; denoising the data of the corresponding head and tail stations and calculating the signal average, so as to turn the signals into positive and negative signals; establishing two equal length moving windows arranged at a preset interval for the data, and enabling the moving windows to move with preset step lengths, so as to cut out the two equal length time domain signals, performing power spectrum analysis on the two signals respectively to compare the two signals and calculate the corresponding characteristic indexes; detecting the threshold value of sensitivity according to the set corresponding leakage, and setting to be 1 or zero clearing the leakage diagnosis mark of the upper and lower streams; and performing the leakage diagnosis according to the leakage diagnosis mark of the upper and lower streams. The pipeline leakage diagnosis method can effectively reduce mistaken reporting and completely avoid missing reporting.

Description

A kind of pipeline leakage diagnosis method

Technical field

The present invention relates to a kind ofly for gas pipeline, oil product conveyance conduit, the water pipe equal pressure fluid line pipeline leakage testing field based on dynamic pressure transducer, sonic sensor, particularly relate to a kind of pipeline leakage diagnosis method.

Background technology

Be embedded in underground civil gas management, oil product conveyance conduit, water pipe equal pressure fluid line because laying work area is wide, complex circuit, after leaking appears in pipeline, if fail to report the police, just can not in time find its leakage point, thereby cause the loss and waste of resource, and may bring potential safety hazard and environmental pollution.

In existing pipeline leakage detection method, generally adopt feature extraction to realize the diagnosis of leakage in conjunction with the method for neural network model.And the prerequisite that realizes Leak Detection is to obtain the leakage sample data by the method for leaking simulation, and the method for employing model learning is come training pattern and obtained.This method requires that more number of samples is arranged, otherwise may cause the generalization ability of model poor, thereby causes the police that fails to report of leakage; But because study is the manual simulation with sample, and number of samples is limited, and these a limited number of analog samples can not substitute real booster fully, dig brokenly the stronger pipe leakage of signal such as pipeline, fail to report the police thereby cause inevitably.

Therefore, explore a kind of reliably, the leakage diagnosing method of independent of model, significant.

Summary of the invention

The object of the present invention is to provide a kind of pipeline leakage diagnosis method, it is false alarm reduction effectively, stops to fail to report.

A kind of pipeline leakage diagnosis method for realizing that the object of the invention provides comprises the steps:

Step S100 installs respectively a dynamic pressure transmitter or sound wave leakage monitoring instrument at pipeline first and last station, monitor in real time, continuously dynamic pressure signal or the acoustic signals of pipe interior.

The cycle of setting leak diagnostics is that NT/2(N is number of data points, and T is signal sampling period), read respectively the N/2 point data that gather from the first and last station every NT/2 cycle, and respectively with nearest N/2 point data, consist of the new data that a frame N is ordered.

Wherein, front N/2 point data are nearest historical data, and rear N/2 point data are the real time data of up-to-date collection;

Step S200 carries out denoising to the N point data at described corresponding first and last station respectively, calculates the signal average, makes it to become positive negative signal;

Step S300, to described historical data and real time data, if the Moving Window of two equal lengths, the preset space length of being separated by moves and intercepts respectively two time-domain signals of equal length with predetermined step-length, to these two signals rate of doing work analysis of spectrums respectively, compare both difference, calculate corresponding characteristic index, according to the threshold value of the corresponding Sensitivity of leak test of setting, respectively the leak diagnostics sign of upstream and downstream is put 1 or zero clearing.

Step S400, the leak diagnostics sign of the upstream and downstream that obtains according to step S300, the diagnosis of leaking: if two signs all are 1, judge that then fault has occured pipeline, otherwise pipe safety.

Beneficial effect of the present invention: pipeline leakage diagnosis method of the present invention, at first at the be separated by signal of two sections equal time length of certain hour interval intercepting of frame data, by their power spectrum signal difference of comparison in choosing frequency range, carry out leak diagnostics, it is very effectively avoided because leak diagnostics model generalization ability, leak the sample size pipe leakage that causes such as few and fail to report the police, and very effectively avoid effectively improving the leak diagnostics accuracy that the unknown is leaked because the pipe leakage of learning and causing of crossing of leak diagnostics model is failed to report the police.

Description of drawings

Fig. 1 is the pipeline leakage diagnosis method process flow diagram of the embodiment of the invention;

Fig. 2 is embodiment of the invention pipeline leakage diagnosis method signal denoising and bipolar processes result schematic diagram;

Fig. 3 is the intercepting of movement-based window method signal and the power spectrum comparison synoptic diagram thereof of embodiment of the invention pipeline leakage diagnosis method;

Fig. 4 is the synchronizing moving synoptic diagram of embodiment of the invention pipeline leakage diagnosis method Moving Window.

Embodiment

In order to make purpose of the present invention, technical scheme and advantage clearer, below in conjunction with accompanying drawing, the realization of pipeline leakage diagnosis method of the present invention is described in further detail.Should be appreciated that specific embodiment described herein only in order to explain the present invention, is not intended to limit the present invention.

The pipeline leakage diagnosis method of the embodiment of the invention, adopt dynamic pressure transducer or sonic sensor continuous monitoring pipe interior dynamic pressure (sound wave) signal, at the be separated by signal of two sections equal time length of certain hour interval intercepting of frame data, by their power spectrum signal difference of comparison in choosing frequency range, realize the diagnosis of leaking.

As a kind of embodiment, the pipeline leakage diagnosis method of the embodiment of the invention comprises the steps:

Step S100 installs respectively a dynamic pressure transmitter or sound wave leakage monitoring instrument at pipeline first and last station, monitor in real time, continuously dynamic pressure signal or the acoustic signals of pipe interior.

The cycle of setting leak diagnostics is that NT/2(N is number of data points, T is signal sampling period), read respectively the N/2 point data that gather from the first and last station every NT/2 cycle, and respectively with nearest N/2 point data, consist of each frame of N point new data (respectively corresponding upstream and downstream signal).

Wherein, front N/2 point data are nearest historical data, and rear N/2 point data are the real time data of up-to-date collection;

Adopting dynamic pressure transducer or sonic sensor continuous monitoring pipe interior dynamic pressure or acoustic signals, is a kind of prior art, therefore, in embodiments of the present invention, describes in detail no longer one by one.

Step S200 carries out denoising to the N point data at described corresponding first and last station respectively, calculates the average of signal, makes it to become positive negative signal.

Described signal being carried out denoising, calculate the signal average in the signal, make it to become positive negative signal, is a kind of prior art, therefore, in embodiments of the present invention, describes in detail no longer one by one.

Step S300, to described historical data and real time data, set the Moving Window of two equal lengths, the preset time interval of being separated by moves and intercepts respectively two time-domain signals of equal length with predetermined step-length, to two time-domain signals intercepting rate of doing work analysis of spectrums respectively, compare both spectrum difference in default frequency range, calculate corresponding power spectrum characteristic, according to the threshold value of the corresponding Sensitivity of leak test of setting, respectively the leak diagnostics sign of upstream and downstream is put 1 or zero clearing.

As a kind of embodiment, described step S300 comprises the steps:

Step S310, if Moving Window length M, Moving Window moving step pitch step, the spacing of former and later two Moving Windows (number of data points of being separated by between the beginning of the end of last Moving Window and a rear Moving Window) span, according to described parameter the resulting positive negative signal of step S200 is carried out continuation, determine the continuation data length according to formula (1).

(LOOP-1)*step+1+(M+span)+(M-1)-N (1)

Wherein LOOP is for making formula (1) for just and the positive integer of numerical value minimum, and to get this LOOP value be the cycle index that power spectrum is compared.

Step S320, according to described Moving Window length M, Moving Window moving step pitch step, the distance s pan of former and later two Moving Windows and cycle index LOOP are according to the edxh finish time of initial time stxf, the edxf finish time of front Moving Window of the front Moving Window of formula (2) ~ (5) calculating, the initial time stxh of rear Moving Window, rear Moving Window.

stxf(k)=(k-1)*step+1 (2)

edxf(k)=stxf(k)+(M-1) (3)

stxh(k)=stxf(k)+(M+span) (4)

edxh(k)=stxh(k)+(M-1) (5)

K is the circulation sequence number in the formula, and wherein, the maximal value of k is cycle index LOOP value.

Step S330 according to initial, finish time of the stxf of former and later two Moving Windows, edxf, stxh, edxh intercepting M length data from N point length data frame respectively, is assigned to array fx and fy.

Step S340, to burst fx, fy is rate of doing work analysis of spectrum respectively.

Step S350, determine the effective ratio power spectrum spectral line starting and ending position fstx corresponding to frequency range of signal, fmid, fend, calculate respectively burst fx according to formula (6) ~ (10), the power spectrum of fy between fstx and fmid cumulative and SumPxM, SumPyM, and the power spectrum between fstx and fend is cumulative and SumPxE, SumPyE.

$\mathrm{SunPxM}=\underset{k=\mathrm{fstx}}{\overset{\mathrm{fmid}}{\mathrm{\Σ}}}\mathrm{Powx}\left(k\right)---\left(6\right)$

$\mathrm{SumPym}=\underset{k=\mathrm{fstx}}{\overset{\mathrm{fmid}}{\mathrm{\Σ}}}\mathrm{Powy}\left(k\right)---\left(7\right)$

$\mathrm{SumPxE}=\underset{k=\mathrm{fstx}}{\overset{\mathrm{fend}}{\mathrm{\Σ}}}\mathrm{Powx}\left(k\right)---\left(8\right)$

$\mathrm{SumPyE}=\underset{k=\mathrm{fstx}}{\overset{\mathrm{fend}}{\mathrm{\Σ}}}\mathrm{Powy}\left(k\right)---\left(9\right)$

Wherein Powx (k) and Powy (k) are respectively the power spectrum of burst fx and fy.

In addition, according to formula (10)

$\mathrm{DPxy}=\underset{j=\mathrm{fstx}}{\overset{\mathrm{fmid}}{\mathrm{\Σ}}}(\mathrm{Powx}\left(j\right)-\mathrm{Powy}\left(j\right))---\left(10\right)$

Calculate burst fx and the power spectrum of fy between fstx and fmid difference cumulative and.

Step S360, calculated characteristics ratio R PC and RPZ;

If DPxy then utilizes formula (11) and (12) ratio calculated RPC and RPZ for just;

$\mathrm{RPC}=\frac{\mathrm{DPxy}}{\mathrm{SumPxM}}---\left(11\right)$

$\mathrm{RPZ}=\frac{\mathrm{SumPxM}}{\mathrm{SumPxE}}---\left(12\right)$

If DPxy then utilizes formula (13) and (14) ratio calculated RPC and RPZ for negative;

$\mathrm{RPC}=\frac{\mathrm{DPxy}}{\mathrm{SumPyM}}---\left(13\right)$

$\mathrm{RPZ}=\frac{\mathrm{SumPyM}}{\mathrm{SumPyE}}---\left(14\right)$

Step S370, the leak diagnostics threshold value of establishing upstream (at first stop) monitor signal is respectively SetVU1 and SetVU2, if RPC 〉=SetVU1, and RPZ=SetVU2, then leakage monitoring signal in upstream has unusually, and upstream line leak flag AlarmFlagU puts 1; Otherwise AlarmFlagU=0.

Step S380, if AlarmFlagU is not 1, then according to cycle index LOOP, and initial, finish time of stxf, edxf, stxh, the edxh of different former and later two Moving Windows of circulation time, stream signal is repeated to do step S330～step S370 process.

Step S390, the leak diagnostics threshold value of setting trip (terminal) monitor signal is respectively SetVD1 and SetVD2, and to downstream data repeating step S310～step S380, and mark downstream line leak flag AlarmFlagD is 1 or 0.

Step S400 according to upstream and downstream (first and last station) the pipe leakage value of statistical indicant that step S300 obtains, judges further whether pipeline leakage (abnormal failure) has occured.

Particularly, as a kind of embodiment, in step S400, if upstream (at first stop) pipe leakage sign A larmFlagU is 1, and downstream (terminal) pipe leakage sign A larmFlagD also is 1, then in the pipeline fault occured, and sends fault alarm; Otherwise pipe safety does not leak.

The below lifts an example, as shown in Figure 1, further describes pipeline leakage diagnosis method of the present invention, and the present invention can realize with any Programming with Pascal Language, and move at corresponding computer.

Step 1. adopts many Bei Xi (Daubechies) db9 small echo that the upstream and downstream original signal is done denoising, and the Wavelet Denoising Method yardstick is 4; Through Wavelet Denoising Method and calculating signal average, obtain ambipolar upstream and downstream signal.Wherein Fig. 2 is the ambipolar upstream and downstream signal behind the Wavelet Denoising Method.

Mark 1 and 2 is respectively the correspondence position of upstream and downstream leakage signal among Fig. 2.

It is 6000 that step 2. is established signal length (number of data points) N, and sampling period T is 20ms, and the Moving Window length M is 2048, and Moving Window moving step pitch step is 512, and the distance s pan of former and later two Moving Windows is 512.Be respectively 144 border end effect as length to the upstream and downstream signal, Fig. 3 (descending) is depicted as through the stream signal after the continuation, and getting cycle index LOOP is 4 times.

Step 3. is according to the Moving Window length M of setting, Moving Window moving step pitch step, and the distance s pan of former and later two Moving Windows and cycle index LOOP calculate initial, finish time of stxf, edxf, stxh, the edxh of former and later two Moving Windows

Be respectively:

Stxf＝{1 513 1025 1537}

Edxf＝{2048 2560 3072 3584}

Stxh={2561 3073 3585 4097}

Edxh={4608 5120 5632 6144}

Step 4. intercepts respectively 1-2048 point data and 2561-4608 point data, is assigned to array fx and fy.As shown in Figure 3.

In Fig. 3 (descending), 3 are 1-2048 point data, and 4 are 2561-4608 point data, the Moving Window of the corresponding M=2048 of 5 and 6 difference, 7 correspondences span(512 point between the two).

Step 5. is respectively to fx and fy rate of doing work analysis of spectrum, as Fig. 3 (on) shown in, 8 and 9 power spectrum of corresponding fx and fy respectively wherein.

Step 6. is by analyzing the power spectrum of contrast normal signal and leakage signal, determine fstx, fmid and fend are respectively 2,6 and 15, calculate SumPxM, SumPyM, SumPxE, SumPyE and DPxy, be respectively 1.3624e+007,3.3060e+008,2.5163e+008,6.8396e+008 and-3.1698e+008

Step 7. is calculated RPC and RPZ, gets RPC=0.9588, RPZ=0.4834.

Leak diagnostics threshold value SetVU1 and SetVU2 that step 8. is established stream signal are respectively 0.87 and 0.3.Because RPC and RPZ are respectively greater than SetVU1 and SetVU2, so stream signal warning mark AlarmFlagU=1.

If the AlarmFlagU of step 9. the first circulation is not 1, then move two windows (window function), get respectively the data of 513-2560 and 3073-5120, correspondence is assigned to fx and fy, and repeating step 4-step 8 re-starts diagnosis; So circulation is until the circulation end.Among Fig. 4 (descending), 10 and 11 distinguish the window function of corresponding 513-2560 and 3073-5120; Otherwise, enter step 10.

Step 10. is for downstream signal, the distance s pan of its data length N, sampling period T, Moving Window length M, Moving Window moving step pitch step, former and later two Moving Windows, border end effect length and cycle index LOOP do the same setting with the upstream, and stxf, edxf, stxh, edxh also equally set.And set downstream leak diagnostics threshold value SetVD1 and SetVD2 is respectively 0.87 and 0.3.SumPxM, SumPyM, SumPxE, SumPyE and the DPxy of the first circulation of its correspondence are respectively 1.8828e+007,1.0954e+006,2.4721e+007,3.1796e+006 and 1.7733e+007.Corresponding RPC and RPZ are respectively 0.9975 and 0.7297, respectively greater than setting value.So AlarmFlagD=1.

Step 11. all is 1 according to AlarmFlagU and AlarmFlagD, can determine to contain fault-signal in the corresponding upstream and downstream signal constantly, sends fault alarm.

The pipeline leakage diagnosis method of the embodiment of the invention, very effectively avoid because the pipe leakage that leak diagnostics model generalization ability causes is failed to report the police, and very effectively avoid effectively improving the leak diagnostics accuracy that the unknown is leaked because the pipe leakage of learning and causing of crossing of leak diagnostics model is failed to report the police.

Should be noted that at last that obviously those skilled in the art can carry out various changes and modification to the present invention and not break away from the spirit and scope of the present invention.Like this, if of the present invention these revise and modification belongs within the scope of claim of the present invention and equivalent technologies thereof, then the present invention also is intended to comprise these changes and modification.

Claims (3)

1. a pipeline leakage diagnosis method is characterized in that, comprises the steps:

Step S100 installs respectively a dynamic pressure transmitter or sound wave leakage monitoring instrument at pipeline first and last station, monitor in real time, continuously dynamic pressure signal or the acoustic signals of pipe interior;

The cycle of setting leak diagnostics is NT/2, reads respectively the N/2 point data that gather from the first and last station every NT/2 cycle, and respectively with nearest N/2 point data, consists of the new data that a frame N is ordered;

Wherein, N is number of data points, and T is signal sampling period; Front N/2 point data are nearest historical data, and rear N/2 point data are the real time data of up-to-date collection;

Step S200 carries out denoising to the N point data at described corresponding first and last station respectively, calculates the signal average, makes it to become positive negative signal;

Step S300, to described historical data and real time data, set the Moving Window of two equal lengths, the preset time interval of being separated by moves and intercepts respectively two time-domain signals of equal length with predetermined step-length, to two signals rate of doing work analysis of spectrums respectively, compare both difference, calculate corresponding characteristic index, according to the threshold value of the corresponding Sensitivity of leak test of setting, respectively the leak diagnostics sign of upstream and downstream is put 1 or zero clearing;

Step S400, the leak diagnostics sign of the upstream and downstream that obtains according to step S300, the diagnosis of leaking.

2. pipeline leakage diagnosis method according to claim 1 is characterized in that, described step S300 comprises the steps:

Step S310 is set as follows parameter: Moving Window length M, Moving Window moving step pitch step, the distance s pan of former and later two Moving Windows;

According to described parameter the resulting positive negative signal of step S200 is carried out continuation, determine the continuation data length according to formula (1);

(LOOP-1)*step+1+(M+span)+(M-1)-N (1)

Wherein LOOP is for making formula (1) for just and the positive integer of numerical value minimum, and to get this LOOP value be the cycle index that power spectrum is compared;

Step S320, according to described Moving Window length M, Moving Window moving step pitch step, the distance s pan of former and later two Moving Windows and cycle index LOOP are according to the edxh finish time of initial time stxf, the edxf finish time of front Moving Window of the front Moving Window of formula (2) ~ (5) calculating, the initial time stxh of rear Moving Window, rear Moving Window;

stxf(k)=(k-1)*step+1 (2)

edxf(k)=stxf(k)+(M-1) (3)

stxh(k)=stxf(k)+(M+span) (4)

edxh(k)=stxh(k)+(M-1) (5)

K is the circulation sequence number in the formula, and wherein, the maximal value of k is cycle index LOOP value;

Step S330 according to initial, finish time of the stxf of former and later two Moving Windows, edxf, stxh, edxh intercepting M length data from N point length data frame respectively, is assigned to array fx and fy;

Step S340, to burst fx, fy is rate of doing work analysis of spectrum respectively;

Step S350, determine the effective ratio power spectrum spectral line starting and ending position fstx corresponding to frequency range of signal, fend, fmid is between fstx and fend, calculate respectively burst fx according to formula (6) ~ (9), the power spectrum of fy between fstx and fmid cumulative and SumPxM, SumPyM, and the power spectrum between fstx and fend is cumulative and SumPxE, SumPyE;

$\mathrm{SumPxM}=\underset{k=\mathrm{fstx}}{\overset{\mathrm{fmid}}{\mathrm{\Σ}}}\mathrm{Powx}\left(k\right)---\left(6\right)$

$\mathrm{SumPxM}=\underset{k=\mathrm{fstx}}{\overset{\mathrm{fmid}}{\mathrm{\Σ}}}\mathrm{Powy}\left(k\right)---\left(7\right)$

$\mathrm{SumPxE}=\underset{k=\mathrm{fstx}}{\overset{\mathrm{fmid}}{\mathrm{\Σ}}}\mathrm{Powx}\left(k\right)---\left(8\right)$

$\mathrm{SumPxE}=\underset{k=\mathrm{fstx}}{\overset{\mathrm{fmid}}{\mathrm{\Σ}}}\mathrm{Powy}\left(k\right)---\left(9\right)$

Wherein Powx (k) and Powy (k) are respectively the power spectrum of burst fx and fy;

In addition, according to formula (10)

$\mathrm{DPxy}=\underset{j=\mathrm{fstx}}{\overset{\mathrm{fmid}}{\mathrm{\Σ}}}(\mathrm{Powx}\left(j\right)-\mathrm{Powy}\left(j\right))---\left(10\right)$

Calculate the cumulative and DPxy of the difference of burst fx and the power spectrum of fy between fstx and fmid;

Step S360, calculated characteristics ratio R PC and RPZ;

If DPxy then utilizes formula (11) and (12) ratio calculated RPC and RPZ for just;

$\mathrm{RPC}=\frac{\mathrm{DPxy}}{\mathrm{SumPxM}}---\left(11\right)$

$\mathrm{RPZ}=\frac{\mathrm{SumPxM}}{\mathrm{SumPxE}}---\left(12\right)$

If DPxy then utilizes formula (13) and (14) ratio calculated RPC and RPZ for negative;

$\mathrm{RPC}=\frac{\mathrm{DPxy}}{\mathrm{SumPyM}}---\left(13\right)$

$\mathrm{RPZ}=\frac{\mathrm{SumPyM}}{\mathrm{SumPyE}}---\left(14\right)$

Step S370, the leak diagnostics threshold value of establishing the upstream monitor signal is respectively SetVU1 and SetVU2, if RPC 〉=SetVU1, and RPZ=SetVU2, then leakage monitoring signal in upstream has unusually, and upstream line leak flag AlarmFlagU puts 1; Otherwise AlarmFlagU=0;

Step S380, if AlarmFlagU is not 1, then according to cycle index LOOP, and initial, finish time of stxf, edxf, stxh, the edxh of different former and later two Moving Windows of circulation time, stream signal is repeated to do step S330～step S370 process;

Step S390, the leak diagnostics threshold value of setting the trip monitor signal is respectively SetVD1 and SetVD2, and to downstream data repeating step S310～step S380, and mark downstream line leak flag AlarmFlagD is 1 or 0.

3. pipeline leakage diagnosis method according to claim 2 is characterized in that, described step S400, and described diagnosis of leaking comprises the steps:

If upstream line leak flag AlarmFlagU is 1, and downstream line leak flag AlarmFlagD also is 1, then judges fault has occured in the pipeline, sends fault alarm; Otherwise the judgement pipe safety does not leak.

Images