CN108488638A - Line leakage system and method based on sound wave suction wave hybrid monitoring - Google Patents

Line leakage system and method based on sound wave suction wave hybrid monitoring Download PDF

Info

Publication number
CN108488638A
CN108488638A CN201810260988.3A CN201810260988A CN108488638A CN 108488638 A CN108488638 A CN 108488638A CN 201810260988 A CN201810260988 A CN 201810260988A CN 108488638 A CN108488638 A CN 108488638A
Authority
CN
China
Prior art keywords
signal
data
leakage
pressure
sound wave
Prior art date
Application number
CN201810260988.3A
Other languages
Chinese (zh)
Other versions
CN108488638B (en
Inventor
马大中
张化光
冯健
汪刚
刘金海
于洋
刘富聪
关勇
Original Assignee
东北大学
Priority date (The priority date is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the date listed.)
Filing date
Publication date
Application filed by 东北大学 filed Critical 东北大学
Priority to CN201810260988.3A priority Critical patent/CN108488638B/en
Publication of CN108488638A publication Critical patent/CN108488638A/en
Application granted granted Critical
Publication of CN108488638B publication Critical patent/CN108488638B/en

Links

Classifications

    • FMECHANICAL ENGINEERING; LIGHTING; HEATING; WEAPONS; BLASTING
    • F17STORING OR DISTRIBUTING GASES OR LIQUIDS
    • F17DPIPE-LINE SYSTEMS; PIPE-LINES
    • F17D5/00Protection or supervision of installations
    • F17D5/02Preventing, monitoring, or locating loss
    • F17D5/06Preventing, monitoring, or locating loss using electric or acoustic means
    • GPHYSICS
    • G01MEASURING; TESTING
    • G01MTESTING STATIC OR DYNAMIC BALANCE OF MACHINES OR STRUCTURES; TESTING OF STRUCTURES OR APPARATUS, NOT OTHERWISE PROVIDED FOR
    • G01M3/00Investigating fluid-tightness of structures
    • G01M3/02Investigating fluid-tightness of structures by using fluid or vacuum
    • G01M3/04Investigating fluid-tightness of structures by using fluid or vacuum by detecting the presence of fluid at the leakage point
    • G01M3/24Investigating fluid-tightness of structures by using fluid or vacuum by detecting the presence of fluid at the leakage point using infrasonic, sonic, or ultrasonic vibrations
    • G01M3/243Investigating fluid-tightness of structures by using fluid or vacuum by detecting the presence of fluid at the leakage point using infrasonic, sonic, or ultrasonic vibrations for pipes
    • GPHYSICS
    • G01MEASURING; TESTING
    • G01MTESTING STATIC OR DYNAMIC BALANCE OF MACHINES OR STRUCTURES; TESTING OF STRUCTURES OR APPARATUS, NOT OTHERWISE PROVIDED FOR
    • G01M3/00Investigating fluid-tightness of structures
    • G01M3/02Investigating fluid-tightness of structures by using fluid or vacuum
    • G01M3/26Investigating fluid-tightness of structures by using fluid or vacuum by measuring rate of loss or gain of fluid, e.g. by pressure-responsive devices, by flow detectors
    • G01M3/28Investigating fluid-tightness of structures by using fluid or vacuum by measuring rate of loss or gain of fluid, e.g. by pressure-responsive devices, by flow detectors for pipes, cables or tubes; for pipe joints or seals; for valves ; for welds
    • G01M3/2807Investigating fluid-tightness of structures by using fluid or vacuum by measuring rate of loss or gain of fluid, e.g. by pressure-responsive devices, by flow detectors for pipes, cables or tubes; for pipe joints or seals; for valves ; for welds for pipes
    • G01M3/2815Investigating fluid-tightness of structures by using fluid or vacuum by measuring rate of loss or gain of fluid, e.g. by pressure-responsive devices, by flow detectors for pipes, cables or tubes; for pipe joints or seals; for valves ; for welds for pipes using pressure measurements

Abstract

The present invention provides a kind of line leakage system and method based on sound wave suction wave hybrid monitoring, is related to pipeline risk profile technical field.The system includes the pressure sensor positioned at pipeline head end and end, sonic sensor, slave computer, interchanger and host computer, slave computer control pressure sensor and sonic sensor acquisition pressure and sonic data, host computer is sent to by interchanger after being pre-processed, host computer executes leakage monitoring program therein, respectively by pressure and sonic data integrate memory module receive and parse through slave computer transmission come pressure and sonic data, by data processing module secondary filtering is carried out to obtaining data, nondimensionalization processing and semi-supervised Fei Sheer differentiations processing, judge whether pipeline leaks and carry out pressure signal and acoustic signals mixed positioning by line leakage module.The present invention can more preferably shield noise jamming, it is ensured that when signal source changes, the accuracy of recovering signal, more acurrate to leak point positioning after filtering.

Description

Line leakage system and method based on sound wave suction wave hybrid monitoring
Technical field
The present invention relates to pipeline risk profile technical field more particularly to a kind of pipes based on sound wave suction wave hybrid monitoring Road leakage monitoring system and method.
Background technology
Institute role of the pipeline transportation in economic development more come with it is important, as city water pipeline, land are former The transport of oil-piping, sea-bottom oil-gas pipeline etc., oil is largely transported in the form of product oil in the duct.With pipe network by Year enlarging, oneself warp of pipeline transportation become the major way that Land petroleum transports.But the aging of pipeline, corrosion, sudden nature Disaster and artificial destruction etc. can all cause the leakage or even rupture of processed oil pipeline, such as find and prevented not in time, not only Energy waste, economic loss, environmental pollution, and entail dangers to personal safety are caused, or even causes catastrophic failure.Therefore, right Oil-gas pipeline carries out real time on-line monitoring, carries out alarm accurately and timely to leakage accident, and accurately estimate the position of leakage point It sets and has great importance.
There are many kinds of pipeline leakage detection methods now, as optical fiber leaks hunting method, acoustic wave detection, pressure gradient method, negative Press wave method etc..Wherein, negative pressure wave method is that most widely used pipeline leakage detection method, this method have anti-in the world in recent years The features such as short between seasonable, detectable leakage rate range is wide, but for slow and smaller flow leakage, due to its unit interval Interior pressure change is slow, and negative pressure wave method is relatively low to its susceptibility, easy tos produce and fails to report, and due to the complexity of pipe-line transportation system Operating condition adjustment, the start and stop of such as main defeated pump of some common operations, the variation of the switch, control valve opening of valve can all cause to bear Wave is pressed, and the suction wave caused with leakage has very high similarity, reduces the accuracy of detection of negative pressure wave method.
Acoustic wave detection leak detection has a preferable monitoring accuracy for slow and smaller flow leakage, but it is to easy Leakage is mutually obscured with Abnormal acoustic wave caused by artificial or environmental factor, and this method cannot detect simultaneous multiple spot Leakage.
Invention content
It is a kind of based on sound wave suction wave the technical problem to be solved by the present invention is in view of the above shortcomings of the prior art, provide The line leakage system and method for hybrid monitoring, can preferably shield noise jamming, and can ensure that when signal source changes, The accuracy of recovering signal after filtering reduces wrong report, has more accurate judgement to the leakage of small flow to greatest extent, to leakage The positioning of point is more accurate.
In order to solve the above technical problems, the technical solution used in the present invention is:
On the one hand, the present invention provides a kind of line leakage system based on sound wave suction wave hybrid monitoring, including pressure Force snesor, sonic sensor, slave computer, interchanger and host computer;
The pressure sensor is contacted in the head end of pipeline and end respectively placement one with pumped (conveying) medium, for real-time The high accuracy data of acquisition reflection pipeline pressure;The sonic sensor is in the head end of pipeline and end respectively placement two, juxtaposition Pressure sensor both sides in head end and end, for acquiring the high-speed data for reflecting pipeline sound wave in real time;Head end or end The distance of two sonic sensors is not less than 20Tv, and T is the period of sound wave, and v is the spread speed of sound wave in the duct;
The slave computer, interchanger and host computer are respectively provided with one group in the head end of pipeline and end, the pressure sensor, The sonic sensor is connect with the slave computer of corresponding one end, and the slave computer passes through cable, interchanger and corresponding one The host computer connection at end;
The slave computer is programmed to execute following steps:Control the pressure sensor and sonic sensor of corresponding one end When acquiring pressure and sonic data, and pre-processed to gathered data, including filtering, be subject to according to the GPS gathers time Between stab;Host computer is sent to by interchanger after pretreated data are packaged, while being stored in the SD card of slave computer and carrying out Backup;
Include the executable leakage monitoring program of computer in the host computer, specifically includes:Pressure and sonic data Integrate memory module, data processing module, line leakage module;
Pressure integrates memory module with sonic data, for receiving and parsing through the pressure and sound wave number that slave computer transmission comes According to and being stored according to the Data Detection time, and by one group of pressure data and two groups of sonic data groups of each pipeline one end It is combined into a third-order matrix;
Data processing module, by integrated from pressure and sonic data obtained in memory module data handled and based on It calculates, including the processing of the secondary filtering of data, nondimensionalization, semi-supervised Fei Sheer differentiations processing, obtains data processing and tied with calculating Fruit;Wherein, it in the processing of the secondary filtering of data, to sound wave and is born using improved complete overall experience mode decomposition ICEEMD Pressure wave signal is decomposed, and according to the approximate entropy for calculating each component after its decomposition, filters out making an uproar in sound wave and negative pressure wave signal Sound, and sound wave interference signal is removed, obtain final filtered signal;Nondimensionalization processing is carried out to the data after secondary filtering In, by the historical pressures and sonic data of acquisition, the semi-supervised Fei Sheer of training differentiates, respectively obtains and belongs to nominal situation letter Number, the databases of gross leak signal, small leakage signal and Operating condition adjustment signal these four operating mode situations;
Line leakage module is integrated for the real time data using pressure and sound wave according to pressure and sonic data The data processing of module and data processing module judges whether pipeline leaks with result of calculation, when being determined as gross leak or small When leakage signal, leakage distance is calculated by pressure signal and acoustic signals mixing, pipe leakage point is positioned.
On the other hand, it the present invention also provides a kind of line leakage method based on sound wave suction wave hybrid monitoring, adopts Realize that this approach includes the following steps with above-mentioned monitoring system:
Step 1:The head end and end slave computer of pipeline control corresponding pressure sensor and sonic sensor collection tube road Pressure and sonic data, including pressure dataAnd sonic dataAnd to gathered data into Row filtering obtains filtered pressure signal X1、X2And acoustic signals Y11、Y12、Y21、Y22;Slave computer is per second by GPS gathers Time data is subject to timestamp to filtered pressure and acoustic signals;
Step 2:Slave computer transmits the pressure for filtering and adding timestamp and sound wave live signal to host computer, simultaneously It is stored in the SD card of slave computer and is backed up;
Step 3:Host computer receives the data that slave computer is sent, by the packet header of identification data packet with packet tail to determine The data packet needed parses data type and data value in data packet, is finally stored into the database of host computer later;
Step 4:Host computer carries out secondary filtering processing to the data received, includes the following steps:
Step 4.1:Using improved complete overall experience mode decomposition, i.e. ICEEMD, to sound wave and negative pressure wave signal into Row decomposes, each intrinsic mode function component after being decomposed;
Step 4.2:The approximate entropy for calculating each intrinsic mode function component, according to the approximate entropy-discriminate pressure signal of each component X1、X2And acoustic signals Y11、Y12、Y21、Y22In noise signal and filter out;
Step 4.3:The sound wave interference in acoustic signals in upstream station and downstream stations station is removed, after obtaining secondary filtering Signal;
Step 5:The processing of standard deviation method nondimensionalization is carried out to the signal after secondary filtering, by the historical pressures of acquisition and Sonic data, the semi-supervised Fei Sheer of training differentiates, respectively obtains and belongs to nominal situation signal, gross leak signal, small leakage signal With the database of Operating condition adjustment signal these four operating mode situations;
Step 6:Acquisition pipeline pressure and sonic data in real time pass through the secondary filtering method described in pretreatment and step 4 After carrying out secondary filtering, in seconds, it is sent into step 5 and trains obtained semi-supervised Fei Sheer discrimination models, according to what is obtained Feature vector is compared with the database for the four kinds of operating mode situations established, and calculates similarity degree between the two;
Step 7:The fault type of real time data is determined according to similarity degree, if it is determined that believing for gross leak signal or Small leak Number when, then leakage point position is calculated by pressure signal and acoustic signals mixing, pipe leakage is positioned, otherwise return to step Rapid 6, continue monitoring in real time.
Sound wave and negative pressure wave signal are decomposed using improved complete overall experience mode decomposition in the step 4.1 Detailed process be:
Step 4.1.1:I group white Gaussian noises are added to signal X (t) and generate I new signals, i-th of new signal is Xi(t)=X (t)+βkwi;Wherein, X (t) is original sound wave or negative pressure wave signal;wiFor i-th group of white Gaussian noise variable, i=1, 2、…、I;βkkstd(rk), rkFor k-th of remainder, εk=0.2;
Step 4.1.2:Loop initialization parameter, enables i=1;
Step 4.1.3:Determine signal Xi(t) all extreme points on, including Local modulus maxima and local minizing point; Fitting extreme point obtains coenvelope lineWith lower envelope lineMake Xi(t) meet:
Step 4.1.4:The average value for calculating Xi (t) two envelopes up and down, is denoted as m (t),
Step 4.1.5:With original Xi(t) signal data subtracts average value m (t) and obtains function hi(t), i.e. hi(t)=Xi (t)-m(t);
Step 4.1.6:Enable Xi(t)=hi(t), judge hi(t) whether meet two conditions of intrinsic mode function IMF:
Condition 1:On at any point in time, the mean value for the envelope that the local maximum and local minimum of IMF define is necessary It is zero, i.e., to arbitrary t, has
Condition 2:In entire data sequence, the quantity of extreme point and the quantity of zero crossing it is equal or at most be not much different in One;
If satisfied, 4.1.7 is thened follow the steps, and otherwise, return to step step 4.1.3;
Step 4.1.7:By hi(t) it is separated from X (t) signals, as IMFiComponent obtains a high frequency division of removal The function r of the difference of amounti(t), i.e. remainder ri(t)=X (t)-IMFi
Step 4.1.8:Judge residue signal ri(t) whether be monotonic function signal, if so, X (t) cannot be decomposed again Go out IMF components, completes decomposable process, obtain each IMF components, be denoted as IMF1~IMFn, n=i, and obtain residue signal to the end For rn(t)=rnn-1(t)-hn(t), signal X (t) is the sum of n intrinsic mode function component and a discrepance, i.e.,Otherwise, X (t)=r is enablediFor new signal, then i=i+1 is enabled, return to step 4.1.3 is carried out next Secondary decomposition.
The detailed process that the approximate entropy of each intrinsic mode function component is calculated in the step 4.2 is:
Step 4.2.1:Data in j-th of intrinsic mode function, which are regarded as, the time series of y point, is denoted as { Zj}= {Zj1、Zj2、…、Zjy};For given threshold value a and pattern dimension g,Calculate the two-value distance matrix B of n*n:
Wherein,
Step 4.2.2:Utilize the element b of matrix BrjCalculate separately the number of elements for being less than a and apart from total n-g+1 and n- The ratio of g+2, is denoted as respectivelyWithShown in following two formula:
Step 4.2.3:According toWithIt is calculatedWithShown in following two formula,
The then approximate entropy A of j-th of intrinsic mode functionjIt is expressed as:
Step 4.2.4:The descending arrangement of approximate entropy that will be obtained, corresponding IMF component of the approximation entropy more than 1 regard It is filtered out for noise signal, the IMF components between being 0.486~0.490 to acoustic signals approximate entropy are that sound wave is believed with pressure Number coupled signal, filtered out as interference signal.
The detailed process of the sound wave interference in acoustic signals in upstream station and downstream stations station is removed in the step 4.3 For:
Step 4.3.1:Acoustic signals after the secondary filtering obtained by step 4.2 judge its each IMF component to each sound The time of wave sensor;
Step 4.3.2:The time that the outside sonic sensor of pipeline head end receives receives earlier than inside sonic sensor To time for receiving of the intrinsic mode function of time and the outside sonic sensor of pipe end sensed earlier than inside sound wave The intrinsic mode function for the time of device received is the sound wave interference in upstream station and downstream stations station, and it is intrinsic to remove these Mode function;
Step 4.3.3:The corresponding remaining intrinsic mode function of acoustic signals is added to obtain and filters out the noise in station Sound-source signal, the as acoustic signals Y of the sonic sensor of upstream station and downstream stations1、Y2
The detailed process of the step 5 is:
Step 5.1:Data after step 4 secondary filtering are subjected to the processing of standard deviation method nondimensionalization, are combined as measuring square Battle array XL;The point that m numerical value was equal to the pressure data values at a upper time point is inserted between each pressure signal, The integer to round up is m, wherein:f1For the frequency of acoustic signals, f2For the frequency of pressure signal;
Step 5.2:Calculation matrix XL=[x1, x2, x3... xN], including normal oil transportation signal, Small leak signal, big Leakage signal and operating mode interference signal include n in kth kind classificationkA sample, k=1,2,3,4;n1+n2+n3+n4=N;
Define SbFor class scatter matrix, SwFor Scatter Matrix in class, distinguish shown in following two formula:
Wherein, weight matrix W(b)With W(w)It is respectively defined as:
Wherein, CkIndicate the set of classification number k, Ck={ 1,2,3,4 };
Define StFor global Scatter Matrix,Wherein,
Step 5.3:It enablesWherein I is unit diagonal matrix, then semi-supervised to take She Er optimization discriminant vectors are obtained by following formula:
Wherein, J is maximum scores parameter, and p is the arbitrary constant not equal to 0, and P is load matrix;
Above formula is equivalent to SrbQ=λ SrwQ, wherein λ are generalized eigenvalues, and q is corresponding generalized eigenvector;
Obtained generalized eigenvalue descending is arranged as λ1≥λ2≥…≥λN, corresponding generalized eigenvector is q1, q2..., qN, vectorial q1, q2..., qNAs semi-supervised Fei Sheer optimizes discriminant vector, and classification capacity also weakens successively;
Assuming that the prior probability that sample belongs to every one kind is equal, it isK is classification sum, then exemplarCondition Probability density function is shown below:
Wherein, Qr=[q1, q2..., qr] it is that preceding r Fei Sheer differentiates feature vector, QrThe space being turned into is the half of r dimensions It supervises Fei Sheer and differentiates subspace,It is CkThe mean vector of class sample;
According to bayesian criterion,The posterior probability calculation formula for belonging to the i-th type is:
The definition for partly superintending and directing Fei Sheer discriminant functions is
Indicate that test data concentrates arbitrary sampleBelong to the discriminant score of k-th of classification in training set;
According to Leakage classification criterion shown in following formula, each test sample in test set is predictedBelieve for normal oil transportation Number, gross leak signal, Small leak signal or operating mode interference signal:
Step 5.4:It is interfered using nominal situation signal, gross leak signal, small leakage signal and the operating mode in historical data The mathematical model that step 5.3 obtains is respectively trained in signal, obtains nominal situation signal, gross leak signal, small leakage signal and work Condition interference signal is correspondingEstablish the database of pipeline leakage testing.
The calculation formula of similarity degree is as follows in the step 6:
In formula, SkFor similarity, C (x)newFor freshly harvested data,For the test sample of kth type,For k One set of class sample,For a set of all test samples;
If Si>=0.85, then current data is kth class data;Otherwise, it is examined in conjunction with the actual analysis of field conditions and artificially It is disconnected to determine the type of failure, and taken in the database of pipeline leakage testing.
The detailed process positioned to pipeline leakage in the step 7 is:
Step 7.1:Leakage point is calculated according to the time difference of upstream station and downstream stations pressure sensor, is shown below,
In formula, Z1For leakage point to the distance of upstream station pressure sensor;l1Between the pressure sensor of pipeline upstream and downstream station Distance;τ1The time difference of upstream and downstream pressure sensor is passed to for suction wave;v1For the spread speed of suction wave in-line;
Step 7.2:Leakage is respectively chosen in the upstream station and downstream stations of leakage point, and first 30 seconds pressure data mean values of point occur With 15 seconds after leakage point pressure data mean values, upstream and downstream station pressure change ratio is calculated, respectively shown in following two formula,
Wherein:δ1、δ2The respectively pressure change ratio of upstream station and downstream stations;X1+、X2+Respectively point preceding 30 occurs for leakage The pressure data mean value of second upstream station and downstream stations;X1-、X2-Respectively the pressure of 15 seconds upstream station and downstream stations after point occurs for leakage Force data mean value;
Then δ is compared in gross pressure variation when pipe leakagepFor:Wherein, μ is by field pipes length and normally The parameter that conveyance conduit pressure size determines;
Step 7.3:To the characteristic value of sound wave in the time difference of upstream station and downstream stations, leakage point is calculated.Such as following formula institute Show,
In formula, Z2For leakage point to the distance of upstream station sonic sensor;l2Between the sonic sensor of pipeline upstream and downstream station Distance;τ2The time difference of upstream and downstream sonic sensor is passed to for sound wave;v2For the spread speed of sound wave in-line;
Step 7.4:Leakage point upstream station and downstream stations respectively choose leakage occur point preceding 5 seconds sonic data mean value with 3 seconds sonic data mean values after leakage point calculate upstream and downstream station sound wave and change ratio, respectively shown in following two formula,
Wherein, δ3、δ4Respectively the sound wave of upstream station and downstream stations changes ratio;Y1+、Y2+Point occurs for respectively leakage first 5 seconds The sonic data mean value of upstream station and downstream stations;Y1-、Y2-Respectively the sound wave of 3 seconds upstream station and downstream stations after point occurs for leakage Data mean value;
Then δ is compared in total sound wave variation when pipe leakagesFor:
Step 7.5:Final distance according to leakage point to upstream station pressure sensor is
Wherein, T is the period of sound wave.
It is using advantageous effect caused by above-mentioned technical proposal:It is provided by the invention a kind of mixed based on sound wave suction wave The line leakage method for closing monitoring, there is following advantage compared with existing system:
(1) the slave computer cost that the present invention uses is lower;
(2) it includes the independent hardware and software filtering to sound wave and pressure signal to use multiple filter, is preferably shielded The interference of noise;
(3) in data preprocessing phase, signal is decomposed by ICEEMD, and its source signal is selected by approximate entropy Corresponding intrinsic mode function, it is ensured that when signal source changes, the accuracy of recovering signal after filtering;
(4) it is interfered by the way that the maskable sound wave in station of alliteration wave sensor is arranged;
(5) the method comprehensive analysis of Fei Sheer discriminant analyses pressure and acoustic signals are used, it is ensured that the feature of leakage The accuracy of value reduces wrong report to greatest extent, has more accurate judgement to the leakage of small flow;
(6) it comprehensively utilizes sound wave and pressure signal judges leakage point, for negative pressure wave method to Small leak amount and sonic method pair Gross leak amount positions inaccurate problem, and it is more accurate to the positioning of leakage point to set up threshold values.
Description of the drawings
Fig. 1 is that the line leakage system structure provided in an embodiment of the present invention based on sound wave suction wave hybrid monitoring is shown It is intended to;
Fig. 2 is the totality of the line leakage method provided in an embodiment of the present invention based on sound wave suction wave hybrid monitoring Algorithm flow chart;
Fig. 3 is the implementation of the line leakage method provided in an embodiment of the present invention based on sound wave suction wave hybrid monitoring Flow chart;
Fig. 4 is secondary filtering method flow diagram provided in an embodiment of the present invention;
Fig. 5 is the ICEEMD decomposition result figures of nominal situation suction wave provided in an embodiment of the present invention;
The ICEEMD decomposition result figures of suction wave when Fig. 6 is leakage provided in an embodiment of the present invention;
Fig. 7 is the filtered recovering signal schematic diagram of pressure signal provided in an embodiment of the present invention;
The ICEEMD decomposition result figures of acoustic signals when Fig. 8 is nominal situation provided in an embodiment of the present invention;
The ICEEMD decomposition result figures of acoustic signals when Fig. 9 is leakage provided in an embodiment of the present invention;
Figure 10 is the filtered recovering signal schematic diagram of acoustic signals provided in an embodiment of the present invention.
In figure:1, pipeline head end pressure sensor;2, pipe end pressure sensor;3, the outside sound wave of pipeline head end passes Sensor;4, the inside sonic sensor of pipeline head end;5, the inside sonic sensor of pipe end;6, the outside sound of pipe end Wave sensor;7, slave computer;8, interchanger;9, host computer;10, enter the station valve;11, it pumps.
Specific implementation mode
With reference to the accompanying drawings and examples, the specific implementation mode of the present invention is described in further detail.Implement below Example is not limited to the scope of the present invention for illustrating the present invention.
A kind of line leakage system based on sound wave suction wave hybrid monitoring, as shown in Figure 1, including pressure sensing Device, sonic sensor, slave computer 7, interchanger 8 and host computer 9.
Pressure sensor is contacted in the head end of pipeline and end respectively placement one with pumped (conveying) medium, for acquiring in real time Reflect the high accuracy data of pipeline pressure.Pipeline head end and the pressure sensor of end are respectively such as 1 and 2 in Fig. 1.
In the present embodiment, Rosemount 3051s pressure transmitters, major parameter is as follows:
(1) the measurement pressure limit of pressure transmitter is 0~8MPa;
(2) signal resolution 0.015%, accuracy ± 0.075%, renewal rate 50Hz;
(3) output signal is 4~20mADC (two-wire system), and carrying load ability is not less than 700 Ω, power supply 24VDC;
(4) there is 1.5 times of overload capacity for bearing maximum range;
(5) environment temperature often changes the influence of 50 °F (28 DEG C) and is better than:± (the 0.025% range upper limit+0.125% is measured Journey);
(6) static pressure often changes the influence of 1000psi (6.9MPa) and is better than:± 0.1% range upper limit.
Sonic sensor is placed in head end and the pressure sensor two of end in the head end of pipeline and end respectively placement two Side, for acquiring the high-speed data for reflecting pipeline sound wave in real time.Two sonic sensors of pipeline head end are respectively such as 3 in Fig. 1 With 4, two sonic sensors of road end are respectively such as 5 and 6 in Fig. 1, wherein 3 and 6 be outside, and 4 and 5 be inside.Head end or The distance of two sonic sensors of end is not less than 20Tv, and T is the period of sound wave, and v is the spread speed of sound wave in the duct.
In the present embodiment, sonic sensor has technical indicator below using CT1000 series:
(1) voltage sensibility:947.47mv/g;
(2) measurement frequency range:0.2~1200Hz;
(3) range:5g;
(4) linearity:≤ 1%;
(5) operating temperature:- 20~120 DEG C;
(6) shock resistance:10g;
(7) way of output:Top M5;
(8) exciting current:2~10mA, driving voltage:10-24VDC.
Slave computer 7, interchanger 8 and host computer 9 are respectively provided with one group in the head end of pipeline and end, pressure sensor, sound wave Sensor is connect with the slave computer of corresponding one end 7, and slave computer 7 is connected by cable, interchanger 8 and the host computer 9 of corresponding one end It connects.
Slave computer 7 is programmed to execute following steps:The pressure sensor and sonic sensor for controlling corresponding one end are adopted Collect pressure and sonic data, and gathered data is pre-processed, including filters, carried out being subject to the time according to the GPS gathers time Stamp;Host computer 9 is sent to by interchanger 8 after pretreated data are packaged, while being stored in the SD card of slave computer 7 and carrying out Backup.
Technical requirement in the present embodiment is as follows:
(1) sample frequency range:0-1.2KHz;
(2) drive signal:3.3V and 5V;
(3) port number of sampled signal:At least 4 tunnels;
(4) sampling precision:0.001;
(5) communication mode:Ethernet communication;
(6) power supply mode:24V direct current supplys.
In the present embodiment, 50Hz is chosen for pressure signal sample frequency, even to the sample frequencys of acoustic signals 1200Hz。
Include the executable leakage monitoring program of computer in host computer 9, specifically includes:Pressure is integrated with sonic data Memory module, data processing module, line leakage module.
Pressure integrates memory module with sonic data, for receiving and parsing through the pressure and sound wave number that the transmission of slave computer 7 comes According to and being stored according to the Data Detection time, and by one group of pressure data and two groups of sonic data groups of each pipeline one end It is combined into a third-order matrix;
Data processing module, by integrated from pressure and sonic data obtained in memory module data handled and based on It calculates, including the processing of the secondary filtering of data, nondimensionalization, Fei Sheer differentiations processing, obtains data processing and result of calculation;Its In, in the secondary filtering processing of data, using improved complete overall experience mode decomposition ICEEMD to sound wave and suction wave Signal is decomposed, and according to the approximate entropy for calculating its each component after decomposing, filters out the noise in sound wave and negative pressure wave signal, and Sound wave interference signal is removed, final filtered signal is obtained;Data after secondary filtering are carried out in nondimensionalization processing, are led to The historical pressures and sonic data of acquisition are crossed, the semi-supervised Fei Sheer of training differentiates, respectively obtains and belongs to nominal situation signal, lets out greatly The database of leakage signal, small leakage signal and Operating condition adjustment signal these four operating mode situations;
Line leakage module is integrated for the real time data using pressure and sound wave according to pressure and sonic data The data processing of module and data processing module judges whether pipeline leaks with result of calculation, when being determined as gross leak or small When leakage signal, leakage distance is calculated by pressure signal and acoustic signals mixing, pipe leakage point is positioned.
Line leakage method based on sound wave suction wave hybrid monitoring, totality are realized using above-mentioned monitoring system Algorithm as shown in Fig. 2, specific implementation flow chart as shown in figure 3, that the specific method is as follows is described.
Step 1:The head end and end slave computer of pipeline control corresponding pressure sensor and sonic sensor collection tube road Pressure and sonic data, including pressure dataAnd sonic dataAnd to gathered data into Row filtering obtains filtered pressure signal X1、X2And acoustic signals Y11、Y12、Y21、Y22
Due to pipeline long transmission distance, the rate of decay is fast in the duct for the high-frequency noise that pipe leakage generates, therefore slave computer Collect the sound wave that the high frequency section of acoustic signals can be considered noise signal, therefore low-pass filtering rejection frequency is used to be more than 200Hz Signal.Filter out the interference such as gross error, the high-frequency noise in slave computer pressure signal.
The slave computer time data per second by GPS gathers is subject to timestamp to filtered pressure and acoustic signals.
Step 2:Slave computer transmits the pressure for filtering and adding timestamp and sound wave live signal to host computer, simultaneously It is stored in the SD card of slave computer and is backed up.
The data within nearest 60 minutes are stored in SD card.When host computer because the reasons such as network delay do not receive certain for the moment Between section data packet when, send and ask to slave computer, at next second bottom chance by the data packet of current time together with host computer The data of missing are sent to host computer together.
Step 3:Host computer receives the data that slave computer is sent, by the packet header of identification data packet with packet tail to determine The data packet needed parses data type and data value in data packet, is finally stored into the database of host computer later.
According to sound wave and the pipeline leakage signal of suction wave generation and the mechanism of transmission analysis it is found that leakage when sound wave Signal and negative pressure wave signal have fuctuation within a narrow range in frequency domain, and traditional algorithm applied in pipeline leakage testing is not right It is optimized, and can be positioned to subsequent leakage and be generated interference, influence its positioning accuracy.The present embodiment is using ICEEMD to sound Involve negative pressure wave signal to be decomposed, according to the approximate entropy for calculating its each component after decomposing, choose final filtered signal, It is as follows.
Step 4:Host computer carries out secondary filtering processing to the data received, as shown in figure 4, including the following steps:
Step 4.1:Using improved complete overall experience mode decomposition, i.e. ICEEMD, to sound wave and negative pressure wave signal into Row decomposes, each intrinsic mode function component after being decomposed;Detailed process is:
Step 4.1.1:I group white Gaussian noises are added to signal X (t) and generate I new signals, i-th of new signal is Xi(t)=X (t)+βkwi;Wherein, X (t) is original sound wave or negative pressure wave signal;wiFor i-th group of white Gaussian noise variable, i=1, 2、…、I;βkkstd(rk), rkFor k-th of remainder, εk=0.2;
In the present embodiment, according to the analysis to ICEEMD and IMF approximate entropies, sound wave and pressure signal to experiment acquisition into The Gaussian noise that 200 groups of signal-to-noise ratio are 5 is added in row data processing, and the maximum iteration choosing of EMD is then 500 times.
Step 4.1.2:Loop initialization parameter, enables i=1;
Step 4.1.3:Determine signal Xi(t) all extreme points on, including Local modulus maxima and local minizing point; Fitting extreme point obtains coenvelope lineWith lower envelope lineMake Xi(t) meet:
Step 4.1.4:Calculate Xi(t) average value of two envelopes up and down, is denoted as m (t),
Step 4.1.5:With original Xi(t) signal data subtracts average value m (t) and obtains function hi(t), i.e. hi(t)=Xi (t)-m(t);
Step 4.1.6:Enable Xi(t)=hi(t), judge hi(t) whether meet two conditions of intrinsic mode function IMF:
Condition 1:On at any point in time, the mean value for the envelope that the local maximum and local minimum of IMF define is necessary It is zero, i.e., to arbitrary t, has
Condition 2:In entire data sequence, the quantity of extreme point and the quantity of zero crossing it is equal or at most be not much different in One;
If satisfied, 4.1.7 is thened follow the steps, and otherwise, return to step step 4.1.3;
First restrictive condition is to ensure waveform Local Symmetric, and second restrictive condition is approximate traditional stable Gaussian mistake The definition about narrowband of journey;
Step 4.1.7:By hi(t) it is separated from X (t) signals, as IMFiComponent obtains a high frequency division of removal The function r of the difference of amounti(t), i.e. remainder ri(t)=X (t)-IMFi
First from Xi(t) the IMF components, that is, h obtained in1(t), it is denoted as IMFi1, i=1,2 ..., I, to all IMFi1 It takes and is worth to IMF1, remainder is r at this time1=X (t)-IMF1
Step 4.1.8:Judge residue signal ri(t) whether be monotonic function signal, if so, X (t) cannot be decomposed again Go out IMF components, completes decomposable process, obtain each IMF components, be denoted as IMF1~IMFn, n=i, and obtain residue signal to the end For rn(t)=rn-1(t)-hn(t), signal X (t) is the sum of n intrinsic mode function component and a discrepance, i.e.,Otherwise, X (t)=r is enablediFor new signal, then i=i+1 is enabled, return to step 4.1.3 is carried out next Secondary decomposition.
Step 4.2:Calculate each intrinsic mode function component IMF1~IMFnApproximate entropy, sentenced according to the approximate entropy of each component Other pressure signal X1、X2And acoustic signals Y11、Y12、Y21、Y22In noise signal and filter out, detailed process is:
Step 4.2.1:Data in j-th of intrinsic mode function, which are regarded as, the time series of y point, is denoted as { Zj}= {Zj1、Zj2、…、Zjy};For given threshold value a and pattern dimension g,Calculate the two-value distance matrix B of n*n:
Wherein,
In the present embodiment, a takes 0.1~0.2, g to take 2.
Step 4.2.2:Utilize the element b of matrix BrjCalculate separately the number of elements for being less than a and apart from total n-g+1 and n- The ratio of g+2, is denoted as respectivelyWithShown in following two formula:
Step 4.2.3:According toWithIt is calculatedWith(To CrNatural logrithm is taken, then asks being averaged for all r Value), shown in following two formula,
The then approximate entropy A of j-th of intrinsic mode functionjIt is expressed as:
Step 4.2.4:The descending arrangement of approximate entropy that step 4.2.3 is obtained, corresponding approximation entropy are more than 1 IMF components are considered as noise signal and are filtered out, and the IMF components between being 0.486~0.490 to acoustic signals approximate entropy are sound wave With the coupled signal of pressure signal, filtered out as interference signal.
Step 4.3:The sound wave interference in acoustic signals in upstream station and downstream stations station is removed, after obtaining secondary filtering Signal, detailed process is:
Step 4.3.1:Acoustic signals after the secondary filtering obtained by step 4.2 judge its each IMF component to each sound The time of wave sensor;
Step 4.3.2:The time that the outside sonic sensor of pipeline head end receives receives earlier than inside sonic sensor To time for receiving of the intrinsic mode function of time and the outside sonic sensor of pipe end sensed earlier than inside sound wave The intrinsic mode function for the time of device received is the sound wave interference in upstream station and downstream stations station, and it is intrinsic to remove these Mode function;
Step 4.3.3:The corresponding remaining intrinsic mode function of acoustic signals is added to obtain and filters out the noise in station Sound-source signal, the as acoustic signals Y of the sonic sensor of upstream station and downstream stations1、Y2
In the present embodiment, the ICEEMD of nominal situation suction wave is decomposed as shown in figure 5, the corresponding approximation of its each IMF component Entropy is as shown in table 1.By Fig. 5 and table 1 it is found that in normal conditions, the approximate entropy of the IMF1 of pressure signal is 1.2854, and r pairs Effect of signals very little and fluctuating range very little, therefore pressure signal can remove high-frequency noise by IMF2-IMF10 signals revivifications And the very low frequencies signal useless to signal characteristic abstraction.
The approximate entropy of the IMF functions of 1 nominal situation suction wave of table
Components number Approximate entropy Components number Approximate entropy Components number Approximate entropy
IMF1 1.2854 IMF5 0.2562 IMF9 0.0052
IMF2 0.8544 IMF6 0.1275 IMF10 0.0037
IMF3 0.6254 IMF7 0.0426 r 6.3882*e-4
IMF4 0.5747 IMF8 0.0214 - -
The ICEEMD of suction wave is decomposed as shown in fig. 6, the corresponding approximate entropy of its each IMF component is as shown in table 2 when leakage.By Fig. 6 and table 2 it is found that the approximate entropy that the approximate entropy of the IMF1 of pressure signal is 1.7260, IMF2 in the case where leaking operating mode is 1.1563, The two components can be considered as HF noise signal.And the influence very little of IMF11, IMF12 and r these three components to signal, because This pressure signal can be can remove high-frequency noise and believed the useless very low frequencies of signal characteristic abstraction by IMF3-IMF10 signals revivifications Number.Signal after pressure signal filtering reduction is as shown in Figure 7, wherein the recovering signal of pressure signal when figure a is nominal situation, The recovering signal of pressure signal when figure b is leakage.
The approximate entropy of the IMF functions of suction wave when table 2 leaks
Components number Approximate entropy Components number Approximate entropy Components number Approximate entropy
IMF1 1.7260 IMF6 0.0706 IMF11 0.0004
IMF2 1.1563 IMF7 0.0420 IMF12 7.6097*e-4
IMF3 0.5579 IMF8 0.0209 r 2.3579*e-4
IMF4 0.2889 IMF9 0.0018 - -
IMF5 0.0622 IMF10 0.0010 - -
The ICEEMD of acoustic signals is decomposed as shown in figure 8, the corresponding approximate entropy such as table 3 of its each IMF component when nominal situation It is shown.By Fig. 8 and table 3 it is found that the value of the approximate entropy of all IMF components of acoustic signals is respectively less than 1 when nominal situation, and do not deposit In the minimum component of approximate entropy, therefore acoustic signals can be restored by whole IMF component signals in normal conditions.
The approximate entropy of the IMF functions of acoustic signals when 3 nominal situation of table
Components number Approximate entropy Components number Approximate entropy Components number Approximate entropy
IMF1 0.9229 IMF6 0.2007 IMF11 0.0057
IMF2 0.4960 IMF7 0.0973 IMF12 0.0014
IMF3 0.3220 IMF8 0.0444 r 0.0036
IMF4 0.1938 IMF9 0.0261 - -
IMF5 0.2133 IMF10 0.0203 - -
The ICEEMD of acoustic signals is decomposed as shown in figure 9, the corresponding approximate entropy of its each IMF component is as shown in table 4 when leakage. By Fig. 9 and table 4 it is found that noise largely exists in leakage, the approximate entropy that the approximate entropy of IMF1 is 1.2255, IMF2 is 1.0348, it is known that IMF1 and IMF2 is noise signal, and the approximate entropy of surplus r is 7.8322*e-4, the influence to signal is very It is small, therefore IMF1, IMF2 and surplus r must be filtered out when signals revivification, therefore acoustic signals can be by IMF3-IMF11 component signals It is restored.Signal after acoustic signals filtering reduction is as shown in Figure 10, wherein acoustic signals goes back when figure a is nominal situation Original signal, the recovering signal of acoustic signals when figure b is leakage.
The approximate entropy of the IMF functions of acoustic signals when table 4 leaks
Components number Approximate entropy Components number Approximate entropy Components number Approximate entropy
IMF1 1.2255 IMF5 0.4863 IMF9 0.0214
IMF2 1.0348 IMF6 0.2553 IMF10 0.0068
IMF3 0.6796 IMF7 0.1212 IMF11 0.0060
IMF4 0.5723 IMF8 0.0480 r 7.8322*e-4
Step 5:The processing of standard deviation method nondimensionalization is carried out to the signal after secondary filtering, by the historical pressures of acquisition and Sonic data, the semi-supervised Fei Sheer of training differentiates, respectively obtains and belongs to nominal situation signal, gross leak signal, small leakage signal With the database of Operating condition adjustment signal these four operating mode situations;
Based on core Fei Sheer discriminant analyses and according to the theory of semi-supervised learning method, let out with analysis of history by learning Database when dew occurs is come when judging that host computer will monitor Wave anomaly from now on, if is really to leak.Detailed process It is as follows:
Step 5.1:Data after step 4 secondary filtering are subjected to the processing of standard deviation method nondimensionalization, are combined as measuring square Battle array XL;To ensure the time consistency of acoustic signals and pressure signal, needs to expand pressure signal, believe in each pressure The point that m numerical value was equal to the pressure data values at a upper time point is inserted between number, The integer to round up is i.e. For m, wherein:f1For the frequency of acoustic signals, f2For the frequency of pressure signal;
Step 5.2:Calculation matrix XL=[x1, x2, x3... xN], including normal oil transportation signal, Small leak signal, big Leakage signal and operating mode interference signal include n in kth kind classificationkA sample, k=1,2,3,4;n1+n2+n3+n4=N;
Define SbFor class scatter matrix, SwFor Scatter Matrix in class, distinguish shown in following two formula:
Wherein, weight matrix W(b)With W(w)It is respectively defined as:
Wherein, CkIndicate the set of classification number k, Ck={ 1,2,3,4 };
Define StFor global Scatter Matrix,Wherein,
Step 5.3:It enablesWherein I is unit diagonal matrix, then semi-supervised to take She Er optimization discriminant vectors are obtained by following formula:
Wherein, J is maximum scores parameter, and p is the arbitrary constant not equal to 0, and P is load matrix;
Above formula is equivalent to SrbQ=λ SrwQ, wherein λ are generalized eigenvalues, and q is corresponding generalized eigenvector;
Obtained generalized eigenvalue descending is arranged as λ1≥λ2≥…≥λN, corresponding generalized eigenvector is q1, q2..., qN, vectorial q1, q2..., qNAs semi-supervised Fei Sheer optimizes discriminant vector, and classification capacity also weakens successively;This reality It applies and considers calculating speed problem in example, only take first four, i.e. N=4;
Under normal conditions, it is to meet multivariate Gaussian distribution that the data under nominal situation, which may be assumed that, and leak data can also It is considered to meet Gaussian Profile.Assuming that the prior probability that sample belongs to every one kind is equal, it isK is classification sum, then label SampleConditional probability density function be shown below:
Wherein, Qr=[q1, q2..., qr] it is that preceding r Fei Sheer differentiates feature vector, QrThe space being turned into is the half of r dimensions It supervises Fei Sheer and differentiates subspace,It is CkThe mean vector of class sample;
According to bayesian criterion,The posterior probability calculation formula for belonging to the i-th type is:
The definition for partly superintending and directing Fei Sheer discriminant functions is
Indicate that test data concentrates arbitrary sampleBelong to the discriminant score of k-th of classification in training set;That , each sample point in test setBelong to the discriminant value of k-th of classification, can be calculated by above formula;
According to Leakage classification criterion shown in following formula, each test sample in test set is predictedBelieve for normal oil transportation Number, gross leak signal, Small leak signal or operating mode interference signal:
Step 5.4:It is interfered using nominal situation signal, gross leak signal, small leakage signal and the operating mode in historical data The mathematical model that step 5.3 obtains is respectively trained in signal, obtains nominal situation signal, gross leak signal, small leakage signal and work Condition interference signal is correspondingEstablish the database of pipeline leakage testing.
Step 6:Acquisition pipeline pressure and sonic data in real time pass through the secondary filtering method described in pretreatment and step 4 After carrying out secondary filtering, in seconds, it is sent into step 5 and trains obtained semi-supervised Fei Sheer discrimination models, according to what is obtained Feature vector is compared with the database of the pipeline leakage testing of foundation, calculates similarity degree between the two, as follows:
In formula, SkFor similarity, C (x)newFor freshly harvested data,For the test sample of kth type,For k One set of class sample,For a set of all test samples;
If Si>=0.85, then current data is kth class data;Otherwise, any feature in current data and database to The similarity of amount is both less than the threshold value 0.85, then a new failure not being identified is likely to, in conjunction with the reality of field conditions Border is analyzed and artificial diagnosis determines the type of failure, and is taken in the database of pipeline leakage testing.
Step 7:The fault type of real time data is determined according to similarity degree, if it is determined that believing for gross leak signal or Small leak Number when, then leakage point position is calculated by pressure signal and acoustic signals mixing, pipe leakage is positioned, otherwise return to step Rapid 6, continue monitoring in real time.
The detailed process positioned to pipeline leakage is:
Step 7.1:Leakage point is calculated according to the time difference of upstream station and downstream stations pressure sensor, is shown below,
In formula, Z1For leakage point to the distance of upstream station pressure sensor;l1Between the pressure sensor of pipeline upstream and downstream station Distance;τ1The time difference of upstream and downstream pressure sensor is passed to for suction wave;v1For the spread speed of suction wave in-line;
Step 7.2:Leakage is respectively chosen in the upstream station and downstream stations of leakage point, and first 30 seconds pressure data mean values of point occur With 15 seconds after leakage point pressure data mean values, upstream and downstream station pressure change ratio is calculated, respectively shown in following two formula,
Wherein:δ1、δ2The respectively pressure change ratio of upstream station and downstream stations;X1+、X2+Respectively point preceding 30 occurs for leakage The pressure data mean value of second upstream station and downstream stations;X1-、X2-Respectively the pressure of 15 seconds upstream station and downstream stations after point occurs for leakage Force data mean value;
Then δ is compared in gross pressure variation when pipe leakagepFor:Wherein, μ is by field pipes length and normally The parameter that conveyance conduit pressure size determines;
Step 7.3:To the characteristic value of sound wave in the time difference of upstream station and downstream stations, leakage point is calculated.Such as following formula institute Show,
In formula, Z2For leakage point to the distance of upstream station sonic sensor;l2Between the sonic sensor of pipeline upstream and downstream station Distance;τ2The time difference of upstream and downstream sonic sensor is passed to for sound wave;V2 is the spread speed of sound wave in-line;
Step 7.4:Leakage point upstream station and downstream stations respectively choose leakage occur point preceding 5 seconds sonic data mean value with 3 seconds sonic data mean values after leakage point calculate upstream and downstream station sound wave and change ratio, respectively shown in following two formula,
Wherein, δ3、δ4Respectively the sound wave of upstream station and downstream stations changes ratio;Y1+、Y2+Point occurs for respectively leakage first 5 seconds The sonic data mean value of upstream station and downstream stations;Y1-、Y2-Respectively the sound wave of 3 seconds upstream station and downstream stations after point occurs for leakage Data mean value;
Then δ is compared in total sound wave variation when pipe leakagesFor:
Step 7.5:For shielding gross leak when acoustic location low precision, the problem of suction wave positioning accuracy difference when Small leak, Final distance according to leakage point to upstream station pressure sensor is
Wherein, T is the period of sound wave.
One is judged with pressure wave and calculates leakage point, and one is judged with sound wave and calculates leakage point, and two kinds of wave mixing are sentenced The disconnected result obtained is more acurrate.
Finally it should be noted that:The above embodiments are merely illustrative of the technical solutions of the present invention, rather than its limitations;Although Present invention has been described in detail with reference to the aforementioned embodiments, it will be understood by those of ordinary skill in the art that:It still may be used To modify to the technical solution recorded in previous embodiment, either which part or all technical features are equal It replaces;And these modifications or replacements, model defined by the claims in the present invention that it does not separate the essence of the corresponding technical solution It encloses.

Claims (8)

1. a kind of line leakage system based on sound wave suction wave hybrid monitoring, it is characterised in that:The system includes pressure Sensor, sonic sensor, slave computer, interchanger and host computer;
The pressure sensor is contacted in the head end of pipeline and end respectively placement one with pumped (conveying) medium, for acquiring in real time Reflect the high accuracy data of pipeline pressure;The sonic sensor is placed in head in the head end of pipeline and end respectively placement two The pressure sensor both sides at end and end, for acquiring the high-speed data for reflecting pipeline sound wave in real time;Two of head end or end The distance of sonic sensor is not less than 20Tv, and T is the period of sound wave, and v is the spread speed of sound wave in the duct;
The slave computer, interchanger and host computer are respectively provided with one group in the head end of pipeline and end, the pressure sensor, described Sonic sensor is connect with the slave computer of corresponding one end, and the slave computer passes through cable, interchanger and corresponding one end The host computer connection;
The slave computer is programmed to execute following steps:Control the pressure sensor and sonic sensor acquisition of corresponding one end Pressure and sonic data, and gathered data is pre-processed, including filter, carried out being subject to timestamp according to the GPS gathers time; Host computer is sent to by interchanger after pretreated data are packaged, while being stored in the SD card of slave computer and being backed up;
Include the executable leakage monitoring program of computer in the host computer, specifically includes:Pressure is integrated with sonic data Memory module, data processing module, line leakage module;
Pressure integrates memory module with sonic data, for receiving and parsing through the pressure and sonic data that slave computer transmission comes, and It is stored according to the Data Detection time, and one group of pressure data of each pipeline one end and two groups of sonic datas is combined as one A third-order matrix;
Data processing module is handled and is calculated for integrating acquisition data in memory module with sonic data from pressure, wrapped Include data secondary filtering, nondimensionalization processing, semi-supervised Fei Sheer differentiations processing, obtain data processing and result of calculation;Its In, in the secondary filtering processing of data, using improved complete overall experience mode decomposition ICEEMD to sound wave and suction wave Signal is decomposed, and according to the approximate entropy for calculating its each component after decomposing, filters out the noise in sound wave and negative pressure wave signal, and Sound wave interference signal is removed, final filtered signal is obtained;Data after secondary filtering are carried out in nondimensionalization processing, are led to The historical pressures and sonic data of acquisition are crossed, the semi-supervised Fei Sheer of training differentiates, respectively obtains and belongs to nominal situation signal, lets out greatly The database of leakage signal, small leakage signal and Operating condition adjustment signal these four operating mode situations;
Line leakage module integrates module for the real time data using pressure and sound wave according to pressure and sonic data Data processing with data processing module judges whether pipeline leaks with result of calculation, when being determined as gross leak or Small leak When signal, leakage distance is calculated by pressure signal and acoustic signals mixing, pipe leakage point is positioned.
2. a kind of line leakage method based on sound wave suction wave hybrid monitoring is based on sound using described in claim 1 The line leakage system of wave suction wave hybrid monitoring is realized, it is characterised in that:This approach includes the following steps:
Step 1:The head end and end slave computer of pipeline control corresponding pressure sensor and sonic sensor collection tube road pressure And sonic data, including pressure dataAnd sonic dataAnd gathered data is filtered Obtain filtered pressure signal X1、X2And acoustic signals Y11、Y12、Y21、Y22;The slave computer time number per second by GPS gathers It is subject to timestamp according to filtered pressure and acoustic signals;
Step 2:Slave computer transmits the pressure for filtering and adding timestamp and sound wave live signal to host computer, is stored in simultaneously It is backed up in the SD card of slave computer;
Step 3:Host computer receives the data that slave computer is sent, and the packet header by identifying data packet is required to determine with packet tail Data packet, later parse data packet in data type and data value, be finally stored into the database of host computer;
Step 4:Host computer carries out secondary filtering processing to the data received, includes the following steps:
Step 4.1:Using improved complete overall experience mode decomposition, i.e. ICEEMD divides sound wave and negative pressure wave signal Solution, each intrinsic mode function component after being decomposed;
Step 4.2:The approximate entropy for calculating each intrinsic mode function component, according to the approximate entropy-discriminate pressure signal X of each component1、X2 And acoustic signals Y11、Y12、Y21、Y22In noise signal and filter out;
Step 4.3:The sound wave interference in acoustic signals in upstream station and downstream stations station is removed, the letter after secondary filtering is obtained Number;
Step 5:The processing of standard deviation method nondimensionalization is carried out to the signal after secondary filtering, passes through the historical pressures and sound wave of acquisition Data, the semi-supervised Fei Sheer of training differentiates, respectively obtains and belongs to nominal situation signal, gross leak signal, small leakage signal and work Condition adjusts the database of these four operating mode situations of signal;
Step 6:Acquisition pipeline pressure and sonic data in real time are carried out by the secondary filtering method described in pretreatment and step 4 After secondary filtering, in seconds, it is sent into step 5 and trains obtained semi-supervised Fei Sheer discrimination models, according to obtained feature Vector is compared with the database for the four kinds of operating mode situations established, and calculates similarity degree between the two;
Step 7:The fault type of real time data is determined according to similarity degree, if it is determined that being gross leak signal or Small leak signal When, then leakage point position is calculated by pressure signal and acoustic signals mixing, pipe leakage is positioned, otherwise return to step 6, continue monitoring in real time.
3. the line leakage method according to claim 2 based on sound wave suction wave hybrid monitoring, it is characterised in that: The specific mistake that sound wave and negative pressure wave signal are decomposed using improved complete overall experience mode decomposition in the step 4.1 Cheng Wei:
Step 4.1.1:I group white Gaussian noises are added to signal X (t) and generate I new signals, i-th of new signal is Xi(t) =X (t)+βkwi;Wherein, X (t) is original sound wave or negative pressure wave signal;wiFor i-th group of white Gaussian noise variable, i=1,2 ..., I; βkkstd(rk), rkFor k-th of remainder, εk=0.2;
Step 4.1.2:Loop initialization parameter, enables i=1;
Step 4.1.3:Determine signal Xi(t) all extreme points on, including Local modulus maxima and local minizing point;Fitting Extreme point obtains coenvelope lineWith lower envelope lineMake Xi(t) meet:
Step 4.1.4:Calculate Xi(t) average value of two envelopes up and down, is denoted as m (t),
Step 4.1.5:With original Xi(t) signal data subtracts average value m (t) and obtains function hi(t), i.e. hi(t)=Xi(t)-m (t);
Step 4.1.6:Enable Xi(t)=hi(t), judge hi(t) whether meet two conditions of intrinsic mode function IMF:
Condition 1:On at any point in time, the mean value for the envelope that the local maximum and local minimum of IMF define is necessary for zero, I.e. to arbitrary t, have
Condition 2:In entire data sequence, the quantity of extreme point and the quantity of zero crossing are equal or be at most not much different in one It is a;
If satisfied, 4.1.7 is thened follow the steps, and otherwise, return to step step 4.1.3;
Step 4.1.7:By hi(t) it is separated from X (t) signals, as IMFiComponent obtains one and removes high fdrequency component The function r of differencei(t), i.e. remainder ri(t)=X (t)-IMFi
Step 4.1.8:Judge residue signal ri(t) whether be monotonic function signal, if so, X (t) cannot decomposite IMF again Component completes decomposable process, obtains each IMF components, be denoted as IMF1~IMFn, n=i, and the residue signal obtained to the end is rn (t)=rn-1(t)-hn(t), signal X (t) is the sum of n intrinsic mode function component and a discrepance, i.e.,Otherwise, X (t)=r is enablediFor new signal, then i=i+1 is enabled, return to step 4.1.3 is carried out next Secondary decomposition.
4. the line leakage method according to claim 2 based on sound wave suction wave hybrid monitoring, it is characterised in that: The detailed process that the approximate entropy of each intrinsic mode function component is calculated in the step 4.2 is:
Step 4.2.1:Data in j-th of intrinsic mode function, which are regarded as, the time series of y point, is denoted as { Zj}={ Zj1、 Zj2、…、Zjy};For given threshold value a and pattern dimension g,Calculate the two-value distance matrix B of n*n:
Wherein,
Step 4.2.2:Utilize the element b of matrix BrjCalculate separately the number of elements for being less than a and apart from total n-g+1 and n-g+2 Ratio, be denoted as respectivelyWithShown in following two formula:
Step 4.2.3:According toWithIt is calculatedWithShown in following two formula,
The then approximate entropy A of j-th of intrinsic mode functionjIt is expressed as:
Step 4.2.4:The descending arrangement of approximate entropy that will be obtained, corresponding IMF component of the approximation entropy more than 1, which is considered as, makes an uproar Acoustical signal is filtered out, and the IMF components between being 0.486~0.490 to acoustic signals approximate entropy are sound wave and pressure signal Coupled signal is filtered out as interference signal.
5. the line leakage method according to claim 2 based on sound wave suction wave hybrid monitoring, it is characterised in that: The detailed process that the sound wave in acoustic signals in upstream station and downstream stations station interferes is removed in the step 4.3 is:
Step 4.3.1:Acoustic signals after the secondary filtering obtained by step 4.2 judge that its each IMF component is passed to each sound wave The time of sensor;
Step 4.3.2:What the time that the outside sonic sensor of pipeline head end receives received earlier than inside sonic sensor The time that the intrinsic mode function of time and the outside sonic sensor of pipe end receive is earlier than inside sonic sensor The intrinsic mode function of the time received is the sound wave interference in upstream station and downstream stations station, removes these intrinsic mode Function;
Step 4.3.3:It is added the corresponding remaining intrinsic mode function of acoustic signals to obtain the sound for filtering out the noise in station The acoustic signals Y of the sonic sensor of source signal, as upstream station and downstream stations1、Y2
6. the line leakage method according to claim 2 based on sound wave suction wave hybrid monitoring, it is characterised in that: The detailed process of the step 5 is:
Step 5.1:Data after step 4 secondary filtering are subjected to the processing of standard deviation method nondimensionalization, are combined as calculation matrix XL; The point that m numerical value was equal to the pressure data values at a upper time point is inserted between each pressure signal, Four houses Five integers entered are m, wherein:f1For the frequency of acoustic signals, f2For the frequency of pressure signal;
Step 5.2:Calculation matrix XL=[x1, x2, x3... xN], including normal oil transportation signal, Small leak signal, gross leak Signal and operating mode interference signal include n in kth kind classificationkA sample, k=1,2,3,4;n1+n2+n3+n4=N;
Define SbFor class scatter matrix, SwFor Scatter Matrix in class, distinguish shown in following two formula:
Wherein, weight matrix W(b)With W(w)It is respectively defined as:
Wherein, CkIndicate the set of classification number k, Ck={ 1,2,3,4 };
Define StFor global Scatter Matrix,Wherein,
Step 5.3:It enablesWherein I is unit diagonal matrix, then semi-supervised Fei Sheer Optimization discriminant vector is obtained by following formula:
Wherein, J is maximum scores parameter, and p is the arbitrary constant not equal to 0, and P is load matrix;
Above formula is equivalent to SrbQ=λ SrwQ, wherein λ are generalized eigenvalues, and q is corresponding generalized eigenvector;
Obtained generalized eigenvalue descending is arranged as λ1≥λ2≥…≥λN, corresponding generalized eigenvector is q1, q2..., qN, Vectorial q1, q2..., qNAs semi-supervised Fei Sheer optimizes discriminant vector, and classification capacity also weakens successively;
Assuming that the prior probability that sample belongs to every one kind is equal, it isK is classification sum, then exemplarConditional probability it is close Degree function is shown below:
Wherein, Qr=[q1, q2..., qr] it is that preceding r Fei Sheer differentiates feature vector, QrThe space being turned into is the semi-supervised of r dimensions Fei Sheer differentiates subspace,It is CkThe mean vector of class sample;
According to bayesian criterion,The posterior probability calculation formula for belonging to the i-th type is:
The definition for partly superintending and directing Fei Sheer discriminant functions is
Indicate that test data concentrates arbitrary sampleBelong to the discriminant score of k-th of classification in training set;
According to Leakage classification criterion shown in following formula, each test sample in test set is predictedFor normal oil transportation signal, greatly Leakage signal, Small leak signal or operating mode interference signal:
Step 5.4:Utilize nominal situation signal, gross leak signal, small leakage signal and the operating mode interference signal in historical data The mathematical model that step 5.3 obtains is respectively trained, it is dry to obtain nominal situation signal, gross leak signal, small leakage signal and operating mode It is corresponding to disturb signalEstablish the database of pipeline leakage testing.
7. the line leakage method according to claim 6 based on sound wave suction wave hybrid monitoring, it is characterised in that: The calculation formula of similarity degree is as follows in the step 6:
In formula, SkFor similarity, C (x)newFor freshly harvested data,For the test sample of kth type,For k class samples This set,For a set of all test samples;
If Si0.85, then current data be kth class data;Otherwise, it is determined in conjunction with the actual analysis of field conditions and artificial diagnosis The type of failure, and taken in the database of pipeline leakage testing.
8. the line leakage method according to claim 2 based on sound wave suction wave hybrid monitoring, it is characterised in that: The detailed process positioned to pipeline leakage in the step 7 is:
Step 7.1:Leakage point is calculated according to the time difference of upstream station and downstream stations pressure sensor, is shown below,
In formula, Z1For leakage point to the distance of upstream station pressure sensor;l1Between the pressure sensor of pipeline upstream and downstream station away from From;τ1The time difference of upstream and downstream pressure sensor is passed to for suction wave;v1For the spread speed of suction wave in-line;
Step 7.2:Leakage is respectively chosen in the upstream station and downstream stations of leakage point first 30 seconds pressure data mean values of point occur and let out 15 seconds pressure data mean values after leak source calculate upstream and downstream station pressure change ratio, respectively shown in following two formula,
Wherein:δ1、δ2The respectively pressure change ratio of upstream station and downstream stations;X1+、X2+Respectively leakage occurs to put on first 30 seconds The pressure data mean value at trip station and downstream stations;X1-、X2-Respectively the number pressure of 15 seconds upstream station and downstream stations after point occurs for leakage According to mean value;
Then δ is compared in gross pressure variation when pipe leakagepFor:Wherein, μ is by field pipes length and normal conveying The parameter that pipeline pressure size determines;
Step 7.3:To the characteristic value of sound wave in the time difference of upstream station and downstream stations, leakage point is calculated, is shown below,
In formula, Z2For leakage point to the distance of upstream station sonic sensor;l2Between the sonic sensor of pipeline upstream and downstream station away from From;τ2The time difference of upstream and downstream sonic sensor is passed to for sound wave;v2For the spread speed of sound wave in-line;
Step 7.4:Leakage is respectively chosen in the upstream station and downstream stations of leakage point, and first 5 seconds sonic data mean values of point and leakage occurs 3 seconds sonic data mean values after point calculate upstream and downstream station sound wave and change ratio, respectively shown in following two formula,
Wherein, δ3、δ4Respectively the sound wave of upstream station and downstream stations changes ratio;Y1+、Y2+Respectively the preceding 5 seconds upstreams of point occur for leakage It stands and the sonic data mean value of downstream stations;Y1-、Y2-Respectively the sonic data of 3 seconds upstream station and downstream stations after point occurs for leakage Mean value;
Then δ is compared in total sound wave variation when pipe leakagesFor:
Step 7.5:Final distance according to leakage point to upstream station pressure sensor is
Wherein, T is the period of sound wave.
CN201810260988.3A 2018-03-28 2018-03-28 Line leakage system and method based on sound wave suction wave hybrid monitoring CN108488638B (en)

Priority Applications (1)

Application Number Priority Date Filing Date Title
CN201810260988.3A CN108488638B (en) 2018-03-28 2018-03-28 Line leakage system and method based on sound wave suction wave hybrid monitoring

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
CN201810260988.3A CN108488638B (en) 2018-03-28 2018-03-28 Line leakage system and method based on sound wave suction wave hybrid monitoring

Publications (2)

Publication Number Publication Date
CN108488638A true CN108488638A (en) 2018-09-04
CN108488638B CN108488638B (en) 2019-08-20

Family

ID=63316736

Family Applications (1)

Application Number Title Priority Date Filing Date
CN201810260988.3A CN108488638B (en) 2018-03-28 2018-03-28 Line leakage system and method based on sound wave suction wave hybrid monitoring

Country Status (1)

Country Link
CN (1) CN108488638B (en)

Cited By (6)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN109058771A (en) * 2018-10-09 2018-12-21 东北大学 The pipeline method for detecting abnormality of Markov feature is generated and is spaced based on sample
CN109115884A (en) * 2018-09-27 2019-01-01 广州市建筑科学研究院有限公司 A kind of foundation pile integrity detection system based on sound wave transmission method
CN109506135A (en) * 2018-11-06 2019-03-22 三川智慧科技股份有限公司 Pipe leakage independent positioning method and device
CN109555977A (en) * 2018-11-23 2019-04-02 水联网技术服务中心(北京)有限公司 The equipment and recognition methods of leak noise measuring
CN109556797A (en) * 2018-11-19 2019-04-02 浙江工业大学 The pipeline leakage detection and location method with convolutional neural networks is decomposed based on spline local mean value
CN110529746A (en) * 2019-09-05 2019-12-03 北京化工大学 Detection method, device and the equipment of pipe leakage

Citations (8)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN101968162A (en) * 2010-09-30 2011-02-09 东北大学 Pipeline leakage positioning system and method based on collaborative detection with negative pressure wave and sound wave
CN203147291U (en) * 2013-03-27 2013-08-21 黄鹏 System capable of monitoring pipeline leakage by means of infrasonic waves, flow balance and negative pressure waves
CN103438359A (en) * 2013-08-06 2013-12-11 毛振刚 Oil pipeline leakage detection and positioning system
CN203477909U (en) * 2013-11-07 2014-03-12 李文杰 Pipeline leakage automatic monitoring positioning device based on low-frequency sound waves and negative-pressure waves
CN105156905A (en) * 2015-07-09 2015-12-16 南京声宏毅霆网络科技有限公司 Leakage monitoring system, method and device for pipeline and server
CN105840987A (en) * 2016-04-25 2016-08-10 北京宏信环科科技发展有限公司 Pipeline leakage weighted positioning method and device based on pressure waves and sound waves
CN106523928A (en) * 2016-11-24 2017-03-22 东北大学 Pipeline leakage detection device and method based on secondary screening of sound wave real-time data
CN107218518A (en) * 2017-04-17 2017-09-29 昆明理工大学 A kind of detection method of detection means for drain line blockage failure

Patent Citations (8)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN101968162A (en) * 2010-09-30 2011-02-09 东北大学 Pipeline leakage positioning system and method based on collaborative detection with negative pressure wave and sound wave
CN203147291U (en) * 2013-03-27 2013-08-21 黄鹏 System capable of monitoring pipeline leakage by means of infrasonic waves, flow balance and negative pressure waves
CN103438359A (en) * 2013-08-06 2013-12-11 毛振刚 Oil pipeline leakage detection and positioning system
CN203477909U (en) * 2013-11-07 2014-03-12 李文杰 Pipeline leakage automatic monitoring positioning device based on low-frequency sound waves and negative-pressure waves
CN105156905A (en) * 2015-07-09 2015-12-16 南京声宏毅霆网络科技有限公司 Leakage monitoring system, method and device for pipeline and server
CN105840987A (en) * 2016-04-25 2016-08-10 北京宏信环科科技发展有限公司 Pipeline leakage weighted positioning method and device based on pressure waves and sound waves
CN106523928A (en) * 2016-11-24 2017-03-22 东北大学 Pipeline leakage detection device and method based on secondary screening of sound wave real-time data
CN107218518A (en) * 2017-04-17 2017-09-29 昆明理工大学 A kind of detection method of detection means for drain line blockage failure

Non-Patent Citations (3)

* Cited by examiner, † Cited by third party
Title
任娇等: "基于负压波和音波的成品油管道泄漏定位综合分析", 《当代化工》 *
李文杰: "基于低频声波和负压波的管道泄漏监测系统", 《油气田地面工程》 *
马大中等: "一种基于多传感器信息融合的故障诊断方法", 《智能系统学报》 *

Cited By (6)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN109115884A (en) * 2018-09-27 2019-01-01 广州市建筑科学研究院有限公司 A kind of foundation pile integrity detection system based on sound wave transmission method
CN109058771A (en) * 2018-10-09 2018-12-21 东北大学 The pipeline method for detecting abnormality of Markov feature is generated and is spaced based on sample
CN109506135A (en) * 2018-11-06 2019-03-22 三川智慧科技股份有限公司 Pipe leakage independent positioning method and device
CN109556797A (en) * 2018-11-19 2019-04-02 浙江工业大学 The pipeline leakage detection and location method with convolutional neural networks is decomposed based on spline local mean value
CN109555977A (en) * 2018-11-23 2019-04-02 水联网技术服务中心(北京)有限公司 The equipment and recognition methods of leak noise measuring
CN110529746A (en) * 2019-09-05 2019-12-03 北京化工大学 Detection method, device and the equipment of pipe leakage

Also Published As

Publication number Publication date
CN108488638B (en) 2019-08-20

Similar Documents

Publication Publication Date Title
CN105700518B (en) A kind of industrial process method for diagnosing faults
US9797799B2 (en) Intelligent adaptive system and method for monitoring leakage of oil pipeline networks based on big data
CN105846780B (en) A kind of photovoltaic module method for diagnosing faults based on decision-tree model
CN106443310B (en) A kind of transformer fault detection method based on SOM neural network
CN102521613B (en) Method for fault diagnosis of automobile electronic system
US7080290B2 (en) Exception analysis for multimissions
CN103996077B (en) A kind of electrical equipment malfunction Forecasting Methodology based on multidimensional time-series
CN104061445B (en) A kind of pipeline leakage detection method based on neutral net
CN100520425C (en) Post-wavelet analysis treating method and device for electric power transient signal
CN101251564B (en) Method for diagnosis failure of power transformer using extension theory and rough set theory
Romano et al. Automated detection of pipe bursts and other events in water distribution systems
CN103776654B (en) The method for diagnosing faults of multi-sensor information fusion
Wang et al. Pipeline leak detection by using time-domain statistical features
US8868985B2 (en) Supervised fault learning using rule-generated samples for machine condition monitoring
CN105373700B (en) A kind of mechanical failure diagnostic method based on comentropy and evidence theory
CN100427961C (en) Virtual instrument technique based gas insulation combined electric appliances online detecting method
CN103144937B (en) System and method for intelligently monitoring belt-type conveyer for coal mine steel wire rope core
CN104573850A (en) Method for evaluating state of thermal power plant equipment
US20040068392A1 (en) Control system and method for detecting plugging in differential pressure cells
CN102132138B (en) A method and apparatus for real time enhancing of the operation of a fluid transport pipeline
CN104132250A (en) Pipeline leakage feature vector extraction method based on improved wavelet packet
CN100552668C (en) Leakage detecting and locating method based on pressure and sound wave information fusion
CN107461611B (en) The leakage detection method and leak detecting device combined is reconstructed based on small echo and EMD
CN104280791B (en) Meteorological data quality control and processing method and system
CN100504846C (en) Electrical energy power quality disturbance automatic identification method and system based on information fusion

Legal Events

Date Code Title Description
PB01 Publication
PB01 Publication
SE01 Entry into force of request for substantive examination
SE01 Entry into force of request for substantive examination
GR01 Patent grant
GR01 Patent grant