CN102242872B - Oil transportation pipeline network leakage detection method based on generalized fuzzy hyperbolic model - Google Patents

Abstract

The invention provides an oil transportation pipeline network leakage detection method based on a generalized fuzzy hyperbolic model, belonging to the technical field of pipeline detection. The method comprises the following steps: 1, a negative pressure wave signal is collected and sent to a signal conditioning plate for calculating an initial position generated by the negative pressure wave; 2, the negative pressure wave source can be classified by using the generalized fuzzy hyperbolic model, and whether the generation of the negative pressure wave is caused by leakage, pressure beyond station or working condition adjustment can be judged; 3, if the negative pressure wave comes from the pressure beyond station, the previous section pipeline can be detected for leaks, and the step 1 is returned; if the negative pressure wave comes from leakage, and the step 4 is implemented; if the negative pressure wave comes from the working condition adjustment, and the step 5 is implemented; 4, a leakage alarm is provided. The invention has the following advantages: the generalized fuzzy hyperbolic model is used to distinguish the source of the negative pressure wave; valve opening, pump state, flow, temperature, pressure and density are taken as the input variables for the generalized fuzzy hyperbolic model and the input values can be used to judge if a leakage happens, thus false alarms can be avoided.

Description

Flow circuit leakage detection method based on generalized fuzzy hyperbolic model

Technical field:

The invention belongs to the pipe detection technical field, relate to a kind of flow circuit leakage detection method based on generalized fuzzy hyperbolic model.

Background technique:

Using the pipeline transport fluid is a kind of economy, means of transportation easily, compares with other means of transportation, and it has efficiently, safety, economy, be convenient to the multiple advantages such as control and management, therefore occupies an important position in oil and the conveying of other fluids.Petroleum pipeline is not only long, and the overlay area is also very large.Annual because pipe-line equipment is aging, the geographical conditions variation causes crude oil leakage and artificial drilling hole of oil stolen to bring massive losses to country and enterprise, and causes environmental pollution.The way that early stage pipeline adopts manual segmentation to make an inspection tour mostly although the shortcoming of this method is constantly to make an inspection tour round the clock, because pipeline is long, still can not in time be found to leak; And, carry out the Leak testtion of oil transport pipeline in the mode of artificial enquiry, also expended a large amount of human and material resources and financial resource.

In recent decades, the development of pipeline industry is very fast, and it is particularly important that the research topic of pipe monitoring aspect seems, and the major issue of pipe monitoring is Leak testtion and the location of system.It is early stage that pipeline industry develops, and leakage detection method biases toward hardware approach, as detecting ball method, pipe section pressure test method etc. in magnaflux, the pipe.According to the Leak testtion principle, the method that is used at present Leak testtion can be divided into direct Detection Method and indirect detection method: direct Detection Method namely detects according to the medium that leaks, and the earth's surface vestige that spills such as according to oil and gas leakage time the and the smell that distributes etc. detect; The indirect detection rule is that the variation of the relevant parameter of the pipeline fed sheet of a media that causes according to leakage is inferred.Existing detection and localization method are divided into hardware based method and substantially based on the method two large classes of software in the world.Hardware based method refers to leakage is directly detected, such as direct observational method, leak detection cables method, radioactive-tracer method, light leak detecting etc.Refer to utilize modern control theory, signal to process and computer technology etc. gathers, processes and estimate the impact (such as pressure, flow etc.) that causes because of leakage based on the method for software, leakage is detected and locates.

Summary of the invention

For the deficiencies in the prior art, the invention provides a kind of flow circuit leakage detection method based on generalized fuzzy hyperbolic model.

The hardware system that the present invention relies on is the SCADA system, the SCADA system comprises DSP unit, A/D module, electrical level transferring chip, signal regulating panel and upper-position unit, and wherein the DSP unit comprises dsp chip, power circuit, reset circuit, clock circuit, jtag interface and memory interface;

Power circuit connects dsp chip, is chip power supply; Reset circuit, clock circuit, memory interface, jtag interface are connected with dsp chip respectively; The serial peripheral interface of DSP connects A/D module output terminal; The A/D module input connects the signal regulating panel output terminal; Signal regulating panel input end connection data acquisition module output terminal; The serial communication interface of DSP connects level switch module; Level switch module connects upper-position unit.

A kind of flow circuit leakage detection method based on generalized fuzzy hyperbolic model of the present invention, as follows:

Step 1, the pressure transducer that is installed in the data acquisition unit at segment pipe two ends in the pipe network pick up negative pressure wave signal, pass to signal regulating panel, and the time difference that arrives two ends according to suction wave is calculated the initial position that suction wave produces;

The time difference that arrives the single conduit two ends according to suction wave is calculated the initial position that suction wave produces;

$X=\frac{L-\mathrm{\α}(t_2-t_1)}{2}---\left(1\right)$

The distance of X-Pipeline Leak point head end in the formula, m;

L-oil transport pipeline length, m;

The velocity of propagation of α-suction wave in oil transport pipeline, m/s;

t ₁-suction wave arrives head end time, s;

t ₂-suction wave arrives terminal time, s;

Have the mixed defeated situation of diesel oil, gasoline in the pipeline, the speed of velocity of wave this moment in two kinds of liquid is different, calculates the initial position formula that suction wave produces when detecting in this case suction wave and existing and is converted to:

$X=\frac{(t_1-t_2-\frac{S}{{\mathrm{\α}}_1}){\mathrm{\α}}_2+S+L}{2}---\left(2\right)$

T in the formula ₁-suction wave arrives head end time, s;

t ₂-suction wave arrives terminal time, s;

Gasoline length in the S-oil transport pipeline, m;

α ₁-suction wave is velocity of propagation in gasoline, m;

α ₂-suction wave is velocity of propagation in diesel oil, m;

L-oil transport pipeline length, m;

X-Pipeline Leak point is apart from the distance of head end, m;

Suction wave is to be produced by upper certain point of this section leaks, other pipeline sections leak generation pressure overreach or Operating condition adjustment on step 2, a certain pipeline section, utilize generalized fuzzy hyperbolic model to be classified in the suction wave source, the generation of judging suction wave is because leakage, pressure overreach or Operating condition adjustment;

It is as follows that generalized fuzzy hyperbolic model is carried out assorting process to suction wave source:

1), the model output value is 1 o'clock, representative is leaked and is occured, suction wave derives from leakage;

2), the model output value is 0.5 o'clock, represent suction wave and derive from the pressure overreach;

3), the model output value is 0 o'clock, represent suction wave and derive from Operating condition adjustment.

The form of generalized fuzzy hyperbolic model l bar fuzzy rule is:

$R^l:\mathrm{IF}(x_1-d_{11})\mathrm{is}{F}_{x_{11}}\mathrm{and}(x_1-d_{12})\mathrm{is}{F}_{x_{12}}\mathrm{and}...\mathrm{and}(x_1-d_{{1w}_1})\mathrm{is}{F}_{x_{1{\mathrm{<figure-callout\; id="1"\; label="w"\; filenames="HDA0000070256430000011.png"\; state="\{\{state\}\}">w</figure-callout>}}_1}\mathrm{and}}$

$(x_2-d_{21})\mathrm{is}{F}_{x_{21}}\mathrm{and}(x_2-d_{22})\mathrm{if}{F}_{x_{22}}\mathrm{and}...\mathrm{and}(x_2-d_{{2w}_2})\mathrm{is}{F}_{x_{2w_2}\mathrm{and}...}$

$\mathrm{and}(x_n-d_{n1})\mathrm{is}{F}_{x_{\mathrm{<figure-callout\; id="1"\; label="n"\; filenames="HDA0000070256430000011.png"\; state="\{\{state\}\}">n</figure-callout>}1}}\mathrm{and}(x_n-d_{n2})\mathrm{is}{F}_{x_{n2}}\mathrm{and}...\mathrm{and}(x_n-d_{{\mathrm{nw}}_n})\mathrm{is}{F}_{x_{n{w}_n}}$

$\mathrm{THEN}{y}^l=c_{F_{11}}+c_{F_{12}}+...c_{F_{1{\mathrm{<figure-callout\; id="1"\; label="w"\; filenames="HDA0000070256430000011.png"\; state="\{\{state\}\}">w</figure-callout>}}_1}}+c_{F_{21}}+c_{F_{22}}+...c_{F_{2w_2}}+...$

$+{\mathrm{<figure-callout\; id="1"\; label="c\; F\; n"\; filenames="HDA0000070256430000011.png"\; state="\{\{state\}\}">c</figure-callout>}}_{F_n}+c_{F_{n2}}+...c_{F_{n{w}_n}}---\left(3\right)$

In the formula, w _iFor with x _iThe number of linear transformation, i=1 ..., n; d _IjBe x _iThe linear transformation point, i=1 ..., n; J=1 ... w _i

For with

Corresponding fuzzy subset comprises two Linguistic Values of positive P and negative N, when

During for positive P,

For

When

During for negative N,

For

Be with

Corresponding output constant, among the IF among input variable and the THEN output constant item all be optional, but output item With input variable be one to one, if namely partly comprise at IF

Then should comprise in the THEN part

; On the contrary, if IF part do not comprise

Then do not comprise in the THEN part yet

;

Given one group of broad sense hyperbolic tangential type fuzzy rule base, definition broad sense input variable

x _i＝x _z-d _zj (4)

d _ZjBe x _zThe linear transformation point, j=1 ..., w _zIf W in the formula _zFor with x _zThe number of linear transformation, z=1 wherein ..., n gets fuzzy set corresponding to broad sense input variable

With

Membership function be

With

$\left\{\begin{array}{c}{\mathrm{\&mu;}}_{P_{x_i}}\left(x_i\right)=e^{-\frac{1}{2}{(x_i-k_{x_i})}^2}\\ {\mathrm{\&mu;}}_{N_{x_i}}\left(x_i\right)=e^{-\frac{1}{2}{(x_i+k_{x_i})}^2}\end{array}\right.---\left(5\right)$

In the formula,

Be constant, will

Be abbreviated as

Be abbreviated as

Be abbreviated as k _i, draw such as generalized fuzzy hyperbolic model according to the fuzzy rule base:

$f\left(x\right)=\underset{i=1}{\overset{m}{\mathrm{\&Sigma;}}}\frac{c_{P_i}{e}^{k_i{x}_i}+c_{N_i}{e}^{-k_i{x}_i}}{e^{k_i{x}_i}+e^{-k_i{x}_i}}---\left(6\right)$

In the formula,

Be the fuzzy rules that adds up to,

Be with

Corresponding output constant,

With

Corresponding output constant;

x ₁, Λ, x _nBe the data that collection from pipeline comes, the change amount of the state of pressure, flow, temperature, valve opening, pump etc., f (x) is model output, judges suction wave source in the pipe network according to f (x);

If step 3 suction wave derives from the pressure overreach, then detect on pipeline the last period and whether leak, get back to step 1; If deriving from, suction wave leaks then execution in step 4; If suction wave derives from then execution in step 5 of Operating condition adjustment;

Step 4, provide leakage alarms;

Step 5, end.

Working principle of the present invention: leak the suction wave that produces because the pressure overreach that other pipeline sections leakages produce and station internal pressure-regulating can produce to be similar to, and also the initial position of suction wave is located and the pipeline end points according to formula (1) or (2).So when detecting suction wave in a certain single hop pipeline in the pipe network and the suction wave initial position is positioned pipeline end points (namely in the station), can not judge station internal leakage, stand internal pressure-regulating or suction wave pressure overreach occur, this moment is in conjunction with generalized fuzzy hyperbolic model, detect suc as formula (3), (4), (5), (6), classified in the suction wave source.The alert probability of this effective false alarm reduction of testing process meeting, the warning degree of accuracy that improves system.

Advantage of the present invention: adopt generalized fuzzy hyperbolic model to distinguish the source of suction wave.With valve opening, the state of pump, flow, temperature, pressure, density is as the input variable of generalized fuzzy hyperbolic model, and judges whether to leak by output value, prevents false alarm.

Description of drawings

Fig. 1 is pipe network structure simplified schematic diagram of the present invention;

Fig. 2 is the mixed defeated schematic representation of gasoline of the present invention, diesel oil;

Fig. 3 is the present invention's suction wave waveform that pipe ends detects when leaking;

Fig. 4 is the flow chart that detects the suction wave source in the pipe network of the present invention;

Fig. 5 is hardware circuit diagram of the present invention;

Fig. 6 is AD of the present invention and TMS320F2812 communication interface catenation principle figure;

Embodiment

The present invention is described in detail with Figure of description in conjunction with specific embodiments.

To choose model be AD7656 to the A/D module in the present embodiment; It is TMS320F2812 that dsp chip is chosen model; It is MAX232 that electrical level transferring chip is chosen model.

The support of the inventive method system is the SCADA system, the SCADA system comprises DSP unit, A/D module, electrical level transferring chip, signal regulating panel and upper-position unit, and wherein the DSP unit comprises dsp chip, power circuit, reset circuit, clock circuit, jtag interface, memory interface.The signal of access signal regulating panel is voltage signal, to comprise valve opening, the state of pump, pipeline flow, mean temperature, pressure, the field datas such as density, the process signal regulating panel amplifies, filtering, inputs to AD7656, and the AD chip carries out analog-to-digital conversion, and digital quantity passed to TMS320F2812, DSP utilizes its powerful data-handling capacity, and data are compressed and a digital filtering, and data were passed to PC after the SCI serial communication interface that utilizes at last DSP will be processed by electrical level transferring chip MAX232.Wherein AD communicates by letter with DSP and adopts the spi bus agreement.This system's lower-position unit circuit block diagram as shown in Figure 5.

The spi bus interface scheme: the SPI interface is comprised of 4 signaling lines: serial data input (the SPISOMIA pin of DSP), serial data output (the SPISIMOA pin of DSP), SCK (the SPICLKA pin of DSP), SS (the SPISTEA pin of DSP).Main equipment flows by shift clock being provided and controlling data from enable signal.An optional control interface signal from enable signal, if do not have special in enable signal, can by whether existing effective shift clock to decide, must always keep enabled state from equipment this moment, and only have one from equipment, block diagram is as shown in Figure 6.

The present invention is a kind of flow circuit leakage detection method based on generalized fuzzy hyperbolic model, can effectively prevent the generation of false alarm.Its step is as follows:

Step 1, the pressure transducer that is installed in the data acquisition unit at segment pipe two ends in the pipe network pick up negative pressure wave signal, pass to signal regulating panel, and upper-position unit calculates the initial position that suction wave produces according to the time difference that suction wave arrives two ends;

When pipeline occurs to leak, will produce transient pressure in leak and fall, form a suction wave, this ripple is propagated to the pipeline upper and lower end with the velocity of wave in the pipeline, and is received by the pressure transducer that is arranged on two sections of pipeline sections.Propagate into the time difference of upstream and downstream and the position that the overpressure velocity of wave propagation just can calculate leakage point according to suction wave.Line construction complicated in the pipe network is different from single conduit, and its Leak testtion and location be many than the one-pipe complexity also, however, still can adopt negative pressure wave method that the pipeline network leak point is monitored and located.In pipe network, (convergence point is two sections tie point in the pipeline to the pressure wave that leak to produce by the pipeline convergence point, such as B, D, E point) time can pass to next pipeline section (or lower several pipeline sections), there is pressure overreach phenomenon, therefore one section generation is leaked and just may be had multistage can detect the existence of suction wave, and the pipeline section that so occurs to leak just might be to the warning message that makes mistake.

Take Fig. 1 as example, suppose that a point leaks, record the time t that suction wave arrives B, D two stations _B, t _D, just can leak location Calculation, suc as formula (7)

$X=\frac{L-\mathrm{\&alpha;}(t_D-t_B)}{2}---\left(7\right)$

The X-Pipeline Leak is put to the distance of B end, m in the formula;

L-oil transport pipeline length, m;

The velocity of propagation of α-suction wave in oil transport pipeline, m/s;

t _B-suction wave arrives B end time, s;

t _D-suction wave arrives D end time, s.

Velocity of wave α is take constant as prerequisite in the formula (7).Velocity of wave is relevant with specific heat, density, pressure and the pipe material of medium, and the oil density that normal temperature is carried changes little along pipeline, and velocity of wave can be regarded as constant.Because piping feeding distance is long, temperature variation is large, and the transmission speed of pressure wave might not be constant.Consider the factors such as density, elasticity and pipe material character of liquid, the suction wave velocity of propagation is revised.The negative pressure velocity of wave propagation can be calculated by following formula:

$\mathrm{\&alpha;}=\sqrt{\frac{k/\mathrm{\&rho;}}{1+\frac{\mathrm{kD}}{\mathrm{Ee}}{c}_1}}---\left(8\right)$

K-liquid volume elasticity coefficient in the formula, m/s;

ρ-fluid density, kg/m ³

E-tubing Young's modulus, Pa;

The D-caliber, m;

The e-pipe thickness, m;

c ₁-pipeline constraint factor.

Have gasoline, the mixed defeated situation of diesel oil in the reality, namely the first half section is that the diesel oil second half section is gasoline in the pipeline, and is perhaps opposite.Because two kinds of oil densities are different, velocity of wave α just can not be considered as constant more so.Gasoline, diesel oil mix defeated simplified schematic diagram as shown in Figure 2 in the single hop pipeline, and at this moment, the leak point positioning formula is converted to:

$X=\frac{(t_1-t_2-\frac{S}{{\mathrm{\&alpha;}}_1}){\mathrm{\&alpha;}}_2+S+L}{2}---\left(9\right)$

T in the formula ₁-suction wave arrives B station time, s;

t ₂-suction wave arrives D station time, s;

Gasoline length in the S-oil transport pipeline, m;

α ₁-suction wave is velocity of propagation in gasoline, m;

α ₂-suction wave is velocity of propagation in diesel oil, m;

L-oil transport pipeline length, m;

X-Pipeline Leak point is apart from the distance of head end B, m;

Figure 3 shows that two detected suction wave waveforms in station when leaking, can try to achieve the leakage point position by the corresponding time difference of two suction wave trailing edges and by formula (7), (8) or (9).

Step 2, utilize generalized fuzzy hyperbolic model that suction wave is classified.

Leak at certain some a place on BD section among Fig. 1, and suction wave propagates into the position that time difference that B, D order and overpressure velocity of wave propagation just can calculate leakage point.Because whole pipeline is communicated with, and has pressure overreach phenomenon, suction wave can pass in other pipeline sections (such as DE, BC, EF).Propagate into time difference and the formula (7), (8) or (9) of two stations between the DE according to suction wave and can locate " leakage point " at the D point; In like manner propagate into two time differences of station between the BC according to suction wave and can locate " leakage point " at the B point, but this moment be not because D point and the caused suction wave of B point leakage, so just false alarm might appear.

Operating mode is very complicated in the actual pipe network, the reason that causes the pressure parameter fluctuation is diversified, the change of the change of the adjustment of the adjustment of valve opening, pump state, the change of flow, temperature, pipeline pressurization, density etc. all can make suction wave change, and suction wave is propagated in pipeline and is also had loss, and the pressure surge of transferring pump, transferring valve, termination of pumping to cause is similar with the pressure surge that leakage causes, therefore need to make a distinction leaking with complicated, normal Operating condition adjustment, prevent the generation of reporting by mistake, failing to report and leakage point is accurately located.

Adopt generalized fuzzy hyperbolic model to differentiate to above two kinds of situations, distinguish with leakage.The form of generalized fuzzy hyperbolic model l bar fuzzy rule is:

$R^l:\mathrm{IF}(x_1-d_{11})\mathrm{is}{F}_{x_{11}}\mathrm{and}(x_1-d_{12})\mathrm{is}{F}_{x_{12}}\mathrm{and}...\mathrm{and}(x_1-d_{{1w}_1})\mathrm{is}{F}_{x_{1{\mathrm{<figure-callout\; id="1"\; label="w"\; filenames="HDA0000070256430000011.png"\; state="\{\{state\}\}">w</figure-callout>}}_1}\mathrm{and}}$

$(x_2-d_{21})\mathrm{is}{F}_{x_{21}}\mathrm{and}(x_2-d_{22})\mathrm{if}{F}_{x_{22}}\mathrm{and}...\mathrm{and}(x_2-d_{{2w}_2})\mathrm{is}{F}_{x_{2w_2}\mathrm{and}...}$

$\mathrm{and}(x_n-d_{n1})\mathrm{is}{\mathrm{<figure-callout\; id="1"\; label="F\; x\; n"\; filenames="HDA0000070256430000011.png"\; state="\{\{state\}\}">F</figure-callout>}}_{x_n}\mathrm{and}(x_n-d_{n2})\mathrm{is}{F}_{x_{n2}}\mathrm{and}...\mathrm{and}(x_n-d_{{\mathrm{nw}}_n})\mathrm{is}{F}_{x_{n{w}_n}}$

$\mathrm{THEN}{y}^l=c_{F_{11}}+c_{F_{12}}+...c_{F_{1{\mathrm{<figure-callout\; id="1"\; label="w"\; filenames="HDA0000070256430000011.png"\; state="\{\{state\}\}">w</figure-callout>}}_1}}+c_{F_{21}}+c_{F_{22}}+...c_{F_{2w_2}}+...$

$+{\mathrm{<figure-callout\; id="1"\; label="c\; F\; n"\; filenames="HDA0000070256430000011.png"\; state="\{\{state\}\}">c</figure-callout>}}_{F_n}+c_{F_{n2}}+...c_{F_{n{w}_n}}---\left(10\right)$

In the formula, w _i(i=1 ..., be with x n) _iThe number of linear transformation; d _Ij(i=1 ..., n; J=1 ... w _i) be x _iThe linear transformation point;

For with

Corresponding fuzzy subset is just comprising (P) and negative (N) two Linguistic Values, when

During for just (P),

For

When

During for negative (N),

For Be with

Corresponding output constant.Among the IF among input variable and the THEN output constant item all be optional, but output item

With input variable be one to one, if namely partly comprise at IF

Then should comprise in the THEN part

; On the contrary, if IF part do not comprise

Then do not comprise in the THEN part yet

.

If given one group of broad sense hyperbolic tangential type fuzzy rule base at first defines the broad sense input variable

x _i＝x _z-d _zj (11)

d _Zj(j=1 ..., w _z) be x _zThe linear transformation point; If

W in the formula _z(z=1 ..., be with x n) _zThe number of linear transformation.Get fuzzy set corresponding to broad sense input variable

With

Membership function be With

$\left\{\begin{array}{c}{\mathrm{\&mu;}}_{P_{x_i}}\left(x_i\right)=e^{-\frac{1}{2}{(x_i-k_{x_i})}^2}\\ {\mathrm{\&mu;}}_{N_{x_i}}\left(x_i\right)=e^{-\frac{1}{2}{(x_i+k_{x_i})}^2}\end{array}\right.---\left(12\right)$

In the formula,

Be constant.Will

Be abbreviated as

Be abbreviated as

Be abbreviated as k _i, draw such as generalized fuzzy hyperbolic model according to the fuzzy rule base:

$f\left(x\right)=\underset{i=1}{\overset{m}{\mathrm{\&Sigma;}}}\frac{c_{P_i}{e}^{k_i{x}_i}+c_{N_i}{e}^{-k_i{x}_i}}{e^{k_i{x}_i}+e^{-k_i{x}_i}}---\left(13\right)$

In the formula,

Be the fuzzy rules that adds up to,

Be with

Corresponding output constant,

With

Corresponding output constant.

Take pipe network structure shown in Figure 1 as example, as for the DE section, establish x ₁Be D station valve opening change amount, x ₂Be E station valve opening change amount, x ₃Be the state change amount of D station pump, x ₄Be the state change amount of E station pump, x ₅Be DE section mean flowrate change amount, x ₆Be DE section mean temperature change amount; x ₇Be DE section hydrodynamic pressure change amount, x ₈Be DE section average fluid density change amount.Output y is DE segment pipe operation conditions, and y represented to have to leak to occur in 1 o'clock, represented not leak generation during y=0.Every section has 8 input variables, an output variable.Can obtain following broad sense hyperbolic tangential type fuzzy rule base according to formula (10):

$R^1:\mathrm{IF}{x}_1-d_{11}\mathrm{is}{P}_{x_{11}}\mathrm{and}{x}_1-d_{12}\mathrm{is}{P}_{x_{12}}\mathrm{and}$

$x_2-d_{21}\mathrm{is}{P}_{x_{21}}\mathrm{and}{x}_2-d_{22}\mathrm{is}{P}_{x_{22}}\mathrm{and}$

$...$

$x_8-d_{81}\mathrm{is}{P}_{x_{81}}\mathrm{and}{x}_2-d_{82}\mathrm{is}{P}_{x_{82}}$

$\mathrm{<figure-callout\; id="1"\; label="THEN\; y"\; filenames="HDA0000070256430000011.png"\; state="\{\{state\}\}">THEN</figure-callout>}{y}^{}{}^1=c_{P_{11}}+c_{P_{12}}+c_{P_{21}}+c_{P_{22}}......+c_{P_{81}}+c_{P_{82}}$

$R^2:\mathrm{IF}{x}_1-d_{11}\mathrm{is}{N}_{x_{11}}\mathrm{and}{x}_1-d_{12}\mathrm{is}{P}_{x_{12}}\mathrm{and}$

$x_2-d_{21}\mathrm{is}{P}_{x_{21}}\mathrm{and}{x}_2-d_{22}\mathrm{is}{P}_{x_{22}}\mathrm{and}$

$...$

$x_8-d_{81}\mathrm{is}{P}_{x_{81}}\mathrm{and}{x}_2-d_{82}\mathrm{is}{P}_{x_{82}}$

$\mathrm{THEN}{y}^2=c_{N_{11}}+c_{P_{12}}+c_{P_{21}}+c_{P_{22}}......+c_{P_{81}}+c_{P_{82}}$

$...$

$...$

$R^{2^{16}}:\mathrm{IF}{x}_1-d_{11}\mathrm{is}{N}_{x_{11}}\mathrm{and}{x}_1-d_{12}\mathrm{is}{N}_{x_{12}}\mathrm{and}$

$x_2-d_{21}\mathrm{is}{N}_{x_{21}}\mathrm{and}{x}_2-d_{22}\mathrm{is}{N}_{x_{22}}\mathrm{and}$

$...$

$x_8-d_{81}\mathrm{is}{N}_{x_{81}}\mathrm{and}{x}_2-d_{82}\mathrm{is}{N}_{x_{82}}$

$\mathrm{THEN}{y}^{2^{16}}=c_{N_{11}}+c_{N_{12}}+c_{N_{21}}+c_{N_{22}}......+c_{N_{81}}+c_{N_{82}}---\left(14\right)$

Get

With

(m=1,2......, 8; N=1,2) membership function is

$\left\{\begin{array}{c}{\mathrm{\&mu;}}_{P_{\mathrm{m\; n}}}\left(x_{\mathrm{mn}}\right)=e^{-\frac{1}{2}{(x_m-k_{\mathrm{mn}})}^2}\\ {\mathrm{\&mu;}}_{P_{\mathrm{m\; n}}}\left(x_{\mathrm{mn}}\right)=e^{-\frac{1}{2}{(x_m+k_{\mathrm{mn}})}^2}\end{array}\right.---\left(15\right)$

D in the formula (14) _Ij(i=1,2......, 8; J=1,2) get suitable value,

With

(i=1,2 ... 8; J=1,2) value rule of thumb,

Constant k in the formula (15) _MnGet appropriate value, can obtain output value y suc as formula shown in (16).

Size according to output value y is judged leakage, pressure overreach or Operating condition adjustment.

Step 3, for the suction wave that is positioned at the pipeline end points, if generalized fuzzy hyperbolic model output is 1, then expression is leaked and is occurred in end points; If model is output as 0, then represent Operating condition adjustment; If model is output as 0.5, illustrate that then suction wave derives from the pressure overreach, then detect on pipeline the last period whether to leak, get back to step 1; If pump and valve etc. carried out adjustment, then suction wave derives from Operating condition adjustment, forwards step 5 to.

Step 4, provide leakage alarms.

Step 5, end.

The implementation of the inventive method is as follows:

When a point leaked among Fig. 1, A, C, E, F, G point all may detect the suction wave trailing edge, owing to have loss in the suction wave propagation process, therefore far away apart from the leakage point distance, the suction wave loss is just larger.Suppose that A point, C point, E point, F point, G point have all detected the existence of trailing edge.Detect trailing edge as example take the G point now, detected the time difference t of suction wave by E, G two ends, and follow according to formula (7) (8) and leak point positioning can be arrived as near the E point, this moment is in conjunction with generalized fuzzy hyperbolic model, each state of EG pipeline section as input variable, is judged that according to the model output variable it is that this section leakage, Operating condition adjustment or other pipeline sections leak the pressure wave overreach that causes on earth that suction wave produces.Be this section leakage when detecting, then provide warning message; Be Operating condition adjustment when detecting, then be failure to actuate; If detect as the front pipeline section leaks the pressure overreach that causes, whether the pipeline section that rejudges E point front that then uses the same method leaks, until find leakage point.This example should be the pressure overreach according to the output judged result.Flow chart as shown in Figure 4.

Claims (1)

1. the flow circuit leakage detection method based on generalized fuzzy hyperbolic model is characterized in that, as follows:

Step 1, the pressure transducer that is installed in the data acquisition unit at segment pipe two ends in the pipe network pick up negative pressure wave signal, pass to signal regulating panel, and the time difference that arrives two ends according to suction wave is calculated the initial position that suction wave produces;

The time difference that arrives the single conduit two ends according to suction wave is calculated the initial position that suction wave produces; Formula is as follows,

$X=\frac{L-\mathrm{\&alpha;}(t_2-t_1)}{2}---\left(1\right)$

The distance of X-Pipeline Leak point head end in the formula, m;

L-oil transport pipeline length, m;

The velocity of propagation of α-suction wave in oil transport pipeline, m/s;

t ₁-suction wave arrives head end time, s;

t ₂-suction wave arrives terminal time, s;

Have the mixed defeated situation of diesel oil, gasoline in the pipeline, the speed of velocity of wave this moment in two kinds of liquid is different, calculates the initial position formula that suction wave produces when detecting in this case suction wave and existing and is converted to:

$X=\frac{(t_1-t_2-\frac{S}{{\mathrm{\&alpha;}}_1}){\mathrm{\&alpha;}}_2+S+L}{2}---\left(2\right)$

T in the formula ₁-suction wave arrives head end time, s;

t ₂-suction wave arrives terminal time, s;

Gasoline length in the S-oil transport pipeline, m;

α ₁-suction wave is velocity of propagation in gasoline, m;

α ₂-suction wave is velocity of propagation in diesel oil, m;

L-oil transport pipeline length, m;

X-Pipeline Leak point is apart from the distance of head end, m;

Suction wave is to be produced by upper certain point of this section leaks, other pipeline sections leak generation pressure overreach or Operating condition adjustment on step 2, a certain pipeline section, utilize generalized fuzzy hyperbolic model to be classified in the suction wave source, the generation of judging suction wave is because leakage, pressure overreach or Operating condition adjustment;

It is described that to utilize generalized fuzzy hyperbolic model that sorting technique is carried out in suction wave source as follows:

1), the model output value is 1 o'clock, representative is leaked and is occured, suction wave derives from leakage;

2), the model output value is 0.5 o'clock, represent suction wave and derive from the pressure overreach;

3), the model output value is 0 o'clock, represent suction wave and derive from Operating condition adjustment;

The form of generalized fuzzy hyperbolic model l bar fuzzy rule is:

$R^l:\mathrm{IF}(x_1-d_{11})\mathrm{is}{F}_{x_{11}}\mathrm{and}(x_1-d_{12})\mathrm{is}{F}_{x_{12}}\mathrm{and}\&CenterDot;\&CenterDot;\&CenterDot;\mathrm{and}(x_1-d_{1w_1})\mathrm{is}{F}_{x_{1w_1}}\mathrm{and}$

$(x_2-d_{21})\mathrm{is}{F}_{x_{21}}\mathrm{and}(x_2-d_{22})\mathrm{is}{F}_{x_{22}}\mathrm{and}\&CenterDot;\&CenterDot;\&CenterDot;\mathrm{and}(x_2-d_{2w_2})\mathrm{is}{F}_{x_{2w_2}}\mathrm{and}\&CenterDot;\&CenterDot;\&CenterDot;$

$\mathrm{and}(x_n-d_{n1})\mathrm{is}{F}_{x_{n1}}\mathrm{and}(x_n-d_{n2})\mathrm{is}{F}_{x_{n2}}\mathrm{and}\&CenterDot;\&CenterDot;\&CenterDot;\mathrm{and}(x_n-d_{n{w}_n})\mathrm{is}{F}_{x_{n{w}_n}}$

$\mathrm{THEN}{y}^l=c_{F_{11}}+c_{F_{12}}+\&CenterDot;\&CenterDot;\&CenterDot;c_{F_{1w_1}}+c_{F_{21}}+c_{F_{22}}+\&CenterDot;\&CenterDot;\&CenterDot;c_{F_{2w_2}}+\&CenterDot;\&CenterDot;\&CenterDot;$

$+c_{F_{n1}}+c_{F_{n2}}+\&CenterDot;\&CenterDot;\&CenterDot;c_{F_{n{w}_n}}---\left(3\right)$

In the formula, w _iFor with x _iThe number of linear transformation, i=1 ..., n; d _IjBe x _iThe linear transformation point, i=1 ..., n; J=1 ... w _i

For with Corresponding fuzzy subset comprises two Linguistic Values of positive P and negative N, when During for positive P,

For

When

During for negative N,

For

Be with

Corresponding output constant, among the IF among input variable and the THEN output constant item all be optional, but output item

With input variable be one to one, if namely partly comprise at IF

Then should comprise in the THEN part

; On the contrary, if IF part do not comprise Then do not comprise in the THEN part yet

;

Given one group of broad sense hyperbolic tangential type fuzzy rule base, definition broad sense input variable

x _i＝x _z-d _zj (4)

d _ZjBe x _zThe linear transformation point, j=1 ..., w _zIf W in the formula _zFor with x _zThe number of linear transformation, z=1 wherein ..., n gets fuzzy set corresponding to broad sense input variable

With

Membership function be

With

$\left\{\begin{array}{c}{\mathrm{\&mu;}}_{P_{x_i}}\left(x_i\right)=e^{-\frac{1}{2}{(x_i-k_{x_i})}^2}\\ {\mathrm{\&mu;}}_{N_{x_i}}\left(x_i\right)=e^{-\frac{1}{2}{(x_i+k_{x_i})}^2}\end{array}\right.---\left(5\right)$

In the formula,

Be constant, will

Be abbreviated as

Be abbreviated as

Be abbreviated as k _i, draw such as generalized fuzzy hyperbolic model according to the fuzzy rule base:

$f\left(x\right)=\underset{i=1}{\overset{m}{\mathrm{\&Sigma;}}}\frac{c_{P_i}{e}^{k_i{x}_i}+c_{N_i}{e}^{-k_i{x}_i}}{e^{k_i{x}_i}+e^{-k_i{x}_i}}---\left(6\right)$

In the formula,

Be the fuzzy rules that adds up to,

Be with

Corresponding output constant,

With

Corresponding output constant;

x ₁..., x _nBe the data that collection from pipeline comes, the change amount of the state of pressure, flow, temperature, valve opening, pump etc., f (x) is model output, judges suction wave source in the pipe network according to f (x);

If step 3 suction wave derives from the pressure overreach, then detect on pipeline the last period and whether leak, get back to step 1; If deriving from, suction wave leaks then execution in step 4; If suction wave derives from then execution in step 5 of Operating condition adjustment;

Step 4, provide leakage alarms;

Step 5, end.

Images