CN105527226A - Photoelectric diode array sensor-based ductule gas-liquid two-phase parameter measurement device and method - Google Patents

Abstract

The invention discloses a photoelectric diode array sensor-based ductule gas-liquid two-phase parameter measurement device and method. The device comprises a laser diode, a beam expander, a slit, a glycerinum groove, a transparent ductule, a photoelectric diode array sensor, a data acquisition module and a microcomputer. The photoelectric diode array sensor is adopted to acquire a light intensity distribution signal reflecting flowing information of a two-phase flow; for four typical flow patterns, Fisher is adopted to judge, analyze and build three flow pattern classifiers for realizing flow pattern identification; a support vector machine is adopted to respectively build voidage measurement models of the four typical flow patterns for realizing voidage measurement. The method can be used for realizing measurement of ductule gas-liquid two-phase flow parameters by using the photoelectric diode array sensor, the corresponding device has the advantages of low cost, no contact and the like, meanwhile a flow pattern identification result is introduced into voidage measurement, the influence of flow pattern change on voidage measurement is reduced, and a beneficial reference is provided for ductule gas-liquid two-phase parameter measurement.

Description

Based on small pipeline biphase gas and liquid flow parameter measuring apparatus and the method for photodiode array sensor

Technical field

The present invention relates to polyphasic flow parameter measurement field, particularly relate to a kind of small pipeline biphase gas and liquid flow parameter measuring apparatus based on photodiode array sensor and method.

Background technology

In recent years, along with developing rapidly of micro-Chemical Engineering Technology and new material technology, commercial unit progressively presents the trend of microminiaturization, miniaturization.Microminiaturized, miniaturization reactor, heat exchanger, mixing arrangement etc. all have a wide range of applications in various fields such as biology, medical treatment, pharmacy, chemical industry.Therefore, the small pipeline gas-liquid two-phase flow parameter measurement problem in microminiaturized and miniaturization commercial unit causes concern and the attention of more and more researcher, becomes a focus and the branch of current two-phase flow research field.

But relative to conventional pipeline, small pipeline gas-liquid two-phase flow parameter measurement has certain difficulty.Under miniature scale, due to the reduction of pipeline hydraulic diameter, make give prominence to relative with the impact of viscosity effect of surface tension and the impact of Action of Gravity Field weakens relatively, in pipeline, flow characteristics differs from conventional pipeline, the theoretical model obtained under conventional pipeline and experimental formula will be no longer applicable, and the measuring method full-fledged at conventional pipeline is also no longer applicable to small pipeline.

In the last few years, researcher, based on the measuring method of conventional pipeline, had carried out a large amount of correlative studys for small pipeline gas-liquid two-phase flow parameter measurement, and had made some progress.Achievement in research is to the theoretical research of small pipeline biphase gas and liquid flow, and the research of miniaturization and microminiaturized device systems, application and development play an important role.But due to the shortage of acquisition of information and information processing means, obtain the precision of metrical information and the scope of application has certain one-sidedness, also fail to meet the requirement of engineer applied and scientific research.Just current, existing measurement means is also in developing stage mostly, and the Application comparison in engineering real process has limitation, also needs to excavate the measurement method of parameters that effectively can be applied to small pipeline biphase gas and liquid flow further.

At present, the means for small pipeline Parameter Measurement of Gas-liquid Two-phase are less, mainly contain High Speed Photography, capacitance detecting method, conductance detection etc.By contrast, the optical detecting method based on laser has the advantage such as noncontact, low cost, can realize the effective measurement for diphasic stream parameter.Therefore, use the metering system of laser and photodiode array to carry out the exploration of meteor trail echoes, the correlative study for the identification of small pipeline two phase flow pattern has suitable reference value.

Summary of the invention

The object of the invention is to overcome the deficiencies in the prior art, a kind of stable, reliable small pipeline biphase gas and liquid flow parameter measuring apparatus based on photodiode array sensor and method are provided.

Small pipeline biphase gas and liquid flow parameter measuring apparatus based on photodiode array sensor comprises laser diode, beam expanding lens, slit, glycerine groove, transparent small pipeline, photodiode array sensor, data acquisition module and microcomputer.Place the laser diode of common optical axis, beam expanding lens and slit in glycerine groove side, optical axis is perpendicular to glycerine groove.Photodiode array sensor, data acquisition module and microcomputer is placed successively at glycerine groove opposite side.

Transparent pipeline is placed among glycerine groove, fills glycerine in glycerine groove.Because glycerine refractive index and transparent pipeline tube wall refraction rate are similar to, decrease tube wall to the refraction of sheet laser and reflex.Photodiode array sensor comprises 12 × 6 sensing units, fits tightly with glycerine groove one side.Sensor photosensitive surface, perpendicular to laser diode optical axis, is connected with microcomputer with data acquisition module simultaneously successively.

Small pipeline gas-liquid two-phase flow parameter measurement based on photodiode array sensor comprises the steps:

1) use laser diode, beam expanding lens and slit to produce sheet laser, after sheet laser light glycerine groove, be irradiated to transparent small pipeline, after the reflection and refraction action of gas-liquid two-phase medium in piping, form shoot laser;

2) utilize photodiode array sensor to obtain transmission laser signal, signal is converted into voltage signal after data collecting module collected, input microcomputer;

3) eigenwert calculating and extraction are carried out to the voltage signal got, obtain proper vector;

4) utilize the proper vector extracted and obtain, set up classification of flow patterns device in conjunction with Fisher discriminatory analysis.Based on " two-step approach " classificating thought, realize the meteor trail echoes of small pipeline biphase gas and liquid flow typical case flow pattern;

5) the void fraction determination model under utilizing support vector machine to set up often kind of flow pattern, in conjunction with meteor trail echoes result, selects corresponding void fraction determination model realization void fraction determination.

The present invention compared with prior art has beneficial effect:

1) adopt laser diode as lasing light emitter, adopt photodiode array sensor as detecting element, effectively can simplify experimental provision, reduce system cost.

2) transparent glass tube is placed among glycerine groove, reduces the interference effect of tube wall for laser optical path, improve the sensitivity of measuring method.

3) meteor trail echoes result is introduced in void fraction determination, eliminate the impact that variations in flow patterns produces void fraction determination, improve void fraction determination precision.

Accompanying drawing explanation

Fig. 1 is the small pipeline biphase gas and liquid flow parameter measuring apparatus structural representation based on photodiode array sensor;

Fig. 2 is the glycerine groove that the present invention adopts, the structural representation of transparent pipeline and photodiode array sensor;

Fig. 3 is the meteor trail echoes process flow diagram based on " two-step approach " of the present invention;

Fig. 4 is void fraction determination process flow diagram of the present invention;

Fig. 5 is the measuring gas-liquid two-phase flow porosity result under four kinds of different inner diameters;

In figure: laser diode 1, beam expanding lens 2, slit 3, glycerine groove 4, transparent small pipeline 5, photodiode array sensor 6, data acquisition module 7, microcomputer 8.

Embodiment

The present invention is directed to the present situation that small pipeline Parameter Measurement of Gas-liquid Two-phase means lack, utilize laser diode, photodiode array sensor and sophisticated machine learning algorithm, propose a kind of small pipeline biphase gas and liquid flow parameter measuring apparatus based on photodiode array sensor and method.Related device has that structure is simple, cost is low, non-cpntact measurement, measuring accuracy advantages of higher, for the identification of small pipeline two phase flow pattern provides useful reference.

As shown in Figure 1, small pipeline biphase gas and liquid flow parameter measuring apparatus based on photodiode array sensor comprises laser diode 1, beam expanding lens 2, slit 3, glycerine groove 4, transparent small pipeline 5, photodiode array sensor 6, data acquisition module 7 and microcomputer 8, in glycerine groove 4 side, place the laser diode 1 of common optical axis, beam expanding lens 2 and slit 3 successively, place photodiode array sensor 6, data acquisition module 7, microcomputer 8 outward at glycerine groove 4 opposite side Vertical dimension.Transparent small pipeline 5 is placed in glycerine groove 4, and photodiode array sensor 6 fits tightly with glycerine groove 4 one side, is connected successively with data acquisition module 7, microcomputer 8 simultaneously.Be illustrated in figure 2 the structural representation of transparent small pipeline, glycerine groove and photodiode.

The idiographic flow utilizing this device to realize small pipeline gas-liquid two-phase flow parameter measurement is:

1), after the laser that laser diode is launched passes through beam expanding lens and slit, become a sheet laser, after this sheet laser light glycerine groove, vertical irradiation is on transparent pipeline.

2) inner at transparent pipeline, sheet laser is subject to the effect of gas-liquid two-phase medium, produces refraction, reflection and absorption.Under different flow patterns (namely different biphase gas and liquid flow Entropy density deviation), refraction, reflection and absorption have obvious difference.Therefore transmission laser signal can reflect the flowing information of the inner biphase gas and liquid flow of transparent pipeline.

3) utilize photodiode array sensor to obtain transmission laser signal, change into voltage signal, input microcomputer by data acquisition module.By data processing technique, extract the proper vector x of this voltage signal

x＝[m ₁,…,m ₇₂,e ₁,…,e ₇₂] ^T

Wherein m and e is respectively the mean and variance of the voltage signal that each sensing unit obtains.Make u _kfor the voltage signal that the sensing unit of kth in photodiode array sensor obtains, signal length is L, then its average m _kwith variance e _kcan be expressed as:

$m_k=\frac{1}{L}{\mathrm{\Σ}}_{i=1}^L{u}_k\left(i\right)$

$e_k=\sqrt{\frac{1}{L-1}{\mathrm{\Σ}}_{i=1}^L{\left(u_k\right(i)-m_k)}^2}$

4) the meteor trail echoes process flow diagram based on " two-step approach " thought that the present invention proposes is illustrated in figure 3, this meteor trail echoes is for four kinds of typical flow patterns (bubble flow, slug flows, laminar flow and annular flow), its concrete steps are: the proper vector obtained inputted in flow pattern sorter A, realization group one and the differentiation (organize and comprise bubble flow and slug flow, group two comprises laminar flow and annular flow) organizing two; According to the identification result of sorter A, proper vector is inputted in flow pattern sorter B or C, carry out bubble flow and slug flow, or the identification of laminar flow and annular flow; Final acquisition meteor trail echoes result.

Three the classification of flow patterns devices used in meteor trail echoes all adopt Fisher discriminatory analysis to set up, and realize two classification problems respectively.Sorter A realization group one and the classification organizing two, sorter B realizes the classification of bubble flow and slug flow, and sorter C realizes the classification of laminar flow and annular flow.Its discriminant function all can be expressed as:

y(x)＝sign[ω ^Tx+ω ₀]

Wherein, y is class label (y=-1 or 1), and x is proper vector, and ω is Fisher vector, ω ₀it is discrimination threshold.To the proper vector of discriminant function unknown input flow pattern, by judging that the symbol of class label y can realize classification.The Fisher vector ω of discriminant function can obtain by solving following point

$J\left(\mathrm{\ω}\right)={\max }_{\mathrm{\ω}}\frac{{\mathrm{\ω}}^T{S}_b\mathrm{\ω}}{{\mathrm{\ω}}^T{S}_w\mathrm{\ω}}$

Wherein, S _bfor inter _ class relationship matrix, S _wfor scatter matrix within class.Definition X is the set of training set proper vector, x _ibe i-th proper vector.X _qfor belonging to the set of the proper vector of class q, its number of samples is n _q.Definition for class X _qmean vector:

${\stackrel{\‾}{x}}_q=\frac{1}{n_q}\underset{x_{i\∈X_q}}{\Σ}{x}_i$

Then, Mean Matrix S between class _bwith Mean Matrix S in class _wcan be expressed as:

$S_w=\underset{x_i\∈X_{-1}}{\Σ}(x_i-{\stackrel{\‾}{x}}_{-1}){(x_i-{\stackrel{\‾}{x}}_{-1})}^T+\underset{x_i\∈X_1}{\Σ}(x_i-{\stackrel{\‾}{x}}_1){(x_i-{\stackrel{\‾}{x}}_1)}^T$

$S_b=({\stackrel{\‾}{x}}_{-1}-{\stackrel{\‾}{x}}_1){({\stackrel{\‾}{x}}_{-1}-{\stackrel{\‾}{x}}_1)}^T$

Fisher vector ω can be obtained, simultaneously ω by solving J (ω) maximal value ₀also can obtain:

$\mathrm{\ω}=S_w^{-1}({\stackrel{\‾}{x}}_{-1}-{\stackrel{\‾}{x}}_1)$

${\mathrm{\ω}}_0=-\frac{1}{2}{\mathrm{\ω}}^T({\stackrel{\‾}{x}}_{-1}+{\stackrel{\‾}{x}}_1)$

5) process flow diagram of void fraction determination is illustrated in figure 4.Its step comprises: first utilize proper vector and voidage calibration value component model training set; By model training collection input support vector machine, for four kinds of typical flow patterns, set up four void fraction determination models respectively; According to meteor trail echoes result, the void fraction determination model that real-time selection flow pattern is corresponding, obtains void fraction determination value.

Experiment adopts support vector machine to carry out void fraction determination model modeling, and its pattern function can be expressed as:

$\mathrm{\α}\left(x\right)=\underset{i=1}{\overset{p}{\Σ}}({\mathrm{\β}}_i-{\mathrm{\β}}_i^{*})K(x,x_i)+b$

Wherein, α is voidage, and x is proper vector, for model training collection, p is training set number of samples.K (x, x _i) be kernel function, in this experiment, select radial basis function K (x, x _i)=exp (-| x-x _i| ²/ σ ²) as kernel function, σ is one of them parameter.B is a custom parameter.β _iwith β _i* being Lagrange multiplier, can obtaining by solving following optimization problem

$\min \[\frac{1}{2}\underset{i=1,j=1}{\overset{p}{\Σ}}({\mathrm{\β}}_i-{\mathrm{\β}}_i^{*})({\mathrm{\β}}_j-{\mathrm{\β}}_j^{*})K(x_i,x_j)-\underset{i=1}{\overset{p}{\Σ}}({\mathrm{\β}}_i-{\mathrm{\β}}_i^{*}){\mathrm{\α}}_i+\underset{i=1}{\overset{p}{\Σ}}({\mathrm{\β}}_i+{\mathrm{\β}}_i^{*})\mathrm{\ϵ}\]$

$s.t.\left\{\begin{array}{c}0\≤{\mathrm{\β}}_i,{{\mathrm{\β}}_i}^{*}\≤C,i=1,...,n\\ \underset{i=1}{\overset{p}{\Σ}}({\mathrm{\β}}_i-{\mathrm{\β}}_i^{*})=0,i=1,...,n\end{array}\right.$

Wherein, ε is slack variable, and C is penalty factor.

Small pipeline biphase gas and liquid flow parameter measuring apparatus proposed by the invention and method has been utilized to carry out preliminary experiment.Experiment adopts nitrogen as gas phase, adopts regular tap water as liquid phase, carries out respectively in the transparent pipeline of four kinds of different inner diameters.Internal diameter of the pipeline comprises 4.22mm, 3.04mm, 2.16mm and 1.08mm.

Identification of Gas-Liquid Two-Phase result under four kinds of different inner diameters is as shown in table 1-4.Experimental result shows, the small pipeline Identification of Gas-Liquid Two-Phase method that the present invention proposes can identification four kinds of flow patterns effectively, and meteor trail echoes precision is all higher than 90%.

Table 14.22mm internal diameter pipeline Identification of Gas-Liquid Two-Phase result

Table 23.04mm internal diameter pipeline Identification of Gas-Liquid Two-Phase result

Table 32.16mm internal diameter pipeline Identification of Gas-Liquid Two-Phase result

Table 41.08mm internal diameter pipeline Identification of Gas-Liquid Two-Phase result

Measuring gas-liquid two-phase flow porosity result under four kinds of different inner diameters as shown in Figure 5.A () internal diameter is the void fraction determination result of 4.22mm; B () internal diameter is the void fraction determination result of 3.04mm; C () internal diameter is the void fraction determination result of 2.16mm; D () internal diameter is the void fraction determination result of 1.08mm.

As can be seen from the figure the small pipeline measuring gas-liquid two-phase flow porosity method that the present invention proposes is feasible, and the void fraction determination error under four kinds of calibers is all less than 7%.

Claims (8)

1. the small pipeline biphase gas and liquid flow parameter measuring apparatus based on photodiode array sensor, it is characterized in that comprising laser diode (1), beam expanding lens (2), slit (3), glycerine groove (4), transparent small pipeline (5), photodiode array sensor (6), data acquisition module (7), microcomputer (8), in glycerine groove (4) side, place the laser diode (1) of common optical axis successively, beam expanding lens (2), slit (3), at the photodiode array sensor (6) that glycerine groove (4) opposite side Vertical dimension is placed outward, data acquisition module (7), microcomputer (8), transparent small pipeline (5) is placed in glycerine groove (4), and it is vertical with optical path direction, photodiode array sensor (6) and glycerine groove (4) are fitted and are placed, simultaneously with data acquisition module (7), microcomputer (8) is connected successively.

2. " two-step approach " Identification of Gas-Liquid Two-Phase method of device as claimed in claim 1, is characterized in that the method comprises the steps:

1) laser diode, beam expanding lens and slit produce sheet laser, are irradiated to transparent small pipeline after sheet laser light glycerine groove, after the reflection and refraction action of gas-liquid two-phase medium in piping, form shoot laser; Use photodiode array sensor (6) to obtain laser signal, after data acquisition module (7) is converted into voltage signal, is sent to microcomputer computer (8);

2) mean and variance of voltage signal is extracted, composition characteristic vector;

3) " two-step approach " is adopted to realize meteor trail echoes: the first step is according to two phase flow feature, flow pattern is divided into two groups, wherein organize one and comprise bubble flow and slug flow, group two comprises laminar flow and annular flow, Fisher discriminatory analysis is utilized to set up corresponding sorter A, realization group one and the classification organizing two; Second step utilizes Fisher discriminatory analysis to set up sorter B and C, carries out meteor trail echoes respectively, thus realize the identification of four kinds of typical flow patterns in group one and group two respective inside.

3. meteor trail echoes method according to claim 2, is characterized in that described step 2) proper vector be:

x＝[m ₁,m ₂…,m _n,e ₁,e ₂…,e _n] ^T

Wherein m and e is respectively the mean and variance of the voltage signal that each sensing unit obtains, and n is sensing unit number in photodiode array sensor, makes u _kfor the voltage signal that the sensing unit of kth in photodiode array sensor obtains, signal length is L, then its average m _kwith variance e _kcan be expressed as:

$m_k=\frac{1}{L}{\mathrm{\Σ}}_{i=1}^L{u}_k\left(i\right)$

$e_k=\sqrt{\frac{1}{L-1}{\mathrm{\Σ}}_{i=1}^L{\left(u_k\right(i)-m_k)}^2}.$

4. the meteor trail echoes method according to Claims 2 or 3, is characterized in that described step 3) be specially:

Three the classification of flow patterns devices used in meteor trail echoes all adopt Fisher discriminatory analysis to set up, realize two classification problems respectively, sorter A realization group one and the classification organizing two, sorter B realizes the classification of bubble flow and slug flow, sorter C realizes the classification of laminar flow and annular flow, and its discriminant function all can be expressed as:

y(x)＝sign[ω ^Tx+ω ₀]

Wherein, y is class label, and x is proper vector, and ω is Fisher vector, ω ₀be discrimination threshold, to the proper vector of the discriminant function unknown input flow pattern of each sorter, by judging that the symbol of class label y realizes classification, the Fisher vector ω of discriminant function obtains by solving following point:

$J\left(\mathrm{\ω}\right)={\max }_{\mathrm{\ω}}\frac{{\mathrm{\ω}}^T{S}_b\mathrm{\ω}}{{\mathrm{\ω}}^T{S}_w\mathrm{\ω}}$

Wherein, S _bfor inter _ class relationship matrix, S _wfor scatter matrix within class, definition X is the set of training set proper vector, x _ibe i-th proper vector, X _qfor belonging to the set of the proper vector of class q, its number of samples is n _q, definition for class X _qmean vector:

${\stackrel{\‾}{x}}_q=\frac{1}{n_q}\underset{x_{i\∈X_q}}{\Σ}{x}_i$

Then, Mean Matrix S between class _bwith Mean Matrix S in class _wbe expressed as:

$S_w=\underset{x_i\∈X_{-1}}{\Σ}(x_i-{\stackrel{\‾}{x}}_{-1}){(x_i-{\stackrel{\‾}{x}}_{-1})}^T+\underset{x_i\∈X_1}{\Σ}(x_i-{\stackrel{\‾}{x}}_1){(x_i-{\stackrel{\‾}{x}}_1)}^T$

$S_b=({\stackrel{\‾}{x}}_{-1}-{\stackrel{\‾}{x}}_1){({\stackrel{\‾}{x}}_{-1}-{\stackrel{\‾}{x}}_1)}^T$

Fisher vector ω can be obtained, simultaneously ω by solving J (ω) maximal value ₀also can obtain:

$\mathrm{\ω}=S_w^{-1}({\stackrel{\‾}{x}}_{-1}-{\stackrel{\‾}{x}}_1)$

${\mathrm{\ω}}_0=-\frac{1}{2}{\mathrm{\ω}}^T({\stackrel{\‾}{x}}_{-1}+{\stackrel{\‾}{x}}_1).$

5. a measuring gas-liquid two-phase flow porosity method for device as claimed in claim 1, is characterized in that the method comprises the steps:

1) laser diode, beam expanding lens and slit produce sheet laser, are irradiated to transparent small pipeline after sheet laser light glycerine groove, after the reflection and refraction action of gas-liquid two-phase medium in piping, form shoot laser; Use photodiode array sensor (6) to obtain laser signal, after data acquisition module (7) is converted into voltage signal, is sent to microcomputer computer (8);

2) extract the mean and variance of voltage signal, composition characteristic vector, obtains voidage reference value simultaneously;

3) proper vector and voidage reference value are formed training dataset, void fraction determination model respective under adopting support vector machine to set up often kind of flow pattern;

4) in conjunction with meteor trail echoes result, select void fraction determination model, realize void fraction determination.

6. void fraction determination method according to claim 5, is characterized in that described step 2) proper vector be:

x＝[m ₁,m ₂…,m _n,e ₁,e ₂…,e _n] ^T

Wherein m and e is respectively the mean and variance of the voltage signal that each sensing unit obtains, and n is sensing unit number in photodiode array sensor, makes u _kfor the voltage signal that the sensing unit of kth in photodiode array sensor obtains, signal length is L, then its average m _kwith variance e _kcan be expressed as:

$m_k=\frac{1}{L}{\mathrm{\Σ}}_{i=1}^L{u}_k\left(i\right)$

$e_k=\sqrt{\frac{1}{L-1}{\mathrm{\Σ}}_{i=1}^L{\left(u_k\right(i)-m_k)}^2}.$

7. void fraction determination method according to claim 5, is characterized in that described step 3) be specially: adopt support vector machine to carry out void fraction determination model modeling, its pattern function is expressed and is:

$\mathrm{\α}\left(x\right)=\underset{i=1}{\overset{p}{\Σ}}({\mathrm{\β}}_i-{\mathrm{\β}}_i^{*})K(x,x_i)+b$

Wherein, α is voidage, and x is proper vector, for model training collection, p is training set number of samples, K (x, x _i) be kernel function, b is setup parameter, β _iwith β _i* being Lagrange multiplier, obtaining by solving following optimization problem

$\min \[\frac{1}{2}\underset{i=1,j=1}{\overset{p}{\Σ}}\left({\mathrm{\β}}_i-{\mathrm{\β}}_i^{*}\right)\left({\mathrm{\β}}_j-{\mathrm{\β}}_j^{*}\right)K\left(x_i,x_j\right)-\underset{i=1}{\overset{p}{\Σ}}\left({\mathrm{\β}}_i-{\mathrm{\β}}_i^{*}\right){\mathrm{\α}}_i+\underset{i=1}{\overset{p}{\Σ}}\left({\mathrm{\β}}_i+{\mathrm{\β}}_i^{*}\right)\mathrm{\ϵ}\]$

$s.t.\left\{\begin{array}{c}0\≤{\mathrm{\β}}_i,{{\mathrm{\β}}_i}^{*}\≤C,i=1,...,n\\ \underset{i=1}{\overset{p}{\Σ}}({\mathrm{\β}}_i-{\mathrm{\β}}_i^{*})=0,i=1,...,n\end{array}\right.$

Wherein, ε is slack variable, and C is penalty factor.

8. void fraction determination method according to claim 7, is characterized in that described kernel function K (x, x _i) select radial basis function K (x, x _i)=exp (-| x-x _i| ²/ σ ²) as kernel function, σ is setup parameter.