CN100403987C - X-ray CT multi-phase current inspection based on hereditary algorithm - Google Patents
X-ray CT multi-phase current inspection based on hereditary algorithm Download PDFInfo
- Publication number
- CN100403987C CN100403987C CNB2005100862502A CN200510086250A CN100403987C CN 100403987 C CN100403987 C CN 100403987C CN B2005100862502 A CNB2005100862502 A CN B2005100862502A CN 200510086250 A CN200510086250 A CN 200510086250A CN 100403987 C CN100403987 C CN 100403987C
- Authority
- CN
- China
- Prior art keywords
- ray
- genetic algorithm
- projection
- detected
- multiphase flow
- Prior art date
- Legal status (The legal status 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 status listed.)
- Expired - Fee Related
Links
- 238000004422 calculation algorithm Methods 0.000 title claims abstract description 58
- 238000007689 inspection Methods 0.000 title 1
- 230000002068 genetic effect Effects 0.000 claims abstract description 46
- 238000000034 method Methods 0.000 claims abstract description 45
- 238000009826 distribution Methods 0.000 claims abstract description 41
- 230000008569 process Effects 0.000 claims abstract description 21
- 238000001514 detection method Methods 0.000 claims description 15
- 230000035772 mutation Effects 0.000 claims description 8
- 230000004075 alteration Effects 0.000 claims description 6
- 230000000153 supplemental effect Effects 0.000 claims description 5
- 238000013461 design Methods 0.000 claims description 4
- 238000004364 calculation method Methods 0.000 claims description 3
- 238000000205 computational method Methods 0.000 claims description 3
- 238000010353 genetic engineering Methods 0.000 claims description 3
- 238000012545 processing Methods 0.000 claims description 3
- 238000012163 sequencing technique Methods 0.000 claims description 3
- 239000012141 concentrate Substances 0.000 claims description 2
- 230000006872 improvement Effects 0.000 claims description 2
- 239000011159 matrix material Substances 0.000 claims description 2
- 238000005259 measurement Methods 0.000 abstract description 10
- 238000003384 imaging method Methods 0.000 abstract description 7
- 238000005457 optimization Methods 0.000 abstract description 5
- 238000010276 construction Methods 0.000 abstract description 2
- 239000003814 drug Substances 0.000 description 14
- 238000005516 engineering process Methods 0.000 description 12
- 238000004088 simulation Methods 0.000 description 10
- 230000000694 effects Effects 0.000 description 8
- 238000012612 static experiment Methods 0.000 description 6
- 210000004027 cell Anatomy 0.000 description 5
- 238000011160 research Methods 0.000 description 5
- 230000003068 static effect Effects 0.000 description 5
- XLYOFNOQVPJJNP-UHFFFAOYSA-N water Substances O XLYOFNOQVPJJNP-UHFFFAOYSA-N 0.000 description 5
- 230000008901 benefit Effects 0.000 description 4
- 238000002474 experimental method Methods 0.000 description 4
- 239000007787 solid Substances 0.000 description 4
- 238000004519 manufacturing process Methods 0.000 description 3
- 230000001052 transient effect Effects 0.000 description 3
- 230000005587 bubbling Effects 0.000 description 2
- 238000002591 computed tomography Methods 0.000 description 2
- 239000007788 liquid Substances 0.000 description 2
- 239000000203 mixture Substances 0.000 description 2
- 238000005192 partition Methods 0.000 description 2
- 239000002002 slurry Substances 0.000 description 2
- 239000000126 substance Substances 0.000 description 2
- 230000002123 temporal effect Effects 0.000 description 2
- 230000004304 visual acuity Effects 0.000 description 2
- 210000004556 brain Anatomy 0.000 description 1
- 210000000481 breast Anatomy 0.000 description 1
- 239000003990 capacitor Substances 0.000 description 1
- 238000004587 chromatography analysis Methods 0.000 description 1
- 230000007547 defect Effects 0.000 description 1
- 238000011161 development Methods 0.000 description 1
- 238000003745 diagnosis Methods 0.000 description 1
- 230000005670 electromagnetic radiation Effects 0.000 description 1
- 238000010894 electron beam technology Methods 0.000 description 1
- 239000012530 fluid Substances 0.000 description 1
- 230000003993 interaction Effects 0.000 description 1
- 230000007246 mechanism Effects 0.000 description 1
- 238000005272 metallurgy Methods 0.000 description 1
- 210000002487 multivesicular body Anatomy 0.000 description 1
- 238000010845 search algorithm Methods 0.000 description 1
- 230000009897 systematic effect Effects 0.000 description 1
- 238000012360 testing method Methods 0.000 description 1
- 238000003325 tomography Methods 0.000 description 1
- 230000007704 transition Effects 0.000 description 1
- 238000002604 ultrasonography Methods 0.000 description 1
- 238000012795 verification Methods 0.000 description 1
Images
Landscapes
- Apparatus For Radiation Diagnosis (AREA)
Abstract
The present invention relates to an X-ray CT multiphase flow detecting method based on a genetic algorithm, which constructs the concrete genetic algorithm by combining the phase distribution or the medium distribution law of a multiphase flow system to be detected so as to realize the flow field reconstruction of X-ray projection data of a limited angle. The method comprises the concrete steps: 1) the system to be detected is simplified; 2) geometric subdivision is carried out on a region to be detected; 3) the X-ray projection data is collected; 4) an equation set of X-ray projection is discretized; 5) the concrete genetic algorithm is constructed and is applied to the process of image reconstruction; 6) a numerical solution is converted into a phase/medium distribution map of the region to be detected. On one hand the present invention can well complete the process of multiphase flow fast imaging, and on the other hand, the present invention combines the concrete genetic algorithm for the physical characteristic construction of the system, realizes the solving process of constrained optimization, has strong capability of resisting measurement noise interference, effectively controls the ill-posed nature of the process of the image reconstruction, and obviously weakens the influence on a reconstruction result by a measurement error.
Description
Technical field
The present invention relates to a kind of X ray CT multiphase flow detection method, belong to the multiphase flow measurement technical field based on genetic algorithm.
Background technology
The medicine CT technology has been applied to section and the three-dimensional imaging research and the state of an illness diagnosis of human body maturely.So far, medicine CT entered with the electron beam ct machine be representative the 5th generation equipment.Five generation the medicine CT machine single-throw shadow and tomoscan time as shown in the table.
Five generation the medicine CT machine single-throw shadow and tomoscan time
| Sweep time | The first generation | The second filial generation | The third generation | The 4th generation | The 5th generation |
| Single-throw shadow (s) | 1 | 1 | ～0.001 | ～0.25 | ＜0.0004 |
| Tomoscan (s) | ～200 | ～18 | 5 | 1 | 0.01 |
Although medicine CT has obtained important progress aspect fast detecting, medicine CT is primarily aimed at human body, and equipment is huge, and cost is extremely expensive, thereby requires a large amount of scanning angles to guarantee accurate image reconstruction to reach complete data for projection.These factors have all limited medicine CT and have been directly used in industrial process measurement and breadboard research application.On the other hand, except the 5th generation medicine CT have tomoscan speed faster, equal density field information when other CT technology generally is used to obtain.
The existing in theory ripe relatively and complete image reconstructing method of X ray CT, the general exigent data for projection completeness of its image reconstruction algorithm, to guarantee the accurate reproduction to the object to be detected internal information, this point has also limited the temporal resolution of existing medicine CT simultaneously.Desire realizes the rapid CT measurement, then needs expensive equipment, and abundant x-ray source and detector is provided.
Along with the medicine CT principle is applied to industrial engineering more and more, the restriction of the temporal resolution of medicine CT is also little by little outstanding.
In fields such as oil, chemical industry, metallurgy, power, the energy, medicine, food, multiphase flow process ubiquity is as solution-air, system such as gas-solid, liquid-solid, gas-liquid-solid.Owing to have interfacial effect and interaction phase, the multiphase flow process is extremely complicated.Set up process model, and the prediction of the process of carrying out, design and control, at first will solve the measuring technique problem of multiphase flow process.Developing rapidly of scientific research and production technology requires exploitation and upgrades the multiphase flow checkout equipment with high time resolution, high spatial resolution and suitable price, being applied to that transient temperature field distribution, velocity field in two-dimensional section in the heterogeneous flow field or the three dimensions distributes and the obtaining of information such as phase composition distribution, even reach the effect (also being that flow field information transient state " is freezed " and quick reconfiguration) of the online detection of original position.
Process tomographic imaging (PT) technology is that the medicine CT technology is applied to industrial process and produces., mainly should be used for measuring each phase phase content distribution and flow pattern judgement etc. at present owing to but the characteristics such as two-dimensional/three-dimensional distribution of the noiseless acquisition process parameter of its stream field receive much concern.The ultimate principle of PT technology is: arrange plurality of sensors all around according to certain rules in the broken flow field of detecting, sensor array obtains the flow field information of detected object with noncontact or non-intruding mode, utilizes image reconstruction algorithm to calculate the section fluid then and distributes and show.There is the multiple form of obtaining information in the PT system, presses the detection principle of pick off, can be divided into electromagnetic radiation formula, acoustics formula and electric measuring type three major types, serves as typical case's representative with X ray CT (X-CT), ultrasound computed tomography (UCT) and capacitor C T (ECT) respectively.Present PT technology all also in the middle of evolution, has limitation separately.
The major advantage of X-CT is comparatively simple in the more complete situation hypograph reconstruction computing of data for projection, imaging precision higher (be decided by radiographic source and projection number, reach as high as 100 nanometers); Major defect is to adopt third and fourth CT scan mode if consider economic factors in generation, and every tomography radiographic source rotary course is consuming time to reach 1 second even higher magnitude, has had a strong impact on the effect of obtaining of transient state information.The characteristics of the high spatial resolution of X-CT are significant to the meticulous measurement in flow field that realizes multiphase flow, guarantee that its excellent spatial resolution improves the emphasis that its time resolution is present X-CT development again simultaneously.
Simultaneously, CT reconstruct problem can be summed up as the problem of finding the solution of deciding indirect problem based on the discomfort of first kind integral operator, and its ill-posedness is embodied in: reconstructed results is sensitive to the minor variations of measuring projection, is difficult to obtain stable solution.For ECT, the minor variations of electric capacity between detecting electrode can be introduced inevitably and measure noise, and these noises have had a strong impact on the stability of reconstructed image, have caused the distortion of imaging system, have reduced the reliability of reconstructed results.(Chen Deyun such as Chen Deyun, Zheng Guibin, Yu Xiaoyang, Sun Li engraves. based on the research of genetic algorithm capacitance chromatography imaging image reconstruction algorithm, Electric Machines and Control, 2003,7 (3): 207-211.) introduce genetic algorithm, trial overcomes the image reconstruction ill-posedness in conjunction with the flow pattern characteristics with the inherent high randomness of genetic algorithm, adaptivity and complex nonlinear problem globally optimal solution search high robust, has obtained effect preferably.For X-CT, when data for projection incomplete (less), generally adopt solution by iterative method after turning to system of linear equations with projection equation's group is discrete as projection angle, also there is the problem of the ill-posedness that is similar to ECT.
Genetic algorithm is a kind of random search algorithm based on biological natural selection and hereditary mechanism, it is compared with respect to the searching algorithm of traditional based target function minimization, have following characteristics: the search of genetic algorithm starts from a population of separating, rather than the single of traditional optimization separated; Genetic algorithm is only used just when function, and does not use subsidiary conditions such as the gradient of object function or higher derivative; Genetic algorithms use at random, rather than the state transitions rule of determining.Above characteristics make genetic algorithm can carry out the global search under the probability meaning very effectively, thus the optimal solution of the problem that converges on of having the ability.The operation of genetic algorithm generally includes: the generation of coding and initial population; Fitness assessment and judgement convergence; Selection operation, interlace operation and mutation operation.
Genetic algorithm is from arbitrary initial population, by selecting (making excellent individual have more multimachine can pass to filial generation), intersect (embodying the information exchange between excellent individual), variation (is introduced new individuality, the multiformity that keeps population) evolves to an operation population generation generation in the search volume near the optimum point, until converging to optimum solution point.
In addition, consider the physical features of multiphase flow system, as gas-liquid system is that typical local concentration is the system of 0-1, and the gas slurry is that nearly 0-1 system (slurry attitude be considered as intending homogeneous phase mutually), the biphase bubbling process of gas-solid also can be approximately 0-1 system (i.e. breast mutually with steep mutually); Simultaneously, the dynamic behaviour of the large-scale structure in the multiphase flow (as gas-liquid bubbling kinetics etc.) is often to flowing and the reactor behavior is started to control to make and used.At the multiphase flow system of two-value or approximate two-value, fully the physical features of articulated system is simplified the hardware and software requirement of X-CT technology in the multiphase flow measurement application process, and the deep structure research of chemical industry multiphase flow is had important scientific meaning and construction value.
Summary of the invention
The purpose of this invention is to provide a kind of X ray CT multiphase flow detection method based on genetic algorithm.Construct concrete genetic algorithm by distribution mutually or dielectric distribution rule in this way in conjunction with detected multiphase flow system, can realize flow field reconstruct, require (fladellums of hundreds of angles) can finish the fast imaging process better than the collection of the complete data for projection of traditional CT technology to limited angle (as the fladellum of 3～24 angles) X ray data for projection.
Therefore, in order to achieve the above object, the invention provides a kind of X ray CT multiphase flow detection method based on genetic algorithm, this method is carried out as follows:
1) broken detection architecture is simplified processing
In conjunction with the distribution mutually or the dielectric distribution rule of detected multiphase flow system, make it to be reduced to two-value or approximate two-value system, whole spatial phase/dielectric distribution non-0 is 1;
2) the geometry subdivision is carried out in detected zone
The two-dimensional section a certain to be measured of selecting the multiphase flow system is as detected zone, detected zone is split into several unit subregions, the interior media of setting each unit subregion is evenly distributed, and then the dielectric distribution assignment of all unit subregions constitutes an image vector;
3) collection of X ray data for projection
Utilize the fladellum X ray data for projection under x-ray source, X-ray detector and the data acquisition control equipment collection limited angle, the data for projection collection of different angles hockets by the rotation-emission of single x-ray source, and perhaps a plurality of x-ray sources are finished by electrical switch control emission successively;
4) discretization of X ray projection equation group
Calculate the weight coefficient that characterizes mutual relation between each bar X ray and each component of image vector, with the discrete system of linear equations Ax=b that turns to of projection equation's group, sketch map is seen Fig. 1, wherein: A represents the coefficient of relationship matrix, x presentation video vector, b represents projection vector, selects the image reconstruction optiaml ciriterion simultaneously; Described weight coefficient is determined based on one of following four kinds of ray definition modes:
1. ignore beamwidth, when ray passed a unit subregion, the weight coefficient of getting this unit subregion correspondence was 1, otherwise is 0;
2. ignore beamwidth, when ray passes a unit subregion, get this unit subregion and intercept the length of ray as the weight coefficient that characterizes its projection contribution;
3. ray is regarded as the thick line with certain width, get this ray and cover the area of each unit subregion as the weight coefficient that characterizes its projection contribution;
4. ray is regarded as the thick line with certain width, think that subarea, unit valuation of a field concentrates on the center of unit subregion, the weight coefficient of this unit subregion correspondence is 1 during this center of ray process, otherwise is 0;
5) the concrete genetic algorithm of structure is applied to image reconstruction procedure
Distribution mutually or dielectric distribution rule in conjunction with detected multiphase flow system are constructed concrete genetic algorithm, design is based on selection operator, crossover operator and the mutation operator of the physical model that flows, and introduce computational methods of evolving simultaneously on multiple populations, realization is to effective reconstruction calculations of image vector, and the concrete constitution step of described genetic algorithm is as follows:
1. select the selection operator of " sequencing selection " and the crossover operator of " single-point intersection ";
2. introduce mutation operator based on node: carry out per generation during genetic manipulation equiprobability get the some nodes in space at random, add up subregion 0-1 distribution situation in unit in the related zone of node one by one, note N (0) is 0 value cell subregion number, and N (1) is 1 value cell subregion number; As N (0), N (1) is arbitrary when being 0, each unit subregion aberration rate is set at 0 in the related zone; When N (0), N (1) all were not 0, the aberration rate of 0,1 each unit subregion of value was set at respectively in the related zone:
6) numerical solution is converted into the phase/dielectric distribution figure in detected zone
The image vector numerical solution of being tried to achieve is converted into the phase/dielectric distribution figure and the output in detected cross section.
In technique scheme, technical characterictic of the present invention also is: the geometry grid that subdivision adopted in the detected zone described step 2) comprises square net and finite element triangle gridding; Described limited angle in the step 3) comprises 3～36 angles; Optiaml ciriterion described in the step 4) comprises following four kinds: least squared criterion, maximum criterion of homogeneity and level and smooth criterion, maximum entropy criterion, bayesian criterion; Wherein least squared criterion is to lead then, and the three is supplemental provisions in addition, selects some supplemental provisions then to be used in combination with main according to the particular problem needs; Genetic algorithm described in the step 5) comprises the improvement genetic algorithm and the fuzzy genetic algorithm of other modes; Phase/dielectric distribution described in the step 6) comprises phase content distribution, dielectric distribution and flow pattern.
The present invention compares in prior art, has following substantive distinguishing features and remarkable result.On the one hand, inherited the high advantage of x-ray ct technology spatial resolving power, by introducing system priori understanding to reduce the requirement of X ray data for projection completeness, and then improved time resolution, can reach the effect of quick recording projection data, expand the adaptability and the range of application of x-ray ct technology.In addition on the one hand, the articulated system physical features is constructed concrete genetic algorithm, has realized the solution procedure of constrained optimization, has the ability of stronger anti-measurement noise jamming, effectively controlled the ill-posedness of image reconstruction procedure, obviously weakened the influence of measurement error reconstruction result.
Description of drawings
Fig. 1 is projection equation's discretization sketch map.
Fig. 2 gathers sketch map for data for projection.
Fig. 3 is detected regional finite element triangle gridding subdivision sketch map.
Fig. 4 is image reconstruction result of the present invention under the different projected angle number of degrees in the numerical simulation experiment.
Fig. 5 is the image reconstruction result of FBP method under the different projected angle number of degrees in the numerical simulation experiment.
Fig. 6 is image reconstruction result of the present invention under the different projected angle number of degrees in the static experiment.
Fig. 7 is the image reconstruction result of FBP method under the different projected angle number of degrees in the static experiment.
Fig. 8 be static experiment x-ray source projection value under certain three projection angle with the comparison of theoretical value.
Fig. 9 is complicated multivesicular body system's application image reconstruction result of the present invention under the different mesh generations.
The specific embodiment
Below in conjunction with example, concrete enforcement of the present invention is further described.
Broken detection architecture as shown in Figure 2, cross section to be measured is that a diameter is the disc of 90mm, also promptly maximum round region is an initial point with its center of circle, x axle forward level is towards the right side, y axle forward is set up rectangular coordinate system straight up, unit is mm.The cross section is by two phase compositions of air-water: air is distributed in four circles mutually, and home position is respectively (24,10), (22,15), (2 ,-2) and (2 ,-29), and corresponding diameter is respectively 20,16,16 and 20mm; All the other zones are water.
1) detected system is simplified processing
Air-water two-phase system is reduced to 0-1 two-value system, and 0 represents air, and 1 represents water.
2) the geometry subdivision is carried out in detected zone
Carry out the square net subdivision as shown in Figure 1, detected zone is placed in the circumscribed rectangle of disc and goes, the rectangular area evenly is divided into several rectangular cells subregions.Carry out the finite element grid subdivision, during as the employing triangle gridding, the thickness that grid is divided has determined the height of spatial resolving power, and meticulous more grid more can accurate description cross section phase/dielectric distribution.In the algorithm implementation procedure, according to the requirement of spatial resolution, whether decision carries out " zoom operation " to grid, also promptly each unit subregion is carried out subdivision again, as to connect its subdivision of naming a person for a particular job in leg-of-mutton three limits be the triangle of four congruences.
As shown in Figure 3, the triangle gridding subdivision is carried out in zone to be measured one time, common property is given birth to 128 unit subregions, adjusts node location, makes triangle as far as possible near equilateral triangle.On the reconstruction result basis under the above-mentioned subdivision, carry out twice " zoom operation ", then can obtain 2048 reconstruction result under the unit subregion.
3) collection of X ray data for projection
Utilize the fladellum X ray data for projection under x-ray source, X-ray detector and the data acquisition control equipment collection limited angle (as the fladellum of 3～36 angles), the data for projection collection of different angles hockets by the rotation-emission of single x-ray source, and perhaps a plurality of x-ray sources are finished by electrical switch control emission successively.Required radiogenic energy and quantity need be by in advance this system being estimated and being determined.
The data for projection collection divides two kinds of forms: numerical simulation experiment and static experiment.The former accurately calculates projection vector by known cross section phase/dielectric distribution; The flash of light X-ray machine that it is 150 kV that the latter adopts a running voltage and an X ray flat-panel detector, by the rotation static models, substep is realized the data for projection collection of different angles.Equidistantly settle some groups of x-ray sources and detector according to numerical simulation experiment or static experiment, and the two with initial point apart from reservation fixed value (being respectively 180mm and 130mm).Have 196 rays on the X ray sector of the 40 degree subtended angles that each direction is launched, the X ray width is 1mm, and the projector distance on detector is equidistant.All got 24 angles (finishing a projection every 15 degree) in numerical simulation experiment and static experimentation, 3,4,6,8,12 projection angles or whole projection angle data of getting respectively in the restructuring procedure are wherein carried out image reconstruction.
4) discretization of X ray projection equation group
The difference of iterative reconstruction algorithm and backprojection reconstruction algorithm maximum be the former at the very start with cross-sectional image f (r, θ) discretization, as shown in Figure 1, centrifugal pump x (k)=f (r of each subregion inside, unit
k, θ
k(i j), and is a constant to)=pixel, and this centrifugal pump is represented gray scale or density, and wherein the expression way of x (k) is applicable to finite element triangle gridding partition patterns, and (i j) is applicable to the grid partition patterns to pixel.Image vector x is the K dimensional vector.
The total projection value of p bar ray is
R wherein
PkBe the weight coefficient of k unit subregion to the contribution of total projection value.If the projection number is P, then obtain the projection equation of P discretization, remember A={r again
Pk}
P * K, further can obtain system of linear equations Ax=b.Wherein A just determines when data collecting system being set and selecting projection ray, the projection value vector that b is just corresponding.
In the present embodiment, each component of image vector x can only get 0 or 1, the thickness of each component representative all water that every corresponding ray penetrated of projection vector.Image vector dimension K=128 or 2048; The total P=196*NA of ray, wherein NA is the number of fladellum projection.
Because this equation has stronger ill-posedness, solution vector makes the problem solving difficulty very big not in the continuous space territory, and we are converted into constrained optimization problems with it
Its physical significance is: choose x, make thus the quadratic sum minimum of the error of the projection vector that generates and measured value.According to concrete physical process feature, needing increases different supplemental provisions, makes the more realistic process of result of optimizing.
5) the concrete genetic algorithm of structure is applied to image reconstruction procedure
For this problem, the effect of genetic algorithm is continuous search, accepts or rejects x, makes it more and more can satisfy the requirement of dwindling distance between Ax and the b, thereby has avoided directly finding the solution the ill effect that produces in the Ax=b process.
General genetic algorithm all has limitation to this particular problem at aspects such as convergence, effectiveness and efficient, must construct concrete genetic algorithm in conjunction with the distribution mutually or the dielectric distribution rule of detected multiphase flow system, design is based on selection operator, crossover operator and the mutation operator of the physical model that flows, and introduce computational methods of evolving simultaneously on multiple populations, realization is to effective reconstruction calculations of image vector, and the concrete constitution step of described genetic algorithm is as follows:
1. select the selection operator of " sequencing selection " and the crossover operator of " single-point intersection ", wherein the crossing-over rate of " single-point intersection " is set at 0.8;
2. introduce mutation operator based on node: carry out per generation during genetic manipulation equiprobability get the some nodes in space at random, add up subregion 0-1 distribution situation in unit in the related zone of node one by one, note N (0) is 0 value cell subregion number, and N (1) is 1 value cell subregion number; As N (0), N (1) is arbitrary when being 0, each unit subregion aberration rate is set at 0 in the related zone; When N (0), N (1) all were not 0, the aberration rate of 0,1 each unit subregion of value was set at respectively in the related zone:
When implementing " zoom operation ", the reconstruction result under the coarse grid is suitably handled,, helped more reliable, the more effectively convergence of restructuring procedure as the iterative initial value under the refined net.
6) numerical solution is converted into brokenly the phase/dielectric distribution figure of surveyed area
Genetic algorithm has two stop criterions, also promptly multiplies between algebraical sum generation just when deviation, and the former depends on the setting of the character of problem own and selection, intersection, mutation operator, and the latter is depended on the setting just when function.For taking into account the reliability of separating and the efficient of computing, add that this as the random optimization process, must carry out repeatedly reconstruct, determine more excellent operation setting and stop criterion, obtain real stable reliably separating.Afterwards, the image vector numerical solution of being tried to achieve is converted into the phase/dielectric distribution figure and the output in detected cross section.
In this example, according to the complexity of restructuring procedure procreation algebraically being set is 500～10000, gets 10 just when deviation between generation
-6, guarantee that restructuring procedure obtains stable reliably separating.Again image vector is converted into the phase/dielectric distribution figure and the output in detected cross section, shown in Fig. 4,6 series (K=2048), the desired locations that gives bubble among the figure is to do reference, and a, b, c, d, e and f represent the reconstruction result of 3,4,6,8,12 and 24 angle data for projection respectively.
Through numerical simulation experiment and static experimental verification, can find that improved genetic algorithm can control the ill-posedness of image reconstruction really effectively, under bigger noise jamming, can also obtain stable reliably separating.For superiority of the present invention is described, use the filtered back-projection reconstruction algorithm (FBP) of classical CT restructing algorithm to calculate same example, reconstruction result is shown in Fig. 5,7 series, a, b, c, d, e and f represent the reconstruction result of 3,4,6,8,12 and 24 angle data for projection respectively, wherein a, b two figure omit owing to lack physical significance among Fig. 7.
As Fig. 4, by the numerical simulation result of experiment as can be known, from 3 to 24 projection angles, the present invention can reconstruct the position and the general shape of bubble (air phase), and along with the projected angle number of degrees increase, the description on phase border constantly improves, and reconstruction result is quite desirable.As Fig. 5, investigate the reconstruction result of FBP method, under 8 projection angles, the position of bubble and shape still can't be discerned judgement, at least need 12 projection angles cooperation human brain intelligence just can identify the position and the shape of bubble, and reconstructed image is all quite serious at detected pseudo-shadow within and outside the region.The result who contrasts the two as can be known, the FBP method is helpless for the limited angle data reconstruction.Algorithm couples of the present invention the physical features of system, demonstrated and solved the greater advantage that limited angle data reconstruction problem is had under the two-value system.
The numerical simulation description of test potential advantages of algorithm, and the data for projection of being gathered in the actual measurement process will be with systematic error and random error, the feasibility of this method must be through the checking of Physical Experiment.Be subject to experiment condition, the precision of our static experimental facilities is not high, can not show a candle to the projection accuracy of medicine CT device, and the measured value of certain three angle fladellums projection and theoretical value are as shown in Figure 8.Though there is no small deviation, reconstruction result is still very noticeable.As Fig. 6, by static result of experiment as can be known, even under three angles, the present invention also can provide bubble approximate location and shape, and along with the projected angle number of degrees increase, the description on phase border also can constantly improve, and it is desirable that reconstruction result is tending towards; Can notice simultaneously that there are fixed deviation in reconstruction result and expected result under the more projection number, this should be attributed to the imperfection of experimental system.As Fig. 7, investigate the reconstruction result of FBP method, even under 24 projection angles, reconstruction result still is undesirable, the boundary of bubble has than large deviation; And under 12 projection angles, reconstruction result is very fuzzy.The result who contrasts the two as can be known, the present invention has bigger superiority aspect the limited angle data reconstruction problem solving under the two-value system.
At above-mentioned system, the present invention has obtained more satisfactory reconstruct effect in numerical simulation experiment and static experiment, but bubble quantity and bubble may be far away more than 4 in the general system, and the size of bubble also can be smaller.Apply the present invention to the more complicated system that comprises 10 bubbles, adopt 4 projection angles, the numerical simulation result as shown in Figure 9 under different mesh generations.By the result as can be known, the present invention has successfully determined the position and the general shape of all bubbles.
Another one unique distinction of the present invention is: as long as the permission of computer hardware condition, " zoom operation " can constantly carry out, and reconstruction result can constantly be improved, and more becomes desirable.
Therefore, the X ray CT multiphase flow detection method based on genetic algorithm that the present invention proposes is truly feasible, has its original superiority.
Claims (6)
1. X ray CT multiphase flow detection method based on genetic algorithm is characterized in that this method carries out as follows:
1) detected system is simplified processing
In conjunction with the distribution mutually or the dielectric distribution rule of detected multiphase flow system, make it to be reduced to two-value or approximate two-value system, whole spatial phase/dielectric distribution non-0 is 1;
2) the geometry subdivision is carried out in detected zone
The two-dimensional section a certain to be measured of selecting the multiphase flow system is as detected zone, detected region geometry is split into several unit subregions, the interior media of setting each unit subregion is evenly distributed, and then the dielectric distribution assignment of all unit subregions constitutes an image vector;
3) collection of X ray data for projection
Utilize the fladellum X ray data for projection under x-ray source, X-ray detector and the data acquisition control equipment collection limited angle, the data for projection collection of different angles hockets by the rotation-emission of single x-ray source, and perhaps a plurality of x-ray sources are finished by electrical switch control emission successively;
4) discretization of X ray projection equation group
Calculate the weight coefficient that characterizes mutual relation between each bar X ray and each component of image vector, turn to system of linear equations Ax=b with projection equation's group is discrete, wherein: A represents the coefficient of relationship matrix, x presentation video vector, b represents projection vector, selects the image reconstruction optiaml ciriterion simultaneously; Described weight coefficient is determined based on one of following four kinds of ray definition modes:
1. ignore beamwidth, when ray passed a unit subregion, the weight coefficient of getting this unit subregion correspondence was 1, otherwise is 0;
2. ignore beamwidth, when ray passes a unit subregion, get this unit subregion and intercept the length of ray as the weight coefficient that characterizes its projection contribution;
3. ray is regarded as the thick line with certain width, get this ray and cover the area of each unit subregion as the weight coefficient that characterizes its projection contribution;
4. ray is regarded as the thick line with certain width, think that subarea, unit valuation of a field concentrates on the center of unit subregion, the weight coefficient of this unit subregion correspondence is 1 during this center of ray process, otherwise is 0;
5) the concrete genetic algorithm of structure is applied to image reconstruction procedure
Distribution mutually or dielectric distribution rule in conjunction with detected multiphase flow system are constructed concrete genetic algorithm, design is based on selection operator, crossover operator and the mutation operator of the physical model that flows, and introduce computational methods of evolving simultaneously on multiple populations, realization is to effective reconstruction calculations of image vector, and the concrete constitution step of described genetic algorithm is as follows:
1. select the selection operator of " sequencing selection " and the crossover operator of " single-point intersection ";
2. introduce mutation operator based on node: carry out per generation during genetic manipulation equiprobability get the some nodes in space at random, add up subregion 0-1 distribution situation in unit in the related zone of node one by one, note N (0) is 0 value cell subregion number, and N (1) is 1 value cell subregion number; As N (0), N (1) is arbitrary when being 0, each unit subregion aberration rate is set at 0 in the related zone; When N (0), N (1) all were not 0, the aberration rate of 0,1 each unit subregion of value was set at respectively in the related zone:
6) numerical solution is converted into the phase/dielectric distribution figure in detected zone
The image vector numerical solution of being tried to achieve is converted into the phase/dielectric distribution figure and the output in detected cross section.
2. the X ray CT multiphase flow detection method based on genetic algorithm as claimed in claim 1 is characterized in that: the geometry grid that subdivision adopted in the detected zone described step 2) comprises square net and finite element triangle gridding.
3. the X ray CT multiphase flow detection method based on genetic algorithm as claimed in claim 1, it is characterized in that: the described limited angle in the step 3) comprises 3～36 angles.
4. the X ray CT multiphase flow detection method based on genetic algorithm as claimed in claim 1, it is characterized in that: the optiaml ciriterion described in the step 4) comprises following four kinds: least squared criterion, maximum criterion of homogeneity and level and smooth criterion, maximum entropy criterion, bayesian criterion; Wherein least squared criterion is to lead then, and the three is supplemental provisions in addition, selects some supplemental provisions then to be used in combination with main according to the particular problem needs.
5. the X ray CT multiphase flow detection method based on genetic algorithm as claimed in claim 1, it is characterized in that: the genetic algorithm described in the step 5) comprises the improvement genetic algorithm and the fuzzy genetic algorithm of other modes.
6. the X ray CT multiphase flow detection method based on genetic algorithm as claimed in claim 1, it is characterized in that: the phase/dielectric distribution described in the step 6) comprises phase content distribution, dielectric distribution and flow pattern.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CNB2005100862502A CN100403987C (en) | 2005-08-19 | 2005-08-19 | X-ray CT multi-phase current inspection based on hereditary algorithm |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CNB2005100862502A CN100403987C (en) | 2005-08-19 | 2005-08-19 | X-ray CT multi-phase current inspection based on hereditary algorithm |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN1745713A CN1745713A (en) | 2006-03-15 |
| CN100403987C true CN100403987C (en) | 2008-07-23 |
Family
ID=36165546
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CNB2005100862502A Expired - Fee Related CN100403987C (en) | 2005-08-19 | 2005-08-19 | X-ray CT multi-phase current inspection based on hereditary algorithm |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN100403987C (en) |
Families Citing this family (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN103308535B (en) * | 2012-03-09 | 2016-04-13 | 同方威视技术股份有限公司 | For equipment and the method for ray scanning imaging |
| WO2016002084A1 (en) * | 2014-07-04 | 2016-01-07 | 株式会社島津製作所 | Image reconstruction processing method |
| CN105701824B (en) * | 2016-01-14 | 2019-02-19 | 天津大学 | Electricity based on arc-shaped mesh generation chromatographs image rebuilding method |
| CN105701825B (en) * | 2016-01-14 | 2019-02-19 | 天津大学 | The arc-shaped grid cutting algorithm solved for electricity chromatography image sensitivity coefficient |
| CN108711178B (en) * | 2018-05-21 | 2021-09-28 | 北京航空航天大学 | Capacitance tomography image reconstruction method based on closed-loop control principle |
Citations (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| WO2005019779A1 (en) * | 2003-08-22 | 2005-03-03 | Instituto Mexicano Del Petróleo | Method of viewing multiphase flows using electrical capacitance tomography |
-
2005
- 2005-08-19 CN CNB2005100862502A patent/CN100403987C/en not_active Expired - Fee Related
Patent Citations (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| WO2005019779A1 (en) * | 2003-08-22 | 2005-03-03 | Instituto Mexicano Del Petróleo | Method of viewing multiphase flows using electrical capacitance tomography |
Non-Patent Citations (2)
| Title |
|---|
| 基于遗传算法电容层析成像图像重建算法的研究. 陈德运，郑贵滨，于晓洋，孙立镌.电机与控制学报，第7卷第3期. 2003 |
| 基于遗传算法电容层析成像图像重建算法的研究. 陈德运，郑贵滨，于晓洋，孙立镌.电机与控制学报，第7卷第3期. 2003 * |
Also Published As
| Publication number | Publication date |
|---|---|
| CN1745713A (en) | 2006-03-15 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| Chauhan et al. | Processing of rock core microtomography images: Using seven different machine learning algorithms | |
| Neumann et al. | Quantifying the influence of microstructure on effective conductivity and permeability: virtual materials testing | |
| CN100403987C (en) | X-ray CT multi-phase current inspection based on hereditary algorithm | |
| US8559592B2 (en) | System and method for image reconstruction by using multi-sheet surface rebinning | |
| Sharp et al. | Pointtrinet: Learned triangulation of 3d point sets | |
| Gerke et al. | Description and reconstruction of the soil pore space using correlation functions | |
| CN112085829A (en) | Spiral CT image reconstruction method and equipment based on neural network and storage medium | |
| Fend et al. | Reconstruction of highly porous structures from FIB‐SEM using a deep neural network trained on synthetic images | |
| CN101861609A (en) | Reactor dosimetry applications using a parallel 3-D radiation transport code | |
| Zubov et al. | Pore-network extraction using discrete Morse theory: Preserving the topology of the pore space | |
| Röding et al. | Machine learning-accelerated small-angle X-ray scattering analysis of disordered two-and three-phase materials | |
| Zheng et al. | The effect of concave and convex weight adjustments on self-organizing maps | |
| Mishra et al. | A priority based genetic algorithm for limited view tomography | |
| Scheible et al. | Tomosim: Simulation of filamentous cryo-electron tomograms | |
| Yang et al. | A hybrid tomographic reconstruction algorithm for high speed X-ray tomography | |
| Ching et al. | Rotation-as-fast-axis scanning-probe x-ray tomography: the importance of angular diversity for fly-scan modes | |
| Balázs et al. | An evolutionary approach for object-based image reconstruction using learnt priors | |
| Bendito et al. | Computational cost of the Fekete problem I: The Forces Method on the 2-sphere | |
| Hur et al. | Topology optimization based on spline-based meshfree method using topological derivatives | |
| Romaszko et al. | Massive dimensionality reduction for the left ventricular mesh | |
| Lang et al. | Big Data Analytics for the Inspection of Battery Materials | |
| Carosio et al. | Improving efficiency in electrical impedance tomography problem by hybrid parallel genetic algorithm and a priori information | |
| Shi et al. | Application of grid-based k-means clustering algorithm for optimal image processing | |
| Zhang et al. | Image reconstruction of ECT based on second-order hybrid sensitivity matrix and fuzzy nonlinear programming | |
| Bielecki | A genetic algorithm-based method of neutron emissivity tomographic inversion for tokamak plasma |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| C06 | Publication | ||
| PB01 | Publication | ||
| C10 | Entry into substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| C14 | Grant of patent or utility model | ||
| GR01 | Patent grant | ||
| CF01 | Termination of patent right due to non-payment of annual fee |
Granted publication date: 20080723 Termination date: 20170819 |
|
| CF01 | Termination of patent right due to non-payment of annual fee |