CN106644375A - Multi-source information fusion method based on wavelet multi-resolution double-layer complex network and application thereof - Google Patents
Multi-source information fusion method based on wavelet multi-resolution double-layer complex network and application thereof Download PDFInfo
- Publication number
- CN106644375A CN106644375A CN201610888616.6A CN201610888616A CN106644375A CN 106644375 A CN106644375 A CN 106644375A CN 201610888616 A CN201610888616 A CN 201610888616A CN 106644375 A CN106644375 A CN 106644375A
- Authority
- CN
- China
- Prior art keywords
- network
- node
- layer
- wavelet
- average
- 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.)
- Granted
Links
Classifications
-
- G—PHYSICS
- G01—MEASURING; TESTING
- G01M—TESTING STATIC OR DYNAMIC BALANCE OF MACHINES OR STRUCTURES; TESTING OF STRUCTURES OR APPARATUS, NOT OTHERWISE PROVIDED FOR
- G01M10/00—Hydrodynamic testing; Arrangements in or on ship-testing tanks or water tunnels
Landscapes
- Physics & Mathematics (AREA)
- Fluid Mechanics (AREA)
- General Physics & Mathematics (AREA)
- Investigating Or Analyzing Materials By The Use Of Electric Means (AREA)
Abstract
The invention relates to a multi-source information fusion method based on a wavelet multi-resolution double-layer complex network and an application thereof. The multi-source information fusion method comprises the steps of respectively building wavelet coefficient complex networks for multichannel time sequences acquired through a cyclic excitation dual-mode sensor formed by a cyclic excitation multi-electrode conductive sensor and a cyclic excitation multi-electrode capacitive sensor; building a wavelet multi-resolution aggregation network, and respectively drawing a relation diagram containing flow pattern information between diagram energy and the resolution and a relation diagram containing flow pattern information between an interactive clustering coefficient entropy and the resolution; and building a wavelet double-layer complex network, and respectively drawing a relation diagram containing flow pattern information between the mean of node clustering coefficients in each layer of the network and the resolution, the mean of node degrees in each layer of the network and the resolution and the mean of node feature vector centralities in each layer of the network and the resolution. According to the invention, a vertical oil-water two-phase flow experiment conducted by adopting the cyclic excitation dual-mode sensor formed by the cyclic excitation multi-electrode conductive sensor and the cyclic excitation multi-electrode capacitive sensor; and the experiment is conducted by fixing the proportion of an oil phase and a water phase and changing the flow of the oil phase and the water phase. The multi-source information fusion method realizes recognition for a complex flow structure.
Description
Technical field
The present invention relates to a kind of two phase flow bimodal multi-sources Information Fusion Method.It is more particularly to a kind of to be directed to water-oil phase
The Multi-source Information Fusion based on Wavelet Multiresolution Decomposition bilayer complex network of stream cycle motivation dual-modality sensor multi channel signals
Method and application.
Background technology
Oil-water two-phase flow is widely present in oil exploitation and transportation industry.In water-oil phase streaming system, the distribution of each phase
It is being continually changing with space over time, is defining different nowed formings, referred to as flow pattern.The flow pattern of two phase flow is complicated more
Become, local flow information is difficult to accurately seizure so that the measurement of the two-phase flow parameter such as phase content has many difficult points.This is right
In oil exploitation with technological transformation into many impacts.At present, for the research of flow pattern mainly adopts observation method, wavelet character
Analysis and fuzzy C-means clustering, fuzzy logic and genetic algorithm, Digital Image Processing algorithm etc..For phase content measurement more adopt
Conductance method, capacitance method, optical method and ray method etc..Annular conductivity sensor and double helix capacitance sensing that traditional measurement is adopted
Device etc., is single channel sensor, is easily lost the local flow information of microcosmic.And distributing triggers reorganization and excitation cycle swash
Encouraging sensor etc. then can capture more rich micro flow information while gather multi channel signals, be that flow pattern and flow pattern are drilled
Flow mechanism research in change provides important technology support.
Nowadays Complex Networks Theory has been flourished multi-field since the foundation, is research complication system
One important tool, especially in recent years it has made major contribution in time series analysis.Practice have shown that, complex network pair
There is obvious advantage, effect with the important information in Kind of Nonlinear Dynamical System in excavating to be included in Nonlinear Time Series
Significantly.While its succeeding in single channel time series is applied, scientific research personnel is also caused how to begin to focus on it
In being applied to the convergence analysis of multi-channel data.Additionally, the past is concentrated mainly on single layer network for the research of complex network
In analysis, and the appearance of double-layer network then compensate for unicity of the single layer network in terms of fuse information, can more preferably merge not
With the information of attribute, so as to excavate the intrinsic characteristic of more abundant complication system.
The content of the invention
The technical problem to be solved be to provide it is a kind of can realize the identification to complex flow structure, portray
Flow pattern develop in internal motivation mechanism based on Wavelet Multiresolution Decomposition bilayer complex network Multi-source Information Fusion method and answer
With.
The technical solution adopted in the present invention is:A kind of two phase flow multi-source information based on Wavelet Multiresolution Decomposition double-layer network
Fusion method, comprises the steps:
1) to by following for being made up of cycle motivation multi-electrode conductivity sensor and cycle motivation multielectrode capacitance sensor
The multichannel time series that ring excitation dual-modality sensor is obtained builds respectively wavelet coefficient complex network;
2) Wavelet Multiresolution Decomposition converging network is built, and draws the figure energy for including flow pattern information respectively and intersect aggregation
The graph of a relation of coefficient entropy and resolution ratio;
3) small echo bilayer complex network is built, and draws the node rendezvous in the every layer network for including flow pattern information respectively
The graph of a relation of the average of coefficient, the average of the average of node degree and node diagnostic vector center and resolution ratio.
Step 1) include:
(1) S for including two phase flow fluid local flow information obtained by cycle motivation dual-modality sensor is obtained
Group length is the multichannel time series of L
(2) for the time series of each passage, the wavelet transformation of 6 resolution ratio is carried out, each resolution ratio obtains 1
Low frequency coefficient subband is approximation coefficient and 1 high frequency coefficient subband i.e. detail coefficients, and 6 low frequency coefficient subbands and 6 are obtained
High frequency coefficient subband, seeks maximum x of each low-frequency band and each high frequency band1, minimum of a value x2, mean value x3, standard deviation
x4, steepness function x5With kurtosis function x6;So for each resolution ratio, 12 characteristic values of low-frequency band and high frequency band
One characteristic vector of composition
(3) for multichannel time seriesIn each resolution ratio, when calculating every two passage
Between sequence characteristic vector Euclidean distance d=| | X (m)-X (n) | | m=1,2 ..., S n=1,2 ..., S, wherein, | | |
| represent and calculate Euclidean distance, X (m) represents the characteristic vector of passage m, the characteristic vector of X (n) passage n;By the letter of each passage
Number as complex network node, with the company side that two interchannel Euclidean distances determine two nodes in complex networks;Calculate all
Average M of Euclidean distancenAnd standard deviation, then threshold value R=M is obtainedn+ q σ, wherein q=0.15;If two passages
Euclidean distance is more than the threshold value, then do not connect side between two nodes, conversely, then there is even side between two nodes;Thus, at different points
Build wavelet coefficient complex network under resolution respectively;So, for a multichannel time series, 6 wavelet systems are obtained altogether
Number complex network.
Step 2) include:
(1) one is respectively obtained by cycle motivation multi-electrode conductivity sensor and cycle motivation multielectrode capacitance sensor
Multichannel time series, for each multichannel time series obtains a wavelet coefficient complex web in each resolution ratio respectively
Network;
(2) consider cycle motivation multi-electrode conductivity sensor and cycle motivation multielectrode capacitance sensor in each resolution ratio
It is lower to build a wavelet coefficient complex network respectively, so the two wavelet coefficient complex networks obtained under each resolution ratio, will
The total company side of the two wavelet coefficient complex networks retains, and interstitial content is constant, so as to constitute one under each resolution ratio
Wavelet Multiresolution Decomposition converging network and its corresponding adjacency matrix A;
(3) to described Wavelet Multiresolution Decomposition converging network, figure energy indexes E (A) is set:Wherein,
λiThe characteristic value of wavelet coefficient complex network adjacency matrix A new under each resolution ratio is represented, what n was represented is the number of characteristic value
Mesh;Setting convergence factor entropy index EC:
Wherein, TvWhat is represented is the number of the closing triangle for including node v in a new wavelet coefficient complex network
Mesh, kvWhat is represented is the degree of new wavelet coefficient Node Contraction in Complex Networks v, and what C (v) was represented is the convergence factor of node v;
(4) the figure energy for including flow pattern information and the graph of a relation for intersecting convergence factor entropy and resolution ratio are drawn respectively, are led to
The transformation for studying the two indexs in flow pattern evolutionary process is crossed, to recognize complex flow structure, flow pattern evolutionary process is indicated
In internal motivation mechanism.
Step 3) include:
(1) for step 1) two wavelet coefficient complex networks of each resolution ratio for building, it is double-deck multiple respectively as small echo
One layer of miscellaneous network, so as to obtain small echo bilayer complex network, the adjacency matrix of the small echo bilayer complex network is expressed asWherein, α represents the number of plies of double-deck wavelet coefficient complex network, if node i and j have connection in α layers,
Then corresponding element in adjacency matrixOtherwiseInterstitial content in two-tier network is all S;
(2) convergence factor of any layer α interior joint i of small echo bilayer complex network
Wherein, α ' represents another layer that α layers are different from small echo bilayer complex network,Withα is represented respectively
The element value of layer interior joint i and node j in adjacency matrix, the element value of α ' layer interior joint j and m in adjacency matrix, in α layers
The element value of node m and i in adjacency matrix;The convergence factor of each node in each layer is obtained, every layer of interior joint aggregation is calculated
Average C of coefficientα:
(3) degree of the arbitrary node i in the arbitrary α layers in small echo bilayer complex network is calculated Its
In,Represent the element value of α layer interior joint i and j in adjacency matrix;The degree of each node in each layer is obtained, calculates every
Average k of layer interior joint degreeα:
(4) eigenvector centrality of arbitrary α layers interior joint i of small echo bilayer complex networkRepresent the adjacent square of α layers
Battle array A[α]The corresponding characteristic vector of dominant eigenvalue i-th element value;Obtain in each layer in the characteristic vector of each node
Disposition value, calculates average E of every layer of interior joint eigenvector centralityα:
(5) average C of the node rendezvous coefficient in the every layer network for including flow pattern information is drawn respectivelyα, node degree
Average kαWith average E of node diagnostic vector centerαWith the graph of a relation of resolution ratio;Cycle motivation multi-electrode conductivity sensor pair
Locally low oil content measures sensitive height, and cycle motivation multielectrode capacitance sensor is high to the high oil content measurement sensitivity in local, electricity
Conduction holds the effective integration of multi-source metrical information and realizes complementary two phase flow spatial flow infomation detection, and by research stream is included
Average C of the node rendezvous coefficient in every layer network of the type information per layer networkα, node degree average kαWith node diagnostic vector
Central average EαWith the graph of a relation of resolution ratio, average C of the node rendezvous coefficient in per layer of every layer network is studiedα, node
Average k of degreeαWith average E of node diagnostic vector centerαThe transformation of three indexs in flow pattern evolutionary process, realizes to double
The fusion of modal information, indicates the internal motivation mechanism in flow pattern evolutionary process.
A kind of application of the two phase flow Multi-source Information Fusion method based on Wavelet Multiresolution Decomposition double-layer network, using sharp by circulation
The cycle motivation dual-modality sensor for encouraging multi-electrode conductivity sensor and cycle motivation multielectrode capacitance sensor composition is hung down
Straight oil-water two-phase flow experiment;The proportioning of fixed oil phase and water phase, the flow for changing oil phase and water phase is tested;Experimentation bag
Include following steps:
1) proportioning of fixed water phase and oil phase, is passed through a certain amount of water, then gradually to pipeline toward vertical ascent pipeline
In be passed through oil phase, when water-oil phase fully merge and it is gradually stable after, using cycle motivation conductivity sensor and cycle motivation electricity
Hold sensor and measure multichannel signal respectively, and flow pattern is recorded with high-speed camera instrument;
2) after once collection terminates, the flow of oil phase and water phase is changed, by step 1) process continues collection, until fixing
Proportioning under designed operating mode all complete;
3) change the proportioning of oil phase and water phase, repeat step 1 again) to step 2) process complete the measurement of this wheel, directly
All operating modes to design are all measured;
4) it is how electric for cycle motivation multi-electrode conductivity sensor and cycle motivation based on the multi-channel measurement signal for obtaining
The wavelet coefficient complex network that electrode capacitance sensor is set up respectively under different resolution, for two-tier network, to obtain a small echo more
Resolution ratio converging network;
5) figure energy and convergence factor entropy are calculated, draw respectively include flow pattern information different operating mode figure below energy,
Convergence factor entropy portrays the internal motivation mechanism in the evolutionary process of different flow patterns with the graph of a relation of change resolution;
6) for two-tier network structure small echo bilayer complex network, the node rendezvous Coefficient Mean of each layer network of calculating,
Node degree average, node diagnostic vector center average, draw respectively node rendezvous Coefficient Mean, the section for including flow pattern information
The graph of a relation of point degree average, 3 indexs of node diagnostic vector center average and change resolution, develops to study in flow pattern
During, the change of complex flow structure and internal motivation mechanism.
The Multi-source Information Fusion method based on Wavelet Multiresolution Decomposition bilayer complex network of the present invention and application, by by following
Ring encourages multi-electrode conductivity sensor and the cycle motivation dual-modality sensor of cycle motivation multielectrode capacitance sensor composition to obtain
To multi-channel data using the thought of wavelet transformation set up Wavelet Multiresolution Decomposition bilayer complex network method carry out multi-source letter
Breath fusion, extraction can indicate that the double-deck complex network index that flow pattern is generated and developed, and realize the identification to complex flow structure,
Portray the internal motivation mechanism during flow pattern develops.Have the advantages that:
(1) a kind of oil-water two-phase flow Multi-source Information Fusion side based on Wavelet Multiresolution Decomposition bilayer complex network is proposed
Method;
(2) the method can portray the internal motivation mechanism in oil-water two-phase flow flow pattern evolutionary process.
Description of the drawings
Fig. 1 is analysis schematic diagram of the present invention based on the Multi-source Information Fusion method of Wavelet Multiresolution Decomposition bilayer complex network.
Specific embodiment
With reference to the multi-source information based on Wavelet Multiresolution Decomposition bilayer complex network of embodiment and accompanying drawing to the present invention
Fusion method and application are described in detail.
The two phase flow Multi-source Information Fusion method based on Wavelet Multiresolution Decomposition double-layer network of the present invention, proposes that a kind of small echo is more
Resolution ratio bilayer complex network networking method, by adopting small echo to the multi-channel data that cycle motivation dual-modality sensor is obtained
The thought of conversion sets up the method for double-deck complex network and carries out Multi-source Information Fusion, extracts double-deck complex network index and portrays in stream
In type evolutionary process, internal motivation mechanism.
The two phase flow Multi-source Information Fusion method based on Wavelet Multiresolution Decomposition double-layer network of the present invention, comprises the steps:
1) to by following for being made up of cycle motivation multi-electrode conductivity sensor and cycle motivation multielectrode capacitance sensor
The multichannel time series that ring excitation dual-modality sensor is obtained builds respectively wavelet coefficient complex network;Including:
(1) S for including two phase flow fluid local flow information obtained by cycle motivation dual-modality sensor is obtained
Group length is the multichannel time series of L
(2) for the time series of each passage, the wavelet transformation of 6 resolution ratio is carried out, it is little using Daubechies2 ranks
Ripple (db2) decomposes to it, and Decomposition order is the resolution ratio of 6 layers, i.e., 6.Each resolution ratio obtains 1 low frequency coefficient subband
I.e. approximation coefficient and 1 high frequency coefficient subband are detail coefficients, be obtained after the Time Series 6 low frequency coefficient subbands and
6 high frequency coefficient subbands, seek maximum x of each low-frequency band and each high frequency band1, minimum of a value x2, mean value x3, standard
Deviation x4, steepness function x5With kurtosis function x6;So for each resolution ratio, 12 spies of low-frequency band and high frequency band
Value indicative constitutes a characteristic vector
(3) for multichannel time seriesIn each resolution ratio, when calculating every two passage
Between sequence characteristic vector Euclidean distance d=| | X (m)-X (n) | | m=1,2 ..., S n=1,2 ..., S, wherein, | | |
| represent and calculate Euclidean distance, X (m) represents the characteristic vector of passage m, the characteristic vector of X (n) passage n;By the letter of each passage
Number as complex network node, with the company side that two interchannel Euclidean distances determine two nodes in complex networks;Calculate all
Average M of Euclidean distancenAnd standard deviation, make coefficient q=0.15, then threshold value R=M is obtainedn+ q σ, wherein q=
0.15;If the Euclidean distance of two passages is more than the threshold value, side is not connected between two nodes, conversely, then having between two nodes
Lian Bian;Thus, wavelet coefficient complex network is built respectively under different resolution;So, for a multichannel time series,
6 wavelet coefficient complex networks are obtained altogether.
2) it is Wavelet Multiresolution Decomposition converging network to build Wavelet Multiresolution Decomposition aggregation network, and is painted respectively
System includes the figure energy of flow pattern information and intersects the graph of a relation of convergence factor entropy and resolution ratio;Including:
(1) one is respectively obtained by cycle motivation multi-electrode conductivity sensor and cycle motivation multielectrode capacitance sensor
Multichannel time series, for each multichannel time series obtains a wavelet coefficient complex web in each resolution ratio respectively
Network;
(2) consider cycle motivation multi-electrode conductivity sensor and cycle motivation multielectrode capacitance sensor in each resolution ratio
It is lower to build a wavelet coefficient complex network respectively, so the two wavelet coefficient complex networks obtained under each resolution ratio, will
The total company side of the two wavelet coefficient complex networks retains, and interstitial content is constant, so as to constitute one under each resolution ratio
Wavelet Multiresolution Decomposition converging network and its corresponding adjacency matrix A;
(3) to described Wavelet Multiresolution Decomposition converging network, figure energy indexes E (A) is set:Its
In, λiThe characteristic value of wavelet coefficient complex network adjacency matrix A new under each resolution ratio is represented, what n was represented is characteristic value
Number;Setting convergence factor entropy index EC:
Wherein, TvWhat is represented is the number of the closing triangle for including node v in a new wavelet coefficient complex network
Mesh, kvWhat is represented is the degree of new wavelet coefficient Node Contraction in Complex Networks v, and what C (v) was represented is the convergence factor of node v;
(4) the figure energy for including flow pattern information and the graph of a relation for intersecting convergence factor entropy and resolution ratio are drawn respectively, are led to
The transformation for studying the two indexs in flow pattern evolutionary process is crossed, to recognize complex flow structure, flow pattern evolutionary process is indicated
In internal motivation mechanism.
3) small echo bilayer complex network is built, and draws the node rendezvous in the every layer network for including flow pattern information respectively
The graph of a relation of the average of coefficient, the average of the average of node degree and node diagnostic vector center and resolution ratio, including:
(1) for step 1) two wavelet coefficient complex networks of each resolution ratio for building are double-deck complicated respectively as small echo
One layer of network, so as to obtain small echo bilayer complex network, the adjacency matrix of the small echo bilayer complex network is expressed asWherein, α represents the number of plies of small echo bilayer complex network, if node i and j have connection in α layers,
Corresponding element in adjacency matrixOtherwiseInterstitial content in two-tier network is all S;
(2) convergence factor of any layer α interior joint i of small echo bilayer complex network
Wherein, α ' represents another layer that α layers are different from small echo bilayer complex network,Withα is represented respectively
The element value of layer interior joint i and node j in adjacency matrix, the element value of α ' layer interior joint j and m in adjacency matrix, in α layers
The element value of node m and i in adjacency matrix;The convergence factor of each node in each layer is obtained, every layer of interior joint aggregation is calculated
Average C of coefficientα:
(3) degree of the arbitrary node i in the arbitrary α layers in small echo bilayer complex network is calculated Its
In,Represent the element value of α layer interior joint i and j in adjacency matrix;The degree of each node in each layer is obtained, calculates every
Average k of layer interior joint degreeα:
(4) eigenvector centrality of arbitrary α layers interior joint i of small echo bilayer complex networkRepresent the adjacent square of α layers
Battle array A[α]The corresponding characteristic vector of dominant eigenvalue i-th element value;Obtain in each layer in the characteristic vector of each node
Disposition value, calculates average E of every layer of interior joint eigenvector centralityα:
(5) average C of the node rendezvous coefficient in the every layer network for including flow pattern information is drawn respectivelyα, node degree
Average kαWith average E of node diagnostic vector centerαWith the graph of a relation of resolution ratio;Cycle motivation multi-electrode conductivity sensor pair
Locally low oil content measures sensitive height, and cycle motivation multielectrode capacitance sensor is high to the high oil content measurement sensitivity in local, electricity
Conduction holds the effective integration of multi-source metrical information and realizes complementary two phase flow spatial flow infomation detection, and by research stream is included
Average C of the node rendezvous coefficient in every layer network of the type information per layer networkα, node degree average kαWith node diagnostic vector
Central average EαWith the graph of a relation of resolution ratio, average C of the node rendezvous coefficient in per layer of every layer network is studiedα, node
Average k of degreeαWith average E of node diagnostic vector centerαThe transformation of three indexs in flow pattern evolutionary process, realizes to double
The fusion of modal information, indicates the internal motivation mechanism in flow pattern evolutionary process.
The application of the two phase flow Multi-source Information Fusion method based on Wavelet Multiresolution Decomposition double-layer network of the present invention, using by following
Ring encourages multi-electrode conductivity sensor and the cycle motivation dual-modality sensor of cycle motivation multielectrode capacitance sensor composition to enter
The vertical oil-water two-phase flow experiment of row;The proportioning of fixed oil phase and water phase, the flow for changing oil phase and water phase is tested;Each biography
Sensor is all made up of 16 electrodes, and every time in measurement, one of electrode is grounded as excitation end, an electrode, remaining 14
Electrode is received, the measurable signal for obtaining 16 × 14=224 passages of one cycle.Because electrode cycle encourages speed relative to stream
It is very fast for the speed of body flowing, it is possible to which that the signal for being equivalent to this 224 passage is measurement simultaneously, thus can effectively be caught
Catch abundant local flow information.Experimentation comprises the steps:
1) proportioning of fixed water phase and oil phase, is passed through a certain amount of water, then gradually to pipeline toward vertical ascent pipeline
In be passed through oil phase, when water-oil phase fully merge and it is gradually stable after, using cycle motivation conductivity sensor and cycle motivation electricity
Hold sensor and measure multichannel signal respectively, and flow pattern is recorded with high-speed camera instrument;
2) after once collection terminates, the flow of oil phase and water phase is changed, by step 1) process continues collection, until fixing
Proportioning under designed operating mode all complete;
3) change the proportioning of oil phase and water phase, repeat step 1 again) to step 2) process complete the measurement of this wheel, directly
All operating modes to design are all measured;
4) it is how electric for cycle motivation multi-electrode conductivity sensor and cycle motivation based on the multi-channel measurement signal for obtaining
The wavelet coefficient complex network that electrode capacitance sensor is set up respectively under different resolution, for two-tier network, to obtain a small echo more
Resolution ratio converging network;
5) figure energy and convergence factor entropy are calculated, draw respectively include flow pattern information different operating mode figure below energy,
Convergence factor entropy portrays the internal motivation mechanism in the evolutionary process of different flow patterns with the graph of a relation of change resolution;
6) for two-tier network structure small echo bilayer complex network, the node rendezvous Coefficient Mean of each layer network of calculating,
Node degree average, node diagnostic vector center average, draw respectively node rendezvous Coefficient Mean, the section for including flow pattern information
The graph of a relation of point degree average, 3 indexs of node diagnostic vector center average and change resolution, develops to study in flow pattern
During, the change of complex flow structure and internal motivation mechanism.
The present invention is divided by the multi-channel data obtained to cycle motivation dual-modality sensor using the method for wavelet transformation
Wavelet Multiresolution Decomposition converging network that Gou Jian be under different resolution and small echo bilayer complex network, respectively under different resolution
Complex network indices are calculated, drafting includes the different complex network index of flow pattern information with the pass of change resolution
System's figure, to study the internal motivation mechanism in flow pattern evolutionary process.
Above to the description of the present invention and embodiment, it is not limited to which this, the description in embodiment is only the reality of the present invention
One of mode is applied, it is any to design and the technical scheme without creative in the case of without departing from the invention objective
Similar structure or embodiment, belong to protection scope of the present invention.
Claims (5)
1. a kind of two phase flow Multi-source Information Fusion method based on Wavelet Multiresolution Decomposition double-layer network, it is characterised in that including as follows
Step:
1) circulation by being made up of cycle motivation multi-electrode conductivity sensor and cycle motivation multielectrode capacitance sensor is swashed
Encourage the multichannel time series that dual-modality sensor obtains and build wavelet coefficient complex network respectively;
2) Wavelet Multiresolution Decomposition converging network is built, and draws the figure energy for including flow pattern information respectively and intersect convergence factor
The graph of a relation of entropy and resolution ratio;
3) small echo bilayer complex network is built, and draws the node rendezvous coefficient in the every layer network for including flow pattern information respectively
Average, the average of the average of node degree and node diagnostic vector center and resolution ratio graph of a relation.
2. the two phase flow Multi-source Information Fusion method based on Wavelet Multiresolution Decomposition double-layer network according to claim 1, it is special
Levy and be, step 1) include:
(1) the S groups for including two phase flow fluid local flow information obtained by cycle motivation dual-modality sensor are obtained long
Degree is the multichannel time series of L
(2) for the time series of each passage, the wavelet transformation of 6 resolution ratio is carried out, each resolution ratio obtains 1 low frequency
Coefficient subband is approximation coefficient and 1 high frequency coefficient subband i.e. detail coefficients, and 6 low frequency coefficient subbands and 6 high frequencies are obtained
Coefficient subband, seeks maximum x of each low-frequency band and each high frequency band1, minimum of a value x2, mean value x3, standard deviation x4、
Steepness function x5With kurtosis function x6;So for each resolution ratio, 12 eigenvalue clusters of low-frequency band and high frequency band into
One characteristic vector
(3) for multichannel time seriesIn each resolution ratio, every two channel times sequence is calculated
Euclidean distance d=of the characteristic vector of row | | X (m)-X (n) | | m=1,2 ..., S n=1,2 ..., S, wherein, | | | | represent
Euclidean distance is calculated, X (m) represents the characteristic vector of passage m, the characteristic vector of X (n) passage n;Using the signal of each passage as
The node of complex network, with the company side that two interchannel Euclidean distances determine two nodes in complex network;Calculate it is all it is European away from
From average MnAnd standard deviation, then threshold value R=M is obtainedn+ q σ, wherein q=0.15;If the Euclidean of two passages away from
From more than the threshold value, then side is not connected between two nodes, conversely, then there is even side between two nodes;Thus, under different resolution
Wavelet coefficient complex network is built respectively;So, for a multichannel time series, 6 wavelet coefficient complexity are obtained altogether
Network.
3. the two phase flow Multi-source Information Fusion method based on Wavelet Multiresolution Decomposition double-layer network according to claim 1, it is special
Levy and be, step 2) include:
(1) respectively obtained by cycle motivation multi-electrode conductivity sensor and cycle motivation multielectrode capacitance sensor and led to more one
Road time series, for each multichannel time series obtains a wavelet coefficient complex network in each resolution ratio respectively;
(2) consider that cycle motivation multi-electrode conductivity sensor and cycle motivation multielectrode capacitance sensor divide under each resolution ratio
Not Gou Jian a wavelet coefficient complex network, so the two wavelet coefficient complex networks obtained under each resolution ratio, by this two
The total company side of individual wavelet coefficient complex network retains, and interstitial content is constant, so as to constitute a small echo under each resolution ratio
Multiresolution converging network and its corresponding adjacency matrix A;
(3) to described Wavelet Multiresolution Decomposition converging network, figure energy indexes E (A) is set:Wherein, λiTable
Show the characteristic value of wavelet coefficient complex network adjacency matrix A new under each resolution ratio, what n was represented is the number of characteristic value;
Setting convergence factor entropy index EC:
Wherein, TvRepresent be include in a new wavelet coefficient complex network node v closing triangle number, kv
What is represented is the degree of new wavelet coefficient Node Contraction in Complex Networks v, and what C (v) was represented is the convergence factor of node v;
(4) the figure energy for including flow pattern information and the graph of a relation for intersecting convergence factor entropy and resolution ratio are drawn respectively, by grinding
Transformation of the two indexs in flow pattern evolutionary process is studied carefully, to recognize complex flow structure, in indicating flow pattern evolutionary process
Internal motivation mechanism.
4. the two phase flow Multi-source Information Fusion method based on Wavelet Multiresolution Decomposition double-layer network according to claim 1, it is special
Levy and be, step 3) include:
(1) for step 1) two wavelet coefficient complex networks of each resolution ratio for building, respectively as the one of small echo bilayer complex network
Layer, so as to obtain small echo bilayer complex network, the adjacency matrix of the small echo bilayer complex network is expressed as
Wherein, α represents the number of plies of double-deck wavelet coefficient complex network, corresponding in adjacency matrix if node i and j have connection in α layers
ElementOtherwiseInterstitial content in two-tier network is all S;
(2) the convergence factor C of any layer α interior joint i of small echo bilayer complex networki [α]:
Wherein, α ' represents another layer that α layers are different from small echo bilayer complex network,WithRepresent respectively in α layers
The element value of node i and node j in adjacency matrix, the element value of α ' layer interior joint j and m in adjacency matrix, α layer interior joints
The element value of m and i in adjacency matrix;The convergence factor of each node in each layer is obtained, every layer of interior joint convergence factor is calculated
Average Cα:
(3) degree of the arbitrary node i in the arbitrary α layers in small echo bilayer complex network is calculated Wherein,Represent the element value of α layer interior joint i and j in adjacency matrix;The degree of each node in each layer is obtained, per layer is calculated
Average k of interior joint degreeα:
(4) eigenvector centrality of arbitrary α layers interior joint i of small echo bilayer complex networkRepresent the adjacency matrix A of α layers[α]The corresponding characteristic vector of dominant eigenvalue i-th element value;Obtain the eigenvector centrality of each node in each layer
Property value, calculate every layer of interior joint eigenvector centrality average Eα:
(5) average C of the node rendezvous coefficient in the every layer network for including flow pattern information is drawn respectivelyα, node degree average kα
With average E of node diagnostic vector centerαWith the graph of a relation of resolution ratio;Cycle motivation multi-electrode conductivity sensor is low to local
Oil content measures sensitive height, and cycle motivation multielectrode capacitance sensor is high to the high oil content measurement sensitivity in local, conduction capacity
The effective integration of multi-source metrical information realizes complementary two phase flow spatial flow infomation detection, and by research flow pattern information is included
Average C of the node rendezvous coefficient in every layer network per layer networkα, node degree average kαWith node diagnostic vector center
Average EαWith the graph of a relation of resolution ratio, average C of the node rendezvous coefficient in per layer of every layer network is studiedα, node degree it is equal
Value kαWith average E of node diagnostic vector centerαThe transformation of three indexs in flow pattern evolutionary process, realizes believing bimodal
The fusion of breath, indicates the internal motivation mechanism in flow pattern evolutionary process.
5. the two phase flow Multi-source Information Fusion method based on Wavelet Multiresolution Decomposition double-layer network described in a kind of claim 1 should
With, it is characterised in that using what is be made up of cycle motivation multi-electrode conductivity sensor and cycle motivation multielectrode capacitance sensor
Cycle motivation dual-modality sensor carries out vertical oil-water two-phase flow experiment;The proportioning of fixed oil phase and water phase, changes oil phase and water
The flow of phase is tested;Experimentation comprises the steps:
1) proportioning of fixed water phase and oil phase, toward vertical ascent pipeline a certain amount of water is passed through, then gradually logical in pipeline
Enter oil phase, after water-oil phase fully merges and gradually stablizes, passed using cycle motivation conductivity sensor and cycle motivation electric capacity
Sensor measures respectively multichannel signal, and records flow pattern with high-speed camera instrument;
2) after once collection terminates, the flow of oil phase and water phase is changed, by step 1) process continues collection, until matching somebody with somebody in fixed
All complete than lower designed operating mode;
3) change the proportioning of oil phase and water phase, repeat step 1 again) to step 2) process complete the measurement of this wheel, until setting
All operating modes of meter are all measured;
4) based on the multi-channel measurement signal for obtaining, for cycle motivation multi-electrode conductivity sensor and cycle motivation multi-electrode electricity
Hold the wavelet coefficient complex network that sensor is set up respectively under different resolution, for two-tier network obtains a wavelet multiresolution
Rate converging network;
5) figure energy and convergence factor entropy are calculated, different operating mode figure below energy, the aggregation for including flow pattern information is drawn respectively
Coefficient entropy portrays the internal motivation mechanism in the evolutionary process of different flow patterns with the graph of a relation of change resolution;
6) for two-tier network builds small echo bilayer complex network, node rendezvous Coefficient Mean, the node of each layer network are calculated
Degree average, node diagnostic vector center average, draw respectively node rendezvous Coefficient Mean, the node degree for including flow pattern information
The graph of a relation of average, 3 indexs of node diagnostic vector center average and change resolution, to study in flow pattern evolutionary process
In, the change of complex flow structure and internal motivation mechanism.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201610888616.6A CN106644375B (en) | 2016-10-11 | 2016-10-11 | Multi-source Information Fusion method and application based on Wavelet Multiresolution Decomposition bilayer complex network |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201610888616.6A CN106644375B (en) | 2016-10-11 | 2016-10-11 | Multi-source Information Fusion method and application based on Wavelet Multiresolution Decomposition bilayer complex network |
Publications (2)
Publication Number | Publication Date |
---|---|
CN106644375A true CN106644375A (en) | 2017-05-10 |
CN106644375B CN106644375B (en) | 2018-12-28 |
Family
ID=58856383
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN201610888616.6A Active CN106644375B (en) | 2016-10-11 | 2016-10-11 | Multi-source Information Fusion method and application based on Wavelet Multiresolution Decomposition bilayer complex network |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN106644375B (en) |
Cited By (1)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN112767296A (en) * | 2021-01-20 | 2021-05-07 | 山东大学 | Multi-sensor data fusion method and system based on time series imaging |
Citations (5)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
US4063153A (en) * | 1976-08-31 | 1977-12-13 | Auburn International, Inc. | Vapor liquid fraction determination |
CN103760197A (en) * | 2014-01-23 | 2014-04-30 | 天津大学 | Two-phase flow measuring system based on distributed conductivity sensor |
CN103776876A (en) * | 2014-01-23 | 2014-05-07 | 天津大学 | Structural parameter optimization method of distributed conductivity sensor |
CN104049001A (en) * | 2014-05-27 | 2014-09-17 | 天津大学 | Vertical oil-water phase content measurement and verification method based on frequency complex network |
CN104049000A (en) * | 2014-05-27 | 2014-09-17 | 天津大学 | Gas-liquid phase content measurement method based on modal migration complex network and verification method thereof |
-
2016
- 2016-10-11 CN CN201610888616.6A patent/CN106644375B/en active Active
Patent Citations (5)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
US4063153A (en) * | 1976-08-31 | 1977-12-13 | Auburn International, Inc. | Vapor liquid fraction determination |
CN103760197A (en) * | 2014-01-23 | 2014-04-30 | 天津大学 | Two-phase flow measuring system based on distributed conductivity sensor |
CN103776876A (en) * | 2014-01-23 | 2014-05-07 | 天津大学 | Structural parameter optimization method of distributed conductivity sensor |
CN104049001A (en) * | 2014-05-27 | 2014-09-17 | 天津大学 | Vertical oil-water phase content measurement and verification method based on frequency complex network |
CN104049000A (en) * | 2014-05-27 | 2014-09-17 | 天津大学 | Gas-liquid phase content measurement method based on modal migration complex network and verification method thereof |
Non-Patent Citations (2)
Title |
---|
ZHONG-KE GAO ET AL: "Complex network analysis of phase dynamics underlying oil-water two phase flows", 《SCIENTIFIC REPORTS》 * |
高忠科等: "两相流流型复杂网络社团结构及其统计特性", 《物理学报》 * |
Cited By (2)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN112767296A (en) * | 2021-01-20 | 2021-05-07 | 山东大学 | Multi-sensor data fusion method and system based on time series imaging |
CN112767296B (en) * | 2021-01-20 | 2022-06-28 | 山东大学 | Multi-sensor data fusion method and system based on time series imaging |
Also Published As
Publication number | Publication date |
---|---|
CN106644375B (en) | 2018-12-28 |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
CN106485325B (en) | Two phase flow multivariate information fusion method based on complex network and deep learning and application | |
CN106503800A (en) | Deep learning model based on complex network and the application in measurement signal analysis | |
CN106373397B (en) | Remote sensing images road situation analysis method based on fuzzy neural network | |
CN106023156B (en) | The method for registering of point cloud model and CAD model based on detection feature | |
CN104089985B (en) | Multiphase flow visual testing method based on electricity Yu ultrasonic sensitive principle | |
CN103177450B (en) | A kind of image scene segmentation of Component-Based Development centralized procurement sample combines method for solving with layering | |
CN102116754B (en) | Multiphase flow measurement method based on double-section impedance type long waist cone sensor | |
CN107563349A (en) | A kind of Population size estimation method based on VGGNet | |
CN107437339A (en) | Variable information advices plate control method for coordinating and system under a kind of information guidance | |
CN104048997B (en) | Profit phase content based on multiple phase spatial complex network is measured and verification method | |
CN106650929A (en) | Recursive-graph-based deep learning model and its application in oil-water phase rate measurement | |
CN105120517B (en) | Indoor WLAN signal plan structure and localization method based on multi-dimentional scale MDS analyses | |
CN107967489A (en) | A kind of method for detecting abnormality and system | |
CN108304887A (en) | Naive Bayesian data processing system and method based on the synthesis of minority class sample | |
CN110533103A (en) | A kind of lightweight wisp object detection method and system | |
CN106127263A (en) | The human brain magnetic resonance image (MRI) classifying identification method extracted based on three-dimensional feature and system | |
CN110584596B (en) | Sleep stage classification method based on dual-input convolutional neural network and application thereof | |
CN108765961A (en) | A kind of floating car data processing method based on modified amplitude limit average filter | |
CN105869354B (en) | A kind of Falls Among Old People detection method based on attractor propagation algorithm | |
CN109582754A (en) | The method for carrying out urban subject functional areas central detector using POI data | |
CN106650818A (en) | Resting state function magnetic resonance image data classification method based on high-order super network | |
CN105843829A (en) | Big data credibility measurement method based on layering model | |
CN106055580A (en) | Radviz-based fuzzy clustering result visualization method | |
CN106121622A (en) | A kind of Multiple faults diagnosis approach of Dlagnosis of Sucker Rod Pumping Well based on indicator card | |
CN104049001B (en) | Vertical profit phase content based on frequency complex network is measured and verification method |
Legal Events
Date | Code | Title | Description |
---|---|---|---|
PB01 | Publication | ||
PB01 | Publication | ||
SE01 | Entry into force of request for substantive examination | ||
SE01 | Entry into force of request for substantive examination | ||
GR01 | Patent grant | ||
GR01 | Patent grant |