CN105139025A - Non-linear analysis method based online intelligent identification method for flow pattern of gas-solid fluidized bed - Google Patents
Non-linear analysis method based online intelligent identification method for flow pattern of gas-solid fluidized bed Download PDFInfo
- Publication number
- CN105139025A CN105139025A CN201510461151.1A CN201510461151A CN105139025A CN 105139025 A CN105139025 A CN 105139025A CN 201510461151 A CN201510461151 A CN 201510461151A CN 105139025 A CN105139025 A CN 105139025A
- Authority
- CN
- China
- Prior art keywords
- cluster
- gas
- fluidized bed
- solid fluidized
- pressure fluctuation
- 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
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/23—Clustering techniques
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V10/00—Arrangements for image or video recognition or understanding
- G06V10/70—Arrangements for image or video recognition or understanding using pattern recognition or machine learning
- G06V10/74—Image or video pattern matching; Proximity measures in feature spaces
- G06V10/75—Organisation of the matching processes, e.g. simultaneous or sequential comparisons of image or video features; Coarse-fine approaches, e.g. multi-scale approaches; using context analysis; Selection of dictionaries
Landscapes
- Engineering & Computer Science (AREA)
- Theoretical Computer Science (AREA)
- Data Mining & Analysis (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Evolutionary Computation (AREA)
- Artificial Intelligence (AREA)
- General Physics & Mathematics (AREA)
- Physics & Mathematics (AREA)
- General Engineering & Computer Science (AREA)
- Evolutionary Biology (AREA)
- Bioinformatics & Computational Biology (AREA)
- Bioinformatics & Cheminformatics (AREA)
- Life Sciences & Earth Sciences (AREA)
- Health & Medical Sciences (AREA)
- Computing Systems (AREA)
- Databases & Information Systems (AREA)
- General Health & Medical Sciences (AREA)
- Medical Informatics (AREA)
- Software Systems (AREA)
- Multimedia (AREA)
- Fluidized-Bed Combustion And Resonant Combustion (AREA)
Abstract
The present invention discloses a non-linear analysis method based online intelligent identification method for a flow pattern of a gas-solid fluidized bed. The method comprises: performing sampling and analysis of a pressure pulsation signal on the gas-solid fluidized bed and extracting a feature value; performing objective clustering by means of a fuzzy cluster algorithm according to the feature value and the pressure pulsation signal; and establishing and training a flow pattern identification system according to the pressure pulsation signal, the feature value and an objective clustering result, and embedding the system into a computer to implement online intelligent identification of the flow pattern of the gas-solid fluidized bed. The method can avoid the influence of subjective factors on flow pattern identification accuracy, and perform dynamic analysis and flow pattern identification on a transient state parameter.
Description
Technical field
The present invention relates to a kind of gas-solid fluidized bed flow type identification method, belong to gas-solid fluidized bed technical field.
Background technology
Gas-solid fluidized bedly to be used widely at multiple industrial circle such as chemical industry, the energy because it has good flowing heat transfer characteristic.Measurement and the flow pattern of gas-solid fluidized bed parameter are closely related, and how to carry out in real time accurately identification to gas-solid fluidized bed flow pattern is an important topic of current gas-solid fluidized bed area research.
At present direct observational method and Parameter analysis method are mainly contained to the method for gas-solid fluidized bed Flow Regime Ecognition.Direct observational method needs reaction unit visual or utilize image documentation equipment to carry out image acquisition, and the gas-solid fluidized bed apparatus in commercial Application, because of the restriction by rapidoprint, is difficult to adopt direct observational method to carry out Flow Regime Ecognition accurately.Parameter analysis method needs to obtain relevant parameter by the method such as capacitance method, optical fiber probe mensuration, pressure fluctuation signal collection, then to process acquisition parameter in conjunction with existing flow pattern decision criteria and method and analyzes.These two kinds of methods all cannot avoid the impact of subjective factor convection identification accuracy, are also difficult to carry out performance analysis and Flow Regime Ecognition to instantaneous state parameter.
Summary of the invention
Object of the present invention: in order to overcome the deficiencies in the prior art, the invention provides a kind of gas-solid fluidized bed flow pattern online intelligent recognition method based on nonlinear analysis method, this method can avoid the impact of subjective factor convection identification accuracy, can carry out performance analysis and Flow Regime Ecognition to instantaneous state parameter simultaneously.
Technical scheme: for achieving the above object, the technical solution used in the present invention is:
Based on a gas-solid fluidized bed flow pattern online intelligent recognition method for nonlinear analysis method, by carrying out pressure fluctuation signal sampling analysis to gas-solid fluidized bed and extract eigenwert; Objective cluster is carried out by fuzzy clustering algorithm according to pressure fluctuation signal characteristic of correspondence value; Set up according to the result of pressure fluctuation signal, eigenwert and objective cluster and training Flow Regime Ecognition system, then this system embedment computing machine is realized the online intelligent recognition of gas-solid fluidized bed flow pattern.
Specifically comprise the following steps:
Step 1, gathers the carrying out of gas-solid fluidized bed middle pressure fluctuation signal;
Step 2, is processed the pressure fluctuation signal that step 1 gathers by nonlinear analysis method, extracts eigenwert;
Step 3, introduces fuzzy c clustering algorithm by the eigenwert extracted in step 2, carries out objective cluster to the pressure fluctuation signal gathered in step 1;
Step 4, introduces neural network by the cluster result of step 3 and carries out the building of Flow Regime Ecognition system, trains;
Step 5, is loaded into Flow Regime Ecognition and output system by the Flow Regime Ecognition system trained, realizes gas-solid fluidized bed objective, accurate, real-time Flow Regime Ecognition.
The acquisition method of pressure fluctuation signal in described step 1: in the gas-solid fluidized bed middle collection utilizing differential pressure transmitter (7) to carry out pressure fluctuation signal; The signal that differential pressure transmitter (7) gathers after the conversion of A/D converter (8) at the characteristics extraction carrying out step 2.
Nonlinear analysis method in described step 2 is Hilbert-Huang transform analysis method.
Described Hilbert-Huang transform analysis method comprises empirical mode decomposition and Hilbert transform, specifically comprises the following steps:
Step 21, first itself intrinsic gang's Intrinsic mode function is successively decomposed from high to low by the frequency component of empirical mode decomposition to the pressure fluctuation signal pressing force fluctuating signal gathered, this Intrinsic mode function embodies the characteristic information of pressure fluctuation signal, and the most HFS of first Intrinsic mode function representative pressure fluctuating signal;
Step 22, carries out Hilbert transform to each Intrinsic mode function, obtains its instantaneous frequency and instantaneous amplitude, and then obtains its eigenwert.
In described step 21, Intrinsic mode function number is more than one, and Intrinsic mode function is designated as IMF, is defined as follows by IMF component energy:
In formula, E is the energy of IMF component, and n is sample total amount, a
it () is the instantaneous amplitude that i-th IMF is corresponding;
In described step 22, eigenwert refers to and analyzes the pressure difference signal of different bed material in different gas speed current downflow, calculates the energy size of each rank IMF component, according to the frequency range of each rank IMF component, is divided into high band E
h, Mid Frequency E
mwith low-frequency range E
l, E
h, E
m, E
lbe extracted eigenwert.
Gas-solid fluidized bed flow pattern online intelligent recognition method based on nonlinear analysis method according to claim 6, is characterized in that: the pressure fluctuation signal gathered in step 1 in described step 3 carries out the method for objective cluster:
Step 31, eigenwert, cluster numbers, cluster centre and the pressure fluctuation signal extracted according to the pressure fluctuation signal gathered in step 1, step 2 and this cluster centre degree of membership set up cluster object module;
Step 32, according to the pressure fluctuation signal gathered in step 1, step 2 extract eigenwert and choose initial cluster center, calculated the cluster desired value of this initial cluster center by the cluster object module of step 3; The Euclidean distance of the eigenwert that calculation procedure 2 is extracted and initial cluster center, pressure fluctuation signal is included in corresponding class, obtain new cluster centre according to the sample average of each class, calculated the cluster desired value of this new cluster centre by the cluster object module of step 3; And judge the cluster desired value of initial cluster center and the cluster desired value of new cluster centre, if these two cluster desired values are variant, again calculate the Euclidean distance of each eigenwert and new cluster centre, then reclassify, the cluster desired value of this new cluster centre is calculated by the cluster object module of step 3, and judge the cluster desired value of previous cluster centre and the cluster desired value of a rear cluster centre, until these two cluster desired values no longer change, the cluster of complete paired samples.
The cluster object module set up in described step 31:
Wherein, J is cluster objective function, X={x
1, x
2x
nbe sample, each x
i=[x
i1, x
i2... x
in] all including n dimensional feature value, c is cluster numbers, chooses the individual different x of c
ias initial cluster center v
k, k=1,2 ..., c, U is degree of membership, and V is cluster centre, and m is weight coefficient,
d
ki＝||x
i-v
k||，d
ji＝||x
i-v
j||。
In described step 21, Intrinsic mode function number is determined by frequency component, is generally 7 or 8, is respectively IMF1-IMF7 or IMF1-IMF8; The recognition system trained is connected with gas-solid fluidized bed pressure fluctuation signal acquisition system by described step 5, carries out data processing by the direct calling data of program, show in output system this moment gas-solid fluidized bed in flow pattern state.
Beneficial effect: a kind of gas-solid fluidized bed flow pattern online intelligent recognition method based on nonlinear analysis method provided by the invention, compared to existing technology, has following beneficial effect:
By carrying out pressure fluctuation signal sampling analysis to gas-solid fluidized bed and extract eigenwert; Objective cluster is carried out by fuzzy clustering algorithm according to eigenwert; Set up according to the result of pressure fluctuation signal, eigenwert and objective cluster and training Flow Regime Ecognition system, again this system embedment computing machine is realized the online intelligent recognition of gas-solid fluidized bed flow pattern, therefore this method can avoid the impact of subjective factor convection identification accuracy, can carry out performance analysis and Flow Regime Ecognition to instantaneous state parameter simultaneously.
Accompanying drawing explanation
Fig. 1 is gas-solid fluidized bed flow pattern ONLINE RECOGNITION system equipment system diagram of the present invention.
Fig. 2 is the IMF component map obtained after example of the present invention carries out data processing.
In Fig. 1,1 is air compressor; 2 is flow control systems; 3 is gas-solid fluidized bed bodies; 4 is air compartments; 5 is distribution grids; 6 is pressure taps; 7 is differential pressure transmitters; 8 is A/D converters; 9 is computing machines.
Embodiment
By reference to the accompanying drawings a detailed introduction is done to the present invention, typical case of the present invention but non-limiting example is as follows:
Based on a gas-solid fluidized bed flow pattern online intelligent recognition method for nonlinear analysis method, by carrying out pressure fluctuation signal sampling analysis to gas-solid fluidized bed and extract eigenwert; Objective cluster is carried out by fuzzy clustering algorithm according to eigenwert; Set up according to the result of pressure fluctuation signal, eigenwert and objective cluster and training Flow Regime Ecognition system, then this system embedment computing machine is realized the online intelligent recognition of gas-solid fluidized bed flow pattern.
Wherein, recognition system as shown in Figure 1: comprise air compressor 1, flow regulation is 2, gas-solid fluidized bed body 3, air compartment 4, distribution grid 5, pressure tap 6, differential pressure transporting 7, A/D conversion 8, calculate 9; Described air compartment 4 is arranged at the bottom of gas-solid fluidized bed body 3, and described distribution grid 5 is arranged on air compartment 4, the pressure fluctuation signal of gas-solid fluidized bed body 3 is measured on one drive test buret road of described differential pressure transporting 7 by pressure tap 6, another drive test buret road is arranged in air compartment 4, and between gas-solid fluidized bed body 3 and distribution grid 5, and described differential pressure transporting 7 carries out signal conversion with calculating by A/D conversion 89 is connected, and the air of described air compressor 1 compression is 2 to enter in air compartment 4 by flow regulation.
Recognition methods of the present invention specifically comprises the following steps:
Step 1, gathers the carrying out of gas-solid fluidized bed middle pressure fluctuation signal; By in the gas-solid fluidized bed middle collection utilizing differential pressure transmitter (7) to carry out pressure fluctuation signal; The signal that differential pressure transmitter (7) gathers after the conversion of A/D converter (8) at the characteristics extraction carrying out step 2.
Step 2, is processed the pressure fluctuation signal that step 1 gathers by nonlinear analysis method, extracts eigenwert.Described nonlinear analysis method is Hilbert-Huang transform analysis method, mainly comprise two aspects, empirical mode decomposition (EmpiricalModeDecomposition is called for short EMD) and Hilbert transform (HilbertTransportation is called for short HT).Wherein EMD is core, first adopt EMD method, from pressure reduction fluctuating signal, extract itself intrinsic gang's Intrinsic mode function (IntrinsicModeFunction is called for short IMF), its quantity is limited, includes real physical significance.EMD extracts IMF from original signal, and the IMF obtained embodies the characteristic information of original signal, and first IMF represents the most HFS of original signal.Then Hilbert transform is carried out to each IMF, obtain its instantaneous frequency and instantaneous amplitude.EMD decomposition is carried out to gathered pressure reduction fluctuating signal, successively the frequency component of pressure fluctuation signal is decomposed out from high to low, extract 7 or 8, be respectively IMF1-IMF7 or IMF1-IMF8.For analyzing the Changing Pattern of different range instantaneous frequency better, each IMF component energy is defined as follows:
In formula, a
it () is the instantaneous amplitude that i-th IMF is corresponding.
The pressure difference signal of different bed material in different gas speed current downflow is analyzed, calculates the energy size of each rank IMF component, according to the frequency range of each rank IMF component, be divided into high band E
h(E1,2,3), Mid Frequency E
m(E4 ~ 6) and low-frequency range E
l(E7,8), E
h, E
m, E
lthe i.e. eigenwert of step 2 by extracting after nonlinear analysis method processing pressure fluctuating signal.
Step 3, introduces fuzzy c clustering algorithm by the eigenwert extracted in step 2, carries out objective cluster to the pressure fluctuation signal gathered in step 1.Cluster is carried out to the eigenwert that step 2 is extracted, the concrete flow pattern that contrast gas is fast and image sampling determination cluster is corresponding.
The pressure fluctuation signal gathered in step 1 in described step 3 carries out the method for objective cluster:
Step 31, eigenwert, cluster numbers, cluster centre and the pressure fluctuation signal characteristic of correspondence value extracted according to the pressure fluctuation signal gathered in step 1, step 2 and this cluster centre degree of membership set up cluster object module;
The cluster object module set up in described step 31
Wherein, J is cluster objective function, X={x
1, x
2x
nbe sample, each x
i=[x
i1, x
i2... x
in] all including n dimensional feature value, c is cluster numbers, chooses the individual different x of c
ias initial cluster center v
k, k=1,2 ..., c, U is degree of membership, and V is cluster centre, and m is weight coefficient,
μ
ki, v
icalculated by following formula and determine, d
ki=|| x
i-v
k||, d
ji=|| x
i-v
j||:
Step 32, according to the pressure fluctuation signal gathered in step 1, step 2 extract eigenwert and choose initial cluster center, calculated the cluster desired value of this initial cluster center by the cluster object module of step 3; The Euclidean distance of the eigenwert that calculation procedure 2 is extracted and initial cluster center, pressure fluctuation signal is included in corresponding class, obtain new cluster centre according to the sample average of each class, calculated the cluster desired value of this new cluster centre by the cluster object module of step 3; And judge the cluster desired value of initial cluster center and the cluster desired value of new cluster centre, if these two cluster desired values are variant, again calculate the Euclidean distance of each eigenwert and new cluster centre, then reclassify, the cluster desired value of this new cluster centre is calculated by the cluster object module of step 3, and judge the cluster desired value of previous cluster centre and the cluster desired value of a rear cluster centre, until these two cluster desired values no longer change, the cluster of complete paired samples, by X={x
1, x
2x
nin x
ithe center of being grouped into is v
kall kinds of in.
Step 4, introduces neural network by the cluster result of step 3 and carries out the building of Flow Regime Ecognition system, trains.
The eigenwert of cluster is carried out building and training of Flow Regime Ecognition system as the training data of neural network.Neural network algorithm used in step 4 is made up of input layer node, output layer neuron node and hidden layer neuron node.The step of standard genetic algorithm is as follows:
1) number, filial generation number individual in population is determined;
2) individual fitness is calculated;
3) select according to ideal adaptation degree or eliminate individual;
4) according to certain intersection, mutation probability and cross method, new individuality is generated;
5) population of a new generation is produced by crossover and mutation;
6) reach setting filial generation number and stop algorithm.
Step 5, is loaded into Flow Regime Ecognition and output system by the Flow Regime Ecognition system trained, realizes gas-solid fluidized bed objective, accurate, real-time Flow Regime Ecognition.
Above-mentioned steps 1 to step 4 is training and the foundation of convection recognition system, the recognition system trained is connected with gas-solid fluidized bed pressure fluctuation signal acquisition system by step 5, carry out data processing by the direct calling data of program, show in output system this moment gas-solid fluidized bed in flow pattern state.
Example
Step one: the collection of pressure fluctuation signal, as shown in Figure 1, gas-solid fluidized bed 3 carry out data acquisition by differential pressure transmitter 7 and A/D digital signal conversion collector 8, and institute's image data record is in the computer 9; Also recording instrument without paper or other signals conversion acquisition system can be adopted.Ensure that sample frequency is 100Hz, adopt the cycle more than 10 seconds.Respectively pressure fluctuation signal collection is carried out to the operating mode of different gas speed and different thickness of bed layer, in this example, adopts the mixture of silica sand and pelletized biomass different proportion, sample 120 groups altogether.
Step 2: processed the pressure fluctuation signal that step one gathers by nonlinear analysis method, extracts eigenwert: process sampled data according to preceding method, obtain the IMF energy interval value corresponding to each group of data.Often organize the corresponding three-dimensional array of pressure fluctuation signal in this example, this array is the energy value of IMF low-frequency range, IMF Mid Frequency and IMF high band.Fig. 2 is the IMF component map of a certain group of data.
Step 3: fixed bed, bubbling bed, joint can be divided into gush bed, turbulent bed and fast bed according to gas-solid fluidized bed classical taxonomy mode, fast bed generally realizes in recirculating fluidized bed, limit by fluidized bed size in this example, only front 4 kinds of flow patterns are classified, if cluster numbers is 4.Sampled data set is 120 groups, random selecting 4 group data set is as initial cluster center, calculate each the group eigenwert of data and Euclidean distance of cluster centre, data are included in the class corresponding to the minimum cluster centre of Euclidean distance, then carrying out average by the eigenwert of all data to each class asks calculation to obtain new cluster centre, again calculate the eigenwert of each group of data and the Euclidean distance of cluster centre and data are reclassified, until cluster centre no longer changes, cluster completes.The flow velocity corresponding according to sampled data, can by cluster and flow pattern one_to_one corresponding.In this example, 120 groups of data clusters results are: 31 groups, fixed bed, bubbling bed 38 groups, joint gush bed 24 groups, turbulent flow group 27 groups.
Step 4: cluster result step 3 obtained is as population training data.If Population Size is 100, filial generation number is 200, and crossover probability is 0.5, and mutation probability is 0.001.Neural network after training is verified by all the other 20 groups of data having neither part nor lot in training, and have 18 groups of identifications correct, accuracy rate is 90%.The selection of parameter reasonably can adjust according to actual conditions, and parameter is determined after test of many times herein.
Step 5: the network trained carries out integrated as a module of ONLINE RECOGNITION system and data processing section, and again access in gas-solid fluidized bed system and run, the pressure fluctuation signal of every 10 seconds carries out online Flow Regime Ecognition as one group of data.Recognition result is directly shown by the display of computing machine 10.
Claims (9)
1. based on a gas-solid fluidized bed flow pattern online intelligent recognition method for nonlinear analysis method, it is characterized in that: by carrying out pressure fluctuation signal sampling analysis to gas-solid fluidized bed and extract eigenwert; Objective cluster is carried out by fuzzy clustering algorithm according to pressure fluctuation signal characteristic of correspondence value; Set up according to the result of pressure fluctuation signal, eigenwert and objective cluster and training Flow Regime Ecognition system, then this system embedment computing machine is realized the online intelligent recognition of gas-solid fluidized bed flow pattern.
2. the gas-solid fluidized bed flow pattern online intelligent recognition method based on nonlinear analysis method according to claim 1, is characterized in that, comprise the following steps:
Step 1, gathers the carrying out of gas-solid fluidized bed middle pressure fluctuation signal;
Step 2, is processed the pressure fluctuation signal that step 1 gathers by nonlinear analysis method, extracts eigenwert;
Step 3, introduces fuzzy c clustering algorithm by the eigenwert extracted in step 2, carries out objective cluster to the pressure fluctuation signal gathered in step 1;
Step 4, introduces neural network by the cluster result of the eigenwert of step 3 and carries out the building of Flow Regime Ecognition system, trains;
Step 5, is loaded into Flow Regime Ecognition and output system by the Flow Regime Ecognition system trained, realizes gas-solid fluidized bed objective, accurate, real-time Flow Regime Ecognition.
3. the gas-solid fluidized bed flow pattern online intelligent recognition method based on nonlinear analysis method according to claim 2, is characterized in that: the acquisition method of pressure fluctuation signal in described step 1: in the gas-solid fluidized bed middle collection utilizing differential pressure transmitter (7) to carry out pressure fluctuation signal; The signal that differential pressure transmitter (7) gathers after the conversion of A/D converter (8) at the characteristics extraction carrying out step 2.
4. the gas-solid fluidized bed flow pattern online intelligent recognition method based on nonlinear analysis method according to claim 3, is characterized in that: the nonlinear analysis method in described step 2 is Hilbert-Huang transform analysis method.
5. the gas-solid fluidized bed flow pattern online intelligent recognition method based on nonlinear analysis method according to claim 4, it is characterized in that: described Hilbert-Huang transform analysis method comprises empirical mode decomposition and Hilbert transform, specifically comprises the following steps:
Step 21, first itself intrinsic gang's Intrinsic mode function is successively decomposed from high to low by the frequency component of empirical mode decomposition to the pressure fluctuation signal pressing force fluctuating signal gathered, this Intrinsic mode function embodies the characteristic information of pressure fluctuation signal, and the most HFS of first Intrinsic mode function representative pressure fluctuating signal;
Step 22, carries out Hilbert transform to each Intrinsic mode function, obtains its instantaneous frequency and instantaneous amplitude, and then obtains its eigenwert.
6. the gas-solid fluidized bed flow pattern online intelligent recognition method based on nonlinear analysis method according to claim 5, it is characterized in that: in described step 21, Intrinsic mode function number is more than one, Intrinsic mode function is designated as IMF, is defined as follows by IMF component energy:
In formula, E is the energy of IMF component, and n is sample total amount, a
it () is the instantaneous amplitude that i-th IMF is corresponding;
In described step 22, eigenwert refers to and analyzes the pressure difference signal of different bed material in different gas speed current downflow, calculates the energy size of each rank IMF component, according to the frequency range of each rank IMF component, is divided into high band E
h, Mid Frequency E
mwith low-frequency range E
l, E
h, E
m, E
lbe extracted eigenwert.
7. the gas-solid fluidized bed flow pattern online intelligent recognition method based on nonlinear analysis method according to claim 6, is characterized in that: the pressure fluctuation signal gathered in step 1 in described step 3 carries out the method for objective cluster:
Step 31, eigenwert, cluster numbers, cluster centre and the pressure fluctuation signal characteristic of correspondence value extracted according to the pressure fluctuation signal gathered in step 1, step 2 and this cluster centre degree of membership set up cluster object module;
Step 32, according to the pressure fluctuation signal gathered in step 1, step 2 extract eigenwert and choose initial cluster center, calculated the cluster desired value of this initial cluster center by the cluster object module of step 3; The Euclidean distance of the eigenwert that calculation procedure 2 is extracted and initial cluster center, pressure fluctuation signal is included in corresponding class, obtain new cluster centre according to the sample average of each class, calculated the cluster desired value of this new cluster centre by the cluster object module of step 3; And judge the cluster desired value of initial cluster center and the cluster desired value of new cluster centre, if these two cluster desired values are variant, again calculate the Euclidean distance of each eigenwert and new cluster centre, then reclassify, the cluster desired value of this new cluster centre is calculated by the cluster object module of step 3, and judge the cluster desired value of previous cluster centre and the cluster desired value of a rear cluster centre, until these two cluster desired values no longer change, the cluster of complete paired samples.
8. the gas-solid fluidized bed flow pattern online intelligent recognition method based on nonlinear analysis method according to claim 7, is characterized in that: the cluster object module set up in described step 31:
Wherein, J is cluster objective function, X={x
1, x
2x
nbe sample, each x
i=[x
i1, x
i2... x
in] all including n dimensional feature value, c is cluster numbers, chooses the individual different x of c
ias initial cluster center v
k, k=1,2 ..., c, U is degree of membership, and V is cluster centre, and m is weight coefficient,
i=1,2 ..., n, μ
ki∈ (0,1);
1≤k≤c, 1≤i≤N;
1≤i≤c, d
ki=|| x
i-v
k||, d
ji=|| x
i-v
j||.
9. the gas-solid fluidized bed flow pattern online intelligent recognition method based on nonlinear analysis method according to claim 8, it is characterized in that: in described step 21, Intrinsic mode function number is determined by frequency component, be generally 7 or 8, be respectively IMF1-IMF7 or IMF1-IMF8; The recognition system trained is connected with gas-solid fluidized bed pressure fluctuation signal acquisition system by described step 5, carries out data processing by the direct calling data of program, show in output system this moment gas-solid fluidized bed in flow pattern state.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201510461151.1A CN105139025B (en) | 2015-07-30 | 2015-07-30 | Gas-solid fluidized bed flow pattern online intelligent recognition method based on nonlinear analysis method |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201510461151.1A CN105139025B (en) | 2015-07-30 | 2015-07-30 | Gas-solid fluidized bed flow pattern online intelligent recognition method based on nonlinear analysis method |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN105139025A true CN105139025A (en) | 2015-12-09 |
| CN105139025B CN105139025B (en) | 2019-03-05 |
Family
ID=54724368
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201510461151.1A Active CN105139025B (en) | 2015-07-30 | 2015-07-30 | Gas-solid fluidized bed flow pattern online intelligent recognition method based on nonlinear analysis method |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN105139025B (en) |
Cited By (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN111829929A (en) * | 2020-07-30 | 2020-10-27 | 安徽工业大学 | Method for identifying binary particle separation or mixing state in fluidized bed |
| CN114417647A (en) * | 2021-11-19 | 2022-04-29 | 西南石油大学 | Reciprocating compressor dynamic compression flow domain system airflow pulsation calculation method |
Citations (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US5045240A (en) * | 1989-05-01 | 1991-09-03 | Westinghouse Electric Corp. | Contaminated soil restoration method |
| US20100158359A1 (en) * | 2008-12-18 | 2010-06-24 | Yue Qiao | Image data clustering and color conversion |
| CN102338732A (en) * | 2011-07-06 | 2012-02-01 | 浙江大学 | Method for measuring fluidization parameter of gas-solid fluidized bed |
| CN103646354A (en) * | 2013-11-28 | 2014-03-19 | 国家电网公司 | Effective index FCM and RBF neural network-based substation load characteristic categorization method |
| CN103742901A (en) * | 2013-12-24 | 2014-04-23 | 广州市恒力安全检测技术有限公司 | Method for determining consumption difference analysis optimized target value system of circulating fluidized bed unit |
-
2015
- 2015-07-30 CN CN201510461151.1A patent/CN105139025B/en active Active
Patent Citations (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US5045240A (en) * | 1989-05-01 | 1991-09-03 | Westinghouse Electric Corp. | Contaminated soil restoration method |
| US20100158359A1 (en) * | 2008-12-18 | 2010-06-24 | Yue Qiao | Image data clustering and color conversion |
| CN102338732A (en) * | 2011-07-06 | 2012-02-01 | 浙江大学 | Method for measuring fluidization parameter of gas-solid fluidized bed |
| CN103646354A (en) * | 2013-11-28 | 2014-03-19 | 国家电网公司 | Effective index FCM and RBF neural network-based substation load characteristic categorization method |
| CN103742901A (en) * | 2013-12-24 | 2014-04-23 | 广州市恒力安全检测技术有限公司 | Method for determining consumption difference analysis optimized target value system of circulating fluidized bed unit |
Non-Patent Citations (4)
| Title |
|---|
| 周树孝: "小波分析和神经网络理论及其在气-固流化床中的应用", 《中国优秀博硕士学位论文全文数据库(硕士)工程科技Ⅰ辑》 * |
| 王晓萍: "基于现代非线性信息处理技术的气固流化床流型识别方法与实验研究", 《中国优秀博硕士学位论文全文数据库(博士) 工程科技Ⅰ辑》 * |
| 范九伦，赵凤，吴青等: "《模式识别导论》", 31 December 2012 * |
| 钟文琪 等: "喷动流化床流动结构的SHANNON信息熵模糊聚类分析", 《中国电机工程学报》 * |
Cited By (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN111829929A (en) * | 2020-07-30 | 2020-10-27 | 安徽工业大学 | Method for identifying binary particle separation or mixing state in fluidized bed |
| CN114417647A (en) * | 2021-11-19 | 2022-04-29 | 西南石油大学 | Reciprocating compressor dynamic compression flow domain system airflow pulsation calculation method |
| CN114417647B (en) * | 2021-11-19 | 2024-05-14 | 西南石油大学 | Dynamic compression river basin system airflow pulsation calculation method of reciprocating compressor |
Also Published As
| Publication number | Publication date |
|---|---|
| CN105139025B (en) | 2019-03-05 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN109685314B (en) | Non-intrusive load decomposition method and system based on long-term and short-term memory network | |
| CN101630376A (en) | Soft-sensing modeling method and soft meter of multi-model neural network in biological fermentation process | |
| CN106022377A (en) | Online prediction method for iron ore sintering bed permeability states | |
| CN106933778A (en) | A kind of wind power combination forecasting method based on climbing affair character identification | |
| CN103065202B (en) | Wind power plant ultrashort term wind speed prediction method based on combination kernel function | |
| CN102661759B (en) | Method for identifying and predicting nonlinear multivariable key parameters of circulating fluidized bed boiler | |
| CN107480705A (en) | A kind of running water pipe leakage detection method | |
| CN104751229A (en) | Bearing fault diagnosis method capable of recovering missing data of back propagation neural network estimation values | |
| CN110824586B (en) | Rainfall prediction method based on improved decision tree algorithm | |
| CN102693452A (en) | Multiple-model soft-measuring method based on semi-supervised regression learning | |
| CN104462850A (en) | Multi-stage batch process soft measurement method based on fuzzy gauss hybrid model | |
| CN110298085A (en) | Analog-circuit fault diagnosis method based on XGBoost and random forests algorithm | |
| CN103942422B (en) | Granular-computation-based long-term prediction method for converter gas holder positions in metallurgy industry | |
| CN106874950A (en) | A kind of method for identifying and classifying of transient power quality recorder data | |
| CN110619084B (en) | Method for recommending books according to borrowing behaviors of library readers | |
| CN112446091A (en) | Artificial neural network-based pulsating pressure prediction method | |
| CN103226728B (en) | High density polyethylene polymerization cascade course of reaction Intelligent Measurement and yield optimization method | |
| CN103745271A (en) | Forecasting method for induction thermal deposition calcium-phosphate coating process on basis of neural network | |
| CN107657274A (en) | A kind of y-bend SVM tree unbalanced data industry Fault Classifications based on k means | |
| CN107341359A (en) | Stalk fermentation produces the flexible measurement method of ethanol process key parameters | |
| CN114429152A (en) | Rolling bearing fault diagnosis method based on dynamic index antagonism self-adaption | |
| CN116127409A (en) | Non-invasive load identification method based on Gram angle difference field feature fusion | |
| CN109921426A (en) | Wind-electricity integration system probability load flow calculation method based on CV-KDE | |
| CN105139025A (en) | Non-linear analysis method based online intelligent identification method for flow pattern of gas-solid fluidized bed | |
| CN109376903B (en) | PM2.5 concentration value prediction method based on game neural network |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| C06 | Publication | ||
| PB01 | Publication | ||
| C10 | Entry into substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant |