CN104034794A - Extreme learning machine-based pipeline magnetic flux leakage defect detection method - Google Patents
Extreme learning machine-based pipeline magnetic flux leakage defect detection method Download PDFInfo
- Publication number
- CN104034794A CN104034794A CN201410261496.8A CN201410261496A CN104034794A CN 104034794 A CN104034794 A CN 104034794A CN 201410261496 A CN201410261496 A CN 201410261496A CN 104034794 A CN104034794 A CN 104034794A
- Authority
- CN
- China
- Prior art keywords
- pipeline
- magnetic flux
- defect
- learning machine
- flux leakage
- 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
Landscapes
- Investigating Or Analyzing Materials By The Use Of Magnetic Means (AREA)
Abstract
The invention relates to an extreme learning machine-based pipeline magnetic flux leakage defect detection method. The method comprises the following steps: establishing an extreme learning machine model according to data of the length, the width and the depth of a known pipeline magnetic flux leakage defect and a magnetic flux leakage signal waveform characteristic value; training the data of the length, the width and the depth of the known pipeline magnetic flux leakage defect in sample data, wherein the data is taken as the input of the model; selecting the number of nodes of a hidden layer by a cut-and-trial method; and calculating an output matrix and an output weight value of the hidden layer, wherein the magnetic flux leakage signal waveform characteristic value is taken as the output of the model; when a pipeline suffers from magnetic flux leakage, acquiring the waveform of a magnetic flux leakage signal of an unknown magnetic flux leakage shape, and performing pipeline magnetic flux leakage defect detection by the extreme learning machine model. According to the method, the pipeline defect shape is subjected to intelligent inversion by the extreme learning machine model; the method has the advantages of high learning speed, high generalization performance and the like; the shape of the defect can be constructed quickly and accurately by virtue of the waveform of the detected defect, so that the severity of the defect is learnt, a pipeline risk can be foreknown, and the leakage of the pipeline is prevented.
Description
Technical field
The invention belongs to pipe detection technical field, be specifically related to a kind of pipe leakage defect inspection method based on extreme learning machine.
Background technology
In the face of defect of pipeline research field, defect of pipeline shape data amount is large, requires the fast feature of detection speed.For these difficulties, the finite element algorithm extensively adopting at present, finite element algorithm is a kind of high-effect, conventional computing method, and it can, by the discrete set that turns to the cell cube of several limited sizes continuously, can be applied in the described all kinds of physical fields of any differential equation.But finite element algorithm is in the face of the larger feature of data volume, and speed is slow, and elapsed time length can not meet the demands.In addition also has support vector machine, support vector machine (Support Vector Machine, SVM) be a kind of outstanding machine learning algorithm occurring in recent years, it is that the structure risk minimum principle and the VC that are based upon statistical theory tie up on theoretical basis, according to limited sample data information, between the complicacy (i.e. the study precision to specific training sample data) of model and learning ability (identifying error-free the ability of arbitrary sample data), seek the best approach, in the hope of obtaining best Generalization Ability.A lot of advantages of support vector machine are embodied in it and solve in high dimensional pattern, non-linear etc. pattern recognition problem.Now, as a kind of outstanding machine learning algorithm, it has become the study hotspot of international machine learning, artificial intelligence field.But support vector machine, for the large feature of defect of pipeline data volume, can highlight its speed slow, the shortcoming of length consuming time, does not meet defect of pipeline research field and requires iteratively faster, the fast feature of detection speed.
Extreme learning machine (extreme learning machine) ELM is a kind of use, effective single hidden layer feedforward neural network SLFNS learning algorithm of being simple and easy to.2004 associate professors Huang Guangbin of Nian You Nanyang Technolohy University propose.Traditional Learning Algorithm (as BP algorithm) need to artificially arrange a large amount of network training parameters, and is easy to produce locally optimal solution.Extreme learning machine only need to arrange the hidden node number of network, does not need to adjust the input weights of network and the biasing of hidden unit, and produce unique optimum solution in algorithm implementation, therefore has advantages of that pace of learning is fast and Generalization Capability good.This algorithm is different from traditional learning algorithm, and this learning method is when guaranteeing that network has good Generalization Capability, and greatly degree has improved the pace of learning of feedforward neural network, simultaneously the advantages such as pace of learning is fast again, Generalization Capability is good.ELM algorithm produces the threshold value that is connected weights and hidden layer neuron of input layer and hidden layer at random, at training process, without adjustment, the number of hidden layer neuron only need to be set, and just can obtain unique optimum solution.
Summary of the invention
The deficiency existing for prior art, the invention provides a kind of pipe leakage defect inspection method based on extreme learning machine.
Technical scheme of the present invention is:
A pipe leakage defect inspection method based on extreme learning machine, comprises the following steps:
Step 1: obtain the form parameter of known pipeline pipeline magnetic flux leakage defect, comprise length, width, the depth data of pipe leakage defect, and the magnetic leakage signal waveform at known pipeline pipeline magnetic flux leakage defect place is carried out to eigenwert extraction, extract magnetic leakage signal waveform character value.
Step 2: using the length of known pipeline pipeline magnetic flux leakage defect, width, depth data and magnetic leakage signal waveform character value as sample data, sample data is divided into training sample data and test sample book data.
Step 3: for training sample data, set up extreme learning machine model, the length of the known pipeline pipeline magnetic flux leakage defect in training sample data, width, depth data are as the input of this model, use method of trial and error to choose hidden layer node number, calculate hidden layer output matrix and output weights, magnetic leakage signal waveform character value is as the output of this model.
Step 4: utilize test sample book data to proofread and correct extreme learning machine model: by the length of known pipeline pipeline magnetic flux leakage defect in test sample book, width, depth data input limits learning machine model, average relative error to the magnetic leakage signal waveform character value of the output of extreme learning machine model and test sample book data judges: if average relative error meets the expection average relative error of setting, current extreme learning machine model is final extreme learning machine model, execution step 5, otherwise return to step 3.
Step 5: during pipeline generation leakage field, obtain the magnetic leakage signal waveform of unknown pipeline magnetic flux leakage defect shape, limit of utilization learning machine model carries out pipe leakage defects detection.
Step 5.1: produce at random original shape parameter, i.e. the length of initial tract pipeline magnetic flux leakage defect, width, depth data.
Step 5.2: by form parameter input limits learning machine model, obtain signal waveform eigenwert corresponding to form parameter.
Step 5.3: the fitness value that calculates the magnetic leakage signal waveform of magnetic leakage signal waveform corresponding to this signal waveform eigenwert and unknown defect shape, if this fitness value is more than or equal to the maximum adaptation degree value of setting, determine that current form parameter is the form parameter of pipe leakage defect, execution step 5.5, otherwise, execution step 5.4.
Step 5.4: current form parameter application genetic algorithm is upgraded to the form parameter making new advances, returned to step 5.3.
Step 5.5: the form parameter of current pipe leakage defect is pipe leakage defects detection result.
The invention has the beneficial effects as follows: the present invention is based on extreme learning machine model defect of pipeline shape is carried out to Intelligent Inversion, set up extreme learning machine model, can carry out defect shape structure fast to defect of pipeline shape.Compare with the Finite Element Method extensively adopting at present, extreme learning machine model method has the advantages such as pace of learning is fast, Generalization Capability is good.For the defective waveform that uses pipeline detection robot to detect, can quick and precisely construct the shape of defect, thereby learn the seriousness of defect, for precognition pipeline risk, prevent that pipe leakage is significant.
Accompanying drawing explanation
Fig. 1 is the pipe leakage defect inspection method process flow diagram based on extreme learning machine in the specific embodiment of the invention;
Fig. 2 is pipe leakage flaw indication waveform character value schematic diagram in the pipe leakage defect inspection method based on extreme learning machine in the specific embodiment of the invention;
Fig. 3 is extreme learning machine model training process flow diagram flow chart in the pipe leakage defect inspection method based on extreme learning machine in the specific embodiment of the invention;
Fig. 4 is that the signal waveform of the unknown defect shape of utilization in the specific embodiment of the invention utilizes genetic algorithm pipeline to be carried out to the process flow diagram of pipe leakage defects detection.
Embodiment
Below in conjunction with accompanying drawing to the specific embodiment of the invention detailed explanation in addition.
A pipe leakage defect inspection method based on extreme learning machine, as shown in Figure 1, comprises the following steps:
Step 1: obtain the form parameter of known pipeline pipeline magnetic flux leakage defect, comprise length, width, the depth data of pipe leakage defect, and the magnetic leakage signal waveform at known pipeline pipeline magnetic flux leakage defect place is carried out to eigenwert extraction, extract magnetic leakage signal waveform character value.
Step 1.1: obtain the form parameter of known pipeline pipeline magnetic flux leakage defect, comprise length, width, the depth data of pipe leakage defect, obtain the flaw indication oscillogram of known pipeline pipeline magnetic flux leakage defect, obtain the transverse and longitudinal coordinate data of flaw indication waveform.
Step 1.2: the transverse and longitudinal coordinate data to many strip defects signal waveform of each pipe leakage fault location is added up.
Step 1.3: the magnetic leakage signal waveform at each the known pipeline pipeline magnetic flux leakage defect place after being combined carries out eigenwert extraction: crest and the trough data of calculating each the pipe leakage fault location signal waveform after merging, extract the most obvious waveform of each fault location waveform fluctuation, obtain flaw indication waveform character value, i.e. the peak-to-valley value of waveform, peak-valley difference, Gu Gu difference and paddy valley.
As shown in Figure 2, pipe leakage flaw indication waveform character value is mainly divided double peak to valley, two kinds of situations of unimodal pair of paddy, poor, the paddy valley of peak-to-valley value, peak-valley difference, Gu Gu of definition is in the definition of having nothing in common with each other of double peak to valley, two kinds of situations of unimodal pair of paddy, but the feature of base natural reaction defect.
Peak-to-valley value represents that crest is to longitudinal difference of trough, and paddy valley represents the longitudinal difference between two troughs, and peak-valley difference represents the transversal discrepancy between crest and trough, the poor transversal discrepancy representing between trough and trough of Gu Gu.
Peak-to-valley value in double peak to valley situation, peak-valley difference, Gu Gu are poor, paddy valley ratio is easier to determine, in present embodiment, peak-to-valley value has 4, peak-valley difference has 4, Gu Gu is poor, paddy valley only has 1, as for peak-to-peak value, can get difference by two peak-to-valley values and determine, therefore, no longer singly list as a feature.And peak-to-peak is poor, also can be determined by two peak-valley differences, therefore also no longer single-row out as a feature.
In unimodal pair of paddy situation, peak-to-valley value, peak-valley difference, Gu Gu are poor, the definition of paddy valley is because only have a peak, therefore distinguish to some extent with the situation of double peak to valley.Peak-to-valley value has 2, and peak-valley difference has 2, and Gu Gu is poor, paddy valley only has 1.In addition, if desired peak-to-peak is poor, because only have a peak value, think peak-to-peak poor=0.
Choosing of flaw indication waveform character value can be extracted according to actual conditions.
Step 2: using the length of known pipeline pipeline magnetic flux leakage defect, width, depth data and magnetic leakage signal waveform character value as sample data, sample data is divided into training sample data and test sample book data.
Step 3: as shown in Figure 3, for training sample data, set up extreme learning machine model, the length of the known pipeline pipeline magnetic flux leakage defect in training sample data, width, depth data are as the input of this model, use method of trial and error to choose hidden layer node number, calculate hidden layer output matrix and output weights, magnetic leakage signal waveform character value is as the output of this model.
Step 3.1: setting initial hidden layer node number is N
*, initial hidden layer node number is generally got half of training sample data number N.Obtain at random initial input weight w
iwith biasing b
i, i=1, N
*.
Step 3.2: calculate hidden layer output matrix H.
The hidden layer node number that only needs definite neural network for extreme learning machine model, do not need to adjust the input weights of network and the biasing of hidden unit and other parameters, therefore adopt method of trial and error to choose suitable hidden layer node number: for given N different training sample data N={ (x
i, t
i) | x
i∈ R
n, t
i∈ R
m, i=1 ... N}, N value is generally greater than 50, gets N=100, wherein x in present embodiment
ifor n dimension input variable, the input variable in present embodiment is length, width, the depth data of the known pipeline pipeline magnetic flux leakage defect in training sample data, is 3 dimensions, i.e. x
i∈ R
3, n=3, t
ifor m dimension output variable, the output variable in present embodiment is the signal waveform eigenwert of pipe leakage defect, is respectively that two peak-to-valley values of signal waveform, two peak-valley differences, paddy paddy are poor, a paddy valley, is therefore 6 dimensions, i.e. t
i∈ R
6, i is training sample data numberings, m=6.
Setting activation function is g (x), a unlimited differentiable function of the general selection of activation function, can choose the non-linear activation functions such as sine function, S type function and function of functions, also can use non-differentiable function, or even discontinuous function be as activation function.
Calculate hidden layer output matrix H, formula is as follows:
Step 3.3: calculate output weights β, formula is as follows:
β=H
-1T (2)
Wherein, T=[t
1t
n], H
-1it is the Moore-Penrose generalized inverse matrix of hidden layer output matrix H.
Step 4: utilize test sample book data to proofread and correct extreme learning machine model: by the length of known pipeline pipeline magnetic flux leakage defect in test sample book, width, depth data input limits learning machine model, average relative error to the magnetic leakage signal waveform character value of the output of extreme learning machine model and test sample book data judges: if average relative error meets the expection average relative error of setting, current extreme learning machine model is final extreme learning machine model, execution step 5, otherwise return to step 3.
Utilize test sample book data to proofread and correct extreme learning machine model: length, width, the depth data of the known pipeline pipeline magnetic flux leakage defect in test sample book data are inputted to this extreme learning machine model, obtain the output of extreme learning machine model, i.e. the signal waveform eigenwert o of output
i∈ R
m, i=1 ... M, wherein o
ifor m dimension output variable, M is test sample book data amount check.
By the signal waveform eigenwert o of output
iwith the signal waveform eigenwert t in test sample book data
iuse average relative error to carry out the judgement of accuracy rate, average relative error computing formula is as follows:
Wherein M is test sample book data amount check, and the average relative error ζ obtaining is compared with expection average relative error ζ ', guarantees that in M variable, maximum average relative error is less than prediction relative average error, i.e. ζ=max (ζ
i)≤ζ ', wherein ζ ' is the expection average relative error of setting, the expection average relative error of setting in present embodiment is in 10%.
If the average relative error obtaining is less than or equal to the expection average relative error of setting, current extreme learning machine model is final extreme learning machine model, otherwise, reselect hidden layer node number, the hidden layer node number reselecting changes near initial hidden layer node number, returns to step 3.
Step 5: during pipeline generation leakage field, obtain the magnetic leakage signal waveform of unknown pipeline magnetic flux leakage defect shape, limit of utilization learning machine model carries out pipe leakage defects detection, as shown in Figure 4.
Step 5.1: produce at random original shape parameter, i.e. the length of initial tract pipeline magnetic flux leakage defect, width, depth data.
Step 5.2: by form parameter input limits learning machine model, obtain signal waveform eigenwert corresponding to form parameter.
Step 5.3: the fitness value that calculates the magnetic leakage signal waveform of magnetic leakage signal waveform corresponding to this signal waveform eigenwert and unknown defect shape.
Fitness function F formula is as follows:
Wherein, N is length, the width of initial tract pipeline magnetic flux leakage defect, the sample data number of depth data, and C is a constant, span generally 1 to 100, B
i messuredthe magnetic flux density of the corresponding defect of signal waveform of unknown defect shape, B
i eLMit is the magnetic flux density of the corresponding defect of signal waveform corresponding to form parameter of prediction.Because work as
time, it is 1 that F reaches maximal value, so 0<F≤1, therefore can learn when J equal 0 between constant time the size of F depend on the size of constant C.
The maximum adaptation degree value F setting
max, its span is 0.9~1, in present embodiment, value is 0.99, F
maxvalue is different, and test findings has difference, and test of many times, selects suitable value as requested.
The maximum adaptation degree value F of the fitness value F relatively calculating and setting
maxmagnitude relationship, if this fitness value F is more than or equal to the maximum adaptation degree value F of setting
max, determine that current form parameter is the form parameter of pipe leakage defect, execution step 5.5, otherwise, execution step 5.4.
Step 5.4: current form parameter application genetic algorithm is upgraded to the form parameter making new advances, returned to step 5.3.
Step 5.5: the form parameter of current pipe leakage defect is pipe leakage defects detection result.
Claims (1)
1. the pipe leakage defect inspection method based on extreme learning machine, is characterized in that, comprises the following steps:
Step 1: obtain the form parameter of known pipeline pipeline magnetic flux leakage defect, comprise length, width, the depth data of pipe leakage defect, and the magnetic leakage signal waveform at known pipeline pipeline magnetic flux leakage defect place is carried out to eigenwert extraction, extract magnetic leakage signal waveform character value;
Step 2: using the length of known pipeline pipeline magnetic flux leakage defect, width, depth data and magnetic leakage signal waveform character value as sample data, sample data is divided into training sample data and test sample book data;
Step 3: for training sample data, set up extreme learning machine model, the length of the known pipeline pipeline magnetic flux leakage defect in training sample data, width, depth data are as the input of this model, use method of trial and error to choose hidden layer node number, calculate hidden layer output matrix and output weights, magnetic leakage signal waveform character value is as the output of this model;
Step 4: utilize test sample book data to proofread and correct extreme learning machine model: by the length of known pipeline pipeline magnetic flux leakage defect in test sample book, width, depth data input limits learning machine model, average relative error to the magnetic leakage signal waveform character value of the output of extreme learning machine model and test sample book data judges: if average relative error meets the expection average relative error of setting, current extreme learning machine model is final extreme learning machine model, execution step 5, otherwise return to step 3;
Step 5: during pipeline generation leakage field, obtain the magnetic leakage signal waveform of unknown pipeline magnetic flux leakage defect shape, limit of utilization learning machine model carries out pipe leakage defects detection;
Step 5.1: produce at random original shape parameter, i.e. the length of initial tract pipeline magnetic flux leakage defect, width, depth data;
Step 5.2: by form parameter input limits learning machine model, obtain signal waveform eigenwert corresponding to form parameter;
Step 5.3: the fitness value that calculates the magnetic leakage signal waveform of magnetic leakage signal waveform corresponding to this signal waveform eigenwert and unknown defect shape, if this fitness value is more than or equal to the maximum adaptation degree value of setting, determine that current form parameter is the form parameter of pipe leakage defect, execution step 5.5, otherwise, execution step 5.4;
Step 5.4: current form parameter application genetic algorithm is upgraded to the form parameter making new advances, returned to step 5.3;
Step 5.5: the form parameter of current pipe leakage defect is pipe leakage defects detection result.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201410261496.8A CN104034794B (en) | 2014-06-12 | 2014-06-12 | A kind of pipe leakage defect inspection method based on extreme learning machine |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201410261496.8A CN104034794B (en) | 2014-06-12 | 2014-06-12 | A kind of pipe leakage defect inspection method based on extreme learning machine |
Publications (2)
Publication Number | Publication Date |
---|---|
CN104034794A true CN104034794A (en) | 2014-09-10 |
CN104034794B CN104034794B (en) | 2017-01-04 |
Family
ID=51465643
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN201410261496.8A Active CN104034794B (en) | 2014-06-12 | 2014-06-12 | A kind of pipe leakage defect inspection method based on extreme learning machine |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN104034794B (en) |
Cited By (18)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN104503420A (en) * | 2014-09-19 | 2015-04-08 | 北京化工大学 | Non-linear process industry fault prediction method based on novel FDE-ELM and EFSM |
CN104514987A (en) * | 2014-12-19 | 2015-04-15 | 清华大学 | Three-dimensional pipeline flux leakage imaging defect quantizing method |
CN105005693A (en) * | 2015-07-08 | 2015-10-28 | 中国科学院合肥物质科学研究院 | Genetic material specificity based tumor cell drug sensitivity evaluation method |
CN105929216A (en) * | 2016-04-08 | 2016-09-07 | 南京航空航天大学 | Satellite power source main bus-bar current interval prediction method |
CN106018545A (en) * | 2016-06-29 | 2016-10-12 | 东北大学 | Pipeline defect magnetic flux leakage inversion method based on Adaboost-RBF synergy |
CN106770627A (en) * | 2016-12-16 | 2017-05-31 | 北京华航无线电测量研究所 | A kind of axial magnetic leakage signal length quantization method |
CN107024532A (en) * | 2017-04-12 | 2017-08-08 | 东北大学 | A kind of leakage field defect of pipeline position extracting method based on forms feature |
CN107024531A (en) * | 2017-03-22 | 2017-08-08 | 清华大学 | Magnetic Flux Leakage Inspecting cell defect is stretched reconstructing method |
CN107102056A (en) * | 2017-03-22 | 2017-08-29 | 清华大学 | The flexible construction method of the unit of defect and magnetic leakage signal |
CN107256453A (en) * | 2016-12-05 | 2017-10-17 | 东北大学 | A kind of quality of hollow billet forecasting procedure based on improvement ELM algorithms |
CN108730776A (en) * | 2018-06-22 | 2018-11-02 | 大连理工大学 | A kind of Oil Pipeline Leakage Detection based on the fusion of extreme learning machine information |
CN109613109A (en) * | 2018-12-19 | 2019-04-12 | 智云安科技(北京)有限公司 | A kind of Pipeline Magnetic Flux Leakage Inspection automatic data analysis system |
CN109632942A (en) * | 2019-02-21 | 2019-04-16 | 东北大学 | A kind of inversion method of the defect of pipeline size based on SL |
CN111448453A (en) * | 2017-10-06 | 2020-07-24 | 罗森瑞士股份公司 | Method for determining the geometry of a defect and method for determining the load capacity limit |
CN111815561A (en) * | 2020-06-09 | 2020-10-23 | 中海石油(中国)有限公司 | Pipeline defect and pipeline assembly detection method based on depth space-time characteristics |
CN111861985A (en) * | 2020-06-09 | 2020-10-30 | 中海油能源发展装备技术有限公司 | Magnetic flux leakage defect deep identification method based on self-adaptive fuzzy neural network |
US11003518B2 (en) | 2016-09-29 | 2021-05-11 | Hewlett-Packard Development Company, L.P. | Component failure prediction |
CN115081485A (en) * | 2022-07-04 | 2022-09-20 | 中特检深燃安全技术服务(深圳)有限公司 | AI-based automatic analysis method for magnetic flux leakage internal detection data |
Citations (4)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
US7215834B1 (en) * | 2002-06-07 | 2007-05-08 | Magnachip Semiconductor, Ltd. | Congfigurable image processing driver |
CN102254196A (en) * | 2011-06-22 | 2011-11-23 | 江苏奥博洋信息技术有限公司 | Method for identifying handwritten Chinese character by virtue of computer |
CN102944769A (en) * | 2012-10-18 | 2013-02-27 | 上海市电力公司 | Fault diagnosis method of power transformer based on extreme learning machine |
CN103593670A (en) * | 2013-10-14 | 2014-02-19 | 浙江工业大学 | Copper sheet and strip surface defect detection method based on-line sequential extreme learning machine |
-
2014
- 2014-06-12 CN CN201410261496.8A patent/CN104034794B/en active Active
Patent Citations (4)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
US7215834B1 (en) * | 2002-06-07 | 2007-05-08 | Magnachip Semiconductor, Ltd. | Congfigurable image processing driver |
CN102254196A (en) * | 2011-06-22 | 2011-11-23 | 江苏奥博洋信息技术有限公司 | Method for identifying handwritten Chinese character by virtue of computer |
CN102944769A (en) * | 2012-10-18 | 2013-02-27 | 上海市电力公司 | Fault diagnosis method of power transformer based on extreme learning machine |
CN103593670A (en) * | 2013-10-14 | 2014-02-19 | 浙江工业大学 | Copper sheet and strip surface defect detection method based on-line sequential extreme learning machine |
Non-Patent Citations (1)
Title |
---|
遇炳杰等: "加权极限学习机在变压器故障诊断中的应用", 《计算机工程与设计》 * |
Cited By (28)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN104503420B (en) * | 2014-09-19 | 2017-02-15 | 北京化工大学 | Non-linear process industry fault prediction method based on novel FDE-ELM and EFSM |
CN104503420A (en) * | 2014-09-19 | 2015-04-08 | 北京化工大学 | Non-linear process industry fault prediction method based on novel FDE-ELM and EFSM |
CN104514987A (en) * | 2014-12-19 | 2015-04-15 | 清华大学 | Three-dimensional pipeline flux leakage imaging defect quantizing method |
CN104514987B (en) * | 2014-12-19 | 2017-02-22 | 清华大学 | Three-dimensional pipeline flux leakage imaging defect quantizing method |
CN105005693B (en) * | 2015-07-08 | 2017-12-01 | 中国科学院合肥物质科学研究院 | One kind is based on the specific tumour cell drug susceptibility appraisal procedure of inhereditary material |
CN105005693A (en) * | 2015-07-08 | 2015-10-28 | 中国科学院合肥物质科学研究院 | Genetic material specificity based tumor cell drug sensitivity evaluation method |
CN105929216A (en) * | 2016-04-08 | 2016-09-07 | 南京航空航天大学 | Satellite power source main bus-bar current interval prediction method |
CN105929216B (en) * | 2016-04-08 | 2019-04-19 | 南京航空航天大学 | A kind of satellite power supply main bus-bar electric current section prediction technique |
CN106018545A (en) * | 2016-06-29 | 2016-10-12 | 东北大学 | Pipeline defect magnetic flux leakage inversion method based on Adaboost-RBF synergy |
CN106018545B (en) * | 2016-06-29 | 2019-05-14 | 东北大学 | A kind of pipeline defect and magnetic leakage inversion method based on Adaboost-RBF collaboration |
US11003518B2 (en) | 2016-09-29 | 2021-05-11 | Hewlett-Packard Development Company, L.P. | Component failure prediction |
CN107256453A (en) * | 2016-12-05 | 2017-10-17 | 东北大学 | A kind of quality of hollow billet forecasting procedure based on improvement ELM algorithms |
CN107256453B (en) * | 2016-12-05 | 2020-10-23 | 东北大学 | Capillary quality forecasting method based on improved ELM algorithm |
CN106770627A (en) * | 2016-12-16 | 2017-05-31 | 北京华航无线电测量研究所 | A kind of axial magnetic leakage signal length quantization method |
CN106770627B (en) * | 2016-12-16 | 2019-12-13 | 北京华航无线电测量研究所 | Axial magnetic flux leakage signal length quantization method |
CN107102056A (en) * | 2017-03-22 | 2017-08-29 | 清华大学 | The flexible construction method of the unit of defect and magnetic leakage signal |
CN107024531A (en) * | 2017-03-22 | 2017-08-08 | 清华大学 | Magnetic Flux Leakage Inspecting cell defect is stretched reconstructing method |
CN107024531B (en) * | 2017-03-22 | 2020-03-27 | 清华大学 | Defect expansion reconstruction method for magnetic flux leakage detection unit |
CN107024532A (en) * | 2017-04-12 | 2017-08-08 | 东北大学 | A kind of leakage field defect of pipeline position extracting method based on forms feature |
CN111448453A (en) * | 2017-10-06 | 2020-07-24 | 罗森瑞士股份公司 | Method for determining the geometry of a defect and method for determining the load capacity limit |
CN108730776A (en) * | 2018-06-22 | 2018-11-02 | 大连理工大学 | A kind of Oil Pipeline Leakage Detection based on the fusion of extreme learning machine information |
CN109613109A (en) * | 2018-12-19 | 2019-04-12 | 智云安科技(北京)有限公司 | A kind of Pipeline Magnetic Flux Leakage Inspection automatic data analysis system |
CN109632942A (en) * | 2019-02-21 | 2019-04-16 | 东北大学 | A kind of inversion method of the defect of pipeline size based on SL |
CN111815561A (en) * | 2020-06-09 | 2020-10-23 | 中海石油(中国)有限公司 | Pipeline defect and pipeline assembly detection method based on depth space-time characteristics |
CN111861985A (en) * | 2020-06-09 | 2020-10-30 | 中海油能源发展装备技术有限公司 | Magnetic flux leakage defect deep identification method based on self-adaptive fuzzy neural network |
CN111815561B (en) * | 2020-06-09 | 2024-04-16 | 中海石油(中国)有限公司 | Pipeline defect and pipeline assembly detection method based on depth space-time characteristics |
CN111861985B (en) * | 2020-06-09 | 2024-04-16 | 中海油能源发展装备技术有限公司 | Magnetic flux leakage defect depth identification method based on self-adaptive fuzzy neural network |
CN115081485A (en) * | 2022-07-04 | 2022-09-20 | 中特检深燃安全技术服务(深圳)有限公司 | AI-based automatic analysis method for magnetic flux leakage internal detection data |
Also Published As
Publication number | Publication date |
---|---|
CN104034794B (en) | 2017-01-04 |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
CN104034794A (en) | Extreme learning machine-based pipeline magnetic flux leakage defect detection method | |
CN101923083B (en) | Sewage chemical oxygen demand soft measuring method based on support vector machine and neural network | |
EP4216117B1 (en) | Method and apparatus for training water-and-sediment prediction model for reservoir and method and apparatus for predicting water-and-sediment in reservoir | |
CN103033856B (en) | Rainfall frequency estimation method based on hydrometeorology consistency geographical analysis | |
CN103065202B (en) | Wind power plant ultrashort term wind speed prediction method based on combination kernel function | |
CN103886373A (en) | High-speed cold-rolling mill third octave flutter prediction method based on BP neural network | |
CN104239659A (en) | Carbon steel corrosion rate prediction method of back propagation (BP) neural network | |
CN104376214A (en) | Fluctuating wind velocity simulation method based on data driving | |
CN102663495B (en) | Neural net data generation method for nonlinear device modeling | |
CN102364501A (en) | Method for reproducing two-dimensional defect of petroleum pipeline PSO-BP (Particle Swarm Optimization-Back-Propagation) neural network | |
CN111664823B (en) | Method for detecting thickness of scale layer of voltage-sharing electrode based on difference of medium heat conduction coefficients | |
CN103198215A (en) | Method for predicting subsidence of soil outside pit based on differential evolution support vector machine | |
Adnan et al. | 5 hours flood prediction modeling using improved NNARX structure: case study Kuala Lumpur | |
CN113762625A (en) | Power distribution network state evaluation method and system based on graph convolution network | |
CN105005708A (en) | Generalized load characteristic clustering method based on AP clustering algorithm | |
CN102542126A (en) | Soft measurement method based on half supervision learning | |
CN106320255A (en) | Overbank flow floodplain and channel excess flow computing method | |
CN106405683A (en) | Wind speed forecasting method and device based on G-L mixed noise characteristic kernel ridge regression technology | |
Wang et al. | Hybrid learning based Gaussian ARTMAP network for tool condition monitoring using selected force harmonic features | |
CN107341336A (en) | A kind of tank product geometric accuracy method for evaluating consistency | |
CN116757321A (en) | Solar direct radiation quantity prediction method, system, equipment and storage medium | |
Xie et al. | A method of flood forecasting of chaotic radial basis function neural network | |
Lee et al. | Neural network approach to coastal high and low water level prediction | |
Jiang et al. | Discharge estimation based on machine learning | |
CN103822960A (en) | Online multi-metal ion concentration detection method utilizing polarography |
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 |