CN110116254A - Oil-gas gathering and transportation composite bimetal pipe girth joint failure prediction and control method - Google Patents
Oil-gas gathering and transportation composite bimetal pipe girth joint failure prediction and control method Download PDFInfo
- Publication number
- CN110116254A CN110116254A CN201910368418.0A CN201910368418A CN110116254A CN 110116254 A CN110116254 A CN 110116254A CN 201910368418 A CN201910368418 A CN 201910368418A CN 110116254 A CN110116254 A CN 110116254A
- Authority
- CN
- China
- Prior art keywords
- defect
- welding
- data
- prediction
- weld
- 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
-
- B—PERFORMING OPERATIONS; TRANSPORTING
- B23—MACHINE TOOLS; METAL-WORKING NOT OTHERWISE PROVIDED FOR
- B23K—SOLDERING OR UNSOLDERING; WELDING; CLADDING OR PLATING BY SOLDERING OR WELDING; CUTTING BY APPLYING HEAT LOCALLY, e.g. FLAME CUTTING; WORKING BY LASER BEAM
- B23K9/00—Arc welding or cutting
- B23K9/0026—Arc welding or cutting specially adapted for particular articles or work
-
- B—PERFORMING OPERATIONS; TRANSPORTING
- B23—MACHINE TOOLS; METAL-WORKING NOT OTHERWISE PROVIDED FOR
- B23K—SOLDERING OR UNSOLDERING; WELDING; CLADDING OR PLATING BY SOLDERING OR WELDING; CUTTING BY APPLYING HEAT LOCALLY, e.g. FLAME CUTTING; WORKING BY LASER BEAM
- B23K9/00—Arc welding or cutting
- B23K9/095—Monitoring or automatic control of welding parameters
-
- B—PERFORMING OPERATIONS; TRANSPORTING
- B23—MACHINE TOOLS; METAL-WORKING NOT OTHERWISE PROVIDED FOR
- B23K—SOLDERING OR UNSOLDERING; WELDING; CLADDING OR PLATING BY SOLDERING OR WELDING; CUTTING BY APPLYING HEAT LOCALLY, e.g. FLAME CUTTING; WORKING BY LASER BEAM
- B23K9/00—Arc welding or cutting
- B23K9/235—Preliminary treatment
-
- B—PERFORMING OPERATIONS; TRANSPORTING
- B23—MACHINE TOOLS; METAL-WORKING NOT OTHERWISE PROVIDED FOR
- B23K—SOLDERING OR UNSOLDERING; WELDING; CLADDING OR PLATING BY SOLDERING OR WELDING; CUTTING BY APPLYING HEAT LOCALLY, e.g. FLAME CUTTING; WORKING BY LASER BEAM
- B23K9/00—Arc welding or cutting
- B23K9/32—Accessories
-
- B—PERFORMING OPERATIONS; TRANSPORTING
- B23—MACHINE TOOLS; METAL-WORKING NOT OTHERWISE PROVIDED FOR
- B23K—SOLDERING OR UNSOLDERING; WELDING; CLADDING OR PLATING BY SOLDERING OR WELDING; CUTTING BY APPLYING HEAT LOCALLY, e.g. FLAME CUTTING; WORKING BY LASER BEAM
- B23K2101/00—Articles made by soldering, welding or cutting
- B23K2101/04—Tubular or hollow articles
- B23K2101/06—Tubes
Abstract
Oil-gas gathering and transportation composite bimetal pipe girth joint failure prediction and control method, based on actual welding defect a situation arises big data, comprehensive machine learning and Optimization Theory method, sorting algorithm, regression algorithm or clustering algorithm is respectively adopted, the bimetal tube welding defect prediction model by welding condition as dependent variable, all kinds of defective datas of welding as variable is established, optimization is then implemented using genetic algorithm.Present invention combination On-site Welding Technology parameter and defects detection distributed data statistics, weld defect prediction model is established by Monte Carlo simulation, according to establishing desired defect Con trolling index the considerations of taking into account quality and cost, using genetic algorithm, to be expected weld defect Con trolling index as optimization aim, using welding condition as optimal design parameter, optimize calculating, obtain the welding condition optimum results that control defect occurs, so as to improve the accuracy of weld defect Occurrence forecast, the probability of happening of defect is effectively controlled.
Description
Technical field
The present invention relates to the defeated equipment technology fields of petroleum gas collection, in particular to oil-gas gathering and transportation composite bimetal pipe ring
Plumb joint failure prediction and control method.
Background technique
Dual-metal clad steel pipe basic principle: outer base tube is responsible for the effect of pressure-bearing and pipeline rigid support, and internal lining pipe undertakes
Corrosion resistant effect.Oil gas field collection is defeated to generally use mechanical stitch method, explosion composite method, drawing composite algorithm with composite bimetal pipe
And hydraulic composite algorithm manufacture.Gathering line is connected by site welding mode, and site girth welding seam is the weak ring of pipeline integrity
Section, weldquality seriously affect the military service safety of gathering line.Bimetallic collector and delivery pipe site welding prevailing quality defect includes
The following aspects: naked eyes or low power magnifying glass can detect and be located at face of weld defect, as weld size is not inconsistent rule
Model, undercut, overlap, arc crater, surface pores, slag inclusion, face crack etc.;Destructive testing or special non-destructive testing side must be used
Method just detectable tubing internal flaw, such as stomata, slag inclusion, underbead crack, lack of penetration, incomplete tusion.Each type of welding
Defect generation is influenced by the complicated coupling of many factors, such as: the factor for leading to weld size defect includes that weld groove processing is flat
Directly spend that poor, bevel angle is improper, fit-up gap size is uneven, welding current is excessive causes that welding rod fusing is too fast, electric current is too small
Cause welding rod bonding etc.;The causing factors of undercut have, and welding current is excessive, electric arc is too long, welding rod angle grasps improper, arc manipulation
Speed it is improper etc.;The influence factor of crackle has, weld heat-affected zone shrink after generate big stress, base material containing hardened structure compared with
Lead to easily raw crackle after cooling down, higher, other harmful element impurity effects of hydrogen concentration etc. in weld seam more;Arc crater generates main cause
Have, the blow-out time is too short etc. when welding ends;Slag inclusion reason has, and the cleaning of weld seam base material is not clean, welding current is too small, fusing is golden
Category solidification is too fast, slag has little time emersion etc.;Stomata reason has, welding rod itself is inferior, welding rod dampness is not dried by prescribed requirement,
Electrode coating is rotten or peels off, core wire corrodes etc.;Arc scar producing cause has, and welding operation is lack of standardization, and maintenance tool is not in place etc.;
Lack of penetration reason includes, the angle of the groove or gap of group pair is too small, the oxidation that do not clean out is welded in groove two sides and interlayer
Object and slag hinder sufficiently to fuse between metal, select welding current is excessive cause welding rod prior to base material fusing, select too little current
When arc manipulation excessive velocities, the angle of welding rod incorrect cause to melt amesiality and to form part lack of penetration etc..In addition to above-mentioned
Except, the special defect producing cause complicated difficult in some bimetal tube welding processes is surveyed, such as: welding process base carbon content is high
In lining, make diluting effect of the lining by base, causes weld seam austenite former to reduce and increase with phosphorus content, be also easy to produce
Solidification cracking;Welding heat effect makes composite material local melting, and alloying element penetrates into weld seam, geneva is easily generated near melt run
Body tissue, to cause the melting area of welding seam brittle.
Adjustment repeatedly and optimization are carried out often through welding scene technique trial and error in the prior art, causes to weld in engineering practice
Connect defect producing cause difficulty look for, workaround lack.
Summary of the invention
In order to solve the problems in the prior art, the object of the present invention is to provide oil-gas gathering and transportation composite bimetal pipes
Girth joint failure prediction and control method effectively control welding defect and produce to improve the accuracy of welding defect Occurrence forecast
It is raw.
To achieve the above object, technical solution provided by the invention is as follows:
Oil-gas gathering and transportation composite bimetal pipe girth joint failure prediction and control method are sent out based on actual welding defect
Raw situation big data, comprehensive machine learning and Optimization Theory method are respectively adopted classification and calculate according to different welding defects
Method, regression algorithm or clustering algorithm are established by welding condition as dependent variable, all kinds of defective datas of welding as variable
Then bimetal tube welding defect prediction model implements optimization using genetic algorithm, realize oil-gas gathering and transportation composite bimetal pipe
The prediction and control of girth joint welding defect.
A further improvement of the present invention lies in that implementing welding technology optimization and welding using genetic algorithm, it is double to realize that oil-gas gathering and transportation is used
Metal composite pipe girth joint welding defect prediction with control process include genetic coding generation, random initial population it is true
The Optimized Iterative process that fixed, Fitness analysis and progeny population calculate, on the basis of assessing all individual Pareto values,
The mean value and standard deviation for introducing individual Pareto value assess optimization aim gradient pair by the mean value and standard deviation of individual Pareto value
The sensitivity of individual surrounding population density information makes offspring's breeding of individual tend to sluggish region far from optimization;Establish with
It reduces welding defect and occurs as optimization aim, using welding condition as the Optimized model of optimization design variable, realize and change weldering
Connect control of the technological parameter to welding defect is reduced.
A further improvement of the present invention lies in that specifically includes the following steps:
S1, be stored in advance multiple groups unlike material, grade of steel, specification and bimetallic compound type On-site Welding Technology parameter and
Corresponding weld seam actual defects data;
Weld seam actual defects data are statisticallyd analyze respectively, or carry out the book of final entry according to with/without situation, or according to size
Or quantity carries out continuous data recording, and carries out classification grading;
Then the bimetal tube welding of the several groups material of the same race, specification, grade of steel and bimetallic compound type of storage is chosen
Technological parameter data and its corresponding weld seam actual defects data, and normalization is carried out to the data of selection and is abstracted as vector,
Data after one vectorization of rule are divided into defect training data sample and verify data sample;
S2, for a given group welding technological parameter and defective data, according to point of welding input and defect output relation
Analysis judges that defect welds defect type or composite bimetal pipe welding defect type for metal;
S3 welds defect type for metal, carries out machine learning using supervised learning mode, by sorting algorithm or returns
Reduction method is handled, and classification prediction model or regressive prediction model are obtained;
S4 carries out machine learning using unsupervised learning mode, and carry out for composite bimetal pipe welding defect type
Organized cluster and clustering, establish Forecast model;
S5, welding defect data are corresponding with welding condition, uniformly it is input to classification prediction model, regression forecasting
Model or Forecast model, by Model Self-Learning, to obtain all kinds of as dependent variable, welding by welding condition
Bimetal tube welding defect prediction model of the defective data as variable;
S6 carries out the bimetal tube welding defect prediction model in step S5 using the verify data sample in step S1
Cross validation, the model prediction for inaccuracy is as a result, then more new database, obtains defect training data sample and verify data
Verify data sample is re-entered the bimetal tube welding defect prediction model into step S5 and carries out cross validation by sample,
The correctness of model prediction is continuously improved;
Step S6 is repeated several times in S7, finally obtains welding defect prediction model, and to lacking under specified welding condition
It falls into generation degree to be predicted, exports weld defect predicted value;
Weld defect predicted value and target defect controlling value are compared difference as objective function, welding condition by S8
As design variable, iteration is optimized with genetic algorithm, finally meets the requirement of Engineering Control target, output control defect
The welding condition Optimum Design Results of degree, to control the generation of defect.
A further improvement of the present invention lies in that On-site Welding Technology supplemental characteristic includes: welding current, weldering in step S1
Stitch groove straightness, bevel angle, fit-up gap, arc manipulation speed, base material and wlding chemical component, the welding end of a period blow-out time,
Weld seam base material clean-up performance and welding rod degree of drying;Weld seam actual defects data include: that naked eyes or low power magnifying glass can detect
To and the tubing internal flaw that detects of defect, destructive testing or lossless detection method that is located at face of weld.
It is that naked eyes or low power magnifying glass can detect and be located at weld seam a further improvement of the present invention lies in that in step S2
The defect on surface includes that weld size is not inconsistent specification, undercut, overlap, arc crater, surface pores, slag inclusion and face crack;
In step S2, the tubing internal flaw that lossless detection method detects includes stomata, slag inclusion, underbead crack, does not weld
Saturating and incomplete tusion.
A further improvement of the present invention lies in that sorting algorithm uses k- nearest neighbor algorithm in step S3, k- nearest neighbor algorithm includes
Following steps: (1) weld seam detection tested point is calculated at a distance from the point that known weld defect categorical data is concentrated;(2) according to away from
From ascending sort;(3) it chooses with weld seam detection tested point apart from the smallest k point;(4) the affiliated defect classification of k point before calculating
The frequency of occurrences;(5) using the highest defect classification of frequency as the classification of weld seam tested point, classification prediction model is obtained;
Regression algorithm in step S3 specifically: be fitted, returned for the data point in welding defect database
Return prediction model;
Metal welding defect type in step S3 includes weld geometry sizes problem, undercut defect, heat affected area contraction
Crackle, arc crater, slag inclusion, stomata and lack of penetration;
For heat affected area contraction crack or incomplete penetration defect problem, characterization is given to defect using sorting algorithm, specifically
Are as follows: defect existence or non-existence is predicted according to weld seam range estimation and non-destructive testing value, output is 0 or 1 two discrete value,
In, 0 represents zero defect, 1 represent it is defective;
For weld seam deviation, undercut, arc crater, slag inclusion and gas hole defect problem, defect is fitted using regression algorithm
And it continuously characterizes.
A further improvement of the present invention lies in that composite bimetal pipe welding defect type includes dilution of the lining by base
It is crisp that weld seam crystallization crackle caused by effect, composite material local melting cause alloying element to penetrate into the melting area of welding seam caused by weld seam
Change.
A further improvement of the present invention lies in that the detailed process of step S4 the following steps are included:
(1) k object is arbitrarily chosen as in initial cluster from n technological parameter mathematic vector for being actually formed weld defect
The heart,
(2) according to the mean value of object in cluster, each object is calculated at a distance from these cluster centers, by each object assignment
To most like cluster;
(3) cluster mean value is updated, that is, calculates the mean value of object in each cluster;
(4) circulation step (2) and step (3) obtain Forecast when each cluster criterion function is no longer changed
Model.
A further improvement of the present invention lies in that in step S6, for sorting algorithm prediction output with or without result it is different
It causes, is then inaccuracy, is unanimously then accurate;
Threshold value for regression algorithm and clustering algorithm using 5% error range as accuracy of judgement with inaccuracy.
A further improvement of the present invention lies in that the process for carrying out cross validation is as follows in step S6:
1) from the sample of random selection s is remaining as training set train in random selection in whole training data S
As test set test;
2) by obtaining assuming function or model to test set training;
3) in test set to each sample according to function or model is assumed, obtain the category of training set, find out classification
Accuracy;
4) selection has the model or hypothesis of maximum classification accuracy rate.
Compared with prior art, the invention has the advantages that: the present invention is based on the welderings of actually detected bimetal tube
It connects technological parameter data and corresponding welding and actually generates weld defect data, according to bimetal tube welding defect specific features,
Implement machine learning respectively by classification, recurrence and improvement clustering algorithm method, establish bimetal tube welding defect prediction model,
It is carried out according to generation degree of the bimetal tube welding defect prediction model to weld defect under the conditions of default welding condition
Prediction;It is counted in conjunction with On-site Welding Technology parameter and defects detection distributed data, weld defect is established by Monte Carlo simulation
Prediction model, according to desired defect Con trolling index is established the considerations of taking into account quality and cost, using Revised genetic algorithum, with pre-
Phase weld defect Con trolling index is that optimization aim using welding condition as optimal design parameter optimizes calculating, is obtained pair
Control the welding condition optimum results that defect occurs.So as to improve the accuracy of weld defect Occurrence forecast, effectively
Control the probability of happening of defect.
The present invention is directed to the rule and feature that composite bimetal pipe girth joint welding defect occurs, and is based on actual welding
Defect a situation arises big data, comprehensive machine learning and Optimization Theory method distinguish supervision according to different welding defects
With non-supervisory class Machine Learning Problems, the comprehensive weldering for implementing based on knowledge engineering using classification, recurrence and improved clustering algorithm
Failure prediction is connect, welding defect control optimization is implemented using Revised genetic algorithum, proposes oil-gas gathering and transportation composite bimetal pipe
The prediction and control method of girth joint welding defect.
Detailed description of the invention
Fig. 1 is the prediction and control method flow chart of composite bimetal pipe welding defect provided in an embodiment of the present invention.
Specific embodiment
To describe the technical solutions in the embodiments of the present invention more clearly, make required in being described below to embodiment
Attached drawing is briefly described, it should be apparent that, drawings in the following description are only some embodiments of the invention, for
For those of ordinary skill in the art, without creative efforts, it can also be obtained according to these attached drawings other
Attached drawing.
Based on actual welding defect a situation arises big data, comprehensive machine learning and Optimization Theory method, according to
Sorting algorithm, regression algorithm or clustering algorithm is respectively adopted in different welding defects, establishes and is used as by welding condition because becoming
Then the bimetal tube welding defect prediction model measured, weld all kinds of defective datas as variable is implemented excellent using genetic algorithm
Change, realizes the prediction and control of oil-gas gathering and transportation composite bimetal pipe girth joint welding defect.
For rule and feature that composite bimetal pipe girth joint welding defect occurs, sent out based on actual welding defect
Raw situation big data, comprehensive machine learning and Optimization Theory method ask supervision class according to different welding defects
Topic, using classification and regression algorithm.
For non-supervisory class problem, is predicted using the welding defect that clustering algorithm carries out based on knowledge engineering, then used
Genetic algorithm implements welding defect control optimization, realizes the prediction of oil-gas gathering and transportation composite bimetal pipe girth joint welding defect
With control.
The present invention implements welding technology optimization and welding using genetic algorithm, and process includes genetic coding generation, random initial population
It determines, the Optimized Iterative process that Fitness analysis and progeny population calculate.Although genetic operator Fitness analysis process can be with
It is continuouslyd optimize by the Pareto value that non-dominated ranking obtains all heredity individuals, but the population around individual can not be assessed
Density information greatly reduces excellent to will appear the different individual of optimization aim gradient hereditary chance still having the same
Change computational efficiency;For this problem genetic algorithm is proposed to improve: on the basis of assessing all individual Pareto values, be introduced
The mean value and standard deviation of individual Pareto value, thus assess optimization aim gradient to around individual population density information it is sensitive
Degree makes offspring's breeding of individual tend to sluggish region far from optimization, achievees the purpose that improve optimization computational efficiency.
Referring to Fig. 1, the present invention specifically includes the following steps:
S1 constructs database:
The On-site Welding Technology parameter of multiple groups unlike material, grade of steel, specification and bimetallic compound type and right is stored in advance
The weld seam actual defects data answered.
The On-site Welding Technology supplemental characteristic specifically includes that welding current, weld groove straightness, bevel angle, dress
It is dry with gap, arc manipulation speed, base material and wlding chemical component, welding end of a period blow-out time, weld seam base material clean-up performance, welding rod
Degree etc..Weld seam actual defects data, comprising: naked eyes or low power magnifying glass can detect and be located at face of weld defect,
Such as weld size is not inconsistent specification, undercut, overlap, arc crater, surface pores, slag inclusion, face crack;Destructive testing or lossless inspection
The tubing internal flaw that survey method detects, such as stomata, slag inclusion, underbead crack, lack of penetration, incomplete tusion.To actual defects number
The book of final entry is carried out according to statisticalling analyze respectively, or according to with/without situation, or carries out continuous data recording according to size or quantity,
And carry out classification grading.These are knowledge, there is prediction rule;But defect peculiar for two metal solder, it is difficult to predict
Rule needs to be classified automatically using clustering algorithm and found rule, obtains prediction model.
Then the bimetal tube welding of the several groups material of the same race, specification, grade of steel and bimetallic compound type of storage is chosen
Technological parameter data and its corresponding welding actual defects analyze data, and to the data of selection carry out normalization and be abstracted as to
Data after one vectorization of rule are divided into defect training data sample and verify data sample by amount;
S2 carries out statistic of classification analysis to welding defect problem:
More clear for defect producing cause, i.e., a given group welding technological parameter and defective data can be according to theories
The specific welding input of understanding exports particular kind of relationship with defect, thus it is speculated that the case where going out welding defect result judges defect for gold
Belong to welding defect type or composite bimetal pipe welding defect type;
S3 welds defect type for metal, carries out machine learning using supervised learning mode, by sorting algorithm or returns
Reduction method is handled, and classification prediction model or regressive prediction model are obtained;The problem of meeting this kind of situation includes such as: weld seam
Geometric dimension problem, undercut defect, heat affected area contraction crack, arc crater, slag inclusion, stomata, lack of penetration etc.;
For supervising class problem, is handled according to two methods of sorting algorithm and regression algorithm, obtain classification prediction mould
Type and regressive prediction model;Discrete defect classification results are exported in classification problem according to welding parameter input variable, such as right
Weldquality has the defects of crackle seriously affected, lack of penetration, statistical analysis bead crack, lack of penetration etc. occur with or without
Situation;Continuous defective data structure is exported according to welding parameter input variable in regression problem, such as welding is not inconsistent standard
Geometrical deviation size, undercut geometric parameter, arc crater quantity and size, slag inclusion quantity and size, stomata quantity and size etc.;
For crackle, it is lack of penetration the defects of problem, characterization is given to defect using sorting algorithm, is generally estimated according to weld seam
And non-destructive testing value predicts defect existence or non-existence, output is 0 or 1 two discrete value.(0 represents zero defect, and 1 represents
It is defective);
Regression algorithm uses k- nearest neighbor algorithm (KNN), and the method using sample characteristics data range measurement is classified.From
Welding defect sample data concentration is specified crackle existing for each defective data or lack of penetration equal labels in sample set, that is, is known
The corresponding relationship of each group of defective data and affiliated classification in sample set.After inputting the not new data of label, by the every of data
Corresponding with the sample intensive data feature of a feature is carried out apart from calculating, then extracts in sample set the most like data of feature (most
Neighbour) tag along sort.K most like data are selected, the most classification of selection frequency of occurrence in from k number, as new
The classification of data.Key step includes: the point that (1) calculates that weld seam detection tested point and known weld defect categorical data are concentrated
Distance;(2) according to apart from ascending sort;(3) it chooses with weld seam detection tested point apart from the smallest k point;(4) k before calculating
The frequency of occurrences of defect classification belonging to point;(5) classification of the highest defect classification of return frequency as weld seam tested point, it obtains
Classification prediction model;
The defects of for weld seam deviation, undercut, arc crater, slag inclusion and stomata problem, using regression algorithm to defect characterize, needle
Data point in welding defect database is fitted, regressive prediction model is obtained,
To typical linear regression model y=1x1i+2x2i+…+mxmi, using least square method, and it is based on training data sample
This, determines aj, wherein xji substitutes into the data of j-th of influence factor in i-th group of training data sample, yiSubstitute into i-th group of instruction
Practice data sample in weld seam actual defects data, j=1,2 ..., m;Specific defect data include: geometrical deviation size, undercut
Geometric parameter, arc crater quantity and size, slag inclusion quantity and size, stomata quantity and size etc.;
S4 carries out statistic of classification analysis to welding defect problem, other than supervising class problem, for defect producing cause
Experience and theoretic knowledge achievement indefinite, can not use for reference, welding input without clear particular kind of relationship, that is, are given with defect output
The case where determining a group welding technological parameter data, cannot suppose that out welding defect result welds composite bimetal pipe and lacks
Type is fallen into, machine learning is carried out using unsupervised learning mode, and carry out organized cluster and clustering, establishes Forecast
Model;
The problem of meeting this kind of situation includes such as: the bimetal tube that some reason complicated difficults are surveyed welds special defect, packet
Include weld seam crystallization crackle caused by diluting effect of the lining by base, composite material local melting causes alloying element to penetrate into weld seam
Caused by the melting area of welding seam embrittlement etc.;
For unsupervised class problem, the inner link that welding procedure and defect generate can be found by machine learning, be based on
Welding parameter and actual welding defect big data analysis constantly refine key element, and carry out organized cluster and clustering,
Establish the bug prediction model of new based on knowledge engineering;Detailed process are as follows: bimetal tube welds solidification cracking and weld material
The defects of brittle, is influenced by welding procedure multidimensional data complicated coupling, and the prior art is insufficient, it is difficult to determine inherent mechanism.According to non-
Supervised learning issue handling implements machine learning using the principle of K-means clustering algorithm and prediction models, and detailed process is such as
Under:
(1) k object is arbitrarily chosen as in initial cluster from n technological parameter mathematic vector for being actually formed weld defect
The heart,
(2) according to the mean value of object in cluster, each object is calculated at a distance from these cluster centers, by each object assignment
To most like cluster,
(3) cluster mean value is updated, that is, calculates the mean value of object in each cluster,
(4) circulation step (2) and step (3) are no longer changed until each cluster criterion function, obtain Forecast mould
Type.
It is more quick to the initial value and selection number at initial cluster center in view of the prediction accuracy of K-means clustering algorithm
Sense, for this problem propose improve: to select the K point of characteristic value spacing as far as possible as initial cluster as principle, first with
Machine selects a sample number strong point as first initial classes cluster central point, then selects that point farthest apart from the data point
As second initial classes cluster central point, then minimum distance of the reselection apart from the first two data point is maximum data point work
For the central point of third initial classes cluster, and so on, until selecting K object as initial cluster center.
S5, the welding defect data of standardization are corresponding with welding condition, be uniformly input to classification prediction model,
Regressive prediction model or Forecast model are enriched constantly knowledge base by Model Self-Learning, improve precision of prediction;To
Mould is predicted as the bimetal tube welding defect of variable as dependent variable, all kinds of defective datas of welding to by welding condition
Type;
Using sorting algorithm: there is the crackle seriously affected to weldquality, crackle only has presence that cannot tolerate, so
It can be divided into or without two classes.The influence factor of crackle has, and weld heat-affected zone generates big stress, base material containing hardening after shrinking
Organize it is more cause it is cooling after easily raw crackle, higher, other harmful element impurity effects of hydrogen concentration etc. in weld seam.These factors are
Input variable obtains prediction model using classification method by detection data collection.
Using classification method, for it is lack of penetration the defects of: the angle of the groove or gap of group pair is too small, groove two sides and layer
Between weld the oxide that do not clean out and slag and hinder sufficiently to fuse between metal, select welding current is excessive to cause welding rod prior to mother
Material fusing, excessive velocities, the angle of welding rod of arc manipulation are incorrect when selecting too little current leads to melt amesiality and formation office
Portion is lack of penetration etc..
For defect regression problem, continuous defective data structure is exported according to welding parameter input variable, such as weld
Be not inconsistent geometrical deviation size, undercut geometric parameter, arc crater quantity and the size of standard, slag inclusion quantity and size, stomata quantity and
Size etc.;)
S6 carries out the bimetal tube welding defect prediction model in step S5 using the verify data sample in step S1
Cross validation, the model prediction for inaccuracy is as a result, then more new database, obtains defect training data sample and verify data
Verify data sample is re-entered the bimetal tube welding defect prediction model into step S5 and carries out cross validation by sample,
The correctness of model prediction is continuously improved;
The step of cross validation, is as follows: 1, from the sample conduct for randomly choosing s in whole training data S in random selection
Training set train, remaining conduct test set test.2, by obtaining assuming function or model to test set training.3, exist
Test set, according to function or model is assumed, obtains the category of training set, finds out classification accuracy rate to each sample.4, it selects
Model or hypothesis with maximum classification rate.This method is known as hold-out cross validation or is known as letter
Characteristics for Single Staggered verifying.Since test set and training set are separated, the phenomenon that avoiding over-fitting.)
For sorting algorithm prediction output with or without result it is inconsistent, then be inaccuracy, be unanimously then accurate;
For regression algorithm and clustering algorithm, using 5% error range as accuracy of judgement with inaccuracy threshold value.
Step S6 is repeated several times in S7, finally obtains welding defect prediction model, and to lacking under specified welding condition
It falls into generation degree to be predicted, exports weld defect predicted value;Using the present invention, can be built based on the practical big data of welding defect
Vertical bug prediction model, improves the accuracy of composite bimetal pipe welding defect prediction;
Weld defect predicted value and target defect controlling value are compared difference as objective function, welding condition by S8
As design variable, iteration is optimized with genetic algorithm, finally meets the requirement of Engineering Control target, output control defect
The welding condition Optimum Design Results of degree, to control the generation of defect.
The present invention can carry out the prediction of welding defect, and can also by the up-to-standard target of site welding be constructed into
This control comprehensive analysis determines the objective function of bimetal tube site welding quality control, i.e., will be calculated using the above method
Weld defect predicted value and target defect controlling value compare difference as objective function, welding condition is as design change
Amount, optimizes iteration with Revised genetic algorithum, finally meets the requirement of Engineering Control target, output control defect level
Welding condition Optimum Design Results, effectively control defect generation;It can be established by Monte Carlo simulation based on response
The weld defect prediction model of surface function takes into account the precision and efficiency of weld defect optimal control.
Claims (10)
1. oil-gas gathering and transportation composite bimetal pipe girth joint failure prediction and control method, which is characterized in that based on actual
Welding defect a situation arises big data, comprehensive machine learning and Optimization Theory method, according to different welding defects, respectively
Using sorting algorithm, regression algorithm or clustering algorithm, establish by welding condition as dependent variable, all kinds of defective datas of welding
As the bimetal tube welding defect prediction model of variable, optimization is then implemented using genetic algorithm, it is double to realize that oil-gas gathering and transportation is used
The prediction and control of metal composite pipe girth joint welding defect.
2. oil-gas gathering and transportation according to claim 1 composite bimetal pipe girth joint failure prediction and control method,
It is characterized in that, welding technology optimization and welding is implemented using genetic algorithm, realize that oil-gas gathering and transportation is welded with composite bimetal pipe girth joint
The process of prediction and the control of defect includes genetic coding generation, the determination of random initial population, Fitness analysis and filial generation kind
The Optimized Iterative process that group calculates introduces the mean value of individual Pareto value on the basis of assessing all individual Pareto values
And standard deviation, by mean value and standard deviation the assessment optimization aim gradient of individual Pareto value to population density information around individual
Sensitivity makes offspring's breeding of individual tend to sluggish region far from optimization;Foundation occurs as optimizing to reduce welding defect
Target realizes that change welding condition lacks to welding is reduced using welding condition as the Optimized model of optimization design variable
Sunken control.
3. oil-gas gathering and transportation according to claim 1 composite bimetal pipe girth joint failure prediction and control method,
It is characterized in that, specifically includes the following steps:
The On-site Welding Technology parameter and correspondence of multiple groups unlike material, grade of steel, specification and bimetallic compound type is stored in advance in S1
Weld seam actual defects data;
Weld seam actual defects data are statisticallyd analyze respectively, or carry out the book of final entry according to with/without situation, or according to size or number
Amount carries out continuous data recording, and carries out classification grading;
Then the several groups material of the same race, specification, the bimetal tube welding procedure of grade of steel and bimetallic compound type of storage are chosen
Supplemental characteristic and its corresponding weld seam actual defects data, and normalization is carried out to the data of selection and is abstracted as vector, it will advise
Data after one vectorization are divided into defect training data sample and verify data sample;
S2 sentences a given group welding technological parameter and defective data according to the analysis of welding input and defect output relation
Breakthrough, which falls into, welds defect type or composite bimetal pipe welding defect type for metal;
S3 welds defect type for metal, carries out machine learning using supervised learning mode, by sorting algorithm or returns calculation
Method is handled, and classification prediction model or regressive prediction model are obtained;
S4 carries out machine learning using unsupervised learning mode, and carried out group for composite bimetal pipe welding defect type
Cluster and clustering are knitted, Forecast model is established;
S5, welding defect data are corresponding with welding condition, uniformly it is input to classification prediction model, regressive prediction model
Or Forecast model, by Model Self-Learning, to obtain by welding condition as dependent variable, all kinds of defects of welding
Bimetal tube welding defect prediction model of the data as variable;
S6 intersects the bimetal tube welding defect prediction model in step S5 using the verify data sample in step S1
Verifying, the model prediction for inaccuracy is as a result, then more new database, obtains defect training data sample and verify data sample
This, re-enters the bimetal tube welding defect prediction model into step S5 for verify data sample and carries out cross validation, no
The disconnected correctness for improving model prediction;
Step S6 is repeated several times in S7, finally obtains welding defect prediction model, and produce to the defect under specified welding condition
Raw degree is predicted, weld defect predicted value is exported;
Weld defect predicted value and target defect controlling value are compared difference as objective function, welding condition conduct by S8
Design variable optimizes iteration with genetic algorithm, finally meets the requirement of Engineering Control target, output control defect level
Welding condition Optimum Design Results, to control the generation of defect.
4. oil-gas gathering and transportation according to claim 3 composite bimetal pipe girth joint failure prediction and control method,
Be characterized in that, in step S1, On-site Welding Technology supplemental characteristic include: welding current, weld groove straightness, bevel angle,
Fit-up gap, arc manipulation speed, base material and wlding chemical component, welding end of a period blow-out time, weld seam base material clean-up performance and weldering
Degree of drying;Weld seam actual defects data include: naked eyes or low power magnifying glass can detect and be located at face of weld lack
The tubing internal flaw that sunken, destructive testing or lossless detection method detect.
5. oil-gas gathering and transportation according to claim 4 composite bimetal pipe girth joint failure prediction and control method,
Be characterized in that, in step S2, naked eyes or low power magnifying glass can detect and be located at face of weld defect include weld size
It is not inconsistent specification, undercut, overlap, arc crater, surface pores, slag inclusion and face crack;
In step S2, the tubing internal flaw that lossless detection method detects include stomata, slag inclusion, underbead crack, it is lack of penetration with
And incomplete tusion.
6. oil-gas gathering and transportation according to claim 4 composite bimetal pipe girth joint failure prediction and control method,
It is characterized in that, in step S3, sorting algorithm uses k- nearest neighbor algorithm, and k- nearest neighbor algorithm is the following steps are included: (1) calculates weld seam inspection
Tested point is surveyed at a distance from the point that known weld defect categorical data is concentrated;(2) according to apart from ascending sort;(3) it chooses and welds
Seam detection tested point is apart from the smallest k point;(4) frequency of occurrences of the affiliated defect classification of k point before calculating;(5) most by frequency
Classification of the high defect classification as weld seam tested point obtains classification prediction model;
Regression algorithm in step S3 specifically: be fitted for the data point in welding defect database, obtain returning pre-
Survey model;
In step S3 metal welding defect type include weld geometry sizes problem, undercut defect, heat affected area contraction crack,
Arc crater, slag inclusion, stomata and lack of penetration;
For heat affected area contraction crack or incomplete penetration defect problem, characterization is given to defect using sorting algorithm, specifically: root
Defect existence or non-existence is predicted according to weld seam range estimation and non-destructive testing value, and output is 0 or 1 two discrete value, wherein 0 generation
Table zero defect, 1 represent it is defective;
For weld seam deviation, undercut, arc crater, slag inclusion and gas hole defect problem, defect is fitted and is connected using regression algorithm
Continued sign.
7. oil-gas gathering and transportation according to claim 4 composite bimetal pipe girth joint failure prediction and control method,
It is characterized in that, composite bimetal pipe welding defect type includes that weld seam crystallization caused by diluting effect of the lining by base is split
Line, composite material local melting cause alloying element to penetrate into the embrittlement of the melting area of welding seam caused by weld seam.
8. oil-gas gathering and transportation according to claim 7 composite bimetal pipe girth joint failure prediction and control method,
Be characterized in that, the detailed process of step S4 the following steps are included:
(1) k object is arbitrarily chosen as initial cluster center from n technological parameter mathematic vector for being actually formed weld defect,
(2) according to the mean value of object in cluster, each object is calculated at a distance from these cluster centers, by each object assignment to most
Similar cluster;
(3) cluster mean value is updated, that is, calculates the mean value of object in each cluster;
(4) circulation step (2) and step (3) obtain Forecast model when each cluster criterion function is no longer changed.
9. oil-gas gathering and transportation according to claim 4 composite bimetal pipe girth joint failure prediction and control method,
Be characterized in that, in step S6, for sorting algorithm prediction output with or without result it is inconsistent, then be inaccuracy, be unanimously then
Accurately;
Threshold value for regression algorithm and clustering algorithm using 5% error range as accuracy of judgement with inaccuracy.
10. oil-gas gathering and transportation according to claim 4 composite bimetal pipe girth joint failure prediction and control method,
It is characterized in that, in step S6, the process for carrying out cross validation is as follows:
1) from the sample of s is randomly choosed in whole training data S in random selection as training set train, remaining conduct
Test set test;
2) by obtaining assuming function or model to test set training;
3) in test set to each sample according to function or model is assumed, obtain the category of training set, it is correct to find out classification
Rate;
4) selection has the model or hypothesis of maximum classification accuracy rate.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201910368418.0A CN110116254B (en) | 2019-05-05 | 2019-05-05 | Method for predicting and controlling defects of annular welding joint of bimetal composite pipe for oil and gas gathering and transportation |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201910368418.0A CN110116254B (en) | 2019-05-05 | 2019-05-05 | Method for predicting and controlling defects of annular welding joint of bimetal composite pipe for oil and gas gathering and transportation |
Publications (2)
Publication Number | Publication Date |
---|---|
CN110116254A true CN110116254A (en) | 2019-08-13 |
CN110116254B CN110116254B (en) | 2021-07-02 |
Family
ID=67520300
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN201910368418.0A Active CN110116254B (en) | 2019-05-05 | 2019-05-05 | Method for predicting and controlling defects of annular welding joint of bimetal composite pipe for oil and gas gathering and transportation |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN110116254B (en) |
Cited By (18)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN111168195A (en) * | 2019-11-29 | 2020-05-19 | 常熟理工学院 | IPv 6-based electric welding machine cluster monitoring control method |
CN111178156A (en) * | 2019-12-10 | 2020-05-19 | 网络通信与安全紫金山实验室 | Time sequence characteristic effective window extraction method applied to machine learning |
CN111291071A (en) * | 2020-01-21 | 2020-06-16 | 北京字节跳动网络技术有限公司 | Data processing method and device and electronic equipment |
RU2732546C1 (en) * | 2019-12-30 | 2020-09-21 | федеральное государственное бюджетное образовательное учреждение высшего образования "Московский государственный технический университет имени Н.Э. Баумана (национальный исследовательский университет)" (МГТУ им. Н.Э. Баумана) | Device for monitoring and controlling complex of pulsed laser deposition |
CN112052554A (en) * | 2020-07-23 | 2020-12-08 | 中国石油天然气集团有限公司 | Method for establishing self-height prediction model of pipeline buried defects |
CN112184693A (en) * | 2020-10-13 | 2021-01-05 | 东北大学 | Intelligent detection method for weld defects of ray industrial negative |
CN112733406A (en) * | 2021-01-20 | 2021-04-30 | 昆山六丰机械工业有限公司 | Method for establishing friction stir welding structure defect prediction model by using finite element method |
CN113076817A (en) * | 2021-03-17 | 2021-07-06 | 上海展湾信息科技有限公司 | Weld pore defect real-time detection method and system |
CN113486457A (en) * | 2021-06-04 | 2021-10-08 | 宁波海天金属成型设备有限公司 | Die casting defect prediction and diagnosis system |
CN113642253A (en) * | 2021-08-30 | 2021-11-12 | 佛山众陶联供应链服务有限公司 | Method and system for judging loss of ceramic tile semi-finished product in processing |
CN113762410A (en) * | 2021-09-17 | 2021-12-07 | 哈尔滨工业大学 | Key welding process parameter prediction method based on machine learning |
CN114216915A (en) * | 2021-12-15 | 2022-03-22 | 江门市浩远科技有限公司 | Method for detecting and classifying types of stains and defects of circuit board based on class level |
CN114345839A (en) * | 2021-12-28 | 2022-04-15 | 北京航星机器制造有限公司 | Method and system for cleaning longitudinal seam of titanium alloy cylinder |
CN114378422A (en) * | 2020-10-21 | 2022-04-22 | 天津大学 | Comprehensive adjusting method for assembling quality of friction stir welding joint |
CN115609180A (en) * | 2022-12-01 | 2023-01-17 | 江苏新恒基特种装备股份有限公司 | Strengthening parameter generation method and system based on welding process prediction and storage medium |
CN116300475A (en) * | 2023-05-11 | 2023-06-23 | 苏州先准电子科技有限公司 | Metal rolling control method and system |
CN117436769A (en) * | 2023-12-20 | 2024-01-23 | 山东方垠智能制造有限公司 | Structural part welding quality monitoring method, system, storage medium and equipment |
CN112052554B (en) * | 2020-07-23 | 2024-04-30 | 中国石油天然气集团有限公司 | Method for establishing height prediction model of buried defect of pipeline |
Families Citing this family (1)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
JP2024010290A (en) * | 2022-07-12 | 2024-01-24 | 三菱重工業株式会社 | Welding support device, welding support method and program |
Citations (6)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN102201020A (en) * | 2010-03-22 | 2011-09-28 | 爱发股份有限公司 | Methods and systems for numerically predicting surface imperfections on stamped sheet metal parts |
US20140282288A1 (en) * | 2013-03-15 | 2014-09-18 | Globalfoundries Singapore Pte. Ltd. | Design-for-manufacturing - design-enabled-manufacturing (dfm-dem) proactive integrated manufacturing flow |
CN107391890A (en) * | 2017-09-01 | 2017-11-24 | 东营市永利精工石油机械制造有限公司 | A kind of oil bushing threaded connector machines prediction and the optimal control method for line defect of quivering |
CN107784293A (en) * | 2017-11-13 | 2018-03-09 | 中国矿业大学(北京) | A kind of Human bodys' response method classified based on global characteristics and rarefaction representation |
CN108090461A (en) * | 2017-12-29 | 2018-05-29 | 浙江大学宁波理工学院 | Three-dimensional face identification method based on sparse features |
CN109711474A (en) * | 2018-12-24 | 2019-05-03 | 中山大学 | A kind of aluminium material surface defects detection algorithm based on deep learning |
-
2019
- 2019-05-05 CN CN201910368418.0A patent/CN110116254B/en active Active
Patent Citations (6)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN102201020A (en) * | 2010-03-22 | 2011-09-28 | 爱发股份有限公司 | Methods and systems for numerically predicting surface imperfections on stamped sheet metal parts |
US20140282288A1 (en) * | 2013-03-15 | 2014-09-18 | Globalfoundries Singapore Pte. Ltd. | Design-for-manufacturing - design-enabled-manufacturing (dfm-dem) proactive integrated manufacturing flow |
CN107391890A (en) * | 2017-09-01 | 2017-11-24 | 东营市永利精工石油机械制造有限公司 | A kind of oil bushing threaded connector machines prediction and the optimal control method for line defect of quivering |
CN107784293A (en) * | 2017-11-13 | 2018-03-09 | 中国矿业大学(北京) | A kind of Human bodys' response method classified based on global characteristics and rarefaction representation |
CN108090461A (en) * | 2017-12-29 | 2018-05-29 | 浙江大学宁波理工学院 | Three-dimensional face identification method based on sparse features |
CN109711474A (en) * | 2018-12-24 | 2019-05-03 | 中山大学 | A kind of aluminium material surface defects detection algorithm based on deep learning |
Cited By (27)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN111168195B (en) * | 2019-11-29 | 2021-11-23 | 常熟理工学院 | IPv 6-based electric welding machine cluster monitoring control method |
CN111168195A (en) * | 2019-11-29 | 2020-05-19 | 常熟理工学院 | IPv 6-based electric welding machine cluster monitoring control method |
CN111178156A (en) * | 2019-12-10 | 2020-05-19 | 网络通信与安全紫金山实验室 | Time sequence characteristic effective window extraction method applied to machine learning |
RU2732546C1 (en) * | 2019-12-30 | 2020-09-21 | федеральное государственное бюджетное образовательное учреждение высшего образования "Московский государственный технический университет имени Н.Э. Баумана (национальный исследовательский университет)" (МГТУ им. Н.Э. Баумана) | Device for monitoring and controlling complex of pulsed laser deposition |
CN111291071A (en) * | 2020-01-21 | 2020-06-16 | 北京字节跳动网络技术有限公司 | Data processing method and device and electronic equipment |
CN111291071B (en) * | 2020-01-21 | 2023-10-17 | 北京字节跳动网络技术有限公司 | Data processing method and device and electronic equipment |
CN112052554A (en) * | 2020-07-23 | 2020-12-08 | 中国石油天然气集团有限公司 | Method for establishing self-height prediction model of pipeline buried defects |
CN112052554B (en) * | 2020-07-23 | 2024-04-30 | 中国石油天然气集团有限公司 | Method for establishing height prediction model of buried defect of pipeline |
CN112184693A (en) * | 2020-10-13 | 2021-01-05 | 东北大学 | Intelligent detection method for weld defects of ray industrial negative |
CN112184693B (en) * | 2020-10-13 | 2023-10-24 | 东北大学 | Intelligent detection method for welding line defects of ray industrial negative film |
CN114378422A (en) * | 2020-10-21 | 2022-04-22 | 天津大学 | Comprehensive adjusting method for assembling quality of friction stir welding joint |
CN114378422B (en) * | 2020-10-21 | 2023-03-14 | 天津大学 | Comprehensive adjusting method for assembling quality of friction stir welding joint |
CN112733406A (en) * | 2021-01-20 | 2021-04-30 | 昆山六丰机械工业有限公司 | Method for establishing friction stir welding structure defect prediction model by using finite element method |
CN113076817A (en) * | 2021-03-17 | 2021-07-06 | 上海展湾信息科技有限公司 | Weld pore defect real-time detection method and system |
CN113076817B (en) * | 2021-03-17 | 2022-11-04 | 上海展湾信息科技有限公司 | Weld pore defect real-time detection method and system |
CN113486457A (en) * | 2021-06-04 | 2021-10-08 | 宁波海天金属成型设备有限公司 | Die casting defect prediction and diagnosis system |
CN113642253A (en) * | 2021-08-30 | 2021-11-12 | 佛山众陶联供应链服务有限公司 | Method and system for judging loss of ceramic tile semi-finished product in processing |
CN113642253B (en) * | 2021-08-30 | 2024-04-02 | 佛山众陶联供应链服务有限公司 | Loss judging method and system for ceramic tile semi-finished product in processing |
CN113762410A (en) * | 2021-09-17 | 2021-12-07 | 哈尔滨工业大学 | Key welding process parameter prediction method based on machine learning |
CN113762410B (en) * | 2021-09-17 | 2023-12-08 | 哈尔滨工业大学 | Key welding process parameter prediction method based on machine learning |
CN114216915A (en) * | 2021-12-15 | 2022-03-22 | 江门市浩远科技有限公司 | Method for detecting and classifying types of stains and defects of circuit board based on class level |
CN114216915B (en) * | 2021-12-15 | 2024-03-29 | 江门市浩远科技有限公司 | Method for detecting and classifying class levels based on stains and defect types of circuit board |
CN114345839A (en) * | 2021-12-28 | 2022-04-15 | 北京航星机器制造有限公司 | Method and system for cleaning longitudinal seam of titanium alloy cylinder |
CN115609180A (en) * | 2022-12-01 | 2023-01-17 | 江苏新恒基特种装备股份有限公司 | Strengthening parameter generation method and system based on welding process prediction and storage medium |
CN116300475B (en) * | 2023-05-11 | 2023-08-22 | 苏州先准电子科技有限公司 | Metal rolling control method and system |
CN116300475A (en) * | 2023-05-11 | 2023-06-23 | 苏州先准电子科技有限公司 | Metal rolling control method and system |
CN117436769A (en) * | 2023-12-20 | 2024-01-23 | 山东方垠智能制造有限公司 | Structural part welding quality monitoring method, system, storage medium and equipment |
Also Published As
Publication number | Publication date |
---|---|
CN110116254B (en) | 2021-07-02 |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
CN110116254A (en) | Oil-gas gathering and transportation composite bimetal pipe girth joint failure prediction and control method | |
CN109086804B (en) | Hydraulic equipment early failure prediction method based on fusion of multi-source state monitoring information and reliability characteristics | |
Dhas et al. | A review on optimization of welding process | |
CN107391890B (en) | Prediction and optimal control method for oil casing threaded joint machining chatter defect | |
CN113469241B (en) | Product quality control method based on process network model and machine learning algorithm | |
CN109308556A (en) | A kind of pipeline evaluation method and device based on interior detection data | |
WO2009110795A1 (en) | Method for non-destructive inspection | |
CN115063051B (en) | Steel pipe performance evaluation method and system based on metal element analysis | |
CN109977477A (en) | Based on the Utility Boiler Superheater health state evaluation method for improving Fuzzy Level Analytic Approach | |
Zelmati et al. | A probabilistic approach to estimate the remaining life and reliability of corroded pipelines | |
CN112633611A (en) | Submarine cable state maintenance strategy optimization method and system based on big data analysis | |
CN115689114A (en) | Submarine cable running state prediction method based on combined neural network | |
Sreeraj et al. | Simulation and parameter optimization of GMAW process using neural networks and particle swarm optimization algorithm | |
CN116822342A (en) | Process identification and performance prediction method for zirconium alloy laser cutting | |
Vasudevan et al. | Genetic algorithm for optimisation of A-TIG welding process for modified 9Cr–1Mo steel | |
Yatian et al. | Residual life prediction method for remanufacturing sucker rods based on magnetic memory testing and a support vector machine model | |
CN102289597A (en) | Method for identifying pre-stressed secondary tensioning inflection point | |
Roy et al. | An approach for solving multi characteristics optimization of submerged are welding process parameters by using grey based genetic algorithm | |
CN113554336A (en) | Evaluation method of ship manufacturing process | |
Seneviratne et al. | In-service inspection of static mechanical equipment: Use of a fuzzy inference system for maintaining the quality of an inspection program | |
CN110490452B (en) | Multi-index comprehensive grading method for aging state of overhead line | |
Kariyawasam et al. | A Data Driven Validation of a Defect Assessment Model and its Safe Implementation | |
Lee et al. | Study on Real-Time Porosity Defect Detection Through Neural Network Structure Optimization using Genetic Algorithm in GMAW | |
Uruchurtu-Chavarin et al. | Artificial intelligence for the assessment on the corrosion conditions diagnosis of transmission line tower foundations | |
CN117592223B (en) | Intelligent design method of hole machining tool for aerospace materials |
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 |