CN109977511A - Method based on artificial intelligence big data prediction Pressurized Plastic Pipes long term life - Google Patents
Method based on artificial intelligence big data prediction Pressurized Plastic Pipes long term life Download PDFInfo
- Publication number
- CN109977511A CN109977511A CN201910201622.3A CN201910201622A CN109977511A CN 109977511 A CN109977511 A CN 109977511A CN 201910201622 A CN201910201622 A CN 201910201622A CN 109977511 A CN109977511 A CN 109977511A
- Authority
- CN
- China
- Prior art keywords
- plastic pipes
- pressurized plastic
- prediction
- big data
- life
- 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.)
- Pending
Links
- 229920003023 plastic Polymers 0.000 title claims abstract description 101
- 239000004033 plastic Substances 0.000 title claims abstract description 101
- 238000000034 method Methods 0.000 title claims abstract description 64
- 230000007774 longterm Effects 0.000 title claims abstract description 29
- 238000013473 artificial intelligence Methods 0.000 title claims abstract description 25
- 239000000463 material Substances 0.000 claims abstract description 30
- 238000002474 experimental method Methods 0.000 claims abstract description 11
- 230000008569 process Effects 0.000 claims abstract description 11
- 238000007405 data analysis Methods 0.000 claims abstract description 10
- 238000004458 analytical method Methods 0.000 claims abstract description 9
- 238000007418 data mining Methods 0.000 claims abstract description 5
- 230000006378 damage Effects 0.000 claims description 22
- 238000012549 training Methods 0.000 claims description 20
- 230000007797 corrosion Effects 0.000 claims description 18
- 238000005260 corrosion Methods 0.000 claims description 18
- 230000007547 defect Effects 0.000 claims description 14
- 241000288113 Gallirallus australis Species 0.000 claims description 10
- 238000001746 injection moulding Methods 0.000 claims description 9
- 238000012360 testing method Methods 0.000 claims description 9
- 230000000694 effects Effects 0.000 claims description 8
- 238000004519 manufacturing process Methods 0.000 claims description 8
- 238000005516 engineering process Methods 0.000 claims description 7
- 238000005070 sampling Methods 0.000 claims description 7
- 238000012423 maintenance Methods 0.000 claims description 6
- 238000000714 time series forecasting Methods 0.000 claims description 6
- 238000011156 evaluation Methods 0.000 claims description 5
- 238000005457 optimization Methods 0.000 claims description 5
- 238000007789 sealing Methods 0.000 claims description 5
- 230000002159 abnormal effect Effects 0.000 claims description 4
- 238000013461 design Methods 0.000 claims description 4
- 208000025274 Lightning injury Diseases 0.000 claims description 3
- 238000004364 calculation method Methods 0.000 claims description 3
- 238000010276 construction Methods 0.000 claims description 3
- 230000003628 erosive effect Effects 0.000 claims description 3
- 230000009395 genetic defect Effects 0.000 claims description 3
- 238000003466 welding Methods 0.000 claims description 3
- 238000012731 temporal analysis Methods 0.000 claims description 2
- 238000000700 time series analysis Methods 0.000 claims description 2
- 210000001367 artery Anatomy 0.000 claims 1
- 230000005518 electrochemistry Effects 0.000 claims 1
- 238000000605 extraction Methods 0.000 claims 1
- 210000003462 vein Anatomy 0.000 claims 1
- 238000009434 installation Methods 0.000 abstract description 3
- 239000000284 extract Substances 0.000 abstract description 2
- 230000035882 stress Effects 0.000 description 24
- 208000037656 Respiratory Sounds Diseases 0.000 description 10
- 208000010392 Bone Fractures Diseases 0.000 description 9
- 206010017076 Fracture Diseases 0.000 description 9
- 238000013213 extrapolation Methods 0.000 description 7
- 238000010586 diagram Methods 0.000 description 6
- 235000013399 edible fruits Nutrition 0.000 description 4
- 230000002706 hydrostatic effect Effects 0.000 description 4
- 230000007246 mechanism Effects 0.000 description 4
- 230000006399 behavior Effects 0.000 description 3
- 230000015572 biosynthetic process Effects 0.000 description 3
- 238000007796 conventional method Methods 0.000 description 3
- 239000002994 raw material Substances 0.000 description 3
- 239000000126 substance Substances 0.000 description 3
- 230000009471 action Effects 0.000 description 2
- 230000003679 aging effect Effects 0.000 description 2
- 230000001066 destructive effect Effects 0.000 description 2
- 238000001514 detection method Methods 0.000 description 2
- 230000005611 electricity Effects 0.000 description 2
- 230000006870 function Effects 0.000 description 2
- 230000006872 improvement Effects 0.000 description 2
- 230000000977 initiatory effect Effects 0.000 description 2
- 238000012417 linear regression Methods 0.000 description 2
- 238000004643 material aging Methods 0.000 description 2
- 230000010349 pulsation Effects 0.000 description 2
- 238000011160 research Methods 0.000 description 2
- 239000007787 solid Substances 0.000 description 2
- 238000012706 support-vector machine Methods 0.000 description 2
- 238000003786 synthesis reaction Methods 0.000 description 2
- XLYOFNOQVPJJNP-UHFFFAOYSA-N water Substances O XLYOFNOQVPJJNP-UHFFFAOYSA-N 0.000 description 2
- 241001269238 Data Species 0.000 description 1
- 208000013201 Stress fracture Diseases 0.000 description 1
- 238000009825 accumulation Methods 0.000 description 1
- 230000032683 aging Effects 0.000 description 1
- 230000002902 bimodal effect Effects 0.000 description 1
- 239000004566 building material Substances 0.000 description 1
- 239000004568 cement Substances 0.000 description 1
- 238000006243 chemical reaction Methods 0.000 description 1
- 238000003776 cleavage reaction Methods 0.000 description 1
- 238000005336 cracking Methods 0.000 description 1
- 230000007423 decrease Effects 0.000 description 1
- 238000011161 development Methods 0.000 description 1
- 230000018109 developmental process Effects 0.000 description 1
- 239000013013 elastic material Substances 0.000 description 1
- 239000000835 fiber Substances 0.000 description 1
- 229920001903 high density polyethylene Polymers 0.000 description 1
- 238000011534 incubation Methods 0.000 description 1
- 238000011900 installation process Methods 0.000 description 1
- 239000002184 metal Substances 0.000 description 1
- 238000005065 mining Methods 0.000 description 1
- 238000012986 modification Methods 0.000 description 1
- 230000004048 modification Effects 0.000 description 1
- 210000005036 nerve Anatomy 0.000 description 1
- 238000006116 polymerization reaction Methods 0.000 description 1
- 238000012545 processing Methods 0.000 description 1
- 230000007017 scission Effects 0.000 description 1
- 239000000243 solution Substances 0.000 description 1
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F30/00—Computer-aided design [CAD]
- G06F30/10—Geometric CAD
- G06F30/17—Mechanical parametric or variational design
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F30/00—Computer-aided design [CAD]
- G06F30/20—Design optimisation, verification or simulation
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F2119/00—Details relating to the type or aim of the analysis or the optimisation
- G06F2119/06—Power analysis or power optimisation
Abstract
The invention belongs to data analysis technique fields, disclose a kind of method based on artificial intelligence big data prediction Pressurized Plastic Pipes long term life, collect Pressurized Plastic Pipes failure case;The mechanical property parameters of typical plastics pressure pipeline material are obtained by experiment;Extract the basic parameter for influencing Pressurized Plastic Pipes service life;The basic parameter of Pressurized Plastic Pipes service life be will affect as parameter, failure mode and remaining life is inputted as output parameter, form big data analysis file;By big data analysis software WEKA, data mining analysis is carried out, obtains prediction failure mode and remaining life;Selection is suitable for the prediction algorithm of Pressurized Plastic Pipes;The prediction to Pressurized Plastic Pipes failure mode and remaining life is completed according to the algorithm of selection.The present invention predicts that the precision of Pressurized Plastic Pipes long term life, reliability are higher, and the service life of Pressurized Plastic Pipes can be improved, the safety of installation, use process can be improved.
Description
Technical field
The invention belongs to data analysis technique fields, more particularly to a kind of artificial intelligence big data that is based on to predict plastics pressure
The method of hydraulic piping long term life.
Background technique
As China reinforces the popularization and application dynamics to chemical building material, plastic conduit is widely used, has goed deep into
To the every field of social production.Compared with the pipelines such as traditional metal tube, pipe of cement, plastic tube has energy-saving material-saving, light
Inexpensive, corrosion-resistant, easy to install, the advantages that service life is long, flexibility is good, thus increasingly by the favor of engineering circles.It is macro in China
Under the pulling for seeing rapid economic development, China has become the most country of plastic conduit yield, and the kind of pipeline is more.
On the other hand, the working environment of plastic conduit is generally long-term pressure-bearing, high-temperature-hot-water, buried outer load etc., these
Severe application conditions determine that plastic conduit should have excellent performance.And the plastic conduit low quality of China's production, and
A large amount of to use the raw material without long-term hydrostatic evaluation test, this is that the quality problem being likely to occur from now on has buried hidden danger.
Currently, the theory of plastic conduit life prediction work is also immature, method is still not perfect, and does not catch up with the hair of plastic conduit much
Open up speed.Therefore, how evaluating the long-term mechanical property of plastic conduit and carrying out life prediction just seems especially urgent.At present
Existing method is as follows:
Linear elastic fracture mechanics (LEFM)
LEFM mainly describes the behavior of tip crackle in linear elastic materials.The destructive process of material is idealized as applying by it
The result that the fracture toughness of the stress intensity factor (SIF) and tubing that add balances each other.According to LEFM theory, destructive process can be with
It is divided into 3 stages: is applying the incubation period between load and crackle sprouting, SCG and the SIF when application are more than the disconnected of material
The brittle break occurred when splitting toughness.As shown in Figure 1, Fig. 1 is typical crack length with load time variation diagram.
But predict that pipeline life has a following limitation using LEFM method: first, lack crackle on pipeline and sprouts
The size and shape information of position, and these will affect crackle and sprout and original crack extension phase.Second, in fracture mechanics
In sample, body of material is met with stresses and is strained within the scope of linear viscoelasticity (lower than 1~2MPa), and actual interior pressure pipe
For the stress level that road is born generally in non-linear viscoelasticity range (being higher than 4MPa), this is likely to result in LEFM theory breaks.And
And PE tubing raw material newly developed has bigger Crack Tip Plastic, which also limits the validity of LEFM theory.Third,
When using LEFM method, it is necessary to correctly be assessed the aging effect of material.
Standard extrapolation (SEM)
The failure behaviour of plastic tube can be analyzed according to stress rupture curve (being obtained by hydrostatic experiment).It is hydrostatic
Experiment is sample pipe to be placed in the environment of certain temperature (such as water or air), and selected medium is full of in pipe sample, is applied to it
Different pressure values records the time to rupture of sample.According to many experiments, it can be deduced that Pressurized Plastic Pipes mainly have 3 kinds of mistakes
Effect mode.As shown in Fig. 2, logS is circumference stress, logtfFor the out-of-service time.
When using standard extrapolation, it should be noted that the influence of material aging and fault in material.In tubing manufacture, installation process
In, it will cause the initial imperfection of tubing, such as Brown has determined that the largest random defect average-size of business pipeline is about
100μm.When carrying out pipe design using standard extrapolation, it is contemplated that the influence of these defects should introduce safety coefficient.Phase
For standard extrapolation, Farshad proposes limiting strain extrapolation (USEM) and distortion energy extrapolation (DEEM), and points out
USEM is suitable for brittleness and fibre reinforced materials and DEEM is then suitable for extensive material type.Wherein, USEM loses according to strain
Criterion is imitated, and DEEM is according to distortion energy corresponding with failure stress.
Elastic-plastic fracture mechanics (EPFM)
Biggish plastic deformation occurs when destroying for toughness material (such as PE), and at this moment EPFM is applicable.EPFM method
It carries out being based primarily upon description Crack-tip parameters-cracks deflection and bridge (J) integral when failure prediction.For nonlinear elasticity solid,
J integral is the energy release of unit crackle area, is equivalent to the cracks deflection and bridge of unit width crack front.JRCharacterize crackle
Increase resistance, can be measured by the experimental method that European structural intergrity association describes.The J of 0.2mm crack propagation occursRValue is
J needed for crack propagationRCritical value, the toughness J corresponding to elastoplasticity crack initiationC.Elastic-plastic fracture mechanics thinks, works as unit
The cracks deflection and bridge J of width is more than JCWhen, that is, crack initiation occurs.After crack initiation, 2 kinds of possible failures will occur.When having
When imitating crack length more than pipe thickness, stationary crack will occur first and increase, when stationary crack rises to JR~a curve with
JRUnstable crack growth occurs when~a contact of a curve, failure criteria is as shown in Figure 3.
Crazing mechanism (CM)
Although single factor test fracture mechanics (LEFM or EPFM) can be applied in many occasions, polymerization newly developed
Object (such as bimodal PE material) can form fibrillated crazing region in crack tip, this may cause these theory breaks.
Fig. 4 is the crazing that PE material is formed.For certain component, creep will lead to stress and redistribute, and the stress in crazing area then may be used
To be assessed by " Reference Stress " theory.Using Reference Stress theory, Pandya is obtained: tough for the fracture for bearing constant pressure
Property sample (SENB style), prediction crackle sprout time be consistent with what is measured.But crazing mechanism method assumes material
The service life of material terminates after crackle sprouting, and the subsequent crack propagation time is ignored, this lead to that this method predicts when
Between be shorter than expected.Therefore, crazing mechanism method also needs further to develop, with simulating crack expansion process.
Other methods
Ben etc. is proposed creeping solids fracture mechanics method (FMCS), which thinks that creep loading parameter (C*) is and failure
Time (tR) relevant major parameter, and the relational expression of the two is established, which is considered as a very promising engineering
Application method.Laiarinandrasana etc. then depicts correlation curve by experiment, and has studied pvc pipe in conjunction with FMCS method
Influence of the road aging to its creep behaviour, while having carried out residual life evaluation.Choi etc. is theoretical according to crackle layer (CL) and ties
It closes experimental data and life prediction has been carried out to PE-HD pipeline, and pointing out from now on will research On Creep Crack Growth (CCG) and fatigue
Relationship between crack propagation (FCG) dynamics.Meanwhile the researcher is according to same theoretical to the expansion of stress corrosion cracking (SCC) at a slow speed
Exhibition process is modeled, and discusses the algorithm for assessing the plastic conduit service life under stress and corrosive environment.Khelif
Etc. a kind of probabilistic method is proposed, time-based Life Prediction Model and the life prediction mould based on creep strain are compared
Type, and reliability assessment has been carried out to the pipeline under various operating conditions.Hoang etc. is using this method and hydrostatic is combined to be compacted
It tests and has carried out PE100 grades of aqueduct life appraisals, while pointing out that this method is suitable for other plastic conduits.
In conclusion problem of the existing technology is:
Existing plastic conduit life-span prediction method lacks the size and shape information that crackle on pipeline sprouts position, nothing
Method correctly assesses the aging effect of material, lacks the assessment of the Crack Tip Plastic to PE tubing raw material newly developed,
Do not consider material aging and fault in material influence, the plastic deformation of toughness material is required it is larger, to certain of application material
Kind characteristic has special requirement, and then influences the precision of prediction.
Summary of the invention
In view of the problems of the existing technology, the present invention provides one kind predicts plastics pressure based on artificial intelligence big data
The method of hydraulic piping long term life.
The invention is realized in this way a kind of predict Pressurized Plastic Pipes long term life based on artificial intelligence big data
Method, the method based on artificial intelligence big data prediction Pressurized Plastic Pipes long term life include:
Step 1 collects a large amount of Pressurized Plastic Pipes failure case;
Step 2 obtains the mechanical property parameters of typical plastics pressure pipeline material by experiment;
Step 3, extracts the basic parameter for influencing Pressurized Plastic Pipes service life, such as Initial crack length, and fatigue is answered
Power width, the scratch for squeezing the prestressing force of injection molding, surface;
Step 4, will affect the basic parameter of Pressurized Plastic Pipes service life as input parameter, failure mode and
Remaining life forms big data analysis file as output parameter;With Initial crack length length, Fatigue Stress Amplitude
Amplitude, squeeze the prestressing force prestressing of injection molding, the scratch mark on surface is input parameter, if it is defeated for failing
Parameter out, parameter include failure and safe both of which;Ultimately form the arff file that weka can be identified;
Step 5 optimizes the training sample of acquisition, including also using SMOTE synthesis minority class oversampling technique
I.e. oversampling technique manually enhances existing training set, class balance, lack sampling, rejects unusual training sample technology, realizes
Optimization training to training sample;Analysis is carried out to minority class sample and is added according to the artificial synthesized new samples of minority class sample
Into data set;
Step 6 selects different algorithms to carry out data mining analysis by big data analysis software WEKA, and optimization is each
Kind algorithm parameter, obtains prediction failure mode and remaining life;SMOreg recurrence kernel selection Puk algorithm,
RegOptimizer selects RegSMOImproved algorithm;
Step 7, more different algorithm predictions as a result, carrying out ballot vote to algorithms of different prediction result;
Step 8 completes the prediction to Pressurized Plastic Pipes failure mode and remaining life by voting results.
Further, the Pressurized Plastic Pipes failure case of the step 1 includes:
(1) misoperation causes service condition to deteriorate, including superpressure, overtemperature, corrosive media are exceeded, pressure and temp is different
Often pulsation, generate alternating load, caused by Pressurized Plastic Pipes fatigue failure;
(2) it designs, manufacture, construction existing defects, pipeline flexible is undesirable, and material selection is improper or with material mistake,
There are welding defect, weld or assemble it is unreasonable cause stress excessive, pipe-supporting system is unreasonable etc.;Pipeline is before coming into operation
Existing genetic defects will cause the low stress brittle fracture of material;
(3) maintenance is made mistakes, and the major defect or damage on pipeline fail to be detected discovery, or lack scientific evaluation, and
Unreasonable maintenance process causes new defect and damage;
(4) extrinsic damage damages, earthquake, strong wind, flood, lightning stroke and other machinery damage and artificial destruction etc.;
(5) corrosion failure, the corrosion of pressure pipeline are due to the chemistry or electricity by interior media and external environment medium
Chemical action and the destruction occurred, also including machinery etc. the collective effect of reasons as a result, unreasonable operation leads to concentration of medium
Variation, aggravate corrosion failure;
(6) erosion attack, the long-term of pipeline interior media, flow at high speed can make piping component inner wall be thinned or seal
Pair wrecks, and influences pipeline compressive resistance and sealing performance, causes the leakage of pipeline.
Further, the SMOTE over-sampling of the step 5 specifically includes:
1) for each sample x in minority class, using Euclidean distance as criterion calculation the institute into minority class sample set Smin
There is the distance of sample, obtains its k neighbour;
2) multiplying power N is sampled to determine according to one oversampling ratio of sample imbalance ratio setting, for each minority class
Sample x randomly chooses several samples from its k neighbour, and the neighbour selected is xn;
3) for the neighbour xn that each is selected at random, new sample is constructed according to following formula with original sample respectively:
xnew=x+rand (0,1) * | x-xn |;
LOF basic thought are as follows: the part of point p peels off factor representation are as follows:
Indicate the average of the ratio between the local reachability density of the neighborhood point Nk (p) of point p and the local reachability density of point p;Such as
This ratio of fruit illustrates that its neighborhood dot density of p is similar, p may belong to cluster with neighborhood closer to 1;If this ratio
More less than 1, illustrate that the density of p is higher than its neighborhood dot density, p is point off density;If this ratio is more greater than 1, illustrate that p's is close
Degree is less than its neighborhood dot density, and p more may be abnormal point.
Further, the step 6 specifically includes: first according to Pressurized Plastic Pipes primary condition, forming time series
The arff file of analysis;With time Date, Initial crack length length, Fatigue Stress Amplitude amplitude, squeeze injection molding
Prestressing force prestressing, surface scratch mark be input parameter, formed WEKA can identify time series arff analysis
File;Then WEKA deadline sequence prediction obtains the new mechanics parameter of subsequent time, and new mechanics parameter, which is formed, to be surveyed
Arff file on probation;Then use the arff of step 4 as the prediction of the complete pairwise testing arff of training set: if prediction result
To fail, the then moment corresponds to Pressurized Plastic Pipes out-of-service time node, obtains the service life of Pressurized Plastic Pipes;Otherwise such as
The step of fruit result is safety, then continues time series forecasting, repeat front, until the result judgement of prediction is to lose
Effect.
Plastics are predicted based on artificial intelligence big data using described another object of the present invention is to provide a kind of
The Pressurized Plastic Pipes long term life predicting platform of the method for pressure pipeline long term life.
Advantages of the present invention and good effect are as follows:
The present invention is analyzed using Pressurized Plastic Pipes failure case of the artificial intelligence big data to collection, acquisition
Sample size, type are more, improve the reliability of data, improve Pressurized Plastic Pipes failure mode and remaining life
Predict precision.Conventional method is to roots such as Pressurized Plastic Pipes possible failure modes such as Rapid Crack, creep, brittle failure
This is unpredictable, and also only uses experience or semiempirical formula to the prediction of Pressurized Plastic Pipes remaining life.However plastics
Pressure pipeline remaining life is multiple influence factor nonlinearity effects as a result, this also results in existing empirical equation
Prediction result and actual result are quite different, and artificial intelligence big data analysis method is especially good at processing nonlinearity and is returned
Return problem, many times prediction result and actual result difference greatly meet engineering precision demand less than 5%.Table 1 is
The comparison of common big data algorithm regression forecasting precision and conventional method.
1 conventional method of table and artificial intelligence big data algorithm comparison
The Attributions selection function and correlation rule function of big data are utilized simultaneously, and extracting influences Pressurized Plastic Pipes most
Main several factors find the correlation rule influenced between Pressurized Plastic Pipes difference factor.Such as Pressurized Plastic Pipes
Fatigue failure mode, the parameters such as pipeline initial crack and load loading position, size there is certain correlation rule.Pass through
Artificial intelligence big data association rule mining greatly improves plastics pressure so as to targetedly propose related precautionary measures
The service life of hydraulic piping.Existing artificial intelligence big data association rule algorithm suffers from important application in many fields,
For example building detection industry can greatly improve structural life-time using big data association rule algorithm, most can be improved
30% or more.
It is also convenient for establishing the failure mode of Pressurized Plastic Pipes and remaining life data using artificial intelligence big data technology
Library, and then promote supervision ability.The database uses crowdsourcing model, and public participation is made to collect Pressurized Plastic Pipes failure
Case is possibly realized, and greatly enriches database data, database size can be made to expand several times in a short time, is various operating conditions
Accurate Prediction improve may.The present invention predicts that the precision of Pressurized Plastic Pipes long term life, reliability are higher, while can
Using big data analysis method, the service life of Pressurized Plastic Pipes is improved, promotes supervision ability.Moreover, with verifying
The continuous accumulation of data, not only can constantly improve precision of prediction, additionally it is possible to provide the reliability index of prediction error, lead to
The safety check for Pressurized Plastic Pipes is crossed, is also had very actively to the reliability index for improving Pressurized Plastic Pipes safety
Meaning.Finally for the Accurate Prediction of various operating conditions improve may, to reduce the generation of contingency, improve installation,
The safety of use process.
Detailed description of the invention
Fig. 1 is provided in an embodiment of the present invention based on artificial intelligence big data prediction Pressurized Plastic Pipes long term life
The typical crack length of the linear elastic fracture mechanics method of method is with load time variation diagram.
Fig. 2 is provided in an embodiment of the present invention based on artificial intelligence big data prediction Pressurized Plastic Pipes long term life
The internal pressure pipe stress broken curve figure of the standard extrapolation of method.
Fig. 3 is provided in an embodiment of the present invention based on artificial intelligence big data prediction Pressurized Plastic Pipes long term life
The J of the elastic-plastic fracture mechanics method of method integrates failure criteria figure.
Fig. 4 is provided in an embodiment of the present invention based on artificial intelligence big data prediction Pressurized Plastic Pipes long term life
The flow chart of method.
Fig. 5 is the arff file schematic diagram that formation weka provided in an embodiment of the present invention can be identified.
Fig. 6 is that SMOTE oversampling technique provided in an embodiment of the present invention carries out analysis to minority class sample and according to minority
The artificial synthesized new samples of class sample are added to schematic diagram in data set.
Fig. 7 is inflection point detection schematic diagram provided in an embodiment of the present invention.
Fig. 8 is time series arff file schematic diagram provided in an embodiment of the present invention.
Fig. 9 is time series forecasting flow chart provided in an embodiment of the present invention.
Specific embodiment
In order to make the objectives, technical solutions, and advantages of the present invention clearer, with reference to embodiments, to this hair
It is bright to be further elaborated.It should be appreciated that the specific embodiments described herein are merely illustrative of the present invention, not
For limiting the present invention.
With reference to the accompanying drawing and specific embodiment is further described application principle of the invention.
As shown in figure 4, the method based on artificial intelligence big data prediction Pressurized Plastic Pipes long term life includes following step
It is rapid:
S101: a large amount of Pressurized Plastic Pipes failure case is collected;
S102: the mechanical property parameters of typical plastics pressure pipeline material are obtained by experiment;
S103: the basic parameter for influencing Pressurized Plastic Pipes service life, such as Initial crack length, fatigue stress are extracted
Width, the prestressing force, the scratch on surface that squeeze injection molding etc.;
S104: will affect the basic parameter of Pressurized Plastic Pipes service life as input parameter, failure mode and surplus
The remaining service life as output parameter, forms big data analysis file;
S105: it by big data analysis software WEKA, selects different algorithms to carry out data mining analysis, is predicted
Failure mode and remaining life;
S106: that more different algorithms is predicted as a result, selection is suitable for the prediction algorithm of Pressurized Plastic Pipes;
S107: the prediction to Pressurized Plastic Pipes failure mode and remaining life is completed according to the algorithm of selection.
In a preferred embodiment of the invention, step S104 is specifically included: with Initial crack length length (mm), tired
Labor stress amplitude amplitude (MPa), the scratch mark (mm) for squeezing the prestressing force prestressing (MPa) of injection molding, surface
To input parameter, if fail for output parameter, the parameter two kinds of (success) moulds including failure (failure) and safely
Formula.Ultimately form the arff file (as shown in Figure 5) that weka can be identified.Therefore, collection plastic pressure as much as possible is wanted early period
The Engineering Projects of pipeline failure, the mechanics parameter before being failed, to form the training of Pressurized Plastic Pipes big data analysis
Collection.The step is particularly critical, only collect enough failure cases formed big data analysis training and, could be to unknown feelings
Condition is accurately classified, so that the service life of Pressurized Plastic Pipes can accurately be predicted.
In a preferred embodiment of the invention, step S105 is specifically included: being optimized, is wrapped to the training sample of acquisition
It includes using SMOTE (Synthetic Minority Oversampling Technique) synthesis minority class oversampling technique also
I.e. oversampling technique manually enhances existing training set, class balances (ClassBalancer), lack sampling
(SpreadSubsample), the technologies such as unusual training sample (LOF) are rejected, realize the optimization training to training sample;SMOTE
Oversampling technique basic thought are as follows: be a kind of improvement project based on random over-sampling algorithm, since random over-sampling takes letter
The strategy of single reproduction copies is come the problem of increasing minority class sample, be easy to produce model over-fitting in this way, i.e., so that model learning
The information arrived excessively especially (Specific) and it is not extensive enough (General), the basic thought of SMOTE algorithm is to minority class
Sample analyze and be added in data set according to the artificial synthesized new samples of minority class sample, specifically as shown in fig. 6, algorithm
Process is as follows.
1) for each sample x in minority class, using Euclidean distance as criterion calculation it into minority class sample set Smin
The distance of all samples obtains its k neighbour;
2) multiplying power N is sampled to determine according to one oversampling ratio of sample imbalance ratio setting, for each minority class
Sample x randomly chooses several samples from its k neighbour, it is assumed that the neighbour selected is xn;
3) for the neighbour xn that each is selected at random, new sample is constructed according to following formula with original sample respectively:
xnew=x+rand (0,1) * | x-xn |;
LOF (Local outlier factor) basic thought are as follows:
The part of point p peels off factor representation are as follows:
Indicate the average of the ratio between the local reachability density of the neighborhood point Nk (p) of point p and the local reachability density of point p.Such as
This ratio of fruit illustrates that its neighborhood dot density of p is similar, p may belong to cluster with neighborhood closer to 1;If this ratio
More less than 1, illustrate that the density of p is higher than its neighborhood dot density, p is point off density;If this ratio is more greater than 1, illustrate that p's is close
Degree is less than its neighborhood dot density, and p more may be abnormal point, as shown in Figure 7.
In a preferred embodiment of the invention, step S106 is specifically included: passing through big data analysis software WEKA, selection
Different algorithms carries out data mining analysis, optimizes various algorithm parameters, obtains prediction failure mode and residue uses the longevity
Life;Basic ideas are as follows: first according to Pressurized Plastic Pipes primary condition, formed time series analysis arff file (such as Fig. 8,
Preferably collect enough initial condition parameters).With time Date (yyyy-MM-dd HH), Initial crack length length
(mm), Fatigue Stress Amplitude amplitude (MPa), squeeze the scratch of the prestressing force prestressing (MPa) of injection molding, surface
Mark (mm) is input parameter, forms the time series arff Study document that WEKA can be identified;Then WEKA deadline sequence
Column prediction, obtains the new mechanics parameter of subsequent time, and new mechanics parameter forms test arff file.Then step is used
Prediction of the arff of S104 as the complete pairwise testing arff of training set: if prediction result be failure (failure) if this when
Corresponding Pressurized Plastic Pipes out-of-service time node is carved, to obtain the service life of Pressurized Plastic Pipes.If instead result is
The step of safety (success) then continues time series forecasting, repeats front, until the result judgement of prediction is failure
(failure)。
Be exactly time series forecasting as shown in figure 9, crucial, time series forecasting it is accurate whether directly influence energy
Whether no accurate judgement Pressurized Plastic Pipes are realized.And the key factor of influence time sequence prediction is its regression algorithm,
The common algorithm for multiple regression includes: linear regression (LinearRegression), support vector machines (SVM), nerve
Network (MultilayerPerceptron), multiple regression (SMOreg), behind 3 kinds of homing method especially SMOreg be related to
The problem of parameter arrived is very more, this relates to various parameters optimization.Through a large number of experiments research shows that SMOreg is returned
Kernel selects Puk algorithm, regOptimizer selection RegSMOImproved algorithm to imitate for most of time forecasting problem
Fruit is good.
As the preferred embodiment of the present invention, the big data of the Pressurized Plastic Pipes failure case of the S101 includes real
Production, application and the experimentation on border possess the largely data about Pressurized Plastic Pipes failure, such as various mechanical property
The fail datas such as tough crisp conversion, Rapid Crack in energy test experiments data (fatigue, creep, fracture etc.), operational process,
Parameter etc. even in manufacture, installation, transport, specifically includes that
(1) misoperation causes service condition to deteriorate, including superpressure, overtemperature, corrosive media are exceeded, pressure and temp is different
Often pulsation etc., to generate alternating load, caused by Pressurized Plastic Pipes fatigue failure;
(2) it designs, manufacture, construction existing defects, if pipeline flexible is undesirable, material selection is improper or wrong with material
Accidentally, there are welding defect, weld or assemble it is unreasonable cause stress excessive, pipe-supporting system is unreasonable etc.;Pipeline is coming into operation
Preceding existing genetic defects will cause the low stress brittle fracture of material;
(3) maintenance is made mistakes, and the major defect or damage on pipeline fail to be detected discovery, or lack scientific evaluation, and
Unreasonable maintenance process causes new defect and damage etc.;
(4) extrinsic damage damages, such as earthquake, strong wind, flood, lightning stroke and other machinery damage and artificial destruction;
There are many Destroy type of pressure pipeline.Ductile failure (ductile fracture) can be divided by macroscopic deformation amount when destroying and brittleness is broken
It is disconnected can be divided into dimple fracture, cleavage fracture, grain boundary fracture and fatigue by the Micro-fracture Mechanism of material when destroying for bad two major classes
The patterns such as split, in general, classifying at the scene using the method that macroscopic view classification and fracture characteristic combine, flexible destroys, is crisp
Property destruction, corrosion failure, fatigue rupture, creep rupture etc.;
(5) corrosion failure, the corrosion of pressure pipeline are due to the chemistry or electricity by interior media and external environment medium
Chemical action and the destruction occurred, also including machinery etc. reasons collective effect as a result, unreasonable operation to will lead to medium dense
The variation of degree aggravates corrosion failure;The form of the corrosion failure of pressure pipeline have general corrosion, local corrosion, stress corrosion,
Corrosion fatigue etc., wherein stress corrosion often occurs in the case where no tendency suddenly, therefore its harmfulness is bigger;
(6) erosion attack, the long-term of pipeline interior media, flow at high speed can make piping component inner wall be thinned or seal
Pair wrecks, and influences its compressive resistance and sealing performance.With the extension of use time, pressure resistance caused by being thinned as inner wall
The leakage that ability decline or sealing pair are damaged and formed will be as the root of accident.
The foregoing is merely illustrative of the preferred embodiments of the present invention, is not intended to limit the invention, all in essence of the invention
Made any modifications, equivalent replacements, and improvements etc., should all be included in the protection scope of the present invention within mind and principle.
Claims (5)
1. a kind of method based on artificial intelligence big data prediction Pressurized Plastic Pipes long term life, which is characterized in that this is based on
Artificial intelligence big data predict Pressurized Plastic Pipes long term life method the following steps are included:
Step 1 collects a large amount of Pressurized Plastic Pipes failure case;
Step 2 obtains the mechanical property parameters of typical plastics pressure pipeline material by experiment;
Step 3, the basic parameter of extraction influence Pressurized Plastic Pipes service life, such as Initial crack length, Fatigue Stress Amplitude,
Squeeze the prestressing force of injection molding, the scratch on surface;
Step 4 will affect the basic parameter of Pressurized Plastic Pipes service life as input parameter, failure mode and residue
Service life as output parameter, forms big data analysis file;With Initial crack length length, Fatigue Stress Amplitude amplitude,
The scratch mark of the prestressing force prestressing, surface that squeeze injection molding are input parameter, if are failed for output parameter, parameter
Including failure and safe both of which;Ultimately form the arff file that weka can be identified;
Step 5 optimizes the training sample of acquisition, including is synthesized minority class oversampling technique using SMOTE that is, crossed and adopted
Sample technology manually enhances existing training set, class balance, lack sampling, rejects unusual training sample technology, realizes to training
The optimization training of sample;Minority class sample analyze and is added to data set according to the artificial synthesized new samples of minority class sample
In;
Step 6 selects different algorithms to carry out data mining analysis, optimizes various algorithms by big data analysis software WEKA
Parameter obtains prediction failure mode and remaining life;SMOreg recurrence kernel selection Puk algorithm,
RegOptimizer selects RegSMOImproved algorithm;
Step 7, more different algorithm predictions as a result, carrying out ballot vote to algorithms of different prediction result;
Step 8 completes the prediction to Pressurized Plastic Pipes failure mode and remaining life by voting results.
2. the method as described in claim 1 based on artificial intelligence big data prediction Pressurized Plastic Pipes long term life, feature
It is, the Pressurized Plastic Pipes failure case of the step 1 includes:
(1) misoperation causes service condition to deteriorate, including superpressure, overtemperature, corrosive media are exceeded, the abnormal arteries and veins of pressure and temp
It is dynamic, generate alternating load, caused by Pressurized Plastic Pipes fatigue failure;
(2) it designs, manufacture, construction existing defects, pipeline flexible is undesirable, and material selection is improper or with material mistake, exists
Welding defect, weld or assemble it is unreasonable cause stress excessive, pipe-supporting system is unreasonable etc.;Pipeline is existing before coming into operation
Genetic defects will cause the low stress brittle fracture of material;
(3) maintenance is made mistakes, and the major defect or damage on pipeline fail to be detected discovery, or lack scientific evaluation, and do not conform to
The maintenance process of reason causes new defect and damage;
(4) extrinsic damage damages, earthquake, strong wind, flood, lightning stroke and other machinery damage and artificial destruction etc.;
(5) corrosion failure, the corrosion of pressure pipeline are due to the chemistry or electrochemistry by interior media and external environment medium
Effect and occur destruction, also including machinery etc. the collective effect of reasons as a result, unreasonable operation leads to the change of concentration of medium
Change, aggravates corrosion failure;
(6) erosion attack, the long-term of pipeline interior media, flow at high speed can make piping component inner wall be thinned or sealing pair by
It destroys, influences pipeline compressive resistance and sealing performance, cause the leakage of pipeline.
3. the method as described in claim 1 based on artificial intelligence big data prediction Pressurized Plastic Pipes long term life, feature
It is, the SMOTE over-sampling of the step 5 specifically includes:
1) for each sample x in minority class, using Euclidean distance as criterion calculation all samples into minority class sample set Smin
This distance, obtains its k neighbour;
2) multiplying power N is sampled to determine according to one oversampling ratio of sample imbalance ratio setting, for each minority class sample
X randomly chooses several samples from its k neighbour, and the neighbour selected is xn;
3) for the neighbour xn that each is selected at random, new sample is constructed according to following formula with original sample respectively:
xnew=x+rand (0,1) * | x-xn |;
LOF basic thought are as follows: the part of point p peels off factor representation are as follows:
Indicate the average of the ratio between the local reachability density of the neighborhood point Nk (p) of point p and the local reachability density of point p;If this
A ratio illustrates that its neighborhood dot density of p is similar, p may belong to cluster with neighborhood closer to 1;If this ratio is smaller
In 1, illustrate that the density of p is higher than its neighborhood dot density, p is point off density;If this ratio is more greater than 1, illustrate that the density of p is less than
Its neighborhood dot density, p more may be abnormal point.
4. the method as described in claim 1 based on artificial intelligence big data prediction Pressurized Plastic Pipes long term life, feature
It is, the step 6 specifically includes: first according to Pressurized Plastic Pipes primary condition, forms the arff text of time series analysis
Part;With time Date, Initial crack length length, Fatigue Stress Amplitude amplitude, the prestressing force for squeezing injection molding
Prestressing, surface scratch mark be input parameter, form the time series arff Study document that can identify of WEKA;So
WEKA deadline sequence prediction afterwards, obtains the new mechanics parameter of subsequent time, and new mechanics parameter forms test and uses arff
File;Then use the arff of step 4 as the prediction of the complete pairwise testing arff of training set: if prediction result is failure
The moment corresponds to Pressurized Plastic Pipes out-of-service time node, obtains the service life of Pressurized Plastic Pipes;If instead result is
The step of safety then continues time series forecasting, repeats front, until the result judgement of prediction is failure.
5. a kind of predict Pressurized Plastic Pipes based on artificial intelligence big data using described in described in claim 1-4 any one
The Pressurized Plastic Pipes long term life predicting platform of the method for long term life.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201910201622.3A CN109977511A (en) | 2019-03-18 | 2019-03-18 | Method based on artificial intelligence big data prediction Pressurized Plastic Pipes long term life |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201910201622.3A CN109977511A (en) | 2019-03-18 | 2019-03-18 | Method based on artificial intelligence big data prediction Pressurized Plastic Pipes long term life |
Publications (1)
Publication Number | Publication Date |
---|---|
CN109977511A true CN109977511A (en) | 2019-07-05 |
Family
ID=67079208
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN201910201622.3A Pending CN109977511A (en) | 2019-03-18 | 2019-03-18 | Method based on artificial intelligence big data prediction Pressurized Plastic Pipes long term life |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN109977511A (en) |
Cited By (7)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN110705176A (en) * | 2019-09-02 | 2020-01-17 | 北京市燃气集团有限责任公司 | Method and device for predicting residual life of gas pipeline |
CN110968618A (en) * | 2019-11-07 | 2020-04-07 | 华中科技大学 | Method for mining quantitative association rule of welding parameters and application |
CN111310314A (en) * | 2020-01-21 | 2020-06-19 | 北京瑞风协同科技股份有限公司 | Method and system for determining service life of electromechanical device based on artificial intelligence |
CN112052624A (en) * | 2020-08-13 | 2020-12-08 | 五邑大学 | Method for estimating and predicting bridge life based on big data |
CN115130496A (en) * | 2022-05-24 | 2022-09-30 | 北京化工大学 | Pipeline pressure signal anomaly detection method based on Bagging and RM-LOF integrated single classifier |
CN115330094A (en) * | 2022-10-14 | 2022-11-11 | 成都秦川物联网科技股份有限公司 | Intelligent gas pipeline service life prediction method, internet of things system, device and medium |
CN117236087A (en) * | 2023-11-16 | 2023-12-15 | 苏州顶材新材料有限公司 | Method and system for evaluating service life of thermoplastic elastomer of blending type interpenetrating network |
Citations (5)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN101162234A (en) * | 2007-11-23 | 2008-04-16 | 华东理工大学 | Remanent service life prediction method of steam turbine high-temperature component |
CN102081020A (en) * | 2010-01-26 | 2011-06-01 | 上海海事大学 | Material fatigue-life predicting method based on support vector machine |
CN104462806A (en) * | 2014-12-02 | 2015-03-25 | 中国石油天然气集团公司 | Non-metallic pipe life prediction method |
CN105653791A (en) * | 2015-12-29 | 2016-06-08 | 中国石油天然气集团公司 | Data mining based corrosion failure prediction system for on-service oil pipe column |
CN109446189A (en) * | 2018-10-31 | 2019-03-08 | 成都天衡智造科技有限公司 | A kind of technological parameter outlier detection system and method |
-
2019
- 2019-03-18 CN CN201910201622.3A patent/CN109977511A/en active Pending
Patent Citations (5)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN101162234A (en) * | 2007-11-23 | 2008-04-16 | 华东理工大学 | Remanent service life prediction method of steam turbine high-temperature component |
CN102081020A (en) * | 2010-01-26 | 2011-06-01 | 上海海事大学 | Material fatigue-life predicting method based on support vector machine |
CN104462806A (en) * | 2014-12-02 | 2015-03-25 | 中国石油天然气集团公司 | Non-metallic pipe life prediction method |
CN105653791A (en) * | 2015-12-29 | 2016-06-08 | 中国石油天然气集团公司 | Data mining based corrosion failure prediction system for on-service oil pipe column |
CN109446189A (en) * | 2018-10-31 | 2019-03-08 | 成都天衡智造科技有限公司 | A kind of technological parameter outlier detection system and method |
Non-Patent Citations (3)
Title |
---|
叶伟文等: "塑料及其衬里制压力容器的剩余寿命评价方法研究", 《广东化工》 * |
杨毅,卢诚波,徐根海: "面对不平衡数据集的一种精化Borderline-SMOTE方法", 《复旦学报(自然科学版)》 * |
武新丽等: "压力管道破坏的形式、原因与防止", 《中国科技信息》 * |
Cited By (11)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN110705176A (en) * | 2019-09-02 | 2020-01-17 | 北京市燃气集团有限责任公司 | Method and device for predicting residual life of gas pipeline |
CN110968618A (en) * | 2019-11-07 | 2020-04-07 | 华中科技大学 | Method for mining quantitative association rule of welding parameters and application |
CN110968618B (en) * | 2019-11-07 | 2023-04-11 | 华中科技大学 | Method for mining quantitative association rule of welding parameters and application |
CN111310314A (en) * | 2020-01-21 | 2020-06-19 | 北京瑞风协同科技股份有限公司 | Method and system for determining service life of electromechanical device based on artificial intelligence |
CN112052624A (en) * | 2020-08-13 | 2020-12-08 | 五邑大学 | Method for estimating and predicting bridge life based on big data |
CN115130496A (en) * | 2022-05-24 | 2022-09-30 | 北京化工大学 | Pipeline pressure signal anomaly detection method based on Bagging and RM-LOF integrated single classifier |
CN115130496B (en) * | 2022-05-24 | 2023-06-16 | 北京化工大学 | Pipeline pressure signal abnormality detection method based on Bagging and RM-LOF integrated single classifier |
CN115330094A (en) * | 2022-10-14 | 2022-11-11 | 成都秦川物联网科技股份有限公司 | Intelligent gas pipeline service life prediction method, internet of things system, device and medium |
US11898704B2 (en) | 2022-10-14 | 2024-02-13 | Chengdu Qinchuan Iot Technology Co., Ltd. | Methods and Internet of Things systems for smart gas pipeline life prediction based on safety |
CN117236087A (en) * | 2023-11-16 | 2023-12-15 | 苏州顶材新材料有限公司 | Method and system for evaluating service life of thermoplastic elastomer of blending type interpenetrating network |
CN117236087B (en) * | 2023-11-16 | 2024-01-26 | 苏州顶材新材料有限公司 | Method and system for evaluating service life of thermoplastic elastomer of blending type interpenetrating network |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
CN109977511A (en) | Method based on artificial intelligence big data prediction Pressurized Plastic Pipes long term life | |
CN104268383B (en) | Safety evaluation method for high-temperature pressure pipeline containing crack defects | |
CN108388724A (en) | A kind of GM-Markov submarine pipeline corrosion prediction methods based on parameter optimization | |
Taiwo et al. | Development of sustainable water infrastructure: A proper understanding of water pipe failure | |
CN108536926A (en) | Based on the corrosion oil-gas pipeline Residual Life method for improving adaptive GEV distributions | |
Yeter et al. | Structural integrity assessment of fixed support structures for offshore wind turbines: A review | |
CN101853320A (en) | Fuzzy comprehensive evaluation method suitable for aircraft structure corrosion damage | |
Ziegler et al. | Brief communication: Structural monitoring for lifetime extension of offshore wind monopiles: can strain measurements at one level tell us everything? | |
Zhang et al. | Phenomenological modelling of tensile fracture in PE pipe by considering damage evolution | |
Kibey et al. | Tensile strain capacity equations for strain-based design of welded pipelines | |
Lu et al. | A feature selection–based intelligent framework for predicting maximum depth of corroded pipeline defects | |
CN103838931A (en) | Method for evaluating remanufacturing access period of engineering mechanical arm rest class structure | |
CN101769147A (en) | Method for evaluating oilfield fireflood scheme | |
Muda et al. | Burst pressure strength of corroded subsea pipelines repaired with composite fiber-reinforced polymer patches | |
Liu et al. | Cracking risk analysis of face slabs in concrete face rockfill dams during the operation period | |
CN108615098A (en) | Water supply network pipeline burst Risk Forecast Method based on Bayesian survival analysis | |
Jiang et al. | An integrated risk analysis model for corroded pipelines subjected to internal pressures: Considering the interacting effects | |
CN108710946A (en) | The human factors analysis balancing method of deepwater riser system risk-based maintenance decision optimization | |
Mulenga et al. | Investigating the root causes of major failures of critical components–With a case study of asbestos cement pipes | |
Viswanathan et al. | Life-assessment technology for power-plant components | |
Zhang et al. | Reliability assessment of corroded pipeline considering multiple defects interaction based on an artificial neural network method | |
Tsatsis et al. | Fragility analysis of a pipeline under slope failure-induced displacements occurring parallel to its axis | |
Gao | Systematic review for water network failure models and cases | |
Clair | Development of a Novel Performance Index and a Performance Prediction Model for Metallic Drinking Water Pipelines | |
CN115081279A (en) | PCCP pipeline structure damage evaluation method under rockfall impact |
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 | ||
RJ01 | Rejection of invention patent application after publication |
Application publication date: 20190705 |
|
RJ01 | Rejection of invention patent application after publication |