CN115186590A - Method for predicting residual life of pipeline corrosion - Google Patents
Method for predicting residual life of pipeline corrosion Download PDFInfo
- Publication number
- CN115186590A CN115186590A CN202210829744.9A CN202210829744A CN115186590A CN 115186590 A CN115186590 A CN 115186590A CN 202210829744 A CN202210829744 A CN 202210829744A CN 115186590 A CN115186590 A CN 115186590A
- Authority
- CN
- China
- Prior art keywords
- corrosion
- pipeline
- detection
- predicting
- residual 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
- 230000007797 corrosion Effects 0.000 title claims abstract description 129
- 238000005260 corrosion Methods 0.000 title claims abstract description 129
- 238000000034 method Methods 0.000 title claims abstract description 31
- 238000001514 detection method Methods 0.000 claims abstract description 58
- 239000002689 soil Substances 0.000 claims abstract description 29
- 238000013528 artificial neural network Methods 0.000 claims abstract description 27
- 238000002474 experimental method Methods 0.000 claims abstract description 4
- 239000000523 sample Substances 0.000 claims description 41
- 230000006870 function Effects 0.000 claims description 38
- 238000012360 testing method Methods 0.000 claims description 20
- 238000012549 training Methods 0.000 claims description 13
- 239000000463 material Substances 0.000 claims description 9
- 230000004913 activation Effects 0.000 claims description 6
- 230000007547 defect Effects 0.000 claims description 6
- 210000002569 neuron Anatomy 0.000 claims description 6
- 230000033116 oxidation-reduction process Effects 0.000 claims description 6
- 150000003839 salts Chemical class 0.000 claims description 6
- XLYOFNOQVPJJNP-UHFFFAOYSA-N water Substances O XLYOFNOQVPJJNP-UHFFFAOYSA-N 0.000 claims description 6
- 229910000831 Steel Inorganic materials 0.000 claims description 5
- 239000010959 steel Substances 0.000 claims description 5
- 238000004210 cathodic protection Methods 0.000 claims description 3
- 239000011248 coating agent Substances 0.000 claims description 3
- 238000000576 coating method Methods 0.000 claims description 3
- 238000009933 burial Methods 0.000 claims description 2
- 230000003213 activating effect Effects 0.000 claims 1
- 230000003628 erosive effect Effects 0.000 claims 1
- 239000010410 layer Substances 0.000 description 53
- 230000001133 acceleration Effects 0.000 description 5
- 238000004364 calculation method Methods 0.000 description 5
- 230000009286 beneficial effect Effects 0.000 description 3
- 239000011159 matrix material Substances 0.000 description 3
- 230000008901 benefit Effects 0.000 description 2
- 230000015556 catabolic process Effects 0.000 description 2
- 238000006731 degradation reaction Methods 0.000 description 2
- 238000006056 electrooxidation reaction Methods 0.000 description 2
- -1 i.e. Substances 0.000 description 2
- 238000003062 neural network model Methods 0.000 description 2
- 230000008569 process Effects 0.000 description 2
- 238000012795 verification Methods 0.000 description 2
- 210000001367 artery Anatomy 0.000 description 1
- 230000008859 change Effects 0.000 description 1
- 238000010276 construction Methods 0.000 description 1
- 238000010586 diagram Methods 0.000 description 1
- 230000000694 effects Effects 0.000 description 1
- 238000005516 engineering process Methods 0.000 description 1
- 238000011156 evaluation Methods 0.000 description 1
- 230000004927 fusion Effects 0.000 description 1
- 230000007774 longterm Effects 0.000 description 1
- 230000007246 mechanism Effects 0.000 description 1
- 238000012986 modification Methods 0.000 description 1
- 230000004048 modification Effects 0.000 description 1
- 238000012545 processing Methods 0.000 description 1
- 239000011241 protective layer Substances 0.000 description 1
- 239000000126 substance Substances 0.000 description 1
Images
Classifications
-
- 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
- G06F30/27—Design optimisation, verification or simulation using machine learning, e.g. artificial intelligence, neural networks, support vector machines [SVM] or training a model
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/04—Architecture, e.g. interconnection topology
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/08—Learning methods
- G06N3/084—Backpropagation, e.g. using gradient descent
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F2113/00—Details relating to the application field
- G06F2113/14—Pipes
-
- 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
- G06F2119/04—Ageing analysis or optimisation against ageing
Landscapes
- Engineering & Computer Science (AREA)
- Theoretical Computer Science (AREA)
- Physics & Mathematics (AREA)
- Evolutionary Computation (AREA)
- Software Systems (AREA)
- Artificial Intelligence (AREA)
- General Physics & Mathematics (AREA)
- General Engineering & Computer Science (AREA)
- General Health & Medical Sciences (AREA)
- Mathematical Physics (AREA)
- Computational Linguistics (AREA)
- Molecular Biology (AREA)
- Computing Systems (AREA)
- Biophysics (AREA)
- Biomedical Technology (AREA)
- Data Mining & Analysis (AREA)
- Life Sciences & Earth Sciences (AREA)
- Health & Medical Sciences (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Medical Informatics (AREA)
- Computer Hardware Design (AREA)
- Geometry (AREA)
- Testing Resistance To Weather, Investigating Materials By Mechanical Methods (AREA)
Abstract
The invention discloses a method for predicting the residual life of pipeline corrosion. The method aims to solve the problems that in the prior art, the calculated amount is large, and the relation between data detected inside and outside a pipeline is not fully considered; the invention comprises the following steps: s1: carrying out comprehensive detection on the pipeline, and dividing the pipeline into different pipe sections according to comprehensive detection data; s2: respectively carrying out soil corrosivity detection on the pipe sections, and establishing a detection data set; s3: constructing a neural network for judging soil corrosivity, and determining a corroded pipe section; s4: carrying out a buried slice experiment near the corroded pipe section, calculating the corrosion rate, and establishing corrosion rate data sets of the pipe sections with different corrosion grades; s5: and constructing an extreme learning model for predicting the corrosion rate of the pipeline, calculating the maximum allowable corrosion depth according to the standard, and calculating the corrosion residual life of the pipeline according to the corrosion rate and the maximum allowable depth. The extreme learning machine is faster on the premise of ensuring the learning precision.
Description
Technical Field
The invention relates to the field of pipeline life prediction, in particular to a pipeline corrosion residual life prediction algorithm based on a neural network and an extreme learning machine.
Background
Oil and gas pipelines are the main arteries of national economy, and the safe operation of pipelines is also closely related to the life of people. With the increase of the running time of the pipeline, the problem of pipeline corrosion is more serious, and the pipeline is inevitably corroded and the like. Eventually causing the tubing to leak. Pipeline corrosion is one of the important potential safety hazards in long-distance oil and gas pipeline transportation.
The pipeline corrosion residual life prediction is used for evaluating the residual strength of the pipeline and predicting the residual life of the pipeline, so that the whole life cycle of the pipeline can be known and long-term planning of pipeline operation can be guided. With the increasing use and time of oil and gas pipelines that have been put into service, the safety assessment of pipelines in service and at an old age is imminent. The prediction of the residual life of the pipeline corrosion is an important content of safety evaluation.
The technology for predicting the residual service life of the corrosion of the pipeline can provide the most direct and accurate basis for the decision of a pipeline operator, so that the pipeline is maintained and repaired in a planned and targeted manner, the safety and reliability of the oil-gas pipeline in the service period are improved, and the continuous, stable, reasonable and efficient construction and operation of the oil-gas pipeline are realized.
The prior art mainly comprises an empirical formula method, a pitting corrosion-induced first leakage model prediction method and an electrochemical corrosion residual life prediction method.
The empirical formula method is to approximate the relationship between the corrosion depth change and the service life of the pipeline by using an exponential function, obtain undetermined coefficients by adopting a least square method according to the corrosion depth data of the pipeline actually measured on site, and then predict the service life of the pipeline. The scheme depends on expert experience, and human influence factors are large.
The first leakage model prediction caused by the outer wall pitting is that after an outer protective layer of the buried pipeline fails, a mathematical prediction model of the first leakage life caused by the pitting is established, and a semi-experience can be obtained by regression from statistical data of the pitting growth rate to obtain a first leakage life model of the outer wall pitting of the buried pipeline based on the corrosion state index.
The method for predicting the residual life of the electrochemical corrosion utilizes chemical knowledge to carry out corrosion mechanism modeling and predicts the residual life by predicting the corrosion rate.
For example, a method for predicting the residual life and reliability of a corroded oil and gas pipeline based on data fusion, disclosed in chinese patent literature, has a publication number CN108460230A, and comprises the following steps: 1) Obtaining a probability density function, a reliability function and a residual life function of the service life T of the corroded oil and gas pipeline; 2) Performing a double-stress constant accelerated degradation test, and collecting degradation data of a corroded oil and gas pipeline sample under each accelerated stress S0, S1, ss; 3) Finding out an acceleration model according to the type of acceleration stress by using a method for processing constant acceleration service life data in an acceleration service life test; 4) Acquiring a likelihood function according to the degraded data; 5) Simulating parameters in mu, sigma and an acceleration model by adopting Bayesian Monte Carlo; 6) And substituting the simulated result into the probability density function, the reliability function and the residual life function to obtain the probability density function, the reliability function and the residual life function of the service life T of the corroded oil and gas pipeline. However, the scheme is computationally intensive and does not adequately account for the correlation between the data detected inside and outside the pipeline.
Disclosure of Invention
The method mainly solves the problems that the prior art is large in calculated amount and does not fully consider the relation between data detected inside and outside the pipeline; the pipeline corrosion residual life prediction method is characterized in that a BP neural network is utilized to divide pipeline corrosion grades, an extreme learning machine is respectively used for predicting the pipeline corrosion rate of each pipeline grade, the pipeline corrosion residual life is calculated by calculating the maximum allowable corrosion depth of a pipeline and the pipeline corrosion rate predicted by the extreme learning machine, and the calculation speed is very high.
The technical problem of the invention is mainly solved by the following technical scheme:
a method for predicting the residual life of pipeline corrosion comprises the following steps:
s1: carrying out comprehensive detection on the pipeline, and dividing the pipeline into different pipe sections according to comprehensive detection data;
s2: respectively carrying out soil corrosivity detection on the pipe sections, and establishing a detection data set;
s3: constructing a neural network for judging soil corrosivity according to the detection data set, and determining a corroded pipe section;
s4: carrying out a buried slice experiment near the corroded pipe section, calculating the corrosion rate, and establishing corrosion rate data sets of the pipe sections with different corrosion grades;
s5: and constructing an extreme learning machine model for predicting the corrosion rate of the pipeline, calculating the maximum allowable corrosion depth according to the standard, and calculating the corrosion residual life of the pipeline according to the corrosion rate and the maximum allowable depth.
The scheme utilizes the BP neural network to divide the corrosion grade of the pipeline, uses the extreme learning machine to predict the corrosion rate of the pipeline for each grade of the pipeline, and calculates the residual life of the pipeline corrosion by calculating the maximum allowable corrosion depth of the pipeline and the corrosion rate of the pipeline predicted by the extreme learning machine. Thereby being beneficial to the safe operation of the pipeline. The pipeline sections are divided through comprehensive detection, the residual life is further estimated through soil corrosivity detection, and the relation between data detected inside and outside the pipeline is fully considered.
According to the scheme, the BP neural network is used for dividing the corrosion grade of the pipeline, and then the corrosion rate of the pipeline at each grade is predicted, so that the result is more targeted and more accurate. The connection weight of the input layer and the hidden layer of the extreme learning machine and the threshold value of the hidden layer can be randomly set, and the adjustment is not needed after the setting is finished, and only the weight of the output layer needs to be calculated. The use of the extreme learning machine can ensure that the generalization performance of the model is good, and the learning speed is higher than that of the traditional learning algorithm on the premise of ensuring the learning precision.
Preferably, the comprehensive detection comprises cathodic protection condition detection, stray current interference condition detection, anticorrosive coating damage point detection, pipeline buried depth detection, running pressure bearing capacity detection and steel pipe material detection. And performing full-attribute division on the pipeline into different pipe sections through the data.
Preferably, the soil corrosivity detection characteristics sequentially include: soil resistivity, oxidation-reduction potential, pH detection, water content and salt content;
the test data set was established as:
(x 1i ，...，x ni ，y i )
wherein x is 1i 1 st feature of the ith sample;
x ni is the nth feature of the ith sample;
n is the characteristic quantity of soil corrosivity detection;
y i the corrosion grade of the pipe of the ith sample.
And taking 80% of data in the data set as a training set for training the neural network model, taking 10% of data in the data set as a verification set for optimizing the structure of the model, and taking 10% of data in the data set as a test set for evaluating the performance of the selected model.
Preferably, the neural network for judging soil corrosivity sequentially comprises an input layer, two hidden layers and an output layer; n nodes corresponding to the characteristic quantity are arranged in the input layer; m neurons are respectively arranged in the hidden layer; the output layer is provided with k neurons corresponding to the number of corrosion levels. The input layer corresponds to the characteristic quantity, and the output layer corresponds to the corrosion grade quantity, so that the relevance of the internal and external detection data of the pipeline is embodied. And selecting the pipe sections of all grades according to the pipe section corrosion grades obtained by the neural network, determining the weak pipeline, excavating, and finding out the corroded pipe sections.
Preferably, the activation function of the hidden layer adopts a Sigmoid function; to add non-linearity to the neural network. The activation function of the output layer adopts a Softmax function; this transforms the output of the neural network into a probability distribution, so that the distance between the predicted probability distribution and the probability distribution of the true answer can be calculated by cross entropy. The loss function of the neural network adopts a cross entropy loss function.
Preferably, the calculation process of the corrosion rate is as follows:
wherein V is the corrosion rate;
λ is a proportionality coefficient;
W 0 is the weight of the test sample before corrosion;
W 1 the weight of the test sample after corrosion;
s is the surface area of the experimental sample exposed in the corrosive environment;
rho is the density of the experimental sample material;
t is the experimental time.
And (3) carrying out a buried test near the corroded pipe section, wherein the test piece is made of a steel pipe material the same as the material of the pipeline, so that the buried depth is consistent with the actual buried depth of the pipeline, and the thickness and the compactness of the backfill soil are the same as those of the original soil. After several days, the test piece was freed from corrosion products, washed, dried and weighed with an electronic balance. The corrosion rate was calculated.
Preferably, the corrosion rate data sets for the different corrosion grade pipe sections are:
(x 1i ，...，x ni ，y i )
wherein x is 1i 1 st feature of the ith sample;
x ni is the nth characteristic of the ith sample;
n is a characteristic number;
y vi the pipe corrosion rate for the ith sample. For establishing an extreme learning machine model.
Preferably, the output of the extreme learning machine is:
wherein β is an output weight between the hidden layer and the output layer;
β＝[β 1 ，...，β L ] T
h i (x) The output of the ith node of the hidden layer;
c is the number of hidden layer nodes;
the output of the hidden layer of the extreme learning machine is:
H(x)＝[h 1 (x)，...，h c (x)]
wherein h is 1 (x) The output of the 1 st node of the hidden layer;
h c (x) Is the output of the c-th node of the hidden layer.
The hidden layer node parameters w, b are randomly generated (independent of training data) according to any continuous probability distribution, rather than being determined through training, thereby causing great advantages in efficiency compared with the conventional BP neural network.
Preferably, the minimum square error is obtained through H (x) beta and the sample label B and is used as an objective function for evaluating the training error, and the minimum solution of the objective function is the optimal solution. Beta according to good results is obtained.
Preferably, the ASME B31G-1984 standard is used to determine the maximum allowable depth of corrosion for a pipe section:
wherein, d max The maximum allowable corrosion depth of the pipeline;
p rum the operating pressure capacity borne by the pipeline;
l is the length value of the corrosion defect part;
d is the depth value of the corrosion defect part;
σ flow is the flow stress;
d is the diameter of the pipeline;
m is Folias expansion coefficient;
e is the pipe wall thickness.
The maximum allowable etch depth is calculated.
The invention has the beneficial effects that:
1. the BP neural network is utilized to divide the corrosion grade of the pipeline, and then the corrosion rate of the pipeline at each grade is predicted, so that the result is more targeted and more accurate.
2. The connection weight of the input layer and the hidden layer of the extreme learning machine and the threshold value of the hidden layer can be randomly set, and the adjustment is not needed after the setting is finished, and only the weight of the output layer needs to be calculated. The use of the extreme learning machine can ensure that the generalization performance of the model is good, and the learning speed is higher than that of the traditional learning algorithm on the premise of ensuring the learning precision.
3. The pipeline sections are divided through comprehensive detection, the residual life is further estimated through soil corrosivity detection, and the relation between data detected inside and outside the pipeline is fully considered.
Drawings
FIG. 1 is a flow chart of the method for predicting the residual life of corrosion of a pipeline according to the present invention.
Fig. 2 is a schematic diagram of the neural network structure of the present invention.
Detailed Description
The technical scheme of the invention is further specifically described by the following embodiments and the accompanying drawings.
The embodiment is as follows:
the method for predicting the remaining life of corrosion of a pipeline in the embodiment, as shown in fig. 1, includes the following steps:
a method for predicting the residual life of pipeline corrosion comprises the following steps:
s1: and carrying out comprehensive detection on the pipeline, and dividing the pipeline into different pipe sections according to comprehensive detection data.
The comprehensive detection comprises cathodic protection condition detection, stray current interference condition detection, anticorrosive coating damage point detection, pipeline burial depth detection, running pressure bearing capacity detection and steel pipe material detection.
And performing full-attribute division on the pipeline into different pipeline sections through the data.
S2: and respectively carrying out soil corrosivity detection on the pipe sections, and establishing a detection data set.
And carrying out more detailed soil corrosivity detection on each divided pipe section. In this embodiment, the soil corrosivity detection features sequentially include: soil resistivity, oxidation-reduction potential, pH detection, water content and salt content.
The test data set was established as:
(x 1i ，x 2i ，x 3i ，x 4i ，x 5i ，y i )
wherein x is 1i The 1 st characteristic of the ith sample, namely the soil resistivity.
x 2i Is the 2 nd characteristic of the ith sample, i.e., the oxidation-reduction potential.
x 3i Is the 3 rd characteristic of the ith sample, namely pH detection.
x 4i Is the 4 th characteristic of the ith sample, i.e., water cut.
x 5i Is the 5 th characteristic of the ith sample, i.e., salt content.
y i The corrosion grade of the pipe is the ith sample.
In this example, the corrosion grade of the pipeline is determined by experts and is divided into 5 corrosion grades of weak, medium, strong and strong.
Taking 80% of data in the test data set as a training set for training a neural network model; 10% of the data is used as a verification set and used for optimizing the model structure; 10% was used as a test set to evaluate the performance of the selected model.
S3: and constructing a neural network for judging soil corrosivity according to the detection data set, and determining the corroded pipe section.
In this embodiment, the number of network layers of the neural network is four.
The neural network for judging soil corrosivity sequentially comprises an input layer, two hidden layers and an output layer.
5 nodes are arranged in the input layer and correspond to 5-dimensional features;
the hidden layers are respectively provided with 10 neurons. The activation function of the hidden layer uses a Sigmoid function to add non-linearity to the neural network.
The output layer is provided with 5 neurons corresponding to 5 corrosion levels of extremely weak, medium, strong and extremely strong. The activation function of the output layer adopts a Softmax function; this transforms the output of the neural network into a probability distribution, so that the distance between the predicted probability distribution and the probability distribution of the true answer can be calculated by cross entropy.
The loss function of the neural network adopts a cross entropy loss function.
The input layer corresponds to the characteristic quantity, and the output layer corresponds to the corrosion grade quantity, so that the relevance of the internal and external detection data of the pipeline is embodied.
And selecting the pipe sections of all grades according to the pipe section corrosion grades obtained by the neural network, determining the weak pipeline, excavating, and finding out the corroded pipe sections.
S4: and (4) carrying out a buried slice experiment near the corroded pipe section, calculating the corrosion rate, and establishing corrosion rate data sets of the pipe sections with different corrosion grades.
And (3) carrying out a buried piece test near the corroded pipe section, wherein the test piece material is a steel pipe material the same as the material of the pipeline, so that the buried depth is consistent with the actual buried depth of the pipeline, and the thickness and the compactness of the backfill soil are the same as those of the original soil. After several days, the test piece was freed from corrosion products, washed, dried and weighed with an electronic balance. The corrosion rate was calculated.
The calculation process of the corrosion rate is as follows:
wherein V is the corrosion rate.
Lambda is a proportionality coefficient; in this implementation, the scaling factor is 8.76.
W 0 The weight of the test specimen before corrosion.
W 1 The weight of the test sample after corrosion.
And S is the surface area of the experimental sample exposed to the corrosive environment.
ρ is the material density of the experimental sample.
t is the experimental time.
Establishing a corrosion rate data set of pipe sections with different corrosion grades as follows:
(x 1i ，x 2i ，x 3i ，x 4i ，x 5i ，y i )
wherein, x 1i The 1 st characteristic of the ith sample, namely the soil resistivity.
x 2i Is the 2 nd characteristic of the ith sample, i.e., the oxidation-reduction potential.
x 3i Is the 3 rd characteristic of the ith sample, namely pH detection.
x 4i Is the 4 th characteristic of the ith sample, i.e., water cut.
x 5i Is the 5 th characteristic of the ith sample, i.e., salt content.
y vi The corrosion rate of the pipeline is obtained by calculation for the ith sample.
S5: and constructing an extreme learning model for predicting the corrosion rate of the pipeline, calculating the maximum allowable corrosion depth according to the standard, and calculating the corrosion residual life of the pipeline according to the corrosion rate and the maximum allowable depth.
The output of the extreme learning machine is:
where β is the output weight between the hidden layer and the output layer.
β＝[β 1 ，...，β c ] T
h i (x) Is the output of the ith node of the hidden layer.
c is the number of hidden layer nodes. In this embodiment, the number of hidden layer nodes is 5, so c =5.
The output H (x) of the ELM hidden layer of the extreme learning machine is:
H(x)＝[h 1 (x)，...，h c (x)]
wherein h is 1 (x) Is the output of the 1 st node of the hidden layer.
h c (x) Is the output of the c-th node of the hidden layer.
h i (x)＝g(w i ，b i ，x)＝g(w i x+b i )
w i ∈R
b i ∈R
Wherein, g (w) i ，b i X) is an activation function Sigmoid, which may add non-linearity to the network.
w i Are weights of the hidden layer.
b i Is the biasing of the hidden layer.
The hidden layer node parameters w, b are randomly generated (independent of the training data) according to an arbitrary continuous probability distribution, rather than being determined through training, thereby causing a great advantage in efficiency compared with the conventional BP neural network.
In order to obtain β with good effect on the training sample set, it is necessary to ensure that its training error is minimal.
And solving the minimum square error through H (x) beta and the sample label B to serve as an objective function for evaluating the training error, wherein the minimum solution of the objective function is the optimal solution. Namely, the output weight is solved by a method of minimizing the approximate square difference, and the objective function is as follows: min | | | H (x) beta-B | | Y 2
Wherein H (x) is the output of the hidden layer;
beta is the output weight;
b is a sample label.
Deriving the optimal solution beta of the objective function through the knowledge of the line generation and the matrix theory * Comprises the following steps:
wherein, the first and the second end of the pipe are connected with each other,
Moore-Penrose generalized inverse matrix, which is the matrix H (x).
And after the output weight beta is obtained through calculation, the extreme learning machine is designed.
The ASME B31G-1984 standard was used to determine the maximum allowable depth of corrosion for a pipe section:
wherein d is max The maximum allowable depth of corrosion of the pipe.
p rum The operating pressure capability that the pipeline withstands.
L is the length of the corrosion defect.
d is the depth value of the corrosion defect part.
σ flow Is the rheological stress.
D is the diameter of the pipeline.
M is Folias expansion coefficient.
e is the pipe wall thickness.
And acquiring the soil resistivity, the oxidation-reduction potential, the pH detection, the water content and the salt content of the pipeline, using the acquired data as the input of a previously trained extreme learning machine to obtain the corrosion rate of the pipeline, and then calculating the maximum allowable corrosion depth/corrosion rate of the pipeline to obtain the corrosion residual life of the pipeline.
According to the scheme of the embodiment, the BP neural network is utilized to divide the corrosion grade of the pipeline, the extreme learning machine is respectively used for predicting the corrosion rate of the pipeline for each grade of the pipeline, and the corrosion residual life of the pipeline is calculated by calculating the maximum allowable corrosion depth of the pipeline and the corrosion rate of the pipeline predicted by the extreme learning machine. Thereby being beneficial to the safe operation of the pipeline.
The BP neural network is utilized to divide the corrosion grade of the pipeline, and then the corrosion rate of the pipeline at each grade is predicted, so that the result is more targeted and more accurate. The connection weight of the input layer and the hidden layer of the extreme learning machine and the threshold value of the hidden layer can be randomly set, and the adjustment is not needed after the setting is finished, and only the weight of the output layer needs to be calculated. The use of the extreme learning machine can ensure that the generalization performance of the model is good, and the learning speed is higher than that of the traditional learning algorithm on the premise of ensuring the learning precision.
It should be understood that the examples are for illustrative purposes only and are not intended to limit the scope of the present invention. Further, it should be understood that various changes or modifications of the present invention may be made by those skilled in the art after reading the teaching of the present invention, and such equivalents may fall within the scope of the present invention as defined in the appended claims.
Claims (10)
1. A method for predicting the residual life of pipeline corrosion is characterized by comprising the following steps:
s1: carrying out comprehensive detection on the pipeline, and dividing the pipeline into different pipe sections according to comprehensive detection data;
s2: respectively carrying out soil corrosivity detection on the pipe sections, and establishing a detection data set;
s3: constructing a neural network for judging soil corrosivity according to the detection data set, and determining a corroded pipe section;
s4: carrying out a buried slice experiment near the corroded pipe section, calculating the corrosion rate, and establishing corrosion rate data sets of the pipe sections with different corrosion grades;
s5: and constructing an extreme learning model for predicting the corrosion rate of the pipeline, calculating the maximum allowable corrosion depth according to the standard, and calculating the corrosion residual life of the pipeline according to the corrosion rate and the maximum allowable corrosion depth.
2. The method as claimed in claim 1, wherein the comprehensive detection includes cathodic protection detection, stray current interference detection, anticorrosive coating damage point detection, pipeline burial depth detection, running pressure bearing capacity detection and steel pipe material detection.
3. The method for predicting the residual life of corrosion of a pipeline according to claim 1 or 2, wherein the characteristics of soil corrosivity detection sequentially comprise: soil resistivity, oxidation-reduction potential, pH detection, water content and salt content;
the test data set was established as:
(x 1i ，...，x ni ，y i )
wherein x is 1i 1 st feature of the ith sample;
x ni is the nth characteristic of the ith sample;
n is the characteristic quantity of soil corrosivity detection;
y i the corrosion grade of the pipe is the ith sample.
4. The method for predicting the residual life of pipeline corrosion according to claim 3, wherein the neural network for judging soil corrosivity sequentially comprises an input layer, two hidden layers and an output layer; n nodes corresponding to the characteristic quantity are arranged in the input layer; m neurons are respectively arranged in the hidden layer; the output layer is provided with k neurons corresponding to the number of corrosion levels.
5. The method for predicting the residual life of pipeline corrosion according to claim 4, wherein the activation function of the hidden layer is a Sigmoid function; the activating function of the output layer adopts a Softmax function; the loss function of the neural network adopts a cross entropy loss function.
6. The method for predicting the residual life of the pipeline corrosion according to claim 1, wherein the corrosion rate is calculated by the following steps:
wherein V is the corrosion rate;
λ is a proportionality coefficient;
W 0 is the weight of the test sample before corrosion;
W 1 is the weight of the test sample after corrosion;
s is the surface area of the experimental sample exposed in the corrosive environment;
rho is the density of the experimental sample material;
t is the experimental time.
7. The method for predicting the residual life of pipeline corrosion according to claim 1 or 6, wherein the corrosion rate data sets of the pipeline sections with different corrosion grades are as follows:
(x 1i ，...，x ni ，y i )
wherein x is 1i 1 st feature of the ith sample;
x ni is the nth characteristic of the ith sample;
n is a characteristic number;
y vi is the pipe erosion rate for the ith sample.
8. The method of claim 1, wherein the output of the extreme learning machine is:
wherein β is an output weight between the hidden layer and the output layer;
β＝[β 1 ，...，β c ] T
h i (x) The output of the ith node of the hidden layer;
c is the number of hidden layer nodes;
the output of the hidden layer of the extreme learning machine is:
H(x)＝[h 1 (x)，...，h c (x)]
wherein h is 1 (x) Is the output of the 1 st node of the hidden layer;
h c (x) Is the output of the c-th node of the hidden layer.
9. The method for predicting the residual life of pipeline corrosion according to claim 1 or 8, wherein a minimum squared error is obtained through H (x) beta and a sample label B and is used as an objective function for evaluating a training error, and a minimum solution of the objective function is an optimal solution.
10. The method of claim 1, wherein the ASME B31G-1984 standard is used to determine the maximum allowable corrosion depth of the pipe section:
wherein d is max The maximum allowable corrosion depth of the pipeline;
p rum the operating pressure capacity borne by the pipeline;
l is the length value of the corrosion defect part;
d is the depth value of the corrosion defect part;
σ flow is the rheological stress;
d is the diameter of the pipeline;
m is Folias expansion coefficient;
e is the pipe wall thickness.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202210829744.9A CN115186590A (en) | 2022-07-13 | 2022-07-13 | Method for predicting residual life of pipeline corrosion |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202210829744.9A CN115186590A (en) | 2022-07-13 | 2022-07-13 | Method for predicting residual life of pipeline corrosion |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN115186590A true CN115186590A (en) | 2022-10-14 |
Family
ID=83518642
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202210829744.9A Pending CN115186590A (en) | 2022-07-13 | 2022-07-13 | Method for predicting residual life of pipeline corrosion |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN115186590A (en) |
Cited By (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN115639135A (en) * | 2022-10-24 | 2023-01-24 | 广州市万保职业安全事务有限公司 | Steel corrosion safety detection method and system based on machine vision |
| CN115983116A (en) * | 2022-12-22 | 2023-04-18 | 新疆敦华绿碳技术股份有限公司 | Carbon dioxide miscible flooding corrosion detection method |
| US11879599B2 (en) * | 2022-12-16 | 2024-01-23 | Chengdu Qinchuan Iot Technology Co., Ltd. | Methods, Internet of Things systems, and mediums for assessing electrochemical corrosion of smart gas pipeline |
-
2022
- 2022-07-13 CN CN202210829744.9A patent/CN115186590A/en active Pending
Cited By (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN115639135A (en) * | 2022-10-24 | 2023-01-24 | 广州市万保职业安全事务有限公司 | Steel corrosion safety detection method and system based on machine vision |
| CN115639135B (en) * | 2022-10-24 | 2023-10-10 | 广州市万保职业安全事务有限公司 | Steel corrosion safety detection method and system based on machine vision |
| US11879599B2 (en) * | 2022-12-16 | 2024-01-23 | Chengdu Qinchuan Iot Technology Co., Ltd. | Methods, Internet of Things systems, and mediums for assessing electrochemical corrosion of smart gas pipeline |
| CN115983116A (en) * | 2022-12-22 | 2023-04-18 | 新疆敦华绿碳技术股份有限公司 | Carbon dioxide miscible flooding corrosion detection method |
| CN115983116B (en) * | 2022-12-22 | 2024-05-24 | 新疆敦华绿碳技术股份有限公司 | Carbon dioxide miscible flooding corrosion detection method |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| Valor et al. | Reliability assessment of buried pipelines based on different corrosion rate models | |
| CN115186590A (en) | Method for predicting residual life of pipeline corrosion | |
| Ji et al. | Probabilistic physical modelling of corroded cast iron pipes for lifetime prediction | |
| Caleyo et al. | Probability distribution of pitting corrosion depth and rate in underground pipelines: A Monte Carlo study | |
| Gomes et al. | Optimal inspection planning for onshore pipelines subject to external corrosion | |
| Caleyo et al. | Markov chain modelling of pitting corrosion in underground pipelines | |
| Sinha et al. | Probabilistic neural network for reliability assessment of oil and gas pipelines | |
| Ayello et al. | Quantitive assessment of corrosion probability—a Bayesian network approach | |
| CN107063991B (en) | Dynamic safety assessment method and device for internal corrosion defects of conveying pipeline | |
| Heidary et al. | A review of data-driven oil and gas pipeline pitting corrosion growth models applicable for prognostic and health management | |
| Liu et al. | Corrosion failure probability analysis of buried gas pipelines based on subset simulation | |
| CN113468757B (en) | Method for evaluating reliability of corroded natural gas pipeline based on fuzzy random theory | |
| CN114841080A (en) | Gas pipeline internal corrosion prediction method based on BP neural network and SD coupling | |
| CN114638164B (en) | Method for predicting high-temperature creep life of power station pressure pipeline | |
| Wang et al. | Novel method for prediction of corrosion current density of gas pipeline steel under stray current interference based on hybrid LWQPSO-NN model | |
| Li et al. | Pipelines reliability assessment considering corrosion-related failure modes and probability distributions characteristic using subset simulation | |
| CN115358102B (en) | Corrosion pipeline failure probability analysis method under earthquake action of introduced internal detection data | |
| Taiwo et al. | Predicting wall thickness loss in water pipes using machine learning techniques | |
| Nasser et al. | A review on oil and gas pipelines corrosion growth rate modelling incorporating artificial intelligence approach | |
| Guan et al. | Application of probabilistic model in pipeline direct assessment | |
| Han et al. | A dynamic condition assessment model of aging subsea pipelines subject to corrosion-fatigue degradation | |
| Zangenehmadar et al. | Application of neural networks in predicting the remaining useful life of water pipelines | |
| CN111667133B (en) | Comprehensive evaluation method and system for integrity of pipeline | |
| Uruchurtu-Chavarin et al. | Artificial intelligence for the assessment on the corrosion conditions diagnosis of transmission line tower foundations | |
| Alqarni et al. | Predicting Corrosion Growth and Reliability of the Oil Pipeline |
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 | ||
| TA01 | Transfer of patent application right |
Effective date of registration: 20231212 Address after: 310000 Yuhang Tang Road, Xihu District, Hangzhou, Zhejiang 866 Applicant after: ZHEJIANG University Applicant after: Zhejiang Provincial Natural Gas Development Co.,Ltd. Applicant after: ZHEJIANG ENERGY GROUP CO.,LTD. Address before: 310012 No. 38, Zhejiang Road, Hangzhou, Zhejiang, Xihu District Applicant before: ZHEJIANG University Applicant before: ZHEJIANG ZHENENG NATURAL GAS OPERATION CO.,LTD. Applicant before: ZHEJIANG ENERGY GROUP CO.,LTD. |
|
| TA01 | Transfer of patent application right |