CN113052207A - Stainless steel environment cracking judgment method and system based on Hybrid idea - Google Patents
Stainless steel environment cracking judgment method and system based on Hybrid idea Download PDFInfo
- Publication number
- CN113052207A CN113052207A CN202110258345.7A CN202110258345A CN113052207A CN 113052207 A CN113052207 A CN 113052207A CN 202110258345 A CN202110258345 A CN 202110258345A CN 113052207 A CN113052207 A CN 113052207A
- Authority
- CN
- China
- Prior art keywords
- stainless steel
- classification model
- classification
- hybrid
- parameters
- 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
- 229910001220 stainless steel Inorganic materials 0.000 title claims abstract description 76
- 239000010935 stainless steel Substances 0.000 title claims abstract description 76
- 238000005336 cracking Methods 0.000 title claims abstract description 31
- 238000000034 method Methods 0.000 title claims abstract description 31
- 238000013145 classification model Methods 0.000 claims abstract description 154
- 239000002245 particle Substances 0.000 claims abstract description 55
- 230000007613 environmental effect Effects 0.000 claims abstract description 34
- 238000012549 training Methods 0.000 claims abstract description 27
- 238000005457 optimization Methods 0.000 claims abstract description 18
- ZAMOUSCENKQFHK-UHFFFAOYSA-N Chlorine atom Chemical compound [Cl] ZAMOUSCENKQFHK-UHFFFAOYSA-N 0.000 claims description 9
- QVGXLLKOCUKJST-UHFFFAOYSA-N atomic oxygen Chemical compound [O] QVGXLLKOCUKJST-UHFFFAOYSA-N 0.000 claims description 9
- 239000000460 chlorine Substances 0.000 claims description 9
- 229910052801 chlorine Inorganic materials 0.000 claims description 9
- 239000001301 oxygen Substances 0.000 claims description 9
- 229910052760 oxygen Inorganic materials 0.000 claims description 9
- 238000003066 decision tree Methods 0.000 claims description 8
- 230000009466 transformation Effects 0.000 claims description 5
- 238000012795 verification Methods 0.000 claims description 5
- 238000012360 testing method Methods 0.000 claims description 4
- 238000010606 normalization Methods 0.000 claims description 3
- 238000012850 discrimination method Methods 0.000 claims 1
- 238000012706 support-vector machine Methods 0.000 description 19
- 238000005260 corrosion Methods 0.000 description 7
- 230000007797 corrosion Effects 0.000 description 7
- 230000006870 function Effects 0.000 description 6
- 238000010801 machine learning Methods 0.000 description 4
- 230000006399 behavior Effects 0.000 description 3
- 230000000694 effects Effects 0.000 description 3
- 230000001133 acceleration Effects 0.000 description 2
- 230000003044 adaptive effect Effects 0.000 description 2
- 238000013528 artificial neural network Methods 0.000 description 2
- 238000007781 pre-processing Methods 0.000 description 2
- 230000035945 sensitivity Effects 0.000 description 2
- 230000009471 action Effects 0.000 description 1
- 229910000963 austenitic stainless steel Inorganic materials 0.000 description 1
- 238000003889 chemical engineering Methods 0.000 description 1
- 230000000295 complement effect Effects 0.000 description 1
- 230000008878 coupling Effects 0.000 description 1
- 238000010168 coupling process Methods 0.000 description 1
- 238000005859 coupling reaction Methods 0.000 description 1
- 238000002790 cross-validation Methods 0.000 description 1
- 230000003247 decreasing effect Effects 0.000 description 1
- 230000001066 destructive effect Effects 0.000 description 1
- 238000011161 development Methods 0.000 description 1
- 238000010586 diagram Methods 0.000 description 1
- 238000004090 dissolution Methods 0.000 description 1
- 238000006056 electrooxidation reaction Methods 0.000 description 1
- 238000011156 evaluation Methods 0.000 description 1
- 239000000463 material Substances 0.000 description 1
- 230000007246 mechanism Effects 0.000 description 1
- 238000007254 oxidation reaction Methods 0.000 description 1
- 239000003208 petroleum Substances 0.000 description 1
- 230000008569 process Effects 0.000 description 1
- 230000000750 progressive effect Effects 0.000 description 1
- 238000011160 research Methods 0.000 description 1
- XLYOFNOQVPJJNP-UHFFFAOYSA-N water Substances O XLYOFNOQVPJJNP-UHFFFAOYSA-N 0.000 description 1
Images
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/24—Classification techniques
- G06F18/241—Classification techniques relating to the classification model, e.g. parametric or non-parametric approaches
- G06F18/2411—Classification techniques relating to the classification model, e.g. parametric or non-parametric approaches based on the proximity to a decision surface, e.g. support vector machines
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/21—Design or setup of recognition systems or techniques; Extraction of features in feature space; Blind source separation
- G06F18/214—Generating training patterns; Bootstrap methods, e.g. bagging or boosting
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/004—Artificial life, i.e. computing arrangements simulating life
- G06N3/006—Artificial life, i.e. computing arrangements simulating life based on simulated virtual individual or collective life forms, e.g. social simulations or particle swarm optimisation [PSO]
Landscapes
- Engineering & Computer Science (AREA)
- Theoretical Computer Science (AREA)
- Data Mining & Analysis (AREA)
- Physics & Mathematics (AREA)
- Life Sciences & Earth Sciences (AREA)
- Artificial Intelligence (AREA)
- General Physics & Mathematics (AREA)
- General Engineering & Computer Science (AREA)
- Evolutionary Computation (AREA)
- Bioinformatics & Computational Biology (AREA)
- Evolutionary Biology (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Bioinformatics & Cheminformatics (AREA)
- Computational Linguistics (AREA)
- Biomedical Technology (AREA)
- Biophysics (AREA)
- Health & Medical Sciences (AREA)
- General Health & Medical Sciences (AREA)
- Molecular Biology (AREA)
- Computing Systems (AREA)
- Mathematical Physics (AREA)
- Software Systems (AREA)
- Testing Resistance To Weather, Investigating Materials By Mechanical Methods (AREA)
Abstract
The invention discloses a stainless steel environment cracking judgment method and system based on a Hybrid idea. The method comprises the following steps: optimizing preset parameters of a plurality of types of classification models by adopting a particle swarm optimization algorithm to obtain optimal classification models based on the respective optimal preset parameters, wherein the fitness of particles is determined by adopting the accuracy of the classification models obtained by training the particles, samples used for training the classification models are environmental parameters, and a label is whether stainless steel cracks or not; predicting whether stainless steel cracks or not by adopting each optimal classification model according to the environmental parameters of the stainless steel to obtain the prediction result of each optimal classification model; determining whether the stainless steel cracks or not based on each prediction result. The method has higher prediction accuracy and higher generalization capability on the cracking condition of the stainless steel. In addition, the particle swarm optimization algorithm is introduced to determine the optimal parameters required by each classification model, so that the unscientific nature of artificially given parameters is avoided.
Description
Technical Field
The invention relates to the field of stainless steel crack discrimination, in particular to a method and a system for discriminating stainless steel environmental crack based on a Hybrid idea.
Background
The stainless steel is widely applied to industries such as nuclear power, petroleum, chemical engineering and the like due to excellent heat resistance, corrosion resistance and mechanical property, and plays an irreplaceable role in national economy development. Stress Corrosion Cracking (SCC) refers to the failure Cracking of a material under the combined action of Stress and the environment. SCC is extremely complex, destructive, and prone to sudden and catastrophic results, usually in low stress conditions with no obvious signs, and has been a major safety hazard during stainless steel service. Therefore, the research on the prediction of SCC is particularly important, and the results thereof can give certain inspiration to enterprises so as to take corresponding security measures to deal with risks in advance and ensure the safety of lives and properties of people.
Compared with the models for predicting the SCC crack propagation rate based on the slip-oxidation/dissolution, Shoji, environmental coupling Cracking (CEFM), and the like proposed by the SCC mechanism, the empirical model represented by machine learning provides us with a new choice. Machine learning is applied to stainless steel stress corrosion cracking sensitivity prediction by partial scholars at home and abroad, such as a BP neural network algorithm belonging to a link meaning and a Support Vector Machine (SVM) belonging to a behavior analogism. The results show that these models perform well, demonstrating that machine learning is feasible in predicting SCC. However, these algorithms have their own disadvantages, such as poor interpretability of the BP neural network, easy falling into local extrema, obvious shortcomings of SVMs in handling large samples and multiple classes of problems, and easy overfitting of them.
Disclosure of Invention
The invention aims to provide a stainless steel environment crack distinguishing method and system based on the Hybrid idea, which are high in classification accuracy and strong in generalization capability.
In order to achieve the purpose, the invention provides the following scheme:
a stainless steel environment cracking judgment method based on a Hybrid idea comprises the following steps:
optimizing preset parameters of a plurality of types of classification models by adopting a particle swarm optimization algorithm to obtain optimal classification models based on the respective optimal preset parameters, wherein the fitness of particles is determined by adopting the accuracy of the classification models obtained by training the particles, samples used for training the classification models are environmental parameters, and a label is whether stainless steel cracks or not;
predicting whether stainless steel cracks or not by adopting each optimal classification model according to the environmental parameters of the stainless steel to obtain the prediction result of each optimal classification model;
determining whether the stainless steel cracks or not based on each prediction result.
Optionally, the classification model includes at least one of an RF classification model, an XGBoost classification model, and an SVM classification model.
Optionally, the preset parameters of the RF classification model and the XGBoost classification model both include the number of decision trees and the maximum depth of the trees; the preset parameters of the SVM classification model comprise tolerance of errors and a kernel function self-contained parameter sigma.
Optionally, before determining whether the stainless steel cracks based on each prediction result, the method further includes:
and predicting whether the stainless steel cracks or not according to the environmental parameters of the stainless steel by adopting an NBM classification model and/or an ELM classification model respectively to obtain the prediction result of each classification model.
Optionally, the method for calculating the particle fitness includes:
dividing a training set into n parts, wherein n is an integer greater than or equal to 2;
taking n-1 parts of training classification models and 1 part of verification models in turn, and calculating n times of tests to obtain average accuracy, wherein preset parameters in the classification models are provided by the particles, the average accuracy is the fitness of the particles, and the higher the fitness, the more excellent the particles are.
Optionally, the determining whether the stainless steel cracks based on each prediction result specifically includes:
and giving a final prediction result by hard voting according to each prediction result.
Optionally, the environmental parameters include temperature, chlorine content, and oxygen content.
Optionally, before training the classification model, the method further includes:
carrying out logarithmic transformation on the chlorine content and the oxygen content in the sample data;
and respectively carrying out normalization treatment on the temperature in the sample data and the chlorine content and the oxygen content after logarithmic transformation.
The invention also provides a stainless steel environment cracking discrimination system based on the Hybrid idea, which is characterized by comprising the following steps:
the optimized classification model determining module is used for optimizing preset parameters of the classification models by adopting a particle swarm algorithm to obtain the optimal classification models based on the respective optimal preset parameters, wherein the fitness of particles is determined by adopting the accuracy of the classification models obtained by training the particles, samples used for training the classification models are environmental parameters, and labels are whether stainless steel cracks or not;
the classification model prediction module is used for predicting whether the stainless steel cracks or not according to the environment parameters of the stainless steel by adopting each optimal classification model to obtain the prediction result of each optimal classification model;
and the cracking condition determining module is used for determining whether the stainless steel cracks or not based on each prediction result.
Alternatively to this, the first and second parts may,
the classification model in the optimization classification model determination module comprises at least one of an RF classification model, an XGboost classification model and an SVM classification model; the preset parameters of the RF classification model and the XGboost classification model comprise the number of decision trees and the maximum depth of the trees; the preset parameters of the SVM classification model comprise tolerance of errors and a kernel function self-contained parameter sigma;
the classification model prediction module is further used for predicting whether the stainless steel cracks or not according to the environmental parameters of the stainless steel by adopting an NBM classification model and/or an ELM classification model respectively to obtain the prediction results of the classification models.
According to the specific embodiment provided by the invention, the invention discloses the following technical effects: the stainless steel environmental cracking distinguishing method and system based on the Hybrid idea provided by the invention are based on the Hybrid idea, a plurality of types of base models in machine learning are selected to form the Hybrid model, and the final prediction result of the Hybrid model on the stainless steel cracking condition is determined by the classification prediction results of the base models forming the Hybrid model, so that different base models are complementary to each other to a certain extent, and the stainless steel stress corrosion cracking prediction model with higher accuracy and higher generalization capability is obtained. In addition, the particle swarm optimization algorithm is introduced to determine the optimal parameters required by the base model, so that unscientific manual parameter setting is avoided.
Drawings
In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the drawings needed to be used in the embodiments will be briefly described below, and it is obvious that the drawings in the following description are only some embodiments of the present invention, and it is obvious for those skilled in the art to obtain other drawings without inventive exercise.
FIG. 1 is a schematic flow chart of a method for discriminating stainless steel environmental cracking based on Hybrid concept provided in embodiment 1 of the present invention;
FIG. 2 is a flowchart of the Hybrid model in embodiment 1 of the present invention;
FIG. 3 is a graph showing the variation of the accuracy of the Hybrid model and the basic model thereof in example 1 of the present invention;
FIG. 4 is a graph showing the accuracy before and after the PSO optimization of the base model in example 1 of the present invention;
FIG. 5 is a graph showing the comparison of the accuracy of the PSO-Hybrid model and the accuracy of the Hybrid model in example 1 of the present invention;
fig. 6 is a schematic structural diagram of a stainless steel environment cracking determination system based on the Hybrid idea provided in embodiment 2 of the present invention.
Detailed Description
The technical solutions in the embodiments of the present invention will be clearly and completely described below with reference to the drawings in the embodiments of the present invention, and it is obvious that the described embodiments are only a part of the embodiments of the present invention, and not all of the embodiments. All other embodiments, which can be derived by a person skilled in the art from the embodiments given herein without making any creative effort, shall fall within the protection scope of the present invention.
The invention aims to provide a stainless steel environment crack distinguishing method and system based on the Hybrid idea, which are high in classification accuracy and strong in generalization capability.
In order to make the aforementioned objects, features and advantages of the present invention comprehensible, embodiments accompanied with figures are described in further detail below.
The embodiment is used for judging the cracking condition of the stainless steel pipeline, namely judging the cracking possibility of the stainless steel pipeline by acquiring environmental parameters (such as temperature, chlorine content, oxygen content and the like) of the environment where the stainless steel pipeline is located and based on the environmental parameters.
Example 1
Referring to fig. 1, the present embodiment provides a method for determining stainless steel environmental cracking based on the Hybrid idea, which includes the following steps:
step 101: and optimizing preset parameters of a plurality of types of classification models by adopting a particle swarm optimization algorithm to obtain an optimal classification model based on the respective optimal preset parameters, wherein the fitness of the particles is determined by adopting the accuracy of the classification model obtained by training the particles, a sample used for training the classification model is an environmental parameter, and a label is whether stainless steel cracks or not.
In this embodiment, the classification model that needs to optimize the preset parameters includes, but is not limited to, at least one of an RF classification model, an XGBoost classification model, and an SVM classification model. The preset parameters of the RF classification model may be the number of decision trees and the maximum depth of the trees, the preset parameters of the XGBoost classification model may also be the number of decision trees and the maximum depth of the trees, and the preset parameters of the SVM classification model may be the tolerance of an error and the parameter σ of the kernel function itself. The particle swarm algorithm can optimize two parameters at the same time. Although the NBM classification model and the ELM classification model have fewer optimizable parameters, in other embodiments, the parameters in the NBM classification model and the ELM classification model can be optimized by means of a particle swarm optimization. In this embodiment, as an implementation manner of this embodiment, since the NBM classification model and the ELM classification model have fewer optimizable parameters, the NBM classification model and the ELM classification model may not perform parameter optimization, and the parameters are directly given by human.
In this embodiment, the specific process of parameter optimization may be as follows:
(1) initializing a particle swarm:
and determining an optimized parameter object, wherein the parameter object of the SVM classification model is the tolerance C of the error and the parameter sigma of the kernel function. The parameter objects of the RF classification model and the XGboost classification model are the number of decision trees n _ estimators and the maximum depth of the trees max _ depth.
The basic parameters of the particle swarm are set artificially, and the number of particles is set to be 50 in the embodiment; the number of iterations is 300; local particle acceleration factor C 12; global particle acceleration factor C 22; the inertial weight factor ω is 0.6. In addition, there is also a two-dimensional search space for the particle swarm, i.e., the range of values of the parameter min _ value, max _ value]. For the SVM classification model, the optimization range of the parameters is error tolerance C to [0.001, 15%]The kernel function has parameters of 0.001,15]For the RF classification model and the XGboost classification model, the optimization ranges of the parameters are n _ estimators to [1,30 ] respectively]And max _ depth to [1,10 ]]. 50 particles of random position and velocity were generated as in equations (1) and (2), where xiThe horizontal and vertical coordinates, namely the position, of the ith particle after initialization; r is1、r2And r3Is [0,1]]The random number of (2).
vi＝r3 (2)
(2) Calculating the fitness of each particle and determining pbestiAnd gbest. The fitness is an index for evaluating the quality of the particles, and the accuracy capable of visually reflecting the classification condition is selected by the method. The evaluation method is to verify the training set by using the parameters found by the 50 particles, and the higher the average accuracy, the better the particles are. The current position of the i particle is the individual optimal position pbestiI.e. the local optimal solution; highest fitness pbestiAnd assigning to the gbest, wherein the gbest is the initial global optimal solution. Specifically, as an implementation manner of this embodiment, the method for calculating the particle fitness includes: dividing a training set into n parts, wherein n is an integer greater than or equal to 2; taking n-1 parts of training classification models and 1 part of verification models in turn, and calculating n times of tests to obtain average accuracy, wherein preset parameters in the classification models are provided by the particles, the average accuracy is the fitness of the particles, and the higher the fitness, the more excellent the particles are. Preferably, n may be 5, i.e. 5 fold cross validation: and (3) randomly equally dividing the training set into 5 parts, taking 4 parts of training models and 1 part of verification models in turn, and finally taking the average accuracy of 5 tests.
(3) The velocity and position of each particle is updated. And (3) updating the speed and the position of the particle according to the following formulas (3) and (4) in each iteration.
vi n＝ωvi n-1+c1r4(pbesti-xi n-1)+c2r5(gbest-xi n-1) (3)
xi n＝xi n-1+vi n (4)
In the formula: v. ofi nAnd xi nRespectively the speed and position of the ith particle at the nth iteration; v. ofi n-1And xi n-1Respectively the speed and the position of the ith particle in the (n-1) th iteration; r is4And r5Is the interval [0,1]The random number of (2).
(4) Calculating an adaptive value of each particle and comparing the adaptive value with its respective individual extreme value pbestiComparing, if excellent, updating pbestiOtherwise, the original value is reserved. Updating pbestiAnd comparing with the global optimal value gbest, if so, updating the gbest, otherwise, keeping the original value.
(5) And judging whether the termination condition is met. The set termination condition in this embodiment is iteration number 300, if the maximum iteration number is reached (or other termination conditions may be set according to actual conditions), the optimal parameter of the classification model finally found by the PSO is output, the classification model based on the optimal parameter is the optimal classification model, otherwise, the step (2) is returned to.
Step 102: and predicting whether the stainless steel cracks or not by adopting each optimal classification model according to the environmental parameters of the stainless steel to obtain the prediction result of each optimal classification model.
Step 103: determining whether the stainless steel cracks or not based on each prediction result.
The specific implementation of steps 102 and 103 may be as follows:
(1) the method comprises the steps of establishing a Hybrid model, wherein the Hybrid model can be composed of an RF classification model, an XGboost classification model and an SVM classification model which are optimized by PSO parameters, or can be composed of the RF classification model, the XGboost classification model, the SVM classification model, an NBM classification model and an ELM classification model which are optimized by the PSO parameters, and in the embodiment, the Hybrid model is established and composed of the NBM classification model and the ELM classification model which are not optimized by the PSO parameters, the RF classification model, the XGboost classification model and the SVM classification model which are optimized by the PSO parameters.
(2) And determining a prediction result of the Hybrid model, wherein the prediction result is given by the NBM classification model, the ELM classification model, the RF classification model optimized by the PSO parameters, the XGboost classification model and the SVM classification model in a hard voting mode, and referring to fig. 2.
In this embodiment, the environmental parameters may include, but are not limited to, temperature, chlorine content, and oxygen content.
In this embodiment, before model training is performed by using training samples, preprocessing needs to be performed on the samples, and the preprocessing may include the following two aspects:
(1) and (3) logarithmic transformation: according to the basic principle of electrochemical corrosion, the influence of the ion level on the corrosion behavior is generally exponential, so that the contents of chlorine and oxygen are logarithmically transformed, as shown in formula (5).
X'＝ln(X) (5)
(2) Normalization treatment: all samples are then normalized to map the values between intervals [0,1], as shown in equation (6).
The effects of the present invention are verified below
The verification data is derived from 71 groups of experimental data of stress corrosion behavior of certain austenitic stainless steel in high-temperature water environment. The data set has four parameters: temperature (241-. Accuracy (Accuracy) and the Matthews correlation coefficient (Mcc) are chosen as indicators for evaluating the classification performance of the model. The accuracy can visually reflect the condition that the sample is classified correctly, the Mcc represents a correlation coefficient between an actual value and a predicted value in the classification, the value range of the correlation coefficient is [ -1,1], and the closer the numerical value is to 1, the better the performance of the model is represented.
To verify the feasibility and generalization ability of the PSO-Hybrid model, the code was looped 15 times, with 7: 3, randomly dividing the data set according to the proportion, and comparing the performance of the PSO-Hybrid model with that of the other models. As shown in fig. 3 and table 1, the performance of each model without PSO optimization is shown, and a smaller variance represents a more stable model.
TABLE 1 comparison of Hybrid model and its base model Performance (without PSO Algorithm)
As can be seen from table 1, the average accuracy and the average mahius correlation coefficient of the Hybrid model are 0.867 and 0.718, respectively, which are superior to those of the basic models constituting the Hybrid model, and the variance of the Hybrid model is the lowest, which is 0.090, which indicates that the Hybrid prediction model has good stability. Therefore, the Hybrid idea obtains good effect on solving the two-classification problem, and has certain feasibility.
FIG. 4, FIG. 5 and Table 2 show the performance comparison of the PSO-Hybrid model before and after PSO optimization and the relevant partial model under the same 15 cycles as described above.
TABLE 2 comparison of PSO-Hybrid model and partial model Performance
From table 2, the PSO algorithm improves the performance of the two basis models, SVM and RF, the average accuracy of the SVM is improved from 0.809 to 0.830, the average mausus correlation coefficient is increased by 0.053, and the model is more stable; the average accuracy of RF is improved from 0.782 to 0.821, the average mahius correlation coefficient is increased by 0.09, but the stability is deteriorated. The average accuracy of the XGBoost model decreased by 0.009 and the stability deteriorated, but the average mahius correlation coefficient increased by 0.003. The PSO-Hybrid is superior to a Hybrid model in performance, the average accuracy is as high as 0.888, the average Maries correlation coefficient is improved by 0.061, and the stability is better.
In conclusion, the PSO-Hybrid model provided by the invention is excellent in performance, high in accuracy and strong in generalization capability based on the Hybrid idea and by using the PSO optimization algorithm, provides a certain technical support for predicting the stress corrosion cracking sensitivity of the stainless steel, and has certain reliability, feasibility and scientificity.
Example 2
Referring to fig. 6, the present embodiment provides a stainless steel environmental cracking determination system based on Hybrid idea, which includes:
the optimized classification model determining module 601 is configured to optimize preset parameters of a plurality of classification models by using a particle swarm optimization algorithm to obtain optimal classification models based on the respective optimal preset parameters, wherein the fitness of particles is determined by using the accuracy of the classification models obtained by training the particles, samples used for training the classification models are environmental parameters, and a label is whether stainless steel cracks or not. The classification model in the optimized classification model determining module 601 includes at least one of an RF classification model, an XGBoost classification model, and an SVM classification model, and the preset parameters of the RF classification model and the XGBoost classification model both include the number of decision trees and the maximum depth of trees; the preset parameters of the SVM classification model comprise tolerance of errors and a kernel function self-contained parameter sigma.
And the classification model prediction module 602 is configured to predict whether stainless steel cracks or not according to the environmental parameters of the stainless steel by using each optimal classification model, so as to obtain a prediction result of each optimal classification model. The classification model prediction module 602 is further configured to predict whether the stainless steel cracks according to environmental parameters of the stainless steel by using an NBM classification model and/or an ELM classification model, respectively.
And a cracking condition determining module 603, configured to determine whether the stainless steel cracks based on each prediction result.
The embodiments in the present description are described in a progressive manner, each embodiment focuses on differences from other embodiments, and the same and similar parts among the embodiments are referred to each other. For the system disclosed by the embodiment, the description is relatively simple because the system corresponds to the method disclosed by the embodiment, and the relevant points can be referred to the method part for description.
The principles and embodiments of the present invention have been described herein using specific examples, which are provided only to help understand the method and the core concept of the present invention; meanwhile, for a person skilled in the art, according to the idea of the present invention, the specific embodiments and the application range may be changed. In view of the above, the present disclosure should not be construed as limiting the invention.
Claims (10)
1. A stainless steel environment cracking distinguishing method based on a Hybrid idea is characterized by comprising the following steps:
optimizing preset parameters of a plurality of types of classification models by adopting a particle swarm optimization algorithm to obtain optimal classification models based on the respective optimal preset parameters, wherein the fitness of particles is determined by adopting the accuracy of the classification models obtained by training the particles, samples used for training the classification models are environmental parameters, and a label is whether stainless steel cracks or not;
predicting whether stainless steel cracks or not by adopting each optimal classification model according to the environmental parameters of the stainless steel to obtain the prediction result of each optimal classification model;
determining whether the stainless steel cracks or not based on each prediction result.
2. The Hybrid idea-based stainless steel environment crack discrimination method according to claim 1, wherein the classification model comprises at least one of an RF classification model, an XGBoost classification model, and an SVM classification model.
3. The Hybrid idea-based stainless steel environment crack judging method of claim 2, wherein the preset parameters of the RF classification model and the XGBoost classification model each include the number of decision trees and the maximum depth of trees; the preset parameters of the SVM classification model comprise tolerance of errors and a kernel function self-contained parameter sigma.
4. The method for discriminating environmental cracking of stainless steel according to claim 2, wherein before determining whether cracking of the stainless steel occurs based on each prediction result, the method further comprises:
and predicting whether the stainless steel cracks or not according to the environmental parameters of the stainless steel by adopting an NBM classification model and/or an ELM classification model respectively to obtain the prediction result of each classification model.
5. The method for discriminating stainless steel environmental cracking based on the Hybrid idea of any one of claims 1 to 4, wherein the method for calculating the particle fitness comprises the following steps:
dividing a training set into n parts, wherein n is an integer greater than or equal to 2;
taking n-1 parts of training classification models and 1 part of verification models in turn, and calculating n times of tests to obtain average accuracy, wherein preset parameters in the classification models are provided by the particles, the average accuracy is the fitness of the particles, and the higher the fitness, the more excellent the particles are.
6. The method for discriminating environmental cracking of stainless steel according to claim 1 or 3 based on the Hybrid idea, wherein the determining whether the stainless steel cracks based on the prediction results specifically comprises:
and giving a final prediction result by hard voting according to each prediction result.
7. The method for discriminating stainless steel environmental cracking based on the Hybrid idea of claim 1, wherein the environmental parameters include temperature, chlorine content and oxygen content.
8. The method for discriminating stainless steel environmental cracking based on Hybrid idea of claim 7, further comprising, before training the classification model:
carrying out logarithmic transformation on the chlorine content and the oxygen content in the sample data;
and respectively carrying out normalization treatment on the temperature in the sample data and the chlorine content and the oxygen content after logarithmic transformation.
9. A stainless steel environment crack discrimination system based on Hybrid idea is characterized by comprising:
the optimized classification model determining module is used for optimizing preset parameters of the classification models by adopting a particle swarm algorithm to obtain the optimal classification models based on the respective optimal preset parameters, wherein the fitness of particles is determined by adopting the accuracy of the classification models obtained by training the particles, samples used for training the classification models are environmental parameters, and labels are whether stainless steel cracks or not;
the classification model prediction module is used for predicting whether the stainless steel cracks or not according to the environment parameters of the stainless steel by adopting each optimal classification model to obtain the prediction result of each optimal classification model;
and the cracking condition determining module is used for determining whether the stainless steel cracks or not based on each prediction result.
10. The Hybrid concept-based stainless steel environmental cracking determination system of claim 9,
the classification model in the optimization classification model determination module comprises at least one of an RF classification model, an XGboost classification model and an SVM classification model; the preset parameters of the RF classification model and the XGboost classification model comprise the number of decision trees and the maximum depth of the trees; the preset parameters of the SVM classification model comprise tolerance of errors and a kernel function self-contained parameter sigma;
the classification model prediction module is further used for predicting whether the stainless steel cracks or not according to the environmental parameters of the stainless steel by adopting an NBM classification model and/or an ELM classification model respectively to obtain the prediction results of the classification models.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202110258345.7A CN113052207A (en) | 2021-03-10 | 2021-03-10 | Stainless steel environment cracking judgment method and system based on Hybrid idea |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202110258345.7A CN113052207A (en) | 2021-03-10 | 2021-03-10 | Stainless steel environment cracking judgment method and system based on Hybrid idea |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN113052207A true CN113052207A (en) | 2021-06-29 |
Family
ID=76511125
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202110258345.7A Pending CN113052207A (en) | 2021-03-10 | 2021-03-10 | Stainless steel environment cracking judgment method and system based on Hybrid idea |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN113052207A (en) |
Citations (9)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN104634879A (en) * | 2015-02-04 | 2015-05-20 | 北京科技大学 | Metallic material fatigue loading testing and fatigue damage nondestructive testing analytical method |
| CN108637020A (en) * | 2018-05-09 | 2018-10-12 | 北京科技大学 | A kind of TSP question PSO-BP neural networks strip profile prediction technique |
| EP3406376A1 (en) * | 2017-05-25 | 2018-11-28 | Camaga Srl | Method for optimization of machining productivity |
| WO2019055538A1 (en) * | 2017-09-12 | 2019-03-21 | Arconic Inc. | Systems and methods for additive manufacture |
| CN110044554A (en) * | 2019-04-16 | 2019-07-23 | 重庆大学 | A kind of online test method of the metal pressure container leakage based on acoustic emission signal |
| CN111340754A (en) * | 2020-01-18 | 2020-06-26 | 中国人民解放军国防科技大学 | Method for detecting and classifying surface defects based on aircraft skin |
| CN111444629A (en) * | 2020-04-15 | 2020-07-24 | 中国二冶集团有限公司 | Reinforcing steel bar corrosion parameter prediction method based on support vector machine |
| CN111948977A (en) * | 2020-08-21 | 2020-11-17 | 湖北工业大学 | Multi-objective optimization method and system for stainless steel processing |
| CN112257942A (en) * | 2020-10-29 | 2021-01-22 | 中国特种设备检测研究院 | Stress corrosion cracking prediction method and system |
-
2021
- 2021-03-10 CN CN202110258345.7A patent/CN113052207A/en active Pending
Patent Citations (9)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN104634879A (en) * | 2015-02-04 | 2015-05-20 | 北京科技大学 | Metallic material fatigue loading testing and fatigue damage nondestructive testing analytical method |
| EP3406376A1 (en) * | 2017-05-25 | 2018-11-28 | Camaga Srl | Method for optimization of machining productivity |
| WO2019055538A1 (en) * | 2017-09-12 | 2019-03-21 | Arconic Inc. | Systems and methods for additive manufacture |
| CN108637020A (en) * | 2018-05-09 | 2018-10-12 | 北京科技大学 | A kind of TSP question PSO-BP neural networks strip profile prediction technique |
| CN110044554A (en) * | 2019-04-16 | 2019-07-23 | 重庆大学 | A kind of online test method of the metal pressure container leakage based on acoustic emission signal |
| CN111340754A (en) * | 2020-01-18 | 2020-06-26 | 中国人民解放军国防科技大学 | Method for detecting and classifying surface defects based on aircraft skin |
| CN111444629A (en) * | 2020-04-15 | 2020-07-24 | 中国二冶集团有限公司 | Reinforcing steel bar corrosion parameter prediction method based on support vector machine |
| CN111948977A (en) * | 2020-08-21 | 2020-11-17 | 湖北工业大学 | Multi-objective optimization method and system for stainless steel processing |
| CN112257942A (en) * | 2020-10-29 | 2021-01-22 | 中国特种设备检测研究院 | Stress corrosion cracking prediction method and system |
Non-Patent Citations (2)
| Title |
|---|
| 姚舜才，孙传猛编著: "《机器学习基础教程》", 31 March 2020, 西安电子科学技术大学出版社, pages: 201 - 206 * |
| 陈琛;潘文超;: "岸桥金属材料与焊接结构寿命预测", 机械科学与技术, no. 10 * |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN109922032B (en) | Method, device, equipment and storage medium for determining risk of logging in account | |
| CN112460495B (en) | Monitoring point layout method and system for leakage monitoring positioning | |
| CN102185735B (en) | Network security situation prediction method | |
| CN109523021B (en) | Dynamic network structure prediction method based on long-time and short-time memory network | |
| CN108831121B (en) | Early warning method and device for mine safety production | |
| CN107231382A (en) | A kind of Cyberthreat method for situation assessment and equipment | |
| Jeon et al. | Development of surrogate model using CFD and deep neural networks to optimize gas detector layout | |
| CN109598052B (en) | Intelligent ammeter life cycle prediction method and device based on correlation coefficient analysis | |
| CN103955714A (en) | Navy detection model construction method and system and navy detection method | |
| CN114124460B (en) | Industrial control system intrusion detection method and device, computer equipment and storage medium | |
| CN115049019B (en) | Method and device for evaluating arsenic adsorption performance of metal organic framework and related equipment | |
| CN111428419A (en) | Suspended sediment concentration prediction method and device, computer equipment and storage medium | |
| CN112257942A (en) | Stress corrosion cracking prediction method and system | |
| CN111931047A (en) | Artificial intelligence-based black product account detection method and related device | |
| CN114266289A (en) | Complex equipment health state assessment method | |
| CN116341901A (en) | Integrated evaluation method for landslide surface domain-monomer hazard early warning | |
| CN115688581A (en) | Oil gas gathering and transportation station equipment parameter early warning method, system, electronic equipment and medium | |
| CN117520995B (en) | Abnormal user detection method and system in network information platform | |
| CN113988676A (en) | Safety management method and system for water treatment equipment | |
| CN113052207A (en) | Stainless steel environment cracking judgment method and system based on Hybrid idea | |
| CN114092472B (en) | Method, device and medium for detecting uncertain samples in defect detection | |
| CN111340975A (en) | Abnormal data feature extraction method, device, equipment and storage medium | |
| Zhai et al. | Parameter estimation method of mixture distribution for construction machinery | |
| CN114329966A (en) | Method and system for evaluating health degree of remote control system of natural gas pipeline | |
| CN114417977A (en) | Fault diagnosis method and system based on transformer data sample expansion |
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 |