CN116484675A - Crack propagation life prediction method and system for ship engine blade - Google Patents
Crack propagation life prediction method and system for ship engine blade Download PDFInfo
- Publication number
- CN116484675A CN116484675A CN202310376423.2A CN202310376423A CN116484675A CN 116484675 A CN116484675 A CN 116484675A CN 202310376423 A CN202310376423 A CN 202310376423A CN 116484675 A CN116484675 A CN 116484675A
- Authority
- CN
- China
- Prior art keywords
- neural network
- engine blade
- crack
- sample data
- stress intensity
- 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
- 238000000034 method Methods 0.000 title claims abstract description 52
- 230000004044 response Effects 0.000 claims abstract description 53
- 238000009826 distribution Methods 0.000 claims abstract description 44
- 238000004458 analytical method Methods 0.000 claims abstract description 37
- 238000012549 training Methods 0.000 claims abstract description 32
- 238000005070 sampling Methods 0.000 claims abstract description 23
- 238000004364 calculation method Methods 0.000 claims abstract description 14
- 238000013528 artificial neural network Methods 0.000 claims description 48
- 210000002569 neuron Anatomy 0.000 claims description 37
- 230000006870 function Effects 0.000 claims description 31
- 238000003062 neural network model Methods 0.000 claims description 18
- 238000012546 transfer Methods 0.000 claims description 12
- 230000009977 dual effect Effects 0.000 claims description 10
- 239000011159 matrix material Substances 0.000 claims description 7
- 230000000007 visual effect Effects 0.000 claims description 7
- 230000008569 process Effects 0.000 description 11
- 238000004088 simulation Methods 0.000 description 10
- 238000012423 maintenance Methods 0.000 description 9
- 238000010586 diagram Methods 0.000 description 6
- 238000013178 mathematical model Methods 0.000 description 5
- 238000011160 research Methods 0.000 description 5
- 238000004590 computer program Methods 0.000 description 4
- 239000000463 material Substances 0.000 description 4
- 230000000977 initiatory effect Effects 0.000 description 3
- 239000000956 alloy Substances 0.000 description 2
- 238000004422 calculation algorithm Methods 0.000 description 2
- 239000000306 component Substances 0.000 description 2
- 238000012937 correction Methods 0.000 description 2
- 125000004122 cyclic group Chemical group 0.000 description 2
- 238000011161 development Methods 0.000 description 2
- 238000000605 extraction Methods 0.000 description 2
- 238000013473 artificial intelligence Methods 0.000 description 1
- 230000009286 beneficial effect Effects 0.000 description 1
- 238000006243 chemical reaction Methods 0.000 description 1
- 238000005094 computer simulation Methods 0.000 description 1
- 238000010276 construction Methods 0.000 description 1
- 238000013527 convolutional neural network Methods 0.000 description 1
- 239000008358 core component Substances 0.000 description 1
- 238000013135 deep learning Methods 0.000 description 1
- 230000007547 defect Effects 0.000 description 1
- 238000005516 engineering process Methods 0.000 description 1
- 230000007613 environmental effect Effects 0.000 description 1
- 238000011156 evaluation Methods 0.000 description 1
- 239000000284 extract Substances 0.000 description 1
- 238000010801 machine learning Methods 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
- 238000003058 natural language processing Methods 0.000 description 1
- 238000005457 optimization Methods 0.000 description 1
- 239000002245 particle Substances 0.000 description 1
- 238000012545 processing Methods 0.000 description 1
- 230000008439 repair process Effects 0.000 description 1
- 238000003860 storage Methods 0.000 description 1
- 238000012916 structural analysis Methods 0.000 description 1
- 238000012706 support-vector machine Methods 0.000 description 1
- 238000003466 welding Methods 0.000 description 1
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
- G06F30/23—Design optimisation, verification or simulation using finite element methods [FEM] or finite difference methods [FDM]
-
- 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
- G06N3/045—Combinations of networks
-
- 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
- G06F2119/00—Details relating to the type or aim of the analysis or the optimisation
- G06F2119/02—Reliability analysis or reliability optimisation; Failure analysis, e.g. worst case scenario performance, failure mode and effects analysis [FMEA]
-
- Y—GENERAL TAGGING OF NEW TECHNOLOGICAL DEVELOPMENTS; GENERAL TAGGING OF CROSS-SECTIONAL TECHNOLOGIES SPANNING OVER SEVERAL SECTIONS OF THE IPC; TECHNICAL SUBJECTS COVERED BY FORMER USPC CROSS-REFERENCE ART COLLECTIONS [XRACs] AND DIGESTS
- Y02—TECHNOLOGIES OR APPLICATIONS FOR MITIGATION OR ADAPTATION AGAINST CLIMATE CHANGE
- Y02T—CLIMATE CHANGE MITIGATION TECHNOLOGIES RELATED TO TRANSPORTATION
- Y02T90/00—Enabling technologies or technologies with a potential or indirect contribution to GHG emissions mitigation
Landscapes
- Engineering & Computer Science (AREA)
- Physics & Mathematics (AREA)
- Theoretical Computer Science (AREA)
- Evolutionary Computation (AREA)
- General Engineering & Computer Science (AREA)
- General Physics & Mathematics (AREA)
- Artificial Intelligence (AREA)
- Software Systems (AREA)
- Biophysics (AREA)
- Computer Hardware Design (AREA)
- Mathematical Physics (AREA)
- Geometry (AREA)
- Health & Medical Sciences (AREA)
- Life Sciences & Earth Sciences (AREA)
- Biomedical Technology (AREA)
- Computing Systems (AREA)
- Computational Linguistics (AREA)
- Data Mining & Analysis (AREA)
- General Health & Medical Sciences (AREA)
- Molecular Biology (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Medical Informatics (AREA)
- Testing Of Devices, Machine Parts, Or Other Structures Thereof (AREA)
Abstract
The invention provides a crack propagation life prediction method and a crack propagation life prediction system for a ship engine blade, wherein the method comprises the following steps: taking parameters affecting crack extension life of the engine blade as input random variables, and carrying out finite element analysis on the engine blade to obtain stress distribution of the engine blade; presetting micro cracks at the maximum stress position of an engine blade based on stress distribution, taking stress intensity factors at different stages of crack expansion as output responses, and sampling random variables to obtain first sample data; training a stress intensity factor prediction model based on the first sample data and the corresponding actual output response; sampling the random variables to obtain second sample data, inputting the second sample data into a stress intensity factor prediction model, outputting a prediction result, inputting the prediction result into a reliability calculation model, outputting crack extension life distribution and reliability of engine blades, and improving accuracy and efficiency of crack extension life prediction of the ship engine.
Description
Technical Field
The invention belongs to the technical field of reliability evaluation, and particularly relates to a crack propagation life prediction method and system for a ship engine blade.
Background
Due to the particularity of the ship task, the ship needs to be far away from the shore when the ship performs the task, the sailing time is long, various task states are alternately performed, and the supply and guarantee are difficult. The ship engine blade is a core component of a ship, and if the ship engine blade is cracked and destroyed once, serious accident loss can be caused, so that the method has important significance for predicting the crack propagation life of the ship engine blade.
The study of crack growth life has been an important research direction in the fields of material science and engineering. At present, research on crack growth life is mainly focused on the aspects of influence of material properties on crack growth life, research on crack growth mechanism, influence of environmental factors on crack growth life and the like, the prediction of the crack growth life of the blade still depends on judgment through artificial experience after the occurrence of the crack of the blade, the subjectivity is strong, and the prediction result is inaccurate.
Disclosure of Invention
Aiming at the defects of the prior art, the invention aims to provide a crack propagation life prediction method and system for a ship engine blade, and aims to solve the problem of low prediction accuracy of the existing crack propagation life prediction method.
In order to achieve the above object, in a first aspect, the present invention provides a crack propagation life prediction method for a ship engine blade, including:
s101, taking parameters affecting crack extension life of an engine blade as input random variables, and carrying out finite element analysis on the engine blade to obtain stress distribution of the engine blade;
s102, presetting micro cracks at the maximum stress position of an engine blade based on stress distribution, taking stress intensity factors at different stages of crack expansion as output responses, and sampling random variables to obtain first sample data;
s103, training a stress intensity factor prediction model based on the first sample data and the corresponding actual output response to obtain a trained stress intensity factor prediction model;
s104, sampling the random variables to obtain second sample data, inputting the second sample data into the trained stress intensity factor prediction model, outputting a prediction result, inputting the prediction result into the reliability calculation model, and outputting crack extension life distribution and reliability of the engine blade.
In an alternative example, the output response of the stress intensity factor prediction model is obtained based on a dual response surface function of different stages of crack propagation; the dual response surface function specifically comprises:
wherein W is jk Is the connection weight between the input layer neuron j and the hidden layer neuron k; w (W) ij Is the connection weight between the hidden layer neuron i and the output layer neuron j; b j Is the hidden layer jth threshold; b k Is the output layer kth threshold; f (f) 2 Is an implicit layer transfer function; f (f) 1 Is the output layer transfer function; m is an inputThe number of layer neurons; n is the number of hidden layer neurons; y is the output response; the superscript (1) and (2) represent the variables in the 1 st and 2 nd response surface functions, respectively.
In an alternative example, the stress intensity factor prediction model is constructed based on an RBM-BP neural network model.
In an alternative example, step S103 specifically includes:
optimizing parameters of the RBM neural network based on the first sample data until a training termination condition is met, and obtaining an optimized RBM neural network; the parameters comprise a connection weight matrix, a visual layer offset variable and an implicit layer offset variable;
based on the optimized RBM neural network, obtaining sample characteristics corresponding to the first sample data;
training the BP neural network based on the sample characteristics and the corresponding actual output response to obtain a trained BP neural network;
and obtaining a trained stress intensity factor prediction model based on the optimized RBM neural network and the trained BP neural network.
In a second aspect, the present invention provides a crack propagation life prediction system for a ship engine blade, comprising:
the finite element analysis module is used for carrying out finite element analysis on the engine blade by taking parameters affecting the crack extension life of the engine blade as input random variables to obtain the stress distribution of the engine blade;
the sampling module is used for presetting micro cracks at the maximum stress position of the engine blade based on stress distribution, taking stress intensity factors at different stages of crack expansion as output responses, and sampling random variables to obtain first sample data;
the model training module is used for training the stress intensity factor prediction model based on the first sample data and the corresponding actual output response to obtain a trained stress intensity factor prediction model;
the life prediction module is used for sampling the random variable to obtain second sample data, inputting the second sample data into the trained stress intensity factor prediction model, outputting a prediction result, inputting the prediction result into the reliability calculation model, and outputting crack extension life distribution and reliability of the engine blade.
In an alternative example, the output response of the stress intensity factor prediction model in the model training module is obtained based on a dual response surface function of different stages of crack propagation; the dual response surface function specifically comprises:
wherein W is jk Is the connection weight between the input layer neuron j and the hidden layer neuron k; w (W) ij Is the connection weight between the hidden layer neuron i and the output layer neuron j; b j Is the hidden layer jth threshold; b k Is the output layer kth threshold; f (f) 2 Is an implicit layer transfer function; f (f) 1 Is the output layer transfer function; m is the number of neurons in the input layer; n is the number of hidden layer neurons; y is the output response; the superscript (1) and (2) represent the variables in the 1 st and 2 nd response surface functions, respectively.
In an alternative example, the stress intensity factor prediction model in the model training module is constructed based on an RBM-BP neural network model.
In an alternative example, the model training module is specifically configured to:
optimizing parameters of the RBM neural network based on the first sample data until a training termination condition is met, and obtaining an optimized RBM neural network; the parameters comprise a connection weight matrix, a visual layer offset variable and an implicit layer offset variable;
based on the optimized RBM neural network, obtaining sample characteristics corresponding to the first sample data;
training the BP neural network based on the sample characteristics and the corresponding actual output response to obtain a trained BP neural network;
and obtaining a trained stress intensity factor prediction model based on the optimized RBM neural network and the trained BP neural network.
In general, the above technical solutions conceived by the present invention have the following beneficial effects compared with the prior art:
the invention provides a method and a system for predicting crack propagation life of a ship engine blade, which are characterized in that cracks are preset on a model through finite element analysis of a ship engine blade geometric model through obtained stress distribution, a crack propagation process is simulated through building a neural network model to obtain a predicted result of stress intensity factors obtained through mass simulation, and the predicted result is used for reliability analysis to calculate life distribution and reliability of the predicted result, so that accuracy and efficiency of ship engine crack propagation life prediction and reliability assessment are improved, labor and time cost are reduced, and the method and the system can be used for guiding a maintenance process decision based on the predicted result.
Drawings
FIG. 1 is a schematic flow chart of a method for predicting crack growth life of a ship engine blade according to an embodiment of the present invention;
FIG. 2 is a second flow chart of a method for predicting crack growth life of a ship engine blade according to an embodiment of the present invention;
FIG. 3 is a general flow diagram of a method for predicting crack growth life of a ship engine blade provided by an embodiment of the present invention;
FIG. 4 is a schematic general flow chart of a method for obtaining a stress intensity factor prediction model according to an embodiment of the present invention;
FIG. 5 is one of the architecture diagrams of a crack propagation life prediction system for a ship engine blade provided by an embodiment of the present invention;
FIG. 6 is a second architecture diagram of a crack propagation life prediction system for a ship engine blade according to an embodiment of the present invention.
Detailed Description
The present invention will be described in further detail with reference to the drawings and examples, in order to make the objects, technical solutions and advantages of the present invention more apparent. It should be understood that the specific embodiments described herein are for purposes of illustration only and are not intended to limit the scope of the invention.
Neural networks are a computational model inspired by an artificial neuron system. The method can be regarded as a multi-level nonlinear function approximator, realizes the identification and prediction of input data by learning the mode and rule in the data set, and has a plurality of models and algorithms based on the neural network, such as convolutional neural network, cyclic neural network, deep learning and the like, along with the development of the neural network; nowadays, neural networks have become one of important technologies in the field of artificial intelligence, and are widely used in various fields such as machine learning, computer vision, natural language processing, intelligent control, and the like.
Based on this, the present invention provides a method for predicting the crack propagation life of a ship engine blade, and fig. 1 is one of flow diagrams of a method for predicting the crack propagation life of a ship engine blade according to an embodiment of the present invention, as shown in fig. 1, where the method includes:
s101, taking parameters affecting crack extension life of an engine blade as input random variables, and carrying out finite element analysis on the engine blade to obtain stress distribution of the engine blade;
step S102, presetting micro cracks at the maximum stress position of an engine blade based on stress distribution, taking stress intensity factors at different stages of crack expansion as output responses, and sampling random variables to obtain first sample data;
step S103, training a stress intensity factor prediction model based on the first sample data and the corresponding actual output response to obtain a trained stress intensity factor prediction model;
step S104, sampling the random variables to obtain second sample data, inputting the second sample data into the trained stress intensity factor prediction model, outputting a prediction result, inputting the prediction result into the reliability calculation model, and outputting crack extension life distribution and reliability of the engine blade.
Here, the parameters affecting the crack propagation life of the engine blade may include deterministic influencing factors and normal distributed random input variables, which are specifically: deterministic influencing factors: life safety threshold, inner wall convection coefficient, outer wall convection coefficient, density, stretching limit, hydrodynamic parameters, number of blades, nominal stress correction coefficient and alloy material basic mechanical property; normal distribution random input variables (variable, mean, standard deviation): rotational speed, modulus of elasticity, fatigue strength coefficient, fatigue ductility coefficient, fatigue strength index, fatigue ductility index. The stress intensity factor prediction model is a neural network model.
It should be noted that, in the existing method, crack propagation life prediction is performed through artificial experience after the occurrence of cracks, but in the method provided by the invention, finite element analysis is performed on the blade, a crack is preset at the place with the maximum stress output by the finite element analysis result in a simulation system, simulation is performed through the crack, and small-batch sampling is adopted to obtain first sample data to perform training fitting on the neural network model, so that the predicted result output by the neural network model and the actual stress intensity factor approximation of crack propagation at different stages are realized, the finally trained neural network model can be subjected to mass simulation, one-time sampling simulation is not needed, and the predicted result obtained by mass simulation can be used for reliability analysis, thereby obtaining crack propagation life distribution and reliability of the engine blade.
Crack propagation life corresponds to fatigue life, which refers to the life time of a crack from its occurrence to its development to the breaking of a blade, that is, the cycle time from the occurrence of a crack to the failure of the final blade. By predicting the crack propagation life of the ship engine blade, reasonable maintenance decisions can be made according to the life, the maintenance efficiency of the crack is improved, and the serious consequences caused by long-time non-treatment of the crack are avoided.
According to the method provided by the embodiment of the invention, the crack is preset on the model through finite element analysis of the ship engine blade geometric model through the obtained stress distribution, the crack expansion process is simulated through the establishment of the neural network model, the prediction result of the stress intensity factor obtained through mass simulation is obtained, and the prediction result is used for reliability analysis to calculate the service life distribution and reliability of the ship engine blade geometric model, so that the accuracy and efficiency of the ship engine crack expansion service life prediction and reliability assessment are improved, the labor and time cost is reduced, and the maintenance process decision can be guided based on the prediction result.
In an alternative example, the output response of the stress intensity factor prediction model is obtained based on a dual response surface function of different stages of crack propagation; the dual response surface function is specifically:
wherein W is jk Is the connection weight between the input layer neuron j and the hidden layer neuron k; w (W) ij Is the connection weight between the hidden layer neuron i and the output layer neuron j; b j Is the hidden layer jth threshold; b k Is the output layer kth threshold; f (f) 2 Is an implicit layer transfer function; f (f) 1 Is the output layer transfer function; m is the number of neurons in the input layer; n is the number of hidden layer neurons; y is the output response; the superscript (1) and (2) represent the variables in the 1 st and 2 nd response surface functions, respectively.
It will be appreciated that stress concentration at the crack tip can lead to crack propagation, which can be divided into three stages: crack initiation, crack propagation and failure. Since the crack is unstable in the third stage, namely the breaking stage, the stress intensity factor of the stage is not needed, and therefore, the 1 st response surface function and the 2 nd response surface function are respectively constructed aiming at the crack initiation stage, namely the short crack stage and the crack expansion stage, namely the long crack stage.
In an alternative example, the stress intensity factor prediction model is built based on an RBM-BP neural network model.
In an alternative example, step S103 specifically includes:
optimizing parameters of the RBM neural network based on the first sample data until a training termination condition is met, and obtaining an optimized RBM neural network; the parameters comprise a connection weight matrix, a visual layer offset variable and an implicit layer offset variable;
based on the optimized RBM neural network, obtaining sample characteristics corresponding to the first sample data;
training the BP neural network based on the sample characteristics and the corresponding actual output response to obtain a trained BP neural network;
and obtaining a trained stress intensity factor prediction model based on the optimized RBM neural network and the trained BP neural network.
Here, whether the training termination condition is satisfied may be determined by whether the reconstruction error satisfies the requirement, or may be determined by whether the maximum training number is reached, which is not particularly limited in the embodiment of the present invention. It is understood that in the stress intensity factor prediction model, the output of the RBM neural network is characterized by the input of the BP neural network.
According to the method provided by the embodiment of the invention, the stress intensity factor prediction model is constructed based on the RBM-BP neural network model, and the RBM neural network and the BP neural network are combined and trained based on the first sample data and the corresponding actual output response, so that the prediction result of the stress intensity factor prediction model is more accurate and more practical.
In an alternative example, fig. 2 is a second flowchart of a method for predicting the crack growth life of a ship engine blade according to an embodiment of the present invention, and fig. 3 is a general flowchart of a method for predicting the crack growth life of a ship engine blade according to an embodiment of the present invention, as shown in fig. 2 and 3, where the method includes the following steps:
s1, acquiring relevant data of an engine blade, determining relevant influence factors influencing crack propagation life, establishing a blade finite element model, setting relevant constraint boundary conditions for finite element analysis, and extracting analysis results to obtain blade stress distribution and strain distribution;
the method comprises the steps of acquiring relevant data of an engine blade, including working states, material properties, load environments, component performances and the like, wherein the relevant data comprise deterministic influence factors and normal distribution random input variables, and the method comprises the following steps of:
deterministic influencing factors: life safety threshold, inner wall convection coefficient, outer wall convection coefficient, density, stretching limit, hydrodynamic parameters, number of blades, nominal stress correction coefficient and alloy material basic mechanical property;
normal distribution random input variables (variable, mean, standard deviation): rotational speed, modulus of elasticity, fatigue strength coefficient, fatigue ductility coefficient, fatigue strength index, fatigue ductility index.
The blade finite element analysis is established to output stress and strain profiles, which are specifically: importing a geometric model of an engine blade and meshing, regularly defining geometric variables and input random variables of the geometric model, taking relevant parameters affecting crack propagation of the blade as the input random variables, establishing a finite element model of the blade, and taking actual working conditions of the blade into consideration to carry out deterministic simulation analysis to obtain the maximum stress and maximum strain distribution rule of the blade; and carrying out subsequent research analysis on preset cracks at the maximum stress position.
Here, the relevant parameters include deterministic influencing factors and normally distributed random input variables. The value of the deterministic influence factor is fixed and can be directly reserved, and the value of the normal distribution random input variable is in the range of normal distribution, so that the normal distribution random input variable needs to be sampled.
S2, presetting cracks, determining output response, sampling random variables, and obtaining a small sample data set;
according to a deterministic analysis result, presetting micro cracks at the position with maximum stress, and taking stress intensity factors at different stages of crack expansion as output responses; further, the random variable may be sampled using Gibbs sampling techniques, as follows:
1) For each variable i, fixing the current values of other variables, and sampling according to the conditional probability distribution of the variable to obtain a new value of the variable i;
2) Repeating the steps until the new values of all variables have been sampled;
3) The above steps are repeated with the new value as the starting point for the next iteration until the desired number of samples is obtained.
S3, establishing an RBM-BP neural network model, and obtaining a crack propagation reliability mathematical model, namely a stress intensity factor prediction model, by training the neural network;
FIG. 4 is a schematic general flow chart of a method for obtaining a stress intensity factor prediction model according to an embodiment of the present invention; as shown in fig. 4, the building and training of the RBM-BP neural network model comprises the following specific steps:
s31, constructing a response surface function according to the BP neural network model:
establishing a double-response-surface mathematical model according to different response surfaces of crack propagation:
wherein W is jk Is the connection weight between the input layer neuron j and the hidden layer neuron k; w (W) ij Is the connection weight between the hidden layer neuron i and the output layer neuron j; b j Is the hidden layer jth threshold; b k Is the output layer kth threshold; f (f) 2 Is an implicit layer transfer function; f (f) 1 Is the output layer transfer function; m is the number of neurons in the input layer; n is the number of hidden layer neurons; (1) (2) respectively representing the 1 st weight and the 2 nd weight response surface models;
s32, initializing parameters of the RBM neural network;
s33, inputting sample data generated by sampling;
s34, calculating RBM network output, adjusting parameters (a connection weight matrix, a visual layer offset variable and an implicit layer offset variable), and calculating a reconstruction error;
s35, judging whether the reconstruction error meets the requirement or reaches the maximum training times, if so, turning to a step S36, otherwise, turning to a step S33;
s36, storing related parameters (a connection weight matrix, a visual layer offset variable and an implicit layer offset variable) and constructing a new feature database;
s37, performing BP neural network training to obtain a trained RBM-BP neural network model which can be used for reliability prediction, namely a stress intensity factor prediction model.
S4, sampling the random variables in a large scale, substituting the random variables into a stress intensity factor prediction model to predict, and carrying out reliability analysis by taking a prediction result into a reliability calculation model to output life distribution and reliability;
wherein the crack growth life is determined by:
stress concentration at the crack tip can lead to crack propagation; the crack propagation process can be divided into three phases: crack initiation, crack propagation and failure; whether or not the crack can propagate depends on the cyclic stress intensity factor Δk of the crack tip, and in addition, there is a stress intensity threshold Δk th When DeltaK < DeltaK th When the crack is rapidly expanded, but as the crack length increases, the expansion is decelerated and stopped to reach the critical crack size a 0 The formula is as follows:
wherein DeltaK is the stress intensity factor range of the long crack threshold and can be obtained according to the stress intensity factors of different stages of crack growth predicted by the RBM-BP neural network model, deltasigma e Is a fatigue limit; the stress threshold and the stress intensity factor threshold are expressed as follows:
wherein b is the sum of the crack half-field and the blocked slip width, a is the crack half-length,for frictional resistance +.>Is a sliding belt top stress intensity factor; generally, when a crack passes 10 7 When the expansion is not more than 0.1mm, the ΔK at this time is regarded as ΔK th The method comprises the steps of carrying out a first treatment on the surface of the The propagation rate formula for a long crack propagation but not including the small crack propagation phase is as follows:
wherein C and s are material parameters, R is stress ratio, K c For fracture toughness, N is the number of cycles of the work cycle, i.e., the crack growth life that needs to be determined, can be obtained by solving the expansion rate formula described above.
And S5, carrying out maintenance process decision according to the reliability analysis output result.
In an alternative example, fig. 5 is one of architecture diagrams of a crack propagation life prediction system for a ship engine blade according to an embodiment of the present invention, as shown in fig. 5, where the system includes: the input module is used for acquiring relevant parameters of the engine blade and a geometric model of a research object, and collecting technological parameters and data of a target to be optimized; acquiring relevant data of a field ship engine blade repairing process, such as processing power, pressure parameters, welding speed and the like, optimizing and finishing data such as strength, rigidity, vibration stability and the like of targets; the input module specifically comprises: a deterministic factor set, a life failure mode set, a normal distribution random input variable set, a geometric model management module, a database management module and a data query module;
the parameterized finite element analysis module realizes structural analysis parameterization in a command stream writing mode, and converts an analysis object geometric model into a grid file for finite element analysis; the parameterized finite element analysis module specifically comprises: the system comprises a parameterized network management module, a parameterized finite element management module, an environment generation module and a data acquisition module;
the structure analysis result extraction module is used for extracting the finite element analysis result and generating stress distribution and strain distribution of the blade; the structure analysis result extraction module specifically comprises: a situation display module and a stress and strain distribution output module;
the mathematical model management module performs random variable small batch sampling through preset cracks, takes stress intensity factors at different stages of crack propagation as output responses, establishes a neural network and trains to obtain a mathematical model;
the reliability calculation module samples the random variables in a large batch, substitutes the random variables into a mathematical model to calculate, and brings the calculation result into the reliability calculation model to carry out reliability analysis; the reliability calculation module specifically comprises: a reliability calculation method library, an agent module library and a simulation deduction module;
the output module extracts and analyzes the reliability calculation result and finally outputs life distribution and reliability; the output module specifically comprises: the system comprises a data statistics module, a data query module, a rule management module and a data conversion output module;
the maintenance process decision module carries out regression fitting and optimization on data by using a regression type support vector machine and a multi-target particle swarm algorithm to obtain a group of Pareto front solution sets, and then carries out decision analysis on the Pareto solution sets by combining a hierarchical analysis method and an improved TOPSIS method, so that relevant technicians can conveniently select the most suitable solution to guide on-site repair; the maintenance process decision module specifically comprises: the system comprises a maintenance rule management module, a guarantee scheme management module, a strength compiling construction module and a task scheduling module.
In an alternative example, fig. 6 is a second architecture diagram of a crack propagation life prediction system for a ship engine blade according to an embodiment of the present invention, as shown in fig. 6, where the system includes:
the finite element analysis module 610 is configured to perform finite element analysis on the engine blade with a parameter affecting a crack propagation life of the engine blade as an input random variable, so as to obtain stress distribution of the engine blade;
the sampling module 620 is configured to preset a micro crack at a stress maximum position of the engine blade based on stress distribution, take stress intensity factors at different stages of crack propagation as output responses, and sample random variables to obtain first sample data;
the model training module 630 is configured to train the stress intensity factor prediction model based on the first sample data and the corresponding actual output response, and obtain a trained stress intensity factor prediction model;
the life prediction module 640 is configured to sample the random variable to obtain second sample data, input the second sample data to the trained stress intensity factor prediction model, output a prediction result, input the prediction result to the reliability calculation model, and output crack growth life distribution and reliability of the engine blade.
According to the system provided by the embodiment of the invention, the crack is preset on the model through finite element analysis of the ship engine blade geometric model through the obtained stress distribution, the crack expansion process is simulated through the establishment of the neural network model, the prediction result of the stress intensity factor obtained through mass simulation is obtained, and the prediction result is used for reliability analysis to calculate the service life distribution and reliability of the prediction result, so that the accuracy and efficiency of the ship engine crack expansion service life prediction and reliability assessment are improved, the labor and time cost is reduced, and the maintenance process decision can be guided based on the prediction result.
It can be understood that the detailed functional implementation of each module may be referred to the description in the foregoing method embodiment, and will not be repeated herein.
In addition, the embodiment of the invention provides another crack propagation life prediction device of a ship engine blade, which comprises the following components: a memory and a processor;
the memory is used for storing a computer program;
the processor is configured to implement the method in the above-described embodiments when executing the computer program.
Furthermore, the present invention provides a computer-readable storage medium having stored thereon a computer program which, when executed by a processor, implements the method in the above embodiments.
Based on the method in the above embodiments, an embodiment of the present invention provides a computer program product, which when run on a processor causes the processor to perform the method in the above embodiments.
It will be readily appreciated by those skilled in the art that the foregoing description is merely a preferred embodiment of the invention and is not intended to limit the invention, but any modifications, equivalents, improvements or alternatives falling within the spirit and principles of the invention are intended to be included within the scope of the invention.
Claims (8)
1. The crack propagation life prediction method of the ship engine blade is characterized by comprising the following steps of:
s101, taking parameters affecting crack extension life of an engine blade as input random variables, and carrying out finite element analysis on the engine blade to obtain stress distribution of the engine blade;
s102, presetting micro cracks at the maximum stress position of an engine blade based on stress distribution, taking stress intensity factors at different stages of crack expansion as output responses, and sampling random variables to obtain first sample data;
s103, training a stress intensity factor prediction model based on the first sample data and the corresponding actual output response to obtain a trained stress intensity factor prediction model;
s104, sampling the random variables to obtain second sample data, inputting the second sample data into the trained stress intensity factor prediction model, outputting a prediction result, inputting the prediction result into the reliability calculation model, and outputting crack extension life distribution and reliability of the engine blade.
2. The method of claim 1, wherein the output response of the stress intensity factor prediction model is obtained based on a dual response surface function of different stages of crack propagation; the dual response surface function specifically comprises:
wherein W is jk Is the connection weight between the input layer neuron j and the hidden layer neuron k; w (W) ij Is the connection weight between the hidden layer neuron i and the output layer neuron j; b j Is the hidden layer jth threshold; b k Is the output layer kth threshold; f (f) 2 Is an implicit layer transfer function; f (f) 1 Is the output layer transfer function; m is the number of neurons in the input layer; n is the number of hidden layer neurons; y is the output response; the superscript (1) and (2) represent the variables in the 1 st and 2 nd response surface functions, respectively.
3. The method according to claim 1 or 2, wherein the stress intensity factor prediction model is constructed based on an RBM-BP neural network model.
4. A method according to claim 3, wherein step S103 specifically comprises:
optimizing parameters of the RBM neural network based on the first sample data until a training termination condition is met, and obtaining an optimized RBM neural network; the parameters comprise a connection weight matrix, a visual layer offset variable and an implicit layer offset variable;
based on the optimized RBM neural network, obtaining sample characteristics corresponding to the first sample data;
training the BP neural network based on the sample characteristics and the corresponding actual output response to obtain a trained BP neural network;
and obtaining a trained stress intensity factor prediction model based on the optimized RBM neural network and the trained BP neural network.
5. A crack growth life prediction system for a ship engine blade, comprising:
the finite element analysis module is used for carrying out finite element analysis on the engine blade by taking parameters affecting the crack extension life of the engine blade as input random variables to obtain the stress distribution of the engine blade;
the sampling module is used for presetting micro cracks at the maximum stress position of the engine blade based on stress distribution, taking stress intensity factors at different stages of crack expansion as output responses, and sampling random variables to obtain first sample data;
the model training module is used for training the stress intensity factor prediction model based on the first sample data and the corresponding actual output response to obtain a trained stress intensity factor prediction model;
the life prediction module is used for sampling the random variable to obtain second sample data, inputting the second sample data into the trained stress intensity factor prediction model, outputting a prediction result, inputting the prediction result into the reliability calculation model, and outputting crack extension life distribution and reliability of the engine blade.
6. The system of claim 5, wherein the output response of the stress intensity factor predictive model in the model training module is obtained based on a dual response surface function of different stages of crack propagation; the dual response surface function specifically comprises:
wherein W is jk Is the connection weight between the input layer neuron j and the hidden layer neuron k; w (W) ij Is the connection weight between the hidden layer neuron i and the output layer neuron j; b j Is the hidden layer jth threshold; b k Is the output layer kth threshold; f (f) 2 Is an implicit layer transfer function; f (f) 1 Is the output layer transfer function; m is the number of neurons in the input layer; n is the number of hidden layer neurons; y is the output response; the superscript (1) and (2) represent the variables in the 1 st and 2 nd response surface functions, respectively.
7. The system of claim 5 or 6, wherein the stress intensity factor predictive model in the model training module is constructed based on an RBM-BP neural network model.
8. The system of claim 7, wherein the model training module is specifically configured to:
optimizing parameters of the RBM neural network based on the first sample data until a training termination condition is met, and obtaining an optimized RBM neural network; the parameters comprise a connection weight matrix, a visual layer offset variable and an implicit layer offset variable;
based on the optimized RBM neural network, obtaining sample characteristics corresponding to the first sample data;
training the BP neural network based on the sample characteristics and the corresponding actual output response to obtain a trained BP neural network;
and obtaining a trained stress intensity factor prediction model based on the optimized RBM neural network and the trained BP neural network.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN202310376423.2A CN116484675A (en) | 2023-04-10 | 2023-04-10 | Crack propagation life prediction method and system for ship engine blade |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN202310376423.2A CN116484675A (en) | 2023-04-10 | 2023-04-10 | Crack propagation life prediction method and system for ship engine blade |
Publications (1)
Publication Number | Publication Date |
---|---|
CN116484675A true CN116484675A (en) | 2023-07-25 |
Family
ID=87216949
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN202310376423.2A Pending CN116484675A (en) | 2023-04-10 | 2023-04-10 | Crack propagation life prediction method and system for ship engine blade |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN116484675A (en) |
Cited By (1)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN117332688A (en) * | 2023-09-27 | 2024-01-02 | 中国石油大学(华东) | Method and system for predicting fatigue crack growth of X80 pipeline steel under random load effect |
-
2023
- 2023-04-10 CN CN202310376423.2A patent/CN116484675A/en active Pending
Cited By (2)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN117332688A (en) * | 2023-09-27 | 2024-01-02 | 中国石油大学(华东) | Method and system for predicting fatigue crack growth of X80 pipeline steel under random load effect |
CN117332688B (en) * | 2023-09-27 | 2024-04-16 | 中国石油大学(华东) | Method and system for predicting fatigue crack growth of X80 pipeline steel under random load effect |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
CN113205207A (en) | XGboost algorithm-based short-term power consumption load fluctuation prediction method and system | |
CN109214708B (en) | Electric power system risk assessment method based on cross entropy theory optimization support vector machine | |
CN110598929B (en) | Wind power nonparametric probability interval ultrashort term prediction method | |
CN113569504B (en) | Prediction method and prediction system for creep fatigue life of aero-engine combustion chamber | |
CN111784061B (en) | Training method, device and equipment for power grid engineering cost prediction model | |
CN112632794A (en) | Power grid reliability evaluation method based on cross entropy parameter subset simulation optimization | |
CN113821931B (en) | Fan output power prediction method and system | |
CN111310990A (en) | Improved gray combination model-based track quality prediction method and system | |
CN116484675A (en) | Crack propagation life prediction method and system for ship engine blade | |
CN111897240B (en) | Simulation method and system based on nuclear power system operation | |
CN111025041A (en) | Electric vehicle charging pile monitoring method and system, computer equipment and medium | |
CN114781692A (en) | Short-term power load prediction method and device and electronic equipment | |
CN114119273A (en) | Park comprehensive energy system non-invasive load decomposition method and system | |
CN115982141A (en) | Characteristic optimization method for time series data prediction | |
CN105389442A (en) | Reverse design method for coupling genetic algorithm, neural network and numerical simulation | |
CN113762591A (en) | Short-term electric quantity prediction method and system based on GRU and multi-core SVM counterstudy | |
CN116861256A (en) | Furnace temperature prediction method, system, equipment and medium for solid waste incineration process | |
CN115409291B (en) | Wind power prediction method and system for correcting wind speed | |
CN115967092A (en) | Data-driven non-parameter probability optimal power flow prediction-analysis integrated method for new energy power system | |
CN116307139A (en) | Wind power ultra-short-term prediction method for optimizing and improving extreme learning machine | |
CN115965177A (en) | Improved autoregressive error compensation wind power prediction method based on attention mechanism | |
CN115860232A (en) | Steam load prediction method, system, electronic device and medium | |
CN112581311B (en) | Method and system for predicting long-term output fluctuation characteristics of aggregated multiple wind power plants | |
Bouzem et al. | Probabilistic and Reliability Analysis of an Intelligent Power Control for a Doubly Fed Induction Generator-Based Wind Turbine System | |
CN113642784A (en) | Wind power ultra-short term prediction method considering fan state |
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 |