CN114117684A - Selective laser melting forming 316L stainless steel abrasion prediction method based on machine learning - Google Patents

Selective laser melting forming 316L stainless steel abrasion prediction method based on machine learning Download PDF

Info

Publication number
CN114117684A
CN114117684A CN202111477638.0A CN202111477638A CN114117684A CN 114117684 A CN114117684 A CN 114117684A CN 202111477638 A CN202111477638 A CN 202111477638A CN 114117684 A CN114117684 A CN 114117684A
Authority
CN
China
Prior art keywords
laser melting
selective laser
stainless steel
remelting
neural network
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
Application number
CN202111477638.0A
Other languages
Chinese (zh)
Inventor
丁红钦
袁梓骏
祝毅
张超
谢海波
杨华勇
Current Assignee (The listed assignees may be inaccurate. Google has not performed a legal analysis and makes no representation or warranty as to the accuracy of the list.)
High End Equipment Research Institute Of Zhejiang University
Original Assignee
High End Equipment Research Institute Of Zhejiang University
Priority date (The priority date 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 date listed.)
Filing date
Publication date
Application filed by High End Equipment Research Institute Of Zhejiang University filed Critical High End Equipment Research Institute Of Zhejiang University
Priority to CN202111477638.0A priority Critical patent/CN114117684A/en
Publication of CN114117684A publication Critical patent/CN114117684A/en
Pending legal-status Critical Current

Links

Images

Classifications

    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06F—ELECTRIC DIGITAL DATA PROCESSING
    • G06F30/00—Computer-aided design [CAD]
    • G06F30/10—Geometric CAD
    • G06F30/17—Mechanical parametric or variational design
    • B—PERFORMING OPERATIONS; TRANSPORTING
    • B22—CASTING; POWDER METALLURGY
    • B22F—WORKING METALLIC POWDER; MANUFACTURE OF ARTICLES FROM METALLIC POWDER; MAKING METALLIC POWDER; APPARATUS OR DEVICES SPECIALLY ADAPTED FOR METALLIC POWDER
    • B22F10/00—Additive manufacturing of workpieces or articles from metallic powder
    • B22F10/20—Direct sintering or melting
    • B22F10/28—Powder bed fusion, e.g. selective laser melting [SLM] or electron beam melting [EBM]
    • B—PERFORMING OPERATIONS; TRANSPORTING
    • B22—CASTING; POWDER METALLURGY
    • B22F—WORKING METALLIC POWDER; MANUFACTURE OF ARTICLES FROM METALLIC POWDER; MAKING METALLIC POWDER; APPARATUS OR DEVICES SPECIALLY ADAPTED FOR METALLIC POWDER
    • B22F10/00—Additive manufacturing of workpieces or articles from metallic powder
    • B22F10/80—Data acquisition or data processing
    • B—PERFORMING OPERATIONS; TRANSPORTING
    • B33—ADDITIVE MANUFACTURING TECHNOLOGY
    • B33Y—ADDITIVE MANUFACTURING, i.e. MANUFACTURING OF THREE-DIMENSIONAL [3-D] OBJECTS BY ADDITIVE DEPOSITION, ADDITIVE AGGLOMERATION OR ADDITIVE LAYERING, e.g. BY 3-D PRINTING, STEREOLITHOGRAPHY OR SELECTIVE LASER SINTERING
    • B33Y10/00—Processes of additive manufacturing
    • B—PERFORMING OPERATIONS; TRANSPORTING
    • B33—ADDITIVE MANUFACTURING TECHNOLOGY
    • B33Y—ADDITIVE MANUFACTURING, i.e. MANUFACTURING OF THREE-DIMENSIONAL [3-D] OBJECTS BY ADDITIVE DEPOSITION, ADDITIVE AGGLOMERATION OR ADDITIVE LAYERING, e.g. BY 3-D PRINTING, STEREOLITHOGRAPHY OR SELECTIVE LASER SINTERING
    • B33Y50/00—Data acquisition or data processing for additive manufacturing
    • 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
    • G06N20/00—Machine learning
    • 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
    • G06N3/044—Recurrent networks, e.g. Hopfield 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
    • G06F2113/00—Details relating to the application field
    • G06F2113/10—Additive manufacturing, e.g. 3D printing
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06F—ELECTRIC DIGITAL DATA PROCESSING
    • G06F2119/00—Details relating to the type or aim of the analysis or the optimisation
    • G06F2119/04—Ageing analysis or optimisation against ageing
    • 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
    • G06F2119/18—Manufacturability analysis or optimisation for manufacturability
    • 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
    • Y02P—CLIMATE CHANGE MITIGATION TECHNOLOGIES IN THE PRODUCTION OR PROCESSING OF GOODS
    • Y02P10/00—Technologies related to metal processing
    • Y02P10/25—Process efficiency

Abstract

The invention discloses a method for predicting the abrasion of 316L stainless steel formed by selective laser melting based on machine learning, which comprises the following steps of firstly, establishing a training data set based on technological parameters of the selective laser melting forming process and the abrasion rate of 316L stainless steel formed by selective laser melting; then, carrying out standardization processing on the training data set; constructing a machine learning prediction model of 316L stainless steel abrasion of selective laser melting forming based on a neural network, and training the model by adopting a standardized training data set; and finally, carrying out the same standardization on the technological parameters of the new selective laser melting forming process, and inputting the technological parameters into the trained machine learning prediction model to obtain the corresponding wear rate of the 316L stainless steel. The method has high prediction precision.

Description

Selective laser melting forming 316L stainless steel abrasion prediction method based on machine learning
Technical Field
The invention relates to the technical field of additive manufacturing, in particular to a method for predicting the abrasion of 316L stainless steel formed by selective laser melting based on machine learning.
Background
With the rapid development of machining technology, additive manufacturing technology has attracted people's attention as a new production mode. Compared with the traditional material reducing manufacturing mode, the material increasing manufacturing mode has the advantages of short production period, no limitation of part structures, effective cost saving and the like. Therefore, the additive manufacturing technology is widely applied to the fields of aerospace, biomedicine, automobile manufacturing and the like. The selective laser melting technique is one of metal additive manufacturing techniques, which first slices according to a CAD model, then uses a fiber laser to selectively scan pre-laid metal powder, builds parts layer by layer, and finally forms the required components. Due to the flexibility of selective laser melting manufacturing techniques, hard composite materials with high strength and high wear resistance are often selected when machining parts with specific structural and functional requirements.
However, since the selective laser melting process is accompanied by laser interaction with the material, the continuous cycling of heating and cooling produces unique thermal effects that cause phenomena such as bath extension, incomplete melting of the material, Marangoni convection, and powder oxidation. Therefore, the selective laser melting process can generate defects such as pores, sticky powder, cracks, stress concentration and the like which are different from the traditional manufacturing process, and the microstructure of the material is also changed, so that the mechanical property and the surface property of the material are changed in response. Therefore, in the actual engineering field, the formed part prepared by the selective laser melting process may aggravate the problem of abrasion damage due to the manufacturing defects on the surface of the formed part, and in severe cases, the surface material of the formed part may be peeled off, the working performance may be reduced, and the service life may be reduced, which is directly related to the reliability of the formed part. Therefore, the abrasion problem of the selective laser melting process formed piece is more serious than that of the formed piece in the traditional mode, and the abrasion problem needs to be studied intensively. Generally, post-processing techniques are important means to enhance the surface properties of selected laser melt-formed parts. However, due to the complexity of the post-processing process, particularly for shaped parts having complex structures and geometries, the post-processing process may not be suitable. Therefore, in order to enable the formed part prepared by the selective laser melting process to meet the requirement of surface performance, the selective laser melting process parameters are adjusted, and the method is also an effective way for improving the surface performance of the formed part. The process parameters (such as laser power, scanning speed, layer thickness, scanning distance, etc.) and scanning strategies involved in the selective laser melting process affect the processing process, and any change in the parameters may change the microstructure of the material, thereby causing a change in the surface properties.
Although many studies have been made to reveal the relationship between the selective laser melting process parameters and the surface properties of the formed part, the research process often requires a lot of experiments, especially the wear properties of the surface of the formed part, a lot of experimental samples are required to be processed, and a lot of repeated experiments are required to obtain the wear properties of the selective laser melting formed part. Therefore, it is necessary to reduce the number of experiments, reduce the cost of research, and improve the efficiency of obtaining the performance of the formed part. However, additive manufacturing is a complex process with multiple physical fields and multiple scales, how to establish a reliable physical model of process parameters, microstructure and surface performance still faces huge challenges, and how to quickly obtain the wear performance of a selective laser melting forming piece is also an important problem to be solved urgently. The appearance of machine learning provides a new idea for scientific research personnel, the machine learning is a subset of artificial intelligence, and the surface performance of a formed part can be predicted by means of data fitting according to past process parameters. The application of machine learning in additive manufacturing can not only help practitioners to optimize the manufacturing process and improve the product quality, but also save a large amount of time and cost, and enable non-professionals to participate in the parameter optimization process. Therefore, the machine learning algorithm is of great significance in prediction of the surface performance of the selected laser melting forming part, and an effective way is provided for revealing the wear prediction of the 316L stainless steel formed by the selected laser melting forming part.
Disclosure of Invention
Aiming at the defects of the prior art, the invention provides a method for predicting the abrasion of the 316L stainless steel formed by selective laser melting based on machine learning, which can rapidly and accurately predict the abrasion rate of the 316L stainless steel formed by selective laser melting.
The purpose of the invention is realized by the following technical scheme:
a selected-region laser melting forming 316L stainless steel abrasion prediction method based on machine learning comprises the following steps:
(1) establishing a corresponding relation between the technological parameters of the selective laser melting forming process and the wear rate of the selective laser melting forming 316L stainless steel based on the technological parameters of the selective laser melting forming process and the wear rate of the selective laser melting forming 316L stainless steel, extracting characteristic data and creating a training data set;
(2) carrying out standardization processing on the training data set;
(3) constructing a machine learning prediction model of 316L stainless steel abrasion of selective laser melting forming based on a neural network, and training the model by adopting a standardized training data set;
(4) and (3) after the new technological parameters of the selective laser melting forming process are standardized as in the step (2), inputting the new technological parameters into the trained machine learning prediction model obtained in the step (3), and obtaining the wear rate of the corresponding 316L stainless steel.
Further, the training data set comprises feature data X and a target attribute Y, wherein the target attribute Y is a wear rate; the characteristic data X comprises point distance, exposure time, scanning speed, scanning distance, energy density, load, sliding distance and a forming mode; the point distance, the exposure time, the scanning speed, the scanning distance and the energy density are technological parameters in the selective laser melting forming process, the load and the sliding distance are corresponding working condition data of a friction and wear experiment, and the forming modes are three, including non-remelting, parallel remelting and vertical remelting; when the forming mode is not remelting, the technological parameters of the selective laser melting forming process are the parameters of the substrate layer; when the forming mode is parallel remelting, the technological parameters of the selective laser melting forming process are parameters of the remelting layer, and the energy density is equal to the energy density of the substrate layer plus 0.83 times of the energy density of the remelting layer; when the forming mode is vertical remelting, the technological parameters of the zone-selection laser melting forming process are parameters of the remelting layer, and the energy density is equal to the energy density of the base layer plus 0.75 times of the energy density of the remelting layer.
Further, the neural network selects a BP neural network, and the training process of the BP neural network specifically includes:
(1) setting the number of neuron layers of a neural network, determining an activation function and a batch _ size, and establishing a loss function MSELoss, wherein adjustable key parameters in a neural network model comprise a learning rate learning _ rate, a regularization coefficient weight _ decay and cycle times; and constructing an evaluation standard R for the accuracy of the model prediction result2For representing the degree of reproduction of the observation by the model, R2Closer to 1 indicates higher prediction accuracy; the R is2The calculation formula of (a) is as follows:
Figure BDA0003394074050000031
wherein a isiMeans for indicating the corresponding fact of the ith dataRate of wear, piRepresenting a predicted wear rate for the ith data, a representing an average of actual wear rates, and n representing the number of data in the current data set;
(2) after completing one round of training, calculating R2When R is2When the parameter is less than the set threshold value, after the parameter is adjusted, the BP neural network continues to train until R2And stopping training when the parameter is larger than or equal to the set threshold value, and obtaining the BP neural network with the optimal parameters.
The invention has the beneficial effects that:
the method for predicting the wear of the 316L stainless steel formed by the selective laser melting based on the machine learning realizes the differential representation of the manufacturing states of non-remelting, vertical remelting and parallel remelting, can be used for a machine learning model, fully considers the incidence relation between the technological parameters of the selective laser melting process and the working condition parameters of a friction and wear experiment, establishes and trains the machine learning prediction model of the 316L stainless steel formed by the selective laser melting based on a neural network, and predicts the wear rate of the 316L stainless steel under the technological parameters of a new selective laser melting forming process after finishing the training and storing of the machine learning prediction model of the 316L stainless steel formed by the selective laser melting based on the neural network. The method has high prediction precision.
Drawings
FIG. 1 is a flow chart of a machine learning based selective laser melting forming 316L stainless steel wear prediction method of the present invention.
Fig. 2 is a flow chart of training a BP neural network.
FIG. 3 is a schematic view of parallel reflow and vertical reflow, wherein the left drawing is a schematic view of parallel reflow and the right drawing is a schematic view of vertical reflow.
Fig. 4 is a schematic diagram of the architecture of the BP neural network.
FIG. 5 is a diagram of the calculated prediction results after the neural network has been trained.
Detailed Description
The present invention will be described in detail below with reference to the accompanying drawings and preferred embodiments, and the objects and effects of the present invention will become more apparent, it being understood that the specific embodiments described herein are merely illustrative of the present invention and are not intended to limit the present invention.
In the embodiment, the material of the selected area laser melting forming sample is 316L stainless steel, and the chemical components of the selected area laser melting forming sample are less than or equal to 0.03 percent of C, 10.00-14.00 percent of Ni, less than or equal to 2.00 percent of Mn, less than or equal to 0.03 percent of S, less than or equal to 0.045 percent of P, 16.00-18.00 percent of Cr, less than or equal to 0.20 percent of Cu, 2.00-3.00 percent of Mo, and the balance of Fe.
As shown in fig. 1, the method for predicting the abrasion of 316L stainless steel formed by selective laser melting based on machine learning specifically comprises the following steps:
the method comprises the following steps: establishing a corresponding relation between the technological parameters of the selective laser melting forming process and the wear rate of the selective laser melting forming 316L stainless steel based on the technological parameters of the selective laser melting forming process and the wear rate of the selective laser melting forming 316L stainless steel, extracting characteristic data, and establishing a wear database of the selective laser melting forming 316L stainless steel, wherein the detailed steps are as follows:
(1) establishing a three-dimensional model of a 316L stainless steel sample, wherein the model is a cube with the size of 10 multiplied by 10 mm;
(2) setting laser process parameters of a sample formed by the selective laser melting technology, wherein the first group is a non-remelting sample, the processing power is 200W, the scanning strategy is Meander, the exposure time is respectively 80 mu s and 150 mu s, the layer thickness is 50 mu m, the dot pitch is respectively 10 mu m, 20 mu m, 30 mu m and 60 mu m, the scanning pitch is respectively 10 mu m, 20 mu m, 30 mu m and 110 mu m, the forming direction is horizontal forming, and the protective gas is argon. As shown in fig. 3, the second set is a parallel remelted sample, the laser process parameters of the substrate layer: the processing power is 200W, the scanning strategy is Meander, the exposure time is 80 Mus, the layer thickness is 50 Mum, the dot pitch is 60 Mum, the scanning distance is 110 Mum, the forming direction is horizontal forming, the protective gas is argon, and the laser process parameters of the remelting layer are as follows: the processing power was 200W, the scanning strategy was "0", the exposure time was 80. mu.s, the layer thickness was 50 μm, the dot pitch was 60 μm, and the scanning pitches were 10 μm, 15 μm, 20 μm, 30 μm, and 110 μm, respectively. The third group is a vertical remelting sample, and the laser process parameters of the basal layer are as follows: the processing power is 200W, the scanning strategy is Meander, the exposure time is 80 Mus, the layer thickness is 50 Mum, the dot pitch is 60 Mum, the scanning distance is 110 Mum, the forming direction is horizontal forming, the protective gas is argon, and the laser process parameters of the remelting layer are as follows: the processing power was 200W, the scanning strategy was "0", the exposure time was 80. mu.s and 150. mu.s, respectively, the layer thickness was 50 μm, the dot pitch was 30 μm, 60 μm and 80 μm, respectively, and the scanning pitch was 10 μm, 15 μm, 20 μm, 30 μm and 110 μm, respectively.
(3) And forming the 316L stainless steel sample according to the set process parameters of the selective laser melting forming sample.
(4) The friction and wear test is carried out on the 316L stainless steel sample formed by selective laser melting, and the specific test working conditions are as follows: the load is 1N, 3N and 5N respectively, the experimental time is 15min, and the sliding reciprocating distance is set to be 1-7 mm.
(5) The sample is weighed before and after the experiment, then a confocal microscope is used for observing the grinding mark of the sample after the experiment, the depth and the length of the grinding mark are recorded, and the wear rates of different experimental samples are calculated. The wear rate δ is calculated as follows:
Figure BDA0003394074050000041
wherein V is the wear volume of the test piece, and Sigma W is the accumulated friction work.
(6) Calculating the energy density of different samples, wherein the calculation formula of the energy density E is as follows:
Figure BDA0003394074050000051
where P denotes the laser power, v denotes the scanning speed, h denotes the scanning pitch, tlThickness of the layer, d dot pitch, teThe exposure time is indicated.
In this example, different energy density calculation methods were used to distinguish between non-remelted, vertically remelted, and parallel remelted selective laser melt-formed 316L stainless steel samples, specifically:
the parameters of the basal layer are selected as the technological parameters (the dot pitch, the exposure time, the scanning speed and the scanning interval) of the selective laser melting forming process corresponding to the non-remelting sample, and the parameters of the remelting layer are selected as the technological parameters (the dot pitch, the exposure time, the scanning speed and the scanning interval) of the selective laser melting forming process corresponding to the remelting sample;
the energy density of the vertically remelted sample is calculated by the following method: the energy density of the basal layer +0.75 × energy density of the remelted layer and the energy density of the parallel remelted sample are calculated by the following steps: base layer energy density +0.83 × remelted layer energy density.
(7) Extracting technological parameters, friction and wear experiment working condition parameters and forming modes of the selective laser melting forming process of all samples as characteristic data, and taking the wear rate as a target output value; the technological parameters of the selective laser melting forming process in the characteristic data comprise: dot pitch, exposure time, scanning speed, scanning pitch, energy density; the friction and wear test working condition parameters comprise: load, sliding distance. The forming mode comprises non-remelting, parallel remelting and vertical remelting. Finally, a wear database of 316L stainless steel formed by selective laser melting is obtained, and the total number of the data pairs is 42. And the data set is divided into a training data set and a verification data set by using a train _ test _ split function in python, and the parameter is set to test _ size ═ 0.3.
Step two: carrying out standardization processing on the training data set;
in this embodiment, a StandardScaler function in python is used to perform normalization processing on a training data set.
Step three: the method comprises the following steps of constructing a machine learning prediction model of 316L stainless steel abrasion of selective laser melting forming based on a neural network, and training the model by adopting a standardized training data set, and specifically comprising the following substeps:
(1) learning and training by using a back propagation neural network (BP for short) framework in python; the BP neural network is a multilayer feedforward neural network based on an error back propagation algorithm, and compared with other neural network algorithms, the BP neural network has strong identification and classification capabilities on external input samples and strong nonlinear mapping capabilities, so that the BP neural network can well predict a highly nonlinear system combining additive manufacturing and frictional wear.
(2) Setting the number of neuron layers and the number of neuron layers of a neural network, as shown in fig. 4, which are an input layer, a 3-layer hidden layer and an output layer, respectively, where the number of neuron layers is 1-13-13-8-1, determining an activation function as a Sigmoid function, and a batch _ size as 30, establishing a loss function mselos, where a parameter reduction is 'mean', key parameters adjustable in a neural network model include a learning rate learning _ rate, a regularization coefficient weight _ decay and a cycle number, where the learning rate learning _ rate is 0.002, the regularization coefficient weight _ decay is 0.001, and the cycle number is 4000.
(3) Establishing an evaluation standard R for the accuracy of the model prediction result2Representing the degree of reproduction of the observation by the prediction model, R2Closer to 1 indicates higher prediction accuracy; after completing one round of training, calculating R2When R is2When the parameter is less than the set threshold value, after the parameter is adjusted, the BP neural network continues to train until R2And stopping training when the parameter is larger than or equal to the set threshold value, and obtaining the BP neural network with the optimal parameters. R2The calculation formula of (a) is as follows:
Figure BDA0003394074050000061
wherein a isiRepresents the actual wear rate, p, corresponding to the ith dataiRepresenting a predicted wear rate for the ith data, a representing an average of actual wear rates, and n representing the number of data in the current data set;
step four: after the training of the machine learning prediction model for the selective laser melting forming 316L stainless steel wear is completed and stored, the present embodiment adopts a verification set to predict the wear rate of the 316L stainless steel under the new technological parameters of the selective laser melting forming process. The method comprises the following steps:
(1) for the verification data set, carrying out standardization processing on the feature data X and the target attribute Y by using a StandardScaler function in python;
(2) inputting the data of the verification data set into the trainedAnd (5) a neural network prediction model to obtain a 316L stainless steel wear rate prediction result. And comparing the predicted result with the test data, as shown in FIG. 5, calculating the trend goodness of fit of the predicted result and the test result, and calculating R of the predicted result2And the prediction accuracy of the obtained prediction model on the whole verification data set is 0.91, so that the expected effect is achieved.
In order to further verify the advantages of the BP neural network, the BP neural network is adopted to predict the wear rate of the 316L stainless steel formed by selective laser melting, machine learning algorithms such as KNN-nearest neighbor, support vector machine regression (SVR), regression tree and the like are also used to predict the wear rate, model parameters are adjusted and optimized, and the R of the KNN-nearest neighbor algorithm is calculated according to the prediction result20.63, R of the regression Tree Algorithm20.81, R of regression algorithm of support vector machine20.83, which are lower than the prediction accuracy of the BP neural network algorithm, so in this embodiment, the BP neural network has the highest prediction accuracy compared to other machine learning algorithms.
It will be understood by those skilled in the art that the foregoing is only a preferred embodiment of the present invention, and is not intended to limit the invention, and although the invention has been described in detail with reference to the foregoing examples, it will be apparent to those skilled in the art that various changes in the form and details of the embodiments may be made and equivalents may be substituted for elements thereof. All modifications, equivalents and the like which come within the spirit and principle of the invention are intended to be included within the scope of the invention.

Claims (3)

1. A selected-region laser melting forming 316L stainless steel abrasion prediction method based on machine learning is characterized by comprising the following steps:
(1) establishing a corresponding relation between the technological parameters of the selective laser melting forming process and the wear rate of the selective laser melting forming 316L stainless steel based on the technological parameters of the selective laser melting forming process and the wear rate of the selective laser melting forming 316L stainless steel, extracting characteristic data and creating a training data set;
(2) carrying out standardization processing on the training data set;
(3) constructing a machine learning prediction model of 316L stainless steel abrasion of selective laser melting forming based on a neural network, and training the model by adopting a standardized training data set;
(4) and (3) after the new technological parameters of the selective laser melting forming process are standardized as in the step (2), inputting the new technological parameters into the trained machine learning prediction model obtained in the step (3), and obtaining the wear rate of the corresponding 316L stainless steel.
2. The machine-learning-based selective laser melting shaping 316L stainless steel wear prediction method of claim 1, wherein the training data set comprises feature data X and a target attribute Y, wherein the target attribute Y is a wear rate; the characteristic data X comprises point distance, exposure time, scanning speed, scanning distance, energy density, load, sliding distance and a forming mode; the point distance, the exposure time, the scanning speed, the scanning distance and the energy density are technological parameters in the selective laser melting forming process, the load and the sliding distance are corresponding working condition data of a friction and wear experiment, and the forming modes are three, including non-remelting, parallel remelting and vertical remelting; when the forming mode is not remelting, the technological parameters of the selective laser melting forming process are the parameters of the substrate layer; when the forming mode is parallel remelting, the technological parameters of the selective laser melting forming process are parameters of the remelting layer, and the energy density is equal to the energy density of the substrate layer plus 0.83 times of the energy density of the remelting layer; when the forming mode is vertical remelting, the technological parameters of the zone-selection laser melting forming process are parameters of the remelting layer, and the energy density is equal to the energy density of the base layer plus 0.75 times of the energy density of the remelting layer.
3. The machine learning-based selective laser melting forming 316L stainless steel wear prediction method of claim 1, wherein the neural network is a BP neural network, and the process of training the BP neural network specifically comprises:
(1) setting the number of neuron layers and the number of neuron elements of the neural network, and determiningActivating a function and batch _ size, establishing a loss function MSELoss, wherein adjustable key parameters in the neural network model comprise a learning rate learning _ rate, a regularization coefficient weight _ decay and cycle times; and constructing an evaluation standard R for the accuracy of the model prediction result2For representing the degree of reproduction of the observation by the model, R2Closer to 1 indicates higher prediction accuracy; the R is2The calculation formula of (a) is as follows:
Figure FDA0003394074040000021
wherein a isiRepresents the actual wear rate, p, corresponding to the ith dataiRepresenting a predicted wear rate for the ith data, a representing an average of actual wear rates, and n representing the number of data in the current data set;
(2) after completing one round of training, calculating R2When R is2When the parameter is less than the set threshold value, after the parameter is adjusted, the BP neural network continues to train until R2And stopping training when the parameter is larger than or equal to the set threshold value, and obtaining the BP neural network with the optimal parameters.
CN202111477638.0A 2021-12-06 2021-12-06 Selective laser melting forming 316L stainless steel abrasion prediction method based on machine learning Pending CN114117684A (en)

Priority Applications (1)

Application Number Priority Date Filing Date Title
CN202111477638.0A CN114117684A (en) 2021-12-06 2021-12-06 Selective laser melting forming 316L stainless steel abrasion prediction method based on machine learning

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
CN202111477638.0A CN114117684A (en) 2021-12-06 2021-12-06 Selective laser melting forming 316L stainless steel abrasion prediction method based on machine learning

Publications (1)

Publication Number Publication Date
CN114117684A true CN114117684A (en) 2022-03-01

Family

ID=80366947

Family Applications (1)

Application Number Title Priority Date Filing Date
CN202111477638.0A Pending CN114117684A (en) 2021-12-06 2021-12-06 Selective laser melting forming 316L stainless steel abrasion prediction method based on machine learning

Country Status (1)

Country Link
CN (1) CN114117684A (en)

Cited By (1)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN115415542A (en) * 2022-07-19 2022-12-02 闽都创新实验室 Method for predicting performance of duplex stainless steel 3D printing piece based on response surface method

Cited By (2)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN115415542A (en) * 2022-07-19 2022-12-02 闽都创新实验室 Method for predicting performance of duplex stainless steel 3D printing piece based on response surface method
CN115415542B (en) * 2022-07-19 2024-05-03 闽都创新实验室 Prediction method for performance of duplex stainless steel 3D printing piece based on response surface method

Similar Documents

Publication Publication Date Title
Ren et al. Thermal field prediction for laser scanning paths in laser aided additive manufacturing by physics-based machine learning
Rajmohan et al. Grey-fuzzy algorithm to optimise machining parameters in drilling of hybrid metal matrix composites
Benardos et al. Predicting surface roughness in machining: a review
Zhang et al. Modeling and optimization of medium-speed WEDM process parameters for machining SKD11
CN112632720B (en) Multidimensional data fusion and quantitative modeling method for metal additive manufacturing process system
CN112651562B (en) Laser cladding process optimization method based on signal-to-noise ratio and improved gray correlation degree
Ladani Applications of artificial intelligence and machine learning in metal additive manufacturing
CN110321658A (en) A kind of prediction technique and device of plate property
CN114117684A (en) Selective laser melting forming 316L stainless steel abrasion prediction method based on machine learning
Shafaie et al. Modified GTN parameters calibration in additive manufacturing of Ti-6Al-4 V alloy: a hybrid ANN-PSO approach
Mouralova et al. Analysis of cut orientation through half-finished product using WEDM
Lu et al. Knowledge transfer using Bayesian learning for predicting the process-property relationship of Inconel alloys obtained by laser powder bed fusion
Vashistha et al. Machine learning assisted optimization of tribological parameters of Al–Co–Cr–Fe–Ni high-entropy alloy
CN113240096B (en) Casting cylinder cover micro-structure prediction method based on rough set and neural network
ÇELİK et al. Modelling of the clearance effects in the blanking process of CuZn30 sheet metal using neural network− a comparative study
Gupta et al. Multi-objective optimization of friction stir welding of aluminium alloy using grey relation analysis with entropy measurement method
Zhang et al. Roughness Prediction and Performance Analysis of Data-Driven Superalloy Belt Grinding
Marrey et al. A Novel Framework for Predictive Modeling and Optimization of Powder Bed Fusion Process
Gusel et al. HARDNESS MODELLING OF DEFORMED CW106C ALLOY BY A GENETIC PROGRAMMING.
Chen et al. A just-in-time learning-based compensation prediction method for hull plate welding
Raza et al. A Consolidated Approach towards Application of Machine Learning Principles in Additive Manufacturing
Wang et al. Optimization of laser cladding process parameters based on genetic algorithm and neural networks
Era et al. Prediction of mechanical behaviors of additively manufactured SS 316L via machine learning
Zhou et al. Advancements in machine learning for material design and process optimization in the field of additive manufacturing
Cui Fabrication, characterization of high-entropy alloys and deep learning-based inspection in metal additive manufacturing

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