CN113182709B - Plate laser cutting process parameter optimization method considering sharp corner ablation - Google Patents

Plate laser cutting process parameter optimization method considering sharp corner ablation Download PDF

Info

Publication number
CN113182709B
CN113182709B CN202110525071.3A CN202110525071A CN113182709B CN 113182709 B CN113182709 B CN 113182709B CN 202110525071 A CN202110525071 A CN 202110525071A CN 113182709 B CN113182709 B CN 113182709B
Authority
CN
China
Prior art keywords
value
cutting
laser cutting
sharp corner
steps
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.)
Active
Application number
CN202110525071.3A
Other languages
Chinese (zh)
Other versions
CN113182709A (en
Inventor
黄彬
朱圣杰
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.)
Fuzhou University
Original Assignee
Fuzhou 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 Fuzhou University filed Critical Fuzhou University
Priority to CN202110525071.3A priority Critical patent/CN113182709B/en
Publication of CN113182709A publication Critical patent/CN113182709A/en
Application granted granted Critical
Publication of CN113182709B publication Critical patent/CN113182709B/en
Active legal-status Critical Current
Anticipated expiration legal-status Critical

Links

Images

Classifications

    • B—PERFORMING OPERATIONS; TRANSPORTING
    • B23—MACHINE TOOLS; METAL-WORKING NOT OTHERWISE PROVIDED FOR
    • B23K—SOLDERING OR UNSOLDERING; WELDING; CLADDING OR PLATING BY SOLDERING OR WELDING; CUTTING BY APPLYING HEAT LOCALLY, e.g. FLAME CUTTING; WORKING BY LASER BEAM
    • B23K26/00—Working by laser beam, e.g. welding, cutting or boring
    • B23K26/36—Removing material
    • B23K26/38—Removing material by boring or cutting
    • 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
    • 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
    • Y02P90/00—Enabling technologies with a potential contribution to greenhouse gas [GHG] emissions mitigation
    • Y02P90/30—Computing systems specially adapted for manufacturing

Abstract

The invention relates to a method for optimizing parameters of a plate laser cutting process by considering sharp corner ablation, which comprises the steps of measuring workpiece quality indexes corresponding to each group of tests through linear cutting and sharp corner cutting orthogonal tests, and training a laser cutting quality BP neural network model according to obtained test data. And (4) taking the obtained neural network model as a fitness function of the genetic algorithm to optimize the process parameters to obtain the optimal process parameters. The invention simultaneously takes the linear cutting quality and the sharp corner cutting quality into consideration, and the process obtained by optimizing the obtained process parameters has better kerf width, section roughness, heat affected zone width and slag adhering length and does not generate sharp corner ablation.

Description

Method for optimizing parameters of plate laser cutting process considering sharp corner ablation
Technical Field
The invention relates to the field of production and manufacturing, in particular to a method for optimizing parameters of a plate laser cutting process considering sharp corner ablation.
Background
Compared with the traditional cutting, the laser cutting has the advantages of no cutting force and high processing efficiency, so that the laser cutting is widely applied to plate processing and occupies an important position in the plate processing technology. However, the laser cutting of the plate involves more process parameters including: laser power, assist gas pressure, cutting speed, defocus amount, duty cycle, and pulse frequency. The optimal set of process parameters selected from the combination of the process parameters has important significance for obtaining better laser cutting quality of the plate.
At present, methods related to laser cutting process parameter optimization are three major types, firstly, a model between process parameters and cutting quality is established by utilizing a traditional physical modeling method, then, the process parameters are optimized, the accuracy rate of the method greatly depends on a series of preset mechanical, material, optical and thermal parameters, and in the actual laser cutting process, the parameters often have coupling property and time-varying property, so that the judgment method based on physical modeling is inaccurate. And secondly, an experiment-based method, which takes laser cutting process parameters as experiment factors, performs a large number of laser cutting experiments, obtains the change condition of the laser cutting quality along with each process parameter according to experiment data, and gives suggestions for optimizing the process parameters, but each process parameter has a complex coupling relationship with the cutting quality, if the laser power is increased, the roughness of a tangent plane is reduced, but the quality of a cutting seam is also deteriorated, and the experiment-based method is difficult to give reasonable suggestions for optimizing the process parameters for the situation of the complex coupling relationship. And thirdly, a laser cutting process parameter optimization method based on machine learning, the method can learn the coupling relation between the laser cutting process parameters and the cutting quality through a machine learning model, and the optimization effect is better than that of the former two methods.
At present, the laser cutting process parameter optimization method based on machine learning accounts for more and more, but has the following defects:
most laser cutting process parameter optimization methods are only suitable for linear cutting, and in actual processing, workpieces processed by process parameters obtained based on linear cutting optimization may have sharp corner ablation at a sharp corner, so that the quality of the workpieces is reduced.
Most methods adopt the weighted summation of laser cutting quality evaluation indexes (kerf width, section roughness, heat affected zone width and slag adhering length) to be unified into one evaluation index, and then process parameter optimization is carried out, but the method cannot reasonably evaluate the condition that each index value is extremely distributed.
Disclosure of Invention
In view of the above, the present invention aims to provide a method for optimizing parameters of a laser cutting process of a plate material in consideration of sharp corner ablation, and simultaneously, linear cutting quality and sharp corner cutting quality are taken into consideration, and the process obtained by optimizing the obtained process parameters can have good kerf width, section roughness, heat affected zone width and slag adhering length, and does not cause sharp corner ablation.
The invention is realized by adopting the following scheme: a method for optimizing parameters of a plate laser cutting process considering sharp corner ablation comprises the following steps:
step S1: performing linear cutting and sharp corner cutting orthogonal tests on plates with different thicknesses, measuring workpiece quality indexes corresponding to each set of tests, wherein the linear cutting quality indexes comprise joint cutting width, section roughness, heat affected zone width and slag adhering length, and the sharp corner cutting quality index is sharp corner ablation amount, and then calculating a comprehensive quality index according to the measured indexes;
step S2: the method comprises the following steps of taking plate laser cutting process parameters and workpiece shape characteristics as input data, taking laser cutting quality comprehensive evaluation indexes as output, and constructing a thin plate laser cutting quality BP neural network model; the plate laser cutting process parameters comprise laser power, gas pressure, cutting speed, defocusing amount, duty ratio and pulse frequency; the workpiece shape characteristics comprise plate thickness, a sharp angle and a sharp angle arc;
and step S3: and (3) taking the neural network model trained in the step (S2) as a fitness function of the genetic algorithm, and optimizing the laser cutting quality comprehensive evaluation index value by applying the genetic algorithm to obtain the laser cutting process parameter corresponding to the model when the optimal laser cutting quality comprehensive evaluation index value is obtained, so that the optimization of the laser cutting process parameter is realized.
Further, the step S1 specifically includes the following steps:
step S11: performing a linear laser cutting orthogonal test on plates with different thicknesses, wherein the test factors comprise laser power, cutting speed, gas pressure, duty ratio, pulse frequency and defocusing amount, and the kerf width measured by the nth test is set as w n Roughness of tangent plane is r n The heat-affected zone is ha n And the length of attached slag is lg n ；
Step S12: performing a sharp corner cutting orthogonal test on plates with different thicknesses, wherein the test factors comprise laser power, cutting speed, gas pressure, a sharp corner angle and a sharp corner arc, and setting the ablation amount of the sharp corner measured in the nth test as H n ；
Step S13: calculating the comprehensive quality index of linear cutting;
step S14: calculating the comprehensive quality index Sc of the cutting data of the nth sharp corner n The method specifically comprises the following steps:
Figure BDA0003065383880000021
wherein max (H) and min (H) are respectively the maximum value and the minimum value in the kerf width data;
step S15: and (3) setting the comprehensive quality evaluation index of the straight line and the sharp angle as a neural network tag set D = [ S Sc ], wherein S represents the comprehensive quality evaluation index with a Kini penalty term.
Further, the step S13 specifically includes the following steps:
step S131: calculating the value wz of the kerf width in the nth cutting quality index data after MaxMin normalization n ：
Figure BDA0003065383880000022
Wherein max (w) and min (w) are respectively the maximum value and the minimum value in the kerf width data;
step S132: calculating the value rz of the tangent plane roughness normalized by MaxMin in the nth cutting quality index data n ：
Figure BDA0003065383880000031
Wherein max (r) and min (r) are respectively the maximum value and the minimum value in the section roughness data;
step S133: calculating the value haz of the thermal influence width normalized by MaxMin in the nth cutting quality index data n ：
Figure BDA0003065383880000032
Wherein max (ha) and min (ha) are respectively the maximum value and the minimum value in the kerf width data;
step S134: calculating the value lgz of the slag adhering length in the nth cutting quality index data after the normalization of MaxMin n ：
Figure BDA0003065383880000033
Wherein max (lg) and min (lg) are respectively the maximum value and the minimum value in the slag adhering length data;
step S135: defining the kini index C as:
Figure BDA0003065383880000034
wherein, when i =1,
Figure BDA0003065383880000035
when the ratio of i =2, the ratio of the total of the number of the bits is set to be lower than the value of i =2,
Figure BDA0003065383880000036
when the ratio of i =3, the ratio of the total of the number of the bits is set to be lower than the value of i =3,
Figure BDA0003065383880000037
when the ratio of i =4, the ratio of the total of the number of the coils is set to be lower than the value of i =4,
Figure BDA0003065383880000038
step S136: calculating a comprehensive quality evaluation index S with a Gini penalty term:
Figure BDA0003065383880000039
wherein, ω is 1 -ω 4 Is a weight coefficient, and ω 1 +ω 2 +ω 3 +ω 4 =1; k is a cut-through coefficient, when the plate is cut through, K is 0, the index value is 0, when the plate can be cut through, K is 1, the index value is larger than 1.87, and the larger the index value is, the better the comprehensive cutting quality is; the index can be used for making global judgment on the quality of the linear cutting laser and punishing the phenomenon of two-stage differentiation of index values.
Further, the step S2 specifically includes the following steps:
step S21: setting the input vector of the BP neural network model as follows:
X＝(x 1 ,x 2 ,...,x 9 )
wherein x is 1 -x 9 The method comprises the following steps that 9 input nodes of a neural network respectively receive input of plate thickness, sharp angle angles, sharp angle arcs, laser power, cutting speed, auxiliary gas pressure, defocusing amount, duty ratio and pulse frequency data;
step S22: randomly generating weight matrix between k hidden layer neuron node and input layer
Figure BDA0003065383880000041
And a weight matrix between the hidden layer and the output layer->
Figure BDA0003065383880000042
Step S23: let the input vector of the kth hidden layer neuron node:
Figure BDA0003065383880000043
h 1 -h 9 a weighting result representing output values of the 9 input layer nodes;
step S24: setting output vectors of hidden layer neuron nodes: z = (Z) 1 ,z 2 ,...,z k ) K number of hidden layer neuron nodes; z is a radical of k An output value for each node;
step S25: setting the activation function to a linear rectification function:
Figure BDA0003065383880000044
step S26: setting a learning rate eta (eta is more than 0) and an error threshold E (E is more than 0) of the neural network, wherein the learning rate represents the learning rate, and the error threshold represents the fitting degree of the model;
step S27: calculating the output Y of the neural network:
Figure BDA0003065383880000045
step S28: calculating the error e of the output value of the neural network and the comprehensive evaluation index value of the real laser cutting quality:
Figure BDA0003065383880000046
where n is the number of data, D i Labeling the ith piece of data in the set for the neural network;
step S29: stopping training if the training error E is smaller than the error threshold E, and taking the obtained neural network model as a fitness function of the genetic algorithm in the step S3, otherwise continuing training according to the step S210;
step S210: calculating an error factor delta for the ith output layer i ：
Figure BDA0003065383880000051
Step S211: error factor delta of k-th input layer k ：
Figure BDA0003065383880000052
Step S212: updating a weight matrix between the hidden layer and the output layer according to the following formula:
Figure BDA0003065383880000053
step S213: updating the weight matrix between the input layer and the hidden layer according to the following formula:
Figure BDA0003065383880000054
step S214: the process returns to step S27.
Further, the step S3 specifically includes the following steps:
step S31: determining a parameter value range;
step S32: the optimization model is established as follows:
Figure BDA0003065383880000055
x 4min ≤x 4 ≤x 4max
x 5min ≤x 5 ≤x 5max
x 6min ≤x 6 ≤x 6max
x 7min ≤x 7 ≤x 7max
x 8min ≤x 8 ≤x 8max
x 9min ≤x 9 ≤x 9max
wherein, f n Representing the nth objective function, F representing the trained neural network model, u n 、v n 、w n Respectively representing the plate thickness, the sharp angle and the sharp angle arc of the cutting stage corresponding to the nth objective function;
step S33: setting binary coding length l, population size M, cross probability Pc, variation probability Pm and iteration times T;
step S34: solving each objective function f separately 1 To f n Maximum value of (d);
step S35: solving each objective function (f) separately 1 To f n ) Is measured.
Step S36: and solving the optimal process parameters by a TOPSIS method.
Further, the step S31 specifically includes the following steps:
step S311: the value range of the laser cutting power is as follows:
x 4min ≤x 4 ≤x 4max
step S312: the laser cutting speed value range is as follows:
x 5min ≤x 5 ≤x 5max
step S313: the auxiliary gas pressure value range is as follows:
x 6min ≤x 6 ≤x 6max
step S314: the defocus measurement value range is as follows:
x 7min ≤x 7 ≤x 7max
step S315: duty cycle span, i.e.:
x 8min ≤x 8 ≤x 9max
step S316: the range of the pulse frequency is as follows:
x 9min ≤x 9 ≤x 9max 。
further, in the step S34, any one objective function f is processed n The method comprises the following steps:
step S341: initializing a population; a mode of randomly generating an initial population is adopted, namely, the value of the gene on the chromosome is randomly selected from 0 and 1;
step S342: calculating the fitness; after the initialized population is decoded, substituting the decoded initialized population into a BP neural network, and taking an output value F of the BP neural network as fitness; selecting individuals with highest fitness from the population, directly inheriting the individuals to the next generation, and performing selection, crossing and mutation operations on the rest individuals;
step S343: selecting operation; generating a new chromosome by adopting a roulette selection mode, wherein the roulette selection mode is related to the fitness value of the chromosome, and the probability of selection is high when the fitness value of the chromosome is larger;
step S344: performing cross operation; random crossing operation is adopted, namely two individuals are randomly selected without being replaced, and one allele is randomly selected as a crossing point to carry out crossing;
step S345: mutation; adopting random variation operation, namely selecting an individual and generating a random number for the individual, and if the number is less than the variation probability Pc, randomly selecting a gene of the individual to negate the gene;
step S346: repeating the steps S342 to S345 until the algebra reaches the set iteration number T;
step S347: for the objective function f 1 To f n The maximum objective function values are respectively marked as F 1 + To
Figure BDA0003065383880000061
Further, the step S35 specifically includes the following steps:
step S351: for any one objective function f n Taking the negative number thereof, namely-F, as a fitness function and iterating according to the steps S341 to S346;
step S352: for the objective function f 1 To f n The minimum objective function values are respectively marked as F 1 - To
Figure BDA0003065383880000071
Further, the step S36 specifically includes the following steps:
step S361: the distances from the individual to the positive and negative ideal points are respectively set as
Figure BDA0003065383880000072
Figure BDA0003065383880000073
Figure BDA0003065383880000074
Wherein the content of the first and second substances,
Figure BDA0003065383880000075
a jth objective function value for a kth individual;
step S362: let the relative closeness d of an individual to an ideal point:
Figure BDA0003065383880000076
step S363: repeating the step S341 to the step S346 by taking the d as a fitness function for iteration;
step S364: and taking the individual with the maximum fitness in all the individuals in the T generation as the optimal process parameter individual.
Compared with the prior art, the invention has the following beneficial effects:
1. the laser process parameter optimization method simultaneously takes the linear cutting quality and the sharp corner cutting quality into consideration, and the process obtained by optimizing the process parameter processing can have good kerf width, section roughness, heat affected zone width and slag adhering length and does not generate sharp corner ablation.
2. The method adopts the mode that a Kini penalty term is introduced into a laser cutting comprehensive index to form a new index, and the index can carry out overall judgment on the quality and also can carry out punishment on the phenomenon of two-stage differentiation of index values.
Drawings
FIG. 1 is a flow chart of a process for optimizing process parameters according to an embodiment of the present invention.
FIG. 2 is a macroscopic view of a 1mm thick sheet in a straight line cutting test according to an embodiment of the present invention.
FIG. 3 is a macroscopic view of a 1mm thin plate in a corner cutting test according to an embodiment of the present invention.
Fig. 4 is a diagram of an aircraft engine locking plate according to an embodiment of the present invention, where 1 is a first straight line, 2 is a second straight line, 3 is a third straight line, 4 is a fourth straight line, 5 is a fifth straight line, 6 is a sixth straight line, 7 is a seventh straight line, 8 is an arc, 9 is a ninth straight line, 10 is a tenth straight line, and 11 is a full circle.
Detailed Description
The invention is further explained below with reference to the drawings and the embodiments.
It should be noted that the following detailed description is exemplary and is intended to provide further explanation of the disclosure. Unless defined otherwise, all technical and scientific terms used herein have the same meaning as commonly understood by one of ordinary skill in the art to which this application belongs.
It is noted that the terminology used herein is for the purpose of describing particular embodiments only and is not intended to be limiting of exemplary embodiments according to the present application. As used herein, the singular forms "a", "an" and "the" are intended to include the plural forms as well, and it should be understood that when the terms "comprises" and/or "comprising" are used in this specification, they specify the presence of stated features, steps, operations, devices, components, and/or combinations thereof, unless the context clearly indicates otherwise.
As shown in fig. 1, the present embodiment provides a method for optimizing parameters of a laser cutting process of a plate material in consideration of sharp corner ablation, which includes the following steps:
step S1: performing linear cutting and sharp corner cutting orthogonal tests on plates with different thicknesses, measuring workpiece quality indexes corresponding to each set of tests, wherein the linear cutting quality indexes comprise joint cutting width, section roughness, heat affected zone width and slag adhering length, and the sharp corner cutting quality index is sharp corner ablation amount, and then calculating a comprehensive quality index according to the measured indexes;
step S2: the method comprises the following steps of taking plate laser cutting process parameters and workpiece shape characteristics as input data, taking laser cutting quality comprehensive evaluation indexes as output, and constructing a thin plate laser cutting quality BP neural network model; the plate laser cutting process parameters comprise laser power, gas pressure, cutting speed, defocusing amount, duty ratio and pulse frequency; the workpiece shape characteristics comprise plate thickness, a sharp angle and a sharp arc;
and step S3: and (3) taking the neural network model trained in the step (S2) as a fitness function of the genetic algorithm, and optimizing the laser cutting quality comprehensive evaluation index value by using the genetic algorithm to obtain the laser cutting process parameter corresponding to the model when the optimal laser cutting quality comprehensive evaluation index value is obtained, so that the optimization of the laser cutting process parameter is realized.
In this embodiment, the step S1 specifically includes the following steps:
step S11: performing a linear laser cutting orthogonal test on plates with different thicknesses, wherein the test factors comprise laser power, cutting speed and gasThe body pressure, duty ratio, pulse frequency and defocusing amount are set as w, and the kerf width measured by the nth test is set as n Roughness of tangent plane is r n The heat-affected zone is ha n And the length of attached slag is lg n ；
Step S12: performing a sharp corner cutting orthogonal test on plates with different thicknesses, wherein the test factors comprise laser power, cutting speed, gas pressure, a sharp corner angle and a sharp corner arc, and setting the ablation amount of the sharp corner measured in the nth test as H n ；
Step S13: calculating the comprehensive quality index of linear cutting;
step S14: calculating the comprehensive quality index Sc of the cutting data of the nth sharp corner n The method specifically comprises the following steps:
Figure BDA0003065383880000081
wherein max (H) and min (H) are respectively the maximum value and the minimum value in the kerf width data;
step S15: and (3) setting the comprehensive quality evaluation index of the straight line and the sharp angle as a neural network tag set D = [ S Sc ], wherein S represents the comprehensive quality evaluation index with a Kini penalty term.
In this embodiment, the step S13 specifically includes the following steps:
step S131: calculating the value wz of the kerf width in the nth cutting quality index data after MaxMin normalization n ：
Figure BDA0003065383880000091
Wherein max (w) and min (w) are respectively the maximum value and the minimum value in the kerf width data;
step S132: calculating the value rz of the tangent plane roughness in the nth cutting quality index data after MaxMin normalization n ：
Figure BDA0003065383880000092
Wherein max (r) and min (r) are respectively the maximum value and the minimum value in the section roughness data;
step S133: calculating the value haz of the thermal influence width normalized by MaxMin in the nth cutting quality index data n ：
Figure BDA0003065383880000093
Wherein max (ha) and min (ha) are respectively the maximum value and the minimum value in the kerf width data;
step S134: calculating the value lgz of the slag adhering length in the nth cutting quality index data after the normalization of MaxMin n ：
Figure BDA0003065383880000094
Wherein max (lg) and min (lg) are respectively the maximum value and the minimum value in the slag adhering length data;
step S135: defining the kini index C as:
Figure BDA0003065383880000095
wherein, when i =1,
Figure BDA0003065383880000096
when the ratio of i =2, the ratio of the total of the number of the bits is set to be lower than the value of i =2,
Figure BDA0003065383880000097
when the ratio of i =3, the ratio of the total of the number of the bits is set to be lower than the value of i =3,
Figure BDA0003065383880000101
when the ratio of the sum of the i =4,
Figure BDA0003065383880000102
step S136: calculating a comprehensive quality evaluation index S with a Gini penalty term:
Figure BDA0003065383880000103
wherein, ω is 1 -ω 4 Is a weight coefficient, and ω 1 +ω 2 +ω 3 +ω 4 =1; k is a cut-through coefficient, when the plate is cut through, K is 0, the index value is 0, when the plate can be cut through, K is 1, the index value is larger than 1.87, and the larger the index value is, the better the comprehensive cutting quality is; the index can make global judgment on the quality of the linear cutting laser and punish the occurrence of two-stage differentiation phenomena of index values.
In this embodiment, the step S2 specifically includes the following steps:
step S21: setting the input vector of the BP neural network model as follows:
X＝(x 1 ,x 2 ,...,x 9 )
wherein x is 1 -x 9 The method comprises the following steps that 9 input nodes of a neural network respectively receive input of plate thickness, sharp angle angles, sharp angle arcs, laser power, cutting speed, auxiliary gas pressure, defocusing amount, duty ratio and pulse frequency data;
step S22: randomly generating weight matrix between kth hidden layer neuron node and input layer
Figure BDA0003065383880000104
And a weight matrix between the hidden layer and the output layer>
Figure BDA0003065383880000105
Step S23: let the input vector of the kth hidden layer neuron node:
Figure BDA0003065383880000106
h 1 -h 9 a weighting result representing output values of the 9 input layer nodes;
step S24: setting output vectors of hidden layer neuron nodes: z = (Z) 1 ,z 2 ,...,z k ) K number of hidden layer neuron nodes; z is a radical of formula k An output value for each node;
step S25: setting the activation function to a linear rectification function:
Figure BDA0003065383880000107
step S26: setting a learning rate eta (eta is more than 0) and an error threshold E (E is more than 0) of the neural network, wherein the learning rate represents the learning rate, and the error threshold represents the fitting degree of the model;
step S27: calculating the output Y of the neural network:
Figure BDA0003065383880000111
step S28: calculating the error e of the output value of the neural network and the comprehensive evaluation index value of the real laser cutting quality:
Figure BDA0003065383880000112
where n is the number of data, D i Labeling the ith piece of data in the set for the neural network;
step S29: stopping training if the training error E is smaller than the error threshold E, and taking the obtained neural network model as a fitness function of the genetic algorithm in the step S3, otherwise continuing training according to the step S210;
step S210: calculating an error factor delta of the ith output layer i ：
Figure BDA0003065383880000113
Step S211: the kth inputError factor delta for in-layer k ：
Figure BDA0003065383880000114
Step S212: updating a weight matrix between the hidden layer and the output layer according to the following formula:
Figure BDA0003065383880000115
step S213: updating the weight matrix between the input layer and the hidden layer according to the following formula:
Figure BDA0003065383880000116
step S214: the process returns to step S27.
In this embodiment, the step S3 specifically includes the following steps:
step S31: determining a parameter value range;
step S32: the optimization model is established as follows:
Figure BDA0003065383880000121
x 4min ≤x 4 ≤x 4max
x 5min ≤x 5 ≤x 5max
x 6min ≤x 6 ≤x 6max
x 7min ≤x 7 ≤x 7max
x 8min ≤x 8 ≤x 8max
x 9min ≤x 9 ≤x 9max
wherein, f n Representing the nth objective function, F representing the trained neural network model, u n 、v n 、w n Respectively representThe plate thickness, the sharp angle and the sharp angle arc of the cutting stage corresponding to the n objective functions;
step S33: setting binary coding length l, population size M, cross probability Pc, variation probability Pm and iteration times T;
step S34: solving each objective function f separately 1 To f n Maximum value of (d);
step S35: solving each objective function (f) separately 1 To f n ) Is measured.
Step S36: and solving the optimal process parameters by using a TOPSIS method.
In this embodiment, the step S31 specifically includes the following steps:
step S311: the value range of the laser cutting power is as follows:
x 4min ≤x 4 ≤x 4max
step S312: the laser cutting speed value range is as follows:
x 5min ≤x 5 ≤x 5max
step S313: the auxiliary gas pressure value range is as follows:
x 6min ≤x 6 ≤x 6max
step S314: the defocus measurement value range is as follows:
x 7min ≤x 7 ≤x 7max
step S315: the duty cycle value range, namely:
x 8min ≤x 8 ≤x 9max
step S316: the range of the pulse frequency is as follows:
x 9min ≤x 9 ≤x 9max 。
in this embodiment, in step S34, any one objective function f is processed n The method comprises the following steps:
step S341: initializing a population; a mode of randomly generating an initial population is adopted, namely, the value of the gene on the chromosome is randomly selected from 0 and 1;
step S342: calculating the fitness; after the initialized population is decoded, substituting the decoded initialized population into a BP neural network, and taking an output value F of the BP neural network as fitness; selecting individuals with highest fitness from the population, directly inheriting the individuals to the next generation, and performing selection, crossing and mutation operations on the rest individuals;
step S343: selecting operation; generating a new chromosome by adopting a roulette selection mode, wherein the roulette selection mode is related to the fitness value of the chromosome, and the probability of selection is high when the fitness value of the chromosome is larger;
step S344: performing cross operation; random crossing operation is adopted, namely two individuals are randomly selected without being replaced, and one allele is randomly selected as a crossing point to carry out crossing;
step S345: mutation; adopting random variation operation, namely selecting an individual and generating a random number for the individual, and if the number is less than the variation probability Pc, randomly selecting a gene of the individual to negate the gene;
step S346: repeating the steps S342 to S345 until the algebra reaches the set iteration number T;
step S347: for the objective function f 1 To f n And the maximum objective function values thereof are respectively marked as F 1 + To
Figure BDA0003065383880000131
In this embodiment, the step S35 specifically includes the following steps:
step S351: for any one objective function f n Taking the negative number thereof, namely-F, as a fitness function and iterating according to the steps S341 to S346;
step S352: for the objective function f 1 To f n The minimum objective function values are respectively marked as F 1 - To
Figure BDA0003065383880000132
In this embodiment, the step S36 specifically includes the following steps:
step S361: set a body toThe distances between the positive and negative ideal points are respectively
Figure BDA0003065383880000133
Figure BDA0003065383880000134
Figure BDA0003065383880000135
Wherein, the first and the second end of the pipe are connected with each other,
Figure BDA0003065383880000136
a jth objective function value for a kth individual;
step S362: let the relative closeness d of an individual to an ideal point:
Figure BDA0003065383880000137
step S363: repeating the steps S341 to S346 by taking the d as a fitness function for iteration;
step S364: and taking the individual with the maximum fitness in all the individuals in the T generation as the optimal process parameter individual.
Preferably, aiming at the defects of the prior art, the linear cutting quality and the sharp corner cutting quality are taken as the consideration factors for optimizing the parameters of the laser cutting process, the kerf width, the section roughness, the width of a heat affected zone and the slag adhering length are taken as linear cutting quality indexes, the sharp corner ablation amount is taken as an index of the sharp corner cutting quality, and a comprehensive evaluation index of the laser cutting quality is provided.
(1) And measuring workpiece quality indexes corresponding to each group of tests through a linear cutting and sharp corner cutting orthogonal test, wherein the linear cutting quality indexes comprise joint cutting width, section roughness, heat affected zone width and slag adhering length, the sharp corner cutting quality index is sharp corner ablation amount, and then calculating a comprehensive quality index according to the measured indexes.
(2) And (3) training a laser cutting quality BP neural network model according to the test data obtained in the step (1).
(3) And (3) taking the neural network model obtained in the step (2) as a fitness function of the genetic algorithm, and optimizing the process parameters to obtain the optimal process parameters.
In order to verify the feasibility of the embodiment, a Q235 plate is selected as a case for laser cutting, the specific technical method is applied, a BP neural network model is trained by collecting the process parameters when the Q235 plate is cut by the laser cutting machine, and the laser cutting process parameters are optimized by using a genetic algorithm.
The test was carried out by using a thin Q235 carbon steel plate having a length and width of 200X 200mm and a material thickness of 1, 2 and 3 mm. A ZT-J-6060M metal laser cutting machine produced by Zhengtian company is selected in the test, the laser generator is an optical fiber laser generator, the maximum power is 500w, the auxiliary gas is oxygen, and the cutting mode is continuous laser cutting. The test is divided into two parts, the first part is a test of laser cutting straight lines, and the second part is a test of laser cutting sharp angles.
1. Straight cut orthogonal test. In the test of this section, a Q235 thin plate 1, 2, and 3mm thick was cut in a straight line, the cut length was 80mm, and six factors of laser power, cutting speed, gas pressure, duty ratio, pulse frequency, and defocus, which were used as test cutting parameters, were divided into 5 levels, specifically as shown in table 1, and the six factors were represented by letters a to F, and the levels of each factor were represented by numerals 1 to 5. The L25 (56) orthogonal test Table designed according to Table 1 is shown in Table 2, where the test numbers are represented by 1 to 25, the factors are represented by A to F corresponding to the factors of Table 1, and the levels are represented by 1 to 5 corresponding to the levels of Table 1.
TABLE 1 control factors and level table for straight line cutting orthogonal test
Figure BDA0003065383880000141
Table 2 straight line cutting orthogonal test table
Figure BDA0003065383880000142
Figure BDA0003065383880000151
According to the test arrangement of table 2, linear cutting is respectively performed on the Q235 thin plates with the thickness of 1mm, 2mm and 3mm, the macroscopic appearance of the cut thin plates with the thickness of 1mm is shown in fig. 2, and each cutting seam corresponds to one linear cutting.
In the part of the test, four cutting quality indexes of kerf width, section roughness, a heat affected zone and adhering slag need to be measured. Table 3 lists the kerf width, section roughness, heat affected zone and dross pick up length measured by laser cutting a 1mm thick Q235 sheet. The process parameters taken by each number in the table are the process parameters of the corresponding number in the table 2, and the numbers 5, 10, 19 and 25 are blank rows, which indicate that the cutting result is the non-cutting.
TABLE 3 straight line cutting test data of 1mm steel plate
Figure BDA0003065383880000152
Figure BDA0003065383880000161
2. And (3) performing sharp corner cutting test. In the test of this section, sharp-angled cutting was performed on 1, 2, and 3mm Q235 sheets, respectively, using laser power, cutting speed, gas pressure, sharp-angled angle, and sharp-angled arc radius as test cutting factors, each of which was divided into 4 levels, as shown in table 4, with five factors being represented by letters a to E and the level of each factor being represented by a number from 1 to 4. L16 (4) designed according to Table 4 5 ) The orthogonality test table is shown in Table 5The tests were performed 16 times, with the test numbers being indicated by 1 to 16, the factors being indicated by a to E, the levels being indicated by 1 to 4, the factors and levels of table 5 corresponding to those in table 4.
TABLE 4 Sharp corner cutting control factors and level table
Figure BDA0003065383880000162
TABLE 5 Angle cutting orthogonal test Table
Figure BDA0003065383880000163
Figure BDA0003065383880000171
The sharp-corner cuts were made on 1, 2, and 3mm thick Q235 sheets according to the test schedule of table 5, and the macroscopic appearance of the cut 1mm thick sheets is shown in fig. 3.
In this part of the test, it was necessary to measure the degree of ablation of the sharp corners, and Table 6 lists the amount of ablation measured in the test of laser cutting a 1mm thick Q235 sheet. The process parameters taken by each number are the process parameters of the corresponding numbers in table 4.
TABLE 6 cutting test data of sharp corners of steel plates of 1mm
Figure BDA0003065383880000172
And 3. Training the BP neural network. Before training of the neural network model, super parameters of the model are set, the learning rate represents the updating magnitude of the weight of each iteration, the convergence rate is slow due to too small learning rate, the model is not converged due to too large learning rate, and the comprehensive consideration of the learning rate is proper to be 0.05; the error threshold value of the neural network represents the fitting degree of the model and the training data, the error threshold value is too large to cause under-fitting, the error threshold value is too small to cause over-fitting, and the error threshold value is selected to be 0.5 to be more suitable in comprehensive consideration; hidden layer of neural networkThe convergence performance of the model can be influenced by the value of the number of the element nodes, and tests show that the number of the element nodes in the hidden layer is 13 and is more suitable; weight omega of comprehensive evaluation index 1 ＝ω 2 ＝ω 3 ＝ω 4 =0.25. And training a neural network by using data obtained by the orthogonal test to obtain a statistical model reflecting laser cutting process parameters and the shape of the workpiece and the comprehensive quality of laser cutting.
4. And optimizing process parameters by using a genetic algorithm. In the scheme, the result obtained by BP neural network training is a fitness function of a genetic algorithm and is used for optimizing the laser cutting process parameters of the aircraft engine lock plate (1 mm thick). As shown in fig. 4, the locking plate is divided into 11 parts, which are represented by numbers 1 to 11, wherein except that the part represented by the number 8 is an arc, the part represented by the number 11 is a full circle, the rest parts are straight lines, the joint of the adjacent straight lines may be a sharp corner, the sharp corners of the locking plate can be divided into 2 types, the sharp corners between the second straight line 2 and the third straight line 3, and the sharp corners between the sixth straight line 6 and the seventh straight line 7 are a type, which is called a sharp corner 1, and the angle of the sharp corner is 90 degrees and the arc of the sharp corner is 0mm; the sharp angle between the fourth line 4 and the fifth line 5 is another type, called sharp angle 2, with a sharp angle of 60 ° and a sharp arc of 0.5mm. Research shows that when the diameter of the circular arc exceeds half of the plate thickness, the circular arc can be regarded as a straight line in the actual cutting process, and therefore, the circular arc and the whole circular part of the locking plate can be regarded as straight line cutting in the laser cutting process. Thus, in this case, the genetic algorithm comprises a total of 3 objective functions f 1 、f 2 、f 3 Wherein an objective function f describing the cutting quality during the linear cutting phase 1 As shown in the following formula:
Figure BDA0003065383880000181
objective function f describing cutting quality of sharp-corner 1 cutting stage 2 Comprises the following steps:
Figure BDA0003065383880000182
objective function f describing cutting quality of sharp-corner 2 cutting stage 3 Comprises the following steps:
Figure BDA0003065383880000183
the range of technological parameters of the ZT-J-6060M metal laser cutting machine is shown in a table 7, and the value ranges of laser power, cutting speed, gas pressure, defocusing amount, duty ratio and pulse frequency are listed in the table. From these ranges, constraints of a laser cut optimization model of an aircraft engine locking tab can be determined.
TABLE 7 ZT-J-6060M Process parameter Range Table
Figure BDA0003065383880000184
Therefore, the laser cutting optimization model of the aircraft engine locking plate is as follows:
Figure BDA0003065383880000185
0≤x 4 ≤500
0≤x 5 ≤8000
0≤x 6 ≤5
0≤x 7 ≤2
0≤x 8 ≤1
0≤x 9 ≤5000
the basic parameter settings of the genetic algorithm are as follows: the binary code length is 8 bits, the population size M =500, the cross probability Pc =0.8, the variation probability Pm =0.05, and the iteration number is 1000. According to the method of the embodiment, positive and negative ideal points are obtained, which are respectively: (8.2, 8, 7.5), (7.6, 5.9, 6), the process parameters corresponding to the chromosome closest to the ideal point are shown in Table 8, and the laser power, cutting speed, gas pressure, defocus amount, duty ratio and pulse frequency are 348W,2535mm/min,2bar,1.2mm,7.2% and 5000Hz, respectively. The measured quality of the locking plate processed by the process parameters listed in Table 8 is shown in Table 9, the cutting quality of the straight line part is that the width of a cutting seam is 0.18mm, the surface roughness is 3 mu m, the width of a heat affected zone is 115mm, and the length of adhering slag is 0mm; the ablation of the sharp corner 1 is 0mm and the ablation of the sharp corner 2 is 0.8mm.
TABLE 8 Process parameter Table corresponding to chromosome with the greatest relative closeness
Figure BDA0003065383880000191
Table 9 lock plate actual measurement quality table
Figure BDA0003065383880000192
As can be seen from the comparison of tables 3, 6 and 9, the cutting seam width, the surface roughness, the heat affected zone width, the slag adhering length and the sharp angle ablation amount of the locking plate obtained in the table 9 in the actually measured processing quality are all superior, which indicates that the process parameter optimization method considering the sharp angle ablation is effective in optimizing the laser cutting process parameters and obtaining better cutting quality.
The above description is only a preferred embodiment of the present invention, and all equivalent changes and modifications made in accordance with the claims of the present invention should be covered by the present invention.

Claims (8)

1. A method for optimizing parameters of a plate laser cutting process considering sharp corner ablation is characterized by comprising the following steps: the method comprises the following steps:
step S1: performing linear cutting and sharp corner cutting orthogonal tests on plates with different thicknesses, measuring workpiece quality indexes corresponding to each set of tests, wherein the linear cutting quality indexes comprise joint cutting width, section roughness, heat affected zone width and slag adhering length, and the sharp corner cutting quality index is sharp corner ablation amount, and then calculating a comprehensive quality index according to the measured indexes;
step S2: the method comprises the steps of taking plate laser cutting process parameters and workpiece shape characteristics as input data, taking laser cutting quality comprehensive evaluation indexes as output, and constructing a thin plate laser cutting quality BP neural network model; the plate laser cutting process parameters comprise laser power, gas pressure, cutting speed, defocusing amount, duty ratio and pulse frequency; the workpiece shape characteristics comprise plate thickness, a sharp angle and a sharp arc;
and step S3: the neural network model trained in the step S2 is used as a fitness function of the genetic algorithm, the genetic algorithm is applied to optimize the laser cutting quality comprehensive evaluation index value, and when the optimal laser cutting quality comprehensive evaluation index value is obtained, the laser cutting process parameter corresponding to the model is obtained, so that the optimization of the laser cutting process parameter is realized;
the step S1 specifically includes the steps of:
step S11: performing a linear laser cutting orthogonal test on plates with different thicknesses, wherein the test factors comprise laser power, cutting speed, gas pressure, duty ratio, pulse frequency and defocusing amount, and the kerf width measured by the nth test is set as w n Roughness of tangent plane is r n The heat-affected zone is ha n And the length of attached slag is lg n ；
Step S12: performing a sharp corner cutting orthogonal test on plates with different thicknesses, wherein the test factors comprise laser power, cutting speed, gas pressure, a sharp corner angle and a sharp corner arc, and setting the ablation amount of the sharp corner measured in the nth test as H n ；
Step S13: calculating the comprehensive quality index of linear cutting;
step S14: calculating the comprehensive quality index Sc of the cutting data of the nth sharp corner n The method specifically comprises the following steps:
Figure FDA0004032763920000021
wherein max (H) and min (H) are respectively the maximum value and the minimum value in the kerf width data;
step S15: and (3) setting the comprehensive quality evaluation index of the straight line and the sharp angle as a neural network tag set D = [ S Sc ], wherein S represents the comprehensive quality evaluation index with a Kini penalty term.
2. The method for optimizing parameters of the laser cutting process of the plate material considering the sharp corner ablation, according to claim 1, is characterized in that: the step S13 specifically includes the following steps:
step S131: calculating the value wz of the kerf width in the nth cutting quality index data after MaxMin normalization n ：
Figure FDA0004032763920000022
Wherein max (w) and min (w) are respectively the maximum value and the minimum value in the kerf width data;
step S132: calculating the value rz of the tangent plane roughness normalized by MaxMin in the nth cutting quality index data n ：
Figure FDA0004032763920000023
Wherein max (r) and min (r) are respectively the maximum value and the minimum value in the section roughness data;
step S133: calculating the value haz of the thermal influence width normalized by MaxMin in the nth cutting quality index data n ：
Figure FDA0004032763920000024
/>
Wherein max (ha) and min (ha) are respectively the maximum value and the minimum value in the kerf width data;
step S134: calculating the value lgz of the slag adhering length in the nth cutting quality index data after the normalization of MaxMin n ：
Figure FDA0004032763920000031
Wherein max (lg) and min (lg) are respectively the maximum value and the minimum value in the slag adhering length data;
step S135: defining the kini index C as:
Figure FDA0004032763920000032
wherein, when i =1, the input signal is converted into a signal,
Figure FDA0004032763920000033
when the ratio of i =2, the ratio of the total of the number of the bits is set to be lower than the value of i =2,
Figure FDA0004032763920000034
when the ratio of i =3, the ratio of the total of the number of the bits is set to be lower than the value of i =3,
Figure FDA0004032763920000035
when the ratio of the sum of the i =4,
Figure FDA0004032763920000036
step S136: calculating a comprehensive quality evaluation index S with a Gini penalty term:
Figure FDA0004032763920000037
wherein, ω is 1 -ω 4 Is a weight coefficient, and ω 1 +ω 2 +ω 3 +ω 4 =1; k is a cut-through coefficient, when the plate is cut through, K is 0, the index value is 0, when the plate can be cut through, K is 1, the index value is larger than 1.87, and the larger the index value is, the better the comprehensive cutting quality is; the index can be used for making global judgment on the quality of the linear cutting laser and punishing the phenomenon of two-stage differentiation of index values.
3. The method for optimizing parameters of the laser cutting process of the plate material considering the sharp corner ablation, according to claim 1, is characterized in that: the step S2 specifically includes the following steps:
step S21: setting the input vector of the BP neural network model as follows:
X＝(x 1 ,x 2 ,...,x 9 )
wherein x is 1 -x 9 The method comprises the following steps that 9 input nodes of a neural network respectively receive input of plate thickness, sharp angle angles, sharp angle arcs, laser power, cutting speed, auxiliary gas pressure, defocusing amount, duty ratio and pulse frequency data;
step S22: randomly generating weight matrix between k hidden layer neuron node and input layer
Figure FDA0004032763920000041
And a weight matrix between the hidden layer and the output layer->
Figure FDA0004032763920000042
Step S23: let the input vector of the kth hidden layer neuron node:
Figure FDA0004032763920000043
h 1 -h 9 a weighting result representing output values of the 9 input layer nodes;
step S24: setting output vectors of hidden layer neuron nodes: z = (Z) 1 ,z 2 ,...,z k ) K number of hidden layer neuron nodes; z is a radical of formula k An output value for each node;
step S25: setting the activation function to a linear rectification function:
Figure FDA0004032763920000044
step S26: setting a learning rate eta (eta is more than 0) and an error threshold E (E is more than 0) of the neural network, wherein the learning rate represents the learning rate, and the error threshold represents the fitting degree of the model;
step S27: calculating the output Y of the neural network:
Figure FDA0004032763920000045
step S28: calculating the error e of the output value of the neural network and the comprehensive evaluation index value of the real laser cutting quality:
Figure FDA0004032763920000051
where n is the number of data, D i Labeling the ith piece of data in the set for the neural network;
step S29: stopping training if the training error E is smaller than the error threshold E, and taking the obtained neural network model as a fitness function of the genetic algorithm in the step S3, otherwise continuing training according to the step S210;
step S210: calculating an error factor delta of the ith output layer i ：
Figure FDA0004032763920000052
Step S211: error factor delta of k-th input layer k ：
Figure FDA0004032763920000053
Step S212: updating a weight matrix between the hidden layer and the output layer according to the following formula:
Figure FDA0004032763920000054
step S213: updating the weight matrix between the input layer and the hidden layer according to the following formula:
Figure FDA0004032763920000055
step S214: the process returns to step S27.
4. The method for optimizing parameters of the laser cutting process of the plate material considering the sharp corner ablation, according to claim 1, is characterized in that: the step S3 specifically includes the following steps:
step S31: determining a parameter value range;
step S32: the optimization model is established as follows:
Figure FDA0004032763920000061
x 4min ≤x 4 ≤x 4max
x 5min ≤x 5 ≤x 5max
x 6min ≤x 6 ≤x 6max
x 7min ≤x 7 ≤x 7max
x 8min ≤x 8 ≤x 8max
x 9min ≤x 9 ≤x 9max
wherein, f n Represents the nth objective function, F represents the trained neural network model, u n 、v n 、w n Respectively representing the plate thickness, the sharp angle and the sharp angle arc of the cutting stage corresponding to the nth objective function;
step S33: setting binary coding length l, population size M, cross probability Pc, variation probability Pm and iteration times T;
step S34: solving each objective function f separately 1 To f n The maximum value of (a);
step S35: solving each objective function f separately 1 To f n Minimum value of (d);
step S36: and solving the optimal process parameters by using a TOPSIS method.
5. The method for optimizing parameters of the laser cutting process of the plate in consideration of sharp corner ablation, as claimed in claim 4, wherein the method comprises the following steps: the step S31 specifically includes the following steps:
step S311: the value range of the laser cutting power is as follows:
x 4min ≤x 4 ≤x 4max
step S312: the laser cutting speed value range is as follows:
x 5min ≤x 5 ≤x 5max
step S313: the auxiliary gas pressure range, namely:
x 6min ≤x 6 ≤x 6max
step S314: the defocus measurement value range is as follows:
x 7min ≤x 7 ≤x 7max
step S315: the duty cycle value range, namely:
x 8min ≤x 8 ≤x 9max
step S316: the range of the pulse frequency is as follows:
x 9min ≤x 9 ≤x 9max 。
6. the method for optimizing parameters of the laser cutting process of the plate in consideration of sharp corner ablation, as claimed in claim 4, wherein the method comprises the following steps: in the step S34, any one objective function f is processed n The method comprises the following steps:
step S341: initializing a population; a mode of randomly generating an initial population is adopted, namely, the value of the gene on the chromosome is randomly selected from 0 and 1;
step S342: calculating the fitness; after the initialized population is decoded, substituting the decoded initialized population into a BP neural network, and taking an output value F of the BP neural network as fitness; selecting individuals with highest fitness from the population, directly inheriting the individuals to the next generation, and performing selection, crossing and mutation operations on the rest individuals;
step S343: selecting operation; generating a new chromosome by adopting a roulette selection mode, wherein the roulette selection mode is related to the fitness value of the chromosome, and the probability of selection is high when the fitness value of the chromosome is larger;
step S344: performing cross operation; random crossing operation is adopted, namely two individuals are selected randomly without putting back, and one allele is selected randomly as a crossing point to carry out crossing;
step S345: performing variation; adopting random variation operation, namely selecting an individual and generating a random number for the individual, and if the number is less than the variation probability Pc, randomly selecting a gene of the individual to negate the gene;
step S346: repeating the steps S342 to S345 until the algebra reaches the set iteration number T;
step S347: for the objective function f 1 To f n The maximum objective function values are respectively marked as F 1 + To
Figure FDA0004032763920000081
7. The method for optimizing parameters of the laser cutting process of the plate in consideration of sharp corner ablation, as claimed in claim 4, wherein the method comprises the following steps: the step S35 specifically includes the following steps:
step S351: for any one objective function f n Taking the negative number thereof, namely-F, as a fitness function and iterating according to the steps S341 to S346;
step S352: for the objective function f 1 To f n And the minimum objective function values thereof are respectively marked as F 1 - To
Figure FDA0004032763920000082
8. The method for optimizing parameters of the laser cutting process of the plate material in consideration of sharp corner ablation, as claimed in claim 4, wherein: the step S36 specifically includes the following steps:
step S361: the distances from the individual to the positive and negative ideal points are respectively set as
Figure FDA0004032763920000083
Figure FDA0004032763920000084
Figure FDA0004032763920000085
Wherein, the first and the second end of the pipe are connected with each other,
Figure FDA0004032763920000086
a jth objective function value for a kth individual;
step S362: let the relative closeness d of an individual to an ideal point:
Figure FDA0004032763920000087
step S363: repeating the step S341 to the step S346 by taking the d as a fitness function for iteration;
step S364: and taking the individual with the maximum fitness in all the individuals in the T generation as the optimal process parameter individual.
CN202110525071.3A 2021-05-14 2021-05-14 Plate laser cutting process parameter optimization method considering sharp corner ablation Active CN113182709B (en)

Priority Applications (1)

Application Number Priority Date Filing Date Title
CN202110525071.3A CN113182709B (en) 2021-05-14 2021-05-14 Plate laser cutting process parameter optimization method considering sharp corner ablation

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
CN202110525071.3A CN113182709B (en) 2021-05-14 2021-05-14 Plate laser cutting process parameter optimization method considering sharp corner ablation

Publications (2)

Publication Number Publication Date
CN113182709A CN113182709A (en) 2021-07-30
CN113182709B true CN113182709B (en) 2023-04-07

Family

ID=76981887

Family Applications (1)

Application Number Title Priority Date Filing Date
CN202110525071.3A Active CN113182709B (en) 2021-05-14 2021-05-14 Plate laser cutting process parameter optimization method considering sharp corner ablation

Country Status (1)

Country Link
CN (1) CN113182709B (en)

Families Citing this family (2)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN114297802A (en) * 2021-12-31 2022-04-08 福州大学 Multi-objective optimization method for laser cutting technological parameters of thin plate
CN114619151A (en) * 2022-01-20 2022-06-14 大族激光科技产业集团股份有限公司 Laser processing control method and device and readable storage medium

Family Cites Families (5)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
IL138926A0 (en) * 2000-10-06 2001-11-25 Notal Vision Ltd Method and system for detecting eye disease
CN107755912B (en) * 2017-10-26 2019-08-30 重庆科技学院 Tubing all positon laser-arc hybrid welding in industry system
JP6975190B2 (en) * 2019-02-26 2021-12-01 ファナック株式会社 Machine learning equipment, laser processing equipment and laser processing system
CN110722285B (en) * 2019-09-30 2021-10-22 湖南科技大学 Laser hot wire welding seam forming quality prediction method, system and medium
CN111625997B (en) * 2020-05-27 2024-01-19 山东天岳先进科技股份有限公司 Crystal cutting process optimization method and device

Also Published As

Publication number Publication date
CN113182709A (en) 2021-07-30

Similar Documents

Publication Publication Date Title
CN113182709B (en) Plate laser cutting process parameter optimization method considering sharp corner ablation
CN109508488B (en) Shot peening forming process parameter prediction method based on genetic algorithm optimization BP neural network
Gaitonde et al. Minimizing burr size in drilling using artificial neural network (ANN)-particle swarm optimization (PSO) approach
Fenggou et al. The study of high efficiency and intelligent optimization system in EDM sinking process
Kumar et al. GA-based optimisation using RSM in WEDM of Nimonic-90: a nickel-based super alloy
CN114297802A (en) Multi-objective optimization method for laser cutting technological parameters of thin plate
CN114897227A (en) Multi-steel-grade mechanical property forecasting method based on improved random forest algorithm
Wu et al. Optimal shape design of an extrusion die using polynomial networks and genetic algorithms
CN101114312A (en) Method for designing ASSEL roll profile based on neurotransmission network technique
Madić et al. Mathematical modelling of the CO2 laser cutting process using genetic programming
CN116776922A (en) Optimizing method for TC4 high-speed milling process parameters
Nayak et al. Parametric appraisal of WEDM using harmony search algorithm
Dutta et al. Optimum process parameters for efficient and quality thin wall machining using firefly algorithm
CN111310884A (en) Optimal layout method of wind turbine generator based on data-driven evolutionary algorithm
Srinivas et al. Application of MQL for developing sustainable EDM and process parameter optimisation using ANN and GRA method
Ren et al. Modeling and process parameter optimization of laser cutting based on artificial neural network and intelligent optimization algorithm
Nayak et al. An intelligent approach for multi-response optimisation of WEDM parameters
Karnik et al. Development of artificial neural network models to study the effect of process parameters on burr size in drilling
CN115994619A (en) Intelligent generation method for part machining process route
CN110334442A (en) A kind of Cutting parameters prediction technique for processing TC4 titanium alloy workpiece
Tiwary et al. Laser Beam Micromarking on Inconel 625 Superalloy.
Dubey A hybrid approach for multi-performance optimization of the electro-chemical honing process
Sen et al. Optimal selection of machining conditions in the electrojet drilling process using hybrid NN-DF-GA approach
CN114003003A (en) Technological parameter optimization and stability control method in laser cladding process
Karnik et al. Integrating Taguchi principle with genetic algorithm to minimize burr size in drilling of AISI 316L stainless steel using an artificial neural network model

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
GR01 Patent grant
GR01 Patent grant