CN114065631A - Energy consumption prediction method and system for laser cutting of plate - Google Patents
Energy consumption prediction method and system for laser cutting of plate Download PDFInfo
- Publication number
- CN114065631A CN114065631A CN202111369000.5A CN202111369000A CN114065631A CN 114065631 A CN114065631 A CN 114065631A CN 202111369000 A CN202111369000 A CN 202111369000A CN 114065631 A CN114065631 A CN 114065631A
- Authority
- CN
- China
- Prior art keywords
- energy consumption
- laser cutting
- data
- consumption prediction
- sparks
- 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.)
- Granted
Links
- 238000005265 energy consumption Methods 0.000 title claims abstract description 81
- 238000003698 laser cutting Methods 0.000 title claims abstract description 73
- 238000000034 method Methods 0.000 title claims abstract description 59
- 238000012549 training Methods 0.000 claims abstract description 51
- 238000004422 calculation algorithm Methods 0.000 claims abstract description 42
- 238000012795 verification Methods 0.000 claims abstract description 23
- 238000012545 processing Methods 0.000 claims abstract description 21
- 238000005457 optimization Methods 0.000 claims abstract description 10
- 238000010200 validation analysis Methods 0.000 claims abstract description 4
- 238000004880 explosion Methods 0.000 claims description 23
- 230000008569 process Effects 0.000 claims description 17
- 238000005520 cutting process Methods 0.000 claims description 15
- 238000005070 sampling Methods 0.000 claims description 13
- 238000004590 computer program Methods 0.000 claims description 12
- 230000002349 favourable effect Effects 0.000 abstract 1
- 238000004519 manufacturing process Methods 0.000 description 11
- 238000010586 diagram Methods 0.000 description 10
- 238000012360 testing method Methods 0.000 description 10
- 239000011159 matrix material Substances 0.000 description 9
- 238000003754 machining Methods 0.000 description 7
- 238000005516 engineering process Methods 0.000 description 6
- 230000006870 function Effects 0.000 description 6
- 238000004364 calculation method Methods 0.000 description 5
- 238000010801 machine learning Methods 0.000 description 5
- 238000013528 artificial neural network Methods 0.000 description 4
- 238000013461 design Methods 0.000 description 4
- 239000007789 gas Substances 0.000 description 4
- 239000000463 material Substances 0.000 description 4
- 230000000717 retained effect Effects 0.000 description 4
- 238000011161 development Methods 0.000 description 3
- 238000003860 storage Methods 0.000 description 3
- 238000012546 transfer Methods 0.000 description 3
- 230000009286 beneficial effect Effects 0.000 description 2
- 230000008859 change Effects 0.000 description 2
- 230000007613 environmental effect Effects 0.000 description 2
- 238000002474 experimental method Methods 0.000 description 2
- 238000012986 modification Methods 0.000 description 2
- 230000004048 modification Effects 0.000 description 2
- 230000004913 activation Effects 0.000 description 1
- QVGXLLKOCUKJST-UHFFFAOYSA-N atomic oxygen Chemical compound [O] QVGXLLKOCUKJST-UHFFFAOYSA-N 0.000 description 1
- 230000008901 benefit Effects 0.000 description 1
- 230000001808 coupling effect Effects 0.000 description 1
- 238000002790 cross-validation Methods 0.000 description 1
- 238000007418 data mining Methods 0.000 description 1
- 230000007547 defect Effects 0.000 description 1
- 238000009826 distribution Methods 0.000 description 1
- 230000006872 improvement Effects 0.000 description 1
- 230000003993 interaction Effects 0.000 description 1
- 238000002955 isolation Methods 0.000 description 1
- 238000013178 mathematical model Methods 0.000 description 1
- 239000002184 metal Substances 0.000 description 1
- 230000003287 optical effect Effects 0.000 description 1
- 239000001301 oxygen Substances 0.000 description 1
- 229910052760 oxygen Inorganic materials 0.000 description 1
- 238000004080 punching Methods 0.000 description 1
- 230000005855 radiation Effects 0.000 description 1
- 230000009467 reduction Effects 0.000 description 1
- 238000011160 research Methods 0.000 description 1
- 238000004088 simulation Methods 0.000 description 1
- 238000010998 test method Methods 0.000 description 1
Images
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F30/00—Computer-aided design [CAD]
- G06F30/20—Design optimisation, verification or simulation
- G06F30/27—Design optimisation, verification or simulation using machine learning, e.g. artificial intelligence, neural networks, support vector machines [SVM] or training a model
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F30/00—Computer-aided design [CAD]
- G06F30/20—Design optimisation, verification or simulation
- G06F30/28—Design optimisation, verification or simulation using fluid dynamics, e.g. using Navier-Stokes equations or computational fluid dynamics [CFD]
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/04—Architecture, e.g. interconnection topology
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/08—Learning methods
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/12—Computing arrangements based on biological models using genetic models
- G06N3/126—Evolutionary algorithms, e.g. genetic algorithms or genetic programming
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F2113/00—Details relating to the application field
- G06F2113/08—Fluids
-
- 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
- G06F2119/14—Force analysis or force optimisation, e.g. static or dynamic forces
Landscapes
- Engineering & Computer Science (AREA)
- Physics & Mathematics (AREA)
- Theoretical Computer Science (AREA)
- General Physics & Mathematics (AREA)
- Evolutionary Computation (AREA)
- Life Sciences & Earth Sciences (AREA)
- Biophysics (AREA)
- Health & Medical Sciences (AREA)
- General Engineering & Computer Science (AREA)
- Software Systems (AREA)
- Mathematical Physics (AREA)
- Artificial Intelligence (AREA)
- Computing Systems (AREA)
- General Health & Medical Sciences (AREA)
- Data Mining & Analysis (AREA)
- Molecular Biology (AREA)
- Computational Linguistics (AREA)
- Biomedical Technology (AREA)
- Computer Hardware Design (AREA)
- Geometry (AREA)
- Evolutionary Biology (AREA)
- Bioinformatics & Computational Biology (AREA)
- Bioinformatics & Cheminformatics (AREA)
- Physiology (AREA)
- Fluid Mechanics (AREA)
- Mathematical Analysis (AREA)
- Mathematical Optimization (AREA)
- Pure & Applied Mathematics (AREA)
- Algebra (AREA)
- Genetics & Genomics (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Medical Informatics (AREA)
- Laser Beam Processing (AREA)
- Management, Administration, Business Operations System, And Electronic Commerce (AREA)
Abstract
The invention relates to a method and a system for predicting energy consumption of plate laser cutting, wherein the method comprises the following steps: (1) dividing laser cutting energy consumption data corresponding to the technological parameters of the laser processing equipment into a training set and a verification set; (2) by passingKThe cross-turn verification method divides the training set intoKA sub-training set; (3) extreme learning machine using firework algorithmThe input weight and the threshold value are optimized to obtain a laser cutting energy consumption prediction model based on a firework algorithm optimization extreme learning machine, and then the sub-training sets are respectively input into the laser cutting energy consumption prediction model to be trained to obtainKA trained laser cutting energy consumption prediction model; (4) computing the by a validation setKAnd (3) the accuracy of the trained laser cutting energy consumption prediction model, and selecting the model with the highest accuracy for laser cutting energy consumption prediction. The method and the system are favorable for accurately predicting the energy consumption of laser cutting.
Description
Technical Field
The invention belongs to the technical field of production and manufacturing, and particularly relates to a method and a system for predicting energy consumption of plate laser cutting.
Background
With the development of scientific technology and economic society, machine tools have become the main production tools in manufacturing industry. The machine tool brings high income to enterprises, and meanwhile, the energy consumption caused by the machine tool is not negligible. Existing studies have shown that more than 90% of the environmental impact of the manufacturing industry is due to the electrical energy consumption of the machine tool during the machining service phase. The machine tool in China is the first global, but the energy utilization rate is lower than 30%. The huge power consumption of the machine tool and the environmental emission problem are increasingly paid attention internationally, the reduction of the energy consumption of machine tool processing can save a large amount of energy for manufacturing enterprises, and the negative influence of the production and manufacturing process of products on the environment can be reduced. Through prediction of machine tool energy consumption, design characteristics with high machining energy consumption can be found in advance in the design stage of a product, and improvement and optimization of product structural design from a design source are guided. In the whole production link of the machine tool, a cutting stage is an important stage for generating the appearance of a product, and the energy consumed in the stage is determined by the combination of machining parameters. 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. Thanks to the rapid development of computer technology, information technology and network technology have been widely used in production plants, and digitization has become the main direction of industrial development in the future. Therefore, the method for predicting the energy consumption in the laser cutting process is a good entry point for building a digital energy-saving and environment-friendly workshop.
In the laser machining process, the interaction between the laser and the material is a complex thermophysical process, and a series of problems including conduction heat transfer, convection heat transfer, radiant heat transfer, hydrodynamics and the like are involved, so that it is difficult to accurately describe the cutting process by using a specific mathematical model. And a result-oriented data driving mode is adopted, a simulation prediction model is established by combining a machine learning algorithm from historical available data, and a precise and complete laser cutting energy consumption prediction model can be obtained under the condition of not clearly analyzing energy flow in the machining process.
At present, two main technologies related to laser cutting energy consumption prediction are available, and firstly, the laser cutting energy consumption is calculated according to a substitute processing diagram and processing technological parameters by establishing an energy consumption database, so that the total laser processing energy consumption is predicted in advance. The accuracy of the method greatly depends on the accuracy of the energy consumption database, and in the actual laser cutting process, the parameters often have coupling property and time-varying property, so that the laser cutting energy consumption prediction based on the method is not accurate. Meanwhile, the method only can predict the processing energy consumption, but cannot optimize the processing energy consumption, so that the method has great limitation. And secondly, a laser cutting energy consumption prediction method based on a machine learning algorithm can learn the change of data in the laser cutting process through the machine learning algorithm, and the prediction result is more accurate than that of the first method. At present, the second method is relatively few in research on prediction of laser cutting energy consumption, so that the invention provides an energy consumption prediction method based on a machine learning algorithm.
In conclusion, how to simply and accurately predict the machine tool energy consumption in the cutting stage according to the processing technological parameters of the laser cutting machine tool has important guiding significance for researching how to optimize the laser cutting technological parameters to reduce the machine tool energy consumption, and the method becomes a technical problem which needs to be solved urgently by technical personnel in the field.
Disclosure of Invention
The invention aims to provide a method and a system for predicting energy consumption of plate laser cutting, which are beneficial to accurately predicting the energy consumption of laser cutting.
In order to achieve the purpose, the invention adopts the technical scheme that: a method for predicting energy consumption of plate laser cutting comprises the following steps:
(1) dividing laser cutting energy consumption data corresponding to laser processing equipment process parameters into a training set CtAnd a verification set Cv；
(2) Dividing training set C by K-fold cross verification methodtFor K sub-training sets Ct1,Ct2,…,Ctk；
(3) Optimizing the input weight and the threshold value of the extreme learning machine by using a firework algorithm to obtain a laser cutting energy consumption prediction model for optimizing the extreme learning machine based on the firework algorithm, and then performing a sub-training set Ct1,Ct2,…,CtkRespectively inputting the laser cutting energy consumption prediction models to train the laser cutting energy consumption prediction models to obtain K trained laser cutting energy consumption prediction models M1,M2,…,Mk；
(4) Pass verification set CvCalculating the K trained laser cutting energy consumption prediction models M1,M2,…,MkAnd selecting the model with the highest accuracy for predicting the energy consumption of laser cutting.
Further, the laser processing equipment process parameters comprise laser power, cutting speed, gas pressure, pulse frequency and focal position.
Further, the step (1) specifically includes the steps of:
(1.1) initializing training set CtAnd a verification set CvAre empty respectively;
(1.2) carrying out N times of putting back sampling on the data, wherein N is the number of data strips; only one piece of data is extracted in each sampling, and the extracted data is added into a training set Ct；
(1.3) after N times of sampling with putting back, putting the data which are not extracted at one time into a verification set Cv。
Further, the step (2) specifically includes the following steps:
(2.1) training set CtDividing into K sub-training sets, K > 1, each containing at most
Bar data, NtFor training set CtThe number of the data pieces is equal to or greater than the number of the data pieces,
is a rounded up symbol;
(2.2) recording the K sub-training sets in (2.1) as Ct1,Ct2,…,Ctk。
Further, the step (3) specifically includes the following steps:
(3.1) initializing a weight value and a threshold value of the extreme learning machine, and setting the group scale Q, the number R of variant sparks, the iteration number counter t which is 0 and the maximum iteration number Iter _ Max of the firework algorithm; setting data in an extreme learning machine, wherein the data mainly comprises population size and iteration times;
(3.2) setting an initial group of the firework algorithm, and calculating the fitness value of the individuals in the group: respectively substituting the weight and the threshold value corresponding to each individual in the initial population into the extreme learning machine, training through a training set, calculating the prediction accuracy of the trained extreme learning machine model by using a verification set, and calculating the individual fitness value according to the prediction accuracy;
(3.3) generating an explosion spark, introducing a Gaussian variation strategy, and generating a variation spark: firstly, calculating the radius and the number of explosion sparks of each firework individual in a group, and generating explosion sparks; then, randomly selecting R fireworks, and respectively generating R variable sparks by adopting a Gaussian variation strategy; calculating an individual fitness value for each explosion spark and each variation spark generated;
(3.4) selecting Q fireworks from the fireworks, the explosion sparks and the variant sparks as next-generation fireworks according to a selection strategy: firstly, according to individual fitness value, selecting one with the maximum fitness from the candidate groups to enter the next-generation firework group, and then selecting Q-1 fireworks, explosion sparks or variant sparks from the candidate groups to enter the next-generation firework group by adopting a roulette selection strategy based on the fitness value;
(3.5) termination judgment: if the iteration times t is larger than the Iter _ Max, terminating the iteration and outputting the optimal individual and the weight value and the threshold value corresponding to the optimal individual; otherwise, t ← t +1, return to step 3.3; after iteration is completed, a laser cutting energy consumption prediction model based on the firework algorithm optimization extreme learning machine is obtained;
(3.6) sub-training set Ct1,Ct2,…,CtkInputting the data into the laser cutting energy consumption prediction model to train the laser cutting energy consumption prediction model to obtain K trained laser cutting energy consumption prediction models M1,M2,…,Mk；
(3.7) inputting validation set data and calculating the K trained laser cutting energy consumption prediction models M1,M2,…,MkThe accuracy of (2).
The invention also provides a system for predicting energy consumption for laser cutting of a sheet material, comprising a memory, a processor and computer program instructions stored on the memory and executable by the processor, wherein the computer program instructions, when executed by the processor, enable the method steps as set forth in the claims.
Compared with the prior art, the invention has the following beneficial effects:
1. a large number of network training parameters need to be manually set in a traditional feedforward neural network (such as a BP neural network), the number of the nodes of an implicit layer is only the parameters needing to be manually set in an extreme learning machine algorithm, manual parameter adjustment is not needed in the algorithm execution process, the process of repeated iteration of the traditional training algorithm is avoided, the fast convergence is achieved, the training time is greatly shortened, the obtained solution is the only optimal solution, and the generalization capability of the network is guaranteed. However, the input layer weight and the threshold value randomly generated by the extreme learning machine have randomness, which may cause that the output matrix of the hidden layer is not a column full-rank matrix, which may reduce the performance of the extreme learning machine and have a large influence on the model prediction accuracy. As a group intelligent algorithm following an optimal solution, the firework algorithm has strong local optimization and global optimization capabilities. According to the energy consumption prediction method for laser cutting of the plate, provided by the invention, the electrical parameters (three-phase power) and the process parameters (laser power, cutting speed, gas pressure, pulse frequency and focal position) of the laser cutting machine during operation are taken as training data together, aiming at the defect that the randomness of the parameters of the extreme learning machine has great influence on the accuracy rate of a prediction model, the parameters of the extreme learning machine are optimized by adopting a firework algorithm, the laser cutting energy consumption prediction model based on the firework algorithm optimized extreme learning machine is established, and the prediction precision and robustness of the model are effectively improved.
2. The method is based on historical available data, combines a machine learning algorithm, constructs a machining energy consumption model of the laser cutting machine tool, obtains an energy consumption prediction model with high prediction precision under the condition that energy flow in the machining process does not need to be clearly analyzed, and establishes conditions for further machine tool energy consumption optimization.
Drawings
FIG. 1 is a schematic diagram of a laser cutting energy consumption prediction model for optimizing an extreme learning machine network based on a firework algorithm in an embodiment of the invention;
FIG. 2 is a macroscopic view of a cut sheet in an embodiment of the present invention;
FIG. 3 is a graph of a portion of an electrical parameter signal collected in an embodiment of the present invention.
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 example 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 embodiment provides a method for predicting energy consumption of laser cutting of a plate based on a firework algorithm optimized extreme learning machine network, which includes the following steps:
(1) dividing the technological parameters of the laser processing equipment and the laser cutting energy consumption data (hereinafter referred to as data) corresponding to the technological parameters into a training set CtAnd a verification set CvThe method comprises the following steps:
(1.1) initializing training set CtAnd a verification set CvAre each empty.
(1.2) carrying out N times of putting back sampling on the data, wherein N is the number of data strips, and N is more than 100. Only one piece of data is extracted in each sampling, and the extracted data is added into a training set Ct。
(1.3) after N times of sampling with putting back, putting the data which are not extracted at one time into a verification set Cv. In all energy consumption data, if a certain piece of data is subjected to N times of sampling with playback, the probability P that the data is not sampled at one timenComprises the following steps:
wherein when the sampling number N tends to be infinite, the expected E which is not sampled at one time∞Comprises the following steps:
therefore, when the sampling times are large enough, the method can ensure that about 36 percent of data is not sampled at one time, and the training set C is ensuredtAnd a verification set CvThe isolation of (2) and the verification of the generalization ability of the model are good.
Wherein the laser processing equipment process parameters comprise laser power, cutting speed, gas pressure, pulse frequency and focus position.
(2) Dividing training set C by K-fold cross verification methodtFor K sub-training sets Ct1,Ct2,…,CtkThe method comprises the following specific steps:
(2.1) training set CtDividing into K (K > 1) sub-training sets, each of which contains at most
Bar data, NtFor training set CtThe number of the data pieces is equal to or greater than the number of the data pieces,
is rounding up the symbol.
(2.2) recording the K sub-training sets in (2.1) as Ct1,Ct2,…,Ctk。
(3) The specific steps of the Fireworks Algorithm (FWA) are as follows:
(3.1) setting firework algorithm parameters, wherein the group scale is Q, the maximum iteration number is Iter _ Max, the number of variant sparks is R, and an iteration number counter t is 0.
(3.2) randomly generating a plurality of fireworks in a specific solution space, each fireworks representing a solution in the solution space according to the fitness function f (X)i) The fitness value of each firework is calculated.
And (3.3) generating sparks in the radiation space of the fireworks according to the actual fireworks properties and by combining the actual situation of the search problem. Each spark represents one solution in the solution space. Each spark in the fireworks algorithm has two attributes: the position and amplitude, the amplitude being related to the intensity value of all sparks, the higher the intensity the smaller the amplitude of the spark and vice versa (where intensity indicates the fitness value of the spark, the brighter the oxygen concentration at this position the better the fitness).
The amplitude of each spark can be calculated by:
wherein A ismaxIs constant and is used for adjusting the overall amplitude; f. ofmin、fmaxRespectively representing a minimum fitness value and a maximum fitness valueA value; f. ofbestRepresenting the fitness value of the optimal firework individual; δ is a very small value to ensure that the denominator is not 0. The number of normal sparks per spark is also determined by its fitness value:
wherein SiIndicating the normal number of sparks to be generated by the ith spark; smaxIs constant and is used to adjust the number of explosion sparks. From the spark number calculation formulas (3) and (4), it can be seen that the more suitable the spark is, the more normal the spark can be generated, and conversely, the worse the spark adaptability is, the less the number of sparks can be generated.
Since the values calculated by the spark number calculation formulas (3) and (4) are decimal numbers, the firework algorithm converts the decimal numbers into integers by using the following formula:
wherein, a and b are respectively an explosion spark number lower limit coefficient and an explosion spark number upper limit coefficient which are used for rounding; sallIs the total number of normal sparks generated, which is a constant. From the spark number calculation equation (7), S is generated in each generationallA normal spark. The position of the normal spark generated is related to the amplitude of the present spark, and it can be seen from the amplitude calculation equations (5), (6) that the more well-adapted spark will have a smaller amplitude and will generate a normal spark around itself, while the less well-adapted spark will have a larger amplitude and will generate a normal spark further away from itself.
Each explosion of the current spark will randomly select z dimension from d dimension search space to update and generate new spark. The position of the normal spark is generated by the following equation:
wherein,
is the current spark position;
the location of a new spark generated for the explosion; z is a uniform random positive integer from 1 to d, and rand (-1,1) represents a uniform random number from-1 to 1. As can be seen from the position equation (8), the position of the normal spark has a direct relationship with its amplitude, and the larger the amplitude, the more distant the new spark is from the present spark.
In order to ensure the diversity of the population, the fireworks need to be subjected to appropriate variation, such as Gaussian variation.
During each iteration, R distinct sparks are generated, i.e., R sparks are randomly selected from the Q sparks, each spark producing a distinct spark. The specific spark is generated by the following equation:
here, randGauss (1,1) represents a random number taken in a gaussian distribution conforming to a mean of 1 and a variance of 1.
From the above process, in each generation, with Q sparks, S will be generatedall+ Q + R normal sparks and R special sparks. But only Q sparks from this spark can be selected in each generation to be retained to the next generation.
Will first be from S each timeallThe optimal spark of the + Q + R sparks is selected to be retained in the next generation, and then Q-1 sparks are selected from the spark. The probability of selecting a spark is as follows:
wherein, p (X)i) Indicating the probability that the spark is selected to be retained to the next generation, and r (x) indicating the sum of the distances of the spark from all other sparks, i.e., the farther away the spark is from the other sparks, the greater the probability that the spark is selected to be retained to the next generation.
And (3.4) calculating the optimal solution of the population, judging whether the optimal solution meets the requirements, stopping searching if the optimal solution meets the requirements, and continuing iteration if the optimal solution does not meet the requirements. The initial value of the iteration is the best solution obtained by this loop and the other solutions selected.
(4) The network structure of the Extreme Learning Machine (ELM) is the same as that of a single hidden layer neural network, but is no longer the gradient descent-based algorithm (back propagation) of the traditional neural network in the training stage, and random input layer weights and deviations are adopted, and the weights of the output layer are obtained by calculation through the generalized inverse matrix theory. And after the weights and the deviations on all the network nodes are obtained, the training of the extreme learning machine can be completed. The whole structure of the extreme learning machine can be divided into three parts: input layer, hidden layer, output layer. Setting m, L and n as the node numbers of an input layer, a hidden layer and an output layer of the ELM network respectively; p different samples (x)i，oi) I is more than or equal to 1 and less than or equal to P, wherein xi＝[xi1,xi2,…,xim]∈Rm，oi＝[oi1,oi2,…,oin]∈Rn. Then the expression of the extreme learning machine network model with L hidden layer nodes is as follows:
in this formula, Wi＝[Wi1,Wi2,…,Wim]As input weight matrix, biTo make a concession thatThreshold value of i-th node of layer, g (x) is activation function, betai＝[βi1,βi2,…,βin]TTo output a weight matrix, tj＝[tj1,tj2,…,tjn]TIs the output value of the ELM network. The above formula is expressed in matrix form:
Hβ＝T (13)
wherein,
after the input weight parameter and the threshold parameter are initialized, an output matrix H can be determined, and the output weight matrix can be calculated by the following formula:
β＝H+T (14)
H+the Moore-depend generalized inverse of the hidden layer output matrix H. Thus, the structure of the entire ELM network is determined.
(5) Optimizing the input weight and the threshold value of the extreme learning machine model by using a firework algorithm to obtain a laser cutting energy consumption prediction model for optimizing the extreme learning machine based on the firework algorithm, and then optimizing a training set CtAnd importing the data into a laser cutting energy consumption prediction model based on a firework algorithm optimization extreme learning machine to train the data.
(5.1) initializing a weight value and a threshold value of the extreme learning machine, and setting the group scale Q, the number R of variant sparks, the iteration number counter t which is 0 and the maximum iteration number Iter _ Max of the firework algorithm; and setting data in the extreme learning machine, wherein the setting mainly comprises a population size, iteration times and the like.
And (5.2) setting an initial group of the firework algorithm, and calculating the fitness value of the individuals in the group.
And respectively substituting the weight and the threshold value corresponding to each individual in the initial population into the extreme learning machine, training through a training set, calculating the prediction accuracy of the trained extreme learning machine model by using a verification set, and calculating the individual fitness value according to the prediction accuracy.
And (5.3) generating an explosion spark, introducing a Gaussian variation strategy, and generating a variation spark.
Firstly, for each firework individual in the group, calculating the radius and the number of explosion sparks by adopting a method in basic fireworks:
generating an explosion spark:
then, randomly selecting R fireworks, and respectively generating R variable sparks by adopting a Gaussian variation strategy:
for each explosion spark and variant spark generated, an individual fitness value is calculated.
And (5.4) selecting Q fireworks from the fireworks, the explosion sparks and the variant sparks according to a selection strategy to serve as the next-generation fireworks.
Firstly, according to the individual fitness value, selecting one with the maximum fitness from the candidate groups to enter the next-generation firework group, and then selecting Q-1 fireworks, explosion sparks or variant sparks from the candidate groups to enter the next-generation firework group by adopting a roulette selection strategy based on the fitness value.
The probability of selecting a spark is as follows:
and (5.5) terminating the judgment rule.
If the iteration algebra t is larger than the Iter _ Max, terminating the iteration and outputting the optimal individual and the weight value and the threshold value corresponding to the optimal individual; otherwise, t ← t +1, return to step 5.3.
(5.6) sub-training set Ct1,Ct2,…,CtkThe data are input into an extreme learning machine, and a network model of the extreme learning machine after the optimization of the firework algorithm is trained:
β＝H+T
(5.7) inputting verification set data and calculating K extreme learning machine prediction models M optimized by firework algorithm1,M2,…,MkThe accuracy of (2).
The accuracy S is calculated as follows:
the embodiment also provides a system for predicting energy consumption of laser cutting of a sheet material, which comprises a memory, a processor and computer program instructions stored on the memory and capable of being executed by the processor, and when the computer program instructions are executed by the processor, the method steps of the method according to the claims can be realized.
In order to verify the feasibility of the implementation of the method, the Q235 plate is selected as an example, the specific technical method is applied, and the prediction of the laser cutting energy consumption is realized by acquiring an electric parameter signal and a process parameter of the Q235 plate cut by the laser cutting machine and training a limit learning machine model optimized by a firework algorithm.
Description of the examples
This example was carried out on a 1mm Q235 plate using a positive ZT-J-6060M metal laser cutting machine, with a linear cut lengthThe degree is 80mm, five factors of laser power, cutting speed, gas pressure, punching frequency and defocusing amount are taken as test cutting parameters, each factor is divided into 5 levels, the specific division is shown in table 1, the five factors are represented by letters from A to E, F is taken as an error column, and the level of each factor is represented by numbers from 1 to 5. L25 (5) designed according to Table 16) Orthogonal tests table as shown in table 2, a total of 25 tests were carried out, each test being indicated by the numbers 1 to 25, the factors being indicated by a to F, the levels being indicated by 1 to 5, the factors and levels corresponding to table 1.
TABLE 1 orthogonal test control factors and horizon table
TABLE 2 orthogonal test Table
A straight cut was made on a 1mm thick Q235 board according to the test arrangement of table 2. The macroscopic appearance of the cut sheet material is shown in fig. 2, and each cutting slit corresponds to one cutting.
In this test, the sheet was cut 25 times, and in order to provide a larger amount of data for the subsequent data mining model, the example repeated the test procedure 3 times, resulting in a total of 75 data (25 × 3). This experiment needs to measure the electrical parameters of the laser cutting equipment during the cutting process. The process parameters taken by each number are the process parameters of the corresponding numbers in table 2. The numbers 2, 5, 6, 7, 14 and 24 are blank lines, indicating that the cut was impervious. FIG. 3 is a graph of electrical signals collected from the 1 st straight cut test of a 1mm Q235 sheet. This experiment was run for 73 seconds and the three phase total apparent power data is shown.
TABLE 31 mmQ235 plate cutting test data
(1) Firework-Extreme Learning machine Algorithm (FWA-ELM) parameter settings
The FWA-ELM algorithm needs to set parameters, the number of sampling times N (step 1.2) needs to be the same as the number of samples, and N is 225; initializing a group scale of a firework algorithm to be Q equal to 5, setting the maximum iteration number to be Iter _ Max equal to 100, setting the number of variant sparks to be R equal to 5, and setting an iteration number counter to be t equal to 0; the number of hidden layer nodes of the initialized limit learning machine is L-6.
(2) FWA-ELM model fitting results
Table 4 lists the accuracy of the candidate FWA-ELM models in the validation set. Since 5-fold cross-validation is employed, there are 5 candidate models, numbered 1 to 5.
TABLE 4 accuracy table for verification set of candidate FWA-ELM models
As shown in Table 4, model 1 was selected as the final FWA-ELM model because of its highest accuracy.
In conclusion, the laser cutting machine energy consumption prediction model based on the firework algorithm improved extreme learning machine network can predict the machine energy consumption in advance, has higher prediction precision, and creates good conditions for further machine energy consumption optimization.
As will be appreciated by one skilled in the art, embodiments of the present application may be provided as a method, system, or computer program product. Accordingly, the present application may take the form of an entirely hardware embodiment, an entirely software embodiment or an embodiment combining software and hardware aspects. Furthermore, the present application may take the form of a computer program product embodied on one or more computer-usable storage media (including, but not limited to, disk storage, CD-ROM, optical storage, and the like) having computer-usable program code embodied therein.
The present application is described with reference to flowchart illustrations and/or block diagrams of methods, apparatus (systems), and computer program products according to embodiments of the application. It will be understood that each flow and/or block of the flow diagrams and/or block diagrams, and combinations of flows and/or blocks in the flow diagrams and/or block diagrams, can be implemented by computer program instructions. These computer program instructions may be provided to a processor of a general purpose computer, special purpose computer, embedded processor, or other programmable data processing apparatus to produce a machine, such that the instructions, which execute via the processor of the computer or other programmable data processing apparatus, create means for implementing the functions specified in the flowchart flow or flows and/or block diagram block or blocks.
These computer program instructions may also be stored in a computer-readable memory that can direct a computer or other programmable data processing apparatus to function in a particular manner, such that the instructions stored in the computer-readable memory produce an article of manufacture including instruction means which implement the function specified in the flowchart flow or flows and/or block diagram block or blocks.
These computer program instructions may also be loaded onto a computer or other programmable data processing apparatus to cause a series of operational steps to be performed on the computer or other programmable apparatus to produce a computer implemented process such that the instructions which execute on the computer or other programmable apparatus provide steps for implementing the functions specified in the flowchart flow or flows and/or block diagram block or blocks.
The foregoing is directed to preferred embodiments of the present invention, other and further embodiments of the invention may be devised without departing from the basic scope thereof, and the scope thereof is determined by the claims that follow. However, any simple modification, equivalent change and modification of the above embodiments according to the technical essence of the present invention are within the protection scope of the technical solution of the present invention.
Claims (6)
1. The energy consumption prediction method for laser cutting of the plate is characterized by comprising the following steps:
(1) dividing laser cutting energy consumption data corresponding to laser processing equipment process parameters into a training set CtAnd a verification set Cv；
(2) Dividing training set C by K-fold cross verification methodtFor K sub-training sets Ct1,Ct2,…,Ctk；
(3) Optimizing the input weight and the threshold value of the extreme learning machine by using a firework algorithm to obtain a laser cutting energy consumption prediction model for optimizing the extreme learning machine based on the firework algorithm, and then performing a sub-training set Ct1,Ct2,…,CtkRespectively inputting the laser cutting energy consumption prediction models to train the laser cutting energy consumption prediction models to obtain K trained laser cutting energy consumption prediction models M1,M2,…,Mk；
(4) Pass verification set CvCalculating the K trained laser cutting energy consumption prediction models M1,M2,…,MkAnd selecting the model with the highest accuracy for predicting the energy consumption of laser cutting.
2. The method for predicting the energy consumption of laser cutting of the plate according to claim 1, wherein the process parameters of the laser processing equipment comprise laser power, cutting speed, gas pressure, pulse frequency and focal position.
3. The method for predicting the energy consumption of laser cutting of the plate according to claim 1, wherein the step (1) specifically comprises the following steps:
(1.1) initializing training set CtAnd a verification set CvAre empty respectively;
(1.2) carrying out N times of putting back sampling on the data, wherein N is the number of data strips; only one piece of data is extracted in each sampling, and the extracted data is added into a training set Ct；
(1.3) after N times of sampling with putting back, putting the data which are not extracted at one time into a verification set Cv。
4. The method for predicting the energy consumption of laser cutting of the plate according to claim 1, wherein the step (2) specifically comprises the following steps:
(2.1) training set CtDividing into K sub-training sets, K > 1, each containing at most
Bar data, NtFor training set CtThe number of the data pieces is equal to or greater than the number of the data pieces,
is a rounded up symbol;
(2.2) recording the K sub-training sets in (2.1) as Ct1,Ct2,…,Ctk。
5. The method for predicting the energy consumption of laser cutting of the plate according to claim 1, wherein the step (3) specifically comprises the following steps:
(3.1) initializing a weight value and a threshold value of the extreme learning machine, and setting the group scale Q, the number R of variant sparks, the iteration number counter t which is 0 and the maximum iteration number Iter _ Max of the firework algorithm; setting data in an extreme learning machine, wherein the data mainly comprises population size and iteration times;
(3.2) setting an initial group of the firework algorithm, and calculating the fitness value of the individuals in the group: respectively substituting the weight and the threshold value corresponding to each individual in the initial population into the extreme learning machine, training through a training set, calculating the prediction accuracy of the trained extreme learning machine model by using a verification set, and calculating the individual fitness value according to the prediction accuracy;
(3.3) generating an explosion spark, introducing a Gaussian variation strategy, and generating a variation spark: firstly, calculating the radius and the number of explosion sparks of each firework individual in a group, and generating explosion sparks; then, randomly selecting R fireworks, and respectively generating R variable sparks by adopting a Gaussian variation strategy; calculating an individual fitness value for each explosion spark and each variation spark generated;
(3.4) selecting Q fireworks from the fireworks, the explosion sparks and the variant sparks as next-generation fireworks according to a selection strategy: firstly, according to individual fitness value, selecting one with the maximum fitness from the candidate groups to enter the next-generation firework group, and then selecting Q-1 fireworks, explosion sparks or variant sparks from the candidate groups to enter the next-generation firework group by adopting a roulette selection strategy based on the fitness value;
(3.5) termination judgment: if the iteration times t is larger than the Iter _ Max, terminating the iteration and outputting the optimal individual and the weight value and the threshold value corresponding to the optimal individual; otherwise, t ← t +1, return to step 3.3; after iteration is completed, a laser cutting energy consumption prediction model based on the firework algorithm optimization extreme learning machine is obtained;
(3.6) sub-training set Ct1,Ct2,…,CtkInputting the data into the laser cutting energy consumption prediction model to train the laser cutting energy consumption prediction model to obtain K trained laser cutting energy consumption prediction models M1,M2,…,Mk；
(3.7) inputting validation set data and calculating the K trained laser cutting energy consumption prediction models M1,M2,…,MkThe accuracy of (2).
6. Energy consumption prediction system for laser cutting of sheets, comprising a memory, a processor and computer program instructions stored on the memory and executable by the processor, which when executed by the processor, are capable of carrying out the method steps according to claims 1-5.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202111369000.5A CN114065631B (en) | 2021-11-18 | 2021-11-18 | Energy consumption prediction method and system for plate laser cutting |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202111369000.5A CN114065631B (en) | 2021-11-18 | 2021-11-18 | Energy consumption prediction method and system for plate laser cutting |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN114065631A true CN114065631A (en) | 2022-02-18 |
| CN114065631B CN114065631B (en) | 2024-06-28 |
Family
ID=80277787
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202111369000.5A Active CN114065631B (en) | 2021-11-18 | 2021-11-18 | Energy consumption prediction method and system for plate laser cutting |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN114065631B (en) |
Citations (6)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN107169565A (en) * | 2017-04-27 | 2017-09-15 | 西安工程大学 | Yarn quality prediction method based on fireworks algorithm improvement BP neural network |
| WO2017197626A1 (en) * | 2016-05-19 | 2017-11-23 | 江南大学 | Extreme learning machine method for improving artificial bee colony optimization |
| US10061300B1 (en) * | 2017-09-29 | 2018-08-28 | Xometry, Inc. | Methods and apparatus for machine learning predictions and multi-objective optimization of manufacturing processes |
| CN110560921A (en) * | 2019-08-22 | 2019-12-13 | 浙江科技学院 | total energy consumption prediction method for laser cutting based on shortest distance |
| CN112257342A (en) * | 2020-10-20 | 2021-01-22 | 李�杰 | Neural network laser cutting quality prediction method |
| CN112949203A (en) * | 2021-03-19 | 2021-06-11 | 福州大学 | Board laser cutting quality judgment method based on electrical parameters and XGBOOST-NN algorithm |
-
2021
- 2021-11-18 CN CN202111369000.5A patent/CN114065631B/en active Active
Patent Citations (6)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| WO2017197626A1 (en) * | 2016-05-19 | 2017-11-23 | 江南大学 | Extreme learning machine method for improving artificial bee colony optimization |
| CN107169565A (en) * | 2017-04-27 | 2017-09-15 | 西安工程大学 | Yarn quality prediction method based on fireworks algorithm improvement BP neural network |
| US10061300B1 (en) * | 2017-09-29 | 2018-08-28 | Xometry, Inc. | Methods and apparatus for machine learning predictions and multi-objective optimization of manufacturing processes |
| CN110560921A (en) * | 2019-08-22 | 2019-12-13 | 浙江科技学院 | total energy consumption prediction method for laser cutting based on shortest distance |
| CN112257342A (en) * | 2020-10-20 | 2021-01-22 | 李�杰 | Neural network laser cutting quality prediction method |
| CN112949203A (en) * | 2021-03-19 | 2021-06-11 | 福州大学 | Board laser cutting quality judgment method based on electrical parameters and XGBOOST-NN algorithm |
Also Published As
| Publication number | Publication date |
|---|---|
| CN114065631B (en) | 2024-06-28 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| Yu et al. | A dynamic all parameters adaptive BP neural networks model and its application on oil reservoir prediction | |
| CN103745273B (en) | Semiconductor fabrication process multi-performance prediction method | |
| CN103440361B (en) | The modeling method of yield is etched in a kind of plasma etch process | |
| CN108171379B (en) | Power load prediction method | |
| CN102331966A (en) | Software test data evolution generation system facing path | |
| CN110221580B (en) | Feed speed optimization method based on main shaft data simulation | |
| CN105608295B (en) | The multi-objective genetic algorithm of coking furnace pressure and RBF neural Optimization Modeling method | |
| CN101782769B (en) | Quick prediction method of average flowing-through time on basis of index compensation | |
| CN104122796A (en) | Intelligent assembly sequence planning method | |
| CN114282646B (en) | Optical power prediction method and system based on two-stage feature extraction and BiLSTM improvement | |
| CN114297802A (en) | Multi-objective optimization method for laser cutting technological parameters of thin plate | |
| CN112271731B (en) | Method for generating and reducing wind power multi-period time sequence scene | |
| CN105975701A (en) | Parallel scheduling disassembly path forming method based on mixing fuzzy model | |
| CN104732067A (en) | Industrial process modeling forecasting method oriented at flow object | |
| CN117334271A (en) | Method for generating molecules based on specified attributes | |
| CN111192158A (en) | Transformer substation daily load curve similarity matching method based on deep learning | |
| Gilan et al. | Sustainable building design: A challenge at the intersection of machine learning and design optimization | |
| CN110929930A (en) | Scheduling and scheduling optimization method for marine crankshaft production line | |
| JPWO2021146361A5 (en) | ||
| CN117556532A (en) | Optimization method for multi-element matching of novel turbine disc pre-rotation system | |
| MirRokni | Applying genetic algorithm in architecture and neural network training | |
| CN114065631A (en) | Energy consumption prediction method and system for laser cutting of plate | |
| Sun et al. | A novel hybrid estimation of distribution algorithm for solving hybrid flowshop scheduling problem with unrelated parallel machine | |
| CN111144569A (en) | Yield improvement applicable model optimization method based on genetic algorithm | |
| JP2009070200A (en) | Plant operation optimization device, plant operation optimization method and plant operation optimization program |
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 |