CN113449930A - Optical fiber preform preparation quality prediction method based on BP neural network - Google Patents
Optical fiber preform preparation quality prediction method based on BP neural network Download PDFInfo
- Publication number
- CN113449930A CN113449930A CN202110847243.9A CN202110847243A CN113449930A CN 113449930 A CN113449930 A CN 113449930A CN 202110847243 A CN202110847243 A CN 202110847243A CN 113449930 A CN113449930 A CN 113449930A
- Authority
- CN
- China
- Prior art keywords
- optical fiber
- neural network
- fiber preform
- fitness
- data
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Pending
Links
- 239000013307 optical fiber Substances 0.000 title claims abstract description 100
- 238000013528 artificial neural network Methods 0.000 title claims abstract description 79
- 238000002360 preparation method Methods 0.000 title claims abstract description 55
- 238000000034 method Methods 0.000 title claims abstract description 24
- 238000012360 testing method Methods 0.000 claims abstract description 82
- 238000012549 training Methods 0.000 claims abstract description 42
- 230000002068 genetic effect Effects 0.000 claims abstract description 32
- 210000000349 chromosome Anatomy 0.000 claims description 84
- 230000006870 function Effects 0.000 claims description 46
- 210000002569 neuron Anatomy 0.000 claims description 25
- 230000003287 optical effect Effects 0.000 claims description 25
- 230000035772 mutation Effects 0.000 claims description 24
- 239000007789 gas Substances 0.000 claims description 23
- 238000004519 manufacturing process Methods 0.000 claims description 17
- 230000009466 transformation Effects 0.000 claims description 16
- VYZAMTAEIAYCRO-UHFFFAOYSA-N Chromium Chemical compound [Cr] VYZAMTAEIAYCRO-UHFFFAOYSA-N 0.000 claims description 12
- 238000004364 calculation method Methods 0.000 claims description 6
- 108090000623 proteins and genes Proteins 0.000 claims description 6
- 238000012546 transfer Methods 0.000 claims description 6
- 230000014759 maintenance of location Effects 0.000 claims description 4
- 238000007781 pre-processing Methods 0.000 claims description 4
- 230000002159 abnormal effect Effects 0.000 claims description 3
- 230000005540 biological transmission Effects 0.000 claims description 3
- 230000001537 neural effect Effects 0.000 claims 1
- 238000002474 experimental method Methods 0.000 description 7
- 239000000463 material Substances 0.000 description 2
- 238000005457 optimization Methods 0.000 description 2
- 238000000576 coating method Methods 0.000 description 1
- 230000001808 coupling effect Effects 0.000 description 1
- 230000007547 defect Effects 0.000 description 1
- 230000000694 effects Effects 0.000 description 1
- 230000006872 improvement Effects 0.000 description 1
- 230000004048 modification Effects 0.000 description 1
- 238000012986 modification Methods 0.000 description 1
- 230000008569 process Effects 0.000 description 1
- 239000002994 raw material Substances 0.000 description 1
- 239000002699 waste material Substances 0.000 description 1
Images
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06Q—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
- G06Q10/00—Administration; Management
- G06Q10/04—Forecasting or optimisation specially adapted for administrative or management purposes, e.g. linear programming or "cutting stock problem"
-
- 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
- G06N3/084—Backpropagation, e.g. using gradient descent
-
- 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
- G06Q—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
- G06Q10/00—Administration; Management
- G06Q10/06—Resources, workflows, human or project management; Enterprise or organisation planning; Enterprise or organisation modelling
- G06Q10/063—Operations research, analysis or management
- G06Q10/0639—Performance analysis of employees; Performance analysis of enterprise or organisation operations
- G06Q10/06395—Quality analysis or management
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06Q—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
- G06Q50/00—Information and communication technology [ICT] specially adapted for implementation of business processes of specific business sectors, e.g. utilities or tourism
- G06Q50/04—Manufacturing
-
- 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
Landscapes
- Engineering & Computer Science (AREA)
- Business, Economics & Management (AREA)
- Physics & Mathematics (AREA)
- Theoretical Computer Science (AREA)
- Human Resources & Organizations (AREA)
- Health & Medical Sciences (AREA)
- General Physics & Mathematics (AREA)
- Life Sciences & Earth Sciences (AREA)
- Strategic Management (AREA)
- Biophysics (AREA)
- Economics (AREA)
- General Health & Medical Sciences (AREA)
- General Engineering & Computer Science (AREA)
- Artificial Intelligence (AREA)
- Evolutionary Computation (AREA)
- Computational Linguistics (AREA)
- Molecular Biology (AREA)
- Computing Systems (AREA)
- Biomedical Technology (AREA)
- Entrepreneurship & Innovation (AREA)
- Mathematical Physics (AREA)
- Software Systems (AREA)
- General Business, Economics & Management (AREA)
- Tourism & Hospitality (AREA)
- Development Economics (AREA)
- Data Mining & Analysis (AREA)
- Marketing (AREA)
- Evolutionary Biology (AREA)
- Bioinformatics & Cheminformatics (AREA)
- Bioinformatics & Computational Biology (AREA)
- Game Theory and Decision Science (AREA)
- Operations Research (AREA)
- Quality & Reliability (AREA)
- Educational Administration (AREA)
- Physiology (AREA)
- Genetics & Genomics (AREA)
- Manufacturing & Machinery (AREA)
- Primary Health Care (AREA)
- Manufacture, Treatment Of Glass Fibers (AREA)
Abstract
The application provides a BP neural network-based optical fiber perform preparation quality prediction method, which comprises the steps of processing test data, constructing a BP neural network-based optical fiber perform preparation quality prediction model (net) and a member Genetic algorithm model genetics, optimizing initial weight and threshold of a BP neural network-based optical fiber perform preparation quality prediction model net, assigning the weight and threshold optimized by the genetic algorithm to the BP neural network, training the BP neural network, testing the BP neural network-based optical fiber perform preparation quality prediction model net, adjusting the weight and threshold of the BP neural network-based optical fiber perform preparation quality prediction model net according to test errors, and finally obtaining the BP neural network-based optical fiber perform preparation quality prediction model.
Description
Technical Field
The invention relates to the technical field of optical fiber perform production, in particular to a preparation quality prediction method of an optical fiber perform based on a BP neural network.
Background
In the field of optical fiber preform production, a controlled variable method is mostly adopted for experiments in the aspect of optical fiber preform manufacturing quality, and in recent years, a neural network and a genetic algorithm are rapidly developed, so that the method is suitable for solving the complex relation and is better applied in the aspects of material preparation and optimization.
At present, the process optimization method in the field of optical fiber preform production needs to be carried out by experiments, and the influence of different gas components on the preparation quality of the optical fiber preform is tested by using a controlled variable method; however, the method has the characteristics of too many variables to be controlled, nonlinearity, strong coupling property and the like, and a great amount of experiments are needed to find a better process scheme, so that a great amount of manpower, material resources and time are consumed.
Disclosure of Invention
The invention aims to make up the defects of the prior art and provides a preparation quality prediction method of an optical fiber preform based on a BP neural network, which adopts the following technical scheme:
a preparation quality prediction method of an optical fiber preform based on a BP neural network comprises the following specific steps:
step 1: collecting original data of optical fiber preform production, collecting gas flow of a blast burner and optical rod quality corresponding to the gas flow when the optical fiber preform is prepared, and respectively using the gas flow as input data and the optical rod quality as output data;
step 2: preprocessing the collected original data, and removing invalid data to obtain a preprocessed data set; the invalid data refers to the data of abnormal production conditions of the optical fiber preform and the quality of the optical fiber preform, and is caused by accidental factors in production;
and step 3: grading the quality of the optical rod, dividing the quality of the optical rod into a plurality of grades according to the optical performance of the optical rod, the transmission performance of the produced optical fiber and the mechanical performance, and grouping the preprocessed data sets according to the grade of the quality of the optical rod;
and 4, step 4: dividing each group of data into a training set and a testing set by adopting a retention method, wherein the data volume ratio of the training set to the testing set is 7 to 3;
and 5: normalizing the training set by using a mapminmax function, normalizing input data and output data of the training set to be between-1 and 1, and outputting normalized input transformation data input train and output transformation data output train;
step 6: constructing a BP neural network-based optical fiber preform preparation quality prediction model (net);
and 7: the component Genetic algorithm model Genetic is used for optimizing the initial weight and the threshold of the BP neural network-based optical fiber preform preparation quality prediction model net to obtain an optimal individual x, wherein the optimal individual x comprises the initial weight and the threshold information of the optimized BP neural network;
and 8: assigning the weight value and the threshold value optimized by the genetic algorithm to a BP neural network; extracting an initial weight and a threshold value of the optimized BP neural network from the optimal individual x solved by the genetic algorithm, and assigning a value to the BP neural network after finishing;
and step 9: BP neural network training: setting training parameters, wherein the iteration times are 100, the learning efficiency is 0.05, and the target precision is 0.00001; inputting normalized input transformation data input train and output transformation data output train by using a train function to complete the training of the neural network, namely completing the training of an optical fiber preform preparation quality prediction model based on the BP neural network;
step 10: normalizing the training set by using a mapminmax function, normalizing the input data of the test set and the output data of the test set to be between-1 and 1, and respectively outputting the normalized input test and output test;
step 11: testing an optical fiber preform preparation quality prediction model net based on a BP neural network; using a sim function, bringing the normalized test set input data input test into a trained optical fiber preform preparation quality prediction model net for testing to obtain a coded test result, and recording the coded test result as an; decoding the optical fiber preform, and recording a decoding result as test _ simul, namely a test result of the optical fiber preform quality prediction model;
step 12: the test result test _ simul is differed from the test set output data output test to obtain the test error of the optical fiber preform preparation quality prediction model; and adjusting the weight and the threshold of the net of the optical fiber preform preparation quality prediction model according to the test error to obtain the BP neural network-based optical fiber preform preparation quality prediction model.
Optionally, in the step 6, constructing a BP neural network-based optical fiber preform quality prediction model (net) includes: the three-layer neural network structure comprises an input layer, a hidden layer and an output layer; the input layer to hidden layer transfer function is "tansig" and the hidden layer to output layer transfer function is "purelin"; the number m of the input layer neurons is adapted to the number of the types of gases of the blast burner during the preparation of the optical fiber preform, and the input layer neurons respectively input the gas flow rates corresponding to the input layer neurons; the number of neurons in the output layer is y, and the quality grade of the corresponding optical rod of the neurons in the output layer is; the number of hidden layer neurons is n, and is adjusted according to the number m of input layers and the number y of output layer neurons, and one of m ═ n + y)1/2+ a, m ═ ㏒ 2n, and m ═ nl) 1/2 is adopted, and a is one of constants between 1 and 10.
Optionally, the number n of hidden layer neurons is determined, and the formula m ═ n + y 1/2+ a is adopted.
Alternatively, step 7, the Genetic algorithm model (Genetic) comprises the steps of:
step 7.1: initializing parameters of a genetic algorithm, and setting parameters of the genetic algorithm according to the data volume of the preprocessed data set, wherein the parameters comprise iteration times maxgen, population scale sizepop, cross probability pcross and variation probability pmutation;
step 7.2: calculating the total number numsum of nodes of the BP neural network optical fiber perform quality prediction model, and adopting a formula numsum which is m + n + n + y; setting a chromosome length lenchrom and a boundary value bound, wherein the length lenchrom of the chromosome is the same as the total number of nodes numsum;
step 7.3: population initialization: inputting the chromosome length lenchrom and the boundary value bound obtained in the previous step into a function code, and randomly generating chromosome individuals to form a population individuals;
step 7.4: calculating the fitness: inputting individual chromosome information and information of a quality prediction model prepared by a BP neural network optical fiber perform into a fitness calculation function (fun) to obtain corresponding fitness indicvidials. Finding out the chromosome bestchrom with the best fitness in each generation and recording the chromosome information, the best fitness value bestfittness and the average fitness avgfittness of the generation;
step 7.5: selecting operation: inputting the population information individuals and the population quantity sizpop into a selection function select, and outputting the selected new population individuals; the selection in the selection function select (e.g. roulette) is based on the fitness value of each individual to determine the probability of selection;
step 7.6: and (3) cross operation: inputting the crossover probability pcross, the chromosome length lenchrom, the chromosome information chrom and the population quantity sizepop into a crossover function cross to obtain a crossover chromosome individual chrom; selecting chromosomes needing cross operation from the new population individuals obtained by the selection operation according to the cross operation probability pcross, randomly pairing every two chromosomes, and performing cross exchange on partial genes;
step 7.7: mutation operation: inputting the variation probability pmutation, the chromosome length lenchrom, the chromosome information chrom, the population quantity sizepop, the current iteration number num, the maximum iteration number maxgen and the individual boundary bound into a variation function Mutation, and outputting the chromosome chrom after variation; a Mutation function Mutation, namely selecting chromosomes needing Mutation operation from the crossed chromosomes according to probability calculation of the Mutation operation, and then mutating certain genes on the chromosomes; chromosomes are randomly selected, and mutation positions are also randomly selected;
step 7.8: and (3) replacement operation: inputting the chromosome population subjected to mutation operation into a fitness function (fun) one by one, recalculating fitness, selecting an optimal individual, outputting and recording the fitness value individuals. fitness (j) of the chromosome, finding out the chromosomes with minimum and maximum fitness and the positions of the chromosomes in the population, and replacing the chromosomes with maximum fitness by the chromosomes with minimum fitness; recording the best fitness and the average fitness in the chromosome population and the fitness of all chromosomes in the population;
step 7.9: performing iterative training, returning to the step 7.5, and performing next iteration; and (5) reaching the set maximum iteration times maxgen, finishing the iteration training to obtain the chromosome with the maximum fitness, and decoding and outputting the optimal initial weight and the threshold.
The invention has the following advantages:
the invention can use the neural network operation to quickly obtain a more accurate prediction result, and replace a large amount of labor for carrying out experiments by controlling variables, thereby avoiding the waste of raw materials such as light bars, coatings and the like, saving the cost and improving the efficiency.
Drawings
In order to more clearly illustrate the technical solutions in the embodiments of the present application, the drawings needed to be used in the embodiments or the prior art descriptions will be briefly described below, and it is obvious that the drawings in the following description are only some embodiments of the present application, and it is obvious for those skilled in the art to obtain other drawings without inventive labor.
FIG. 1 is a flow chart of the algorithm of the present invention.
Detailed Description
In order to make the technical problems, technical solutions and advantageous effects to be solved by the present application clearer, the present application is further described in detail below with reference to the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are merely illustrative of the present application and are not intended to limit the present application.
The method for predicting the preparation quality of the optical fiber preform based on the BP neural network provided by the embodiment of the application is explained. The method for predicting the preparation quality of the optical fiber preform based on the BP neural network is shown in figure 1:
a preparation quality prediction method of an optical fiber preform based on a BP neural network comprises the following specific steps:
step 1: collecting original data of optical fiber preform production, collecting gas flow of a blast burner and optical rod quality corresponding to the gas flow when the optical fiber preform is prepared, and respectively using the gas flow as input data and the optical rod quality as output data;
step 2: preprocessing the collected original data, and removing invalid data to obtain a preprocessed data set; the invalid data refers to the data of abnormal production conditions of the optical fiber preform and the quality of the optical fiber preform, and is caused by accidental factors in production;
and step 3: grading the quality of the optical rod, dividing the quality of the optical rod into a plurality of grades according to the optical performance of the optical rod, the transmission performance of the produced optical fiber and the mechanical performance, and grouping the preprocessed data sets according to the grade of the quality of the optical rod;
and 4, step 4: dividing each group of data into a training set and a testing set by adopting a retention method, wherein the data volume ratio of the training set to the testing set is 7 to 3;
and 5: normalizing the training set by using a mapminmax function, normalizing input data and output data of the training set to be between-1 and 1, and outputting normalized input transformation data input train and output transformation data output train;
step 6: constructing a BP neural network-based optical fiber preform preparation quality prediction model (net);
and 7: the component Genetic algorithm model Genetic is used for optimizing the initial weight and the threshold of the BP neural network-based optical fiber preform preparation quality prediction model net to obtain an optimal individual x, wherein the optimal individual x comprises the initial weight and the threshold information of the optimized BP neural network;
and 8: assigning the weight value and the threshold value optimized by the genetic algorithm to a BP neural network; extracting an initial weight and a threshold value of the optimized BP neural network from the optimal individual x solved by the genetic algorithm, and assigning a value to the BP neural network after finishing;
and step 9: BP neural network training: setting training parameters, wherein the iteration times are 100, the learning efficiency is 0.05, and the target precision is 0.00001; inputting normalized input transformation data input train and output transformation data output train by using a train function to complete the training of the neural network, namely completing the training of an optical fiber preform preparation quality prediction model based on the BP neural network;
step 10: normalizing the training set by using a mapminmax function, normalizing the input data of the test set and the output data of the test set to be between-1 and 1, and respectively outputting the normalized input test and output test;
step 11: testing an optical fiber preform preparation quality prediction model net based on a BP neural network; using a sim function, bringing the normalized test set input data input test into a trained optical fiber preform preparation quality prediction model net for testing to obtain a coded test result, and recording the coded test result as an; decoding the optical fiber preform, and recording a decoding result as test _ simul, namely a test result of the optical fiber preform quality prediction model;
step 12: the test result test _ simul is differed from the test set output data output test to obtain the test error of the optical fiber preform preparation quality prediction model; and adjusting the weight and the threshold of the net of the optical fiber preform preparation quality prediction model according to the test error to obtain the BP neural network-based optical fiber preform preparation quality prediction model.
Optionally, in the step 6, constructing a BP neural network-based optical fiber preform quality prediction model (net) includes: the three-layer neural network structure comprises an input layer, a hidden layer and an output layer; the input layer to hidden layer transfer function is "tansig" and the hidden layer to output layer transfer function is "purelin"; the number m of the input layer neurons is adapted to the number of the types of gases of the blast burner during the preparation of the optical fiber preform, and the input layer neurons respectively input the gas flow rates corresponding to the input layer neurons; the number of neurons in the output layer is y, and the quality grade of the corresponding optical rod of the neurons in the output layer is; the number of hidden layer neurons is n, and is adjusted according to the number m of input layers and the number y of output layer neurons, and one of m ═ n + y)1/2+ a, m ═ ㏒ 2n, and m ═ nl) 1/2 is adopted, and a is one of constants between 1 and 10.
Optionally, the number n of hidden layer neurons is determined, and the formula m ═ n + y 1/2+ a is adopted.
Alternatively, step 7, the Genetic algorithm model (Genetic) comprises the steps of:
step 7.1: initializing parameters of a genetic algorithm, and setting parameters of the genetic algorithm according to the data volume of the preprocessed data set, wherein the parameters comprise iteration times maxgen, population scale sizepop, cross probability pcross and variation probability pmutation;
step 7.2: calculating the total number numsum of nodes of the BP neural network optical fiber perform quality prediction model, and adopting a formula numsum which is m + n + n + y; setting a chromosome length lenchrom and a boundary value bound, wherein the length lenchrom of the chromosome is the same as the total number of nodes numsum;
step 7.3: population initialization: inputting the chromosome length lenchrom and the boundary value bound obtained in the previous step into a function code, and randomly generating chromosome individuals to form a population individuals;
step 7.4: calculating the fitness: inputting individual chromosome information and information of a quality prediction model prepared by a BP neural network optical fiber perform into a fitness calculation function (fun) to obtain corresponding fitness indicvidials. Finding out the chromosome bestchrom with the best fitness in each generation and recording the chromosome information, the best fitness value bestfittness and the average fitness avgfittness of the generation;
step 7.5: selecting operation: inputting the population information individuals and the population quantity sizpop into a selection function select, and outputting the selected new population individuals; the selection in the selection function select (e.g. roulette) is based on the fitness value of each individual to determine the probability of selection;
step 7.6: and (3) cross operation: inputting the crossover probability pcross, the chromosome length lenchrom, the chromosome information chrom and the population quantity sizepop into a crossover function cross to obtain a crossover chromosome individual chrom; selecting chromosomes needing cross operation from the new population individuals obtained by the selection operation according to the cross operation probability pcross, randomly pairing every two chromosomes, and performing cross exchange on partial genes;
step 7.7: mutation operation: inputting the variation probability pmutation, the chromosome length lenchrom, the chromosome information chrom, the population quantity sizepop, the current iteration number num, the maximum iteration number maxgen and the individual boundary bound into a variation function Mutation, and outputting the chromosome chrom after variation; a Mutation function Mutation, namely selecting chromosomes needing Mutation operation from the crossed chromosomes according to probability calculation of the Mutation operation, and then mutating certain genes on the chromosomes; chromosomes are randomly selected, and mutation positions are also randomly selected;
step 7.8: and (3) replacement operation: inputting the chromosome population subjected to mutation operation into a fitness function (fun) one by one, recalculating fitness, selecting an optimal individual, outputting and recording the fitness value individuals. fitness (j) of the chromosome, finding out the chromosomes with minimum and maximum fitness and the positions of the chromosomes in the population, and replacing the chromosomes with maximum fitness by the chromosomes with minimum fitness; recording the best fitness and the average fitness in the chromosome population and the fitness of all chromosomes in the population;
step 7.9: performing iterative training, returning to the step 7.5, and performing next iteration; and (5) reaching the set maximum iteration times maxgen, finishing the iteration training to obtain the chromosome with the maximum fitness, and decoding and outputting the optimal initial weight and the threshold.
Example (b):
the following experiment was performed for the optical fiber preform production using the above prediction method.
Firstly, according to the step 1: collecting 100 groups of gas flow of a blast burner and the quality of an optical rod corresponding to the gas flow when the optical fiber preform is prepared, and respectively using the gas flow as input data and the output data as original data;
according to the step 2: preprocessing 100 groups of collected original data, and obtaining a preprocessed data set if invalid data are not found;
according to the step 3: dividing the quality of the optical fiber preform into 5 grades, and grouping the preprocessed data sets according to the grade of the quality of the optical fiber preform;
according to the step 4: dividing five groups of data according to a ratio of 7 to 3 by adopting a retention method, and dividing a training set and a test set;
according to the step 5: normalizing the training set by using a mapminmax function, normalizing input data and output data of the training set to be between-1 and 1, and outputting normalized input transformation data input train and output transformation data output train;
according to the step 6: constructing a BP neural network-based optical fiber preform preparation quality prediction model (net);
according to the step 7: the component Genetic algorithm model Genetic is used for optimizing the initial weight and the threshold of the BP neural network-based optical fiber preform preparation quality prediction model net to obtain an optimal individual x, wherein the optimal individual x comprises the initial weight and the threshold information of the optimized BP neural network;
according to the step 8: extracting an initial weight and a threshold value of the optimized BP neural network from the optimal individual x solved by the genetic algorithm, and assigning a value to the BP neural network after finishing;
according to the step 9: carrying out BP neural network training: setting training parameters, wherein the iteration times are 100, the learning efficiency is 0.05, and the target precision is 0.00001; inputting normalized input transformation data input train and output transformation data output train by using a train function to complete the training of the neural network, namely completing the training of an optical fiber preform preparation quality prediction model based on the BP neural network;
according to the step 10: normalizing the training set by using a mapminmax function, normalizing the input data of the test set and the output data of the test set to be between-1 and 1, and respectively outputting the normalized input test and output test;
according to the step 11: testing an optical fiber preform preparation quality prediction model net based on a BP neural network; using a sim function, bringing the normalized test set input data input test into a trained optical fiber preform preparation quality prediction model net for testing to obtain a coded test result, and recording the coded test result as an; decoding the optical fiber preform, and recording a decoding result as test _ simul, namely a test result of the optical fiber preform quality prediction model;
according to the step 12: the test result test _ simul is differed from the test set output data output test to obtain the test error of the optical fiber preform preparation quality prediction model; and adjusting the weight and the threshold of the net of the optical fiber preform preparation quality prediction model according to the test error to obtain the BP neural network-based optical fiber preform preparation quality prediction model.
Finally, under the condition of keeping other production conditions unchanged, changing the gas flow of the blowtorch during the preparation of the optical fiber preform, and randomly generating 16 groups of gas flows of the blowtorch during the preparation of the optical fiber preform for combination; respectively inputting the gas flow of the blowtorch when 16 groups of optical fiber preforms are prepared randomly into the trained BP neural network and used for actual production; and recording and comparing the quality of the optical fiber preform obtained by using the same gas component in the two modes, and analyzing the accuracy of predicting the preparation quality of the optical fiber preform by the BP neural network. Data from the above experiment are recorded as follows:
TABLE 1 optical fiber preform quality prediction experiment
As can be seen from table 1, the predicted result of the BP neural network optimized by the genetic algorithm is not greatly deviated from the production experimental result, only one group of data has an error exceeding 0.5, and the remaining errors are all much less than 0.5, and are within an acceptable range. Therefore, the BP neural network optimized by the genetic algorithm can realize accurate prediction of the quality of the optical fiber preform.
The above description is only exemplary of the present application and should not be taken as limiting the present application, as any modification, equivalent replacement, or improvement made within the spirit and principle of the present application should be included in the protection scope of the present application.
Claims (4)
1. A preparation quality prediction method of an optical fiber preform based on a BP neural network comprises the following specific steps:
step 1: collecting original data of optical fiber preform production, collecting gas flow of a blast burner and optical rod quality corresponding to the gas flow when the optical fiber preform is prepared, and respectively using the gas flow as input data and the optical rod quality as output data;
step 2: preprocessing the collected original data, and removing invalid data to obtain a preprocessed data set; the invalid data refers to the data of abnormal production conditions of the optical fiber preform and the quality of the optical fiber preform, and is caused by accidental factors in production;
and step 3: grading the quality of the optical rod, dividing the quality of the optical rod into a plurality of grades according to the optical performance of the optical rod, the transmission performance of the produced optical fiber and the mechanical performance, and grouping the preprocessed data sets according to the grade of the quality of the optical rod;
and 4, step 4: dividing each group of data into a training set and a testing set by adopting a retention method, wherein the data volume ratio of the training set to the testing set is 7 to 3;
and 5: normalizing the training set by using a mapminmax function, normalizing input data and output data of the training set to be between-1 and 1, and outputting normalized input transformation data input train and output transformation data output train;
step 6: constructing a BP neural network-based optical fiber preform preparation quality prediction model (net);
and 7: the component Genetic algorithm model Genetic is used for optimizing the initial weight and the threshold of the BP neural network-based optical fiber preform preparation quality prediction model net to obtain an optimal individual x, wherein the optimal individual x comprises the initial weight and the threshold information of the optimized BP neural network;
and 8: assigning the weight value and the threshold value optimized by the genetic algorithm to a BP neural network; extracting an initial weight and a threshold value of the optimized BP neural network from the optimal individual x solved by the genetic algorithm, and assigning a value to the BP neural network after finishing;
and step 9: BP neural network training: setting training parameters, wherein the iteration times are 100, the learning efficiency is 0.05, and the target precision is 0.00001; inputting normalized input transformation data input train and output transformation data output train by using a train function to complete the training of the neural network, namely completing the training of an optical fiber preform preparation quality prediction model based on the BP neural network;
step 10: normalizing the training set by using a mapminmax function, normalizing the input data of the test set and the output data of the test set to be between-1 and 1, and respectively outputting the normalized input test and output test;
step 11: testing an optical fiber preform preparation quality prediction model net based on a BP neural network; using a sim function, bringing the normalized test set input data input test into a trained optical fiber preform preparation quality prediction model net for testing to obtain a coded test result, and recording the coded test result as an; decoding the optical fiber preform, and recording a decoding result as test _ simul, namely a test result of the optical fiber preform quality prediction model;
step 12: the test result test _ simul is differed from the test set output data output test to obtain the test error of the optical fiber preform preparation quality prediction model; and adjusting the weight and the threshold of the net of the optical fiber preform preparation quality prediction model according to the test error to obtain the BP neural network-based optical fiber preform preparation quality prediction model.
2. The method for predicting the quality of the optical fiber preform based on the BP neural network as claimed in claim 1, wherein the step 6 is to construct the BP neural networkThe model (net) for predicting the quality of the optical fiber preform comprises: the three-layer neural network structure comprises an input layer, a hidden layer and an output layer; the input layer to hidden layer transfer function is "tansig" and the hidden layer to output layer transfer function is "purelin"; the number m of the input layer neurons is adapted to the number of the types of gases of the blast burner during the preparation of the optical fiber preform, and the input layer neurons respectively input the gas flow rates corresponding to the input layer neurons; the number of neurons in the output layer is y, and the quality grade of the corresponding optical rod of the neurons in the output layer is; the number of hidden layer neurons is n, the adjustment is performed according to the number m of input layer neurons and the number y of output layer neurons, and m is (n + y)1/2+ a or m- ㏒2n or m ═ n (nl)1/2A is one of constants between 1 and 10.
3. The method of claim 2, wherein the number of hidden layer neurons n is determined by the formula (m ═ n + y)1/2+a。
4. The method for predicting the quality of an optical fiber preform based on a BP neural network as set forth in claim 2 or 3, wherein said step 7, Genetic algorithm model (Genetic) comprises the steps of:
step 7.1: initializing parameters of a genetic algorithm, and setting parameters of the genetic algorithm according to the data volume of the preprocessed data set, wherein the parameters comprise iteration times maxgen, population scale sizepop, cross probability pcross and variation probability pmutation;
step 7.2: calculating the total number numsum of nodes of the BP neural network optical fiber perform quality prediction model, and adopting a formula numsum which is m + n + n + y; setting a chromosome length lenchrom and a boundary value bound, wherein the length lenchrom of the chromosome is the same as the total number of nodes numsum;
step 7.3: population initialization: inputting the chromosome length lenchrom and the boundary value bound obtained in the previous step into a function code, and randomly generating chromosome individuals to form a population individuals;
step 7.4: calculating the fitness: inputting individual chromosome information and information of a quality prediction model prepared by a BP neural network optical fiber perform into a fitness calculation function (fun) to obtain corresponding fitness indicvidials. Finding out the chromosome bestchrom with the best fitness in each generation and recording the chromosome information, the best fitness value bestfittness and the average fitness avgfittness of the generation;
step 7.5: selecting operation: inputting the population information individuals and the population quantity sizpop into a selection function select, and outputting the selected new population individuals; the selection in the selection function select (e.g. roulette) is based on the fitness value of each individual to determine the probability of selection;
step 7.6: and (3) cross operation: inputting the crossover probability pcross, the chromosome length lenchrom, the chromosome information chrom and the population quantity sizepop into a crossover function cross to obtain a crossover chromosome individual chrom; selecting chromosomes needing cross operation from the new population individuals obtained by the selection operation according to the cross operation probability pcross, randomly pairing every two chromosomes, and performing cross exchange on partial genes;
step 7.7: mutation operation: inputting the variation probability pmutation, the chromosome length lenchrom, the chromosome information chrom, the population quantity sizepop, the current iteration number num, the maximum iteration number maxgen and the individual boundary bound into a variation function Mutation, and outputting the chromosome chrom after variation; a Mutation function Mutation, namely selecting chromosomes needing Mutation operation from the crossed chromosomes according to probability calculation of the Mutation operation, and then mutating certain genes on the chromosomes; chromosomes are randomly selected, and mutation positions are also randomly selected;
step 7.8: and (3) replacement operation: inputting the chromosome population subjected to mutation operation into a fitness function (fun) one by one, recalculating fitness, selecting an optimal individual, outputting and recording the fitness value individuals. fitness (j) of the chromosome, finding out the chromosomes with minimum and maximum fitness and the positions of the chromosomes in the population, and replacing the chromosomes with maximum fitness by the chromosomes with minimum fitness; recording the best fitness and the average fitness in the chromosome population and the fitness of all chromosomes in the population;
step 7.9: performing iterative training, returning to the step 7.5, and performing next iteration; and (5) reaching the set maximum iteration times maxgen, finishing the iteration training to obtain the chromosome with the maximum fitness, and decoding and outputting the optimal initial weight and the threshold.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN202110847243.9A CN113449930A (en) | 2021-07-27 | 2021-07-27 | Optical fiber preform preparation quality prediction method based on BP neural network |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN202110847243.9A CN113449930A (en) | 2021-07-27 | 2021-07-27 | Optical fiber preform preparation quality prediction method based on BP neural network |
Publications (1)
Publication Number | Publication Date |
---|---|
CN113449930A true CN113449930A (en) | 2021-09-28 |
Family
ID=77817301
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN202110847243.9A Pending CN113449930A (en) | 2021-07-27 | 2021-07-27 | Optical fiber preform preparation quality prediction method based on BP neural network |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN113449930A (en) |
Cited By (4)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN114004341A (en) * | 2021-11-15 | 2022-02-01 | 威海长和光导科技有限公司 | Optical fiber preform preparation process optimization method based on genetic algorithm and BP neural network |
CN114462699A (en) * | 2022-01-28 | 2022-05-10 | 无锡雪浪数制科技有限公司 | Optical fiber production qualification index prediction method based on random forest |
CN114707743A (en) * | 2022-04-15 | 2022-07-05 | 西安邮电大学 | Air quality prediction method and system based on adaptive gate control circulation neural network |
CN115796269A (en) * | 2023-02-09 | 2023-03-14 | 中国人民解放军国防科技大学 | Atom cooling parameter online optimization method and device based on artificial neural network |
Citations (5)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN106951983A (en) * | 2017-02-27 | 2017-07-14 | 浙江工业大学 | Injector performance Forecasting Methodology based on the artificial neural network using many parent genetic algorithms |
CN109034388A (en) * | 2018-07-27 | 2018-12-18 | 湖北中烟工业有限责任公司 | A kind of prediction model of cigarette material and mainstream smoke constituents based on Genetic Algorithm Optimized Neural Network |
CN111652402A (en) * | 2019-03-04 | 2020-09-11 | 湖南师范大学 | Optical fiber preform deposition process intelligent optimization method based on big data analysis |
CN112053019A (en) * | 2019-06-06 | 2020-12-08 | 湖南师范大学 | Optical fiber preform deposition process intelligent method based on big data model predictive control framework |
CN112100940A (en) * | 2020-09-17 | 2020-12-18 | 浙江大学 | Method and device for predicting primary stretching technological parameters of optical fiber preform |
-
2021
- 2021-07-27 CN CN202110847243.9A patent/CN113449930A/en active Pending
Patent Citations (5)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN106951983A (en) * | 2017-02-27 | 2017-07-14 | 浙江工业大学 | Injector performance Forecasting Methodology based on the artificial neural network using many parent genetic algorithms |
CN109034388A (en) * | 2018-07-27 | 2018-12-18 | 湖北中烟工业有限责任公司 | A kind of prediction model of cigarette material and mainstream smoke constituents based on Genetic Algorithm Optimized Neural Network |
CN111652402A (en) * | 2019-03-04 | 2020-09-11 | 湖南师范大学 | Optical fiber preform deposition process intelligent optimization method based on big data analysis |
CN112053019A (en) * | 2019-06-06 | 2020-12-08 | 湖南师范大学 | Optical fiber preform deposition process intelligent method based on big data model predictive control framework |
CN112100940A (en) * | 2020-09-17 | 2020-12-18 | 浙江大学 | Method and device for predicting primary stretching technological parameters of optical fiber preform |
Cited By (4)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN114004341A (en) * | 2021-11-15 | 2022-02-01 | 威海长和光导科技有限公司 | Optical fiber preform preparation process optimization method based on genetic algorithm and BP neural network |
CN114462699A (en) * | 2022-01-28 | 2022-05-10 | 无锡雪浪数制科技有限公司 | Optical fiber production qualification index prediction method based on random forest |
CN114707743A (en) * | 2022-04-15 | 2022-07-05 | 西安邮电大学 | Air quality prediction method and system based on adaptive gate control circulation neural network |
CN115796269A (en) * | 2023-02-09 | 2023-03-14 | 中国人民解放军国防科技大学 | Atom cooling parameter online optimization method and device based on artificial neural network |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
CN113449930A (en) | Optical fiber preform preparation quality prediction method based on BP neural network | |
CN110363344B (en) | Probability integral parameter prediction method for optimizing BP neural network based on MIV-GP algorithm | |
CN107229972A (en) | A kind of global optimization based on Lamarch inheritance of acquired characters principle, search and machine learning method | |
Tavakkoli-Moghaddain et al. | Facilities layout design by genetic algorithms | |
CN107346459B (en) | Multi-mode pollutant integrated forecasting method based on genetic algorithm improvement | |
CN111832101B (en) | Construction method of cement strength prediction model and cement strength prediction method | |
CN110059887B (en) | BP neural network risk identification method and system based on adaptive genetic algorithm | |
CN103699812B (en) | Plant variety authenticity identification site selection method based on genetic algorithm | |
CN111639820A (en) | Energy consumption parameter optimization method and system for cement kiln production | |
CN114004153A (en) | Penetration depth prediction method based on multi-source data fusion | |
JP2023519770A (en) | Automatic compression method and platform for pre-trained language models for multitasking | |
CN114004341A (en) | Optical fiber preform preparation process optimization method based on genetic algorithm and BP neural network | |
CN110298506A (en) | A kind of urban construction horizontal forecast system | |
CN113282747A (en) | Text classification method based on automatic machine learning algorithm selection | |
CN112182172B (en) | Group-rolling method, system, device and medium based on particle swarm genetic algorithm | |
CN112347670A (en) | Rockfill material creep parameter prediction method based on neural network response surface | |
CN117350146A (en) | GA-BP neural network-based drainage pipe network health evaluation method | |
CN111310974A (en) | Short-term water demand prediction method based on GA-ELM | |
CN114943866B (en) | Image classification method based on evolutionary neural network structure search | |
CN113743003B (en) | Method for calculating intensity of high-voltage line to ground electric field by considering influence of temperature and humidity | |
CN112132259B (en) | Neural network model input parameter dimension reduction method and computer readable storage medium | |
CN105426960B (en) | Aluminium electroloysis energy-saving and emission-reduction control method based on BP neural network Yu MBFO algorithms | |
CN106503326A (en) | A kind of many two grades of river course maximums of determination enter the Reverse Design of mainstream amount | |
CN109871953B (en) | Wavelet neural network modeling method for heavy oil cracking process of fpRNA genetic algorithm | |
Tsai et al. | The exponential grey forecasting model for CO2 emissions in Taiwan |
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 | ||
RJ01 | Rejection of invention patent application after publication | ||
RJ01 | Rejection of invention patent application after publication |
Application publication date: 20210928 |