CN117392467B - Light bar grade division method based on bar graph characteristics - Google Patents
Light bar grade division method based on bar graph characteristics Download PDFInfo
- Publication number
- CN117392467B CN117392467B CN202311676743.6A CN202311676743A CN117392467B CN 117392467 B CN117392467 B CN 117392467B CN 202311676743 A CN202311676743 A CN 202311676743A CN 117392467 B CN117392467 B CN 117392467B
- Authority
- CN
- China
- Prior art keywords
- optical rod
- optical
- refractive index
- rod
- light bar
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Active
Links
- 238000000034 method Methods 0.000 title claims abstract description 47
- 230000003287 optical effect Effects 0.000 claims abstract description 148
- 238000003066 decision tree Methods 0.000 claims abstract description 26
- 238000000605 extraction Methods 0.000 claims abstract description 24
- 238000005491 wire drawing Methods 0.000 claims description 19
- 238000005253 cladding Methods 0.000 claims description 14
- 239000006185 dispersion Substances 0.000 claims description 8
- 239000012792 core layer Substances 0.000 claims description 5
- 239000010410 layer Substances 0.000 claims description 5
- 238000004458 analytical method Methods 0.000 claims description 4
- 238000013178 mathematical model Methods 0.000 claims description 2
- 238000012545 processing Methods 0.000 claims description 2
- 239000013307 optical fiber Substances 0.000 abstract description 20
- 238000004519 manufacturing process Methods 0.000 abstract description 11
- 238000010586 diagram Methods 0.000 abstract description 3
- 230000008030 elimination Effects 0.000 abstract description 3
- 238000003379 elimination reaction Methods 0.000 abstract description 3
- PCTMTFRHKVHKIS-BMFZQQSSSA-N (1s,3r,4e,6e,8e,10e,12e,14e,16e,18s,19r,20r,21s,25r,27r,30r,31r,33s,35r,37s,38r)-3-[(2r,3s,4s,5s,6r)-4-amino-3,5-dihydroxy-6-methyloxan-2-yl]oxy-19,25,27,30,31,33,35,37-octahydroxy-18,20,21-trimethyl-23-oxo-22,39-dioxabicyclo[33.3.1]nonatriaconta-4,6,8,10 Chemical class C1C=C2C[C@@H](OS(O)(=O)=O)CC[C@]2(C)[C@@H]2[C@@H]1[C@@H]1CC[C@H]([C@H](C)CCCC(C)C)[C@@]1(C)CC2.O[C@H]1[C@@H](N)[C@H](O)[C@@H](C)O[C@H]1O[C@H]1/C=C/C=C/C=C/C=C/C=C/C=C/C=C/[C@H](C)[C@@H](O)[C@@H](C)[C@H](C)OC(=O)C[C@H](O)C[C@H](O)CC[C@@H](O)[C@H](O)C[C@H](O)C[C@](O)(C[C@H](O)[C@H]2C(O)=O)O[C@H]2C1 PCTMTFRHKVHKIS-BMFZQQSSSA-N 0.000 abstract description 2
- 238000005516 engineering process Methods 0.000 description 4
- 238000012797 qualification Methods 0.000 description 4
- 238000004364 calculation method Methods 0.000 description 3
- 230000000694 effects Effects 0.000 description 3
- 238000012681 fiber drawing Methods 0.000 description 3
- 238000012549 training Methods 0.000 description 3
- 239000013598 vector Substances 0.000 description 3
- 238000009826 distribution Methods 0.000 description 2
- 238000011156 evaluation Methods 0.000 description 2
- 239000000284 extract Substances 0.000 description 2
- 238000009472 formulation Methods 0.000 description 2
- 238000005259 measurement Methods 0.000 description 2
- 239000000203 mixture Substances 0.000 description 2
- 238000012795 verification Methods 0.000 description 2
- 239000002699 waste material Substances 0.000 description 2
- 230000001174 ascending effect Effects 0.000 description 1
- 230000009286 beneficial effect Effects 0.000 description 1
- 230000005540 biological transmission Effects 0.000 description 1
- 238000006243 chemical reaction Methods 0.000 description 1
- 238000010276 construction Methods 0.000 description 1
- 230000007547 defect Effects 0.000 description 1
- 230000002349 favourable effect Effects 0.000 description 1
- 238000012423 maintenance Methods 0.000 description 1
- 238000000691 measurement method Methods 0.000 description 1
- 238000012986 modification Methods 0.000 description 1
- 230000004048 modification Effects 0.000 description 1
- 238000012544 monitoring process Methods 0.000 description 1
- 238000003908 quality control method Methods 0.000 description 1
- 238000011160 research Methods 0.000 description 1
- 238000012163 sequencing technique Methods 0.000 description 1
- 238000012360 testing method Methods 0.000 description 1
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V10/00—Arrangements for image or video recognition or understanding
- G06V10/70—Arrangements for image or video recognition or understanding using pattern recognition or machine learning
- G06V10/764—Arrangements for image or video recognition or understanding using pattern recognition or machine learning using classification, e.g. of video objects
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N5/00—Computing arrangements using knowledge-based models
- G06N5/01—Dynamic search techniques; Heuristics; Dynamic trees; Branch-and-bound
-
- 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
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V10/00—Arrangements for image or video recognition or understanding
- G06V10/40—Extraction of image or video features
Landscapes
- Engineering & Computer Science (AREA)
- Theoretical Computer Science (AREA)
- General Physics & Mathematics (AREA)
- Physics & Mathematics (AREA)
- Artificial Intelligence (AREA)
- Business, Economics & Management (AREA)
- Computing Systems (AREA)
- Health & Medical Sciences (AREA)
- Evolutionary Computation (AREA)
- General Health & Medical Sciences (AREA)
- Multimedia (AREA)
- Software Systems (AREA)
- Computational Linguistics (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Databases & Information Systems (AREA)
- Data Mining & Analysis (AREA)
- General Engineering & Computer Science (AREA)
- Mathematical Physics (AREA)
- Manufacturing & Machinery (AREA)
- Medical Informatics (AREA)
- Economics (AREA)
- Human Resources & Organizations (AREA)
- Marketing (AREA)
- Primary Health Care (AREA)
- Strategic Management (AREA)
- Tourism & Hospitality (AREA)
- General Business, Economics & Management (AREA)
- Investigating Or Analysing Materials By Optical Means (AREA)
Abstract
The invention relates to the technical field of optical fiber preforms, and provides a light rod grading method based on rod diagram characteristics, which comprises the following steps: s1: light bar diameter deviation elimination based on median scaling, S2: light bar characteristic extraction based on refractive index, S3: fitting optical wand parameters, and S4: performing optical rod grade division based on the data of fitting prediction; the optical rod is classified by using a decision tree algorithm, the optical rod is classified into four classes (namely ABCD class) according to the quality, the quality and the difference of the optical rod, the decision tree is used for determining the classification according to the sample proportion, the lower the quality of the optical rod is, the purer the classification is, through the process, the automatic classification of the optical rod class is realized, and the complexity of the production flow is reduced.
Description
Technical Field
The invention relates to the technical field of optical fiber preforms, in particular to a method for classifying optical fiber preform grades based on stick diagram features by utilizing a scaling idea.
Background
Optical fiber preforms are one of the key components in the optical fiber manufacturing process, and are used for forming the initial shape and size of an optical fiber, and drawing with an unqualified optical fiber preform causes a series of problems such as unstable optical fiber quality, low drawing efficiency, resource waste, and reduced product quality. Therefore, it is important to pre-judge the quality of the optical fiber preform before drawing, and in the process of drawing the optical fiber, the cut-off wavelength, the zero dispersion wavelength and the 1301-mode field diameter are three important optical parameters of the optical fiber, which have important influences on the transmission characteristics and the performance of the optical fiber, and can be generally used as an important mark for judging whether the optical fiber is qualified or not.
In general, the qualification condition of the optical rod can be known by using refractive index data of the optical rod to predict the grade of the optical rod and combining optical parameters such as cut-off wavelength, zero dispersion wavelength, 1310 mode field diameter and the like after wiredrawing. Aiming at refractive index data, the existing optical rod qualification rate judging method is mainly obtained by comparing refractive index indexes of all positions with standard indexes, and optical parameters such as cut-off wavelength, zero dispersion wavelength, 1310 mode field diameter and the like are obtained after wiredrawing, and mainly has the following defects:
(1) The technical requirements are high: implementing this method requires accurate refractive index measurement techniques and precise instrumentation, which can increase the cost of equipment procurement and maintenance;
(2) The standard formulation is complex: determining proper standard indexes requires deep knowledge and research on the performance, the application and the production process of the product, the standard setting is complicated, and the standard is required to be continuously adjusted for the light bars with different diameters so as to adapt to the changes of markets and technologies;
(3) The production flow is complicated: if online real-time monitoring is to be implemented, refractive index testing needs to be integrated into the production flow, which can increase complexity and risk;
(4) After drawing, qualification rate judgment is carried out, so that some unqualified optical rods can be used for drawing and manufacturing optical fibers, and resource and time are wasted; moreover, if the light bar is found to be unacceptable after drawing, it may be necessary to prepare it again or take other measures, increasing costs and production cycle time.
Therefore, a need has arisen for a method that can utilize the extracted feature training model to fit and predict the main parameters in the optical fiber drawing process, realize the performance of the optical rod to provide evaluation before drawing, quickly divide the grade of the optical rod, screen out high-quality optical rod, and improve the drawing efficiency.
Disclosure of Invention
The invention provides a grading method of optical rods based on rod graph characteristics, which is characterized in that index data such as a core layer, a cladding layer and the like are normalized according to the same standard by utilizing the idea of scaling, deviation caused by non-uniform diameters of the optical rods is eliminated, the rod graph data is subjected to characteristic extraction, the rod graph grades are graded by utilizing the grading method, in addition, main parameters in the optical fiber drawing process are fitted and predicted by utilizing an extracted characteristic training model, the performance of the optical rods can be evaluated before drawing, the grade of the optical rods is graded, high-quality optical rods are screened, the drawing efficiency is improved, and the problem in the background technology is solved.
The specific technical scheme of the invention is as follows:
a light bar classification method based on bar graph features, the light bar classification step comprising:
s1: the method comprises the steps of eliminating optical rod diameter deviation based on median scaling, firstly collecting refractive index data of core layers and cladding layers of a plurality of optical rod samples, then carrying out diameter scaling treatment on the refractive index data of the optical rod samples, and after the diameter scaling treatment, adjusting the refractive index data of all the optical rod samples to the same level, thereby eliminating the refractive index deviation caused by different diameters;
s2: refractive index-based light bar feature extraction, which is to convert complex data into quantifiable features for subsequent analysis and modeling;
s3: fitting optical rod parameters, selecting a proper fitting function or model according to the characteristics and the shape of the optical rod, evaluating the fitting accuracy through fitting error indexes, and calculating the performance and the characteristics of the optical rod described by zero dispersion wavelength, cut-off wavelength and 1310 mode field diameter indexes according to the parameters obtained by fitting;
s4: and (3) carrying out optical rod grading based on the data of fitting prediction, grading the optical rods through a decision tree algorithm, classifying the optical rods into A, B, C and D classes according to characteristic indexes and wiredrawing prediction indexes through a finally generated decision tree model by using the non-purity of the kene as a grading criterion of a decision tree.
As a further improvement of the scheme, in the step S1, the specific process of eliminating the diameter deviation of the optical rod based on median scaling is as follows:
(1) collecting refractive index data of a plurality of optical rod samples, including refractive indexes of a core layer and a cladding layer, and recording diameter data corresponding to each optical rod sample;
(2) the cladding diameter median is calculated, namely the cladding diameter data of all optical rod samples are firstly ordered, and then the value of the intermediate position after the ordering, namely the median, which is the intermediate value in a group of data and is used for reflecting the concentration trend of the optical rod refractive index data, is found;
(3) selecting standard optical rods, namely selecting optical rod samples with a value similar to the median of the diameter of the cladding as standard optical rods, wherein the diameters of the standard optical rods are similar to the median, and the standard optical rods can be used as reference data for subsequent scaling treatment;
(4) diameter scaling, namely performing diameter scaling treatment on refractive index data of other optical rod samples. The method comprises the steps of adjusting refractive index data of other optical rod samples in proportion according to the selected standard optical rod and the corresponding diameter difference; the refractive index data of the optical rod sample with larger diameter generally need to be amplified, and the refractive index data of the optical rod sample with smaller diameter needs to be reduced;
(5) the refractive index deviation is eliminated, namely after diameter scaling treatment, the refractive index data of all optical rod samples are adjusted to the same level, so that the refractive index deviation caused by different diameters is eliminated, further, the optical rods with different diameters have more consistent expression on the refractive index data, and the influence of the refractive index deviation on the subsequent optical rod grading is reduced;
(6) and (3) verifying and analyzing the processed refractive index data to ensure the accuracy and rationality of diameter scaling treatment, wherein the verification method can comprise drawing a refractive index curve graph, comparing refractive index values of different optical rod samples and the like.
As a further improvement of the present solution, in step S2, the refractive index-based light bar feature extraction includes feature extraction and feature combination;
(1) feature extraction is the conversion of complex data into quantifiable features for subsequent analysis and modeling of the light bar, for example, shape feature extraction of the light bar: the feature in the present invention refers to a series of features related to refractive index, such as: maximum, minimum, gu Yu position difference, pattern width position difference, gu Yu and drop zone position difference, peak domain and highest point position difference;
(2) feature combination: integrating the extracted features in the light bar according to indexes such as multiplication, division, square and the like helps to further process the features, so as to obtain new features which are more meaningful and more suitable for modeling, such as feature multiplication, feature division, feature reciprocal taking, feature difference summation, feature square evolution and the like.
As a further improvement of the present solution, in step S3, the specific procedure of the optical wand parameter fitting is as follows:
(1) selecting a fitting function: selecting a proper fitting function or model according to the characteristics and the shape of the light bar, trying different function forms until a better fitting effect is found, selecting a corresponding mathematical model to fit a curve for different types of light bars, such as special shapes of 'W' -type, 'N' -type and the like, fitting original data into a smooth curve by using curve fitting technology (such as polynomial fitting, exponential curve fitting and the like), and describing the shape of the light bar by characteristic points of the curve; after the fitting is completed, the fitting result needs to be evaluated, and the fitting effect is judged to be good or bad, so that the fitting accuracy can be evaluated by using the fitting error index;
wherein the raw data of the fit are: the stick measuring instrument performs a plurality of point position measurement results on the optical stick and extracts some characteristics according to stick diagram data: the main indexes are as follows: refractive index, core diameter, peak threshold difference;
(2) calculating an index: calculating indexes such as zero dispersion wavelength, cut-off wavelength, 1310 mode field diameter and the like to describe the performance and characteristics of the optical rod according to the parameters obtained by fitting;
the characteristic index, the wiredrawing prediction index and the fitting error index are based on the data of the existing optical rod as references; the wire drawing prediction index is based on the existing wire drawing data, the optical rod is subjected to multi-point measurement through a rod measuring instrument to obtain index data of the optical rod, the index data are subjected to feature extraction, the wire drawing index ('zero dispersion wavelength', '1310 mode field diameter', 'cut wavelength') is fitted through the data after the feature extraction, and accuracy of a result is verified, so that a fitting equation is obtained, the wire drawing index is obtained, the optical rod is graded, and the optical rod is used for judging the optical rod before subsequent wire drawing.
As a further improvement of the present invention, in step S4, optical rod grading is performed based on the data of fitting prediction, optical rod grading is performed by using a decision tree algorithm, and a genie unrepeace (Gini purity) is used as a grading criterion of the decision tree; the method comprises the steps that the Indonesia is used for measuring the mixing degree of samples in a data set, and when the Indonesia is low, the samples in the data set are pure and are easy to classify; the dividing flow is as follows:
(1) dividing standard: four category labels are provided: A. b, C and D, corresponding to the four classes of light bars, the division criteria of the decision tree can be expressed as:
wherein p is i Is the sample proportion of category i in the dataset;
(2) node calculation: for the division of a certain characteristic, the weighted base-Ni non-purity of each division point can be calculated, and the division point with the lowest base-Ni non-purity is selected as a node of a decision tree;
(3) iteration: the generation process of the decision tree model enables the reduction of the non-purity of the base Ni after each division to be maximum by recursively selecting optimal division characteristics and division points until stopping conditions (such as the depth of the tree, the number of leaf node samples and the like) are met; and classifying the finally generated decision tree model into A, B, C class and D class according to the characteristic index and the wiredrawing prediction index.
Compared with the prior art, the invention has the following beneficial effects:
1. according to the invention, the diameter deviation of the optical rod based on median scaling is eliminated, the median optical rod of the cladding diameter in refractive index data is found out, and the target optical rod diameter is scaled to the same level by taking the median optical rod as a standard, so that the refractive index deviation caused by different rod diameters is eliminated, and the requirement on precision instruments and equipment is further reduced;
2. the invention extracts light bar refractive index characteristic data based on refractive index data, and comprises construction characteristic statistics, and calculates statistics such as an average value, a variance, a standard deviation, a maximum value, a minimum value, a Gu Yu position difference, a graph width position difference, a Gu Yu and descending region position difference, a peak region and highest point position difference, a peak region and Gu Yu position difference, a peak region and ascending region position difference, and two highest point position differences of the light bar refractive index data, wherein the statistics can reflect integral characteristics of light bar refractive index distribution, and the peak characteristics: detecting the peak value of the refractive index data of the optical rod, wherein the peak value comprises a peak value position, a peak value height and the like, and frequency domain characteristics are as follows: frequency domain features such as peaks, frequency distribution and the like in a frequency domain can be extracted by carrying out Fourier transform on the refractive index data of the optical rod; further solves the problem of complex standard formulation in the existing light bar grading process;
3. the invention carries out the grading of the optical rods based on the fitting prediction, uses a decision tree algorithm to carry out the grading of the optical rods, takes the base non-purity as a grading criterion, recursively selects optimal grading characteristics and grading points according to characteristic indexes and wiredrawing prediction indexes, and ensures that the base non-purity is reduced to the greatest extent; the finally generated decision tree model divides the optical rods into four classes of grades (namely ABCD grade) of excellent grade, good grade, medium grade and bad grade, the decision tree decides the division according to the proportion of the sample, the lower the non-purity of the base is, the purer the division is, through the process, the automatic classification of the optical rod grade is realized, and the complexity of the production flow is reduced;
4. according to the invention, through fitting main parameters of the optical rod, after the characteristics of the optical rod are extracted, a function is selected to fit target variables, so that a fitting function is obtained, the subsequent prediction is facilitated, the performance of the optical rod is evaluated and optimized, the qualification rate judgment is performed after the optical rod is drawn, and the unqualified optical rod is used for drawing and manufacturing optical fibers, so that the waste of resources and time is easily caused.
Drawings
FIG. 1 is a flow chart of the division of light bar grades according to the present invention;
FIG. 2 is a flow chart of light bar classification based on refractive index feature extraction in accordance with the present invention.
Description of the embodiments
Embodiments of the present invention are described in further detail below with reference to the accompanying drawings and examples. The following examples are illustrative of the invention but are not intended to limit the scope of the invention.
The invention provides a light bar grading method based on refractive index feature extraction, which comprises the steps of firstly collecting refractive index data of a plurality of light bar samples and covering refractive index information of different positions; then, for each optical rod sample, refractive index characteristic vectors are established by extracting refractive index data of each optical rod sample at each position, and the characteristic vectors reflect optical properties of the refractive index of the optical rod; then, taking the feature vectors as input to realize grading of the light bars; in addition, the key wire drawing indexes of the optical rod are predicted and fitted, the performance of the optical rod before wire drawing is predicted and evaluated, and then the high-quality optical rod is screened out for the wire drawing process, unqualified optical rods are avoided, the efficiency of the wire drawing process is improved, and accordingly the generation of bad optical fibers is reduced.
As shown in fig. 1-2, the specific implementation steps of the light bar classification flow are as follows:
the first step: optical rod diameter deviation elimination based on median scaling
(1) Firstly, collecting refractive index data of a plurality of optical rod samples (the data is the refractive index of a past qualified optical rod), wherein the refractive index data comprise refractive index information of different positions on the optical rod, including refractive indexes of a core layer and a cladding layer, and diameter data of each optical rod sample;
(2) Calculating the median of the cladding diameters of the optical rods, sequencing the cladding diameter data of all optical rod samples, and then determining the median;
(3) Selecting standard optical rods, and selecting optical rod samples with a value similar to the median of the diameter of the cladding as standard optical rods, wherein the diameters of the standard optical rods are similar to the median, and the standard optical rods are used for reference of subsequent scaling treatment;
(4) After the diameter of the optical rod is scaled, the median of the optical rod and the standard optical rod are determined, refractive index data of other optical rod samples except the standard optical rod are selected for diameter scaling; the specific method comprises the following steps: according to the selected standard optical rod and the corresponding diameter difference, the refractive index data of other optical rod samples are adjusted in proportion, the refractive index data of the optical rod sample with larger diameter needs to be amplified, and the refractive index data of the optical rod sample with smaller diameter needs to be reduced;
(5) The refractive index deviation of the optical rods is eliminated, after the optical rods are subjected to diameter scaling treatment, the refractive index data of all optical rod samples are adjusted to the same level, the refractive index deviation caused by different diameters is eliminated, the optical rods with different diameters of the jernorey have more consistent performance on the refractive index data, and the later feature extraction is facilitated;
(6) And (3) verifying data, namely verifying and analyzing the processed optical rod refractive index data to ensure the accuracy and rationality of diameter scaling treatment, wherein the verification method comprises the steps of drawing a refractive index curve graph, comparing refractive index values of different optical rod samples and the like.
The diameter deviation elimination based on median scaling can enable the optical rods with different diameters to be consistent in refractive index data, and eliminate the influence of the diameters on refractive indexes, so that a more accurate basis is provided for subsequent feature extraction, model training and optical rod grading; the method is favorable for improving the accuracy and stability of the optical rod evaluation, optimizing the optical rod selection and quality control in the optical fiber drawing process, and laying a foundation for the classification of the optical rod.
And a second step of: based on the feature extraction of the refractive index of the light bar, the feature extraction is to convert complex data into quantifiable features so as to facilitate subsequent analysis and modeling of the light bar, and the specific steps of the feature extraction are as follows:
(1) The feature extraction of the light bar comprises feature extraction related to shape and refractive index, and the feature extraction of the light bar comprises position coordinate differences, numerical coordinate differences and the like between extreme points and inflection points; extracting a series of characteristic related to the refractive index of the optical rod, wherein the characteristic comprises a maximum value, a minimum value, a Gu Yu position difference, a graph width position difference, a Gu Yu and descending region position difference and a peak region and highest point position difference; the feature extraction in the invention refers to a series of feature extraction related to the refractive index of the optical rod, and the feature parameter extraction is the prior art and is not described herein;
(2) Integrating the extracted features according to indexes such as multiplication, division, squaring and the like helps further processing the features, and obtaining new features which are more meaningful and more suitable for modeling, including feature multiplication, feature division, feature reciprocal taking, feature difference summation, feature square evolution and the like.
And a third step of: based on the characteristics of the light bar, parameter fitting is carried out, a proper fitting function or model is selected according to the characteristics and the shape of the light bar, and the fitting process mainly comprises the following steps of:
(1) Fitting the original data of the optical rod into a smooth curve by using curve fitting technology (such as polynomial fitting, exponential curve fitting and the like), describing the shape of the optical rod by using characteristic points of the curve, evaluating a fitting result after fitting, judging whether the fitting effect is good or not, and evaluating the fitting accuracy by using indexes such as fitting errors and the like;
(2) According to the parameters obtained by fitting, indexes such as zero dispersion wavelength, cut-off wavelength, 1310 mode field diameter and the like are calculated to describe the performance and the characteristics of the optical rod, and whether the optical rod meets the requirements is further determined.
Fourth step: performing optical rod grading based on the data of fitting prediction, grading the optical rods by utilizing a decision tree algorithm, and using the genie unrepeace (Gini purity) as a grading criterion of the decision tree; the method comprises the steps of measuring the mixing degree of samples in a data set by using the Indonesia, and when the Indonesia is low, indicating that the samples in the data set are purer and easy to classify. The dividing flow of the light bar level by the decision tree algorithm is specifically as follows:
(1) Formulating a division standard, and arranging four category labels: A. b, C and D, corresponding to the four classes of light bars, the division criteria of the decision tree can be expressed as:
wherein p is i Is the sample proportion of category i in the dataset; the wire drawing prediction index is obtained based on standard optical rod data;
(2) Node calculation, namely calculating the weighted base non-purity of each division point for the division of a certain feature, and selecting the division point with the lowest base non-purity as a node of a decision tree;
(3) After the iterative calculation, the division standard and the node determination are completed, the decision tree can be generated, the generation process enables the reduction of the base-Ni unrepeace after each division to be maximum by recursively selecting the optimal division characteristics and the division points until the stop condition is met, and the finally generated decision tree model classifies the optical rods into A, B, C class and D class grades according to the characteristic index and the wiredrawing prediction index, namely the class grades corresponding to the excellent class, the good class, the medium class and the poor class in the figure 2.
The method effectively improves the quality of the optical rod, optimizes the drawing process, improves the production efficiency, and provides an accurate and efficient grading method for optical fiber manufacturing.
The embodiments of the present invention have been presented for purposes of illustration and description, but are not intended to be exhaustive or limited to the invention in the form disclosed, and although the invention has been described in detail with reference to the embodiments, it will be apparent to those skilled in the art that modifications may be made to the technical solutions described in the foregoing embodiments, or equivalents may be substituted for some of the technical features thereof.
Claims (7)
1. The light bar grading method based on the bar graph characteristics is characterized by comprising the following steps of:
s1: the method comprises the steps of eliminating optical rod diameter deviation based on median scaling, firstly collecting refractive index data of core layers and cladding layers of a plurality of optical rod samples, then carrying out diameter scaling treatment on the refractive index data of the optical rod samples, and after the diameter scaling treatment, adjusting the refractive index data of all the optical rod samples to the same level, thereby eliminating the refractive index deviation caused by different diameters;
s2: refractive index-based light bar feature extraction, which is to convert complex data into quantifiable features for subsequent analysis and modeling;
s3: fitting optical rod parameters, selecting a proper fitting function or model according to the characteristics and the shape of the optical rod, evaluating the fitting accuracy through fitting error indexes, and calculating the performance and the characteristics of the optical rod described by zero dispersion wavelength, cut-off wavelength and 1310 mode field diameter indexes according to the parameters obtained by fitting;
s4: and (3) carrying out optical rod grading based on the data of fitting prediction, grading the optical rods through a decision tree algorithm, classifying the optical rods into A, B, C and D classes according to characteristic indexes and wiredrawing prediction indexes through a finally generated decision tree model by using the non-purity of the kene as a grading criterion of a decision tree.
2. The light bar classification method based on bar graph features of claim 1, wherein: in step S1, before the optical rod diameter deviation is eliminated, the cladding diameter data of all optical rod samples are ordered, then the median of the diameters is determined, and an optical rod sample having a value close to the median of the cladding diameters is selected as a standard optical rod for reference of the diameter scaling process.
3. The light bar classification method based on bar graph features of claim 2, wherein: the diameter scaling treatment of the optical rod is to scale the refractive index data of other optical rod samples according to the selected standard optical rod and the corresponding diameter difference, and scale the expansion or reduction of the refractive index according to the diameter of the optical rod.
4. The light bar classification method based on bar graph features of claim 1, wherein: in step S2, the feature extraction of the optical wand is a series of features related to refractive index, including maximum, minimum, gu Yu position difference, pattern width position difference, gu Yu and descending region position difference, peak domain and high point position difference; the extracted features are integrated according to indexes, such as multiplication, division and squaring, so that further processing of the features is facilitated, and new features suitable for modeling are obtained.
5. The light bar classification method based on bar graph features of claim 1, wherein: in step S3, for light bars of different shapes, curve fitting in a mathematical model is selected, the original data is fitted into a smooth curve, and then the shape of the light bar is described by the characteristic points of the curve.
6. The light bar classification method based on bar graph features of claim 1, wherein: in step S4, according to the characteristic index and the wiredrawing prediction index, the division criteria of the decision tree are:
，
wherein p is i Is the sample proportion of class i in the dataset.
7. The light bar classification method based on bar graph features of claim 6, wherein: the decision tree model generation process enables the base-Ni unrepeace after each division to be greatly reduced by recursively selecting the division characteristics and the division points until the stop condition is met.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202311676743.6A CN117392467B (en) | 2023-12-08 | 2023-12-08 | Light bar grade division method based on bar graph characteristics |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202311676743.6A CN117392467B (en) | 2023-12-08 | 2023-12-08 | Light bar grade division method based on bar graph characteristics |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN117392467A CN117392467A (en) | 2024-01-12 |
| CN117392467B true CN117392467B (en) | 2024-02-13 |
Family
ID=89441299
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202311676743.6A Active CN117392467B (en) | 2023-12-08 | 2023-12-08 | Light bar grade division method based on bar graph characteristics |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN117392467B (en) |
Citations (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN109559059A (en) * | 2018-12-17 | 2019-04-02 | 重庆邮电大学 | A kind of optical fiber production Rulemaking method based on regression tree |
| CN117346664A (en) * | 2023-12-04 | 2024-01-05 | 南京烽火星空通信发展有限公司 | Optical rod bow degree calculation method based on rotation stick measurement data |
-
2023
- 2023-12-08 CN CN202311676743.6A patent/CN117392467B/en active Active
Patent Citations (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN109559059A (en) * | 2018-12-17 | 2019-04-02 | 重庆邮电大学 | A kind of optical fiber production Rulemaking method based on regression tree |
| CN117346664A (en) * | 2023-12-04 | 2024-01-05 | 南京烽火星空通信发展有限公司 | Optical rod bow degree calculation method based on rotation stick measurement data |
Also Published As
| Publication number | Publication date |
|---|---|
| CN117392467A (en) | 2024-01-12 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN110245802B (en) | Cigarette empty-head rate prediction method and system based on improved gradient lifting decision tree | |
| CN109543874B (en) | Airport air quality prediction method combining meteorological condition influence | |
| CN107577605A (en) | A kind of feature clustering system of selection of software-oriented failure prediction | |
| CN110826618A (en) | Personal credit risk assessment method based on random forest | |
| CN112749840B (en) | Method for acquiring energy efficiency characteristic index reference value of thermal power generating unit | |
| CN113344019A (en) | K-means algorithm for improving decision value selection initial clustering center | |
| CN103559303A (en) | Evaluation and selection method for data mining algorithm | |
| CN109597757B (en) | Method for measuring similarity between software networks based on multidimensional time series entropy | |
| CN112819299A (en) | Differential K-means load clustering method based on center optimization | |
| CN111401785A (en) | Power system equipment fault early warning method based on fuzzy association rule | |
| CN110555058A (en) | Power communication equipment state prediction method based on improved decision tree | |
| CN111949939A (en) | Intelligent electric meter running state evaluation method based on improved TOPSIS and cluster analysis | |
| CN111860624A (en) | Power grid fault information classification method based on decision tree | |
| CN113420795A (en) | Mineral spectrum classification method based on void convolutional neural network | |
| WO2023108514A1 (en) | Spectral peak automatic detection and reconstruction method and system in tea near-infrared spectral analysis | |
| CN109657287B (en) | Hydrological model precision identification method based on comprehensive scoring method | |
| CN117392467B (en) | Light bar grade division method based on bar graph characteristics | |
| CN108428234B (en) | Interactive segmentation performance optimization method based on image segmentation result evaluation | |
| CN116579842B (en) | Credit data analysis method and system based on user behavior data | |
| CN116432052B (en) | Quality detection method for clamp for new energy automobile die | |
| CN109559059B (en) | Regression decision tree-based optical fiber production rule making method | |
| CN110362911A (en) | A kind of agent model selection method of Design-Oriented process | |
| CN117036781A (en) | Image classification method based on tree comprehensive diversity depth forests | |
| CN116564409A (en) | Machine learning-based identification method for sequencing data of transcriptome of metastatic breast cancer | |
| CN115345846A (en) | Intelligent grading method and system for grain size of medium and low carbon steel |
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 |