CN112836340A - Infrared spectrum-based PE plastic pipe elongation at break identification method - Google Patents
Infrared spectrum-based PE plastic pipe elongation at break identification method Download PDFInfo
- Publication number
- CN112836340A CN112836340A CN202011624968.3A CN202011624968A CN112836340A CN 112836340 A CN112836340 A CN 112836340A CN 202011624968 A CN202011624968 A CN 202011624968A CN 112836340 A CN112836340 A CN 112836340A
- Authority
- CN
- China
- Prior art keywords
- elongation
- break
- model
- correlation
- plastic pipe
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Granted
Links
- 229920003023 plastic Polymers 0.000 title claims abstract description 72
- 239000004033 plastic Substances 0.000 title claims abstract description 72
- 238000002329 infrared spectrum Methods 0.000 title claims abstract description 38
- 238000000034 method Methods 0.000 title claims abstract description 30
- 238000012795 verification Methods 0.000 claims abstract description 59
- 238000012549 training Methods 0.000 claims abstract description 40
- 238000012360 testing method Methods 0.000 claims abstract description 26
- 238000005070 sampling Methods 0.000 claims description 76
- 230000002159 abnormal effect Effects 0.000 claims description 38
- 238000004566 IR spectroscopy Methods 0.000 claims description 13
- 238000010200 validation analysis Methods 0.000 claims description 5
- 238000004519 manufacturing process Methods 0.000 claims description 3
- 230000001502 supplementing effect Effects 0.000 claims description 3
- 238000001514 detection method Methods 0.000 abstract description 7
- 238000005516 engineering process Methods 0.000 abstract description 3
- 239000004698 Polyethylene Substances 0.000 description 51
- 229920000573 polyethylene Polymers 0.000 description 51
- 238000001228 spectrum Methods 0.000 description 3
- 239000002699 waste material Substances 0.000 description 3
- 238000004891 communication Methods 0.000 description 2
- 230000000694 effects Effects 0.000 description 2
- 238000007689 inspection Methods 0.000 description 2
- 239000000463 material Substances 0.000 description 2
- 238000013178 mathematical model Methods 0.000 description 2
- 238000012986 modification Methods 0.000 description 2
- 230000004048 modification Effects 0.000 description 2
- -1 Polyethylene Polymers 0.000 description 1
- 230000009286 beneficial effect Effects 0.000 description 1
- 238000004364 calculation method Methods 0.000 description 1
- 230000008859 change Effects 0.000 description 1
- 230000007547 defect Effects 0.000 description 1
- 230000001066 destructive effect Effects 0.000 description 1
- 238000012423 maintenance Methods 0.000 description 1
- 230000036314 physical performance Effects 0.000 description 1
- 238000000513 principal component analysis Methods 0.000 description 1
- 230000008569 process Effects 0.000 description 1
- 238000003672 processing method Methods 0.000 description 1
- 238000000611 regression analysis Methods 0.000 description 1
- 238000011160 research Methods 0.000 description 1
- 238000009864 tensile test Methods 0.000 description 1
Images
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F30/00—Computer-aided design [CAD]
- G06F30/20—Design optimisation, verification or simulation
-
- G—PHYSICS
- G01—MEASURING; TESTING
- G01N—INVESTIGATING OR ANALYSING MATERIALS BY DETERMINING THEIR CHEMICAL OR PHYSICAL PROPERTIES
- G01N21/00—Investigating or analysing materials by the use of optical means, i.e. using sub-millimetre waves, infrared, visible or ultraviolet light
- G01N21/17—Systems in which incident light is modified in accordance with the properties of the material investigated
- G01N21/25—Colour; Spectral properties, i.e. comparison of effect of material on the light at two or more different wavelengths or wavelength bands
- G01N21/31—Investigating relative effect of material at wavelengths characteristic of specific elements or molecules, e.g. atomic absorption spectrometry
- G01N21/35—Investigating relative effect of material at wavelengths characteristic of specific elements or molecules, e.g. atomic absorption spectrometry using infrared light
- G01N21/3563—Investigating relative effect of material at wavelengths characteristic of specific elements or molecules, e.g. atomic absorption spectrometry using infrared light for analysing solids; Preparation of samples therefor
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/21—Design or setup of recognition systems or techniques; Extraction of features in feature space; Blind source separation
- G06F18/214—Generating training patterns; Bootstrap methods, e.g. bagging or boosting
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F2218/00—Aspects of pattern recognition specially adapted for signal processing
- G06F2218/08—Feature extraction
- G06F2218/10—Feature extraction by analysing the shape of a waveform, e.g. extracting parameters relating to peaks
Abstract
The application discloses PE plastic pipe elongation at break recognition method based on infrared spectrum, relates to plastic pipe detection technology field, and includes: selecting a plurality of PE plastic pipe samples, collecting infrared spectrum data and a breaking elongation test value of at least one sample point of each PE plastic pipe, and randomly dividing the plurality of PE plastic pipe samples into a training sample point set and a verification sample point set; extracting a plurality of correlation factors of the elongation at break according to the infrared spectrum data of the training sample point set; after eliminating the correlation factors of which the variance is greater than a preset variance critical value, respectively performing data fitting modeling on each residual correlation factor and the elongation at break to obtain a plurality of initial models; and selecting and verifying the optimal model, and after the optimal model passes the verification, taking the optimal model as a correlation model to identify the elongation at break of the PE plastic pipe to be detected. According to the identification method, the elongation at break of the PE plastic pipe to be detected can be analyzed and identified through the correlation model, and whether the elongation at break of the PE plastic pipe to be detected reaches the standard or not is judged through fast reading.
Description
Technical Field
The application relates to the technical field of plastic pipe detection, in particular to a PE plastic pipe elongation at break identification method based on infrared spectrum.
Background
At present, the conventional reelable PE plastic pipe is delivered in a delivery length of 500-1000 meters per 1 reel, the hard straight PE plastic pipe is delivered per 6 meters, most actual market orders are tens of thousands of meters or even tens of thousands of meters, and the elongation at break needs to be detected in factory inspection and type inspection. According to industry standard YD/T841.1-5-2016 series standards, the communication Polyethylene (PE) plastic pipe has the index requirement of the elongation at break of more than or equal to 250%.
In the related art, the PE plastic pipes need to be checked, so that the entire tray of products and each plastic pipe in a batch of orders cannot be sampled to form a test strip for testing. However, if the frequency of sampling detection is increased, the whole detection workload is greatly increased, time and labor are wasted, and meanwhile, in the sampling test, human factors and probability factors exist, and quality problems can also occur in the samples which are not sampled or in the disk samples.
Disclosure of Invention
Aiming at one of the defects in the prior art, the application aims to provide the PE plastic pipe breaking elongation recognition method based on the infrared spectrum so as to solve the problems of large overall detection workload, time and labor waste in the related technology.
In order to achieve the purposes, the technical scheme adopted by the application is as follows:
a PE plastic pipe breaking elongation recognition method based on infrared spectrum comprises the following steps:
selecting a plurality of PE plastic pipe samples, collecting infrared spectrum data and a breaking elongation test value of at least one sample point of each PE plastic pipe, and randomly dividing the plurality of PE plastic pipe samples into a training sample point set and a verification sample point set;
extracting a plurality of correlation factors of the elongation at break according to the infrared spectrum data of the training sample set, and acquiring the value of each correlation factor;
after eliminating the correlation factors of which the variance is greater than a preset variance critical value, respectively performing data fitting modeling on each residual correlation factor and the elongation at break to obtain a plurality of initial models;
and selecting an optimal model from the plurality of initial models, verifying by using a verification sample set, and identifying the elongation at break of the PE plastic pipe to be detected as an associated model after the verification is passed.
In some embodiments, the acquiring infrared spectrum data of at least one sampling point of each PE plastic pipe specifically includes:
sampling each sample point respectively, and carrying out the sampling with the wave band of 4000-400cm-1Detecting the infrared spectrum to obtain a plurality of characteristic peak wave bands;
dividing a plurality of characteristic peak bands into known component bands and unknown component bands, and respectively integrating to obtain the peak area of each band.
In some embodiments, extracting a plurality of correlation factors of the elongation at break according to the infrared spectrum data of the training sample set specifically includes:
taking the peak area of each known component wave band as a correlation factor;
taking the peak area of each unknown component wave band as a correlation factor respectively;
taking the sum of peak areas of all known component wave bands as a correlation factor;
taking the sum of peak areas of all unknown component wave bands as a correlation factor;
the sum of the peak areas of all known component wave bands and all unknown component wave bands is taken as the total area, and the ratio of the sum of the peak areas of all unknown component wave bands to the total area is taken as a correlation factor.
Some embodimentsThe unknown component band comprises a first band 1900-950cm-1And a second band of 650-600cm-1The known component bands include 3100--1And a fourth wavelength band of 950-670cm-1。
In some embodiments, when performing the data fitting modeling, the method further includes: and eliminating abnormal sample points, and counting the number of the abnormal sample points eliminated by each initial model.
In some embodiments, the selecting the optimal model from the plurality of initial models specifically includes:
calculating a correlation coefficient and a standard error of each initial model;
and selecting the model with the least number of removed abnormal sampling points, the highest correlation coefficient and the smallest standard error as the optimal model.
In some embodiments, the principle of rejecting the abnormal sampling points is as follows:
rejecting pairwise outliers and obvious outlier sample points until the correlation coefficient of the model is larger than a first threshold, the confidence interval is larger than a second threshold, and the number of rejected sample points is minimum;
the pair of outliers is defined as: and if the difference value of the same related factor values of two sampling points on the same PE plastic pipe is within a preset range, and one of the elongation at break test values of the two sampling points is greater than a standard value and the other is less than the standard value, taking the two sampling points as paired abnormal points.
In some embodiments, the verifying by using the verification sample set specifically includes:
inputting infrared spectrum data of the verification sample point set into the optimal model to obtain a predicted value of the elongation at break of the verification sample point set;
comparing the predicted value of the elongation at break of the verification sample set with the corresponding test value; taking a sample point of which one of the corresponding predicted value and the test value is greater than the standard value and the other one is less than the standard value as a distortion sample point;
when no distortion sampling point exists, taking the optimal model as a correlation model; otherwise, comparing each distorted sample point with the removed abnormal sample points respectively;
if each distorted sample point has an abnormal sample point and has the same data, the optimal model is used as a correlation model;
otherwise, judging that the verification fails, supplementing the distortion sampling points without corresponding abnormal sampling points into the training sampling point set, and carrying out secondary modeling.
In some embodiments, the secondary modeling specifically includes:
if the number of the distortion sampling points supplemented into the training sampling point set exceeds 5% of the verification sampling point set, performing data fitting modeling according to the new training sampling point set to obtain a new optimal model;
otherwise, increasing the correlation factor, and establishing a three-dimensional fitting model according to the new training sample set to serve as a new optimal model.
In some embodiments, after obtaining the new optimal model, the method further includes:
manufacturing a secondary verification sample point set, and acquiring infrared spectrum data and a breaking elongation test value of the secondary verification sample point set;
and carrying out secondary verification on the new optimal model by utilizing the secondary verification sample set until a correlation model is obtained.
The beneficial effect that technical scheme that this application provided brought includes:
according to the PE plastic pipe breaking elongation recognition method based on the infrared spectrum, due to the fact that infrared spectrum data and breaking elongation test values of all sampling points are obtained in advance, then a plurality of PE plastic pipe samples are randomly divided into a training sampling point set and a verification sampling point set, a plurality of correlation factors of the breaking elongation are extracted according to the infrared spectrum data of the training sampling point set, the correlation factors with the variance smaller than or equal to a preset variance critical value are subjected to data fitting modeling with the breaking elongation respectively, an optimal model is selected, the verification sampling point set is used for verification, and the optimal model can be used as a correlation model to recognize the breaking elongation of the PE plastic pipe to be detected after the verification is passed. Therefore, after the infrared spectrum data of the PE plastic pipe to be detected are obtained, the elongation at break of the PE plastic pipe to be detected can be analyzed and identified through the correlation model, and whether the elongation at break of the PE plastic pipe to be detected reaches the standard or not is judged through quick reading.
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 description of the embodiments are briefly introduced 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 based on these drawings without creative efforts.
FIG. 1 is a flow chart of an identification method in an embodiment of the present application;
FIG. 2 is a graph of the first band peak area s1 as a function of the elongation at break L in the examples of the present application;
FIG. 3 is a graph of second band peak area s2 as a function of elongation at break L for the examples of the present application;
FIG. 4 is a graph showing the sum S of peak areas of a first wavelength band and a second wavelength band as a function of elongation at break L in an example of the present application;
FIG. 5 is a graph showing the relationship between the peak area pe1 of the third wavelength band and the elongation at break L in the examples of the present application;
FIG. 6 is a graph showing the relationship between the peak area pe2 of the fourth wavelength band and the elongation at break L in the examples of the present application;
FIG. 7 is a graph showing the sum PE of peak areas of a third band and a fourth band and the elongation at break L in examples of the present application;
FIG. 8 is a graph showing a ratio R of a sum of peak areas of a first wavelength band and a second wavelength band to a total area in the example of the present application with respect to an elongation at break L;
FIG. 9 shows an s1-R1-L interpolation model in an embodiment of the present application.
Detailed Description
In order to make the objects, technical solutions and advantages of the present invention more apparent, the present invention is described in further 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 invention and are not intended to limit the invention. In addition, the technical features involved in the embodiments of the present invention described below may be combined with each other as long as they do not conflict with each other.
The embodiment of the application provides an infrared spectrum-based PE plastic pipe elongation at break identification method, which can solve the problems of large overall detection workload, time waste and labor waste in the related technology.
As shown in figure 1, the identification method of the PE plastic pipe breaking elongation based on the infrared spectrum is used for identifying the PE plastic pipe breaking elongation for underground laying. The identification method comprises the following steps:
s1, selecting a plurality of PE plastic pipe samples, collecting infrared spectrum data and a breaking elongation test value of at least one sample point of each PE plastic pipe, and randomly dividing the plurality of PE plastic pipe samples into a training sample point set and a verification sample point set. Namely, all sampling point sets of part of PE plastic pipe samples are training sampling point sets, and all sampling point sets of the rest PE plastic pipe samples are verification sampling point sets.
And then, respectively testing the elongation at break of each sample point by a material tensile testing machine to obtain infrared spectrum data and the elongation at break test value of each sample point.
And S2, extracting a plurality of correlation factors of the elongation at break according to the infrared spectrum data of the training sample point set, and acquiring the value of each correlation factor. Then, the variance of each correlation factor can be calculated.
And S3, after eliminating the correlation factors of which the variance is greater than a preset variance critical value, performing data fitting modeling on each residual correlation factor and the elongation at break respectively to obtain a plurality of initial models.
When the variance of a certain correlation factor is greater than a preset variance critical value, the fitting degree of the correlation factor and the elongation at break is low, the fitting effect is poor, namely the fitted correlation coefficient is too low, the relative error is large, the distribution of the data of the correlation factor in the orthogonal graph is not clustered, or more clusters are not statistically common, the data distribution is scattered, the distribution area is large, and no rule for research is provided.
And S4, selecting an optimal model from the plurality of initial models, verifying by using a verification sample set, and identifying the elongation at break of the PE plastic pipe to be detected as a correlation model after the verification is passed.
According to the identification method, infrared spectrum data and elongation at break test values of all sampling points are obtained in advance, then a plurality of PE plastic pipe samples are randomly divided into a training sampling point set and a verification sampling point set, a plurality of correlation factors of the elongation at break are extracted according to the infrared spectrum data of the training sampling point set, the correlation factors of which the variance is smaller than or equal to a preset variance critical value are subjected to data fitting modeling with the elongation at break respectively, an optimal model is selected, the verification sampling point set is used for verification, and the optimal model can be used as a correlation model to identify the elongation at break of the PE plastic pipe to be detected after the verification is passed. Therefore, after the infrared spectrum data of the PE plastic pipe to be detected are obtained, the elongation at break of the PE plastic pipe to be detected can be analyzed and identified through the correlation model, and whether the elongation at break of the PE plastic pipe to be detected reaches the standard or not is judged through quick reading.
In this embodiment, in the step S1, the acquiring infrared spectrum data of at least one sampling point of each PE plastic pipe specifically includes:
firstly, sampling each sample point respectively, and carrying out the sampling with the wave band of 4000--1And (4) infrared spectrum detection to obtain a plurality of characteristic peak wave bands.
Then, dividing the characteristic peak bands into known component bands and unknown component bands, and respectively integrating to obtain the peak area of each band.
Wherein the known component waveband is the waveband of the known PE component in the PE plastic pipe, and the unknown component waveband is the waveband of the unknown component in the PE plastic pipe. The same production process for the same batch of PE plastic pipe, and therefore, the same unknown composition.
Preferably, extracting a plurality of correlation factors of the elongation at break according to the infrared spectrum data of the training sample set specifically includes:
the peak area of each known component wave band is used as a correlation factor.
The peak area of each unknown component wave band is taken as a relevant factor.
The sum of the peak areas of all known component bands is taken as a correlation factor.
The sum of the peak areas of all the unknown component bands is taken as a correlation factor.
The sum of the peak areas of all known component wave bands and all unknown component wave bands is taken as the total area, and the ratio of the sum of the peak areas of all unknown component wave bands to the total area is taken as a correlation factor. Therefore, at least 5 correlation factors need to be extracted.
In this embodiment, when performing data fitting modeling on each remaining correlation factor and the elongation at break, the method further includes: and eliminating abnormal sample points, and counting the number of the abnormal sample points eliminated by each initial model.
Preferably, in step S4, the selecting an optimal model from the plurality of initial models specifically includes:
first, the correlation coefficient and standard error of each initial model are calculated.
And then, selecting the model with the least number of removed abnormal sampling points, the highest correlation coefficient and the smallest standard error as the optimal model.
Further, the principle of eliminating the abnormal sampling points is as follows:
and rejecting the paired outliers and the obvious outlier sample points until the correlation coefficient of the model is larger than a first threshold, the confidence interval is larger than a second threshold, and the number of the rejected sample points is minimum.
The pair of outliers is defined as: and if the difference value of the same related factor values of two sampling points on the same PE plastic pipe is within a preset range, and one of the elongation at break test values of the two sampling points is greater than a standard value and the other is less than the standard value, taking the two sampling points as paired abnormal points.
The obvious outlier sample point is a cluster which is obviously far away from the data point group in the relation graph of the correlation factor and the elongation at break, or a point which is far away from the projection distance of the fitting curve.
Preferably, in step S4, the verifying with the verification sample set specifically includes:
firstly, inputting infrared spectrum data of a verification sample set into the optimal model to obtain a predicted value of the elongation at break of the verification sample set.
Then, comparing the predicted breaking elongation value of each sample point in the verification sample point set with the testing value of the breaking elongation thereof; and taking the sampling point of which one of the corresponding predicted value and the test value is larger than the standard value and the other one is smaller than the standard value as a distortion sampling point.
When no distortion sampling point exists, the model is proved to be verified to be passed, and the optimal model is used as a correlation model; otherwise, each distorted sample point is compared with the eliminated abnormal sample points respectively.
If each distorted sample point has an abnormal sample point and has the same data with the abnormal sample point, the model is proved to pass the verification, and the optimal model can be used as a correlation model; otherwise, judging that the verification fails, supplementing the distortion sampling points without corresponding abnormal sampling points into the training sampling point set, and carrying out secondary modeling.
In this embodiment, the secondary modeling specifically includes:
first, whether the number of distorted samples supplemented to the training sample set exceeds 5% of the verification sample set is judged.
And if the number of the distortion sampling points supplemented into the training sampling point set exceeds 5% of the verification sampling point set, performing data fitting modeling according to the new training sampling point set to obtain a new optimal model.
Otherwise, increasing the correlation factor, and establishing a three-dimensional fitting model according to the new training sample set to serve as a new optimal model.
Further, after obtaining a new optimal model through secondary modeling, the method further includes:
firstly, a secondary verification sample set is manufactured, and infrared spectrum data and a breaking elongation test value of the secondary verification sample set are obtained.
And then, carrying out secondary verification on the new optimal model by utilizing the secondary verification sample set until a correlation model is obtained.
In this embodiment, first, peak area signals of all characteristic peaks are found, then, according to the peak area signals of the characteristic peaks, known PE component signals and unknown component signals are distinguished, and grouping statistics is performed on all component signals. Wherein the unknown component band comprises a first band 1900-950cm-1And a second band of 650-600cm-1The known component bands include the third band 3100--1And a fourth wavelength band of 950-670cm-1And the sum of the peak areas of the four wave bands is the total area.
Accordingly, the above-mentioned correlation factors include a first band peak area S1, a second band peak area S2, a third band peak area PE1, a fourth band peak area PE2, a sum S of peak areas of the first band and the second band, a sum PE of peak areas of the third band and the fourth band, and a ratio R between the sum of peak areas of the first band and the second band and the total area. The data for the training sample set is shown in table 1 below.
TABLE 1
As shown in Table 2 below, each correlation can be obtained according to the characteristic peak area signal of the correlation factor in the spectrogram.
TABLE 2
As shown in fig. 2 to 8, by analyzing the correlation relationship between each correlation factor and the elongation at break, the correlation relationship that significantly exceeds the fitting limit can be excluded: PE1-L, PE2-L, and PE-L, thereby reducing the number of fit groups and forming a valid fit group as in table 3 below. And then, a fitting model can be established for the rest incidence relations.
TABLE 3
| Polynomial fitting | Association relation | Number of remaining samples | Number of abnormal sample points | Correlation coefficient | Standard error of |
| |
s1-L | 56 | 7 | 0.9658 | 38.75 |
| |
s2-L | 57 | 6 | 0.9458 | 49.09 |
| |
S-L | 56 | 7 | 0.9622 | 41.17 |
| |
R-L | 57 | 6 | 0.9681 | 36.88 |
As shown in FIG. 2, the initial model of the peak area s1 and elongation at break L for the first band is as follows: l ═ p1 × (s1)2+q1×(s1)+m1
p1＝44.11
q1＝-383.1
m1＝866.6
The correlation coefficient was 0.9658, and the standard error was 38.75.
As shown in FIG. 3, the initial model of the peak area s2 and elongation at break L of the second band is as follows: l ═ p2 × (s2)2+q2×(s2)+m2
p2＝1624
q2＝-2384
m2＝930.8
The correlation coefficient was 0.9458, and the standard error was 49.09.
As shown in fig. 4, the initial model of the sum S of the peak areas of the first band and the second band and the elongation at break L is as follows: l ═ p3 × S2+ q3 × S + m3
p3＝33.95
q3＝-341.2
m3＝902.2
Wherein, the correlation coefficient is 0.9622, and the standard error is 41.17.
As shown in fig. 8, the initial model of the ratio R between the sum of the peak areas of the first band and the second band and the total area and the elongation at break L is as follows:
L＝p4×R2+q4×R+m4
p4＝2.458
q4＝-102.8
m4＝1147
the correlation coefficient was 0.9681, and the standard error was 36.88.
In this embodiment, the total number of the sample points of the training sample point set is 63, and a plurality of abnormal sample points are removed from each model, that is, paired abnormal points and obvious outlier sample points are removed, until the correlation coefficient of the model is greater than a first threshold, the confidence interval is greater than a second threshold, and the number of removed sample points is the minimum.
As in the case of the 2.6 + -0.25 area signal, there are paired outliers in both the S1-L and S-L models, i.e., the paired outliers have elongation at break test values of one greater than 250% and the other less than 250%. The pair abnormal points in the S1-L model are sample points 16 and 29, and the pair abnormal points in the S-L model are sample points 31 and 44.
The influence of the paired abnormal points on the model effect is destructive, and the paired abnormal points belong to different parts of the same PE plastic pipe, so that the PE plastic pipe has poor performance uniformity and has prediction distortion caused by abnormal factors. The PE plastic pipe can be considered to be not in accordance with the physical performance standard requirement, so that the paired abnormal points can be removed under the condition that the model prediction result is ensured to be accurate, namely the model prediction result meets the confidence interval. In addition, the proportion of the pair-abnormal points to be deleted is preferably not more than 6%. The specific rejection number can be calculated according to a statistical formula table look-up so as to accord with a set confidence interval. In this embodiment, the number of pairs of outlier culling in each model is not more than two.
In addition, obvious outlier sampling points exist in each model, and can be eliminated under the condition that the confidence interval and the correlation coefficient of the model are not influenced. In order to continuously ensure the accuracy of the model, the rejected obvious outlier sample points can be screened and compared again in the verification and maintenance of the subsequent model.
In this embodiment, the model with the least number of removed abnormal samples, the highest correlation coefficient, and the smallest standard error is R-L, that is, the R-L is used as the optimal model. The fitting degree of the ratio of the spectrum signals of the unknown components in the sampling points and the elongation at break is optimal, the actual meaning of the correlation factor reflects the spectrum information of the unknown components and also includes the spectrum information of the known components, and the actual condition of the components of the PE material can be reflected to the maximum extent. The R-L model can then be validated using the validation set of samples.
As shown in table 4 below, the number of verification spots in the verification spot set is 60, wherein the elongation at break test value of the 28 th spot is greater than 250%, i.e. the standard requirement should be greater than 250%, but the predicted value of the elongation at break is less than 250%, which is different from the standard requirement, and therefore, the conclusion of the 28 th spot is distorted and is a distorted spot. However, because the distorted sampling point has an abnormal sampling point with the same data as the distorted sampling point, the distorted sampling point does not need to be supplemented into the training sampling point set, the model verification is passed, and the R-L model is judged to be the correlation model.
TABLE 4
If the number of the distorted sampling points which need to be supplemented to the training sampling point set exceeds 5% of the verification sampling point set, the confidence interval of the coefficient 95% in the fitting model is damaged, and data fitting modeling needs to be carried out according to a new training sampling point set after the distorted sampling points are supplemented, so that a new R-L model is obtained and serves as an optimal model.
In other embodiments, if the number of distorted samples supplemented to the training sample set does not exceed 5% of the validation sample set, i.e., the 95% confidence interval in the coefficients of the fit model is not corrupted, its presence causes a decrease in the correlation coefficient of the model, i.e., the correlation coefficient is below 0.9. At the moment, the R-L model cannot cover a newly added distortion sample, the fitted R-L model is established to have failure risk, other related factors with higher related coefficients are added while the original two-dimensional model is kept, and the three-dimensional fitted model is established according to a new training sample set.
In other embodiments, the statistical prediction distortion result set is shown in table 5 below, where 60 groups of verification sample sets have 7 distortion samples, where 4 distortion samples have one removed abnormal sample and have the same data, and the other 3 distortion samples have no corresponding abnormal sample, so that the 3 distortion samples are added to the original training sample set to obtain a new training sample set, that is, the new training sample set includes 66 samples. And establishing a three-dimensional fitting model according to the new training sample point set to serve as a new optimal model.
TABLE 5
| Validation sample | Elongation at break test value | Prediction value | Standard requirements | Conclusion |
| Newly added sample | 682.76％ | 179.05％ | ＞250％ | Distortion |
| Reject sample | 741.14％ | 165.77％ | ＞250％ | Distortion |
| Reject sample | 621.36％ | 94.01％ | ＞250％ | Distortion |
| Reject sample | 596.96％ | 144.06％ | ＞250％ | Distortion |
| Reject sample | 709.56％ | 177.66％ | ＞250％ | Distortion |
| Newly added sample | 294.59％ | 79.53％ | ＞250％ | Distortion |
| Newly added sample | 365.28％ | 126.49％ | ＞250％ | Distortion |
Optionally, the other correlation factor added is the correlation factor with the highest correlation coefficient except for R. As shown in table 6 below, in this embodiment, the newly added correlation factor is s1, and the obtained new optimal model is an s1-R1-L interpolation model, which can be verified again.
As shown in fig. 9, the interpolation model belongs to actual interpolation, and the correlation coefficient thereof is theoretically infinitely close to 1. In the later verification process, if the difference model is found to have the structural characteristics of the regression curve, the difference model can be reduced to a two-dimensional fitting model by deleting the correlation factor s1, and the verification is performed again until the correlation model is obtained.
TABLE 6
According to the identification method, based on the fact that different unknown components in different PE plastic pipes are different, the elongation at break of the PE plastic pipes are also obviously different, the change of the components can be accurately identified through infrared spectroscopy, complex calculation steps are not needed, a mathematical model is established by using a principal component analysis and regression analysis method in chemometrics, the mathematical model is verified and maintained through a reasonable data processing method, the accuracy of model prediction data is improved, and the elongation at break of the PE plastic pipes for communication is quantitatively predicted through infrared spectroscopy.
The present invention is not limited to the above-described embodiments, and it will be apparent to those skilled in the art that various modifications and improvements can be made without departing from the principle of the present invention, and such modifications and improvements are also considered to be within the scope of the present invention.
Claims (10)
1. A PE plastic pipe breaking elongation recognition method based on infrared spectrum is characterized by comprising the following steps:
selecting a plurality of PE plastic pipe samples, collecting infrared spectrum data and a breaking elongation test value of at least one sample point of each PE plastic pipe, and randomly dividing the plurality of PE plastic pipe samples into a training sample point set and a verification sample point set;
extracting a plurality of correlation factors of the elongation at break according to the infrared spectrum data of the training sample set, and acquiring the value of each correlation factor;
after eliminating the correlation factors of which the variance is greater than a preset variance critical value, respectively performing data fitting modeling on each residual correlation factor and the elongation at break to obtain a plurality of initial models;
and selecting an optimal model from the plurality of initial models, verifying by using a verification sample set, and identifying the elongation at break of the PE plastic pipe to be detected as an associated model after the verification is passed.
2. A method of identifying elongation at break of PE plastic tubing based on infrared spectroscopy as recited in claim 1 wherein said collecting infrared spectroscopy data for at least one sample point of each PE plastic tubing comprises:
sampling each sample point respectively, and carrying out the sampling with the wave band of 4000-400cm-1Detecting the infrared spectrum to obtain a plurality of characteristic peak wave bands;
dividing a plurality of characteristic peak bands into known component bands and unknown component bands, and respectively integrating to obtain the peak area of each band.
3. A method of identifying elongation at break of PE plastic tubing based on infrared spectroscopy as claimed in claim 2 wherein extracting a plurality of correlation factors for elongation at break from the infrared spectroscopy data of the set of training samples comprises:
taking the peak area of each known component wave band as a correlation factor;
taking the peak area of each unknown component wave band as a correlation factor respectively;
taking the sum of peak areas of all known component wave bands as a correlation factor;
taking the sum of peak areas of all unknown component wave bands as a correlation factor;
the sum of the peak areas of all known component wave bands and all unknown component wave bands is taken as the total area, and the ratio of the sum of the peak areas of all unknown component wave bands to the total area is taken as a correlation factor.
4. A method of identifying a PE plastic tube elongation at break based on infrared spectroscopy as claimed in claim 3 wherein:
the unknown component wave bands comprise a first wave band 1900-950cm-1And a second band of 650-600cm-1The known component bands comprise a third band of 3100--1And a fourth wavelength band of 950-670cm-1。
5. A method of identifying infrared spectroscopy based elongation at break of PE plastic tubing as claimed in claim 1 wherein said performing data fitting modeling further comprises: and eliminating abnormal sample points, and counting the number of the abnormal sample points eliminated by each initial model.
6. A method for identifying PE plastic tube elongation at break based on infrared spectroscopy as claimed in claim 5 wherein selecting the optimal model from a plurality of initial models comprises:
calculating a correlation coefficient and a standard error of each initial model;
and selecting the model with the least number of removed abnormal sampling points, the highest correlation coefficient and the smallest standard error as the optimal model.
7. A method of identifying PE plastic tube elongation at break based on infrared spectroscopy as claimed in claim 5 wherein the principle of rejecting abnormal sample points is:
rejecting pairwise outliers and obvious outlier sample points until the correlation coefficient of the model is larger than a first threshold, the confidence interval is larger than a second threshold, and the number of rejected sample points is minimum;
the pair of outliers is defined as: and if the difference value of the same related factor values of two sampling points on the same PE plastic pipe is within a preset range, and one of the elongation at break test values of the two sampling points is greater than a standard value and the other is less than the standard value, taking the two sampling points as paired abnormal points.
8. A method of identifying PE plastic tube elongation at break based on infrared spectroscopy as claimed in claim 6 wherein the validation with a set of validation samples comprises:
inputting infrared spectrum data of the verification sample point set into the optimal model to obtain a predicted value of the elongation at break of the verification sample point set;
comparing the predicted value of the elongation at break of the verification sample set with the corresponding test value; taking a sample point of which one of the corresponding predicted value and the test value is greater than the standard value and the other one is less than the standard value as a distortion sample point;
when no distortion sampling point exists, taking the optimal model as a correlation model; otherwise, comparing each distorted sample point with the removed abnormal sample points respectively;
if each distorted sample point has an abnormal sample point and has the same data, the optimal model is used as a correlation model;
otherwise, judging that the verification fails, supplementing the distortion sampling points without corresponding abnormal sampling points into the training sampling point set, and carrying out secondary modeling.
9. A method for identifying an elongation at break of a PE plastic tube based on infrared spectroscopy as claimed in claim 8 wherein the secondary modeling specifically comprises:
if the number of the distortion sampling points supplemented into the training sampling point set exceeds 5% of the verification sampling point set, performing data fitting modeling according to the new training sampling point set to obtain a new optimal model;
otherwise, increasing the correlation factor, and establishing a three-dimensional fitting model according to the new training sample set to serve as a new optimal model.
10. A method of identifying PE plastic tube elongation at break based on infrared spectroscopy as claimed in claim 9 further comprising, after obtaining a new optimal model:
manufacturing a secondary verification sample point set, and acquiring infrared spectrum data and a breaking elongation test value of the secondary verification sample point set;
and carrying out secondary verification on the new optimal model by utilizing the secondary verification sample set until a correlation model is obtained.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202011624968.3A CN112836340B (en) | 2020-12-31 | 2020-12-31 | Infrared spectrum-based PE plastic pipe elongation at break recognition method |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202011624968.3A CN112836340B (en) | 2020-12-31 | 2020-12-31 | Infrared spectrum-based PE plastic pipe elongation at break recognition method |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN112836340A true CN112836340A (en) | 2021-05-25 |
| CN112836340B CN112836340B (en) | 2024-02-02 |
Family
ID=75924455
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202011624968.3A Active CN112836340B (en) | 2020-12-31 | 2020-12-31 | Infrared spectrum-based PE plastic pipe elongation at break recognition method |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN112836340B (en) |
Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US20130305664A1 (en) * | 2010-10-20 | 2013-11-21 | Starlinger & Co Gesellschaft M.B.H. | Coated fabric, a bag produced therefrom, a packaging machine for bags and a method for filling the bags |
| WO2018010352A1 (en) * | 2016-07-11 | 2018-01-18 | 上海创和亿电子科技发展有限公司 | Qualitative and quantitative combined method for constructing near infrared quantitative model |
| CN108595343A (en) * | 2018-06-04 | 2018-09-28 | 北京汉能光伏投资有限公司 | The test method and device of application program |
| CN111811932A (en) * | 2020-07-16 | 2020-10-23 | 西安航天化学动力有限公司 | Near infrared spectrum analysis method for uniaxial tension mechanical property of composite solid propellant |
-
2020
- 2020-12-31 CN CN202011624968.3A patent/CN112836340B/en active Active
Patent Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US20130305664A1 (en) * | 2010-10-20 | 2013-11-21 | Starlinger & Co Gesellschaft M.B.H. | Coated fabric, a bag produced therefrom, a packaging machine for bags and a method for filling the bags |
| WO2018010352A1 (en) * | 2016-07-11 | 2018-01-18 | 上海创和亿电子科技发展有限公司 | Qualitative and quantitative combined method for constructing near infrared quantitative model |
| CN108595343A (en) * | 2018-06-04 | 2018-09-28 | 北京汉能光伏投资有限公司 | The test method and device of application program |
| CN111811932A (en) * | 2020-07-16 | 2020-10-23 | 西安航天化学动力有限公司 | Near infrared spectrum analysis method for uniaxial tension mechanical property of composite solid propellant |
Non-Patent Citations (2)
| Title |
|---|
| 谢锦春等: "乙烯醋酸乙烯酯共聚物断裂伸长率的近红外光谱分析方法", 光谱学与光谱分析, vol. 38, no. 10, pages 113 - 114 * |
| 阿依古丽·塔什波拉提等: "棕色彩棉内在品质近红外光谱无损检测方法研究", 中国棉花, vol. 45, no. 5, pages 30 - 35 * |
Also Published As
| Publication number | Publication date |
|---|---|
| CN112836340B (en) | 2024-02-02 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| US8989888B2 (en) | Automatic fault detection and classification in a plasma processing system and methods thereof | |
| Millan et al. | Image analysis‐based modelling for flower number estimation in grapevine | |
| CN113344134A (en) | Data acquisition abnormity detection method and system for low-voltage power distribution monitoring terminal | |
| WO2020029851A1 (en) | Workflow-based vibration spectrum analysis model optimization method | |
| US9142014B2 (en) | System and method for identifying systematic defects in wafer inspection using hierarchical grouping and filtering | |
| CN113971425A (en) | Abnormity analysis method, abnormity analysis device and storage medium | |
| CN112000081B (en) | Fault monitoring method and system based on multi-block information extraction and Mahalanobis distance | |
| CN105139022A (en) | Plastic identification model establishment method via near-infrared hyper-spectral image technology | |
| CN115035110A (en) | Yarn hairiness quantitative detection method based on pit cutting and chain code tracking | |
| CN108519760A (en) | A kind of Primary Processing stable state recognition methods based on detection of change-point theory | |
| CN117556714A (en) | Preheating pipeline temperature data anomaly analysis method for aluminum metal smelting | |
| CN116777917B (en) | Defect detection method and system for optical cable production | |
| CN112836340A (en) | Infrared spectrum-based PE plastic pipe elongation at break identification method | |
| CN114062305B (en) | Single grain variety identification method and system based on near infrared spectrum and 1D-In-Resnet network | |
| CN114067169A (en) | Raman spectrum analysis method based on convolutional neural network | |
| JP4948238B2 (en) | Distribution analysis method and apparatus, abnormal equipment estimation method and apparatus, program for causing computer to execute distribution analysis method or abnormal equipment estimation method, and computer-readable recording medium recording the program | |
| US11507961B2 (en) | Fabricated data detection method | |
| Wu et al. | Significance test in operational ROC analysis | |
| CN113051307A (en) | Alarm signal analysis method, equipment, storage medium and device | |
| CN117711593B (en) | Intelligent pharmacy medicine inlet and outlet management system | |
| CN116990450B (en) | Defect detection method and system for cornea shaping mirror | |
| CN113177165B (en) | Inspection officer personalized recommendation method based on improved KNN | |
| CN113870265B (en) | Industrial part surface defect detection method | |
| CN117556274B (en) | Temperature data anomaly analysis method for heat pipe backboard | |
| CN117523324B (en) | Image processing method and image sample classification method, device and storage medium |
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 |