CN114689187A - Wavelength point screening method based on multi-sensor spectral data - Google Patents
Wavelength point screening method based on multi-sensor spectral data Download PDFInfo
- Publication number
- CN114689187A CN114689187A CN202210328651.8A CN202210328651A CN114689187A CN 114689187 A CN114689187 A CN 114689187A CN 202210328651 A CN202210328651 A CN 202210328651A CN 114689187 A CN114689187 A CN 114689187A
- Authority
- CN
- China
- Prior art keywords
- wavelength
- data
- points
- point
- screening
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Pending
Links
- 230000003595 spectral effect Effects 0.000 title claims abstract description 60
- 238000012216 screening Methods 0.000 title claims abstract description 47
- 238000000034 method Methods 0.000 title claims abstract description 43
- 238000012512 characterization method Methods 0.000 claims abstract description 41
- 238000001228 spectrum Methods 0.000 claims description 51
- 238000004364 calculation method Methods 0.000 claims description 5
- 230000000717 retained effect Effects 0.000 claims description 5
- 238000002790 cross-validation Methods 0.000 claims description 3
- 238000004458 analytical method Methods 0.000 abstract description 10
- 238000010183 spectrum analysis Methods 0.000 abstract description 9
- 238000005516 engineering process Methods 0.000 abstract description 8
- 238000002329 infrared spectrum Methods 0.000 abstract description 5
- 239000000523 sample Substances 0.000 description 22
- 238000001514 detection method Methods 0.000 description 4
- 230000000694 effects Effects 0.000 description 3
- 238000004497 NIR spectroscopy Methods 0.000 description 2
- 238000007405 data analysis Methods 0.000 description 2
- 239000000463 material Substances 0.000 description 2
- 239000000203 mixture Substances 0.000 description 2
- 238000002360 preparation method Methods 0.000 description 2
- 230000004075 alteration Effects 0.000 description 1
- 230000009286 beneficial effect Effects 0.000 description 1
- 239000002131 composite material Substances 0.000 description 1
- 238000010586 diagram Methods 0.000 description 1
- 238000013139 quantization Methods 0.000 description 1
- 238000005464 sample preparation method Methods 0.000 description 1
- 230000035945 sensitivity Effects 0.000 description 1
- 239000000126 substance Substances 0.000 description 1
- 238000006467 substitution reaction Methods 0.000 description 1
Images
Classifications
-
- G—PHYSICS
- G01—MEASURING; TESTING
- G01J—MEASUREMENT OF INTENSITY, VELOCITY, SPECTRAL CONTENT, POLARISATION, PHASE OR PULSE CHARACTERISTICS OF INFRARED, VISIBLE OR ULTRAVIOLET LIGHT; COLORIMETRY; RADIATION PYROMETRY
- G01J9/00—Measuring optical phase difference; Determining degree of coherence; Measuring optical wavelength
Landscapes
- Physics & Mathematics (AREA)
- Spectroscopy & Molecular Physics (AREA)
- General Physics & Mathematics (AREA)
- Spectrometry And Color Measurement (AREA)
Abstract
The invention relates to a near infrared spectrum analysis technology, and discloses a wavelength point screening method based on multi-sensor spectral data, aiming at reducing the spectral data amount, improving the spectral analysis efficiency and simultaneously improving the accuracy and stability of a spectral model. The method comprises the steps of firstly importing original multi-sensor spectral data, classifying the spectral data according to the number of sensors to obtain single-quantity data, then modeling each single-quantity data by a partial least square method to obtain a root mean square error value of each model, calculating a characterization coefficient of each single-quantity data by combining the root mean square error value, then screening the number of wavelength points of each single-quantity data by combining the number of wavelength points of the single-quantity data and the characterization coefficient, finally screening the wavelength points by combining the number of wavelength points of each single-quantity data, and recombining each single-quantity data after screening to obtain multiple-quantity data.
Description
Technical Field
The invention relates to a near infrared spectrum analysis technology, in particular to a wavelength point screening method based on multi-sensor spectral data.
Background
In recent years, the near infrared spectrum analysis technology is developed rapidly and is applied to a plurality of fields such as chemical industry, pharmacy, military industry, food and the like. The near infrared spectrum technology belongs to the molecular spectrum technology, can indicate material composition and property information on the molecular level, and obtains very high benefit no matter for economic or social influence, thereby having great development potential.
However, most of the existing material composition and property information detection is mainly carried out by using a large laboratory near infrared spectroscopy instrument, although the methods have high quantitative accuracy and sensitivity, the required equipment has huge volume, expensive equipment cost, long sample preparation time and strict sample preparation method, the detection equipment and the sample preparation need professional operation, the detection environment is fixed, the analysis time is long, and the method is not suitable for field detection and is not convenient for popularization and use.
Along with the development of portable near infrared spectroscopy technology, the mainstream large-scale near infrared spectrometer equipment in the market all develops towards small in size, low price's portable direction, and portable near infrared spectrometer in the existing market is mainly single sensor equipment, and single sensor equipment receives the restriction of sensor technology, and it covers the wave band range very limited, and the spectral data stability of gathering the acquisition is relatively poor, and spectral data takes place the skew easily, and then causes the problem that the prediction effect is unstable easily, and the rate of accuracy is low.
In order to increase the coverage of the sensor band, the application scene of the portable near-infrared spectrometer is promoted, the portable near-infrared spectrometer with multiple sensors is produced, although the portable near-infrared spectrometer with multiple sensors can well solve various problems of single sensor equipment, due to the increase of the number of sensors of the portable near-infrared spectrometer with multiple sensors, the number of wavelength points of acquired spectral data is increased by multiple times compared with the number of wavelength points of spectral data of an original single sensor, the acquired data is relatively redundant, and the data information with small correlation with a sample is contained, so that the conventional spectral data analysis method cannot be applied to the spectral data analysis of the multiple sensors, and meanwhile, due to the increase of the spectral data quantity, the analysis efficiency of the portable near-infrared spectrum analysis technology is greatly reduced.
Disclosure of Invention
The technical problem to be solved by the invention is as follows: a wavelength point screening method based on multi-sensor spectral data is provided, and aims to reduce spectral data volume, improve spectral analysis efficiency and improve accuracy and stability of a spectral model.
The technical scheme adopted by the invention for solving the technical problems is as follows:
a wavelength point screening method based on multi-sensor spectral data comprises the following steps:
s1, importing original multi-sensor spectrum data, and classifying the spectrum data according to the number of sensors to obtain single data;
s2, performing partial least square modeling on each group of single data to obtain a root mean square error value of each model;
s3, calculating each single data characterization coefficient group by combining the root mean square error value;
s4, wavelength point number screening is carried out on each group of single data by combining the single data wavelength point number and the characterization coefficient;
and S5, screening the wavelength points by combining the number of the wavelength points of each group of single data, and recombining each group of single data after screening to obtain multiple data.
Further, in step S2, the root mean square error is generated by performing cross validation using leave-one-out method, where the expression is:
wherein M is the number of original samples, yiIs a sample xiThe calibration value of (a) is set,
is a sample xiThe predicted value of (2).
Further, in step S3, the calculating each set of single quantity data characterization coefficients by combining the root mean square error value specifically includes:
the smaller the root mean square error RMESCV value of the spectrum model is, the higher the characterization coefficient is, and the calculation mode is as follows:
wherein alpha isnIs a characterization coefficient of the nth set of single quantity data, AnThe root mean square error RMESCV value of the spectral model established for the nth set of single-dose data.
Further, in step S4, the screening the number of wavelength points of each set of single data by combining the number of wavelength points of the single data and the characterization coefficient specifically includes:
the higher the characterization coefficient corresponding to each group of single data is, the stronger the sample characterization capability of the single data is, the more the number of wavelength points to be retained in the wavelength point number screening is, and the calculation mode is as follows:
Xi＝Round(αi*100％*m)
wherein alpha isiIs the characteristic coefficient of the ith group of single quantity data, m is the sum of the number of wavelength points needing to be screened from all the single quantity data, Round () is a rounding function, XiThe number of wavelength points needed to be reserved in the ith group of single data.
Further, in step S5, the screening the wavelength points in combination with the number of wavelength points of each single-volume data includes:
and (3) verifying the wavelength points in each group of single data one by adopting a stepwise regression method, eliminating the wavelength points with weak representation capability one by one through a root mean square error value, and keeping the wavelength points with strong representation capability.
Further, step S5 specifically includes:
s51, randomly selecting X from the ith group of single dataiOne wavelength point is used as an initial wavelength point, and the rest (z-X) isi) Taking the wavelength points as iteration wavelength points, wherein i is 1;z is the number of wavelength points in the ith group of single data;
s52, carrying out spectrum modeling on the initial wavelength point, and calculating to obtain the root mean square error RMSECV value T of the spectrum model0；
S53, selecting 1 wavelength point from the iterative wavelength points, introducing the wavelength point into the initial wavelength points, and replacing the 1 st to the Xth wavelength points in the initial wavelength points one by oneiThe number of the initial wavelength points is always kept as XiA plurality of;
s54, carrying out spectrum modeling on the replaced initial wavelength points, and calculating to obtain the root mean square error RMSECV value T of the spectrum modelsj(j＝1，2，……Xi)；
S55, comparison T0And TjAnd performing corresponding operations:
if TjAre all greater than T0If the iterative wavelength point is added, the root mean square error RMSECV value of the spectrum model is increased, the characterization capability of the corresponding spectrum model sample is weakened, the iterative wavelength point is abandoned, and the initial wavelength point is unchanged;
if T isjAre all equal to T0If the initial wavelength point is not changed, the iterative wavelength point is discarded if the introduction of the iterative wavelength point does not influence the spectrum model;
if TjIs less than T0If the corresponding initial wavelength point is replaced, the root mean square error RMSECV value of the spectrum model is reduced after the iterative wavelength point is added, the characterization capability of the corresponding spectrum model sample is enhanced, the iterative wavelength point is introduced, and meanwhile, in order to ensure that the number of the initial wavelength points is not changed, the T is addedjIs less than T0Under the condition, the corresponding initial wavelength point is removed;
if TjIn which there are a plurality of values smaller than T0Then the iterative wavelength point is introduced and eliminated simultaneously to make TjWhen the minimum value is reached, the corresponding initial wavelength point is obtained;
s56, repeating the steps S53-S55, introducing the remaining iteration wavelength points one by one until all the iteration wavelengths are traversed, and finally screening to obtain wavelength points which are the screening wavelength points of the ith group of single data;
s57, assigning i to i +1, and returning to the step S51 until all the single-quantity data wavelength point screening is completed;
and S58, recombining the wavelength points screened out by all the single data to obtain multiple data.
The invention has the beneficial effects that:
according to the method, wavelength points of each group of single data are screened through spectral model parameters of the single data, so that original multi-sensor spectral data are reduced to the number of the wavelength points with the same single data on the premise of keeping sample representation, the multi-sensor spectral data can be subjected to a common spectral analysis method, the spectral data amount is reduced, the spectral analysis efficiency is improved, and meanwhile the accuracy and the stability of a spectral model can be greatly improved.
Drawings
FIG. 1 is a flow chart of a method for wavelength point screening based on multi-sensor spectral data in an embodiment of the invention;
fig. 2 is a schematic arrangement diagram of sensors of a portable multi-sensor near-infrared spectrometer in an embodiment of the invention.
Detailed Description
The invention aims to provide a wavelength point screening method based on multi-sensor spectral data, and aims to reduce the spectral data amount, improve the spectral analysis efficiency and improve the accuracy and stability of a spectral model. The method comprises the steps of firstly importing original multi-sensor spectral data, classifying the spectral data according to the number of sensors to obtain single-quantity data, then modeling each group of single-quantity data by a partial least square method to obtain a root mean square error value of each model, calculating a characterization coefficient of each group of single-quantity data by combining the root mean square error value, then screening the number of wavelength points of each group of single-quantity data by combining the number of wavelength points of the single-quantity data and the characterization coefficient, finally screening the wavelength points by combining the number of wavelength points of each group of single-quantity data, and recombining each group of single-quantity data after screening to obtain multiple-quantity data. According to the method, the wavelength points of each group of single data are screened through the spectral model parameters of the single data, so that the spectral data volume can be reduced, the spectral analysis efficiency is improved, and meanwhile, the accuracy and the stability of the spectral model can be greatly improved.
Example (b):
as shown in fig. 1, the method for screening wavelength points based on multi-sensor spectral data in the present embodiment includes the following steps:
step 101: importing original multi-sensor spectral data, and classifying the spectral data according to the number of sensors to obtain single data;
in this step, the essence of the original multi-sensor spectrum data is the collection of each single-sensor spectrum data, the number of the wavelength data points covered by the original multi-sensor spectrum data is the sum of the number of the wavelength points of each single sensor, the original multi-sensor spectrum data is subjected to single quantization processing and then is decomposed into a plurality of single-sensor data, and then the original multi-sensor spectrum data can be subjected to modeling analysis by adopting a conventional single-sensor spectrum analysis method, so that the analysis efficiency is effectively improved.
As a specific example, as shown in fig. 2, the portable multi-sensor near-infrared spectrometer adopted in this example includes 4 sensors, which are respectively 4 sensors of different models, wherein the wavelength ranges of the 4 sensors are 1255nm to 1450nm, 1455nm to 1650nm, 1655nm to 1850nm, 1855nm to 2050nm, the resolution is 5nm, and the distribution of the wavelength points is a wavelength-sharing mode. The wave band range of the sensor 1 is 1255 nm-1450 nm, and the number of the contained wavelength points is M11+ (1450-1255)/5-40; the wave band range of the sensor 2 is 1455 nm-1650 nm, and the number of the included wavelength points is M21+ (1650-1455)/5 is 40; the wave band range of the sensor 3 is 1655 nm-1850 nm, and the number of the included wavelength points is M31+ (1850) -1655)/40; the wave band of the sensor 4 ranges from 1855nm to 2050nm, and the number of the included wavelength points is M41+ (2050-.
From the above, the original spectrum data of the multi-sensor is formed by splicing 4 sections of spectrum data, the specific wavelength range is 1255nm to 2050nm, and the total wavelength contains M5＝M1+M2+M3+M4160 wavelength points. Classifying the multi-sensor spectral data according to the number of the sensors to obtain 4 groups of single-quantity data which are respectively set as single-quantity dataData P1，P2，P3，P4Corresponding to the spectral data of the sensors 1, 2, 3, 4.
Step 102: performing partial least square modeling on each group of single data to obtain a root mean square error value of each model;
in the step, the partial least square method is the most common linear fitting modeling method with excellent effect in spectral analysis; the root mean square error is the most common model index in the spectral modeling analysis process, and the index can directly reflect the quality of the spectral model.
Taking the example of fig. 2 as an example, partial least squares modeling is performed on single quantity data of the sensor 1, the sensor 2, the sensor 3, and the sensor 4, and root mean square error values of the 4 models are further calculated, where a root mean square error (rmesccv) actually reflects a deviation quantity relationship between a predicted calibration value and a predicted value in a model, and is generated by performing cross validation by using a leave-one-out method, and an expression is shown as follows:
assuming that there are M pieces of original spectral data, there are M corresponding original samples, and one sample X is taken from each samplei(the ith sample, i ═ 1, 2, … …, M), modeling with the remaining (M-1) spectral sample data, and predicting the sample X taken with the modeliY of (A) to (B)iValue (calibration value) to obtain the predicted value of the sample
And calculating root mean square errors of all the generated predicted values to obtain an RMSECV value, wherein the RMSECV index has higher accuracy in judging the quality of the model than the conventional MSE and MAE indexes because all samples of the original spectral data are traversed by one-out method, and is more stable and more applicable.
Step 103: calculating each group of single data characterization coefficients by combining the root mean square error value;
in this step, because the spectral data of different sensors cover different wave band ranges, and then the ability of the sensors to characterize samples will also differ, if the same number of wavelength points is simply screened for each group of single data, the problem that the characterization ability of the samples is weak and the spectrum prediction analysis ability is affected is easily caused, the model effect of each group of single data is determined through the root mean square error value, the characterization ability of each group of single data on the samples is further reflected, and then the characterization coefficients are calculated to perform corresponding processing on each group of single data, so that the problem can be effectively solved.
In the present embodiment, the single amount data P is set1Is characterized by a1The RMESCV value of the spectral model established by the spectral data is A1Single amount of data P2Is characterized by a2The RMESCV value of the spectral model established by the spectral data is A2Single amount of data P3Is characterized by a3The RMESCV value of the spectral model established by the spectral data is A3Single amount of data P4Is characterized by a4The RMESCV value of the spectral model established by the spectral data is A4In the capability of the spectrum model for characterizing the sample, the smaller the RMESCV value is, the smaller the deviation between the calibration value and the predicted value is, namely the spectrum model characterization capability is better, the higher the characterization coefficient is, the more the single data with the smaller RMESCV value is, and the specific calculation formula for obtaining the characterization coefficient is:
step 104: screening the number of wavelength points of each group of single data by combining the number of wavelength points of the single data and the characterization coefficient;
in this step, the higher the characterization coefficient corresponding to each set of single quantity data is, the stronger the sample characterization capability of the single quantity data is, the more the number of wavelength points to be retained when the wavelength point number screening is performed is, and the spectral sample characterization capability can be retained to the greatest extent by calculating the number of wavelength points to be retained by each set of single quantity data through the characterization coefficient and the number of wavelength points.
In this embodiment, each set of single data includes 40 wavelength points, that is, each set of single sensor spectrum data includes 40 wavelength points, and the multi-sensor spectrum data includes 160 wavelength points, so that the multi-sensor spectrum data is simplified and can be used in a conventional single sensor spectrum analysis method, the wavelength points of the multi-sensor spectrum data need to be screened to 40, and since the sample characterization capabilities of each set of single data are different, the specific number of wavelength points that each set of single data should contribute can be calculated by combining the characterization coefficients, and the single data P is set1，P2，P3，P4The specific number of wavelength points to be contributed is X1，X2，X3，X4And combining the characterization coefficients to know that the specific number of the wavelength points is as follows:
where Round () is a rounding function, usually (X)1+X2+X3+X4) If present, (X) 401+X2+X3+X4) In the case of 41 or 42, in X1，X2，X3，X4Randomly selecting one or two wavelength points to discard, so that (X)1+X2+X3+X4) This is always true for 40.
Step 105: screening wavelength points by combining the number of the wavelength points of each group of single data, and recombining each group of single data after screening to obtain multiple data;
in this step, after the number of wavelength points in each group of single quantity data is determined, specific wavelength points in each group of single quantity data need to be further determined, wavelength points in each group of single quantity data are verified one by adopting a stepwise regression method, wavelength points with weak representation capability are eliminated one by one through a root mean square error value, and wavelength points with strong representation capability are reserved.
In the present embodiment, the data P is present in a single amount1For example, after screeningThe number of wavelength points of (2) is X1If it is, in the single amount of data P1In selecting X randomly1One wavelength point is used as an initial wavelength point, and the rest (40-X) is1) The wavelength points are used as iteration wavelength points, and the specific stepwise regression step is as follows:
(1) carrying out spectrum modeling on the initial wavelength point, and calculating to obtain the RMSECV value T of the spectrum model0；
(2) Selecting 1 wavelength point from the iterative wavelength points, introducing the selected wavelength point into the initial wavelength points, and replacing the 1 st to the Xth wavelength points in the initial wavelength points one by one1A number of wavelength points always keeping the initial wavelength point as X1A plurality of;
(3) carrying out spectrum modeling on the replaced initial wavelength points, and calculating to obtain the RMSECV values T of the spectrum modelsj(j＝1，2，……X1)；
(4) Comparison T0And TjSize of (c), if TjGreater than T0If so, the RMSECV value of the spectrum model is increased after the iterative wavelength point is added, the characterization capability of the corresponding spectrum model sample is weakened, and the iterative wavelength point is discarded; if T isjIs always equal to T0If the iterative wavelength point is not introduced into the spectral model, discarding the iterative wavelength point; if T isjIs less than T0If the initial wavelength point is removed, the RMSECV value of the spectrum model is reduced after the iterative wavelength point is added, the corresponding spectrum model sample characterization capability is enhanced, the iterative wavelength point is introduced, and meanwhile, in order to ensure that the number of the initial wavelength points is not changed, T is addedjIs less than T0Under the condition, the corresponding initial wavelength point is removed; if TjIn which there are a plurality of values smaller than T0The iterative wavelength point is introduced when TjWhen the minimum value is reached, the corresponding initial wavelength point is removed;
(5) repeating the steps (2), (3) and (4), introducing the rest iterative wavelength points one by one until all iterative wavelength points finish introducing comparison operation, and gradually regressing to obtain the selected spectral wavelength points which are the single data P1The screening wavelength points of (1).
In the same way, the method for preparing the composite material,for single amount data P2，P3，P4And carrying out the same stepwise regression treatment to obtain corresponding screening wavelength points, and recombining each group of single data after screening to obtain multiple data, wherein the multiple data is the multi-sensor spectrum data after screening.
Finally, it should be noted that the above-mentioned embodiments are only preferred embodiments and are not intended to limit the present invention. It should be noted that those skilled in the art can make various changes, substitutions and alterations herein without departing from the spirit of the invention and the scope of the appended claims.
Claims (6)
1. A wavelength point screening method based on multi-sensor spectral data is characterized by comprising the following steps:
s1, importing original multi-sensor spectrum data, and classifying the spectrum data according to the number of sensors to obtain single data;
s2, performing partial least square modeling on each group of single data to obtain a root mean square error value of each model;
s3, calculating each single data characterization coefficient group by combining the root mean square error value;
s4, wavelength point number screening is carried out on each group of single data by combining the single data wavelength point number and the characterization coefficient;
and S5, screening the wavelength points by combining the number of the wavelength points of each group of single data, and recombining each group of single data after screening to obtain multiple data.
2. The method of claim 1, wherein the wavelength point screening method based on multi-sensor spectrum data,
in step S2, the root mean square error is generated by performing cross validation using leave-one method, where the expression is:
wherein M is the number of original samples, yiIs a sample xiThe calibration value of (a) is set,
is a sample xiThe predicted value of (2).
3. The method of claim 1, wherein the wavelength point screening method based on multi-sensor spectrum data,
in step S3, the calculating each single quantity data characterization coefficient by combining the root mean square error value specifically includes:
the smaller the root mean square error RMESCV value of the spectrum model is, the higher the characterization coefficient is, and the calculation mode is as follows:
wherein alpha isnIs a characterization coefficient of the nth set of single quantity data, AnThe root mean square error RMESCV value of the spectral model established for the nth set of single-dose data.
4. The method of claim 1, wherein the wavelength point screening method based on multi-sensor spectrum data,
in step S4, the wavelength point number screening is performed on each set of single quantity data by combining the single quantity data wavelength point number and the characterization coefficient, and specifically includes:
the higher the characterization coefficient corresponding to each group of single data is, the stronger the sample characterization capability of the single data is, the more the number of wavelength points to be retained in the wavelength point number screening is, and the calculation mode is as follows:
Xi＝Round(αi*100％*m)
wherein alpha isiIs the characteristic coefficient of the ith group of single quantity data, m is the sum of the number of wavelength points needing to be screened from all the single quantity data, Round () is a rounding function, XiThe number of wavelength points needed to be reserved in the ith group of single data.
5. The method of claim 1, wherein the wavelength point screening method based on multi-sensor spectrum data,
in step S5, the screening of the wavelength points in combination with the number of wavelength points of each single quantity of data specifically includes:
and (3) verifying the wavelength points in each group of single data one by adopting a stepwise regression method, eliminating the wavelength points with weak representation capability one by one through a root mean square error value, and keeping the wavelength points with strong representation capability.
6. The method of claim 5, wherein the wavelength point screening method based on multi-sensor spectrum data,
step S5 specifically includes:
s51, randomly selecting X from the ith group of single dataiOne wavelength point is used as an initial wavelength point, and the rest (z-X) isi) Taking the wavelength points as iteration wavelength points, wherein i is 1; z is the number of wavelength points in the ith group of single data;
s52, carrying out spectrum modeling on the initial wavelength point, and calculating to obtain the root mean square error RMSECV value T of the spectrum model0；
S53, selecting 1 wavelength point from the iterative wavelength points, introducing the wavelength point into the initial wavelength points, and replacing the 1 st to the Xth wavelength points in the initial wavelength points one by oneiA number of wavelength points always keeping the initial wavelength point as XiA plurality of;
s54, carrying out spectrum modeling on the replaced initial wavelength points, and calculating to obtain the root mean square error RMSECV value T of the spectrum modelsj(j＝1，2，……Xi)；
S55, comparison T0And TjAnd performing corresponding operations:
if TjAre all greater than T0If the iterative wavelength point is added, the root mean square error RMSECV value of the spectral model is increased, the characterization capability of the corresponding spectral model sample is weakened, the iterative wavelength point is abandoned, and the initial time is shortenedThe initial wavelength point is unchanged;
if TjAre all equal to T0If the initial wavelength point is not changed, the iterative wavelength point is discarded if the introduction of the iterative wavelength point does not influence the spectrum model;
if TjIs less than T0If the corresponding initial wavelength point is replaced, the root mean square error RMSECV value of the spectrum model is reduced after the iterative wavelength point is added, the characterization capability of the corresponding spectrum model sample is enhanced, the iterative wavelength point is introduced, and meanwhile, in order to ensure that the number of the initial wavelength points is not changed, the T is addedjIs less than T0Under the condition, the corresponding initial wavelength point is removed;
if TjIn which there are a plurality of values smaller than T0Then the iterative wavelength point is introduced and eliminated simultaneously to make TjWhen the minimum value is reached, the corresponding initial wavelength point is obtained;
s56, repeating the steps S53-S55, and introducing the remaining iteration wavelength points one by one until all the iteration wavelengths are traversed, wherein the wavelength points obtained by screening finally are the screening wavelength points of the ith group of single data;
s57, assigning i to i +1, and returning to the step S51 until all the single-quantity data wavelength point screening is completed;
and S58, recombining the wavelength points screened out by all the single data to obtain multiple data.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202210328651.8A CN114689187A (en) | 2022-03-30 | 2022-03-30 | Wavelength point screening method based on multi-sensor spectral data |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202210328651.8A CN114689187A (en) | 2022-03-30 | 2022-03-30 | Wavelength point screening method based on multi-sensor spectral data |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN114689187A true CN114689187A (en) | 2022-07-01 |
Family
ID=82141699
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202210328651.8A Pending CN114689187A (en) | 2022-03-30 | 2022-03-30 | Wavelength point screening method based on multi-sensor spectral data |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN114689187A (en) |
-
2022
- 2022-03-30 CN CN202210328651.8A patent/CN114689187A/en active Pending
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN109145873B (en) | Spectral Gaussian peak feature extraction algorithm based on genetic algorithm | |
| CN108169165B (en) | Maltose mixture quantitative analysis method based on terahertz spectrum and image information fusion | |
| CN103235095A (en) | Water-injected meat detection method and device | |
| Shenar et al. | The Tarantula Massive Binary Monitoring-VI. Characterisation of hidden companions in 51 single-lined O-type binaries: A flat mass-ratio distribution and black-hole binary candidates | |
| CN109409350B (en) | PCA modeling feedback type load weighting-based wavelength selection method | |
| CN111965140B (en) | Wavelength point recombination method based on characteristic peak | |
| Yang et al. | J-PLUS: Stellar parameters, C, N, Mg, Ca, and [α/Fe] abundances for two million stars from DR1 | |
| CN114676636A (en) | Grassland area soil moisture rapid inversion method integrating vegetation and habitat characteristics | |
| CN113686811B (en) | Spectral data processing method based on double sensors | |
| CN111693487A (en) | Fruit sugar degree detection method and system based on genetic algorithm and extreme learning machine | |
| CN113076692B (en) | Method for inverting nitrogen content of leaf | |
| CN109839362A (en) | IR spectrum quantitative analysis method based on gradual noise-removed technology | |
| CN111795944B (en) | Near infrared spectrum wavelength screening method for improving team progress algorithm | |
| CN108120694B (en) | Multi-element correction method and system for chemical component analysis of sun-cured red tobacco | |
| CN114689187A (en) | Wavelength point screening method based on multi-sensor spectral data | |
| CN109145403B (en) | Near infrared spectrum modeling method based on sample consensus | |
| CN114062306B (en) | Near infrared spectrum data segmentation preprocessing method | |
| CN112881333B (en) | Near infrared spectrum wavelength screening method based on improved immune genetic algorithm | |
| CN114942233A (en) | Near infrared spectrum characteristic wavelength selection method, device, equipment and storage medium | |
| CN112229816B (en) | Wood elastic modulus prediction method based on OPLS-SPA-MIX-PLS | |
| CN111693488B (en) | Fruit grade classification method and system based on DS evidence theory fusion | |
| CN109145887B (en) | Threshold analysis method based on spectral latent variable confusion discrimination | |
| Ni et al. | Transferring near infrared spectral calibration models without standards via multistep wavelength selection | |
| CN115598162A (en) | Soil heavy metal content detection method based on stacked model | |
| CN114707321A (en) | Weighted prediction method based on multi-sensor spectral data |
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 |