CN104062263B - The near-infrared universal model detection method of light physical property close fruit quality index - Google Patents
The near-infrared universal model detection method of light physical property close fruit quality index Download PDFInfo
- Publication number
- CN104062263B CN104062263B CN201410328830.7A CN201410328830A CN104062263B CN 104062263 B CN104062263 B CN 104062263B CN 201410328830 A CN201410328830 A CN 201410328830A CN 104062263 B CN104062263 B CN 104062263B
- Authority
- CN
- China
- Prior art keywords
- fruit
- infrared
- index
- universal model
- spectrum
- 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.)
- Expired - Fee Related
Links
Landscapes
- Investigating Or Analysing Materials By Optical Means (AREA)
Abstract
A kind of near-infrared universal model detection method of smooth physical property close fruit quality index, including setting up near-infrared universal model and utilizing near-infrared universal model to measure fruit quality index two large divisions, it it is the near-infrared universal model set up based on the similarity of near infrared spectrum between the close variety classes fruit of light physical property, its modeling basic ideas are, Pretreated spectra is carried out after spectra collection, the common features wave-length coverage of all kind fruit is filtered out with moving window partial least square method, then in common features wavelength period scope, common features wavelength points is extracted further with SPA algorithm, existing software is finally utilized to set up PLS model or set up MLR model.It is high that the inventive method measures accuracy rate, and feasibility is strong, overcomes the drawback of the necessary classification and Detection of variety classes fruit in existing near infrared detection technology, reduce modeling cost, improve work efficiency, modeling wavelength points is few, it is adaptable to general optical filtering flap-type nir instrument.
Description
Technical field
The present invention relates to the near-infrared technical field of nondestructive testing of fruit, disposably set up near-infrared particularly to one and lead to
Detect the index of quality such as pol index, acidity index or the side of mature indicator of the multiple close fruit of smooth physical property with model simultaneously
Method.
Background technology
Near Infrared Spectroscopy Detection Technology, have non-demolition, quickly, without pre-treatment, the feature such as pollution-free, refer at fruit quality
In target Non-Destructive Testing, have been obtained for using widely.Before carrying out Non-Destructive Testing, it is required for specific material and sets up mould
Type, is predicted unknown sample the most again.Detection limit is generally 0.1%, and for fruit, in spectrum, reflection is chemistry
The information that component content is higher, such as: water, soluble solid etc., for the fruit that physicochemical characteristics is close, near infrared spectrum
Similar, thus close material to set up universal model be feasible.In the Non-Destructive Testing of single variety fruit internal quality,
Near-infrared spectral analysis technology has been obtained for extensively using.Due to different fruit physicochemical properties and the difference of outward appearance, generally adopt
Take and different fruit are set up the strategy each analyzing model, it is therefore apparent that so modeling cost and maintenance cost of later stage model
The highest.Additionally, for optical filtering flap-type nir instrument, limited owing to itself covering wavelength points, cause built-in model
Limited amount, and the model that modeling wavelength points is more can not be comprised, the suitability is poor.
Summary of the invention
Present invention aim at providing a kind of near-infrared universal model utilizing light physical property close fruit quality index to detect
Method, solves to set up near-infrared model for a class fruit, lead to without ripe multi items in existing near infrared detection technology
The foundation with model, thus causing model and the technical problem that maintenance cost is high, work efficiency is relatively low;Also solve existing closely
Wavelength points is limited, cause built-in model limited amount, the scope of application to compare limitation, the suitability relatively owing to itself covers for infrared gear
The technical barrier of difference.
For solving above-mentioned technical problem, the present invention adopts the following technical scheme that
A kind of near-infrared universal model detection method of smooth physical property close multi items fruit quality index, it is characterised in that:
Including near-infrared universal model foundation and utilize near-infrared universal model measure fruit internal quality two large divisions,
Part I, the foundation of near-infrared universal model, specifically include following steps:
Step one, material prepare: preparing the fruit to be measured of the close multi items of light physical property, described smooth physical property is close refers to thing
Have similar physicochemical property between material, the near-infrared original spectrum of any two kinds of fruit therein Euclidean after spectral normalization away from
From no more than 0.2, total kind number of fruit is 2~6 classes.The selection of the material that described smooth physical property is close is extremely important, described light
Physical property is close refers to close physical property and chemical property so that storeroom near-infrared original spectrum is through spectral normalization
After, Euclidean distance is not more than 0.2, and Euclidean distance can represent the similarity degree between signal, and Euclidean distance is the least, the similar journey of spectrum
Spend the highest.Calculate the Euclidean distance between sample according to prior art, specifically comprise the following steps that
1, in each kind, choose sufficient amount of sample, gather its original near infrared spectrum;
2, select representative wave band: the wave band that noise is bigger should be avoided, the fruit variety that the present invention relates to choose 780~
920nm wave band;
3, standard spectrum is calculated: calculate the averaged spectrum of each material, as the standard spectrum of this kind fruit;
4, spectral normalization processes: each material carries out spectral normalization process respectively, and spectral normalization processes
Ultimate principle formula is: X=(A-)/(Amax - Amin);Wherein, X is the value after spectral normalization processes, and A is primary light
The absorbance of spectrum,For the absorbance of standard spectrum, AmaxFor maximum absorbance value, AminFor minimum absorbance;
5, the Euclidean distance of any two kind fruit is calculated: computing formula is Dmn = , wherein, m
Represent fruit m and fruit n respectively with n, p is that total wavelength of near infrared spectrum is counted, and Xmi represents the near infrared spectrum warp of fruit m
Spectral normalization process after the value of i-th, Xni represent the near infrared spectrum of fruit n after spectral normalization processes at the i-th point
Value.
Step 2, choose modeling sample: the fruit to be measured from each kind randomly selects at least 30 sample conducts
Calibration set sample, at least 10 samples integrate the number ratio of sample as 3:1 as checking collection sample, calibration set sample and checking.
Step 3, gather all calibration set samples and checking collection sample original near infrared spectrum.
Step 4, utilize the actual quality index values of all fruit having gathered spectrum of chemical determination.
Step 5, the near infrared spectrum collected in step 3 is carried out pretreatment, use chemo metric software to institute
Spectrum is had to be scattered correction successively, reduce noise and eliminate the process of the time of integration.
Step 6, extract the characteristic wavelength of each kind fruit: respectively with the chemo metric software light to all kinds of fruit
Spectrum processes, and uses chemometrics algorithm such as moving window partial least square method i.e. MWPLS method, extracts such fruit list
Solely characteristic wavelength during modeling.
Step 7, extract the common features wave-length coverage of all kind fruit: treat that the characteristic wavelength of each kind fruit carries
After taking, compare, choose and can cover the wave band of all characteristic wavelengths as common features wave-length coverage.
Step 8, from common features wavelength, extract common features wavelength points: use chemistry strength algorithm such as continuously throwing
Shadow algorithm i.e. SPA algorithm, extracts the common features wavelength points in this wavelength.
Step 9, set up near-infrared universal model: common features wavelength points step 8 obtained is as modeling wavelength, step
Quality index values in rapid four, as standard value, utilizes the pretreated near infrared spectrum of step 5 to set up near-infrared Universal Die
Type.
Step 10, the inspection of near-infrared universal model accuracy: respectively by near infrared spectrum and the reality of checking collection sample
Quality index values substitutes into near-infrared universal model, carries out the inspection of near-infrared universal model accuracy, pre-according to predictor calculation
Surveying standard deviation RMSEP, if its value meets requirement of experiment, representative model is feasible;Otherwise, then repeat step 5~step 10, until
Meet requirement.
Part II, utilizes near-infrared universal model to measure fruit quality index, specifically comprises the following steps that
Step A, near-infrared universal model is imported in corresponding near infrared spectrometer, adjust the associated quad time.
Step B, utilize this near infrared spectrometer collection remain all testing samples original near infrared spectrum, instrument then can
The original near infrared spectrum obtained is input in model, draws quality index values, until all fruit measures complete.
In the present invention, the fruit that in step one, light physical property is close can be Fructus Mali pumilae, Fructus Persicae and pears three kinds, and it has close thing
Reason, chemical property, as be all globular shape, sizableness, skin is thin, moisture and soluble solid content close, and have really
Core, soluble sugar is all made up of sucrose, glucose, fructose and sorbitol, and due near infrared spectrum reflection is to change in material
Learning the information that component content is higher, as being water and soluble solid etc. in fruit, therefore the near infrared spectrum of three exists one
Cause property, pears are 0.124 with the Euclidean distance of Fructus Mali pumilae, and pears are 0.150 with the Euclidean distance of Fructus Persicae, and Fructus Mali pumilae with the Euclidean distance of Fructus Persicae is
0.071, meet the near-infrared universal model that the present invention relates to requirement to material.
As the preferred technical solution of the present invention, the described index of quality can be pol index, acidity index or Maturity
Index.
As present invention further optimization technical scheme, the concrete grammar of near infrared spectra collection in described step 3
For: use K-BA100R type portable near infrared spectrometer, be equipped with collecting fiber adnexa, use CCD-detector, treat that sample is placed
To room temperature, each sample equator carries out in uniform 4 sample area spectra collection, Fructus Mali pumilae, pears and Fructus Persicae spectra collection
Time the time of integration be respectively 100ms, 90ms and 60ms, spectral region is 500nm~1010nm, and resolution is 2nm.
As present invention further optimization technical scheme, in described step 4, when measuring actual quality index values,
1) if the index of quality is pol index, its assay method is: equally distributed four light on fruit sample equator
The centre of spectrum pickup area takes after the square of 20mm*20mm*10mm squeezes juice, uses refractometer to measure fruit internal solvable
The content of property solid content is as actual pol index;
2) if the index of quality is acidity index, its assay method is: carry out acidity assaying with pH potentiometry, at fruit sample
On equator, uniform four regional center positions take sarcocarp 25g, be transferred in 250mL volumetric flask with the water of 80 DEG C, enter after smashing to pieces
Row 30min boiling water bath, then takes out and is cooled to room temperature, and constant volume forms test solution after filtering, and draws test solution 50mL, is placed in beaker,
Add the mixing of 50mL water, be titrated to terminal by the NaOH solution of 0.05mol/L, with the pH value of pH meter monitoring test solution during being somebody's turn to do,
The volume of the volumetric solution that record consumes, finally calculates total acid content;
3) if the index of quality is mature indicator, its assay method is: separately sampled according to above-mentioned actual acidity index and
Measure pol and acidity with the assay method of actual pol index, then calculate sugar-acid ratio, in order to labelling Maturity.Sugar-acid ratio is more
Greatly, illustrate that Maturity is the highest.
As present invention further optimization technical scheme, in described step 5, the concrete grammar of Pretreated spectra is: light
The concrete grammar of spectrum pretreatment is to utilize chemo metric software, cubic polynomial SG using SNV, window size to be 25 successively
Smoothing techniques and second derivative method carry out pretreatment to spectrum;Owing to scattering medium existence is different between fruit, cause sample pair
Scattering of light degree is different, uses SNV to be corrected, improves spectral quality, make the difference caused by scattering between spectrum subtract simultaneously
Little;There is certain noise in spectrum two, can affect modeling effect, uses cubic polynomial item formula, and window size is the SG of 25
The smooth spectrum high-frequency noise that eliminates, raising spectral quality;Additionally, for the accuracy ensureing actual measurement, different fruit, have
The different times of integration, as Fructus Mali pumilae, pears, the time of integration of Fructus Persicae are respectively 100ms, 90ms, 60ms, and during different integration
Between spectrum can be made to drift about, use second dervative can eliminate.
As present invention further optimization technical scheme, in described step 6, use moving window partial least square method
I.e. MWPLS algorithm carries out the extraction of characteristic wavelength respectively to Fructus Mali pumilae, Fructus Persicae and pears, and window size is respectively 20,25 and 30.
As present invention further optimization technical scheme, stating in step 8, the characteristic wavelength point of extraction has 5, respectively
For 840nm, 850nm, 860nm, 886nm, 900nm.
As present invention further optimization technical scheme, in described step 9, use chemo metric software such as TQ
9.0 softwares are set up near-infrared PLS model or use chemo metric software such as IBM SPSS Statistics 20 software to set up
Near-infrared MLR model, as modeling result when setting up near-infrared MLR model is: Y=10.433+22996.898 λ840-
24482.457λ850-4599.339λ860+208314.677λ886-89204.98λ900;Wherein, in formula, λ840、λ850、λ860、λ886、
λ900It is the absorbance at 840nm, 850nm, 860nm, 886nm, 900nm for the near infrared spectrum medium wavelength point after pretreatment
Value.
Compared with prior art, the present invention's it is a technical advantage that:
1, disposable modeling, reduction model are set up and maintenance cost
The present invention establishes the near-infrared universal model being applicable to multiple kind fruit quality Indexs measure, in modeling process
Eliminate the impact of the time of integration by second dervative, the characteristic wavelength of each kind filters out common features wavelength, then uses SPA
Algorithm extracts characteristic wavelength point further, and disposable foundation can detect the near-infrared universal model of multiple types sugar degree, has
Extremely strong feasibility, overcomes variety classes fruit in existing Near Infrared Spectroscopy Detection Technology and must carry out the fraud of classification and Detection
End, both can complete the detection of multiple index of quality of various fruits with a model, greatly reduced modeling cost and model dimension
Protect cost.
2, accuracy rate is measured high
The effectiveness of this method is checked by the present invention by substantial amounts of experiment, including establishing Fructus Mali pumilae, Fructus Persicae and pears
The general near-infrared pol detection model of three, result is as shown in table 1, chooses from common features wavelength 840~918 nm
Five characteristic wavelength points of 840nm, 850nm, 860nm, 886nm, 900nm, set up PLS model, the Rc=0.98 of model, always
REMSEP=0.38, it was predicted that Fructus Mali pumilae, Fructus Persicae, the RMSEP of pears are respectively 0.42,0.32 and 0.41;With the Rc=0.96 of MLR model,
Total RMSEP=0.38, it was predicted that Fructus Mali pumilae, Fructus Persicae, the RMSEP of pears are respectively 0.44,0.31 and 0.40, both of which has good prediction essence
Degree.
3, Model Practical is strong
In modeling process, using second dervative to eliminate the impact of the sample time of integration, model can be used when actually used
The different times of integration is detected;Use MWPLS to combine SPA algorithm, preferably go out common features wavelength points, greatly reduce
The complexity of model, substantially increases the practicality of model, meets practical application request, and near-infrared universal model can use
On easy nir instrument.
Accompanying drawing explanation
Fig. 1 is the original near infrared spectrum of the Fuji apple, water-rich areas and the honey peach that relate in the embodiment of the present invention;
Fig. 2 is Fuji apple, water-rich areas and the honey peach near infrared light after pretreatment related in the embodiment of the present invention
Spectrum;
Fig. 3 and Fig. 4 is the selection result figure of the SPA characteristic wavelength point related in the embodiment of the present invention, and wherein Fig. 3 represents
Being to have minimum sandards error and keep constant during 5 points of selection, what Fig. 4 represented is positions in spectrum, selected 5;
Fig. 5 is the PLS near-infrared universal model related in the embodiment of the present invention 1;
Fig. 6 is the MLR near-infrared universal model related in the embodiment of the present invention 2.
Detailed description of the invention
Below in conjunction with specific embodiment, present invention is further explained, wherein, the embodiment 1 that the present invention relates to
The representative extremely strong fruit of three classes is all selected: Fuji apple, honey peach and water-rich areas, its common feature is: ball-type with embodiment 2
Shape, sizableness, skin is thin, and moisture and soluble solid content are close, and has pit, and the main component of three is solvable
Property sugar is all made up of sucrose, glucose, fructose and Sorbitol, and the near infrared spectrum of three has certain similarity, pears
Being 0.124 with the Euclidean distance of Fructus Mali pumilae, pears are 0.150 with the Euclidean distance of Fructus Persicae, and Fructus Mali pumilae is 0.071 with the Euclidean distance of Fructus Persicae, for
Set up universal model and establish theoretical basis.The present invention is applicable to the mensuration of all close fruit qualities of smooth physical property, described smooth thing
Similar temperament refers to that close physics and chemical property, storeroom near infrared spectrum shape similarity are higher, therein any two
The near-infrared original spectrum of kind fruit Euclidean distance after spectral normalization is not more than 0.2, and the quantity of kind is 2~6 classes, as
Small watermelon and Fructus Melo, Fructus Citri junoris and Citrus;The described index of quality is that pol index, hardness number, acidity index or Maturity refer to
Mark, due to satisfactory different types of fruits all in the present invention or the near-infrared universal model method for building up base of the index of quality
This is consistent, so not enumerating in present invention.This sentences mensuration Fuji apple, honey peach and three kinds of fruit of water-rich areas
Pol as a example by, present disclosure is discussed in detail.
1 materials and methods
1.1 instruments and sample
Experiment uses the K-BA100R type portable near infrared spectrometer of Kubota Co., Ltd. of Japan, is equipped with collecting fiber
Adnexa, uses CCD-detector;The PAL-1 type handheld digital saccharimeter of Atago company of Japan, reading result is Brix degree
(Brix), possesses ATC function.
Each 40 of red fuji apple, honey peach and water-rich areas, totally 120 samples, it is purchased from wholesale market, Beijing.
1.2 spectra collections measure with standard value
Sample is placed to room temperature, carries out spectra collection on uniform 4 regions, each sample equator (being spaced 90 °),
Foundation spectral energy value is respectively 100ms, 90ms in the principle of rated energy scope, the best total of points time of Fructus Mali pumilae, pears and Fructus Persicae
And 60ms.Spectral region is 500nm-1010nm, and resolution is 2nm, totally 256 data points.After spectra collection completes, gathering
Regional center takes the square of length, width and height about 20mm × 20mm × 10mm and squeezes juice measurement pol value.As it is shown in figure 1, be that the present invention implements
The Fuji apple that relates in example, water-rich areas, the original spectrum of honey peach, wave-length coverage is 700~1010nm as can be seen from Figure,
Can be seen that they original spectrums are closely similar.
1.3 chemo metric software
MWPLS, SPA program is in Matlab R2012a (The mathworks Inc., Natick, MA, USA)
Realizing, preprocessing procedures and PLS model realize in TQ 9.0 (Thermo Nicolet Co., USA).MLR mould
Type realizes in IBM SPSS Statistics 20.
1.4 statistical analysis
Using Kennard-Stone algorithm that sample is corrected the division of collection and checking collection, K-S algorithm is based on original
Minimax Euclidean distance between spectrum chooses representative sample composition calibration set.Due to Fructus Mali pumilae in this experiment, Fructus Persicae, pears
There is some difference and overlap, such as Fig. 1 for original spectrum, it is impossible to unified uses K-S algorithm, can only carry out Fructus Mali pumilae, Fructus Persicae, pears respectively
K-S divides, thus ensures the harmony of three kinds of fruit calibration sets.Setting the calibration set ratio with checking collection as 3:1, statistical result is such as
Shown in table 1.
2 results and discussion
2.1 Pretreated spectra
First by standard normal variable conversion (standard normal variate transformation,
SNV), it is used for eliminating the impact on NIR diffusing transmission spectrum of solid particle size, surface scattering and change in optical path length.Use afterwards
Cubic polynomial Savitzky-Golay smooths, and window size is 25, to eliminate the high-frequency noise in spectrum.
2.2 eliminate the impact time of integration
Due near infrared light difference of transmission capacity on different fruit, when different fruit has different acquired integrated
Between.As it is shown in figure 1, owing to the time of integration of Fructus Persicae is the shortest, its absorbance is higher than Fructus Mali pumilae and pears, and due to Fructus Mali pumilae, pears integration time
Between close, cause its spectrum that overlapping occurs, thus the impact of the time of integration must be eliminated, universal model could be set up.
This research uses same Fructus Mali pumilae, same position, in the range of 50ms-150ms, enters at interval of 10ms the time of integration
Experiment of single factor of row, seeks the difference spectrum of every spectrum and averaged spectrum, in addition to the bigger part of noise, be nearly all one straight
Line, the therefore different upper and lower translations that only can cause spectrum the time of integration, just can eliminate long-pending through first derivative or second dervative process
Impact between timesharing.Owing to second dervative can also eliminate lateral light Frequency bias, amplify near infrared region signal intensity.Therefore select
Second dervative processes.Fig. 2 is that the original spectrum of three kinds of fruit carries out pretreated spectrogram, it can be seen that have more preferably
Concordance.
The selection of 2.3 generic features wavelength
In fruit, soluble solid is typically made up of soluble sugar and acid, and the acid content in Fructus Mali pumilae, Fructus Persicae, pears is the lowest
(0.1%), almost without effective information near infrared spectrum, therefore, soluble solid is mainly composed of soluble sugar.Herba Marsileae Quadrifoliae
Really, main soluble sugar is all fructose, glucose sugar, sucrose and sorbitol in Fructus Persicae, pears, and therefore seeking generic features wavelength is can
Row.
Moving window partial least square method (MWPLS), energy optimization information is interval, promotes the predictive ability of PLS model.Herein
Using MWPLS algorithm that Fructus Mali pumilae, Fructus Persicae, pears carry out the selection of characteristic wavelength respectively, window size is respectively 20,25 and 30, result
Such as table 3.
Using minimum RMSEP as selection standard, select Fructus Mali pumilae, Fructus Persicae, the characteristic interval of pears are respectively 880nm~918nm,
852nm~900nm and 840nm~898nm, it can be seen that the similarity of three's characteristic interval.In order to improve the steady of model
Strong property, choosing 840nm~918nm interval herein is that generic features wavelength sets up PLS model.
Owing to MWPLS cannot eliminate the redundancy of selected wave band, and selected 840nm~918nm interval cover Fructus Mali pumilae,
, therefore there is bulk redundancy information in this wave band in Fructus Persicae, the characteristic interval of pears.And successive projection algorithm can make the synteny between variable
Minimize, make redundancy be eliminated.SPA algorithm is used herein, to optimize characteristic area on 840nm~918nm interval
Between, reduced model.Fig. 3 and Fig. 4 is the selection result figure of the SPA characteristic wavelength point that the present invention relates to, and what wherein Fig. 3 represented is choosing
Having minimum sandards error when selecting 5 points and keep constant, what Fig. 4 represented is positions in spectrum, selected 5;Fig. 4
Result shows, selecting wavelength points is 5, respectively 840nm, 850nm, 860nm, 886nm, 900nm.
The foundation of 2.4 universal models and evaluation
General PLS model selection common features wavelength 840~918 nm sets up a model, model Rc=0.98, always
REMSEP=0.38, it was predicted that Fructus Mali pumilae, Fructus Persicae, the RMSEP of pears are respectively 0.42,0.36 and 0.37;
Then from common features wavelength, choose 840nm, five characteristic wavelengths of 850nm, 860nm, 886nm, 900nm
Point modeling, model Rc=0.98, total REMSEP=0.38, it was predicted that Fructus Mali pumilae, Fructus Persicae, the RMSEP of pears are respectively 0.42,0.41 and
0.32, PLS model result is as shown in Figure 5;
As shown in Figure 6, using set up MLR model at these 5, modeling result is:
Y=10.433+22996.898λ840-24482.457λ850-4599.339λ860+208314.677λ886-89204.98
λ900
Wherein, in formula, λ840、λ850、λ860、λ886、λ900For in step 4, wavelength points is 840nm, 850nm, 860nm,
Value at 886nm, 900nm.Its R2=0.96, RMSEC=0.48, RMSEC=0.38.
Result shows, three models all have good precision of prediction, and the RMSEP value of three kinds of fruit is respectively less than 0.5, with
Kubota instrument single variety fruit forecast result of model is close.After the preferred wave point of SPA, greatly reduce variable number, model
Being simplified, the application of model is improved.
3 utilize near-infrared universal model to measure sugar degree
Step A, near-infrared universal model is imported in corresponding near infrared spectrometer, adjust the associated quad time;
Step B, utilize this near infrared spectrometer collection to remain the near infrared spectrum of all testing samples, and utilize built-in
Near-infrared universal model exports pol value automatically, until all fruit measures complete.
Claims (8)
1. a near-infrared universal model detection method for light physical property close fruit quality index, first sets up model, recycles mould
Type measures fruit quality index, it is characterised in that: first setting up near-infrared universal model, recycling near-infrared universal model measures water
The really index of quality;
Described set up comprising the following steps that of near-infrared universal model;
Step one, material prepare, and i.e. prepare the fruit to be measured of the close multiple kinds of light physical property, and described smooth physical property is close refers to thing
Similar physicochemical property, the near-infrared original spectrum of any two kinds of fruit the most therein Euclidean after spectral normalization is had between material
Distance no more than 0.2, fruit total kind number is 2~6 classes;
Step 2, choose modeling sample, from every kind of fruit, i.e. randomly select at least 30 samples as calibration set sample, at least
10 samples are as checking collection sample, and the number ratio of calibration set sample and checking collection sample is 3: 1;
Step 3, acquisition correction collection sample and the original near infrared spectrum of checking collection sample;
Step 4, chemical determination is utilized to gather the actual quality index values of fruit of spectrum;
Step 5, the original near infrared spectrum collected in step 3 is carried out pretreatment, use chemo metric software to institute
Spectrum is had to be scattered correction successively, reduce noise and eliminate the process of the time of integration;
Step 6, extract the characteristic wavelength of every kind of fruit, with chemo metric software, the spectrum of every kind of fruit is carried out the most respectively
Pretreatment, extracts characteristic wavelength when this kind of fruit individually models;
Step 7, extract the common features wave-length coverage of all kind fruit, i.e. treat that the characteristic wavelength of each kind fruit extracts
After, compare, choose and can cover the wave band of all characteristic wavelengths as common features wave-length coverage;
Step 8, from common features wave-length coverage extract common features wavelength points, i.e. with chemo metric software from shared spy
Levy extraction common features wavelength points in wave-length coverage;
Step 9, set up near-infrared universal model, will the common features wavelength points that obtains of step 8 as modeling wavelength, will step
The actual quality index values obtained in rapid four, as standard value, utilizes the pretreated near infrared spectrum of step 5 to set up near-infrared
Universal model;
Step 10, the inspection of near-infrared universal model accuracy, will verify that the near infrared spectrum of collection sample and actual quality refer to
Scale value substitutes into near-infrared universal model, carries out the inspection of near-infrared universal model accuracy, predicts standard according to predictor calculation
Difference RMSEP, if its value meets requirement of experiment, representative model is feasible;Otherwise, then repeating step 5~step 10, wanting until meeting
Ask;
In described step 5, the concrete grammar of Pretreated spectra is to utilize chemo metric software, uses SNV, window big successively
Little be 25 cubic polynomial SG smoothing techniques and second derivative method spectrum is carried out pretreatment;
In described step 9, chemo metric software is used to set up PLS model or MLR model;
Described near-infrared universal model is utilized to measure the comprising the following steps that of fruit quality index;
Step A, near-infrared universal model is imported near infrared spectrometer, adjust the associated quad time;
Step B, utilize this near infrared spectrometer gather fruit to be measured original near infrared spectrum, instrument can by obtain original closely
Infrared spectrum is input in model, draws index of quality predictive value, until all fruit measures complete.
The near-infrared universal model detection method of a kind of smooth physical property the most according to claim 1 close fruit quality index,
It is characterized in that: in described step one, the fruit to be measured of multiple kinds that light physical property is close is Fructus Mali pumilae, pears and Fructus Persicae three kinds, San Zheguang
The time of integration that spectrum gathers is respectively 100ms, 90ms and 60ms, and spectra collection scope is 500nm~1010nm, and resolution is
2nm, the common features wavelength of three is 840~918nm.
The near-infrared universal model detection method of a kind of smooth physical property the most according to claim 2 close fruit quality index,
It is characterized in that: in described step 3, the concrete acquisition method of original near infrared spectra collection is, uses K-BA100R type portable
Formula near infrared spectrometer, is equipped with collecting fiber adnexa, uses CCD-detector, places to room temperature until sample, red at each sample
On road, equally distributed four sample region carry out spectra collection respectively.
The near-infrared universal model detection method of a kind of smooth physical property the most according to claim 3 close fruit quality index,
It is characterized in that: the index of quality in described step 4 is pol index, acidity index or mature indicator.
The near-infrared universal model detection method of a kind of smooth physical property the most according to claim 4 close fruit quality index,
It is characterized in that: in described step 4, when the index of quality is pol index, the method measuring actual quality index values is, at sample
The centre of product equator polishing wax pickup area takes after the square of 20mm*20mm*10mm squeezes juice, uses refractometer to measure juice
The content of internal soluble solid is as actual pol index.
The near-infrared universal model detection method of a kind of smooth physical property the most according to claim 4 close fruit quality index,
It is characterized in that: in described step 4, when the index of quality is acidity index, the concrete grammar measuring actual quality index values is,
Take sarcocarp 20~30g at center, polishing wax sample region, sample equator, after smashing to pieces, be transferred to 250mL volumetric flask with the water of 80 DEG C
In, carrying out 30min boiling water bath, then take out and be cooled to room temperature, constant volume filters, and forms test solution, draws test solution 50mL, adds 50mL
Water mixes, and is titrated to terminal by the NaOH solution of 0.05mol/L, and with the pH value of pH meter monitoring test solution during being somebody's turn to do, record consumes
The volume of volumetric solution, calculate total acid content.
The near-infrared universal model detection method of a kind of smooth physical property the most according to claim 4 close fruit quality index,
It is characterized in that: in described step 4, when the index of quality is mature indicator, measure the concrete grammar of actual quality index values
For, the square of 20mm*20mm*10mm is taken in the centre of fruit sample equator polishing wax pickup area, half is used for squeezing
Refractometer is used to measure the content of fruit internal soluble solid as actual pol index after juice;Second half smash to pieces after with 80
DEG C water be transferred in 250mL volumetric flask, carry out 30min boiling water bath, then take out and be cooled to room temperature, constant volume filter after formed examination
Liquid, draws test solution 50mL, adds the mixing of 50mL water, is titrated to terminal by the NaOH solution of 0.05mol/L, uses pH meter during being somebody's turn to do
The pH value of monitoring test solution, the volume of the volumetric solution that record consumes, calculate total acid content;After measuring pol and acidity respectively, meter
Calculate sugar-acid ratio, in order to labelling Maturity.
8. according to the near-infrared Universal Die of a kind of smooth physical property close fruit quality index described in claim 5~7 any one
Type detection method, it is characterised in that: in described step 8, the common features wavelength points extracted from common features wavelength period is 5
Individual, respectively 840nm, 850nm, 860nm, 886nm, 900nm.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201410328830.7A CN104062263B (en) | 2014-07-11 | 2014-07-11 | The near-infrared universal model detection method of light physical property close fruit quality index |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201410328830.7A CN104062263B (en) | 2014-07-11 | 2014-07-11 | The near-infrared universal model detection method of light physical property close fruit quality index |
Publications (2)
Publication Number | Publication Date |
---|---|
CN104062263A CN104062263A (en) | 2014-09-24 |
CN104062263B true CN104062263B (en) | 2016-09-21 |
Family
ID=51550078
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN201410328830.7A Expired - Fee Related CN104062263B (en) | 2014-07-11 | 2014-07-11 | The near-infrared universal model detection method of light physical property close fruit quality index |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN104062263B (en) |
Families Citing this family (19)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN104568827A (en) * | 2015-01-23 | 2015-04-29 | 北京市农林科学院 | Method for measuring sugar content of watermelon by near-infrared spectrometry |
CN105334172B (en) * | 2015-10-23 | 2018-03-09 | 浙江科技学院 | A kind of reconstructing method of fruit pulp optical properties of tissue |
CN105527244A (en) * | 2015-10-26 | 2016-04-27 | 沈阳农业大学 | Near infrared spectrum-based Hanfu apple quality nondestructive test method |
CN105334186A (en) * | 2015-12-10 | 2016-02-17 | 山东大学 | Infrared spectral analysis method |
CN105675539B (en) * | 2016-01-07 | 2019-05-10 | 北京市农林科学院 | A kind of quality of agricultural product integrated evaluating method |
CN106124447A (en) * | 2016-06-08 | 2016-11-16 | 沈阳农业大学 | A kind of based on the method for soluble solid content in near-infrared spectral analysis technology detection Fructus Fragariae Ananssae |
CN106644957B (en) * | 2016-11-14 | 2019-04-05 | 浙江大学 | A kind of method that loquat adopts rear pulp soluble solid distribution imaging |
CN108195793A (en) * | 2016-12-08 | 2018-06-22 | 中国农业机械化科学研究院 | The universal model construction method of plant-derived feedstuff amino acid content |
CN109164062A (en) * | 2018-11-05 | 2019-01-08 | 黑龙江八农垦大学 | A kind of method of near infrared ray "Hami" melon titratable acid content value |
CN111257277B (en) * | 2018-11-30 | 2023-02-17 | 湖南中烟工业有限责任公司 | Tobacco leaf similarity judgment method based on near infrared spectrum technology |
CN110665842A (en) * | 2019-11-14 | 2020-01-10 | 广西立盛茧丝绸有限公司 | Near infrared spectrum cocoon selection method |
CN113218880B (en) * | 2020-01-21 | 2023-05-16 | 青岛海尔电冰箱有限公司 | Food material detection method of refrigerator, refrigerator and storage medium |
CN113218141B (en) * | 2020-01-21 | 2022-10-28 | 青岛海尔电冰箱有限公司 | Food material detection method for refrigerator, refrigerator and storage medium |
CN111537469A (en) * | 2020-06-04 | 2020-08-14 | 哈尔滨理工大学 | Apple quality rapid nondestructive testing method based on near-infrared technology |
CN111721740A (en) * | 2020-06-23 | 2020-09-29 | 佛山市海天(江苏)调味食品有限公司 | Seasoning physical and chemical index detection method based on calibration model |
CN114384040A (en) * | 2020-10-21 | 2022-04-22 | 江苏康缘药业股份有限公司 | Method for establishing general test model of physical property indexes of solid preparation intermediate |
CN112525855B (en) * | 2020-11-20 | 2021-11-02 | 广东省农业科学院蔬菜研究所 | Detection method and device for quality parameters of pumpkin fruits and computer equipment |
CN112595692A (en) * | 2020-11-24 | 2021-04-02 | 南宁国拓生物科技有限公司 | Establishment method of fruit total sugar content prediction model and fruit total sugar content prediction method |
CN114577751A (en) * | 2022-03-09 | 2022-06-03 | 西北农林科技大学 | Building method for nondestructive testing of internal quality of pear and nondestructive testing method for internal quality of pear |
Family Cites Families (2)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
JPH11304698A (en) * | 1998-04-16 | 1999-11-05 | Natl Food Res Inst | Method and device for measuring amylose content in rice |
CN100429502C (en) * | 2004-12-14 | 2008-10-29 | 中国农业大学 | Non-destructive detection method for quickly detecting inner quality of apple |
-
2014
- 2014-07-11 CN CN201410328830.7A patent/CN104062263B/en not_active Expired - Fee Related
Also Published As
Publication number | Publication date |
---|---|
CN104062263A (en) | 2014-09-24 |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
CN104062263B (en) | The near-infrared universal model detection method of light physical property close fruit quality index | |
Zhang et al. | Nondestructive measurement of soluble solids content in apple using near infrared hyperspectral imaging coupled with wavelength selection algorithm | |
McGlone et al. | Vis/NIR estimation at harvest of pre-and post-storage quality indices for ‘Royal Gala’apple | |
CN104062257B (en) | A kind of based on the method for general flavone content near infrared ray solution | |
CN104792722A (en) | Method for establishing agilawood oil content near infrared spectral prediction model | |
CN103196838B (en) | Hyperspectral remote sensing monitoring method for coastal estuary eutrophication | |
CN104062258B (en) | Method for rapid determination of soluble solids in compound ass-hide glue pulp by near infrared spectroscopy | |
CN105548070B (en) | A kind of apple soluble solid near infrared detection position compensation method and system | |
CN104020129A (en) | Method for discriminating fermentation quality of congou black tea based on near-infrared-spectroscopy-combined amino acid analysis technology | |
CN102937575B (en) | Watermelon sugar degree rapid modeling method based on secondary spectrum recombination | |
Yuan et al. | Non-invasive measurements of ‘Yunhe’pears by vis-NIRS technology coupled with deviation fusion modeling approach | |
Yu et al. | Nondestructive determination of SSC in Korla fragrant pear using a portable near-infrared spectroscopy system | |
CN103293118A (en) | Hogwash oil identification method based on near infrared reflectance spectroscopy | |
CN104237060A (en) | Multi-index quick detection method of honeysuckle | |
CN105044024A (en) | Method for nondestructive testing of grape berries based on near infrared spectrum technology | |
CN106950192A (en) | A kind of method of Contents of Main Components quick detection in vegetable protein beverage based on near-infrared spectral analysis technology | |
CN105758819A (en) | Method for detecting organic components of soil by utilizing near infrared spectrum | |
CN103344597A (en) | Anti-flavored-interference near infrared non-destructive testing method for internal components of lotus roots | |
Zhao et al. | Exploring the use of Near-infrared spectroscopy as a tool to predict quality attributes in prickly pear (Rosa roxburghii Tratt) with chemometrics variable strategy | |
CN110779875B (en) | Method for detecting moisture content of winter wheat ear based on hyperspectral technology | |
CN103353443A (en) | Near infrared spectrum based discrimination method for Zhongning fructus lycii | |
CN110231305A (en) | A method of DPPH free radical scavenging ability in the odd sub- seed of measurement | |
Liu et al. | Measurement of soluble solids content of three fruit species using universal near infrared spectroscopy models | |
Hao et al. | Study on robust model construction method of multi-batch fruit online sorting by near-infrared spectroscopy | |
CN110231302A (en) | A kind of method of the odd sub- seed crude fat content of quick measurement |
Legal Events
Date | Code | Title | Description |
---|---|---|---|
C06 | Publication | ||
PB01 | Publication | ||
C10 | Entry into substantive examination | ||
SE01 | Entry into force of request for substantive examination | ||
C14 | Grant of patent or utility model | ||
GR01 | Patent grant | ||
CF01 | Termination of patent right due to non-payment of annual fee |
Granted publication date: 20160921 Termination date: 20170711 |
|
CF01 | Termination of patent right due to non-payment of annual fee |