CN103175815A - Multi-wavelength LED-induced fluorescence tea quality nondestructive testing method and device - Google Patents

Abstract

The invention relates to a multi-wavelength LED-induced fluorescence tea quality nondestructive testing method and a multi-wavelength LED-induced fluorescence tea quality nondestructive testing device. The device comprises a handheld multi-wavelength LED fluorescence detection probe, a long-pass optical filter, large-core-diameter optical fiber, an LED driving control circuit, a spectrometer, a data acquisition card, and a computer. The method comprises the steps of spectrum acquisition, principal component analysis, linear regression model establishment, and test result comparison and analysis. According to the invention, LED with long service life and low price is adopted as an excitation light source, such that the device is characterized in compact structure and small volume. Also, a method of multi-wavelength excitation of tea fluorescence spectra is adopted, such that more physical and chemical information of tea samples can be obtained, and tea quality testing is facilitated.

Description

The tea quality nondestructive detecting method of multi-wavelength LED induced fluorescence and device

Summary of the invention

The present invention relates to a kind ofly can carry out Quality Detection to multiple tealeaves sample based on multi-wave length illuminating diode (Light Emitting Diode, LED) induced fluorescence detection system, belong to optical sensing technical field and laboratory instrument technical field.

Tealeaves has become one of three maximum large beverages of current consumption as the material of drinking of a kind of health, pure natural, and in China, before the history of drinking tealeaves can be traced back to several thousand.Since nineteen ninety-six, the output of China's tealeaves ranks world's first always, has become tea supplies state the biggest in the world, and the detection of tea leaf quality seems particularly important in China.The quality of tealeaves has a lot of determinatives, growing environment, processing treatment technology and storage method can affect the quality of tealeaves, therefore the evaluation of tea quality is system and complexity, and it is the resultant effect of the factors such as tealeaves color, shape, smell, mouthfeel.

Along with growth in the living standard, in view of the pursuit to health, people are more and more higher to the requirement of tea quality, how reasonably to select tealeaves to be subjected to people's common concern.Simultaneously, in the process that processing is processed and stored, the performance of tealeaves being detected, is also quite necessary.Yet, do not have the method for standardized evaluation tea grades on market, cause the situation that has occurred pretending to be with ordinary tea leaves high-quality tea on tea market in recent years, had a strong impact on the prestige of famous brand tealeaves.Therefore seeking a kind of lossless detection method fast and accurately and device is an urgent demand of tea market.

Current, both at home and abroad the evaluation of tea quality is mainly relied on expert's sensory review, i.e. vision, the sense of taste, sense of touch, sense of smell by the professional, and in conjunction with professional knowledge and rich experience, the metric attribute of tealeaves is evaluated.Therefore, the tealeaves sensory evaluation has certain subjectivity, is difficult to objective, as to demarcate accurately tealeaves true grade, and simultaneously, this method does not satisfy fast, the requirement of Non-Destructive Testing.

In recent years, people have put forward the apparatus and method that some tea leaf qualities detect.Be CN101158657A as the patent No., proposed a kind of tea-leaf producing area identification method based on the XRF technology, adopt the difference of different regions tealeaves contents of heavy metal elements to cause the difference of X-ray fluorescence spectra, the tealeaves of different regions is classified.And for example the patent No. is CN101620180A, has proposed a kind of method of rapidly detecting tea quality through near infrared technology, adopts irreflexive spectral information of tealeaves sample near-infrared region, and the tealeaves sample quality is detected.But these methods have certain drawback, and the X ray electron energy is large, shine for a long time harmful; Near infrared spectroscopy belongs to Testing of Feeble Signals, and sensitivity is lower.

For the deficiency of these technology, the present invention puts forward a kind of simple, effective tealeaves pick-up unit and method.

Tea leaf quality the cannot-harm-detection device of multi-wavelength LED induced fluorescence comprises the multi-wavelength LED fluoroscopic examination probe of hand-held, long pass filter, large core fiber, LED Drive and Control Circuit, spectrometer, data collecting card and computing machine.Be evenly equipped with the LED of six different wave lengths in multi-wavelength LED fluoroscopic examination probe around center probe, be provided with long pass filter and large core fiber at center probe, one of large core fiber is rectified long pass filter, the other end is connected with spectrometer, spectrometer is connected with Computer signal, computing machine is by data acquisition card control LED Drive and Control Circuit, and the LED Drive and Control Circuit provides adjustable steady current for the LED in popping one's head in.

The wavelength coverage of the LED of described different wave length is: 255nm-675nm.

The tea quality nondestructive detecting method of multi-wavelength LED induced fluorescence, specifically:

Step 1. spectra collection: light in turn single ledly, gather the fluorescence spectrum of tealeaves sample 300nm-1100nm, deposit with the form of file.

Step 2. principal component analysis (PCA): respectively the fluorescence data of all the tealeaves samples under each wavelength illumination carried out principal component analysis (PCA), major component is pressed the descending arrangement of variance contribution ratio, according to the requirement to contribution rate of accumulative total, get a front T major component

Step 3. is set up model: adopt leaving-one method to set up linear regression model (LRM):

The tea leaf quality factor of the disallowable sample that obtains by linear regression model (LRM) is expressed as:

Step 4. is estimated: the testing result under different wave length is compared, find best review result, complete the detection of tea quality.

Beneficial effect of the present invention: the present invention adopts the life-span is long, price is low LED as excitation source, make device have compact conformation, characteristics that volume is little, adopt simultaneously the method for multi-wavelength excitation tealeaves fluorescence spectrum, can obtain more Tea Samples physics and chemistry information, more be conducive to the detection of tea leaf quality.

The classification of one: six kind of fried green tealeaves sample of example

Six kinds of tealeaves samples are measured ten times for every kind, total N=60 group spectroscopic data, and tea kinds is demarcated position m=1, and 2,3,4,5,6, represent respectively tealeaves sample CQ, CY, CF, G II, G I, CX.

Wherein The preconditioning matrix of the tea leaf quality that the expression expert is given,

Represent j major component to the contribution of k sample,

The linear coefficient matrix that expression obtains by equation of linear regression (1), N represents total spectrum sample given figure; I represents disallowable sample; M represents the number of variable in spectrum; T represents a front T major component;

Expression regressor and biasing, its first row assignment is 1, is used for elimination 0 value and negative value to the impact of regression model, its residual value is

, represent that j major component is to the contribution of k sample.

The grade distinguishing of two: six kinds of giant Buddha Dragon Well tea of example

The quality factor that represent i sample, the i.e. classification results of i sample;

The contribution of front T the major component of expression to i sample.

Description of drawings

Fig. 1 is structural representation of the present invention.

Fig. 2 is the fluorescence spectrum of a tealeaves sample.

Fig. 3 is the classification results of six kinds of fried green tealeaves.

Fig. 4 is the testing result of six grade giant Buddha Dragon Well tea.

Embodiment

Below in conjunction with accompanying drawing, the present invention is further described, but does not limit protection scope of the present invention.

As shown in Figure 1, tea leaf quality the cannot-harm-detection device of multi-wavelength LED induced fluorescence comprises the multi-wavelength LED fluoroscopic examination probe 1 of hand-held, long pass filter, large core fiber 6, LED Drive and Control Circuit 10, spectrometer 7, data collecting card 9 and computing machine 8.Be evenly equipped with the LED of six different wave lengths in multi-wavelength LED fluoroscopic examination probe around center probe, each LED is luminous in turn, is radiated on Tea Samples 4, and wherein Tea Samples 4 is placed on black aluminium sheet 3.Be provided with long pass filter and large core fiber at center probe, one of large core fiber is rectified long pass filter, the other end is connected with spectrometer light, the exciting light 2 that LED sends shines on Tea Samples, inspire fluorescence 5 signals of Tea Samples through after the long pass filter of 450nm, enter spectrometer by large core fiber, spectrometer is connected with Computer signal, the fluorescence spectrum that collects is stored in computing machine, and adopts Matlab to process.Computing machine is by Labview programming Control data collecting card, and then by data acquisition card control driving circuit, driving circuit provides adjustable steady current for the LED on probe.

The tealeaves sample: according to expert's review result, quality have bad to be followed successively by well fried green (CQ), emerald green bud (CY), Cui Feng (CF),

-aminobutyric acid (second-class G II),

-aminobutyric acid (first-class G I), dawn in spring (CX).

(1) gather LED(375nm) fluorescence spectrum of tealeaves sample under inducing.

Fluorescence under LED induces gathers ten group spectroscopic datas for every kind of tealeaves sample with spectrometer (Ocean Optics USB2000) through after the long pass filter of 450nm, measures simultaneously ten groups of background signals.Can obtain fluorescence spectrum figure as shown in Figure 2 after the removal background signal.Therefore can obtain 60 groups of spectroscopic datas of removing background signal for six kinds of tealeaves samples.

(2) fluorescence data is carried out principal component analysis (PCA).

Fluorescence spectrum to 60 groups of 300nm-1100nm recording carries out principal component analysis (PCA), and major component is pressed the descending arrangement of variance contribution ratio, according to the requirement of contribution rate of accumulative total, gets a front T major component.Present case is got a front T=7 major component.

(3) adopt leaving-one method to set up regression model.

The tea leaf quality factor of the disallowable sample that obtains by linear regression model (LRM) can be expressed as:

For present case, pre-service tea leaf quality matrix:

Quote and return the quality that indexing parameter Q comes the descriptive grade classification.

 

(4) interpretation of result.

Adopt the final classification results of Q value representation, as shown in Figure 3.The Q value is larger, and the result of the larger presentation class of difference between each Q value is better.

(5) multi-wavelength replenishment.

In the unconspicuous situation of single wavelength classification, can adopt multi-wavelength to replenish, final selected best classification results is perhaps classified in conjunction with certain two or some wavelength.The present case wavelength that places an order can obviously be distinguished six kinds of fried green tealeaves samples, does not therefore adopt other wavelength to detect.

The tealeaves sample: the giant Buddha Dragon Well tea of six grades is 1,2,3,4,5,6 by good the demarcation successively to difference.

(1) carry out spectra collection and principal component analysis (PCA) as (1) in example one and (2).

The tea leaf quality factor of the disallowable sample that obtains by linear regression model (LRM) can be expressed as:

Preconditioning matrix can be expressed as:

N represents tea grades, and t represents to measure number of times.

(3) interpretation of result.

The grade distinguishing result as shown in Figure 4.Linearity R=0.975.

Claims (3)

1. tea leaf quality the cannot-harm-detection device of multi-wavelength LED induced fluorescence, the multi-wavelength LED fluoroscopic examination probe that comprises hand-held, long pass filter, large core fiber, the LED Drive and Control Circuit, spectrometer, data collecting card and computing machine, it is characterized in that: the LED that is evenly equipped with six different wave lengths in multi-wavelength LED fluoroscopic examination probe around center probe, be provided with long pass filter and large core fiber at center probe, one of large core fiber is rectified long pass filter, the other end is connected with spectrometer, spectrometer is connected with Computer signal, computing machine is by data acquisition card control LED Drive and Control Circuit, the LED Drive and Control Circuit provides adjustable steady current for the LED in popping one's head in.

2. tea leaf quality the cannot-harm-detection device according to claim 1, it is characterized in that: the wavelength coverage of the LED of described different wave length is: 255nm-675nm.

3. the tea quality nondestructive detecting method of multi-wavelength LED induced fluorescence is characterized in that the method is specifically:

Step 1. spectra collection: light in turn single ledly, gather the fluorescence spectrum of tealeaves sample 300nm-1100nm, deposit with the form of file;

Step 2. principal component analysis (PCA): respectively the fluorescence data of all the tealeaves samples under each wavelength illumination carried out principal component analysis (PCA), major component is pressed the descending arrangement of variance contribution ratio, according to the requirement to contribution rate of accumulative total, get a front T major component

Step 3. is set up model: adopt leaving-one method to set up linear regression model (LRM):

The tea leaf quality factor of the disallowable sample that obtains by linear regression model (LRM) is expressed as:

The quality factor that represent i sample, the i.e. classification results of i sample;

The contribution of front T the major component of expression to i sample;

Step 4. is estimated: the testing result under different wave length is compared, find best review result, complete the detection of tea quality.

Images