A fruit grading quality detection method based on near-infrared diffuse reflectance spectrum characteristics

Abstract

The application discloses a fruit grading quality detection method based on near-infrared diffuse reflection spectrum characteristics, which can be applied to grading detection of partial internal quality of fruits with thin peels in combination with possibility theory. The method constructs reference core area and reference extension area and corresponding center value and boundary value of wavelength and light intensity of each grade standard partial internal quality corresponding wave trough or wave peak through fruit prediction set and fruit calibration set analysis, and then judges whether the data set composed of wavelength and light intensity of wave trough or wave peak of the fruit needing grading quality detection is a subset of the domain of a certain grade standard by possibility theory, and judges the belonging grade standard with the maximum possibility measure, so as to provide a basis for fruit grading from the perspective of partial internal quality.

Description

Technical Field

[0001] This invention belongs to the field of fruit post-harvest processing technology, specifically relating to a method for grading and detecting the internal quality of fruits with thin peels using near-infrared diffuse reflectance spectral characteristics by combining probability theory. Background Technology

[0002] Near-infrared spectroscopy refers to electromagnetic waves with wavelengths ranging from 780 nm to 2526 nm. When a near-infrared light source emits incident light within the near-infrared spectral range towards a substance being tested, resonance occurs when the frequency of interatomic vibrations in the substance matches the frequency of the near-infrared spectral region, resulting in absorption of the spectrum. Because the types and quantities of hydrogen-containing groups vary among different substances, the absorption intensity and position of different chemical groups in the near-infrared spectral region will differ significantly. Therefore, near-infrared spectroscopy can obtain information on the internal chemical composition and structure of the substance being tested, enabling accurate quantitative and qualitative detection of the composition, content, and properties of hydrogen-containing organic compounds. Near-infrared spectroscopy has strong penetrating power into fruits, allowing the acquisition of internal physical and chemical information, enabling effective detection of fruit quality. Mathematical modeling techniques can be used to establish a mathematical relationship between near-infrared spectroscopy and relevant indicators of the tested fruit, thereby accurately predicting the tested indicators. Current research on fruit quality detection based on near-infrared spectroscopy detection technology is divided into three detection methods: diffuse reflectance, diffuse transmissivity, and transmissivity. Among these, diffuse reflectance is often used to obtain information about the surface layer of the fruit. Summary of the Invention

[0003] The purpose of this invention is to analyze the wavelength and absorbance intensity of the troughs or peaks of the fruits corresponding to the internal quality of each grade standard by analyzing the fruit prediction set and the fruit correction set to construct the reference core area and reference extension area, as well as the corresponding center value and boundary value. Then, using probability theory, the data set composed of the wavelength and absorbance intensity of the troughs or peaks of the fruits that need to be graded is used to determine whether it is a subset of the domain of a certain grade standard, and the grade standard to which it belongs is determined by the highest probability measure, thereby providing a basis for fruit grading from the perspective of partial internal quality.

[0004] To achieve the above objectives, the present invention includes the following steps:

[0005] Step 1: Near-infrared spectral data corresponding to the internal quality of each grade standard part in the fruit prediction set and fruit correction set are collected by the near-infrared spectral device of the laboratory and fruit sorting equipment in the diffuse reflection mode. After the near-infrared spectral data is preprocessed, the wavelength and light absorption intensity of the trough or peak of each grade standard are adjusted to construct the reference core area and reference extension area, as well as the corresponding center value and boundary value.

[0006] Step 2: Using probability theory, the universe of discourse and F-set of a certain grade standard are constructed by setting a reference core region for the wavelength and light absorption intensity of the trough or peak corresponding to the quality of a certain grade standard, and a reference edge region for the wavelength and light absorption intensity of the trough or peak. The degree of constraint is set by setting the probability to 1 in the core region and the probability to the power exponent of the deviation from the center value of the core region and the upper and lower boundaries in the extended region.

[0007] Step 3: Using the near-infrared spectroscopy device of the fruit sorting equipment, collect the data set consisting of the wavelength and light absorption intensity of the troughs or peaks of the fruit to be graded and tested in a diffuse reflection manner. Determine whether it is a subset of the domain of discourse of a certain grade standard, and determine the grade standard to which it should be assigned from the perspective of partial internal quality by maximizing the probability measure.

[0008] In one embodiment, the specific method for collecting near-infrared spectral data corresponding to the internal quality of each grade standard portion in the fruit prediction set and the fruit correction set in a diffuse reflectance manner using near-infrared spectral devices in the laboratory and fruit sorting equipment in step 1 is as follows: Let the fruit prediction set Γ A Fruit Correction Set Γ B In this context, subsets corresponding to a certain level standard h can be defined as the corresponding fruit prediction subsets Γ. A,h and fruit correction subset Γ B,h The grading standards can be determined under laboratory conditions based on national or regional standards for fresh fruit quality grading corresponding to a particular type of fruit with a thinner peel, using a fruit prediction set Γ. A Fruit Correction Set Γ B Grading is completed through appearance and internal quality inspection, where h∈H, H represents the grade status that can be divided according to the relevant national or regional standards for fresh fruit quality grading, and the fruit prediction subset Γ A,h and fruit correction subset Γ B,h There is no such fruit that cannot be graded; let the fruit prediction set Γ be... A Near-infrared spectral data with corresponding internal quality characteristics under standard conditions for each grade were collected by a near-infrared spectral device in the laboratory using diffuse reflectance. The fruit prediction subset Γ is then used. A,h Near-infrared spectral data prediction set corresponding to a certain level h Let the fruit correction set Γ B The near-infrared spectral data corresponding to the above relationships are collected by the near-infrared spectral device of the fruit sorting equipment in a diffuse reflectance manner, then the fruit correction subset Γ B,h Near-infrared spectral data correction set corresponding to a certain level h

[0009] In one embodiment, the specific method for adjusting the wavelength and absorption intensity of the trough or peak of the near-infrared spectral data to construct the reference core region and reference extension region, as well as the corresponding center value and boundary value, after preprocessing the near-infrared spectral data includes the following steps:

[0010] The first step is to predict the near-infrared spectral data set at a certain level h after preprocessing. and near-infrared spectral data correction set Find the wavelength λ corresponding to the common troughs or peaks in the wave. h,i Where i∈I, I represents the troughs or peaks that coexist for level h and constitute the wavelength set G of level h. h ={λ h,i}, measuring the fruit prediction subset Γ A,h For fruit k, the wavelength λ h,i Corresponding light absorption intensity k η A,h,i Where fruit k∈Γ A,h Then for a certain level h, it can be obtained by Γ A,h Constituting wavelength λ h,i The corresponding set of absorption intensities E A,h,i ={ k η A,h,i From this, we can obtain the fruit prediction set Γ. A The set of wavelengths and corresponding absorption intensities at different levels;

[0011] The second step is to predict the near-infrared spectral data set. For the wavelength set G of level h h A certain wavelength λ h,i The wavelength λ can be given. h,i The maximum absorption intensity is Corresponding to wavelength λ h,i The minimum absorption intensity is Where max() is the function to find the maximum value and min() is the function to find the minimum value; the wavelength set G of level h is... h Compare with the wavelength sets corresponding to other levels in H; if an intersection exists, set that intersection as G″. h Let G′ be the set that has no intersection with other wavelength sets. h For the wavelength set G′ of class h h Mid-wavelength λ h,i The center value of the absorption intensity The sum() function is used to sum all elements, and the crad() function is used to count the number of elements.

[0012] The third step is to predict the near-infrared spectral data set. Lieutenant General rank h G″ h Let the set of absorption intensities corresponding to E′ be E′. A,h,i Then for G″ h Mid-wavelength λ h,j Where j∈I, the region of light absorption intensity fluctuation In other wavelength sets of different levels, the wavelength is equal to λ. h,j The regions of light absorption intensity fluctuations are compared, and if there are overlapping regions, these regions are set as R″. A,h,j If there is no overlapping region, then it is set as R′. A,h,j For the wavelength set G″ of class h h Mid-wavelength λ h,j The center value of the corresponding absorption intensity

[0013] The fourth step is to predict the near-infrared spectral data set. Lieutenant General G' h Any wavelength λ h,i Corresponding light absorption intensity fluctuation region and G″ h Any wavelength λ h,j The corresponding R′ A,h,j The merger constitutes the core area For G′ h wavelength λ h,i The corresponding upper and lower boundary values ​​of the core area are respectively and For G″ h wavelength λ h,j The corresponding upper and lower boundary values ​​of the core area are max(R′) and max(R′) respectively. A,h,j ) and min(R′ A,h,j ); G″ h Any wavelength λ h,j The corresponding R″ A,h,j The merger forms the extended region V A,h ={(λ h,j ,R″ A,h,j )}, for G″ h wavelength λ A,h,j The corresponding upper and lower boundary values ​​of the extended region are max(R″) and max(R″) respectively. A,h,j ) and min(R″ A,h,j );

[0014] The fifth step is to correct the near-infrared spectral data set at a certain level h after preprocessing. Generate the core region Z following steps one through four. B,h and extended region V B,h Then, in summary and The reference core region Z for level h can be set. h =Z A,h ∩Z B,h For wavelength λ h,i or λ h,j The center value of the corresponding absorption intensity The corresponding upper boundary value of the reference core region is or min[max(R′ A,h,j ),max(R′ B,h,j )], at this time for G′ h wavelength λ h,i The core region is defined by its lower boundary value being max[min(R′). A,h,j ),min(R′ B,h,j )]or For the reference extension region V of level h h =V A,h ∪V B,h ∪(Z A,h -Z h ), corresponding to a wavelength of λ h,j The upper boundary value of the reference extension region is The lower boundary value is Thus complete right Data calibration;

[0015] In one embodiment, in step 2, the universe of discourse and F-set of a certain level standard are constructed by using probability theory to define a reference core region of wavelength and absorption intensity of a certain level standard, and a reference edge region of wavelength and absorption intensity of a certain level standard. The specific method for setting the degree of constraint by setting the probability to 1 in the core region and the probability to a power exponent of the deviation from the center value and upper and lower boundaries of the core region in the extended region is as follows:

[0016] The domain U of the rank standard h h And "h grade" F h For U h The F set on the above can be represented by probability theory using equations (1) and (2), and X can be assumed to be U. h A variable F takes a value from. h The F constraint for X can be expressed as X = u:F h (u), at this time F h (u) is F h The degree of constraint on the value u of X;

[0017] U h ={ h u1,..., h u m ,..., hu M,h u M+1 ,..., h u M+n ,..., h u M+N} (1)

[0018]

[0019] Where M = crad(Z) h ), N = crad(V h ), 1≤m≤M, 1≤n≤N, we know that for Z h middle or (λ) h,j ,R′ A,h,j )exist h u m In a one-to-one correspondence, for V h middle exist h u M+n In a one-to-one correspondence, ε l,j For the tested fruit l, for λ h,j The absorption intensity, if (λ h,j ,ε l,j In V h In the set, corresponding to h u M+n The probability δ(M+n) can be expressed by equation (3) and its value should be between (0,1), where exp() is a power function and α h,j For λ h,j The upper boundary adjustment coefficient, β h,j For λ h,j The lower boundary adjustment coefficient, γ h,j For λ h,j Fading coefficient:

[0020]

[0021] In one embodiment, in step 3, the near-infrared spectroscopy device of the fruit sorting equipment collects data sets consisting of the wavelengths and absorbance intensities of the troughs or peaks of the fruit to be graded and tested using diffuse reflectance. The method for determining whether these data sets constitute a subset of the domain of a certain grade standard, and for using the maximum probability measure to determine the grade standard to which the fruit should be assigned from the perspective of partial internal quality, is as follows:

[0022] The near-infrared spectroscopy device of the fruit sorting equipment collects the wavelengths λ of the troughs or peaks of the fruit (l) requiring grading and quality testing using diffuse reflectance. l,t With the corresponding absorption intensity ε l,t And constitute the data set Λ of the fruit ll ={(λ l,t ,ε l,t Let |1≤t≤T}, where T is the number of troughs or peaks of the fruit l; if the universe of discourse U for the level h h ,exist Therefore, it can be concluded that fruit l cannot be included in grade h; if the domain of discourse U for grade h h ,exist X can be narrowed down to Λ l ∩U h The value is taken above, due to the probability distribution function π related to X. X (u)=F h (u), then the probability measure Poss{fruit l belongs to grade h} can be expressed as: When the probability measure exceeds the probability threshold, the fruit can be considered to be classified as grade h from the perspective of some internal quality; if there are multiple domains and Λ l If there is an intersection, then when the highest probability measure value exceeds the probability threshold, the grade corresponding to the highest probability measure value is the grade given by the fruit j after grading quality testing from the perspective of partial internal quality. Attached Figure Description

[0023] Figure 1 This is a flowchart illustrating the fruit grading and quality detection steps based on near-infrared diffuse reflectance spectral characteristics, as described in this invention.

[0024] Figure 2 This is a flowchart illustrating the construction of the reference core region and reference extension region, as well as the center value and boundary value, according to the present invention. Detailed Implementation

[0025] Figure 1 This is a flowchart illustrating the fruit grading and quality detection steps based on near-infrared diffuse reflectance spectral characteristics according to the present invention. The following detailed description of a specific embodiment for grading and quality detection of Kyoho grape bunches based on near-infrared diffuse reflectance spectral characteristics illustrates the technical solution provided by the present invention, specifically including the following steps:

[0026] Step 1, let the fruit prediction set Γ A Fruit Correction Set Γ B In this context, subsets corresponding to a certain level standard h can be defined as the corresponding fruit prediction subsets Γ. A,h and fruit correction subset Γ B,h The grading standards can be determined under laboratory conditions based on the relevant national or regional standards for fresh fruit quality grading corresponding to the Kyoho variety of table grapes, using a fruit prediction set. A Fruit Correction Set Γ BGrading is completed through appearance and internal quality testing. h∈H, the grades H of Kyoho grapes include premium, first-grade, and second-grade, with soluble solids content ≥18%, ≥16%, and ≥15% for premium, first-grade, and second-grade grapes, respectively. Considering that grape bunches consist of multiple berries, the internal quality indicators of the entire grape bunch can be determined by comprehensively considering the physicochemical indicators of multiple berries from different locations. Let the fruit prediction set Γ be... A Near-infrared spectral data are acquired by a laboratory near-infrared spectroscopy device using diffuse reflectance, then the fruit prediction subset Γ A,h Near-infrared spectral data prediction set corresponding to a certain level h Let the fruit correction set Γ B Near-infrared spectral data are collected by the near-infrared spectral device of the fruit sorting equipment in a diffuse reflectance manner, then the fruit correction subset Γ B,h Near-infrared spectral data correction set corresponding to a certain level h The specific method for constructing the reference core region and reference extension region, as well as the corresponding center and boundary values, by adjusting the wavelength and absorption intensity of the trough or peak of the near-infrared spectral data after preprocessing is as follows: Figure 2 As shown, the steps include:

[0027] The first step is to predict the near-infrared spectral data set at a certain level h after preprocessing. and near-infrared spectral data correction set Find the wavelength λ corresponding to the common troughs or peaks in the wave. h,i Where i∈I, I represents the troughs or peaks that coexist for level h, and the selected spectral wavelengths are 630nm, 705nm, 745nm, 785nm, 830nm, and 925nm, forming the wavelength set G of level h. h ={λ h,i}, measuring the fruit prediction subset Γ A,h For fruit k, the wavelength λ h,i Corresponding light absorption intensity k η A,h,i Where fruit k∈Γ A,h Then for a certain level h, it can be obtained by Γ A,h Constituting wavelength λ h,i The corresponding set of absorption intensities E A,h,i ={ k η A,h,i From this, we can obtain the fruit prediction set Γ. A The set of wavelengths and corresponding absorption intensities at different levels;

[0028] The second step is to predict the near-infrared spectral data set. For the wavelength set G of level h h A certain wavelength λh,i The wavelength λ can be given. h,i The maximum absorption intensity is Corresponding to wavelength λ h,i The minimum absorption intensity is Where max() is the function to find the maximum value and min() is the function to find the minimum value; the wavelength set G of level h is... h Compare with the wavelength sets corresponding to other levels in H; if an intersection exists, set that intersection as G″. h Let G′ be the set that has no intersection with other wavelength sets. h For the wavelength set G′ of class h h Mid-wavelength λ h,i The center value of the absorption intensity The sum() function is used to sum all elements, and the crad() function is used to count the number of elements.

[0029] The third step is to predict the near-infrared spectral data set. Lieutenant General rank h G″ h Let the set of absorption intensities corresponding to E′ be E′. A,h,i Then for G″ h Mid-wavelength λ h,j Where j∈I, the region of light absorption intensity fluctuation In other wavelength sets of different levels, the wavelength is equal to λ. h,j The regions of light absorption intensity fluctuations are compared, and if there are overlapping regions, these regions are set as R″. A,h,j If there is no overlapping region, then it is set as R′. A,h,j For the wavelength set G″ of class h h Mid-wavelength λ h,j The center value of the corresponding absorption intensity

[0030] The fourth step is to predict the near-infrared spectral data set. Lieutenant General G' h Any wavelength λ h,i Corresponding light absorption intensity fluctuation region and G″ h Any wavelength λ h,j The corresponding R′ A,h,j The merger constitutes the core area For G′ h wavelength λ h,i The corresponding upper and lower boundary values ​​of the core area are respectively and For G″ h wavelength λ h,j The corresponding upper and lower boundary values ​​of the core area are max(R′) and max(R′) respectively. A,h,j) and min(R′ A,h,j ); G″ h Any wavelength λ h,j The corresponding R″ A,h,j The merger forms the extended region V A,h ={(λ h,j ,R″ A,h,j )}, for G″ h wavelength λ A,h,j The corresponding upper and lower boundary values ​​of the extended region are max(R″) and max(R″) respectively. A,h,j ) and min(R″ A,h,j );

[0031] The fifth step is to correct the near-infrared spectral data set at a certain level h after preprocessing. Generate the core region Z following steps one through four. B,h and extended region V B,h Then, in summary and The reference core region Z for level h can be set. h =Z A,h ∩Z B,h For wavelength λ h,i or λ h,j The center value of the corresponding absorption intensity The corresponding upper boundary value of the reference core region is or min[max(R′ A,h,j ),max(R′ B,h,j )], at this time for G′ h wavelength λ h,i The core region is defined by its lower boundary value being max[min(R′). A,h,j ),min(R′ B,h,j )]or For the reference extension region V of level h h =V A,h ∪V B,h ∪(Z A,h -Z h ), corresponding to a wavelength of λ h,j The upper boundary value of the reference extension region is The lower boundary value is Thus complete right Data calibration;

[0032] Step 2, here you can set the grade standard h to Special Grade, G′ h This includes 630nm and 925nm, G″ h The range includes 705nm, 745nm, 785nm, and 830nm, then M = crad(Z) hThe value is 2, N = crad(V) h The value of U is 4; h For the domain of the rank standard h, F h For U h Set F on the tangent of X, where X is U h A variable F takes a value from. h The F constraint for X can be expressed as X = u:F h (u), F h (u) is F h The degree of constraint on X taking the value u, then U h and F h It can be represented as:

[0033] U h ={ h u 1,h u 2,h u 3,h u 4,h u 5,h u6}

[0034]

[0035] From this, it can be seen that for Z h middle or (λ) h,j ,R′ A,h,j )exist h u m A one-to-one correspondence exists, 1≤m≤2, 1≤n≤4, for V h middle exist h u M+n In a one-to-one correspondence, ε l,j For the tested fruit l, for λ h,j The absorption intensity, if (λ h,j ,ε l,j In V h In the set, corresponding to h u M+n The probability δ(M+n) can be expressed by equation (3) and its value should be between (0,1);

[0036] Step 3: Using the near-infrared spectroscopy device of the fruit sorting equipment, collect the wavelengths λ of the troughs or peaks of the fruit l that need to be graded and tested in a diffuse reflectance manner. l,t With the corresponding absorption intensity ε l,t The wavelengths of the troughs or peaks measured for the fruit l are set to 630nm, 785nm, and 830nm, and these constitute the data set Λ for the fruit l. l ={(λ l,t ,ε l,t Let U be the universe of discourse for rank h, where |1≤t≤3}.h ,exist Then X can be narrowed down to Λ l ∩U h The value is taken above, due to the probability distribution function π related to X. X (u)=F h (u), from which we can know that the current probability distribution is After inputting the dataset of fruit l, the probability measure Poss{fruit l belongs to grade h} can be expressed as:

[0037]

[0038] It can be seen that when the probability measure exceeds the probability threshold of 0.9, the fruit can be considered to be classified as a premium grade from the perspective of soluble solids.

[0039] Any parts not explicitly stated in this embodiment can be implemented using existing technologies.

Claims

1. A method for grading and quality detection of fruits based on near-infrared diffuse reflectance spectral characteristics, characterized in that, Includes the following steps: Step 1: Using a near-infrared spectroscopy device in the laboratory and fruit sorting equipment, near-infrared spectral data corresponding to the internal quality of each grade standard portion in the fruit prediction set and fruit correction set are collected in a diffuse reflectance manner. After preprocessing the near-infrared spectral data, reference core areas and reference extension areas are constructed by adjusting the wavelength and absorbance intensity of the troughs or peaks for each grade standard, along with the corresponding center and boundary values, including: The first step is to predict the γ set of near-infrared spectral data of a certain level h after preprocessing. A,h and near-infrared spectral data correction set γ B,h Find the wavelength λ corresponding to the common troughs or peaks in the wave. h,i Where i∈I, I represents the troughs or peaks that coexist for level h and constitute the wavelength set of level h. Measuring a subset of fruit predictions For fruit k, the wavelength λ h,i Corresponding light absorption intensity Fruit Then for a certain level h, by Constituting wavelength λ h,i The corresponding set of light absorption intensities This yields a fruit prediction set. The set of wavelengths and corresponding absorption intensities at different levels; The second step is to predict the γ-ray in the near-infrared spectral data set. A,h For the wavelength set G of level h h A certain wavelength λ h,i The wavelength λ can be given. h,i The maximum absorption intensity is Corresponding to wavelength λ h,i The minimum absorption intensity is Where max() is the function to find the maximum value and min() is the function to find the minimum value; the wavelength set G of level h is... h Compare with the wavelength sets corresponding to other levels in H; if an intersection exists, set that intersection as... If a wavelength does not intersect with other wavelength sets, then let it be... For the set of wavelengths of level h Mid-wavelength λ h,i The center value of the absorption intensity , where sum() is the function to sum all elements, and crad() is the function to count the number of elements; The third step is to predict the γ-ray in the near-infrared spectral data set. A,h Lieutenant General rank h The corresponding set of light absorption intensities is set as For Mid-wavelength λ h,j Where j∈I, the region of light absorption intensity fluctuation In other wavelength sets of different levels, the wavelength is equal to λ. h,j The regions of light absorption intensity fluctuations are compared, and if there are overlapping regions, then that region is set as... If there is no overlap, then set it as For the set of wavelengths of level h Mid-wavelength λ h,j The center value of the corresponding absorption intensity ； The fourth step is to predict the γ-ray structure in the near-infrared spectral data set. A,h Lieutenant General Any wavelength λ h,i Corresponding light absorption intensity fluctuation region as well as Any wavelength λ h,j corresponding The merger constitutes the core area ,for wavelength λ h,i The corresponding upper and lower boundary values ​​of the core area are respectively and ,for wavelength λ h,j The corresponding upper and lower boundary values ​​of the core area are respectively and ;Will Any wavelength λ h,j corresponding Merging to form an extension area ,for wavelength λ A,h,j The corresponding upper and lower boundary values ​​of the extended region are respectively and ; The fifth step is to correct the γ set of near-infrared spectral data of a certain level h after preprocessing. B,h Generate the core region Z following steps one through four. B,h and extended region V B,h Then, combined with γ A,h and γ B,h Define the reference core region for level h. For wavelength λ h,i or λ h,j The center value of the corresponding absorption intensity The corresponding upper boundary value of the reference core region is or At this time, for wavelength λ h,i The core area is defined by its lower boundary value. or For the reference extension region of level h , corresponding to a wavelength of λ h,j The upper boundary value of the reference extension region is The lower boundary value is Thus completing γ B,h For γ A,h Data calibration; Step 2: Using probability theory, a reference core region and a reference edge region are constructed based on the wavelength and absorption intensity of the troughs or peaks corresponding to the internal quality of a certain grade standard. The constraint level is set by assigning a probability of 1 to the reference core region and a probability of a power-law deviation from the center value and upper and lower boundaries of the reference core region to the reference extension region. This constraint includes: The domain U of level h h And "h grade" F h For U h The F-set on the above is expressed by probability theory in equations (1) and (2): (1) (2) Where X is U h A variable F takes a value from. h The F constraint for X is represented as X=u: F h (u), F h (u) is F h The degree of constraint on the value u of X. , , , For Z h middle or There exists a one-to-one correspondence for V. h middle exist In a one-to-one correspondence, ε l,j For the fruit being tested l For λ h,j The light absorption intensity, if In V h In the set, corresponding to probability Expressed using equation (3), its value should be between (0,1), where exp() is a power function, α h,j For λ h,j The upper boundary adjustment coefficient, β h,j For λ h,j The lower boundary adjustment coefficient, γ h,j For λ h,j Fading coefficient: (3)； Step 3: Using the near-infrared spectroscopy device of the fruit sorting equipment, collect data sets consisting of wavelengths and absorbance intensities of the troughs or peaks of the fruit requiring grading quality testing in a diffuse reflectance manner. Determine whether these data sets constitute a subset of the universe of discourse for a certain grading standard. Then, use the probability measure to determine the grading standard to which the fruit should be assigned from the perspective of its internal quality. This includes: collecting data on the fruit requiring grading quality testing using the near-infrared spectroscopy device of the fruit sorting equipment in a diffuse reflectance manner. l wavelength of the trough or peak λ l,t With the corresponding absorption intensity ε l,t and constitute the fruit l Data sets Where T represents the fruit l The number of troughs or peaks; if for the universe of discourse U of rank h h ,exist Then fruit l cannot be included in grade h; if the domain of discourse U for grade h h ,exist Narrowing the range of X to The value of X is determined by the probability distribution function related to X. Then, the probability measure Poss{fruit l belongs to level h} is expressed as: When the probability measure exceeds the probability threshold, the fruit l Classified as level h from the perspective of partial internal quality; if multiple domains exist. If there is an intersection with the probability measure, then when the highest probability measure value exceeds the probability threshold, the grade corresponding to the highest probability measure value is the grade given by the fruit j after grading quality testing from the perspective of partial internal quality.

2. The fruit grading and quality detection method based on near-infrared diffuse reflectance spectral characteristics according to claim 1, characterized in that: In step 1, the specific method for collecting near-infrared spectral data corresponding to each grade standard in the fruit prediction set and fruit correction set using a diffuse reflectance method via a near-infrared spectral device in the laboratory and fruit sorting equipment is as follows: Set up a fruit prediction set Fruit Correction Set The subsets constructed by the general within the numerical range of the standard for level h are respectively the subsets corresponding to the fruit prediction subsets. and fruit correction subset The grading standards are based on the national or regional fresh fruit quality grading standards corresponding to a certain type of fruit with thin peel, under laboratory conditions, to create a fruit prediction set for that fruit. Fruit Correction Set Grading is completed through appearance and internal quality inspection. H represents the grade status that can be classified according to the relevant standards for fresh fruit quality grading in a country or region, and is a subset of fruit prediction. and fruit correction subset There is no such fruit that cannot be graded; suppose the fruit prediction set is... Near-infrared spectral data with corresponding internal quality characteristics under standard conditions for each grade were collected by a laboratory near-infrared spectroscopy device using diffuse reflectance. This yielded a fruit prediction subset. The near-infrared spectral data prediction set corresponding to a certain level h γ A,h ; Set up a fruit correction set The near-infrared spectral data corresponding to the above relationships are collected by the near-infrared spectral device of the fruit sorting equipment in a diffuse reflectance manner, then the fruit correction subset... The near-infrared spectral data correction set γ corresponding to a certain level h B,h .

Images