CN114002182B - Integrated meat multi-index rapid nondestructive testing system - Google Patents
Integrated meat multi-index rapid nondestructive testing system Download PDFInfo
- Publication number
- CN114002182B CN114002182B CN202111663645.XA CN202111663645A CN114002182B CN 114002182 B CN114002182 B CN 114002182B CN 202111663645 A CN202111663645 A CN 202111663645A CN 114002182 B CN114002182 B CN 114002182B
- Authority
- CN
- China
- Prior art keywords
- index
- prediction
- sample
- near infrared
- module
- 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.)
- Active
Links
- 235000013372 meat Nutrition 0.000 title claims abstract description 52
- 238000009659 non-destructive testing Methods 0.000 title claims abstract description 29
- 238000001514 detection method Methods 0.000 claims abstract description 85
- 238000002329 infrared spectrum Methods 0.000 claims abstract description 81
- 238000000034 method Methods 0.000 claims abstract description 56
- 238000012937 correction Methods 0.000 claims abstract description 40
- 230000008569 process Effects 0.000 claims abstract description 14
- 238000012545 processing Methods 0.000 claims abstract description 7
- 238000001228 spectrum Methods 0.000 claims description 78
- IJGRMHOSHXDMSA-UHFFFAOYSA-N Atomic nitrogen Chemical compound N#N IJGRMHOSHXDMSA-UHFFFAOYSA-N 0.000 claims description 56
- 150000001412 amines Chemical class 0.000 claims description 28
- 229910052757 nitrogen Inorganic materials 0.000 claims description 28
- XLYOFNOQVPJJNP-UHFFFAOYSA-N water Substances O XLYOFNOQVPJJNP-UHFFFAOYSA-N 0.000 claims description 25
- 238000002310 reflectometry Methods 0.000 claims description 23
- 230000000035 biogenic effect Effects 0.000 claims description 22
- 230000003595 spectral effect Effects 0.000 claims description 20
- 238000000605 extraction Methods 0.000 claims description 18
- 230000001580 bacterial effect Effects 0.000 claims description 16
- 235000020776 essential amino acid Nutrition 0.000 claims description 16
- 239000003797 essential amino acid Substances 0.000 claims description 16
- 230000004927 fusion Effects 0.000 claims description 9
- 238000012163 sequencing technique Methods 0.000 claims description 8
- 238000003307 slaughter Methods 0.000 claims description 7
- 238000012360 testing method Methods 0.000 claims description 6
- 230000005540 biological transmission Effects 0.000 claims description 4
- 229910052782 aluminium Inorganic materials 0.000 claims description 3
- 235000013622 meat product Nutrition 0.000 claims description 2
- XAGFODPZIPBFFR-UHFFFAOYSA-N aluminium Chemical compound [Al] XAGFODPZIPBFFR-UHFFFAOYSA-N 0.000 claims 1
- 229910052736 halogen Inorganic materials 0.000 claims 1
- 150000002367 halogens Chemical class 0.000 claims 1
- 230000010354 integration Effects 0.000 claims 1
- 235000019624 protein content Nutrition 0.000 description 30
- 235000019625 fat content Nutrition 0.000 description 28
- 150000004670 unsaturated fatty acids Chemical class 0.000 description 14
- 235000021122 unsaturated fatty acids Nutrition 0.000 description 14
- 241001494479 Pecora Species 0.000 description 7
- 235000015278 beef Nutrition 0.000 description 4
- 230000000052 comparative effect Effects 0.000 description 3
- 238000010586 diagram Methods 0.000 description 3
- 230000004044 response Effects 0.000 description 3
- 241000272517 Anseriformes Species 0.000 description 2
- 241000283690 Bos taurus Species 0.000 description 2
- -1 aluminum halide Chemical class 0.000 description 2
- HVYWMOMLDIMFJA-DPAQBDIFSA-N cholesterol Chemical compound C1C=C2C[C@@H](O)CC[C@]2(C)[C@@H]2[C@@H]1[C@@H]1CC[C@H]([C@H](C)CCCC(C)C)[C@@]1(C)CC2 HVYWMOMLDIMFJA-DPAQBDIFSA-N 0.000 description 2
- 238000010276 construction Methods 0.000 description 2
- 238000005516 engineering process Methods 0.000 description 2
- 238000011156 evaluation Methods 0.000 description 2
- 238000005259 measurement Methods 0.000 description 2
- 235000016709 nutrition Nutrition 0.000 description 2
- 230000000149 penetrating effect Effects 0.000 description 2
- 238000013441 quality evaluation Methods 0.000 description 2
- 230000001953 sensory effect Effects 0.000 description 2
- 239000000126 substance Substances 0.000 description 2
- 241000287828 Gallus gallus Species 0.000 description 1
- 238000004497 NIR spectroscopy Methods 0.000 description 1
- 240000007594 Oryza sativa Species 0.000 description 1
- 235000007164 Oryza sativa Nutrition 0.000 description 1
- 241000237502 Ostreidae Species 0.000 description 1
- VYPSYNLAJGMNEJ-UHFFFAOYSA-N Silicium dioxide Chemical compound O=[Si]=O VYPSYNLAJGMNEJ-UHFFFAOYSA-N 0.000 description 1
- 241000282887 Suidae Species 0.000 description 1
- 210000001015 abdomen Anatomy 0.000 description 1
- 238000013459 approach Methods 0.000 description 1
- 230000009286 beneficial effect Effects 0.000 description 1
- 238000004364 calculation method Methods 0.000 description 1
- 238000009614 chemical analysis method Methods 0.000 description 1
- 235000013330 chicken meat Nutrition 0.000 description 1
- 235000012000 cholesterol Nutrition 0.000 description 1
- 238000004891 communication Methods 0.000 description 1
- 238000012790 confirmation Methods 0.000 description 1
- 238000010411 cooking Methods 0.000 description 1
- 238000009826 distribution Methods 0.000 description 1
- 230000009977 dual effect Effects 0.000 description 1
- 230000000694 effects Effects 0.000 description 1
- 230000017525 heat dissipation Effects 0.000 description 1
- 238000000265 homogenisation Methods 0.000 description 1
- 238000012905 input function Methods 0.000 description 1
- 238000004519 manufacturing process Methods 0.000 description 1
- 238000012986 modification Methods 0.000 description 1
- 230000004048 modification Effects 0.000 description 1
- 235000020636 oyster Nutrition 0.000 description 1
- 238000004940 physical analysis method Methods 0.000 description 1
- 235000015277 pork Nutrition 0.000 description 1
- 238000005057 refrigeration Methods 0.000 description 1
- 235000009566 rice Nutrition 0.000 description 1
- 238000005070 sampling Methods 0.000 description 1
- 238000010008 shearing Methods 0.000 description 1
- 238000003860 storage Methods 0.000 description 1
- 238000012706 support-vector machine Methods 0.000 description 1
Images
Classifications
-
- G—PHYSICS
- G01—MEASURING; TESTING
- G01N—INVESTIGATING OR ANALYSING MATERIALS BY DETERMINING THEIR CHEMICAL OR PHYSICAL PROPERTIES
- G01N21/00—Investigating or analysing materials by the use of optical means, i.e. using sub-millimetre waves, infrared, visible or ultraviolet light
- G01N21/17—Systems in which incident light is modified in accordance with the properties of the material investigated
- G01N21/25—Colour; Spectral properties, i.e. comparison of effect of material on the light at two or more different wavelengths or wavelength bands
- G01N21/31—Investigating relative effect of material at wavelengths characteristic of specific elements or molecules, e.g. atomic absorption spectrometry
- G01N21/35—Investigating relative effect of material at wavelengths characteristic of specific elements or molecules, e.g. atomic absorption spectrometry using infrared light
- G01N21/359—Investigating relative effect of material at wavelengths characteristic of specific elements or molecules, e.g. atomic absorption spectrometry using infrared light using near infrared light
-
- G—PHYSICS
- G01—MEASURING; TESTING
- G01N—INVESTIGATING OR ANALYSING MATERIALS BY DETERMINING THEIR CHEMICAL OR PHYSICAL PROPERTIES
- G01N21/00—Investigating or analysing materials by the use of optical means, i.e. using sub-millimetre waves, infrared, visible or ultraviolet light
- G01N21/17—Systems in which incident light is modified in accordance with the properties of the material investigated
- G01N21/25—Colour; Spectral properties, i.e. comparison of effect of material on the light at two or more different wavelengths or wavelength bands
- G01N21/31—Investigating relative effect of material at wavelengths characteristic of specific elements or molecules, e.g. atomic absorption spectrometry
- G01N21/35—Investigating relative effect of material at wavelengths characteristic of specific elements or molecules, e.g. atomic absorption spectrometry using infrared light
- G01N21/3554—Investigating relative effect of material at wavelengths characteristic of specific elements or molecules, e.g. atomic absorption spectrometry using infrared light for determining moisture content
-
- G—PHYSICS
- G01—MEASURING; TESTING
- G01N—INVESTIGATING OR ANALYSING MATERIALS BY DETERMINING THEIR CHEMICAL OR PHYSICAL PROPERTIES
- G01N21/00—Investigating or analysing materials by the use of optical means, i.e. using sub-millimetre waves, infrared, visible or ultraviolet light
- G01N21/17—Systems in which incident light is modified in accordance with the properties of the material investigated
- G01N21/25—Colour; Spectral properties, i.e. comparison of effect of material on the light at two or more different wavelengths or wavelength bands
- G01N21/31—Investigating relative effect of material at wavelengths characteristic of specific elements or molecules, e.g. atomic absorption spectrometry
- G01N21/35—Investigating relative effect of material at wavelengths characteristic of specific elements or molecules, e.g. atomic absorption spectrometry using infrared light
- G01N21/3563—Investigating relative effect of material at wavelengths characteristic of specific elements or molecules, e.g. atomic absorption spectrometry using infrared light for analysing solids; Preparation of samples therefor
Abstract
The invention discloses an integrated meat multi-index rapid nondestructive testing system, which comprises a spectrometer, a data acquisition module and a data processing module, wherein the spectrometer is used for acquiring near infrared spectrum data of a sample to be tested; the industrial flat plate comprises: the model embedding module is used for storing a plurality of index prediction models; the model determining module is connected with the model embedding module and used for calling an index prediction model; the index prediction module is connected with the spectrometer and the model determination module and used for receiving the near infrared spectrum data of the sample to be tested and predicting the index data of the sample to be tested by combining the called prediction model; the number of detector elements of the spectrometer is determined according to the final minimum resolution, and the resolution is controlled to be the final minimum resolution in the near infrared spectrum data acquisition process of the sample to be detected. The invention has the advantages of realizing automatic black and white correction, realizing nondestructive prediction of multidimensional indexes, improving the efficiency of model establishment while maintaining the stability of model establishment, reducing the cost and optimizing the volume of a detection system.
Description
Technical Field
The invention relates to the technical field of meat multi-index detection. More specifically, the invention relates to an integrated multi-index rapid nondestructive testing system for meat.
Background
The meat quality mainly comprises edible quality, processing quality, nutritional quality and safety quality. Most of the existing meat quality evaluation methods mainly adopt physical and chemical detection and sensory evaluation methods, the judgment result of the sensory evaluation is not objective, the physical and chemical detection mainly adopts a physical or chemical analysis method to detect the indexes of meat, the method has high detection precision and objective and credible result, but the detection pretreatment is complicated, the sample can be damaged, and the requirement of short-time detection of a large number of samples in actual production is difficult to meet, wherein the indexes for evaluating the edible quality mainly comprise meat color, tenderness and the like; the indexes for evaluating the processing quality mainly include water-binding capacity, pH and the like; the indexes for evaluating the nutritional quality mainly comprise protein content, fat content, moisture, UFA (unsaturated fatty acid), total amount of essential amino acids, etc.; the indexes for evaluating the safety quality mainly comprise the total number of bacterial colonies, volatile basic nitrogen, biogenic amine and the like.
In recent years, near infrared spectroscopy has been successfully used for quality evaluation of meat products, for example, a rapid nondestructive testing method named as fresh beef multi-index, and patent application No. 201510965311.6 discloses that cholesterol, moisture, fat and protein contents, shearing force and water holding capacity of fresh beef are synchronously tested by constructing a multi-index prediction model. Firstly, the problem that the number of elements of a spectrometer is large due to multi-index detection, so that the equipment volume is increased and the equipment cost is increased is solved urgently at present; secondly, the index prediction model corresponding to the least square method resume is applied by utilizing the spectral data of the correction set and the reference value of the sample to be tested, so that the problems of large spectral data capacity and low calculation efficiency in the process of constructing the index prediction model exist, and the problems are more prominent particularly in the process of predicting multiple indexes of meat.
Disclosure of Invention
An object of the present invention is to solve at least the above problems and to provide at least the advantages described later.
The invention also aims to provide an integrated meat multi-index rapid nondestructive testing system, which can realize nondestructive prediction of multi-dimensional indexes, reduce cost and optimize the size of the testing system.
To achieve these objects and other advantages in accordance with the purpose of the invention, there is provided an integrated meat multi-index rapid nondestructive testing system, comprising:
the spectrometer is used for acquiring near infrared spectrum data of a sample to be detected;
an industrial flat panel, comprising:
the model embedding module is used for storing a plurality of index prediction models;
the model determining module is connected with the model embedding module and used for calling at least one index prediction model;
the index prediction module is connected with the spectrometer and the model determination module and used for receiving near infrared spectrum data of a sample to be tested and predicting the index data of the sample to be tested by combining the called prediction model;
the method comprises the following steps of determining the number of detector elements of the spectrometer according to the final minimum resolution, controlling the resolution to be the final minimum resolution in the acquisition process of near infrared spectrum data of a sample to be detected, and determining the final minimum resolution, wherein the determination comprises the following steps:
s1, acquiring spectral wavelength data of each sample in the sample set based on different resolutions;
s2, determining multiple prediction indexes according to multiple index prediction models stored by the model embedding module, establishing an index prediction model of any prediction index under the resolution by applying a chemometrics method according to spectral wavelength data under the resolution and a reference value corresponding to the prediction index under different resolutions, and determining a prediction correlation coefficient of the index prediction model;
s3, determining an inflection point of a prediction correlation coefficient of any prediction index under different resolutions for the resolution, and determining the resolution corresponding to the inflection point as the minimum resolution of the prediction index;
and S4, determining the maximum value of the minimum resolutions of all the prediction indexes as the final minimum resolution.
Preferably, the multiple prediction indexes are determined to be at least 3 of flesh color, tenderness, water binding capacity, pH, protein content, fat content, moisture, colony total number, volatile basic nitrogen, UFA, total amount of essential amino acids and biogenic amine according to the multiple index prediction models;
wherein, when the prediction indexes are flesh color, water content and water content, the collection wave band is 400-1050 nm;
when the prediction indexes are tenderness, pH, protein content and fat content, the acquisition waveband is 900-1700 nm;
when the prediction indexes are total colony count, volatile basic nitrogen, UFA, total essential amino acid and biogenic amine, the collection wave bands are 400-1050nm and 900-1700 nm;
when the acquisition bands corresponding to the multiple prediction indexes simultaneously comprise 400-1050nm and 900-1700nm, the multiple prediction indexes are divided into an A group with the acquisition band comprising 400-1050nm and a B group with the acquisition band comprising 900-1700nm, and the final minimum resolution of all the prediction indexes corresponding to the A group in the acquisition band of 400-1050nm and the final minimum resolution of all the prediction indexes corresponding to the B group in the acquisition band of 900-1700nm are respectively determined.
Preferably, the different resolutions of the predictors of group A are in the range of 0.4 to 8nm and the predictors of group B are in the range of 15 to 45 nm.
Preferably, the method for constructing each index prediction model includes the following steps:
determining a sample set, and determining a reference value of each prediction index corresponding to each sample in the sample set, wherein the attributes of the samples in the sample set are the same as the attributes of the samples to be detected;
determining an acquisition waveband according to a plurality of prediction indexes, and acquiring near infrared spectrum data of each sample in a sample set in a corresponding acquisition waveband, wherein when the prediction indexes are total bacterial count, volatile basic nitrogen, UFA, total essential amino acid or biogenic amine, spectrum fusion is carried out on the near infrared spectrum data of 400-plus 1050nm and 900-plus 1700nm to obtain the near infrared spectrum data of the coverage wavelength of 400-plus 1700 nm;
calculating the reflectivity difference between adjacent wave crests and wave troughs according to the wavelength of each piece of near infrared spectrum data from small to large, sequencing the data in sequence, and constructing a characteristic extraction spectrum taking the sequencing number as a horizontal coordinate and the reflectivity difference as a vertical coordinate;
for any prediction index, establishing an index prediction model corresponding to the prediction index by applying a chemometrics method according to the feature extraction spectrum and a reference value corresponding to the prediction index;
the industrial flat plate also comprises a spectrum processing module which is connected with the spectrometer and used for converting the acquired near infrared spectrum data into a characteristic extraction spectrum, and the index prediction module receives the characteristic extraction spectrum of the sample to be detected and predicts the index data of the sample to be detected by combining with the called prediction model.
Preferably, the samples of the sample set and the samples to be detected are samples of the same genus, the varieties of the samples of the sample set comprise at least 3, the samples of each variety comprise at least 5 months of age, the samples of each month comprise at least 5 parts, and the samples of each part comprise at least 45min, 24h, 72h and 120h time points after slaughter.
Preferably, the industrial flat panel further comprises:
the calibration reference module is used for storing an average spectrum X and a spectrum threshold value of a sample belonging to the same type as the sample to be detected, wherein the average spectrum X, an extremely-low reflectivity spectrum M and an extremely-high reflectivity spectrum M are calculated based on near infrared spectrum data of at least 1000 samples belonging to the same type as the sample to be detected, and the spectrum threshold value is determined according to the average spectrum X, the extremely-low reflectivity spectrum M and the extremely-high reflectivity spectrum M;
and the black-and-white correction module is connected with the correction reference module and the spectrometer and is used for acquiring the No. N +1 near infrared spectrum data after the spectrometer continuously acquires the near infrared spectrum data of the sample to be detected to a set number N, comparing the No. N +1 near infrared spectrum data with the average spectrum X, calculating the Mahalanobis distance, judging whether the Mahalanobis distance is within a spectrum threshold value, if so, determining that the No. N +1 near infrared spectrum data is normal, and if not, re-determining the No. N +1 near infrared spectrum data after correction.
Preferably, the method further comprises the following steps:
the top end of the detection camera bellows is provided with a detection window, the spectrometer is arranged in the detection camera bellows, and a condensing lens and a light source are sequentially arranged in the detection camera bellows from top to bottom below the detection window;
black and white correction subassembly, it is located including the rotation detect in the camera bellows and be located detection window below the disc, be used for the drive disc pivoted motor, be equipped with blank, blackboard on the disc, run through and be equipped with the through-hole, wherein, the motor with black and white correction module connects for control motor is rotatory the disc, so that blank, blackboard, through-hole selectivity and the coaxial setting of detection window.
Preferably, the light source comprises 2 aluminum halide reflector lamp cups symmetrically arranged 15-25mm below the detection window, and the central axis of each reflector lamp cup is 55-65 degrees to the central axis of the detection window.
Preferably, the method further comprises the following steps: and the result display module is connected with the index prediction module and used for receiving and displaying the prediction result.
Preferably, the method further comprises the following steps: and the data transmission module is connected with the index prediction module and is used for receiving and transmitting the predicted index data to the user side.
The invention at least comprises the following beneficial effects:
firstly, embedding a multi-dimensional index prediction model to realize multi-dimensional index nondestructive prediction, arranging a black and white correction component in a system, and arranging a black and white correction module in a matching way to realize black and white correction at the initial or intermediate detection stage of the system, wherein the black and white correction in the middle of detection is judged whether to be carried out according to an average spectrum X and a spectrum threshold value, so that the time and labor are saved when the correction requirement is met and excessive correction is avoided;
secondly, extracting spectrum jump characteristics of the obtained near infrared spectrum data by utilizing the reflectivity intensity of the wave crest and the wave trough of the near infrared spectrum in a certain area, maintaining the stability of model establishment, improving the efficiency of model establishment, and improving the efficiency of later-stage detection and matching prediction of a sample to be detected;
and thirdly, determining minimum resolution based on a single prediction index, and determining the final minimum resolution of the spectrometer by integrating a plurality of indexes to serve as the spectrum proper resolution so as to achieve the purposes of reducing cost and optimizing the volume of a detection instrument (spectrometer), and meanwhile, simplifying the control of detection operation conditions under the condition of not reducing the detection accuracy, especially under the condition of multi-index detection of a large batch of samples.
Additional advantages, objects, and features of the invention will be set forth in part in the description which follows and in part will become apparent to those having ordinary skill in the art upon examination of the following or may be learned from practice of the invention.
Drawings
FIG. 1 is a block diagram of a flow chart of an integrated meat multi-index rapid nondestructive testing system according to one embodiment of the present invention;
FIG. 2 is a block diagram of a flow chart of the integrated meat multi-index rapid nondestructive testing system according to one embodiment of the present invention;
FIG. 3 is a schematic structural diagram of an integrated meat multi-index rapid nondestructive testing system according to one embodiment of the present invention;
FIG. 4 is a graph of a near infrared spectrum according to one embodiment of the present invention;
fig. 5 is a characteristic extraction spectrum according to one embodiment of the present invention.
The reference signs are: 1-a detection window; 2-a disc; 20-blackboard; 21-white board; 22-a through hole; 3-a condenser lens; 4-a light source; 5-a spectrometer; 6-industrial flat plate; 7-detection dark box.
Detailed Description
The present invention is further described in detail below with reference to examples so that those skilled in the art can practice the invention with reference to the description.
As shown in fig. 1 and 3, the present invention provides an integrated meat multi-index rapid nondestructive testing system, comprising:
a spectrometer 5 for acquiring near infrared spectrum data (near infrared spectrum) of a sample to be measured;
an industrial plate 6 comprising:
the model embedding module is used for storing a plurality of index prediction models;
the model determining module is connected with the model embedding module and used for calling at least one index prediction model according to the detection requirement (at least one index type to be determined, wherein the at least one index type to be determined is contained in a plurality of prediction index coverage ranges corresponding to the index prediction models) of the sample to be detected when the model determining module is applied to the detection of the sample to be detected and the called at least one index prediction model is matched with the detection requirement of the sample to be detected;
the index prediction module is connected with the spectrometer 5 and the model determination module, and is used for receiving near infrared spectrum data of the sample to be tested from the spectrometer 5 and predicting the index data of the sample to be tested by combining with the called at least one prediction model;
the number of detector elements of the spectrometer 5 is determined according to the final minimum resolution, the resolution is controlled to be the final minimum resolution in the acquisition process of the near infrared spectrum data of the sample to be detected, and the determination of the final minimum resolution comprises the following steps:
s1, acquiring spectral wavelength data of each sample in the sample set based on different resolutions;
s2, determining multiple prediction indexes according to multiple index prediction models stored by the model embedding module, establishing an index prediction model of any prediction index under the resolution by applying a chemometrics method according to spectral wavelength data under the resolution and a reference value corresponding to the prediction index under different resolutions, and determining a prediction correlation coefficient of the index prediction model;
s3, determining an inflection point of a prediction correlation coefficient of any prediction index under different resolutions for the resolution, and determining the resolution corresponding to the inflection point as the minimum resolution of the prediction index;
and S4, determining the maximum value of the minimum resolutions of all the prediction indexes as the final minimum resolution.
In the technical scheme, the integrated meat multi-index rapid nondestructive detection system can be installed on a cold chain vehicle, in a cold storage and in a supermarket, the index prediction models stored in the integrated meat multi-index rapid nondestructive detection system can be matched and set according to actual application scenes, the detection requirements of corresponding application scenes can be met, after the assembly is finished, a new index prediction model can be embedded according to the later-stage detection requirements on the basis of meeting the final minimum resolution detection requirements, the response range of the spectrometer 5 is 400-plus-1000 nm and/or 900-plus-1700 nm, the embedded index prediction models are mainly determined according to the prediction indexes corresponding to the stored index prediction models, and if the spectrometer 5 contains 900-plus-1700 nm, the system is equipped with TEC refrigeration and fan heat dissipation; the number of detector elements of the spectrometer 5 is determined according to the final minimum resolution, specifically, for example, if the minimum resolution of the 900-1700nm band is determined to be 32nm, the number of detector elements of the spectrometer 5 can be reduced to 25, that is, when the prediction index is determined, the optimal number of detector elements of the spectrometer 5 can be determined according to the final minimum resolution; the chemometrics method can be one of least squares regression or support vector machine regression algorithm; by adopting the technical scheme, multi-dimensional index nondestructive prediction is realized by embedding a multi-dimensional index prediction model, the minimum resolution is determined based on a single prediction index, and the final minimum resolution of the spectrometer 5 is determined by integrating a plurality of indexes and is used as the spectrum proper resolution, so that the purposes of reducing the cost and optimizing the volume of a detection instrument are achieved, meanwhile, the control of detection operation conditions is simplified under the condition of not reducing the detection accuracy, and particularly under the condition of multi-index detection of a large batch of samples.
In another technical scheme, at least 3 of flesh color, tenderness, water retention capacity, pH, protein content, fat content, moisture, total bacterial count, volatile basic nitrogen, UFA, total essential amino acid amount and biogenic amine are determined according to a plurality of index prediction models, namely the index prediction model stored by the integrated meat multi-index rapid nondestructive testing system at least meets the measurement of 3 indexes, and the multi-index rapid nondestructive testing is realized;
wherein, when the prediction indexes are flesh color, water content and water content, the collection wave band is 400-1050 nm;
when the prediction indexes are tenderness, pH, protein content and fat content, the acquisition waveband is 900-1700 nm;
when the prediction indexes are total colony count, volatile basic nitrogen, UFA, total essential amino acid and biogenic amine, the collection wave bands are 400-1050nm and 900-1700 nm;
when the acquisition wave bands corresponding to the multiple prediction indexes simultaneously comprise 400-1050nm and 900-1700nm, the multiple prediction indexes are divided into an A group of which the acquisition wave band comprises 400-1050nm and a B group of which the acquisition wave band comprises 900-1700nm, and the final minimum resolution of all the prediction indexes corresponding to the A group at the acquisition wave band of 400-1050nm and the final minimum resolution of all the prediction indexes corresponding to the B group at the acquisition wave band of 900-1700nm are respectively determined. In the above technical solution, the configuration basis of the spectrometer 5 is as follows:
(1) single detector D1, response range (acquisition band): 400-1050 nm; the predictable indicators include flesh color, water content and moisture;
(2) single detector D2, response range (acquisition band): 900-1700 nm; the predictable indicators include tenderness, pH, protein content, fat content;
(3) dual detectors, which are a collection of single detector D1 and single detector D2, predictive indicators include flesh color, water retention, moisture, tenderness, pH, protein content, fat content, total number of colonies, volatile basic nitrogen, UFA, total amount of essential amino acids, biogenic amines; with this approach, it is used to determine the configuration of the spectrometer 5.
In another technical scheme, in the process of measuring the final minimum resolution corresponding to the prediction indexes of the group A and the group B, for the group A:
a1, setting different resolutions at intervals in the resolution range of 0.4-8 nm;
a2, determining whether the prediction index of the group A comprises one of colony total number, volatile basic nitrogen, UFA, essential amino acid total amount and biogenic amine;
a3, if not, acquiring spectral wavelength data (the acquisition wave band is 400-1050 nm) of each sample in the sample set based on different resolutions, for any prediction index, under different resolutions, according to the spectral wavelength data under the resolution and a reference value corresponding to the prediction index, establishing an index prediction model of the prediction index under the resolution by using a chemometrics method, determining a prediction correlation coefficient of the index prediction model, determining an inflection point of the prediction correlation coefficient of any prediction index under different resolutions, and determining the resolution corresponding to the inflection point as the minimum resolution of the prediction index;
a4, if yes, determining the minimum resolution for the prediction index with the collection wave band of 400-1050nm (A3), and determining the minimum resolution for any one of the total number of colonies, volatile basic nitrogen, UFA, total amount of essential amino acids and biogenic amine, wherein the method comprises the following steps:
a4a, acquiring spectral wavelength data (the acquisition wave band is 400-1050 nm) of each sample in a sample set based on different resolutions, synchronously acquiring the wave band range of 900-1700nm under the set resolution, and setting the resolution to be set according to prediction indexes and experience of technicians in the field, wherein the resolution can be determined by multiple tests;
a4b, performing spectrum fusion on the near infrared spectrum data of 400-1050nm and 900-1700nm to obtain the near infrared spectrum data with the coverage wavelength of 400-1700nm, and determining the prediction correlation coefficient of the index prediction model;
a4c, determining inflection points of prediction correlation coefficients of corresponding prediction indexes under different resolutions for the resolution, and determining the resolution corresponding to the inflection points as the minimum resolution of the prediction indexes;
for group B:
b1, setting different resolutions at intervals in the resolution range of 15-45 nm;
b2, determining whether the prediction indexes of the group B comprise one of colony total number, volatile basic nitrogen, UFA, total essential amino acid amount and biogenic amine;
b3, if not, acquiring spectral wavelength data (the acquisition wavelength band is 900-1700 nm) of each sample in the sample set based on different resolutions, for any prediction index, under different resolutions, according to the spectral wavelength data under the resolution and the reference value corresponding to the prediction index, establishing an index prediction model of the prediction index under the resolution by using a chemometrics method, determining a prediction correlation coefficient of the index prediction model, determining an inflection point of the prediction correlation coefficient of any prediction index under different resolutions to the resolution, and determining the resolution corresponding to the inflection point as the minimum resolution of the prediction index;
b4, if yes, determining the minimum resolution for the prediction index with the acquisition wave band of 900-1700nm as above, and determining the minimum resolution for any one of the colony count, the volatile basic nitrogen, the UFA, the total essential amino acid and the biogenic amine, wherein the method comprises the following steps:
b4a, acquiring spectral wavelength data (the acquisition wave band is 900-1700 nm) of each sample in the sample set based on different resolutions, synchronously acquiring the wave band range of 400-1050nm under the set resolution, and setting the resolution to be set according to the prediction index and the experience of the technicians in the field, wherein the resolution can be determined by multiple tests;
B4B, performing spectrum fusion on the near infrared spectrum data of 400-1050nm and 900-1700nm to obtain the near infrared spectrum data with the coverage wavelength of 400-1700nm, and determining the prediction correlation coefficient of the index prediction model;
and B4c, determining the inflection point of the prediction correlation coefficient of the corresponding prediction index under different resolutions for the resolution, and determining the resolution corresponding to the inflection point as the minimum resolution of the prediction index. By adopting the scheme, different resolution ranges are determined according to different acquisition wave bands, and the minimum resolution of the prediction index corresponding to the acquisition wave band is determined in the resolution range, wherein the coverage wavelength of 400-1700nm is determined in a segmented manner, so that the measurement accuracy of the whole device is improved, and the influence of irrelevant acquisition wave bands on the accuracy of the detection result is avoided.
In another technical scheme, the method for constructing each index prediction model comprises the following steps:
determining a sample set, and determining a reference value of each prediction index corresponding to each sample in the sample set, wherein the attributes of the samples in the sample set are the same as those of the samples to be detected, specifically: the attribute of the beef sample is of the genus of cattle, the attribute of the pork sample is of the genus of pig, the attribute of the mutton sample is of the genus of sheep and the like, namely the beef is divided into pigs, cattle, sheep, chickens, ducks, geese and the like according to the attributes, and the method for determining the reference value of each prediction index is described by specifically referring to the following table 1:
TABLE 1 summary of reference values for meat index
Determining an acquisition waveband according to a plurality of prediction indexes, and acquiring near infrared spectrum data of each sample in a sample set in a corresponding acquisition waveband, wherein when the prediction indexes comprise at least one of total bacterial colony number, volatile basic nitrogen, UFA, total essential amino acid amount and biogenic amine, spectrum fusion is carried out on the near infrared spectrum data of 400-plus 1050nm and 900-plus 1700nm for the corresponding prediction indexes to obtain the near infrared spectrum data with the covering wavelength of 400-plus 1700 nm;
calculating the reflectivity difference between adjacent wave crests and wave troughs according to the wavelength of each piece of near infrared spectrum data from small to large, sequencing the data in sequence, and constructing a characteristic extraction spectrum taking the sequencing number as a horizontal coordinate and the reflectivity difference as a vertical coordinate;
for any prediction index, establishing an index prediction model corresponding to the prediction index by applying a chemometrics method according to the feature extraction spectrum and a reference value corresponding to the prediction index;
the industrial flat plate 6 further comprises a spectrum processing module, which is connected with the spectrometer 5 and used for converting the acquired near infrared spectrum data into a characteristic extraction spectrum;
and the index prediction module is connected with the spectrum processing module and the model determination module and used for receiving the characteristic extraction spectrum of the sample to be tested and predicting the index data of the sample to be tested by combining the called prediction model. In the above technical solution, it is determined that the prediction model to be invoked includes a prediction model corresponding to protein content, a prediction model corresponding to fat content, and a prediction model corresponding to the total number of colonies as an example; the prediction method of the prediction index comprises two methods, specifically as follows:
first, when near infrared spectral data is used in the called predictive model construction process;
when the prediction index is the protein content, predicting the protein content of the sample to be detected by using near infrared spectrum data within the range of 900-;
when the prediction index is the fat content, predicting the fat content of the sample to be detected by using near infrared spectrum data within the range of 900-;
when the prediction index is the total number of the bacterial colonies, performing spectrum fusion on the near infrared spectrum data of 400-1050nm and 900-1700nm to obtain the near infrared spectrum data with the coverage wavelength of 400-1700nm, and predicting the fat content of the sample to be detected by combining the invoked prediction model corresponding to the total number of the bacterial colonies;
second, when the called prediction model construction process uses feature extraction spectra;
when the prediction index is the protein content, extracting a spectrum by using the characteristics within the range of 900-;
when the prediction index is the fat content, extracting a spectrum by using the characteristics within the range of 900-;
and when the prediction index is the total number of the colonies, performing spectrum fusion on the near infrared spectrum data of 400-1050nm and 900-1700nm to obtain the near infrared spectrum data of the coverage wavelength of 400-1700nm, then acquiring a corresponding characteristic extraction spectrum according to the near infrared spectrum data of the coverage wavelength of 400-1700nm, and predicting the fat content of the sample to be detected by combining the invoked prediction model corresponding to the total number of the colonies. By adopting the scheme, the second method is relative to the first method, for the obtained near infrared spectrum data, the spectrum jump characteristics are extracted by utilizing the reflectivity intensity of the wave crest and the wave trough of the near infrared spectrum in a certain area, the model building efficiency is improved while the model building stability is maintained, and the efficiency of the matching prediction of the later-stage detection sample to be detected is improved.
In another technical scheme, the samples of the sample set and the samples to be detected are samples of the same genus, the varieties of the samples of the sample set comprise at least 3, the samples of each variety comprise at least 5 months of age, the samples of each month comprise at least 5 parts, and the samples of each part at least comprise time points of 45min, 24h, 72h and 120h after slaughter. In the technical scheme, the varieties of the samples of the sample set comprise at least 3, the samples of each variety comprise at least 5 months of age, the samples of each month of age corresponding to each variety comprise at least 5 parts, and the samples of each part of each variety and each month of age corresponding to each month of age at least comprise time points of 45min, 24h, 72h and 120h after slaughter, by adopting the scheme, the problem that a traditional majority of meat quality prediction models are only constructed on the basis of a certain sample set with higher homogenization degree, namely the sample set constructed on the basis of single variety, single month of age, single part of meat and single time sampling is solved, so that the applicability and robustness of the constructed prediction models are limited to a certain extent, the application of a near infrared spectrum technology in the meat quality prediction field is limited to a certain extent, the on-line and practicability of a meat quality nondestructive detection system are also limited, and conversely, in the sample set confirmation process, the comprehensiveness of the sample set is determined, and the practicability of the meat quality nondestructive testing system is improved.
In another solution, the industrial flat plate 6 further comprises:
the calibration reference module is used for storing an average spectrum X and a spectrum threshold of a sample which belongs to the same as the sample to be detected, wherein the average spectrum X, an extremely low reflectivity spectrum M and an extremely high reflectivity spectrum M are calculated based on near infrared spectrum data (standard near infrared standard data, particularly, cheap data is not generated) of at least 1000 samples which belong to the same as the sample to be detected, and the spectrum threshold is determined by the average spectrum X, the extremely low reflectivity spectrum M and the extremely high reflectivity spectrum M;
the black-and-white correction module is connected with the correction reference module and the spectrometer 5 and used for acquiring the No. N +1 near infrared spectrum data after the spectrometer 5 continuously acquires the near infrared spectrum data of the sample to be detected to a set number N, comparing the No. N +1 near infrared spectrum data with the average spectrum X, calculating the Mahalanobis distance, judging whether the Mahalanobis distance is within a spectrum threshold value, if so, determining that the No. N +1 near infrared spectrum data is normal, taking the No. N +1 as the No. 1 by the spectrometer 5, continuously acquiring the near infrared spectrum data of the sample to be detected to the set number N, and judging; if not, the No. N +1 near infrared spectrum data is measured again after correction. In the above technical solution, the black-and-white correction module is further used for initial correction, N is preferably 45-55, and more preferably 50, and by adopting this scheme, the system sets the black-and-white correction component, and sets the black-and-white correction module in a supporting manner, for implementing black-and-white correction at the initial or intermediate stage of detection of the system, wherein the intermediate black-and-white correction of the detection is judged whether to be performed according to the average spectrum X and the spectrum threshold, so that time and labor are saved when the correction requirement is met and excessive correction is avoided.
In another technical solution, the integrated meat multi-index rapid nondestructive testing system as illustrated in fig. 2-3 further includes:
the spectrometer 5 is arranged in the detection dark box 7;
the detection window 1 is arranged at the top end of the detection camera bellows 7;
a light source 4 disposed below the detection window 1 in a symmetrical state at a certain angle;
a condenser lens 3 provided between the light source 4 and the detection window 1;
the black-and-white correction assembly comprises a disc 2 and a motor, wherein the disc 2 is rotatably arranged in the detection camera bellows 7 and is positioned below the detection window 1, the motor is used for driving the disc 2 to rotate, and a white board 21, a blackboard 20 and a through hole 22 are arranged on the disc 2 in a penetrating manner;
the motor is connected with the black-and-white correction module and used for controlling the motor to rotate the disc 2, so that the white board 21, the blackboard 20 and the through hole 22 are selectively arranged coaxially with the detection window 1. In the above technical solution, the detection dark box 7 is preferably cube-shaped, the detection window 1 is circular with a diameter of about 40mm, the detection window 1 is made of quartz glass sheet and is flush with the device housing, and the detection window 1 has two functions: the device is a platform for placing a sample to be detected and also a spectral information acquisition window of the sample to be detected, a light source 4 irradiates the sample at a certain angle, diffuse reflection light is collected together through a condensing lens 3 and is transmitted to a spectrometer 5, a black-white correction component is used for being matched with a black-white correction module to be connected to realize initial black-white correction and middle black-white correction of a system, the radius of a through hole 22 is larger than that of a detection window 1, when near infrared spectrum data of the sample to be detected is acquired, a motor drives a disc 2 to rotate, the through hole 22 and the detection window 1 are coaxially arranged, when a white board 21 is corrected, the motor drives the disc 2 to rotate, the white board 21 and the detection window 1 are coaxially arranged, and when the white board 21 is corrected, the motor drives the disc 2 to rotate, the blackboard 20 and the detection window 1 are coaxially arranged; the condenser lens 3 is arranged between the light source 4 and the detection window 1, and is used for converging and collimating the spectrum reflected by or returned from the sample to be detected, so that the reflected spectrum information of the sample to be detected is radiated into the slit of the spectrometer 5 as much as possible, and the effects of improving the signal-to-noise ratio and reducing the exposure time of the spectrometer 5 during working are achieved; in the using process, the method comprises the following steps:
(1) switching on a system power supply;
(2) placing a sample to be detected on a detection window 1;
(3) the method for detecting the start of operation on the operation screen of the industrial flat plate 6 specifically comprises the following steps:
opening detection software on the industrial flat plate 6, automatically detecting the connection condition of each hardware and each module, and selecting the attribute of the sample to be detected and predicting indexes after the connection condition is normal;
the system determines to select a single detector D1, a single detector D2 and double detectors (a single detector D1 and a single detector D2) according to the set attributes of the sample to be detected and the index number and range of the prediction index, wherein if the prediction index comprises one of flesh color, water content and moisture, the single detector D1 is selected; if the prediction index comprises one of tenderness, pH value, protein content and fat content, selecting a single detector D2; if the prediction indexes comprise the total number of colonies, volatile basic nitrogen, UFA, the total amount of essential amino acids and biogenic amine, selecting a double detector, and determining to select a spectrum fusion algorithm for spectrum fusion for the type indexes;
taking prediction indexes including protein content, fat content and total colony number as examples, selecting a double detector;
determining a prediction model to be called (a prediction model corresponding to protein content, a prediction model corresponding to fat content and a prediction model corresponding to the total number of bacterial colonies) according to the determined detection sample and the prediction index, and starting a model embedding module to drop data of the relevant prediction model into a cache;
when the method is started for the first time, clicking a 'start detection' key to perform initial black-and-white correction, and calling a black-and-white correction module to perform initialized black-and-white reference correction;
acquiring near infrared spectrum data of a sample to be detected within a range of 400-1050mm of an acquisition waveband by using a detector D1, wherein in the acquisition process, the resolution of the detector D1 is the final minimum resolution of the acquisition waveband of 400-1050 nm;
acquiring near infrared spectrum data of a sample to be detected within a range of acquisition waveband of 900-;
according to the near infrared spectrum data of the sample to be detected, and by combining at least one called prediction model, predicting index data of the sample to be detected;
the black-white correction assembly comprises a disc 2 and a motor, wherein the disc 2 is rotatably arranged in the detection camera bellows 7 and is positioned below the detection window 1, the motor is used for driving the disc 2 to rotate, and a white board 21, a blackboard 20 and a through hole 22 are arranged on the disc 2 in a penetrating manner; when a near infrared spectrum data acquisition process of a sample to be detected is carried out, the motor drives the disc 2 to rotate, namely, the through hole 22 and the detection window 1 are coaxially arranged, when the white board 21 is corrected, the motor drives the disc 2 to rotate, namely, the white board 21 and the detection window 1 are coaxially arranged, and when the white board 21 is corrected, the motor drives the disc 2 to rotate, namely, the black board 20 and the detection window 1 are coaxially arranged; by adopting the scheme, the motor is matched with the disc 2, so that black and white correction is facilitated.
In another technical scheme, the light source 4 comprises 2 aluminum halide reflector cups symmetrically arranged 15-25mm below the detection window 1, and the central axis of each reflector cup is 55-65 degrees to the central axis of the detection window 1. The power of the reflector lamp cup is 5W to provide near infrared light in the range of 300-2500nm, and the preferred light source 4 is disposed 20mm below the detection window 1. By adopting the scheme, the arrangement of the distribution state of the two reflector lamp cups needs to provide enough space positions for the condenser lens 3 on one hand, and focuses the light sources 4 of the two reflector lamp cups on the detection window 1 to form light spots (specifically, within the range of 10-20 mm) with certain sizes on the other hand.
In another technical solution, the integrated multi-index rapid nondestructive testing system for meat further comprises: and the result display module is connected with the index prediction module and used for receiving and displaying the prediction result. With this solution, the industrial panel 6 has an operation screen which realizes the input function of the relevant signals in the whole detection process and synchronously realizes the function of receiving and displaying the prediction result.
In another technical solution, the integrated multi-index rapid nondestructive testing system for meat further comprises: and the data transmission module is connected with the index prediction module and used for receiving and transmitting the predicted index data to the user side. The data transmission module receives the index data predicted by the index prediction module, transmits the received index data to a user side (a mobile phone, a computer and the like) based on WIFI and Bluetooth, displays the index data on an APP program interface matched with the user side, and can realize the function of detection by using the detection system through the user side remote control industrial panel 6. By adopting the scheme, the communication connection is established with the user side, and the remote receiving and control are convenient.
< example 1>
The meat multi-index rapid nondestructive testing method comprises the following steps:
step one, determining minimum spectral resolution
Determining the multi-dimensional indexes of the prediction model to be constructed, including flesh color, water content, protein content, fat content, moisture, total colony count, volatile basic nitrogen and biogenic amine;
determining the final minimum resolution of 2.8 nm of the acquisition waveband of 400-1050nm according to the multi-dimensional index, wherein the inflection point of most indexes in the corresponding acquisition waveband is as shown in the following table 2;
TABLE 2400-1050 nm acquisition band minimum resolution
| Prediction index | Color of meat | Water binding force | Moisture content |
| Inflection point | 3.8nm | 4.1nm | 4.6nm |
| Prediction index | Total number of colonies | Volatile basic nitrogen | Biogenic amines |
| Inflection point | 3.5nm | 2.8nm | 4.6nm |
Determining the final minimum resolution of 26 nm with the acquisition band of 900-1700nm according to the multi-dimensional index, wherein the inflection points of most indexes in the corresponding acquisition band are shown in the following table 3;
table 3900-1700 nm acquisition waveband minimum resolution
| Prediction index | Protein content | Fat content | |
| Inflection point | 32nm | 32nm | |
| Prediction index | Total number of colonies | Volatile basic nitrogen | Biogenic amines |
| Inflection point | 26nm | 32nm | 40nm |
Step two, determining the number of detector elements of the spectrometer based on minimum resolution, and constructing an integrated meat multi-index rapid nondestructive detection system, wherein an index prediction model corresponding to meat color, water system capacity, protein content, fat content, water, total bacterial count, volatile basic nitrogen and biogenic amine one by one is stored in a model embedding module in an industrial flat plate of the system, and for any prediction index, the method for constructing the index prediction model comprises the following steps:
2.1 selection of samples
Selecting 3 varieties of sheep, randomly selecting 5 sheep with different ages at each month, and collecting the carcass meat samples of the sheep oyster, the dragon rice, the Lin meat, the outer ridge and the belly meat at time points of 45min, 24h, 72h and 120h after slaughtering as sample sets;
2.2, respectively acquiring the near infrared spectrum data of each sample in the acquisition bands of 400-1050nm and 900-1700nm by taking the corresponding final minimum resolution as the acquisition resolution, and fusing the near infrared spectrum data in the wavelength ranges of 400-1700nm according to the near infrared spectrum data in the acquisition bands of 400-1050nm and 900-1700nm to obtain the near infrared spectrum data of the samples in the wavelength ranges of 400-1700 nm;
2.3, determining reference values corresponding to the multidimensional indexes and each sample, wherein the reference values obtained by each sample comprise: meat color, water retention (cooking loss), protein content, fat content, moisture, total number of colonies, volatile basic nitrogen content, biogenic amine content;
2.4, for each sample, corresponding to near infrared spectrum data, calculating the reflectivity difference between adjacent peaks and troughs according to the wavelength from small to large, sequencing the reflectivity difference in sequence, and constructing a characteristic extraction spectrum taking the sequencing number as an abscissa and the reflectivity difference as an ordinate, as shown in fig. 4-5, taking the near infrared spectrum data of a 400-plus 1050nm acquisition band as an example, the adjacent peaks and troughs form a peak-trough pair, as shown in the following table 4:
table 4 feature extraction spectral correlation data
In table 4 above, the reflectance difference range is the range of reflectance difference corresponding to the total near infrared spectrum data under the corresponding peak-valley pair condition;
2.5, for any prediction index, establishing an index prediction model corresponding to the prediction index by applying a chemometrics method according to the feature extraction spectrum and a reference value corresponding to the prediction index;
2.6, embedding the constructed index prediction model into a model embedding module of the integrated meat quality detection system;
2.6; for the samples to be tested, 2 samples to be tested are respectively sample 1 (Ningxia Tan sheep, 8 months old, 45min after slaughter, meat at external spine part); sample 2 (Ningxia Tan sheep, 8 months old, 45min after slaughter, meat on the dragon site);
when the prediction indexes are flesh color, water system capacity and water content, extracting spectra by using the characteristics within the range of 400-;
when the prediction indexes are protein content and fat content, extracting a spectrum by using the characteristics within the range of 900-1700mm, and predicting the protein content and the fat content of the sample to be detected by combining the invoked prediction model corresponding to the protein content and the fat content;
when the prediction indexes are the total number of the bacterial colonies, the volatile basic nitrogen and the biological amine, extracting a spectrum by using the characteristics within the range of 400-1700nm, and predicting the total number of the bacterial colonies, the volatile basic nitrogen and the biological amine of the sample to be tested by combining the invoked prediction model corresponding to the total number of the bacterial colonies, the volatile basic nitrogen and the biological amine;
and constructing a mutton sample protein content lossless quantitative prediction model by using a chemometrics technology based on spectral characteristics.
The method comprises the following steps of (1) acquiring near infrared spectrum data of a mutton sample to be detected, and determining the protein content of the mutton sample by using a mutton sample protein content nondestructive quantitative prediction model, wherein the protein content is specifically shown in the following table 5:
TABLE 5 comparison of index prediction values and reference values for samples 1 and 2 using the method of example 1
< comparative example 1>
The meat multi-index rapid nondestructive testing method comprises the following steps:
step one, determining the minimum spectral resolution, which is the same as the step one of the embodiment 1;
step two, determining the number of detector elements of the spectrometer based on minimum resolution, and constructing an integrated meat multi-index rapid nondestructive testing system, wherein an index prediction model corresponding to flesh color, system water power, protein content, fat content, water, total bacterial count, volatile basic nitrogen and biogenic amine one by one is stored in a model embedding module in an industrial flat plate of the system, and for any prediction index, the method for constructing the index prediction model comprises the following steps:
2.1, selecting a sample as 2.1 of < example 1 >;
2.2, 2.2 as in < example 1 >;
2.3, 2.3 same as < example 1 >;
step 2.4 of < example 1> was not performed;
2.5, for any prediction index, establishing an index prediction model corresponding to the prediction index by applying a chemometrics method according to the near infrared spectrum data and the reference value corresponding to the prediction index;
2.6, embedding the constructed index prediction model into a model embedding module of the integrated meat quality detection system;
2.6; for the sample to be tested
When the prediction indexes are flesh color, water system capacity and water content, predicting the flesh color, water system capacity and water content of the sample to be tested by using near infrared spectrum data in the range of 400-1050mm and combining the invoked prediction model corresponding to the flesh color, the water system capacity and the water content;
when the prediction indexes are protein content and fat content, predicting the protein content and the fat content of the sample to be detected by using near infrared spectrum data in the range of 900-1700mm and combining with the invoked prediction model corresponding to the protein content and the fat content;
when the prediction indexes are the total number of the bacterial colonies, the volatile basic nitrogen and the biological amine, predicting the total number of the bacterial colonies, the volatile basic nitrogen and the biological amine of the sample to be detected by using near infrared spectrum data in the range of 400-1700nm and combining with the called prediction model corresponding to the total number of the bacterial colonies, the volatile basic nitrogen and the biological amine; specifically, the following table 6 shows:
TABLE 6 comparison of index prediction values and reference values for samples 1 and 2 using the method of comparative example 1
As can be seen from tables 5 and 6, the results of the sample index data predicted by the method of example 1 are equivalent to those predicted by the method of comparative example 1.
While embodiments of the invention have been described above, it is not intended to be limited to the details shown, described and illustrated herein, but is to be accorded the widest scope consistent with the principles and novel features herein disclosed, and to such extent that such modifications are readily available to those skilled in the art, and it is not intended to be limited to the details shown and described herein without departing from the general concept as defined by the appended claims and their equivalents.
Claims (10)
1. Many indexs of integration meat product quick nondestructive test system, its characterized in that includes:
the spectrometer is used for acquiring near infrared spectrum data of a sample to be detected;
an industrial flat panel, comprising:
the model embedding module is used for storing a plurality of index prediction models;
the model determining module is connected with the model embedding module and used for calling at least one index prediction model;
the index prediction module is connected with the spectrometer and the model determination module and used for receiving near infrared spectrum data of a sample to be tested and predicting the index data of the sample to be tested by combining the called prediction model;
the method comprises the following steps of determining the number of detector elements of the spectrometer according to the final minimum resolution, controlling the resolution to be the final minimum resolution in the acquisition process of near infrared spectrum data of a sample to be detected, and determining the final minimum resolution, wherein the determination comprises the following steps:
s1, acquiring spectral wavelength data of each sample in the sample set based on different resolutions;
s2, determining multiple prediction indexes according to multiple index prediction models stored by the model embedding module, establishing an index prediction model of any prediction index under the resolution by applying a chemometrics method according to spectral wavelength data under the resolution and a reference value corresponding to the prediction index under different resolutions, and determining a prediction correlation coefficient of the index prediction model;
s3, determining an inflection point of a prediction correlation coefficient of any prediction index under different resolutions for the resolution, and determining the resolution corresponding to the inflection point as the minimum resolution of the prediction index;
and S4, determining the maximum value of the minimum resolutions of all the prediction indexes as the final minimum resolution.
2. The integrated meat multi-index rapid nondestructive testing system of claim 1 wherein the multiple predictive indices are determined to be at least 3 of meat color, tenderness, water retention, pH, protein content, fat content, moisture, colony count, volatile basic nitrogen, UFA, total essential amino acids, biogenic amine according to multiple index predictive models;
wherein, when the prediction indexes are flesh color, water content and water content, the collection wave band is 400-1050 nm;
when the prediction indexes are tenderness, pH, protein content and fat content, the acquisition waveband is 900-1700 nm;
when the prediction indexes are total colony count, volatile basic nitrogen, UFA, total essential amino acid and biogenic amine, the collection wave bands are 400-1050nm and 900-1700 nm;
when the acquisition bands corresponding to the multiple prediction indexes simultaneously comprise 400-1050nm and 900-1700nm, the multiple prediction indexes are divided into an A group with the acquisition band comprising 400-1050nm and a B group with the acquisition band comprising 900-1700nm, and the final minimum resolution of all the prediction indexes corresponding to the A group in the acquisition band of 400-1050nm and the final minimum resolution of all the prediction indexes corresponding to the B group in the acquisition band of 900-1700nm are respectively determined.
3. The integrated meat multi-index rapid nondestructive testing system of claim 1 wherein the different resolutions of the predictors of group a ranges from 0.4 nm to 8nm and the different resolutions of the predictors of group B ranges from 15 nm to 45 nm.
4. The integrated meat multi-index rapid nondestructive testing system of claim 1 wherein each index prediction model is constructed by a method comprising the steps of:
determining a sample set, and determining a reference value of each prediction index corresponding to each sample in the sample set, wherein the attributes of the samples in the sample set are the same as the attributes of the samples to be detected;
determining an acquisition waveband according to a plurality of prediction indexes, and acquiring near infrared spectrum data of each sample in a sample set in a corresponding acquisition waveband, wherein when the prediction indexes are total bacterial count, volatile basic nitrogen, UFA, total essential amino acid or biogenic amine, spectrum fusion is carried out on the near infrared spectrum data of 400-plus 1050nm and 900-plus 1700nm to obtain the near infrared spectrum data of the coverage wavelength of 400-plus 1700 nm;
calculating the reflectivity difference between adjacent wave crests and wave troughs according to the wavelength of each piece of near infrared spectrum data from small to large, sequencing the data in sequence, and constructing a characteristic extraction spectrum taking the sequencing number as a horizontal coordinate and the reflectivity difference as a vertical coordinate;
for any prediction index, establishing an index prediction model corresponding to the prediction index by applying a chemometrics method according to the feature extraction spectrum and a reference value corresponding to the prediction index;
the industrial flat plate also comprises a spectrum processing module which is connected with the spectrometer and used for converting the acquired near infrared spectrum data into a characteristic extraction spectrum, and the index prediction module receives the characteristic extraction spectrum of the sample to be tested and predicts the index data of the sample to be tested by combining with the called prediction model.
5. The integrated meat multi-index rapid nondestructive testing system of claim 4, wherein the samples in the sample set and the samples to be tested are samples of the same genus, the varieties of the samples in the sample set comprise at least 3, the samples of each variety comprise at least 5 months of age, the samples of each month comprise at least 5 parts, and the samples of each part comprise at least 45min, 24h, 72h and 120h time points after slaughter.
6. The integrated meat multi-index rapid nondestructive testing system of claim 1 wherein the industrial flat panel further comprises:
the calibration reference module is used for storing an average spectrum X and a spectrum threshold value of a sample belonging to the same type as the sample to be detected, wherein the average spectrum X, an extremely-low reflectivity spectrum M and an extremely-high reflectivity spectrum M are calculated based on near infrared spectrum data of at least 1000 samples belonging to the same type as the sample to be detected, and the spectrum threshold value is determined according to the average spectrum X, the extremely-low reflectivity spectrum M and the extremely-high reflectivity spectrum M;
and the black-and-white correction module is connected with the correction reference module and the spectrometer and is used for acquiring the No. N +1 near infrared spectrum data after the spectrometer continuously acquires the near infrared spectrum data of the sample to be detected to a set number N, comparing the No. N +1 near infrared spectrum data with the average spectrum X, calculating the Mahalanobis distance, judging whether the Mahalanobis distance is within a spectrum threshold value, if so, determining that the No. N +1 near infrared spectrum data is normal, and if not, re-determining the No. N +1 near infrared spectrum data after correction.
7. The integrated meat multi-index rapid nondestructive testing system of claim 6 further comprising:
the top end of the detection camera bellows is provided with a detection window, the spectrometer is arranged in the detection camera bellows, and a condensing lens and a light source are sequentially arranged in the detection camera bellows from top to bottom below the detection window;
black and white correction subassembly, it is located including the rotation detect in the camera bellows and be located detection window below the disc, be used for the drive disc pivoted motor, be equipped with blank, blackboard on the disc, run through and be equipped with the through-hole, wherein, the motor with black and white correction module connects for control motor is rotatory the disc, so that blank, blackboard, through-hole selectivity and the coaxial setting of detection window.
8. The integrated meat multi-index rapid nondestructive testing system of claim 7 wherein the light source comprises 2 halogen aluminum reflector lamp cups symmetrically disposed 15-25mm below the testing window, the central axis of each reflector lamp cup being 55-65 ° to the central axis of the testing window.
9. The integrated meat multi-index rapid nondestructive testing system of claim 1 further comprising: and the result display module is connected with the index prediction module and used for receiving and displaying the prediction result.
10. The integrated meat multi-index rapid nondestructive testing system of claim 1 further comprising: and the data transmission module is connected with the index prediction module and used for receiving and transmitting the predicted index data to the user side.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202111663645.XA CN114002182B (en) | 2021-12-31 | 2021-12-31 | Integrated meat multi-index rapid nondestructive testing system |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202111663645.XA CN114002182B (en) | 2021-12-31 | 2021-12-31 | Integrated meat multi-index rapid nondestructive testing system |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN114002182A CN114002182A (en) | 2022-02-01 |
| CN114002182B true CN114002182B (en) | 2022-05-06 |
Family
ID=79932474
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202111663645.XA Active CN114002182B (en) | 2021-12-31 | 2021-12-31 | Integrated meat multi-index rapid nondestructive testing system |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN114002182B (en) |
Families Citing this family (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN114894730B (en) * | 2022-05-12 | 2023-06-16 | 广西大学 | Visible light-near infrared-based cane sugar nondestructive detection device and detection method |
Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| WO2009005828A1 (en) * | 2007-07-03 | 2009-01-08 | Tenera Technology, Llc | Imaging method for determining meat tenderness |
| CN102507459A (en) * | 2011-11-23 | 2012-06-20 | 中国农业大学 | Method and system for quick lossless evaluation on freshness of fresh beef |
| CN104330382A (en) * | 2014-11-17 | 2015-02-04 | 中国农业科学院北京畜牧兽医研究所 | Safety classification method of fresh beef |
| CN105548062A (en) * | 2015-12-21 | 2016-05-04 | 中国肉类食品综合研究中心 | A multi-index rapid nondestructive synchronous detection method for fresh beef |
-
2021
- 2021-12-31 CN CN202111663645.XA patent/CN114002182B/en active Active
Patent Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| WO2009005828A1 (en) * | 2007-07-03 | 2009-01-08 | Tenera Technology, Llc | Imaging method for determining meat tenderness |
| CN102507459A (en) * | 2011-11-23 | 2012-06-20 | 中国农业大学 | Method and system for quick lossless evaluation on freshness of fresh beef |
| CN104330382A (en) * | 2014-11-17 | 2015-02-04 | 中国农业科学院北京畜牧兽医研究所 | Safety classification method of fresh beef |
| CN105548062A (en) * | 2015-12-21 | 2016-05-04 | 中国肉类食品综合研究中心 | A multi-index rapid nondestructive synchronous detection method for fresh beef |
Non-Patent Citations (2)
| Title |
|---|
| 基于优选波长的多光谱检测系统快速检测猪肉中挥发性盐基氮的含量;魏文松等;《光学学报》;20171130;第37卷(第11期);第1130003-1至1130003-12页 * |
| 近红外光谱分析技术在肉类产品检测中的应用研究进展;吴习宇等;《食品工业科技》;20130806(第01期);第371-374、380页 * |
Also Published As
| Publication number | Publication date |
|---|---|
| CN114002182A (en) | 2022-02-01 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| Zhang et al. | Nondestructive measurement of soluble solids content in apple using near infrared hyperspectral imaging coupled with wavelength selection algorithm | |
| JP6791972B2 (en) | Methods and Devices for Detecting Interferents in Samples | |
| Barbin et al. | Non-destructive determination of chemical composition in intact and minced pork using near-infrared hyperspectral imaging | |
| ElMasry et al. | Chemical-free assessment and mapping of major constituents in beef using hyperspectral imaging | |
| Cocchi et al. | Classification of bread wheat flours in different quality categories by a wavelet-based feature selection/classification algorithm on NIR spectra | |
| CN109655414B (en) | Electronic equipment, information pushing method and related product | |
| Zamora-Rojas et al. | In-situ Iberian pig carcass classification using a micro-electro-mechanical system (MEMS)-based near infrared (NIR) spectrometer | |
| CN114002182B (en) | Integrated meat multi-index rapid nondestructive testing system | |
| Zhang et al. | A simple identification model for subtle bruises on the fresh jujube based on NIR spectroscopy | |
| CN110702637B (en) | Near-infrared online fusion rapid discrimination method for hot fresh mutton and cold fresh mutton | |
| Wang et al. | Smartphone imaging spectrometer for egg/meat freshness monitoring | |
| CN101907564A (en) | Rapeseed quality non-destructive testing method and device based on near infrared spectrum technology | |
| Hosseinpour et al. | An intelligent machine vision-based smartphone app for beef quality evaluation | |
| CN104049068A (en) | Nondestructive determining device and determining method for freshness of fresh livestock meat | |
| CN107907500A (en) | A kind of detection method in the mutton place of production | |
| Theron et al. | Toward the prediction of PSE-like muscle defect in hams: Using chemometrics for the spectral fingerprinting of plasma | |
| CN102181514A (en) | Method for rapidly and nondestructively detecting colony count of chilled meat | |
| CN110376154A (en) | Fruit online test method and system based on spectrum correction | |
| CN110243805A (en) | Fishbone detection method based on Raman high light spectrum image-forming technology | |
| CN106018292A (en) | Non-destructive testing device for protein conformation in egg white and method of non-destructive testing device | |
| CN112213281A (en) | Comprehensive evaluation method for rapidly determining freshness of freshwater fish based on transmission near infrared spectrum | |
| US11555811B1 (en) | Integrated rapid non-destructive detection system for multi-index of meat quality | |
| US20230089466A1 (en) | Establishment of Identification and Screening Method of Cows with A2 Beta-Casein Genotype of Producing A2 Milk and Applications Thereof | |
| CN107356536B (en) | Wine quality detection method, server and spectral measurement equipment | |
| CN109238973A (en) | A method of detection apple soluble solid content and water content |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant |