CN110263969B - Dynamic prediction system and prediction method for quality of apples with shelf life - Google Patents
Dynamic prediction system and prediction method for quality of apples with shelf life Download PDFInfo
- Publication number
- CN110263969B CN110263969B CN201910376395.8A CN201910376395A CN110263969B CN 110263969 B CN110263969 B CN 110263969B CN 201910376395 A CN201910376395 A CN 201910376395A CN 110263969 B CN110263969 B CN 110263969B
- Authority
- CN
- China
- Prior art keywords
- quality
- apple
- apples
- model
- shelf life
- 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
Images
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/21—Design or setup of recognition systems or techniques; Extraction of features in feature space; Blind source separation
- G06F18/214—Generating training patterns; Bootstrap methods, e.g. bagging or boosting
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/06—Physical realisation, i.e. hardware implementation of neural networks, neurons or parts of neurons
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06Q—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
- G06Q10/00—Administration; Management
- G06Q10/04—Forecasting or optimisation specially adapted for administrative or management purposes, e.g. linear programming or "cutting stock problem"
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06Q—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
- G06Q10/00—Administration; Management
- G06Q10/06—Resources, workflows, human or project management; Enterprise or organisation planning; Enterprise or organisation modelling
- G06Q10/063—Operations research, analysis or management
- G06Q10/0639—Performance analysis of employees; Performance analysis of enterprise or organisation operations
- G06Q10/06395—Quality analysis or management
-
- Y—GENERAL TAGGING OF NEW TECHNOLOGICAL DEVELOPMENTS; GENERAL TAGGING OF CROSS-SECTIONAL TECHNOLOGIES SPANNING OVER SEVERAL SECTIONS OF THE IPC; TECHNICAL SUBJECTS COVERED BY FORMER USPC CROSS-REFERENCE ART COLLECTIONS [XRACs] AND DIGESTS
- Y02—TECHNOLOGIES OR APPLICATIONS FOR MITIGATION OR ADAPTATION AGAINST CLIMATE CHANGE
- Y02P—CLIMATE CHANGE MITIGATION TECHNOLOGIES IN THE PRODUCTION OR PROCESSING OF GOODS
- Y02P90/00—Enabling technologies with a potential contribution to greenhouse gas [GHG] emissions mitigation
- Y02P90/30—Computing systems specially adapted for manufacturing
Abstract
The invention belongs to the technical field of apple quality prediction, and discloses a shelf life apple quality dynamic prediction system and a prediction method, wherein a single-index threshold apple quality model is established, and multi-element large sample data are classified; establishing a quality modeling method for the internal sugar degree value, the acidity value and the hardness value of the apples based on a multiple linear regression method; establishing a shelf life apple quality time sequence prediction model based on internal quality under different classifications of RBF neural networks; constructing a small sample model prediction large sample difference value time sequence quality model based on apple internal quality, detection time t and scanning time t+k according to the shelf life apple quality time sequence prediction model; and establishing a taste evaluation model according to the sugar degree, acidity and hardness values of the apples. The method can feed back the real taste of the apples, and the open interactive application program ensures the freshness of the apple quality evaluation model, thereby providing a good basis for healthy operation of the apple quality dynamic prediction system in shelf life.
Description
Technical Field
The invention belongs to the technical field of apple quality prediction, and particularly relates to a dynamic prediction system and a prediction method for shelf life apple quality.
Background
Currently, the closest prior art:
china has rich land resources, the planting area and the yield of fruits are in the front of the world, the planting area of apples is wide, the planting area reaches 2323.80 kilohectares by 2016 years, the yield of apples is huge and takes on a situation of rising year by year, and the yield of apples in 2017 reaches 4139.00 kilotons. The fresh apples in 2016 only have more than 133 ten thousand tons and account for 3.2 percent of the total output of apples in China, which indicates that most of the fresh apples are consumed by the domestic market, and the average occupation of the apples in 2016 reaches 31.7kg. In terms of nondestructive testing of apple quality, techniques for detecting apple quality using spectroscopy have been quite mature, particularly near infrared spectroscopy. In the prior art, many students also carry out physiological researches on quality information (such as weight loss rate, pulp hardness, skin color, sugar degree, acidity and the like of apples), and it is proved that the internal quality of apples is always changed in the process of storage in a refrigerator, especially apples with shelf life, external conditions are unsuitable for storage, and the internal quality of apples is relatively fast to change, so that even apples in the same lot of shelves are different due to different purchase time. Thus, the internal quality detected when apples are on the shelf is not the quality of apples for purchase or eating. Further, the inspection equipment is not inexpensive, so that consumers cannot perform real-time inspection of the internal quality of apples by purchasing the inspection equipment. Based on the above problems, if apples are stored for too long in shelf life or are not used for a long time after being purchased by consumers, the quality and the price of the apples are inconsistent, and the credit of merchants and the rights and interests of the consumers are affected.
Therefore, a shelf life apple quality time sequence dynamic prediction model is established, the quality of apples in the shelf life is predicted in real time, merchants and consumers are guided to purchase and eat apples, and the method plays a vital role in accelerating the healthy rapid development of apple industry in China. Since the 21 st century, portable apple quality inspection equipment has been rapidly developed, such as commercially available K-BA100R type from KUBOOTA corporation, supNIR-1000 type from light-focusing science and technology corporation, and NIRMagic1100 type portable near infrared spectroscopy instrument from Weichuang graph corporation, however, general quality inspection equipment is often based on a spectrometer, and has problems of high cost, high energy consumption, redundant spectral information, slow inspection speed, low automation degree and the like.
In the prior art, most of the near infrared detectors are manufacturer-oriented and not consumer-oriented, and if consumers purchase the near infrared detectors, not only the economic burden of the consumers is increased, but also the waste of resources is caused, so that the benign development of society is not facilitated.
In summary, the problems of the prior art are:
(1) In the prior art, quality detection equipment has the defects of overhigh cost, high energy consumption, redundant spectrum information, low detection speed and low automation degree, and causes poor dynamic prediction effect of the quality of apples in shelf life. And cannot provide quantitative basis and support for accurate prediction of apple quality.
(2) At present, no model method for predicting the quality of apples in shelf life is applied to actual life, and the best equipment for nondestructive detection of the quality of apples is a near infrared detection instrument. However, it is clearly not possible to push near infrared detectors to the public.
(3) There is no medium capable of recording apple quality information, and the theory is converted into practical results.
(4) The quality requirements of apple by Chinese consumers are different, and it is difficult to unify the apple quality requirements to the same standard.
Meaning of solving the technical problems:
the invention provides a dynamic apple quality prediction system for shelf life based on popularization of existing near infrared nondestructive testing equipment and two-dimensional code technology. When apples are about to be sold on a shelf, quality information of the apples is detected by using apple multi-quality nondestructive detection equipment, and a two-dimensional code is printed and output and is adhered to the apples. And the later-stage consumer scans the two-dimensional code stuck on the apple by using a WeChat applet embedded with a shelf-life apple quality prediction model, and obtains the quality of the apple under the condition of a series of model algorithms. The system well solves the problem that consumers are difficult to purchase high-quality apples, provides the freshest information for the consumers to purchase apples, and assists the consumers to purchase apples suitable for eating by themselves. Moreover, the method digitizes and visualizes the quality information of the apples, is beneficial to the healthy development of food safety, is easy to realize, can be quickly applied to practice, promotes the development of high-end fruit industry in China, and simultaneously provides a new idea for the prediction of the quality of the apples.
Disclosure of Invention
Aiming at the problems existing in the prior art, the invention provides a dynamic prediction system and a prediction method for the quality of apples with shelf life. The invention adopts independently developed apple multi-quality integrated nondestructive testing equipment, and the equipment utilizes quality indexes such as sugar degree, acidity and hardness of apples based on characteristic near infrared spectrum bands for nondestructive testing, and can print two-dimensional codes.
The invention discloses a shelf life apple quality dynamic prediction method, which comprises the following steps:
step one, establishing a single-index threshold apple quality model, and classifying multi-element large sample data;
establishing an apple internal sugar degree value, acidity value and hardness value quality modeling method based on a multiple linear regression method;
thirdly, establishing a shelf life apple quality time sequence prediction model based on internal quality under different classifications of RBF neural networks;
step four, constructing a small sample model prediction large sample difference value time sequence quality model based on apple internal quality, detection time t and scanning time t+k according to shelf life apple quality time sequence prediction models of internal quality under different classifications of RBF neural networks of shelf life;
establishing a taste evaluation model according to the sugar degree, acidity and hardness values of apples; and inquiring the quality information of the apples on the same day according to APP scanning, providing feedback comments after the apples are used on the APP after the consumer eats the apples, and correcting the taste model.
Further, in the first step, the discriminating method includes:
the sugar degree (S) classification S <8; s is more than or equal to 8 and less than or equal to 11; s is more than or equal to 11 and less than or equal to 14; s is more than or equal to 14 and less than 17; s is more than or equal to 17;
acidity (a) class a <0.1; a is more than or equal to 0.1 and less than or equal to 0.3; a is more than or equal to 0.3 and less than or equal to 0.5; a is more than or equal to 0.5 and less than or equal to 0.7; s is more than or equal to 0.7;
hardness (F) classification F <7; f is 7-11; f is more than or equal to 11 and less than or equal to 15; f is 15-19; s is more than or equal to 19;
wherein, the unit of the sugar degree is DEG Brix; acidity unit is mg/g; hardness in N/cm 2 。
Further, in the second step, the discrimination model is:
Y SSC ＝13.793-0.01719 V 420 -0.03557 V 480 -0.00028 V 550 -0.00048 V 580 -0.00032 V 640 -0.00691 V 690 -0.00147 V 700 -0.00225 V 940 -0.00655 V 980 -0.01053 V 1045 ；
Y acid ＝0.10805-0.00811 V 420 +0.06370 V 480 +0.00099 V 550 +0.00015 V 590 -0.00007 V 660 +0.00017 V 690 -0.00008 V 700 +0.00019 V 940 -0.00017 V 980 +0.00006 V 1045 ；
Y firmness ＝11.54901-0.01976 V 420 +0.21150 V 480 -0.00344 V 550 +0.00039 V 590 +0.00050 V 660 ；-0.00494 V 690 -0.00426 V 700 +0.00434 V 940 +0.00618 V 980 -0.00189 V 1045
wherein Y is SSC Is the sugar degree value, the unit is °Brix, Y acid Acidity in mg/g and Y acid Hardness value, in N/cm2; v (V) 420 、V 480 、V 550 、V 580 、V 640 、V 690 、V 700 、V 940 、V 980 、V 1045 The voltage values corresponding to the light intensity values of the lamp beads with the wave bands of 420nm, 480nm, 550nm, 590nm, 660nm, 690nm, 700nm, 940nm, 980nm and 1045nm are sequentially adopted.
In the third step, in the RBF neural network method prediction time sequence model, the input layer node of the RBF neural network only transmits the input signal to the hidden layer, the hidden layer node is composed of radial function functions like Gaussian function, the hidden node is used for adjusting the input signal locally, and the base function is that
x is an n-dimensional input vector; c i Is the center of the ith basis function, a vector having the same dimension as x; sigma (sigma) i Is the i-th perceived variable, determines the width of the base function center point; m is the number of sensing units; x-c i I represents x and c i A distance therebetween; hidden layer implementation x→r i (x) Is non-linear, the output layer implements R i (x)→y k The linear mapping is performed such that,
r is the number of output nodes, ω ik Is a weight;
the sugar degree, acidity and hardness time series model for predicting the internal quality based on the RBF neural network is respectively defined as: f (F) i (t)、Q i (t)、P i (t); the value range of i is 1,2,3,4 and 5.
In the fourth step, a difference time sequence model for predicting apple quality is established based on the shelf life apple internal quality prediction model, the sugar degree, acidity, hardness detection date t and prediction time t+k when the equipment detects apples, and the apple quality prediction model is expressed as follows:
sugar degree value: s is S s (t+k)＝Y ssc +F i (k)-F 1 (t 0 )；
Acidity value: the method comprises the steps of carrying out a first treatment on the surface of the A is that s (t+k)＝Y acid +Q i (k)-Q 1 (t 0 )
Hardness value: h s (t+k)＝Y fimness +P i (k)-P 1 (t 0 )；
t 0 The time of the first data of the experimental sample data is shown.
Further, constructing a small sample model prediction large sample difference value time sequence quality model based on apple internal quality, detection time t and scanning time t+k according to the shelf life apple quality time sequence prediction model;
predicted value of sugar degree at time t+k:
predicted value of acidity at time t+k:
predicted value of hardness at time t+k:
further, establishing a taste evaluation model, wherein the sugar acid ratio is less than 20 and is not good;
the sugar acid ratio is less than 30 and is more than 20;
the sugar acid ratio is less than 30 and less than 50, and is excellent;
the sugar acid ratio is more than 50, and the sweet taste is towards;
g0< hardness <5, is softer;
hardness is less than 5 and less than 12, and is moderate;
12> hardness, which is a relatively hard.
Another object of the invention is to provide a shelf life apple quality dynamic prediction control system for implementing the shelf life apple quality dynamic prediction method.
The invention further aims at providing an APP terminal for implementing the method for dynamically predicting the quality of the apples with the shelf life.
In summary, the invention has the advantages and positive effects that:
the shelf life apple quality dynamic prediction system and the prediction method provided by the invention can predict the quality of the sugar value, the acidity value and the hardness value in the apples in the shelf life, and realize the quick dynamic prediction of the quality index (sugar degree, acidity and hardness) of the apples in the shelf life by adopting a three-layer RBF neural network model; and evaluating the apples by using an apple quality taste evaluation model based on the predicted apple quality index (sugar degree, acidity and hardness) information.
Establishing a threshold apple quality model of an apple single index, and realizing classification of multiple big data, so that dynamic prediction of apple quality is more basis; and constructing an interactive feedback module based on the shelf life apple quality dynamic prediction model and the apple multi-quality integrated nondestructive testing equipment. When apples are about to be put on a shelf, quality information of the apples is detected by using apple multi-quality integrated nondestructive detection equipment, quality information sugar degree, acidity, hardness and detection time t of the apples are stored in a two-dimensional code and are stuck on the surface of the apples, and based on interactive feedback APP and a two-dimensional code technology, information such as sugar degree, acidity, hardness, detection time t and scanning time t+k of the apples in detection is obtained by scanning the two-dimensional code by using a mobile phone. Obtaining 5 parameter indexes based on a mobile phone, inputting the parameter indexes into apple quality information for predicting t+k time based on a three-layer RBF neural network model, and obtaining an apple quality evaluation result based on a quality evaluation model; after the consumer tastes the apples, the taste results are fed back to the system for correcting the apple quality evaluation model. Based on development of near infrared spectrum nondestructive testing technology and internal quality time sequence change rule of apples in shelf life, the invention provides an apple quality dynamic prediction model, an interactive apple quality prediction system is built, apple quality information is provided for consumers, meanwhile, a time sequence model for apple quality prediction is continuously perfected according to mouthfeel information fed back by the consumers, and a quantitative basis and a method support are provided for accurate prediction of apple quality.
On the basis, the theoretical predicted value at the time t+k and the time t and the basic value on the two-dimensional code are taken as inputs, and finally the predicted value of the apple quality index at the time t+k is obtained, and the prediction mode based on the difference theory is more scientific and realistic.
The apple quality dynamic prediction method model further comprises an established taste evaluation model of fresh apples, subjective factors of consumers are referred in the process, and finally, taste evaluation of single index is determined, so that a more pictographic evaluation is displayed for the consumers. The apple quality dynamic prediction method model further comprises an APP feedback system, so that the apple quality dynamic system based on the shelf life is designed to better perfect and correct an apple quality evaluation model and better serve consumers, an application program capable of feeding back the real taste of apples is designed, the freshness of the apple quality evaluation model is guaranteed by an open interactive application program, and a good foundation is provided for healthy operation of the apple quality dynamic prediction system of the shelf life.
Drawings
Fig. 1 is a flowchart of a method for dynamically predicting quality of apples with shelf life according to an embodiment of the invention.
Fig. 2 is a block diagram of an RBF neural network according to an embodiment of the present invention.
Fig. 3 is a schematic diagram of an APP feedback system provided by an embodiment of the present invention.
Fig. 4 is an overall block diagram of a shelf life apple quality dynamic prediction system provided by an embodiment of the present invention.
Detailed Description
The present invention will be described in further detail with reference to the following examples in order to make the objects, technical solutions and advantages of the present invention more apparent. It should be understood that the specific embodiments described herein are for purposes of illustration only and are not intended to limit the scope of the invention.
In the prior art, quality detection equipment has the defects of overhigh cost, high energy consumption, redundant spectrum information, low detection speed and low automation degree, and causes poor dynamic prediction effect of the quality of apples in shelf life. And cannot provide quantitative basis and support for accurate prediction of apple quality.
In order to solve the above problems, the present invention will be described in detail with reference to specific embodiments.
As shown in fig. 1, the method for dynamically predicting the quality of apples in shelf life provided by the embodiment of the invention is based on a visible/near infrared spectrum detection technology, and uses acquired diffuse reflection spectrum data of apples, and the measured hardness, total sugar content and titratable acidity content of apples. The method specifically comprises the following steps:
s101, establishing a single-index threshold apple quality model, and classifying multi-element large sample data.
S102, establishing a modeling method of apple internal quality (sugar degree value, acidity value and hardness value) based on a multiple linear regression method.
S103, establishing a shelf life apple quality time sequence prediction model based on internal quality under different classifications of RBF neural networks.
S104, constructing a small sample model prediction large sample difference value time sequence quality model based on apple internal quality, detection time t and scanning time t+k according to the shelf life apple quality time sequence prediction model.
S105, establishing a taste evaluation model according to the sugar degree, acidity and hardness values of apples; and according to the APP scanning and inquiring the quality information of the current day, providing feedback comments after the apple is used on the APP by consumers after eating the apple, and correcting the taste model.
In step S101, a single-index threshold apple quality model is adopted to classify the multiple large sample data, and the discrimination method is as follows:
the sugar degree (S) classification S <8; s is more than or equal to 8 and less than or equal to 11; s is more than or equal to 11 and less than or equal to 14; s is more than or equal to 14 and less than 17; s is more than or equal to 17;
acidity (a) class a <0.1; a is more than or equal to 0.1 and less than or equal to 0.3; a is more than or equal to 0.3 and less than or equal to 0.5; a is more than or equal to 0.5 and less than or equal to 0.7; s is more than or equal to 0.7;
hardness (F) classification F <7; f is 7-11; f is more than or equal to 11 and less than or equal to 15; f is 15-19; s is more than or equal to 19;
wherein, the unit of the sugar degree is DEG Brix; acidity unit is mg/g; hardness in N/cm 2 。
In step S102, a modeling method of apple internal quality (sugar degree value, acidity value, hardness value) based on a multiple linear regression method is adopted, and a discrimination model is as follows:
Y SSC ＝13.793-0.01719 V 420 -0.03557 V 480 -0.00028 V 550 -0.00048 V 580 -0.00032 V 640 -0.00691 V 690 -0.00147 V 700 -0.00225 V 940 -0.00655 V 980 -0.01053 V 1045 ；
Y acid ＝0.10805-0.00811 V 420 +0.06370 V 480 +0.00099 V 550 +0.00015 V 590 -0.00007 V 660 +0.00017 V 690 -0.00008 V 700 +0.00019 V 940 -0.00017 V 980 +0.00006 V 1045 ；
Y firmness ＝11.54901-0.01976 V 420 +0.21150 V 480 -0.00344 V 550 +0.00039 V 590 +0.00050 V 660 ；-0.00494 V 690 -0.00426 V 700 +0.00434 V 940 +0.00618 V 980 -0.00189 V 1045
wherein Y is SSC Is the sugar degree value, the unit is °Brix, Y acid Acidity in mg/g and Y acid Hardness value, in N/cm2; v (V) 420 、V 480 、V 550 、V 580 、V 640 、V 690 、V 700 、V 940 、V 980 、V 1045 The voltage values corresponding to the light intensity values of the lamp beads with the wave bands of 420nm, 480nm, 550nm, 590nm, 660nm, 690nm, 700nm, 940nm, 980nm and 1045nm are sequentially adopted.
In step S103, the sales period apple quality timing prediction model using internal quality under different classifications based on RBF neural network specifically includes:
the formula of the RBF neural network method prediction time sequence model is as follows:
RBF neural networks are theoretically a preferred network among feed forward networks. The RBF neural network is composed of three layers, and its structure is shown in fig. 2.
The input layer nodes pass only the input signal to the hidden layer, the hidden layer nodes are composed of radial acting functions like gaussian functions, while the output nodes are simple linear functions. The function of action (basis function) in the hidden node will locally affect the input signal, using the basis function:
where x is the n-dimensional input vector; c i Is the center of the ith basis function, a vector having the same dimension as x; sigma (sigma) i Is the i-th perceived variable that determines the width of the base function center point; m is the number of sensing units; x-c i I represents x and c i Distance between them. Hidden layer implementation x→r i (x) Is non-linear, the output layer implements R i (x)→y k I.e.:
where r is the number of output nodes, ω ik Is a weight.
The sugar degree, acidity and hardness time series model for predicting the internal quality based on the RBF neural network is respectively defined as: f (F) i (t)、Q i (t)、P i (t). Wherein the value range of i is 1,2,3,4,5.
In step S104, a difference time sequence model for predicting apple quality is established based on the shelf life apple internal quality prediction model and the sugar degree, acidity, hardness detection date t and prediction time t+k when the equipment detects apples, and the apple quality prediction model is expressed as follows:
sugar degree value: s is S s (t+k)＝Y ssc +F i (k)-F 1 (t 0 )；
Acidity value: the method comprises the steps of carrying out a first treatment on the surface of the A is that s (t+k)＝Y acid +Q i (k)-Q 1 (t 0 )
Hardness value: h s (t+k)＝Y fimness +P i (k)-P 1 (t 0 )；
t 0 The time of the first data of the experimental sample data is shown.
In step S104, classifying the multi-element large sample data by combining the single-index threshold apple quality model in step S101, modeling by combining the internal quality (sugar level value, acidity value, hardness value) of apples by the multi-element linear regression method in step S102, combining the shelf life apple quality time sequence prediction model based on the internal quality of different classifications of RBF neural networks, and combining the sugar level Y when apples are detected based on the shelf life apple internal quality prediction model and equipment ssc Acidity Y acid Hardness Y firmness And detecting the date t and the predicted time t+k, and establishing a difference time sequence model for predicting the apple quality, so as to finally obtain:
predicted value of sugar degree at time t+k:
predicted value of acidity at time t+k:
predicted value of hardness at time t+k:
in step S105, establishing a taste evaluation model based on the sugar degree, acidity and hardness values of the apples; 0<Ratio of sugar to acid<20, is bad, 20<Ratio of sugar to acid<30, is meta-acid, 30<Ratio of sugar to acid<50, is excellent in sugar-acid ratio>50 is sweet, and the sugar acid ratio unit is Brix.g/mg;0<Hardness of<5, is soft and 5<Hardness of<12, is moderate, 12>Hardness is relatively hard, and the hardness unit is N/cm 2 。
And establishing APP scanning and inquiring the quality information of the current day, providing feedback comments after eating the apples on the APP by consumers, and correcting the taste model by utilizing the feedback comments so as to enable the taste model to be continuously perfect.
The invention is further described below in connection with specific embodiments.
Example 1:
the method for dynamically predicting the quality of the apples with the shelf life, provided by the embodiment of the invention, comprises the following steps:
firstly, based on apple quality data obtained in recent years, threshold classification is achieved on apple quality by adopting a frequency distribution method, and an apple quality evaluation model is built based on an apple evaluation system in national standards.
Secondly, based on a visible/near infrared diffuse reflection spectrum test platform, apples with shelf lives are taken as experimental data samples, apple quality prediction models based on RBF neural networks are adopted for classifying apples with different qualities, and apple quality time sequence models with shelf lives are instantiated to realize dynamic prediction of apple quality.
Thirdly, based on the dynamic prediction model of the quality of the apples in the shelf life and the multi-quality integrated nondestructive testing equipment of the apples, an interactive feedback APP is constructed, and an apple quality evaluation model is continuously optimized according to user feedback information.
As a preferred embodiment of the present invention, in the first step, establishing the threshold classification model and the quality evaluation model includes:
based on a large amount of apple quality data in recent years, the quality indexes of apples in the Fuji series shelf life are equally divided into 5 grades by referring to a plurality of national standard files such as GB/T10651-2008 fresh apples and NY/T1075-2006 red Fuji apples and combining probability distribution of 3 quality indexes, and the quality indexes are as follows:
the sugar degree (S) classification S <8; s is more than or equal to 8 and less than or equal to 11; s is more than or equal to 11 and less than or equal to 14; s is more than or equal to 14 and less than 17; s is more than or equal to 17;
acidity (a) class a <0.1; a is more than or equal to 0.1 and less than or equal to 0.3; a is more than or equal to 0.3 and less than or equal to 0.5; a is more than or equal to 0.5 and less than or equal to 0.7; s is more than or equal to 0.7;
hardness (F) classification F <7; f is 7-11; f is more than or equal to 11 and less than or equal to 15; f is 15-19; s is more than or equal to 19;
wherein, the unit of the sugar degree is DEG Brix; acidity unit is mg/g; hardness in N/cm 2 。
According to GB/T10651-2008 fresh apples, NY/T1075-2006 red Fuji apples and other national standard documents and apple quality evaluation articles, establishing an apple-based sugar degree, acidity and hardness value, and further establishing a quality evaluation model; 0< sugar acid ratio <20, 20< sugar acid ratio <30, biatomic acid, 30< sugar acid ratio <50, excellent sugar acid ratio >50 is sweet, sugar acid ratio unit is °brix.g/mg; the hardness is 0< 5, is softer, the hardness is 5< 12, is moderate, the hardness is 12> and the hardness unit is N/cm <2 >.
In the second step, as a preferred embodiment of the present invention, the building of the visible/near infrared spectrum detection platform and the apple quality detection model includes:
according to the detection principle of apple quality diffuse reflection, a quality detection model based on the diffuse reflection principle is built. On the basis, the diffuse reflection spectrum data of apples are collected through an apple diffuse reflection experimental platform, and the data of sugar degree, acidity and hardness of apples are collected through a sugar degree meter, an acid-base titrator and a texture analyzer. And extracting characteristic wave bands by adopting an extraction algorithm of various characteristic wave bands, and finally, establishing a detection model of the apple multi-quality integrated nondestructive detection equipment by adopting a multi-element linear regression mode.
Y SSC ＝13.793-0.01719 V 420 -0.03557 V 480 -0.00028 V 550 -0.00048 V 580 -0.00032 V 640 -0.00691 V 690 -0.00147 V 700 -0.00225 V 940 -0.00655 V 980 -0.01053 V 1045
Y acid ＝0.10805-0.00811 V 420 +0.06370 V 480 +0.00099 V 550 +0.00015 V 590 -0.00007 V 660 +0.00017 V 690 -0.00008 V 700 +0.00019 V 940 -0.00017 V 980 +0.00006 V 1045
Y firmness ＝11.54901-0.01976 V 420 +0.21150 V 480 -0.00344 V 550 +0.00039 V 590 +0.00050 V 660 -0.00494 V 690 -0.00426 V 700 +0.00434 V 940 +0.00618 V 980 -0.00189 V 1045
Wherein Y is SSC Is the sugar degree value, the unit is °Brix, Y acid Acidity in mg/g and Y acid Hardness value in N/cm 2 ；V 420 、V 480 、V 550 、V 580 、V 640 、V 690 、V 700 、V 940 、V 980 、V 1045 The light beads with wave bands of 420nm, 480nm, 550nm, 590nm, 660nm, 690nm, 700nm, 940nm, 980nm and 1045nm are arranged in sequenceA voltage value corresponding to the light intensity value of (a).
As a preferred embodiment of the present invention, in the third step, the RBF-based dynamic prediction model and the difference model include:
in the establishment process of the RBF-based dynamic prediction model, apples with sugar degree smaller than 8 Brix are explained. 30 apples to be put on a shelf with apple sugar degree less than 8 Brix are selected by a nondestructive testing method, and 30 raw data of each measurement are recorded as x for discussion convenience i I=1, 2, 3..30, and average value of sugar data on the same day was recorded
j=1, 2,3, …,25. Let the first day be t 1 The next day is t 2 Until the end of the experiment. The experiment is used for processing data of 30 samples in continuous 25 days, based on the basic principle of the RBF neural network, the number of input nodes of the RBF neural network is designed to be 1, the number of output nodes of the RBF neural network is designed to be 1, and the number of neurons of an hidden layer can be determined in a self-adaptive manner in the network training process. Determining an input vector t= [ T ] 1 ,t 2 ,…t 25 ]The output vector is determined as +.>
The function (basis function) in the hidden node adopts R i (x)＝exp[||x-c i || 2 /2σ i 2 ]And after training until the network convergence is stable, storing the network. Apple quality prediction model based on RBF neural network, the dynamic apple quality prediction model of shelf life also comprises a difference model of apple quality prediction, and in order to enable the prediction model to be more accurate and better adapt to various environments, a difference model S of sugar degree quality indexes is designed s (t+k)＝Y ssc +F 1 (k)-F 1 (t 0 ). Based on the theoretical predicted value of the moment t+k and the moment t and the basic value Y on the two-dimensional code ssc And finally obtaining a predicted value of the apple sugar degree quality index at the time t+k for input, so that a prediction mode based on a difference value theory is more scientific and realistic. />
As shown in fig. 3, as a preferred embodiment of the present invention, the present invention provides an APP feedback system, the principle comprising:
in order to better perfect and correct an apple quality evaluation model and better serve consumers, an application program capable of feeding back the real taste of apples is provided based on a shelf life apple quality dynamic prediction model, and an open interactive application program ensures the instantaneity of the apple quality evaluation model, so that a good foundation is provided for healthy operation of a shelf life apple quality dynamic prediction system. When the apples are about to be put on the shelf, the quality index of the apples is detected by using the apple multi-quality integrated nondestructive detection equipment in the second step, and the two-dimensional code is printed and output and is stuck on the apples. Apple based on sticking two-dimensional code, consumer scans two-dimensional code by using mobile phone to obtain detection time t and sugar degree Y of apple ssc And scanning time t+k, and then bringing the time difference k into the RBF neural network and the difference model to predict the current sugar degree value of the apple. And respectively predicting the other two indexes of the apples based on the apple quality index sugar degree prediction method flow, and obtaining the acidity and hardness of the apples. Further, the apples are reasonably evaluated through the quality evaluation model, and after the consumers taste the apples, the taste information of the apples can be fed back. Based on the obtained apple taste evaluation information, the apple quality model can be perfected and corrected timely.
Example 2
According to the method for dynamically predicting the quality of the apples in the shelf life, provided by the embodiment of the invention, based on a visible/near infrared spectrum detection technology and an apple internal quality nondestructive detection model, when the apples are on the shelf, the quality of the apples is detected by using portable apple multi-quality integrated nondestructive detection equipment, and a two-dimensional code with a detection result (sugar degree, acidity and hardness) and a detection time t is stuck on the apples. Based on the two-dimensional code stored with the data of the sugar degree, the acidity, the hardness and the detection time t of the apples, scanning the two-dimensional code by using a mobile phone of a consumer to obtain the sugar degree, the acidity, the hardness, the detection time and the scanning time t+k of the two-dimensional code of the apples; and 5 obtained data are used as input of an apple quality prediction model, a combined prediction method based on a three-layer RBF neural network is adopted to realize dynamic prediction of apple quality indexes at the time t+k, and an overall block diagram of the dynamic prediction system of the apple quality in the shelf life of a specific embodiment is shown in fig. 4.
The apple quality dynamic prediction method model provided by the embodiment of the invention further comprises the following steps: the single-index threshold apple quality model is used for classifying the apple quality indexes, and the numerical range of the apple quality indexes is divided in detail aiming at the condition that the apple quality indexes are different in shelf life variation range, so that the apple quality indexes are refined, and a practical foundation is laid for realizing accurate prediction. Based on a large amount of apple quality index data obtained in recent years, each apple quality index is divided into 5 threshold ranges by adopting a frequency distribution method, and a single-index-based threshold apple quality model is used as input by taking the obtained apple sugar degree, acidity and hardness indexes as inputs, so that the selection of an apple quality prediction model is realized.
The apple quality dynamic prediction method model provided by the embodiment of the invention further comprises the following steps: the RBF neural network dynamic prediction time sequence model specifically comprises a time sequence model based on RBF neural network combination prediction. Apple sugar degree Y obtained by scanning two-dimensional code by using mobile phone of consumer ssc Acidity Y acid Hardness Y firmness And acquiring theoretical data results of the RBF neural network-based time sequence model at the time t+k of the apple quality index of shelf life by taking the detection time t and the scanning time t+k data of the two-dimensional code as inputs. At the moment, the input node of the RBF neural network is set to be 1, the output node is set to be 1, the number of neurons of an hidden layer can be adaptively determined in the network training process, and an action function (basic function) in the hidden node adopts R i (x)＝exp[||x-c i || 2 /2σ i 2 ]And further obtaining a theoretical predicted value of the apple quality index at the time t+k.
The apple quality dynamic prediction method model provided by the embodiment of the invention further comprises the following steps: the difference model for apple quality prediction is designed based on a difference theory and an actual application scene: s is S s (t+k)＝Y sse +F i (k)-F 1 (t 0 )、A s (t+k)＝Y acid +Q i (k)-Q 1 (t 0 ) And H s (t+k)＝Y fimness +P i (k)-P 1 (t 0 ). On the basis, the theoretical predicted value at the time t+k and the time t and the basic value on the two-dimensional code are taken as inputs, and finally the predicted value of the apple quality index at the time t+k is obtained, and the prediction mode based on the difference theory is more scientific and realistic.
The apple quality dynamic prediction method model provided by the embodiment of the invention further comprises the following steps: and establishing a taste evaluation model of the fresh apples, simultaneously referencing subjective factors of consumers in the process, and finally determining taste evaluation of a single index, so as to show a pictographic evaluation for the consumers.
The apple quality dynamic prediction method model provided by the embodiment of the invention further comprises the following steps: the APP feedback system is used for better perfecting and correcting an evaluation model of apple quality and better serving consumers, so that an application program capable of feeding back real taste of apples is developed based on the apple quality dynamic system of shelf life, the freshness of the apple quality evaluation model is ensured by an open interactive application program, and a good foundation is provided for healthy operation of the apple quality dynamic prediction system of shelf life.
The foregoing description of the preferred embodiments of the invention is not intended to be limiting, but rather is intended to cover all modifications, equivalents, and alternatives falling within the spirit and principles of the invention.
Claims (7)
1. The method for dynamically predicting the quality of the apples in the shelf life is characterized by comprising the following steps of:
step one, establishing a single-index threshold apple quality model, and classifying multi-element large sample data;
establishing an apple internal sugar degree value, acidity value and hardness value quality modeling method based on a multiple linear regression method;
thirdly, establishing a shelf life apple quality time sequence prediction model based on internal quality under different classifications of RBF neural networks;
step four, constructing a small sample model prediction large sample difference value time sequence quality model based on apple internal quality, detection time t and scanning time t+k according to the shelf life apple quality time sequence prediction model;
establishing a taste evaluation model according to the sugar degree, acidity and hardness values of apples; according to the APP scanning and inquiring the quality information of the current day, providing feedback comments after the apple is used on the APP by consumers after eating the apple, and correcting the taste model;
in the first step, the distinguishing method comprises the following steps:
the sugar degree S classification S <8; s is more than or equal to 8 and less than or equal to 11; s is more than or equal to 11 and less than or equal to 14; s is more than or equal to 14 and less than 17; s is more than or equal to 17;
acidity class a <0.1; a is more than or equal to 0.1 and less than or equal to 0.3; a is more than or equal to 0.3 and less than or equal to 0.5; a is more than or equal to 0.5 and less than or equal to 0.7; s is more than or equal to 0.7;
hardness class F <7; f is 7-11; f is more than or equal to 11 and less than or equal to 15; f is 15-19; s is more than or equal to 19;
wherein, the unit of the sugar degree is DEG Brix; acidity unit is mg/g; hardness in N/cm 2 ；
In the second step, the discrimination model is:
Y SSC ＝13.793-0.01719V 420 -0.03557V 480 -0.00028V 550 -0.00048V 580 -0.00032V 640 -0.00691V 690 -0.00147V 700 -0.00225V 940 -0.00655V 980 -0.01053V 1045 ；
Y acid ＝0.10805-0.00811V 420 +0.06370V 480 +0.00099V 550 +0.00015V 590 -0.00007V 660 +0.00017V 690 -0.00008V 700 +0.00019V 940 -0.00017V 980 +0.00006V 1045 ；
Y firmness ＝11.54901-0.01976V 420 +0.21150V 480 -0.00344V 550 +0.00039V 590 +0.00050V 660 ；-0.00494V 690 -0.00426V 700 +0.00434V 940 +0.00618V 980 -0.00189V 1045
wherein Y is SSC Is the sugar degree value, the unit is °Brix, Y acid Acidity in mg/g and Y acid Hardness value in N/cm 2 ；V 420 、V 480 、V 550 、V 580 、V 640 、V 690 、V 700 、V 940 、V 980 、V 1045 The voltage values corresponding to the light intensity values of the lamp beads with the wave bands of 420nm, 480nm, 550nm, 590nm, 660nm, 690nm, 700nm, 940nm, 980nm and 1045nm are sequentially adopted.
2. The method for dynamically predicting the quality of apples on shelf life according to claim 1, wherein in the third step, in the prediction time series model of the RBF neural network method, input layer nodes of the RBF neural network only transmit input signals to hidden layers, the hidden layer nodes are composed of radial action functions like gaussian functions, and the input signals are locally regulated by the action basis functions in the hidden nodes, wherein the basis functions are as follows;
x is an n-dimensional input vector; c i Is the center of the ith basis function, a vector having the same dimension as x; sigma (sigma) i Is the i-th perceived variable, determines the width of the base function center point; m is the number of sensing units; x-c i I represents x and c i A distance therebetween; hidden layer implementation x→r i (x) Is non-linear, the output layer implements R i (x)→y k Linear mapping;
r is the number of output nodes, ω ik Is a weight;
the sugar degree, acidity and hardness time sequence of the internal quality is predicted based on RBF neural networkThe models are defined as: f (F) i (t)、Q i (t)、P i (t); the value range of i is 1,2,3,4 and 5.
3. The method for dynamically predicting the quality of apples in a shelf life according to claim 1, wherein in the fourth step, a difference time sequence model for predicting the quality of apples is established based on a shelf life apple internal quality prediction model and sugar degree, acidity, hardness detection date t and prediction time t+k when equipment detects apples, and the apple quality prediction model comprises:
sugar degree value: s is S s (t+k)＝Y ssc +F i (k)-F 1 (t 0 )；
Acidity value: the method comprises the steps of carrying out a first treatment on the surface of the A is that s (t+k)＝Y acid +Q i (k)-Q 1 (t 0 )；
Hardness value: h s (t+k)＝Y fimness +P i (k)-P 1 (t 0 )；
t 0 The time of the first data of the experimental sample data is shown.
4. The method for dynamically predicting the quality of apples in the shelf life according to claim 1, wherein the method is characterized in that in the step four, a small sample model prediction large sample difference value time sequence quality model based on the internal quality of apples, the detection time t and the scanning time t+k is constructed according to a shelf life apple quality time sequence prediction model;
predicted value of sugar degree at time t+k:
predicted value of acidity at time t+k:
predicted value of hardness at time t+k:
5. the method for dynamically predicting the quality of apples with shelf life according to claim 1, wherein in the fifth step, a taste evaluation model is established, wherein the sugar acid ratio of 0< 20 is poor;
20< sugar acid ratio <30, is meta-acid;
30< sugar acid ratio <50, is excellent;
the sugar-acid ratio is more than 50, and the sweet taste is towards;
g0< hardness <5, soft;
5< hardness <12, moderate;
12> hardness, which is a relatively hard.
6. A shelf life apple quality dynamic prediction control system implementing the shelf life apple quality dynamic prediction method of claim 1.
7. An APP terminal implementing the shelf life apple quality dynamic prediction method of claim 1.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201910376395.8A CN110263969B (en) | 2019-05-07 | 2019-05-07 | Dynamic prediction system and prediction method for quality of apples with shelf life |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201910376395.8A CN110263969B (en) | 2019-05-07 | 2019-05-07 | Dynamic prediction system and prediction method for quality of apples with shelf life |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN110263969A CN110263969A (en) | 2019-09-20 |
| CN110263969B true CN110263969B (en) | 2023-06-02 |
Family
ID=67914338
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201910376395.8A Active CN110263969B (en) | 2019-05-07 | 2019-05-07 | Dynamic prediction system and prediction method for quality of apples with shelf life |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN110263969B (en) |
Families Citing this family (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN110969090A (en) * | 2019-11-04 | 2020-04-07 | 口碑(上海)信息技术有限公司 | Fruit quality identification method and device based on deep neural network |
| CN110882948B (en) * | 2019-11-29 | 2021-06-25 | 嘉应学院 | Grapefruit quality classification method and device |
| CN113128779A (en) * | 2021-04-29 | 2021-07-16 | 华东交通大学 | Decision support system for optimal harvest period of oranges |
Citations (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| JP2004294108A (en) * | 2003-03-25 | 2004-10-21 | Mitsui Mining & Smelting Co Ltd | Apparatus for measuring sugar concentration |
| CN102735621A (en) * | 2011-04-13 | 2012-10-17 | 广东工业大学 | Fruit maturity detection method based on color sensor |
| CN103530808A (en) * | 2013-10-25 | 2014-01-22 | 西安凌科信息技术有限公司 | Whole-process tracing system for quality safety of fruit logistics |
| CN104655810A (en) * | 2015-02-25 | 2015-05-27 | 山东商业职业技术学院 | Method for predicting quality change of cold storage winter jujubes |
| CN108304970A (en) * | 2018-02-05 | 2018-07-20 | 西北农林科技大学 | The method for quick predicting and system of Apple, air conditioned storage monitoring system |
Family Cites Families (8)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| MXPA02009027A (en) * | 2000-03-13 | 2004-08-19 | Richard M Ozanich | Apparatus and method for measuring and correlating characteristics of fruit with visible near infra red spectrum. |
| KR101574895B1 (en) * | 2014-06-26 | 2015-12-04 | 제주대학교 산학협력단 | Method for predicting sugar contents and acidity of citrus using ft-ir fingerprinting combined by multivariate analysis |
| CN105184404B (en) * | 2015-08-31 | 2018-12-18 | 中国科学院广州能源研究所 | Output power classification forecasting system suitable for photovoltaic system Life cycle |
| CN106769970A (en) * | 2016-12-07 | 2017-05-31 | 甘肃萃英大农科技有限公司 | A kind of method of the kind, pol and acidity of Fast nondestructive evaluation apple |
| CN106841103A (en) * | 2017-03-01 | 2017-06-13 | 沈阳农业大学 | Near infrared spectrum detects fruit internal quality method and dedicated test system |
| CN107064047A (en) * | 2017-03-02 | 2017-08-18 | 兰州大学 | A kind of Fuji apple quality damage-free detection method based near infrared spectrum |
| CN109115708A (en) * | 2018-09-29 | 2019-01-01 | 西北农林科技大学 | A kind of more quality integration nondestructive detection systems of apple internal and method |
| CN109325638A (en) * | 2018-11-09 | 2019-02-12 | 电子科技大学 | A kind of SDN method for predicting based on RBF neural |
-
2019
- 2019-05-07 CN CN201910376395.8A patent/CN110263969B/en active Active
Patent Citations (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| JP2004294108A (en) * | 2003-03-25 | 2004-10-21 | Mitsui Mining & Smelting Co Ltd | Apparatus for measuring sugar concentration |
| CN102735621A (en) * | 2011-04-13 | 2012-10-17 | 广东工业大学 | Fruit maturity detection method based on color sensor |
| CN103530808A (en) * | 2013-10-25 | 2014-01-22 | 西安凌科信息技术有限公司 | Whole-process tracing system for quality safety of fruit logistics |
| CN104655810A (en) * | 2015-02-25 | 2015-05-27 | 山东商业职业技术学院 | Method for predicting quality change of cold storage winter jujubes |
| CN108304970A (en) * | 2018-02-05 | 2018-07-20 | 西北农林科技大学 | The method for quick predicting and system of Apple, air conditioned storage monitoring system |
Also Published As
| Publication number | Publication date |
|---|---|
| CN110263969A (en) | 2019-09-20 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN110263969B (en) | Dynamic prediction system and prediction method for quality of apples with shelf life | |
| Cozzolino | Recent trends on the use of infrared spectroscopy to trace and authenticate natural and agricultural food products | |
| CN107609111B (en) | Retrieval method for variety identification, quality grading and maturity judgment of loquat fruits | |
| CN108960315B (en) | Intelligent evaluation system and method for quality of cooked meat product | |
| Moldes et al. | A critical review on the applications of artificial neural networks in winemaking technology | |
| CN104406916A (en) | Method and apparatus for detecting food | |
| Wang et al. | Discrimination of different white chrysanthemum by electronic tongue | |
| Lapchareonsuk et al. | Sensory quality evaluation of rice using visible and shortwave near-infrared spectroscopy | |
| CN106841545A (en) | Based on the tealeaves organoleptic quality evaluations method for improving fuzzy comprehensive evaluation method | |
| Sarkar et al. | Supervised learning aided multiple feature analysis for freshness class detection of Indian gooseberry (Phyllanthus emblica) | |
| Sharma et al. | Rapid ripening stage classification and dry matter prediction of durian pulp using a pushbroom near infrared hyperspectral imaging system | |
| CN104949927A (en) | Hyperspectral detection method of dried scallop moisture content | |
| Tian et al. | Detection of soluble solid content in apples based on hyperspectral technology combined with deep learning algorithm | |
| Wang et al. | Flexible Vis/NIR wireless sensing system for banana monitoring | |
| Wang et al. | Identification of tea quality at different picking periods: A hyperspectral system coupled with a multibranch kernel attention network | |
| Yu et al. | Nondestructive testing of external defects on Nanguo pear | |
| Ren et al. | Intelligent Evaluation of the Appearance Modality of Black Tea Based on Machine Vision and Hyperspectral Imaging | |
| Hu et al. | Nondestructive discrimination of internal defects in jujube (Huizao) of Xinjiang based on visible and near-infrared spectroscopy | |
| Hong et al. | College students’ preferences and willingness to pay for fresh apple varieties in Peru | |
| Qi et al. | Sensory characteristics of “Shine Muscat” grapes based on consumer reviews and human and intelligent sensory evaluation | |
| Gomez‐Carracedo et al. | Evaluation of the pure apple juice content in commercial apple beverages using FTMIR‐ATR and potential curves | |
| Graham-Acquaah et al. | Differences in physicochemical properties of commercial rice from urban markets in West Africa | |
| Baart et al. | Lifestyle behaviours, ethnicity and menstruation have little added value in prediction models for low haemoglobin deferral in whole blood donors | |
| Huang et al. | Rapid screening of sensory attributes of mackerel using big data mining techniques and rapid sensory evaluation methods | |
| Li et al. | Rapid detecting SSC and TAC of peaches based on NIR spectroscopy |
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 |