Note : Les descriptions sont présentées dans la langue officielle dans laquelle elles ont été soumises.
CA 03126171 2021-07-08
LOW-FIELD NUCLEAR MAGNETIC RESONANCE-BASED DEVICE AND
METHOD FOR INTELLIGENT DETECTION OF MICROWAVE-DRIED SPICY
VEGETABLE FLAVOR
TECHNICAL FIELD
The present invention belongs to the technical field of the intelligent
identification of spicy
vegetable drying quality, and relates to a low-field nuclear magnetic
resonance-based device
and method for intelligent detection of a microwave-dried spicy vegetable
flavor.
BACKGROUND
Spicy vegetables can give off an aromatic smell or have a pungent taste, and
are generally
eaten raw, or used as soups, seasonings or dish decorations. Such spicy
vegetables can increase
appetite and also have many medicinal values. Commonly cultivated varieties of
spicy
vegetables in China are coriander, green onions, garlic, ginger, leaf fennel,
etc. Spicy vegetables
contain abundant nutrients. For example, the fresh ginger is rich in nutrients
and contain ginger
oil, gingerol, fatty oil, resin, starch, pentosan, cellulose, protein,
pigment, wax and trace mineral
elements, etc. There are more than 100 species complex chemical have been
discovered, which
can be divided into three categories: gingerols, terpenoid volatile oils and
diphenylheptanes.
Among them, the aroma and flavor of ginger are related to the volatile oil of
ginger essential
oil contained therein, and the spicy flavor of ginger mainly depends on the
gingerols in the non-
volatile oil of ginger oleoresin. In addition, it also contains a variety of
trace elements such as
multiple amino acids, vitamins, copper, iron, manganese, zinc, chromium,
nickel, cobalt, and
multiple functional ingredients. It has the functions of expelling wind and
cold, anti-oxidation,
anti-tumor, lowering cholesterol, lowering blood sugar, detoxification and
sterilization, etc., so
that it has attracted widespread attention from consumers and scholars at home
and abroad.
Fresh garlic contains carbohydrates, proteins, dietary fibers, fats, vitamins,
minerals, and
abundant amino acids. Among them, the amino acid in vegetables is one of the
important
nutrients in vegetables, and its composition and content directly affect the
nutritional value of
vegetables such as white garlic, and are closely related to human taste.
Dehydration drying is
an important technology for long-term storage of agricultural products.
Usually, the drying
method of spicy vegetables is mainly hot air drying, which is simple in
operation and low in
1
Date Recue/Date Received 2021-07-08
CA 03126171 2021-07-08
investment, but which also has disadvantages such as long drying time, low
efficiency and poor
quality. Microwave vacuum drying is an energy-saving, environment-friendly,
modem high-
tech drying technology. The water content of spicy vegetables is high. Due to
its unique
advantages, the new drying technology of microwave vacuum drying has attracted
widespread
attention from scholars at home and abroad in the field of fruit and vegetable
dehydration in
recent years. Microwave vacuum drying can better retain the original color,
fragrance and taste,
vitamins and other heat-sensitive nutrients or biologically active ingredients
of the material
being dried. Spicy vegetables are processed into powder, which can be directly
consumed as
seasoning or solid drink, and can also be used as raw materials of medicinal
materials and health
foods, and as raw materials for bread, candy, biscuits and other foods.
Therefore, the processing
of spicy vegetable powder is of important application value and broad
development prospects.
During the dehydration process, the most important thing for spicy vegetables
is the change
of a flavor. There are many kinds of flavor substances in spicy vegetables,
mainly including
volatile flavor substances such as alcohols, aldehydes, esters, acids,
alkanes, acids, and sulfur-
containing compounds, and non-volatile flavor substances such as soluble
sugars, organic acids,
free amino acids, etc. The former determines the characteristic taste of food
and provides
precursors for the synthesis of the latter, while the latter is
macroscopically manifested as the
food smell. These substances are very small in content and different in odor,
and together form
the flavor system of food. The contribution of the unique aroma of spicy
vegetables to the flavor
is related to its content and threshold. The unique flavor substances can not
only work up
people's appetite, but also promote the secretion of digestive juice, so that
the human body can
digest and absorb nutrients quickly. An electronic nose is used to evaluate
the flavor quality of
spicy vegetables during the drying process, which has the advantages of being
less affected by
subjective factors, high reliability and high repeatability. The electronic
nose can realize the
qualitative and quantitative analysis of different sensitive types of
substances by means of a
metal sensor, and it has been widely used in the fields of food quality
detection, flavor
evaluation, authenticity discrimination and characteristic aroma recognition.
Low-field nuclear magnetic resonance (LF-NMR) uses the spin relaxation
characteristics
of hydrogen nuclei in a magnetic field to explain the changes in distribution
and the migration
of water in the sample from a microscopic point of view through the change of
relaxation time,
and it has the advantages of being fast, accurate, non-destructive, non-
invasive, etc., and has
been widely used in the field of food science in recent years.
Sun Qun et al. (Patent Application Number: CN201711307888.3) disclosed a low-
field
nuclear magnetic resonance non-destructive testing line applicable to dried
shell fruits. Low-
2
Date Recue/Date Received 2021-07-08
CA 03126171 2021-07-08
field nuclear magnetic resonance equipment is integrated with a conveying
device of a dried
shell fruit. Principal component analysis is performed on the existing sample
data, and good
and bad seed area division and boundary equation calculation are performed to
obtain a suitable
mathematical model. In the detection process, low-field NMR technology is used
to detect the
transverse magnetization signal quantity of a dried shell fruit, and then
through comparison
with the constructed mathematical model, the quality of good seeds, mildewed
seeds and moth-
eaten seeds of the dried shell fruit is obtained quickly, and the accuracy
rate can reach 85% or
above.
Guo Tao et al. (Patent Application Number: CN201710984507.9) disclosed a fast
identification method for grape seed oil adulteration based on low-field
nuclear magnetism,
which is applicable to the identification of grape seed oil and adulterated
grape seed oil. A low-
field nuclear magnetic resonance analyzer is used as a main measuring tool,
the difference
between relaxation map data of grape seed oil and relaxation map data of
adulterated grape seed
oil is taken as a main identification basis, a nuclear magnetic resonance
signal is taken as a main
research object, and a mathematical model analysis method is used to realize
quick and accurate
identification of grape seed oil and adulterated grape seed oil.
Li Dajing et al. (Patent Application Number: CN201510967968.6) disclosed a
method for
characterizing a drying end point of far infrared dried Agaricus bisporus
based on water
distribution. In the method, fresh agaricus bisporus slices are selected as
raw materials, far-
infrared drying is performed, and the agaricus bisporus slices during the
drying process are
scanned using low-field nuclear magnetic resonance technology to obtain an
inversion map of
water distribution, and the drying end point is determined according to the
size of the free water
relaxation area.
Tan Mingqian et al. (Patent Application Number: CN201610279790.0) disclosed a
method
for measuring oil content and water content of soybeans using low-field
nuclear magnetic
resonance technology. Soybean samples are measured by using a CPMG sequence of
low-field
nuclear magnetic resonance technology, to obtain relaxation spectrum data of
each soybean
sample, the actual oil content and water content of each soybean sample is
corresponding to the
relaxation spectrum data, and fitting is performed with a chemometric method
to obtain a
prediction model of oil content and water content of the soybeans.
Wang Xin et al. (Patent Application No.: CN201210435185.X) disclosed a low-
field
nuclear magnetic resonance detection method for a frying use limit of soybean
oil. A low-field
nuclear magnetic resonance analyzer is used as a main measuring tool, a
mathematical model
3
Date Recue/Date Received 2021-07-08
CA 03126171 2021-07-08
between multi-component relaxation spectrum data and total polar compound
(TPC) data of
soybean oil is established as a basis, a nuclear magnetic resonance signal
during the frying
process of soybean oil is used as a main observation object, and the frying
use limit of soybean
oil is judged by analyzing multi-component transverse relaxation spectrum data
of soybean oil
in the frying process.
Tan Mingqian et al. (Patent Application Number: CN201610285372.2) disclosed a
method
for rapid and non-destructive detection of the water content in abalone during
the drying and
rehydration process. echo attenuation curve data of fresh and dried abalone
samples are
collected respectively, and the CPMG signals of the samples are collected. The
dried abalone
samples are rehydrated, the CPMG signals of the samples are collected during
the rehydration
process, and the true water content value of each sample is measured. In
correspondence to the
true water content value, a water content prediction model during the drying
and rehydration
process is established. These inventions take advantage of the fact that the
contents of hydrogen
proton-containing ingredients (for example, water in fruits and vegetables,
water in soybean
oil, and water in abalone) are different in sample matrices in different
states, and accordingly,
relaxation profile information in a low-field nuclear magnetic field are
different, so as to make
quick and effective identification and prediction.
Cheng Xinfeng et al. studied the water diffusion characteristics of taro chips
during the
microwave vacuum drying process. A drying test of taro chips is performed
using a microwave
vacuum drying oven at three microwave intensities of 1.5, 2.0 and 2.5 W/g, and
water migration
and distribution in the taro chips during the microwave vacuum drying are
measured using low-
field nuclear magnetic resonance technology. MRI detection shows that MVD taro
loses water
internally and externally at the same time, and the higher the microwave
intensity, the faster the
relaxation signal disappears. This study has revealed the law of water
diffusion in taro during
microwave vacuum drying, that is, the higher the microwave intensity, the
faster the water
diffusion rate and the conversion between different components of water in the
sample.
The change of flavor of spicy vegetables during the drying process is an
important indicator
to measure the drying quality. At present, there is still no device to quickly
detect the flavor
change of spicy vegetables during the drying process. Low-field nuclear
magnetic resonance
technology is widely used in intelligent detection of water content in fruits
and vegetables in
the drying process. In the present invention, the water content and flavor
substance content of
a material in the drying process are in a constantly changing process. With
the detection
principle of nuclear magnetic resonance technology, when hydrogen protons are
excited by a
pulse in a magnetic field to obtain a transverse relaxation time signal, the
intensity of the
4
Date Recue/Date Received 2021-07-08
CA 03126171 2021-07-08
relaxation signal is proportional to the number of nuclei with a fixed
magnetic moment
contained in a tested sample, and the signal attenuation process is closely
related to the
composition and structure of the tested substance, which can provide valuable
information such
as the physical and chemical environment inside the nucleus. A correlation
analysis between
.. flavor characteristics and nuclear magnetic response parameters is carried
out, and an artificial
neural network (BP-ANN) intelligent analysis system is performed, so that the
flavor changes
of microwave vacuum dried fruits and vegetables can be reflected by the
nuclear magnetic
relaxation spectrum information, achieving non-destructive, fast and
intelligent detection.
SUMMARY
The objective of the present invention is to provide a method for
intelligently detecting the
flavor changes of microwave-vacuum-dried spicy vegetables. The use of low-
field nuclear
magnetic resonance detection technology solves the problem of high complexity
of the original
flavor detection technology for spicy vegetables on the basis of ensuring the
shape and nutrients
of the spicy vegetables to the greatest extent, and achieves non-destructive,
convenient and
.. intelligent detection.
A low-field nuclear magnetism-based device for the intelligent detection of a
microwave-
dried spicy vegetable flavor, the device comprising a microwave dryer, a
computer 1, a
temperature sensor 2, a moving slide bar 5, a vacuum chamber 6, a raw material
7, a moving
plate 8, an NMR coil 9, a vacuum controller 10, a microwave controller 11, a
temperature
.. controller 12, a magnetron 13, an NMR box 14, a vacuum tube 15, and a
vacuum pump 16;
where the vacuum chamber 6 is disposed in the microwave dryer, the bottom of
the vacuum
chamber 6 is used to place the raw material 7, the moving slide bar 5 and the
temperature sensor
2 are disposed in the vacuum chamber 6, the moving slide bar 5 is used to move
a drying
chamber, and the temperature sensor 2 is used to measure the temperature in
the microwave
dryer in real-time; the vacuum chamber 6 is connected to the vacuum pump 16
outside the
microwave dryer through the vacuum tube 15; the vacuum controller 10, the
microwave
controller 11, the temperature controller 12 and the magnetron 13 are disposed
on the
microwave dryer; the vacuum controller 10 is used to control the vacuum pump
16 to adjust the
degree of vacuum in the vacuum chamber 6; the microwave controller 11 is used
to control the
.. microwave parameters of the microwave dryer; the temperature controller 12
is used to adjust
the temperature in the microwave dryer; the magnetron 13 is used to convert
energy obtained
from a constant electric field into microwave energy;
5
Date Recue/Date Received 2021-07-08
CA 03126171 2021-07-08
the NMR box 14 is disposed below the microwave dryer by means of the moving
plate 8,
and the vacuum chamber 6 is movable up and down within the microwave dryer and
the NMR
box 14 to ensure real-time sampling; the NMR coil 9 is disposed in the NMR box
14 and is
used to monitor in real-time the NMR parameters during the drying process;
the computer 1 is connected to the temperature sensor 2, the microwave dryer,
and the
NMR box 14, respectively, and is used to transfer a detected data parameter to
the computer 1;
a neural network model is in-built within the computer 1, and the detected
data parameter is
input to the neural network model for real-time analysis of data.
The microwave dryer and the NMR box 14 are connected to the computer through a
microwave dryer data cable 3 and an NMR data cable 4, respectively.
Description of device operation:
First, the material 7 is placed in the vacuum chamber 6 of the microwave
dryer, and the
respective physical parameters of the vacuum controller 10, the microwave
controller 11, the
temperature controller 12, the moving plate 8, and the NMR coil 9 are set up,
the computer 1
and its analysis software are turned on, the vacuum pump 16 is turned on, and
the magnetron
13 (microwave generator) is turn on when the corresponding vacuum degree is
reached, and the
drying begins.
Secondly, the moving plate 8 is opened after the drying reaches the set time,
and the moving
slide bar 5 is operated to send the vacuum chamber 6 and the material 7 into
the NMR box 14.
Sampling is performed by means of the NMR coil 9 for collection of nuclear
magnetic
parameters. After the collection is complete, the slide bar 5 is moved to pull
the drying chamber
up, the moving plate 8 is closed, and the drying continues.
Finally, a neural network model is in-built within the computer 1, and a
detected data
parameter is input to the neural network model for real-time analysis of data.
A low-field nuclear magnetism-based method for intelligent detection of
microwave-dried
spicy vegetable flavor, comprising the following steps:
(1) Pretreatment of spicy vegetables before drying: the raw material of spicy
vegetables is
cleaned up and cut into 1 X 1 x 1 cm cubes, which are placed on a dry tray.
(2) Microwave vacuum drying process: the spicy vegetable raw material is put
into the
vacuum chamber 6 of a microwave vacuum machine, and the vacuum pump 16 is
turned on.
When the vacuum degree reaches 10 MPa, the microwave controller is adjusted
and controlled,
entering the drying stage. During the microwave vacuum drying process, staged
sampling is
6
Date Recue/Date Received 2021-07-08
CA 03126171 2021-07-08
carried out.
(3) Low-field nuclear magnetic resonance analysis of the dried material: low-
field nuclear
magnetic resonance analysis is carried out to obtain various nuclear magnetism
response signal
parameters of samples; where the nuclear magnetic response signal parameters
include
transverse relaxation time and peak area; the transverse relaxation time
includes three types in
total: bound water relaxation time T21, immobile water relaxation time T22,
and free water
relaxation time T23; the peak area includes four types in total: bound water
peak area A21,
immobile water peak area A22, free water peak area A23 and total water peak
area Atotal.
(4) Flavor detection of the dried material: an electronic nose is used to
measure the changes
of different types of flavor substances in the same species of spicy
vegetable, and response
values of a flavor feature sensor of the electronic nose are obtained.
(5) Establishment of a microwave-dried spicy vegetable flavor prediction model
based on
low-field nuclear magnetism: a database for flavor feature sensor response
values of the
electronic nose and corresponding nuclear magnetic response signal parameters
of various
samples are obtained through staged sampling in a single drying experiment and
repeated
drying experiments, and a relationship is established by BP-ANN, to obtain a
microwave-dried
spicy vegetable flavor prediction model.
(6) Intelligent detection of the flavor changes of spicy vegetables during the
microwave
drying process: the spicy vegetable samples during the drying are sampled for
low-field NMR
analysis, and the current changes of flavor substances are predicted by the
microwave-dried
spicy vegetable flavor prediction model obtained in step (5).
Further, in the step (2), the microwave power is 150 W, and sampling is
performed every
10 min until the water content of the spicy vegetable raw material is less
than 10% on the dry
basis.
Further, in the step (3), a CPMG (can-purcell-meiboom-gill) pulse sequence is
used in the
low-field nuclear magnetic resonance analysis for signal collection. The
parameters used in the
CPMG sequence are: number of sampling points TD = 784794, spectral width 100
kHz, number
of echoes 18000, number of repetitive scans NS=4, and sampling repetition time
TW = 4000
ms. The collected signal is processed by nuclear magnetic resonance T2
inversion software to
obtain a T2 inversion spectrum and corresponding NMR parameters.
Further, in the step (4), the changes of different types of flavor substances
in the same
species of spiced vegetable are measured by using the electronic nose, and the
sample (2.0 g on
7
Date Recue/Date Received 2021-07-08
CA 03126171 2021-07-08
the dry basis) is placed in a sealed vial (20 mL) and is allowed to stand for
60 min. The
collection time is 150 s.
Further, in the step (5), when a relationship equation between the peak area
of different
components of water in nuclear magnetic response signal parameters and the
sensor response
value of the electronic nose for the sample is established, the peak area
needs to be normalized
by mass.
The spicy vegetables include, but not limited to, ginger, garlic, green onion,
and pepper.
1. The present invention uses low-field nuclear magnetic resonance to solve
the technical
problems of flavor change detection during the original spicy vegetable drying
process on the
basis of ensuring the shape to the greatest extent, achieves non-destructive,
convenient and
intelligent detection, improves detection work efficiency and product
integrity, and effectively
monitors changes in flavor during the drying process.
2. The present invention has convenient operation, simple process, high
accuracy of
detection results, short time-consuming, and no damage to samples, and can
effectively monitor
in real-time the changes in flavor during the drying process.
3. The method proposed by the present invention can accurately and effectively
judge the
changes of different types of flavor substances in spicy vegetables during the
drying process,
which is very helpful in adjustment and control of the drying process.
BRIEF DESCRIPTION OF THE DRAWINGS
FIG. 1 is a BP-ANN prediction model of ginger flavor during microwave vacuum
drying.
In the figure, (a) is a training set, (b) is a verification set, (c) is a
testing set, and (d) is a
comprehensive set.
FIG. 2 is a BP-ANN prediction model of garlic flavor during microwave vacuum
drying.
In the figure, (a) is a training set, (b) is a verification set, (c) is a
testing set, and (d) is a
comprehensive set.
FIG. 3 is a simplified integration diagram of the device of the present
invention.
FIG. 4 is a diagram of states of sample detection, in which (a) is a state
when microwave
drying is on, (b) is a state when a moving plate is removed, (c) is a state
when a low-field NMR
measurement is performed, and (d) is a state when returning to the microwave
drying.
In the figures: 1 computer; 2 temperature sensor; 3 microwave dryer data
cable; 4 NMR
data cable; 5 moving slide bar; 6 vacuum chamber; 7 raw material; 8 moving
plate; 9 NMR
8
Date Recue/Date Received 2021-07-08
CA 03126171 2021-07-08
coil; 10 vacuum controller; 11 microwave controller; 12 temperature
controller; 13 megnetron;
14 NMR box; 15 vacuum tube; 16 vacuum pump.
DETAILED DESCRIPTION OF THE EMBODIMENTS
The technical solution of the present invention is further illustrated below
with reference
to specific examples and the accompanying drawings.
Example 1: low-field nuclear magnetism-based method and device for the
intelligent
detection of a microwave-vacuum-dried ginger flavor
1. Ginger was cleaned up with water, peeled and cut into lx1x1 cm cubes, which
were
placed on a tray for microwave vacuum drying. The vacuum pump was turned on,
the
microwave heating system was activated when the vacuum degree reached 10 kPa,
and the
microwave power was set to 150 W. Staged sampling was performed for low-field
nuclear
magnetic resonance analysis, to obtain a transverse relaxation time T2 curve
and various
response signal parameters of a sample. And an electronic nose was used to
measure changes
in the flavor of the sample. A feature sensor of the electronic nose was
determined.
2. Model establishment and intelligent adjustment and control: repeated
experiments were
performed many times, to obtain a database for flavor feature sensor response
values and
corresponding nuclear magnetic response signal parameters of a large number of
samples. The
transverse relaxation time and peak area data of ginger during the drying
process were fitted
with the feature sensor through metrology software, and the nuclear magnetic
signal (T21, T22,
T23, A21, A22, A23, and Atotal) was used as input parameters of BP-ANN, the
feature sensor of
the electronic nose was used as as an output parameter, and 70% of the sample
size was
randomly selected as a training set to establish a flavor-related prediction
model (as shown in
Fig. 1). It can be seen from the figure that there is a good correlation
between the predicted
values of the flavor of the ginger sample obtained by the BP-ANN method and
the chemical
values, and the R of the training set is greater than 0.9 in all cases. In
order to verify the accuracy
and stability of the prediction model, 15% of the samples were used as a
verification set and
15% of the samples were used as a test set. The results show that the R of the
verification set is
greater than 0.9 in all cases, indicating that the predictive ability of the
model is very good, and
the comprehensive R is 0.97798, which indicates that the stability of the
prediction model is
good. Twenty groups of samples during the drying process were randomly
selected for low-
field nuclear magnetic resonance analysis by means of an automatic sampling
system. The
established ginger flavor BP-ANN analysis model was used to predict the
current flavor
9
Date Recue/Date Received 2021-07-08
CA 03126171 2021-07-08
conditions. The correlation coefficient R of the model validation set is
0.9688, indicating that
the low-field nuclear magnetic resonance combined with the BP-ANN model can
accurately
predict the flavor changes of ginger during the drying process.
Example 2: low-field nuclear magnetism-based method and device for the
intelligent
detection of a microwave-vacuum-dried garlic flavor
1. Garlic was peeled, cleaned up, and cut into 0.4 cm slices, which were
placed on a tray
for microwave vacuum drying. The vacuum pump was turned on, the microwave
heating
system was activated when the vacuum degree reached 10 kPa, and the microwave
power was
set to 150 W. Staged sampling was performed for low-field nuclear magnetic
resonance
analysis, to obtain a transverse relaxation time T2 curve of a sample and
various response signal
parameters. And an electronic nose was used to measure changes in the flavor
of the sample. A
characteristic sensor of the electronic nose was determined as S4.
2. Model establishment and intelligent adjustment and control: repeated
experiments were
performed many times, to obtain a database for flavor feature sensor response
values and
corresponding nuclear magnetic response signal parameters of a large number of
samples. The
transverse relaxation time and peak area data of garlic during the drying
process were fitted
with the feature sensor through metrology software, and the nuclear magnetic
signal (T21, T22,
T23, A21, A22, A23, and Atotal) was used as input parameters of BP-ANN, the
feature sensor of
the electronic nose was used as as an output parameter, and 70% of the sample
size was
randomly selected as a training set to establish a flavor-related prediction
model (as shown in
Fig. 2). It can be seen from the figure that there is a good correlation
between the predicted
values of the flavor of the different samples obtained by the BP-ANN method
and the chemical
values, and the R of the training set is greater than 0.9 in all cases. In
order to verify the accuracy
and stability of the prediction model, 15% of the samples were used as a
verification set and
15% of the samples were used as a test set. The results show that the R of the
verification set is
greater than 0.9 in all cases, indicating that the predictive ability of the
model is very good, and
the comprehensive R is 0.97581, which indicates that the stability of the
prediction model is
good. Twenty groups of samples during the drying process were randomly
selected for low-
field nuclear magnetic resonance analysis by means of an automatic sampling
system. The
established garlic flavor BP-ANN analysis model was used to predict the
current flavor
conditions. The correlation coefficient R of the model validation set is
0.9589, indicating that
the low-field nuclear magnetic resonance combined with the BP-ANN model can
accurately
predict the flavor changes of garlic during the drying process.
Date Recue/Date Received 2021-07-08
