KR0173674B1 - Non-destructive examination apparatus and magnet box for agricultural products using nmr - Google Patents

Abstract

The present invention relates to a non-destructive testing method of agricultural products using hydrogen nuclear magnetic resonance (NMR) and a magnetic box and a device that is a sensor thereof.

There is a medical diagnostic device using hydrogen nuclear magnetic resonance, but because it uses superconducting magnets to generate high magnetic fields, its structure and operation are complicated, bulky, and expensive to maintain. Inadequately, the present invention provides a non-destructive testing method, a magnetic box, and an apparatus thereof, which can be used practically for agricultural product inspection and the like.

In the case of high magnetic permeability case (2) made of soft magnetic material, plate-shaped permanent magnets (6) and (8) are installed on the upper and lower surfaces to form a magnet box (2) which is a hydrogen nuclear magnetic resonance sensor, but the permanent magnet (6). The magnetic permeability (B ₀ ) is fixed by attaching and fixing high permeability pole faces 16, 18, shimming plates 20, 22, and corner steel 24 to the surfaces and edges of (8), respectively _. ), Improve the performance by improving the uniformity of the cavity, and the pores 10 through which the sample (agricultural product) can pass between the permanent magnets (6) and (8), and then install the RF / sensing coil (34) The formation of (B ₁ ) and the resonance signal of the sample can be detected.

The RF / detection coil 34 includes a transmitter / receiver 42, a preamplifier 54, a receiver and detector 56, and an oscilloscope 58 under the control of a computer 46 and a pulse programmer 48. ) Are sequentially connected to obtain a pure water resonance signal of the sample (agricultural product), and then displayed on an image, and the high frequency power amplifier 52, the high frequency switching unit 50, and the pulse are connected to the other end of the transmitter / converter 42. By connecting the programmer 48 and the computer 46 that controls the system in turn, a magnetic field B ₁ is generated to resonate the sample, thereby making it possible to non-destructively and easily inspect the moisture content and sugar content of the produce. .

Description

Nondestructive testing device for agricultural products using hydrogen nuclear magnetic resonance (NMR) and magnetic box for improving magnetic field strength and uniformity

1 is a view for explaining nuclear magnetic resonance,

(a) the substance is not in a magnetic field.

(b) the substance is in a magnetic field.

2 is a view for explaining nuclear magnetic resonance showing the state of formation of the macroscopic magnetization M.

3 is a diagram for explaining nuclear magnetic resonance, a resonance signal generated by the relaxation of the macroscopic magnetization M.

4 is a B-H curve of the permanent magnet of the present invention.

5, 6 is a longitudinal cross-sectional view of the sensor (magnetic box) of the present invention.

7 is a B-H curve diagram of the sensor (magnetic box) case of the present invention.

8 is a graph analyzing the variation and uniformity of the magnetic field strength according to the width of the shimmering plate of the permanent magnet of the present invention.

9 is a longitudinal sectional view of the present invention RF / sense coil.

10 is a resonant circuit diagram of an RF / sense coil of the present invention.

11 is a circuit block diagram of the present invention.

12 is a FID signal diagram used in the present invention.

Figure 13 is a waveform diagram of the water resonance signal obtained in the rice water content test of the present invention.

FIG. 14 is an envelope waveform diagram of signal processing of the water resonance signal of FIG.

15 and 16 are envelope waveform diagrams obtained from rice samples having different moisture contents in the rice moisture content test of the present invention.

Figure 17 is a graph showing the correlation between the water content and the maximum value in the rice water content test of the present invention.

18 and 19 are graphs showing the relationship between the arithmetic sum of four points after the maximum value and the water content in the rice water content test of the present invention.

20 is a signal waveform diagram for glycerol in the fruit and vegetable sugar test of the present invention.

21 is an envelope waveform diagram obtained by processing the glycerol resonance signal of FIG. 20 according to the present invention.

22, 23, and 24 are envelopes for sugar water samples containing glycerol in the present invention fruit and sugar test.

25 is the relationship between the slope and the sugar content for two points in the sugar test of the present invention.

Figure 26 is a comparison of the envelope for water, apples, kiwi in the sugar test of the present invention.

Figure 27 is a graph showing the results of verifying the sugar content for the remaining 10 samples using a multi-linear regression model in the sugar test of the present invention.

Figure 28 is a sugar graph of the apple by the neural network in the present invention sugar test.

Figure 29 is a graph of the sugar content of the pear by the neural network in the present invention sugar test.

30 is a correlation diagram between FID signal intensity and mass in the present invention sugar test.

Figure 31 is a graph of the sweetness of the kiwi by the neural network in the sugar test of the present invention.

32 is a sugar graph of kumquat by neural network in the present invention sugar test.

33 is a cross-sectional view of another embodiment of the present invention corner steel.

* Explanation of symbols for main parts of the drawings

2: magnet box (sensor) 4: case

6,8 permanent magnet 10 air gap

12,14: air hole 16,18: pole face

20,22: seaming plate 24: corner steel

26,28: Front and rear plates 30,31: Left and right plates

34: RF / detection coil 36: tube

38 tunnel 40 series resonant circuit

42: transmitter and receiver 44: insulating material

46: Computer 48: Pulse Programmer

50: high frequency switching unit 52: high frequency power amplifier

54: Preamplifier 56: Receive and Detector

58: oscilloscope 60: digitizer (A / D converter)

62: image display Bo: magnetic field strength (permanent field) of the permanent magnet

B ₁ : Magnetic field strength (kinetic field) of RF / detection coil

The present invention relates to a non-destructive testing device for agricultural products and a magnetic box (sensor) thereof, and in particular, it is possible to measure the moisture content and sugar content in agricultural products continuously and non-destructively using hydrogen nuclear magnetic resonance (Nuclear Magnetic Resonance) principle. The magnet box is improved to greatly increase the strength and uniformity of the magnetic field, thereby greatly increasing the precision of the accumulation.

The development of a non-destructive moisture content (or sugar) measuring device for agricultural products (or processed foods) is urgently needed for the automation of drying and storage facilities, processing facilities or food processes. Accurate moisture content establishment of agricultural products is the source of thermal energy during drying and storage. Plays an important role in the efficient use and quality decision making.

Non-breakthrough continuous measurement of the moisture content during the drying process can prevent poor drying, such as overdrying or undrying, and also enables uniform drying. Overdrying results in the loss of produce and waste of thermal energy, while less dried produce causes decay during storage or deterioration by microorganisms, leading to degradation and loss of quality.

In the case of rice, it is essential to measure the moisture content of rice in order to minimize the loss of rice during storage in the storage or shipping place.How to measure the moisture content of rice is a direct method including an oven method and a method of measuring the change of dielectric constant according to the moisture content of rice. And non-destructive methods such as measuring changes in electrical resistance.

In recent years, since non-destructive measurement of rice moisture content is required in large-scale rice harvesting facilities, a method of measuring the change in electrical resistance and dielectric constant has been suggested as the method.However, the moisture content measurement by electrical resistance is less representative because of the small number of samples. The water content measurement by has a problem in that it is greatly influenced by the density of the product.

On the other hand, nuclei having an odd sum of protons or neutrons, such as ¹ H, ¹³ C, ¹⁹ F, and ³¹ P, rotate about one axis and consequently have an angular moment and a magnetic moment. Nuclear magnetic resonance (NMR), which absorbs energy, generates resonance in a magnetic field where the moment nucleus changes at the same frequency as the rotational speed of the axis.

Prior to describing a preferred embodiment of the improvement of a magnet box for non-destructively inspecting agricultural products using hydrogen nuclear magnetic resonance (NMR), the principle of hydrogen nuclear magnetic resonance will be described as follows.

If a substance contains a large amount of water, 99.98% of the hydrogen nuclei that make up the water molecule are in the ¹ H state and can be thought of as a small magnet with a magnetic moment. Each of the hydrogen nuclei is randomly arranged, and as a whole, is not magnetic as shown in FIG.

If the material is placed in a strong magnetic field (B ₀ ) of several thousand gauss, the magnetic field of momentum of each hydrogen nucleus is aligned in the direction of the strong magnetic field acting from the outside as shown in FIG. They perform precession movement based on the direction of the magnetic field. Each velocity of the precession has its own value depending on the type of atomic nucleus and changes in proportion to the intensity of the external magnetic field.

That is, the hydrogen atom precesses at a speed of 4.26 MHz in a 1.000gauss magnetic field. The vector sum of the magnetic moments of the entire hydrogen nucleus in the direction of the magnetic field (Z axis) is called macroscopic magnetization or total magnetization M.

In the direction perpendicular to the strong magnetic field B ₀ (X-axis), another relatively weak dozens of Gaussian oscillating magnetic fields B _l are placed at the same frequency as the precessional angular velocity of the hydrogen nucleus. When given, it causes resonance and absorbs energy, causing macroscopic magnetization M to tilt in the Y direction as shown in FIG. When the B ₁ magnetic field is extinguished, the inclined M is returned to its original position, the Z axis, and emits a resonance signal to the outside by emitting energy as shown in FIG. 3.

The resonance signal is captured with a sensing coil to enable chemical qualitative and quantitative analysis of the irradiated material. First, using the magnetic field strength of the permanent magnet, the frequency ν of the resonance signal is calculated. Resonance signals used to find chemical information of a substance go through several processes of electronic signal processing to find the desired signal variables.

Representative signal variables include chemical shift and relaxation time. Even with the same hydrogen atom, the resonance frequency v slightly varies depending on the bonding environment. This difference is called chemical variation, and it is expressed as the difference [ppm] based on the hydrogen resonance frequency from tetramethylsilane (TMS, [Si (CH ₃ ) ₄ ]) and is used as a chemical substance. The composition of can be analyzed.

In addition, the inclined macro magnetization vector M having M _X , M _Y , and M _Z on the X, Y, and Z axes, respectively, has two kinds of relaxation times as it returns to the original Z position. The time for relaxation when M _Z returns to its original size M ₀ is called Spin-Lattice Relaxation Time (T ₁ ), and the time for relaxation when M _X or M _Y disappears is called T ₂ (Spin-Spin Relaxation Time). This relationship is expressed as an expression as follows.

In the above formula, T ₁ is the relaxation time when the relaxation of the atomic nucleus occurs between the environment around the nucleus, and T ₂ is the relaxation time occurring between the nuclei. T ₁ and T ₂ have different values depending on the state or kind of material, and T ₂ is always equal to or less than T ₁ . T ₂ values are shorter in the solid state and longer in the liquid state. Also, the time constants of water and oil have different values.

The advantages of nuclear magnetic resonance can be maintained non-destructively without any chemical or physical damage to the sample to be tested, and also can be re-measured at regular intervals if necessary, so that the diagnostic purpose and the medical field of the medical field. It is used for chemical analysis.

Prior to the practical use of nuclear magnetic resonance (NMR) devices, infrared spectroscopy and mass spectrometers were used mainly for chemical analysis, but the principle of nuclear magnetic resonance was introduced in the late 1970s due to the development of electromagnetism, signal processing, and computer technology. Is widely used in chemical, biological, and medical fields as the most powerful means for structural, quantitative analysis and component inspection of chemical substances in certain substances, and mainly observes nuclear magnetic resonance of hydrogen and carbon. It is widely used in chemical qualitative and quantitative classification.

In addition, the nuclear magnetic resonance signal of hydrogen, the monovalent atom, is the strongest in structural analysis and quantitative measurement of chemical components using nuclear magnetic resonance, and about 99.98% of the total hydrogen atoms in nature are such monovalent atoms.

Above all, since water (H ₂ O) emits the strongest hydrogen resonance signal among the compounds of monovalent hydrogen ( ¹ H), the measurement of moisture content using nuclear magnetic resonance is relatively easier than the quantitative measurement of other compounds. have.

On the other hand, in order to reclassify the already filed file by component, the fruits are imported into the component measurement import system, and then analyzed by the NMR, the component measuring device, to analyze the fruit's sugar ratio, hardness, moisture content, and acid mixture. Product lines and devices, such as fruit, are fed back to the results obtained from the NMR value of the fruit component measurement values, which are programmed in the computational analysis device, so that the fruit can be destructively passed in accordance with each component range of the fruit. A method for evaluating moisture in wire and cable insulators, which has been disclosed in Korean Patent No. 1-75854 and which makes it possible to detect non-destructively and simply and quickly moisture distribution of wires or cables using a nuclear magnetic resonance device (NMR). Japanese Patent Application Laid-Open No. 7-83860 discloses a patent for a non-destructive test using NMR. Although it is possible to detect and measure continuously, the donut-type permanent magnet can be used to continuously move the test object or the measured object. As a result, the uniformity of the magnetic field is reduced and the accuracy (sensitivity) of the measured value is inferior. There was this.

That is, in the case of permanent magnets, the strength of the magnetic field becomes stronger toward the edges or edges, so the magnetic field intensity (magnetic flux density) is nonuniform. Such nonuniformity narrows the bandwidth of the resonant signal that can be detected and attenuates the resonance signal. Since the period is reduced to reduce the overall sensitivity of the resonance signal sensor, the uniformity of the magnetic field strength is absolutely required.

Accordingly, the present invention improves the magnetic box capable of nondestructively measuring components such as moisture content and sugar content in agricultural products (test subjects or measured objects) using the hydrogen nuclear magnetic resonance (NMR) principle as described above. It is an object to make the measurement accuracy (sensitivity) significantly improved by uniformizing the intensity and uniformity distribution.

Another object of the present invention is to provide a practical hydrogen nuclear magnetic resonance magnet box (sensor) suitable for nondestructive inspection of agricultural products.

In order to achieve the above object, the present invention provides a magnetic box that is a sensor for generating a static magnetic field, and an RF / detection coil for transmitting a high frequency signal (Radio Frequency) to generate a magnetic field and detect a resonance signal. A coil, an RF signal processor, and a control computer are included to enable quality measurements without being affected by the density of the produce.

In addition, sensors of nuclear magnetic resonance devices used in general medical diagnostic devices or chemistry analyzers use super conducting magnets to generate magnetic fields of 1T (10,000gauss) or more. In the present invention, a practical magnet box suitable for the present invention is constructed by using a permanent magnet capable of forming a magnetic field of about 1,000 gauss. However, the permanent magnet has a pole face on the surface and side. Face, corner steel and shimming plate are installed to equalize the magnetic field to greatly improve detection sensitivity and detection bandwidth.

In addition, in the present invention, since the control and sensing circuits should be appropriately changed to suit the range of the resonance (resonance) frequency and the design of the tuning circuit must be performed uniquely in consideration of the size of the measurement object, A pulse having a specific period is formed to send a resonance signal to a sample in a permanent magnet as a system, and a 4.1 MHz pulse for resonance is added to this signal, and a magnetic field for nuclear magnetic resonance is switched using a high frequency pulse signal. After amplification and amplification are supplied to the RF / sensing coils, the high frequency pulse signal applied to the RF / sensing coil and the reception of the resonance signal are interrupted. In order to obtain the water (H ₂ O) signal from the resonance signal detected by the RF / detection coil, low-noise amplification is carried out using a pre-amplifier to separate the unwanted noise, and then processed by an image signal with an oscilloscope by signal processing and phase detection. Display and signal the signal in several steps to get the desired signal.

In general, there are many permanent magnets depending on the material used, but ceramic (ceramic 8) series are low-cost permanent magnets that can form a low magnetic field, whereas Nd-Fe-B (Neodymium-Iron-Boron) series and Rare Earth series are permanent. The magnet can form a high magnetic field with a high-priced permanent magnet, and when using this permanent magnet to form a void of 100 mm, which is the distance between the permanent magnets, Nd-Fe-B has a low volume of 4,000 to 5,000 gauss. High magnetic field strength can be obtained, and in the case of ceramic 8 proposed for experiment in the present invention, magnetic field strength of about 800 to 1500 gauss can be obtained.

Therefore, it is possible to develop a device for measuring the internal quality of agricultural products, which is inexpensive and has little maintenance cost using permanent magnets. The material of such a permanent magnet is expressed as μ (Magnetic permeability of the medium), the value of μ also forms a non-linear relationship (BH curve) that varies with the strength of the magnetic field and shows a slight change depending on the ambient temperature. The permanent magnet of 8 series ceramics uses the BH curve based on room temperature (20 ℃) as shown in FIG. Hereinafter, described in detail with reference to the accompanying drawings a preferred embodiment of the present invention.

5 and 6 are longitudinal cross-sectional views of the magnet box (sensor) 2 of the present invention which resonates by applying a high frequency pulse to a sample and then detects a resonance signal of the sample. The case magnet box 4 is generally rectangular in shape. It is easy to design, manufacture and install compared to curved surface.

The permanent magnets 6 and 8 having strong magnetic force are attached to the upper and lower surfaces of the conductive case 4, and the upper and lower permanent magnets 6 and 8 can be spaced apart so that the sample can be placed or passed. Air gaps 10, which are spaces, are formed, and through holes 12 and 14 are formed in the left and right plates of the case 4 so that the sample can pass through the gaps 10. As shown in FIG.

The permanent magnets 6 and 8 use a rectangular magnetic shell in which N and S magnetic poles are formed up and down, and have strong polarities in the Z direction (up and down directions) with opposite polarities. Let a magnetic field form

The performance of the sensor 2 is determined by the intensity of the magnetic field (maganetic flux) and the homogeneity of the magnetic field (homogeneity), these strength and uniformity is the material shape and specification of the permanent magnet (6) (8) and the case (4) The case 4 has a magnetic permeability by using a soft magnetic material with high permeability and low hysteresis loss, so as to reduce the magnetic loss by eliminating the influence or interference of the external magnetic field. Do it.

In the case of the permanent magnets (6) and (8), the strength of the magnetic field becomes stronger toward the edges or edges, so the magnetic field intensity (magnetic flux density) is nonuniform, and this nonuniformity narrows the bandwidth of the resonant signal that can be detected. Since the attenuation period of the resonance signal is reduced to reduce the overall sensitivity of the sensor 2, the uniformity of the magnetic field strength is absolutely required.

Therefore, in the present invention, the entire magnetic surface of the permanent magnets 6 and 8 is formed of a soft magnetic material of the same quality as the case 4 and has a flat pole face such as the permanent magnets 6 and 8. 18) adhesive fixation to form the upper and lower magnetic pole surface to improve the uniformity of the magnetic field.

In addition, shimming plates (20, 22) and corner steel (24) made of soft magnetic materials are additionally installed at the edges and corners of the permanent magnets 6 and 8, which are relatively strong in magnetic field strength. The magnetic field uniformity is further improved to form a static magnetic field (Bo).

In the present invention, the shimming plates 20 and 22 and the corner steel 24 are structures proposed for improving the strength and uniformity of the magnetic field, which are obtained by analyzing the magnetic field distribution of the permanent magnets 6 and 8. By directing the direction of the magnetic field outward from the edges of the faces 16 and 18 toward the center of the cavity 10, the external leakage of the magnetic field is suppressed to the maximum and the uniformity is improved.

The performance of the magnetic box used in the NMR sensor of the present invention is determined by the strength of the magnetic field (Magnetic Flux) and the uniformity of the magnetic field (Homogeneity). Since they are influenced by the material of the permanent magnet, the shape and size of the magnet box, the design of the magnet box of the desired performance is repeated a number of trial and error.

Even if the material of the permanent magnet is inexpensive, if it is accompanied by design and manufacturing through trial and error, the initial development cost will be increased and the development period will be considerably extended. Therefore, in the present invention, if the trial and error is simulated on a computer by applying the finite element analysis technique, not only the development cost can be reduced and the development time can be shortened, but also the appropriate design specification can be selected, and a new shape magnet box is designed. In the present invention, since it helps a lot, the application of such a finite element method to design the magnetic field strength and uniformity of the permanent magnet to the optimal state.

The finite element analysis technique has many applications for solving problems such as static and dynamics of structures, electric and magnetic fields, heat and fluids, and many programs and techniques for this are known. Computer programs) include ABAQUS, NASTRAN, FLUENT, ANSYS, and the general process of finite element analysis is as follows.

(1) Drafting of the magnet box using CAD (design program using computer)

(2) Input characteristic data such as magnet and iron material

(3) Mesh generation

(4) Finite Element Analysis

(5) Post processing

(6) If the design criteria are not satisfied, return to (1).

(7) Precision design of magnet box using computer design program (eg CAD).

In the present invention, finite element analysis was performed using ANSYS (SASI-Swanson Analysis System, Inc.) 5.0, which is a finite element analysis tool for structure, electromagnetic field, and heat and fluid for two-dimensional and three-dimensional shapes. In the present invention, the basic shape of the magnet box was selected in consideration of manufacturing convenience, box shape, a total of 11 design factors were selected as shown in Table 1.

In the present invention, the characteristics of these design factors and their effects on the strength and uniformity of the magnetic field are analyzed by using the finite element analysis technique, and by selecting them optimally, a magnetic box having the desired magnetic field and uniformity is designed. In Table 1, the air gap is a space in which a sample of agricultural products is placed, which acted as a constraint in the design of the magnet box.

The governing equation used in the finite element to analyze the strength and uniformity of the magnetic field of the magnet box is the Maxwell equation for the electromagnetic field such as Equation (1).

The cross-sectional structure of the hydrogen nuclear magnetic resonance magnet box 2 manufactured to test and verify the uniformity of the permanent magnets 6 and 8 in the present invention is the same as that of the fifth and sixth degrees. Same as 2, forming a skeleton of the corner steel 24, the shimming plates 20, 22, the pole faces 16, 18, and the magnet box 2, and serves as a magnetic shielding path. As for the image of the case 4, BH curve is the same as FIG.

In the present invention, the front and rear plates 26 and 28 that are adhesively fixed or bolted to the case 4 are provided with through holes 12 and 14 for entering and exiting the sample, but through the holes 12 and 14, respectively. Since magnetic leakage is generated through the magnetic field, the strength of the magnetic field is weaker than that of the left and right plates 30 and 32 of the case 4, so that the side of the shimmering plates 20 and 22 positioned on the through holes 12 and 14 side. The width K is greatly extended so that the magnetic leakage is compensated for.

In the above, the front and rear plates 26 and 28 are made of a non-magnetic material such as an aluminum plate and electrically conductive, so as not to be affected by the magnetic effect when the sample enters and exits, and the left and right plates 26 and 28 and the case. (4) is to be electrically connected and electrically shielded.

On the other hand, inside the RF / sensing coil 34 is inserted into a rigid tube (36) having a spherical body having a low dielectric constant and then installed so that both ends of the tube (36) protrude into the through holes (12) (14) of the two side plates. Forms a tunnel 38 through which the tube 36 can pass, and the thickness of the tube 36 is formed to a thickness that is about the same as the cross-sectional diameter of the RF / detection coil 34 so that the internal cross-sectional area of the tunnel 38 for sample passage can be obtained. Do not greatly reduce this.

In this case, the RF / sensing coil 34 may be wound around the outer surface of the tube 36 to prevent sagging of the coil, and both ends of the tube 36 are supported by both side plates 12 and 14.

One end of the RF / sensing coil 34 is grounded to a conductive case 4, and the other end of the RF / sensing coil 34 is connected to the series resonant circuit 40 and the transmitter / receiver 42 in turn. The one side plate 32 is insulated with an insulating material 44.

At the bottom of the tunnel 38, a continuous or discontinuous conveying means such as a belt conveyor conveyed by a variable speed drive device is installed to enable continuous or discontinuous measurement of the moisture content of the produce or the sugar content of the fruit or vegetable. The material of the means is preferably nonmagnetic material and insulating material.

On the other hand, in order to measure the magnetic field strength of the sensor 2 of the present invention using a magnetic flux density meter (gauss meter, model name: MG-4D, manufacturer: Walkers co, instrument resolution: ± 0.5gauss or ± 0.05%) diameter 100 After measuring the magnetic field strength in ㎜ circle, the uniformity of the magnetic field was calculated by Equation (2) and showed some error between the measured value and the predicted value. Table 3 below shows the magnetic field strength of XZ plane and YZ plane of sensor (2). And a comparison between the measured value of the uniformity and the result value of the predicted value.

There are many reasons for this error, but first, because finite element analysis is a method of decomposing the original object into finite elements to obtain an approximate solution based on them, it is due to the existence of errors due to the size and shape of the decomposed finite element. .

In addition, since the B-H curve used in the finite element analysis is a characteristic at room temperature (20 ° C.), an error due to the ambient temperature of the magnet at the time of measurement also exists. However, in the case of the permanent magnet (ceramic 8) used in the present invention, the change in the strength of the magnetic field according to the temperature is about 0.19% loss / ℃, the temperature at the time of the measurement is close to room temperature, the error according to the temperature is not so large in the present invention. I can say no.

In order to further improve the uniformity of the permanent magnets (6) and (8) in the present invention, the graph analyzing the magnetic field and the uniformity change by performing finite element analysis while changing the specifications of the shimmering plates (20) and (22) is shown in FIG. The XZ width (d) of the newly prepared shimming plates (20) and (22) was 20.3 mm and the YZ width (k) was 17.8 mm, respectively.

For the newly designed sensor, the uniformity was improved from 1195.84 ppm in the X-Z plane to 345.18 ppm, in the Y-Z plane from 396.84 ppm to 115.74 ppm, and the intensity of the magnetic field was slightly reduced from 958gauss to 956gauss.

In the above state, as a result of finite element analysis by attaching corner steel to four corners on the XZ plane, the magnetic field strength increased from 956gauss to 962gauss, and the uniformity of XZ plane was improved to 103.23ppm. The YZ plane is the same as before because there is no design change.

Meanwhile, in the present invention, by providing the shimmering plates 20 and 22 at each edge of the pole faces 16 and 18, the uniformity of the magnetic field, such as electricity, is realized, and the shimmering plate is installed to verify the improved uniformity. The magnetic field strength and uniformity of the unused magnet box were analyzed by finite element analysis, and the specifications of the magnet box used in the analysis were used as shown in Table 2.

Comparing the intensity and uniformity of the magnetic field calculated by the above analysis with the case where the shimming plates 20 and 22 are installed and not, the table is as shown in Table 4, and the magnetic field when the shimming plates 20 and 22 are installed. The uniformity of is greatly improved, but the intensity of magnetic field is greatly reduced.

That is, in order to increase the strength of the magnetic field reduced by the installation of the shimmering plates 20 and 22, the thickness e, the length i, or the width a of the permanent magnets 6 and 8 must be changed. It can be seen.

In the present invention, the strength of the magnetic field was changed while gradually increasing the thickness (e) of the permanent magnet so as to compensate for the magnetic field while minimizing the influence on the uniformity of the magnetic field. In addition, by increasing the thickness of the permanent magnets 6 and 8 by 10% (7.6 mm), it was possible to design a magnetic box in which the magnetic field in the absence of the shimmering plates 20 and 22 was close to the intensity.

It can be seen that the effect of the shimming plates 20, 22 and corner steel 24 used in the present invention is more pronounced compared to the design of a magnet box of the same uniformity without such structures 20, 22 and 24. have.

That is, in order to improve the uniformity of the permanent magnets 6 and 8, 50% and 100% of the width (a) and the length (i) of the permanent magnet without the shimmering plates 20 and 22 and the corner steel 24 are respectively. The results of the finite element analysis with increasing results are shown in Table 6 below.

In Table 6, if the width and length of the permanent magnets 6 and 8 are increased by 50% and 100%, respectively, the uniformity is considerably improved, but the intensity of the magnetic field is rapidly decreased.

In addition, when the width and length of the permanent magnet is increased by 1.5 times, the strength of the magnetic field similar to that of the magnetic field presented in the present invention can be obtained, and the uniformity shows worse performance than that. The uniformity shows similar performance to the present invention in the XZ plane, but the magnetic field strength is further reduced by 112 gauss, and the uniformity of the YZ plane also falls short of the present invention, compensating for the reduced magnetic field strength. In order to increase the thickness of the permanent magnets 6 and 8 without the shimmering plate 20 and 22 and the corner steel 24, the results are shown in Table 7 below.

That is, even if more than four times the permanent magnet material is used without the shimmering plate 20, 22 and the corner steel 24, the design of the sensor showing the desired strength and uniformity of the magnetic field is impossible, but in the present invention, the permanent magnet of about 20% By adding materials and soft magnetic materials for the shimmering plates 20, 22 and the corner steel 24, it is possible to design the permanent magnet 2 exhibiting the desired strength and uniformity of the magnetic field.

In addition, if the size of the permanent magnet is increased, the volume and weight of the sensor increases, space and installation constraints are also solved.

When the shimmering plate 20, 22 and the corner steel 24 are used to synthesize the experimental results of the sensor 2 of the present invention, the results of comparing the sensors having similar magnetic field strength and uniformity between the two are shown in Table 8 below. Same as

As shown in Table 8, when the shimmering plate 20, 22 and the corner steel 24 are not used, the thickness of the permanent magnet is 40.9% as compared with the case where the structures 20, 22 and 24 are used. It can be seen that the width and length of the permanent magnet are greatly increased by 100%, respectively, and use about six times the material. In the present invention, the structure 20, 22, 24 in the sensor 2 is used. With the material of 6, the desired strength and uniformity of the magnetic field can be obtained, thereby greatly reducing the weight and volume.

On the other hand, since the space 10 of the sensor 2 of the present invention is 152.4 mm, the maximum outer diameter of the RF / detection coil 34 provided in the space 10 is within about 150 mm. Since the magnetic field uniformity of the RF / detection coil 34 determines the transmission characteristics of the entire transmission / reception signal, it should be manufactured so that the magnetic field is uniform without the sensitivity of the resonance signal and the attenuation of the resonance signal.

In addition, the direction of the magnetic field should be designed and installed so as to be perpendicular to the Z axis, the magnetic flux direction of the permanent magnets (6) and (8), and the electrical conductivity is required because it is an air coil installed in the air gap (10). Excellent, rigid, relatively thick conductors, such as copper or aluminum, may be used.

The cross section of the RF / detection coil 34 is shown in FIG. 9 to obtain a uniform magnetic field strength and to make the spacing between coils uniform at equal intervals in order to improve the sense of resonance and prevent signal attenuation. (34) It is preferable that the self capacitance is not induced by being separated by about 1 to 2 times the diameter.

The RF / detection coil 34 receives a high frequency pulse signal (high frequency alternating current signal) to form a magnetic field in the sample to generate nuclear magnetic resonance and to sense (receive) the resonance signal of the sample relaxed by the nuclear magnetic resonance. .

The RF / detection coil 34 may be composed of a coil for applying a high frequency pulse signal to a sample and a coil for detecting a relaxation resonance signal of the sample so as to be in charge of the transmission and reception functions, respectively. In the present invention, since the reception of the signal is impossible, the transmission of the high frequency pulse signal is impossible in the case of the reception of the resonance signal, and the installation difficulties and limitations are constrained. To be able.

The RF / detection coil 34 used in the present invention is wound to a space radius of 60 mm using a copper wire having a cross section radius of 2.54 mm, and the distance between the copper wires is 3.81 mm, which is 1.5 times the cross section radius, and the total number of turns is 14 times. In this case, the inductance (L) of the coil 34 was calculated to be 7.6 μH by equation (3).

Here, Lg, the length of the coil, is 910 mm, the cross-sectional area A of the coil inner space is about 11,204 mm2, and the value obtained by setting the permeability of free space μ to 4π10-7wb / mA. As a result of verifying with an impedance meter (LCR meter), it was shown as 7.9μH at 120μ and 7.7μH at 1㎑.In theory, L value is independent of frequency, but L value increases exponentially with increasing frequency. It tends to decrease.

In the present invention, when the air gap 10 is set to 152.4 mm and the magnetic field strength of the permanent magnets 6 and 8 is set to 965gauss, a high frequency AC signal for obtaining a resonance signal is calculated at 4.1 MHz by the following equation (4). ,

In order to construct a circuit tuned at 4.1 MHz, the strength of the magnetic field generated in the RF / detection coil 34 should be calculated.

In order to obtain a resonance signal from a sample magnetized by the permanent magnets 6 and 8, the weak magnetic field B ₁ , which changes at 4.1 MHz, is passed through the permanent magnet 6 and 8 through the RF / detection coil 34. ) from the magnetic field (B _0), and should form a right angle, and B ₁ to walk a magnetic field by giving the sample vector of (M _0), the degree which becomes from the Z-axis inclined in the XY plane is inclined (θ) is the following formula (6) Is given.

On the other hand, in order to obtain the magnetic field strength (B ₁ ) required for nuclear magnetic resonance from the RF / sense coil 34 it should be able to apply a sufficient current to the RF / sense coil 34 to apply to the RF / sense coil 34 The high frequency pulse signal is amplified to a sufficient intensity to form a high frequency field (RF field).

The power P = I ² R applied to the RF / sensing coil 34 above, I is the current, R is the total impedance including the coil 34 and the leads and should be amplified from several hundred W to several KW through the amplifier. .

In addition, when the air gap 10 of the sensor 2 of the present invention is 152.4 mm and the intensity of the magnetic field is 965 gauss, the frequency of the high frequency pulse signal for obtaining the resonance signal is calculated as 4.1 MHz, as described above. The resonant circuit for resonating the sense coil 34 at 4.1 MHz has a resistor R-variable capacitor AVC-RF / sense coil 34 connected in series to form a series resonant circuit as shown in FIG. Ideally derived equations in the series resonant circuit are shown in Table 9 below.

11 is a circuit block diagram of the measuring apparatus of the present invention. The pulse programmer 48 which controls the transmission and reception duty cycle of the transmission and reception is connected to the computer 46 which is in charge of controlling the apparatus and analyzes the resonance signal.

The transmit / receive operation period is determined by the H pulse of about 10μs and the L pulse of about 100ms, and the transmitting / receiving time is determined. Nuclear magnetic resonance is applied to the coil 34 to generate nuclear magnetic resonance. During the L period of the signal, the resonance signal of the sample is detected by the RF / detection coil 34, detected, and then signal processed to obtain a desired moisture resonance signal. Of course, the reception operation cycle can be appropriately changed depending on the target of the EUT and the desired signal.

In addition, a high frequency switching unit (RF switching section) 50 is connected to the output of the pulse programmer 48 to generate a high frequency pulse signal of 4.1 MHz during the period of operation H of the pulse programmer 48. A high frequency power amplifier 52 is connected to the output of 50 to amplify the 4.1 MHz high frequency pulse signal to generate a magnetic field B of sufficient magnitude for nuclear magnetic resonance through the RF / detection coil 34.

A transmitter / receiver (T / R switch or duplexer: 42) is connected to the output of the high frequency power amplifier 52 so that the transmission / reception state of the RF / detection coil 34 can be switched. 42 is connected to the RF / sensing coil 34 of the sensor 2, and a pre-amplifire 54 is connected to one end of the transmitter / receiver switch 42 to sense it from the RF / sensing coil 34. Low noise amplification of the resonance signal.

The low-noise preamplifier 54 suppresses amplification of unwanted noise in the signal, amplifies the resonance signal into a large signal because the resonance signal input from the RF / detection coil 34 is a few μV, and the front end for signal processing. As a particularly important circuit portion.

Since the signal amplified by the preamplifier 54 is a mixed signal of a resonance signal and a 4.1 MHz carrier, the receiver and the detector 56 are connected to the output of the preamplifier 54, so that several pure signals, which are necessary signals among the mixed signals, are required. Obtain a moisture resonance signal.

A digital oscilloscope (58) is connected to the output of the receiver and detector 56 to display the pure water resonance signal processed, and a high frequency switch 50 is provided at one end of the receiver and detector 56. Operation of the receiver when the reference frequency of 4.1 MHz is input to remove the 4.1 MHz carrier signal from the connected and received signals, and the pulse programmer 48 is connected to one end of the receiver and detector 56 in the transmission state. This can be blocked to protect the receiver.

The oscilloscope 58 has a built-in digitizer (60), converts pure water resonance signals input in an analog form into digital signals, detects digital XY coordinates, and then uses the image display 62 as an XT coordinate information. Output the signal.

In addition, the computer 46 connects the digitizer 60 so that the pure water resonance signal converted into a digital signal that can be recognized by the computer is input so that the signal analysis and signal processing on the computer 46 can be performed. In addition, the reception and detection unit 56 is connected to the signal output unit of the computer 46 so that the signal analyzed and processed by the computer 46 is displayed on the oscilloscope 58. The signal analyzed and processed by the computer 46 may be output not only through the oscilloscope 58 but also through an output device such as a monitor and a print separately connected to the computer 46.

The present invention has a structure in which the entire system is controlled by the computer 46. The high frequency pulse signal generated by the computer program has a pulse width of about 10 μs and a delay time of 100 ms. Therefore, the whole system must have a duty cycle to satisfy this, and a signal-to-noise ratio. It should be possible to make several resonant signals in order to increase the value of.

When the pulse programmer 48 generates a gate signal and sends the gate signal to the high frequency switching unit 50, in the high frequency circuit, the gate signal of the pulse programmer 48 and the Larmor Precession frequency of 4.1 MHz are mixed with the gate signal. To generate a signal, that is, to generate a sinusoidal wave of 4.1 MHz only while the gate signal is present.

The pulse programmer 48 generates a pulse having only a constant width, and two signals are generated by the high frequency switching unit 50, which is a gate signal and a high frequency signal of 4.1 MHz.

Since the signal output from the high frequency switching unit 50 is within several Watts and this signal cannot form a high frequency field, the signal output should be amplified. The part responsible for this function is a high frequency power amplifier ( 52). The amplified signal has an average of 800 to 1000 watts. The amplified signal enters the RF / detection coil 34 to form a high frequency field that forms a magnetic field perpendicular to the static magnetic field in the sample placed in the static magnetic field.

In the past, the NMR apparatus had a part for sending this signal to the RF / sensing coil 34 and a part for detecting a resonance signal. Therefore, two devices for sending and receiving a signal to a sample were required, but the RF / The sensing coil 34 simplifies the device by sending a high frequency field signal and receiving a resonance signal again.

Therefore, there should be a device capable of sending the high frequency pulse signal generated by the high frequency power amplifier 52 to the RF / detection coil 34 and again receiving the resonant signal, which is the transmitter / receiver switch 42. The transmitter / receiver 42 must separately obtain a signal due to resonance following a high frequency signal using a built-in damping circuit.

After the resonance signal is transmitted, a signal generated as a sample is detected by the RF / detection coil 34 and the signal is transmitted to the transmitter / receiver 42. The signal from the transmitter and receiver 42 is input to the preamplifier 54. The strength of the signal input to the RF / sense coil 34 is a small signal within several μV and amplified to a few tens of m Volts through the preamplifier 54.

Since the signal amplified by the preamplifier 54 is a mixed signal of a 4.1 MHz high frequency pulse signal and a sample resonance signal applied to the RF / detection coil 34, the 4.1 MHz signal is separated from the receiving and detecting unit 56. After removal, only pure resonance signals are obtained.

Since the signal obtained above is an analog signal, it must be converted into a digital signal that can be recognized by the computer 46 using the digitizer 60. The digitizer 60 is satisfied if it has a sampling rate more than twice the frequency generated finally for signal analysis, and a digitizer having a sampling rate within several MHz is usually used.

In the signal generation method used in the present invention, a free induction decay (FID) signal using a resonance signal is used as shown in FIG. 12. That is, as described above, the width of the gate signal generated by the pulse programmer 48 is fixed and then the signal is detected after the width ends. The width of the signal is defined by the electrons of the hydrogen nucleus with a constant angle (usually 60 ° to 90 °).

In FID, the signal is usually generated by 90 ° to obtain the signal for the sample, but in the actual experiment, the signal was obtained after having the slope of 40 ° to 60 °, and the signal was easily analyzed within 60 °. . In the present invention, the pulse width was 6.5 μs to have a slope of 60 °, and the delay time between the pulses was 100 ms.

Experimental example and the results of measuring the moisture content of agricultural products and the sweetness of fruits and vegetables using the present invention configured as described above are as follows.

Experimental Example 1

Moisture Content Measurement of Rice

end. Sample production

Ten springs were prepared by varying the water content of rice varieties of rice, and the sample production method was as follows. First, water is added to a container such as an air-blocking bag or a bottle, and then stored for 7 days in a closed space at 40 ° C. so that the rice slowly absorbs moisture, thereby making the overall moisture content of the rice uniform. The sample taken from the container removes moisture absorbed in the ear using a device such as desiccant. In the present invention, the sample is spread on a newspaper for about 1 hour to remove moisture by absorbing it in the ear.

Remove the sample from the desiccator and place it in a test tube of 30 mm to make a sample. Samples were prepared and dried at a temperature of 105 ° C. for 5 hours to measure the moisture content of each sample at the same time. The moisture content of each sample measured by the conventional resistance measuring apparatus and oven method is shown in Table 10.

The results of the measurements showed a very different value, which is considered to be due to the error of these simple measuring instruments and the influence of moisture not completely removed from the ear.

On the other hand, the sample produced by the above process was obtained the moisture resonance signal using the NMR apparatus of the present invention, by verifying the porosity of the device by obtaining the data twice each. First, the entire system was turned on, and then the vacuum variable capacitor (AVC) was tuned and tuned to adjust the tuning circuit to 4.1 MHz, and then a moisture resonance signal was obtained.

I. Signal acquisition

The signal obtained through the moisture resonance signal is shown in FIG. 13, which is data obtained from a sample having a water content of 20.4%. As shown in the drawing, the obtained signal cannot be used as it is, and an envelope must be obtained by signal processing.

FIG. 14 is an envelope waveform diagram obtained from the data of FIG. 13, and FIGS. 15, 16 and 17 are envelope waveform diagrams obtained from six rice samples having different moisture contents.

18 is a graph showing the correlation between the moisture content and the maximum pack. The coefficient of determination is 0.904. 19 shows the relationship between the arithmetic sum of the four points after the maximum value and the water content, and the coefficient of determination is 0.942. 20 shows the relationship between the arithmetic at four points after the maximum value and the water content, and the coefficient of determination is 0.947.

All. Result analysis

The 18th, 19th, and 20th degrees are obtained by obtaining a correlation between the arithmetic sum of the intensity after the FID maximum value and the water content. ) Were 0.904, 0.942, and 0.947, and these coefficients showed a good correlation with the ambient noise and measurement error.

Based on the above results, it is possible to continuously measure the yield of agricultural products by a relatively simple method.

Experimental Example 2

Measurement of sugar content of fruits and vegetables

end. Sample production

As the fruits and vegetables increase in ripening, the total sugars increase as monosaccharides such as starch, fructose and glucose, and sucrose, disaccharides, increase. Most fruits and vegetables are 80-90% water, mono- and disaccharides are usually 10-20%, and some other fiber. Commonly speaking sugar means sucrose.

Sucrose is composed of a combination of monosaccharides, fructose and glucose, and has 22 hydrogen atoms, and 14 sugar samples were prepared for preliminary experiments on fruits and vegetables. Glycerol sample was used to set the optimal resonance state. Experimental samples were made with evenly distributed distilled water in a 30 mm diameter glass tube and by varying the mass of sugar (only pure sucrose in the experiment).

The sugar content for the sugar water used in the experiment was pure distilled water, 3.1% sugar water, 5.8% sugar water, 6.9% sugar water, 8.7% sugar water, 10.5% sugar water, 11.9% sugar water, 15.7. % Sugar water, 17.6% sugar water, 18.9% sugar water, 19.7% sugar water, 22.2% sugar water, 22.5% sugar water, 26.8% sugar water and glycerol.

I. Preliminary Experiment with Glycerol

Experiments on glycerol were performed first. After the entire system was turned on, the tuning was performed by adjusting the vacuum variable capacitor (AVC) to set the tuning circuit to 4.1 MHz, and then obtaining a signal.

Checking for tuning can be done with a measuring instrument such as a frequency counter, Q-meter (quality factor meter) or standing-wave-detector. It was regarded as sympathetic when the signal of.

The equipment for acquiring the experimental data was a digital oscilloscope (58: PH33IPA model of PHILIPS) and obtained 8,192 data with a time base of 10μs to obtain a FID resonance signal of 1.6ms.

Since the signal obtained once is a weak signal to noise, one signal cannot be used and a plurality of signals are averaged on the digital oscilloscope 58. The average method is the old data as shown in Equation (7) below. ), The method of adding the influence of the new input signal (New data) was selected.

In the above, the average factor value is set to 128, and the signal is processed on the digital oscilloscope 58. The number of averaged data is 50 or more signals, and most of the noise can be removed by the average.

21 shows a signal for glycerol in a 1.4 ms period, which appears after about 40 μs. Signals before 40 μs are divided into two classes: the input signal for FID (6.5 μs) and the standby state until the RF / detection coil 34 detects a signal derived from the sample after the input of the signal is cut off. to be. This wait time contains a lot of information, so shortening this time improves the overall system performance.

The acquired signal decreases exponentially and exhibits a pattern of oscillating signal. The value of this damping degree is T ₂ obtained under the uniformity of the given magnetic field, and is obtained as in the following equation (8).

Where A is the output signal, A ₀ is the maximum value of the output signal, T ₂ is the inherent value of the object, and the value varies slightly depending on the uniformity of the magnetic field. Analyzing this value can measure the sugar content and water content of fruits and vegetables, but in the case of semi-solid state in which water and sugar are mixed like fruits and vegetables, two or more T ₂ values are observed and the magnetic field of permanent magnets (6) (8) B ₀ ) The higher the uniformity, the more accurate T ₂ can be measured.

Since the signal obtained above is a signal that vibrates with time, the envelope as shown in FIG. 22 is obtained without using the signal as it is. As shown in the figure, the signal of glycerol shows a form of attenuation at a high initial stage. This is a sign of completion, which means that most signals are present before 800µs.

Experiments with glycerol showed that in order to obtain and analyze meaningful signals, signals before 800 μs must be detected. However, this value can be extended beyond 800 μs by either increasing the uniformity of the magnetic field in the permanent magnets (6) (8), or by making the RF / detection coil (34) more precise to increase the signal granularity or to exactly match the impedance. have.

All. Experiment and analysis of sugar water

After setting the environment of the whole system with glycerol, experiments were conducted on 14 sugar waters, and 23, 24, and 25 degrees are graphs of envelopes of 6 sugar water samples including glycelo.

As shown in 23, 24, and 25 degrees, after 800 μs, high and low sugar samples are attenuated to the same value. Pure water is slowed while high sugar samples are attenuated fast. Thus, T ₂ of pure water does not appear larger than the high sugar sample. This means that the hydrogen atoms in the high sugar sample are less active than in pure water.

Several simple methods have been attempted to easily predict sugar content. For example, the sugar content was predicted by the slope of two points on the envelope without obtaining T ₂ for sugar water.

Fig. 26 shows the relationship between the slope and the sugar content for two points, and sugar can be easily measured for sugar water by using a linear model in a simple manner.

la. Experimental Analysis and Results for Green Water

Fruits and vegetables used as experiments are shown in Table 11 below.

Figure 27 compares the envelope of water, apple, and kiwi, and the signal shape is different for each object, and liquid water has a relatively long decay period compared to semi-solid apple or kiwi.

(A) Experimental analysis and results for apples

Thirty apple samples were made and made with the same mass (25.5g ± 0.2g) to remove any deviation from the mass. The sugar content was measured twice with a digital refractive sugar meter (TRM-100). Values were averaged of two values. When the sugar content of apple samples is distributed within 10.6% to 15.95%, the sugar distribution is as follows. 10% to 11%: 1, 11% to 12%: 1, 12% to 13%: 4, 13% to 14%: 5, 14% to 15%: 8, 15% to 16% : 11 pieces of fruit with a total sugar content of 14% to 16%.

Since apples are semi-solid, unlike sugar water, they used envelope data directly, rather than simply using two-way gradients, such as signal processing in liquid samples such as sugar water.

The predictive model was developed by multiple linear regression using statistical processing program (SAS), which found correlation coefficient between envelope and sugar using 20 data. Decision coefficient R of the predictive model when 18 coefficients are used = 0.931.

However, the results of verifying the sugar content for the remaining 10 samples using the developed multilinear regression model are as shown in Figure 28. In the case of apples with a standard deviation of 2.4% and a sugar range of 10% -16% It was concluded that the sugar prediction using the statistical processing program (SAS) was not valid because it was difficult to classify it into 2 grades if it was determined by measuring the sugar content using this method.

Therefore, the resonance signals of 30 apples were predicted after learning using neural networks. The input coefficients were taken in sugar water, taking the slope of the two points as the initial value and the slope of the neighboring values, the number of hidden nodes in the hidden layer, and the output value in one sugar content path. The number of input patterns was 20 and the remaining 10 were verified.

FIG. 29 shows the results obtained by taking 20 samples from 30 sample data based on the results learned by the neural network, and the possibility of classifying grades using the neural network was predicted based on the verified results. .

Experimental results for apples are as follows.

(A) Finding, predicting, and verifying correlation coefficients using the statistical processing program (SAS) shows that it is not appropriate to classify sugars in a relatively simple linear equation. This is because the RF signal in the surrounding space generates an induced current in the coil to generate noise. Therefore, it is necessary to completely block the RF / sense coil 34 and the tuning circuit from external noise.

(B) The neural network was used to predict and predict the slope and the actual sugar value. The predicted standard error was 0.565%. The measurement error was 1.13% with a reliability of about 95%.

(C) Using the above results, it is judged that apples can be sorted to grade 3 when the sugar content is generally in the range of 10% to 17%.

(B) Experiment and analysis on embryo

Twenty fold samples were made and each mass was made into 40.4g ± 0.2g to reduce the error caused by mass.Also, the sugar content was measured twice with a digital refractive sugar meter (TRM-100) and the average value was measured. The total sugar content was lower than apples, ranging from 9.8% to 13.1% (Brix%), and the sugar content was as follows. 9.8% ~ 10.0%: 3, 10% ~ 11%: 6, 11% ~ 12%: 6, 12% ~ 13.1%: 5 and more evenly distributed sugar than apples.

A neural network was used to train the sample and verified using the learned results. The number of samples was small to 20, and 12 samples were trained. The result of verifying with 8 samples is shown in FIG. The coefficients used in the neural network used seven coefficients extracted from the statistical processing program (SAS) as inputs, seven nodes in the hidden layer, and one sugar value for the output.

Experimental results for the ship are as follows.

(A) The standard error of the result verified by the result learned by the neural network is 0.392%. That is, the measurement error is within 0.78% with 95% confidence.

(B) The mass was larger than the apple sample, which resulted in better results. This is judged because the influence on the ambient noise is relatively small due to the increase of the resonance signal intensity as the mass increases. Using the obtained result, it is judged that grade 3 can be judged according to sugar content.

(C) Experiment and analysis of kiwi

13 kiwi samples were made and the whole was used as a sample instead of cutting out a portion such as an apple or pear. Since the kiwi is longer than the unidirectional direction and has a diameter of about 50 mm in the unidirectional direction, the whole can be put into the RF / sensing coil 34 and tested. The diameter of the manufactured RF / detection coil is 60 mm. The sugar content was measured by one time using a digital refractive sugar meter (model name: TRM-100) and distributed over 8.7% to 12.2%. The sugar distribution was as follows. 8.7% to 9%: 1, 9% to 10%: 3, 10% to 11%: 5, 11% to 12.2%: 4 and lower in sugar than apples.

The kiwi mass is evenly distributed from 55 g to 83 g. In order to find the significance between the mass of the kiwi and the resonance signal, the correlation between the intensity (maximum value of the resonance signal) and the mass of the entire signal was analyzed. (R ) Was 0.902.

The results indicate that the mass can be estimated using the intensity of the FID signal, since the intensity of the resonance signal of the total hydrogen atoms is proportional to the number of hydrogen atoms in the sample.

We tried a method of predicting sugar content using a neural network learning model for kiwis, using eight samples for prediction, and using five samples for verification. Is as follows.

(A) The correlation between the maximum value and mass in the FID signal was obtained. The correlation was 0.902 and showed that the mass can be determined by the intensity of the signal by simple analysis of the resonance signal.

(B) The neural network predicted the sweetness of the kiwi. The predicted standard error is 0.415, and the predicted results will allow the kiwi sugar to be graded in three levels, as in the case of apples and pears.

(D) Experiment and analysis on kumquat

There are 35 samples of kumquats. Like a kiwi, kumquat is a fruit or vegetable smaller than the diameter of the tuning coil, and its diameter is within 20 mm to 30 mm so that it can be put into the RF / sensing coil 34 as a whole. The mass of kumquat is evenly distributed from 10.390g to 18.645g, and the sugar content is distributed from 11.2% to 19.7%. 11.2%-13%: 2, 13%-14%: 1, 14%-15%: 11, 15%-16%: 10, 16%-17%: 7, 17%-18% : 1 piece, 18%-19.7%: 3 pieces.

Prediction experiments were performed by neural networks. The number of input nodes was 10 and the number of correlated data obtained using the statistical processing program (SAS) was input. The number of nodes in the hidden layer was 5 and the output was 1 sugar. In FIG. 33, 25 samples were trained and 10 samples were used to compare sugar with actual measurements. The standard error, which is a measure of the difference between the predicted sugar value and the actual measured value, was 1.461.

The results of the experiment are as follows.

(A) By the neural network model, sugar was predicted using 10 coefficients. The SEP was 1.461 and the reason for the error was rather large. It was considered that the noise was weak due to the weak resonance signal due to the small mass of orange.

(B) It is judged that a classification of about 2 grades is possible using the results given.

(C) It is judged that this small fruit can achieve better results by increasing the strength of the magnetic field and improving the uniformity of the magnetic field by reducing the pores 10 of the magnet box.

In the present invention, the corner steel 24 is formed in a triangular shape, but when the inclined surface (45 ° combed surface) is formed in a concave recessed concave shape as in FIG. 34, the flow of the magnetic field is smooth and the uniformity is increased.

In the present invention, a large amount of heat is generated by the high-frequency high-frequency pulse signal, but the application time of the high-frequency pulse is very short, 10 μs, and heat dissipation is performed through the case (4). ), The diameter of 120mm or 60mm, but if the size of the sample is large, the diameter of the RF / sensing coil 34 may be increased and the strength of the permanent magnets 6 and 8 may be increased.

In the present invention, the non-destructive test subjects, the water content of the agricultural product or the measurement of sugar content of the fruits and vegetables as an example, using the same method can also measure the oil (edible or non-edible oil) content of plants such as sesame, perilla, castor, castor and the like.

As described above, the present invention has the effect of using non-destructive and continuously measuring the moisture content and sugar content of agricultural products using practical hydrogen nuclear magnetic resonance.

Claims (3)

Hydrogen nuclear magnetic resonance (NMR) is obtained by processing a resonance signal obtained by putting a sample into a resonance sensor of a NMR device to obtain a pure water resonance signal or a sugar signal to continuously measure the moisture content or sugar content of agricultural products. In the agricultural non-destructive testing device used; The plate-shaped permanent magnets 6 and 8 are fixed to the inner upper and lower surfaces of the box-shaped case 4 made of a material with high permeability so that the N and S magnetic poles face each other. Forming voids 10, attaching high-permeability pole faces 16 and 18 to opposing surfaces of permanent magnets 6 and 8, and permeability to opposite edge portions of pole faces 16 and 18 The high shearing plates 20 and 22 are attached and fixed, and the high permeability corner steel 24 is attached and fixed to the inner edge of the case 4 to improve the magnetic field strength and uniformity of the static magnetic field B ₀ . Magnet box for agricultural non-destructive testing using hydrogen nuclear magnetic resonance (NMR) to improve the strength and uniformity of the magnetic field by installing an RF / sensing coil 34 connected to the controller (computer) in the space (10). The transmitter / receiver according to claim 1, wherein the transmitter / receiver 42 is connected to one end of the magnet box 2 obtained by applying a high frequency pulse to the sample and then switched to the transmitter / receive state. A preamplifier 54 for amplifying the resonance signal detected by the sensor 2 from the sample is connected to one end of the sample 42, and the receiver and detector 56 are connected to the output of the preamplifier 54 to connect the preamplifier ( Amplifying and detecting the signal amplified from 54) to obtain a pure moisture resonance signal of sufficient magnitude, and connecting the oscilloscope 58 to the output of the receiving and detecting unit 56 so that the pure moisture resonance signal is displayed on the image. The high frequency power amplifier 52, the high frequency switching unit 50, the pulse programmer 48, and the computer 46 for controlling the system are connected to one end of the transmitter / receiver 42 in turn, and the computer 35 and the pulse programmer are connected. 48 and the high frequency switching unit 50 to the receiving and detecting unit 56 Non-destructive testing device using hydrogen nuclear magnetic resonance (NMR) to improve the intensity and uniformity of the magnetic field by interrupting the reception during the transmitter to which the high frequency pulse is applied to the RF / sense coil (34) by connecting each. The magnet box for agricultural product non-destructive inspection using hydrogen nuclear magnetic resonance (NMR) with improved magnetic field strength and uniformity, characterized in that the inclined surface of the corner steel 24 is concave.