Terahertz time-domain spectroscopy-based tobacco shred blending proportion nondestructive testing method

Abstract

The invention discloses a Terahertz time-domain spectroscopy-based tobacco shred blending proportion nondestructive testing method. The method comprises the following steps: respectively preparing single-component tobacco shred samples of cut leaves, cut stems and cut thin sheets; constructing a plurality of groups of tobacco shred mixing systems according to a preset mass ratio gradient to obtain a plurality of groups of mixed tobacco shred samples; obtaining terahertz time-domain spectroscopic data of single-component tobacco shred samples of the cut leaves, the cut stems and the sheet shreds and each group of mixed tobacco shred samples; according to the spectral response difference of the terahertz time-domain spectral data of each group of mixed cut tobacco samples and the single-component cut tobacco samples, constructing a three-cut-tobacco proportion inversion model based on partial least square regression; obtaining terahertz time-domain spectroscopic data of the cut tobacco to be detected; and obtaining the blending proportion of the tobacco shreds to be measured according to the terahertz time-domain spectrum data of the tobacco shreds to be measured and the three-shred proportion inversion model. According to the tobacco shred blending proportion nondestructive testing method based on the terahertz time-domain spectroscopy, nondestructive analysis of the cigarette structure and the tobacco shred proportion can be achieved.

Description

Technical Field

[0001] The present invention relates to the technical field of tobacco detection, and more specifically, to a non-destructive detection method for the blending ratio of cut tobacco based on terahertz time-domain spectroscopy. Background Art

[0002] The quality control of tobacco products is the core link to ensure the sensory quality of products and consumer satisfaction. Among them, the blending ratio and uniformity of the "three kinds of cut tobacco" (leaf cut tobacco, stem cut tobacco and reconstituted tobacco sheet) in cigarettes directly affect the combustion characteristics, smoke release behavior and batch consistency of cigarettes. During the process of optimizing the formulation process in domestic cigarette enterprises, by accurately regulating the blending ratio of the three kinds of cut tobacco and establishing the quantitative effect relationship between the blending amount and smoke release indexes (tar content, nicotine transfer rate, etc.), the coordinated improvement of the physical and chemical stability and consumption comfort of cigarette products can be achieved.

[0003] At present, the production line mainly uses the method of manual sorting to detect the blending ratio of the three kinds of cut tobacco, which is time-consuming and highly subjective.

[0004] Therefore, there is an urgent need for a non-destructive detection method for the blending ratio of cut tobacco based on terahertz time-domain spectroscopy. Summary of the Invention

[0005] The purpose of the present invention is to provide a non-destructive detection method for the blending ratio of cut tobacco based on terahertz time-domain spectroscopy to solve the above problems in the prior art and realize the non-destructive analysis of the cigarette structure and cut tobacco ratio.

[0006] The present invention provides a non-destructive detection method for the blending ratio of cut tobacco based on terahertz time-domain spectroscopy, which includes:

[0007] Prepare single-component cut tobacco samples of leaf cut tobacco, stem cut tobacco and reconstituted tobacco sheet respectively;

[0008] Construct multiple groups of cut tobacco mixing systems according to the preset mass ratio gradient to obtain multiple groups of mixed cut tobacco samples;

[0009] Obtain the terahertz time-domain spectroscopy data of the single-component cut tobacco samples of leaf cut tobacco, stem cut tobacco and reconstituted tobacco sheet and each group of mixed cut tobacco samples;

[0010] Based on the spectral response differences between each group of mixed cut tobacco samples and the single-component cut tobacco samples, construct an inversion model for the ratio of the three kinds of cut tobacco based on partial least squares regression;

[0011] Obtain the terahertz time-domain spectroscopy data of the cut tobacco to be measured;

[0012] According to the terahertz time-domain spectroscopy data of the cut tobacco to be measured and the inversion model for the ratio of the three kinds of cut tobacco, obtain the blending ratio of the cut tobacco to be measured.

[0013] The non-destructive detection method for the blending ratio of cut tobacco based on terahertz time-domain spectroscopy as described above, wherein, preferably, the preparation of single-component cut tobacco samples of cut filler, cut stem and cut lamina respectively includes:

[0014] Perform low-temperature drying pretreatment on cut filler, cut stem and cut lamina respectively;

[0015] Prepare cut tobacco test samples of cut filler, cut stem and cut lamina by using the simple sandwich encapsulation method.

[0016] The non-destructive detection method for the blending ratio of cut tobacco based on terahertz time-domain spectroscopy as described above, wherein, preferably, the low-temperature drying pretreatment of cut filler, cut stem and cut lamina respectively includes:

[0017] Place cut filler, cut stem and cut lamina in an oven, and use dry hot air at 40°C - 50°C for ventilation drying for 3h - 5h until constant weight for standby,

[0018] The preparation of cut tobacco test samples of cut filler, cut stem and cut lamina by using the simple sandwich encapsulation method includes:

[0019] Weigh cut filler, cut stem and cut lamina respectively, and control the total mass to be 4.000 ± 0.005 g;

[0020] Place the three kinds of cut tobacco between two clean glass slides respectively;

[0021] By gently tapping the edge of the glass slide and fine-tuning the horizontal rotation, make the cut tobacco naturally spread out in the central area;

[0022] After visually confirming no significant agglomeration or voids, seal and fix along the periphery of the glass slide with a low-dielectric polyester film tape to form a sandwich sample with a thickness of 0.3mm - 0.5 mm.

[0023] The non-destructive detection method for the blending ratio of cut tobacco based on terahertz time-domain spectroscopy as described above, wherein, preferably, the construction of multiple groups of cut tobacco mixing systems according to a preset mass ratio gradient to obtain multiple groups of mixed cut tobacco samples includes:

[0024] Prepare 10 single-component samples of cut filler, cut stem and cut lamina respectively, and control the mass of each to be 0.200 ± 0.001 g;

[0025] Construct three mixing systems according to a preset mass ratio gradient to obtain glass slide specimens corresponding to mixed sample 1, mixed sample 2 and mixed sample 3 respectively. Among them, the three-wire ratios of cut filler:cut stem:cut lamina corresponding to mixed sample 1, mixed sample 2 and mixed sample 3 are: 1:1:8, 3:2:5 and 1:7:2 respectively.

[0026] The non-destructive detection method for the mixing ratio of cut tobacco based on terahertz time-domain spectroscopy as described above, wherein, preferably, constructing three groups of mixed systems according to a preset mass ratio gradient, and respectively obtaining the glass slide specimens corresponding to the mixed sample 1, the mixed sample 2 and the mixed sample 3, includes:

[0027] Preparing 10 parallel samples for each group of the mixed systems according to the corresponding mixing ratios;

[0028] Using an oscillator to uniformly mix the 10 parallel samples corresponding to each group of the mixed systems;

[0029] Quantitatively clamping the cut tobacco mixture in each group of the mixed systems by using tweezers, spreading it flat between the glass slides, regulating to form a uniformly distributed area by the tapping-rotation method, and sealing and packaging it with a low-loss polytetrafluoroethylene tape to make a standardized sandwich detection unit with a thickness of 0.25±0.05 mm.

[0030] The non-destructive detection method for the mixing ratio of cut tobacco based on terahertz time-domain spectroscopy as described above, wherein, preferably, obtaining the terahertz time-domain spectroscopy data of the single-component cut tobacco samples of cut filler, cut stem and cut lamina and the cut tobacco samples of each group of mixed cut tobacco, includes:

[0031] Using a terahertz time-domain spectroscopy system to detect the single-component cut tobacco samples of cut filler, cut stem and cut lamina and the cut tobacco samples of each group of mixed cut tobacco to obtain the terahertz time-domain spectroscopy data, wherein the scanning frequency spectrum range is 0-3 THz, the spectral resolution is 12.5 GHz, the average number of scans per time is 1024 times, the test environment temperature is 20.5℃-21.5℃, and the environmental relative humidity is 3%-5%.

[0032] The non-destructive detection method for the mixing ratio of cut tobacco based on terahertz time-domain spectroscopy as described above, wherein, preferably, constructing an inversion model for the three-cut ratios based on partial least squares regression according to the spectral response differences between the cut tobacco samples of each group of mixed cut tobacco and the single-component cut tobacco samples, includes:

[0033] Preprocessing the terahertz time-domain spectroscopy data of the cut tobacco samples of each group of mixed cut tobacco and the single-component cut tobacco samples;

[0034] According to the preprocessed terahertz time-domain spectroscopy data of the cut tobacco samples of each group of mixed cut tobacco and the single-component cut tobacco samples, obtaining the terahertz optical parameters of the cut tobacco samples of each group of mixed cut tobacco and the single-component cut tobacco samples;

[0035] According to the spectral response differences of the terahertz optical parameters of the cut tobacco samples of each group of mixed cut tobacco and the single-component cut tobacco samples, based on the principle of linear superposition of optical parameters, constructing the following linear spectral mixture model:

[0036]

[0037] Among them, represents the optical parameter vectors of the mixed samples with unknown doping ratios at characteristic frequency points, and the optical parameter vectors represent the absorption coefficient, refractive index, and dielectric constant. It is expressed as

[0038] ,

[0039] represents the standard spectral matrix of the pure components of cut tobacco, stem cuttings, and reconstituted tobacco sheets, and is expressed as

[0040] ,

[0041] represents the doping ratio matrix vector of cut tobacco, stem cuttings, and reconstituted tobacco sheets in tobacco leaves, and is expressed as

[0042]

[0043] is the proportionality coefficient to be solved. Among them, , , respectively represent the mass fractions of cut tobacco, stem cuttings, and reconstituted tobacco sheets, and , represents the measurement noise;

[0044] For the system of linear equations of multiple groups of mixed tobacco samples, the following formula is used to solve it by the constrained least squares method:

[0045] ,

[0046] The active set algorithm is used for iterative solution, and the convergence threshold is set to to obtain the quantitative inversion result of the blending ratio of tobacco.

[0047] For the non-destructive detection method of the blending ratio of tobacco based on terahertz time-domain spectroscopy as described above, preferably, the terahertz time-domain spectroscopy data of each group of mixed tobacco samples and single-component tobacco samples are preprocessed, including:

[0048] Perform 5-layer discrete wavelet decomposition using the Symlet8 wavelet basis function, and combine the soft threshold method to suppress the high-frequency noise in the time-domain waveform;

[0049] Perform Savitzky-Golay smoothing and airPLS baseline correction on the frequency-domain signal;

[0050] Based on the characteristic absorption peak distribution and the system signal-to-noise ratio threshold, an effective frequency band of 0.1 - 1.5 THz is intercepted, the low-frequency distortion region below 0.1 THz and the high-frequency noise region above 1.5 THz are removed, and more than 95% of the effective spectral information is retained;

[0051] Normalize the absorption coefficient and the imaginary part of the dielectric constant through standard normal transformation.

[0052] The non-destructive detection method for the blending ratio of cut tobacco based on terahertz time-domain spectroscopy as described above, wherein, preferably, the terahertz optical parameters of each group of blended cut tobacco samples and single-component cut tobacco samples are obtained according to the preprocessed terahertz time-domain spectral data of each group of blended cut tobacco samples and single-component cut tobacco samples, including:

[0053] Determine the sensitive frequency band where the cut tobacco thickness has a linear response with the terahertz optical parameters through characteristic band screening, wherein the terahertz optical parameters include the absorption coefficient α(f), the refractive index n(f), and the dielectric constant ε(f);

[0054] Measure the time-domain spectral signals of each single-component sample in the sensitive frequency band respectively, and extract the standard optical parameters through fast Fourier transform;

[0055] For each group of blended cut tobacco samples, synchronously obtain the terahertz time-domain waveform in the sensitive frequency band, and calculate the optical parameter values of each group of blended cut tobacco samples.

[0056] The spectral response differences of the terahertz optical parameters of each group of blended cut tobacco samples and single-component cut tobacco samples include:

[0057] For single-component cut tobacco samples in the 0.5 - 1.1 THz frequency band, cut filler shows significant identification due to its high absorption coefficient in the 0.7 THz - 1.1 THz range and the dielectric constant characteristic peak at 0.55 THz; reconstituted tobacco has the best transmittance in the middle frequency band of 0.4 THz - 0.8 THz;

[0058] For blended cut tobacco samples, the absorption coefficient and the dielectric constant have a strong sorting relationship with the blending ratio in the characteristic frequency bands of 0.53 THz - 0.59 THz and 0.67 THz - 1.1 THz.

[0059] The non-destructive detection method for the blending ratio of cut tobacco based on terahertz time-domain spectroscopy as described above, wherein, preferably, the blending ratio of the cut tobacco to be measured is obtained according to the terahertz time-domain spectral data of the cut tobacco to be measured and the three-thread ratio inversion model, including:

[0060] Obtain the terahertz time-domain spectral data of the cut tobacco to be measured;

[0061] Based on the terahertz time-domain spectroscopy data of the tobacco to be measured, a terahertz optical parameter vector of the tobacco to be measured is obtained;

[0062] The terahertz optical parameter vector of the tobacco to be measured is input into the three-wire ratio inversion model to obtain the blending ratio of the tobacco to be measured.

[0063] The present invention provides a non-destructive detection method for the blending ratio of tobacco based on terahertz time-domain spectroscopy. Based on terahertz time-domain spectroscopy technology (THz-TDS), combined with the linear correlation between the absorption coefficient, dielectric constant and refractive index spectrum of tobacco in a specific band, a quantitative inversion method for the blending ratio of tobacco with multi-parameter fusion is proposed, realizing the rapid non-destructive detection of the blending ratio of three wires; the present invention provides a solution with high precision and high robustness for the non-destructive detection of tobacco blending in the tobacco industry. Description of the Drawings

[0064] To make the objectives, technical solutions and advantages of the present invention clearer, the present invention will be further described below with reference to the drawings, where:

[0065] Figure 1 It is a flowchart of an embodiment of the non-destructive detection method for the blending ratio of tobacco based on terahertz time-domain spectroscopy provided by the present invention;

[0066] Figure 2 It is the comparison of the terahertz time-domain spectroscopy of cut tobacco, stem cuttings and reconstituted tobacco with the reference sample;

[0067] Figure 3 It is the comparison of the terahertz frequency-domain spectroscopy of cut tobacco, stem cuttings and reconstituted tobacco with the reference sample;

[0068] Figure 4 It is the comparison of the terahertz absorption spectra of cut tobacco, stem cuttings and reconstituted tobacco;

[0069] Figure 5 It is the comparison of the dielectric constant spectra of cut tobacco, stem cuttings and reconstituted tobacco;

[0070] Figure 6 It is the comparison of the refractive index spectra of cut tobacco, stem cuttings and reconstituted tobacco;

[0071] Figure 7 It is the comparison of the terahertz time-domain spectroscopy of three groups of blended tobacco samples with the reference sample;

[0072] Figure 8 It is the comparison of the terahertz frequency-domain spectroscopy of three groups of blended tobacco samples with the reference sample;

[0073] Figure 9 It is the comparison of the terahertz absorption spectra of three groups of blended tobacco samples;

[0074] Figure 10 It is the comparison of the terahertz dielectric constant spectra of three groups of blended tobacco samples;

[0075] Figure 11 Comparison of terahertz refractive index spectra of three groups of blended cut tobacco samples. Detailed implementation manners

[0076] Various exemplary embodiments of the present disclosure will now be described in detail with reference to the accompanying drawings. The description of the exemplary embodiments is merely illustrative and in no way limits the present disclosure and its application or use. The present disclosure can be implemented in many different forms and is not limited to the embodiments described herein. These embodiments are provided to make the present disclosure thorough and complete, and to fully convey the scope of the present disclosure to those skilled in the art. It should be noted that: unless otherwise specifically stated, the relative arrangements of components and steps, the components of materials, numerical expressions and values set forth in these embodiments should be construed as merely exemplary and not as limitations.

[0077] The "first", "second", and similar terms used in the present disclosure do not denote any order, quantity, or importance, but are only used to distinguish different parts. Terms such as "comprising" or "including" mean that the elements preceding the term cover the elements enumerated after the term, and do not exclude the possibility of also covering other elements. Terms such as "upper" and "lower" are only used to represent relative positional relationships, and when the absolute position of the object being described changes, the relative positional relationship may also change accordingly.

[0078] In the present disclosure, when it is described that a specific component is located between a first component and a second component, there may or may not be an intermediate component between the specific component and the first component or the second component. When it is described that a specific component is connected to other components, the specific component may be directly connected to the other components without an intermediate component, or may not be directly connected to the other components and have an intermediate component.

[0079] All terms used in the present disclosure (including technical terms or scientific terms) have the same meaning as understood by those of ordinary skill in the art to which the present disclosure pertains, unless otherwise specifically defined. It should also be understood that terms defined in a general dictionary, such as those, should be interpreted as having a meaning consistent with their meaning in the context of the relevant art and should not be interpreted in an idealized or overly formal sense, unless specifically defined as such herein.

[0080] Technologies, methods, and devices known to those of ordinary skill in the relevant art may not be discussed in detail, but where appropriate, the technologies, methods, and devices should be regarded as part of the specification.

[0081] At present, the research on the structure of triple strands and the production control process is relatively in-depth. Through the relevant research on the physical quality indicators and harmful component release amounts of cigarettes under different blending ratios of triple strands, it provides technical support for the optimization of formula design. However, traditional detection methods (such as chemical analysis and physical separation and weighing) require destructive treatment of cigarette sticks, and there are bottlenecks such as long cycle and inability to perform in-situ analysis, making it difficult to meet the requirements of real-time quality control for intelligent production lines in the tobacco industry.

[0082] At present, the detection methods for tobacco formula components mainly include image method and spectroscopy method. The image method focuses on the recognition of appearance characteristics such as the particle morphology of different components. This method has poor recognition effect when the morphology and physical properties of tobacco shred components are similar. The spectroscopy method mainly focuses on near-infrared spectroscopy detection at present. This method does not require complex sample pretreatment (such as chemical dissolution) and can achieve rapid detection of tobacco shred components. However, the detection results are severely affected by the physical state of the sample (the density, particle size and moisture content of tobacco shreds).

[0083] In recent years, terahertz time-domain spectroscopy (THz-TDS) has shown non-destructive analysis advantages in the fields of biology and agricultural detection due to its low photon energy (~4.1 meV 1 THz), high penetration and fingerprint spectrum recognition ability. By emitting ultra-short pulse terahertz waves (0.1 - 10 THz) and analyzing the time-domain attenuation, frequency-domain absorption and phase shift signals, the physical properties (such as density, dielectric constant) and chemical composition information of the sample can be obtained simultaneously. In the field of security detection, terahertz waves can penetrate non-metallic materials such as plastics and papers. Combining its molecular vibration spectrum characteristics, it is used for drug / explosive detection (such as detecting the characteristic absorption peak of heroin at 1.4 THz) and non-destructive security inspection of mail parcels, avoiding the ionization hazards of X-rays. In the field of biomedicine, terahertz imaging technology has been used for early diagnosis of skin cancer (such as the dielectric constant difference of melanoma at 0.5 - 2.5 THz) and drug quality control (such as the detection accuracy of the thickness of the tablet coating layer reaching ±5μm) due to its high resolution (sub-millimeter level) and moisture sensitivity to biological tissues. In industrial manufacturing, terahertz time-domain spectroscopy is integrated into the production line to monitor the delamination defects of composite materials in real time (such as the debonding detection sensitivity of carbon fiber reinforced plastics > 90%) or the coating uniformity (such as the on-line measurement error of the thickness of automotive paint surface < 1μm). In addition, breakthrough progress has also been made in 6G communication (ultra-high-speed wireless transmission, theoretical rate up to 1 Tbps) and cultural heritage protection (non-destructive analysis of ancient painting pigment layers). The application of terahertz time-domain spectroscopy technology in cigarette stick structure and tobacco shred ratio has not been reported yet.

[0084] Such as Figure 1As shown, in the actual implementation process of the non-destructive detection method for the blending ratio of cut tobacco based on terahertz time-domain spectroscopy provided in this embodiment, the specific steps are as follows:

[0085] Step S1: Prepare single-component cut tobacco samples of leaf tobacco, stem tobacco, and reconstituted tobacco respectively.

[0086] Among them, the leaf tobacco, stem tobacco, and reconstituted tobacco used are all provided by China Tobacco Yunnan Industrial Co., Ltd., and the raw materials are directly taken from the cut tobacco production line. In an implementation manner of the non-destructive detection method for the blending ratio of cut tobacco based on terahertz time-domain spectroscopy of the present invention, the step S1 may specifically include:

[0087] Step S11: Perform low-temperature drying pretreatment on leaf tobacco, stem tobacco, and reconstituted tobacco respectively.

[0088] Specifically, place the leaf tobacco, stem tobacco, and reconstituted tobacco in a drying oven, and use dry hot air at 40°C - 50°C (for example, 45°C) for ventilation drying for 3h - 5h (for example, 4h) until constant weight for use, so as to eliminate the influence of excessive moisture on the absorption of terahertz waves.

[0089] Step S12: Prepare cut tobacco test samples of leaf tobacco, stem tobacco, and reconstituted tobacco by using the simple sandwich encapsulation method.

[0090] In an implementation manner of the non-destructive detection method for the blending ratio of cut tobacco based on terahertz time-domain spectroscopy of the present invention, the step S12 may specifically include:

[0091] Step S121: Weigh leaf tobacco, stem tobacco, and reconstituted tobacco respectively, and control the total mass to be 4.000 ± 0.005 g respectively.

[0092] Among them, the weighing accuracy is ±0.1 mg.

[0093] Step S122: Place the three kinds of cut tobacco between two clean glass slides respectively.

[0094] Among them, the size of the glass slide is 75×25×1 mm³.

[0095] Step S123: By gently tapping the edge of the glass slide and horizontally rotating and fine-tuning, make the cut tobacco naturally spread out in the central area.

[0096] Among them, the diameter of the central area is about 10 mm.

[0097] Step S124: After visually confirming that there is no significant agglomeration or void, seal and fix it along the periphery of the glass slide with a low-dielectric polyester film tape to form a sandwich sample with a thickness of 0.3mm - 0.5 mm.

[0098] Among them, the thickness of the low-dielectric polyester film tape is 50 μm, and the dielectric constant .

[0099] In the present invention, by making slide specimens to preserve the form of cut tobacco, and later removing the terahertz time-domain spectroscopy signals of the slides themselves through algorithms, the terahertz time-domain spectroscopy signals closest to those of cut tobacco on the production line can be obtained. Compared with the traditional pelletizing method, this method does not require complex pressing equipment, and the uniform distribution of cut tobacco can be achieved only through manual operation without changing the physical form of cut tobacco, and the influence of tape encapsulation on the transmittance of terahertz waves can be ignored (attenuation rate < 3% ).

[0100] Step S2: Construct multiple groups of cut tobacco mixing systems according to a preset mass ratio gradient to obtain multiple groups of mixed cut tobacco samples.

[0101] In an implementation manner of the method for non-destructive detection of cut tobacco blending ratio based on terahertz time-domain spectroscopy of the present invention, the step S2 may specifically include:

[0102] Step S21: Prepare 10 single-component samples of cut tobacco leaves, cut tobacco stems, and cut tobacco lamina respectively, with the mass of each controlled at 0.200 ± 0.001 g.

[0103] Step S22: Construct three groups of mixing systems according to a preset mass ratio gradient to obtain slide specimens corresponding to mixed sample 1, mixed sample 2, and mixed sample 3 respectively. Among them, the three-wire ratios of cut tobacco leaves: cut tobacco stems: cut tobacco lamina corresponding to mixed sample 1, mixed sample 2, and mixed sample 3 are 1:1:8, 3:2:5, and 1:7:2 respectively.

[0104] In an implementation manner of the method for non-destructive detection of cut tobacco blending ratio based on terahertz time-domain spectroscopy of the present invention, the step S22 may specifically include:

[0105] Step S221: Prepare 10 parallel samples for each group of the mixing systems according to the corresponding blending ratios.

[0106] Step S222: Use an oscillator to mix the 10 parallel samples corresponding to each group of the mixing systems evenly.

[0107] Step S223: Use tweezers to quantitatively pick up cut tobacco mixtures in each group of the mixing systems, lay them flat between glass slides, regulate the uniformly distributed area through the tapping-rotation method, and seal and package them with a low-loss polytetrafluoroethylene tape to make a standardized sandwich detection unit with a thickness of 0.25 ± 0.05 mm.

[0108] The present invention prepares a tobacco cut filler detection sample set through a standardized process. The present invention prepares 10 samples of cut tobacco, cut stem, and reconstituted tobacco each, and prepares three types of mixed tobacco cut filler samples. The mass ratio of the mixed sample 1 is cut tobacco:cut stem:reconstituted tobacco = 1:1:8, the mass ratio of the mixed sample 2 is cut tobacco:cut stem:reconstituted tobacco = 3:2:5, and the mass ratio of the mixed sample 3 is cut tobacco:cut stem:reconstituted tobacco = 1:7:2. Take the three types of tobacco cut filler in proportion, fully mix them, and pick them up with tweezers to make a glass slide specimen.

[0109] Step S3: Obtain the terahertz time-domain spectral data of the single-component tobacco cut filler samples of cut tobacco, cut stem, and reconstituted tobacco and the mixed tobacco cut filler samples of each group.

[0110] Specifically, use a terahertz time-domain spectroscopy system to detect the single-component tobacco cut filler samples of cut tobacco, cut stem, and reconstituted tobacco and the mixed tobacco cut filler samples of each group to obtain terahertz time-domain spectral data. Among them, the scanning frequency spectrum range is 0-3 THz, the spectral resolution is 12.5 GHz, the average number of scans per time is 1024 times, the test environment temperature is 20.5°C - 21.5°C (for example, 21°C), and the environmental relative humidity is 3% - 5% (for example, 4%). In some embodiments of the present invention, a TP15K (model) terahertz time-domain spectroscopy system produced by Dier Times (Tianjin) Technology Co., Ltd. is used to collect terahertz time-domain spectral data.

[0111] Step S4: Construct a three-wire ratio inversion model based on partial least squares regression according to the spectral response differences between the mixed tobacco cut filler samples of each group and the single-component tobacco cut filler samples.

[0112] Among them, the spectral response differences of the terahertz optical parameters between the mixed tobacco cut filler samples of each group and the single-component tobacco cut filler samples include: for the single-component tobacco cut filler samples in the frequency band of 0.5-1.1 THz, cut tobacco shows significant identification due to its high absorption coefficient in the range of 0.7 THz - 1.1 THz and the dielectric constant characteristic peak at 0.55 THz; the transmittance of reconstituted tobacco is optimal in the medium frequency band of 0.4 THz - 0.8 THz; for the mixed tobacco cut filler samples, the absorption coefficient and dielectric constant have a strong sorting relationship with the ratio in the characteristic frequency bands of 0.53 THz - 0.59 THz and 0.67 THz - 1.1 THz.

[0113] Specifically, in terms of the main chemical components, the research data show that there are significant inter-group differences in the total sugar content (cut tobacco: 28.5 ± 3.2%, stem cuttings: 19.8 ± 2.5%, reconstituted tobacco: ≤9.5%); the cellulose distribution shows an opposite trend (cut tobacco: 8.2 ± 1.5%, stem cuttings: 24.3 ± 2.8%, reconstituted tobacco: 21.7 ± 3.1%). Research analysis shows that the three types of tobacco cuttings have significantly heterogeneous microstructures: cut tobacco forms a porous network matrix with a palisade tissue cell skeleton (porosity 62 ± 8%), stem cuttings show the characteristics of dense arrangement of vascular bundles (density 1.32 ± 0.15 g / cm³), and reconstituted tobacco forms an anisotropic network system through fiber orientation cross-linking (fiber diameter 15.6 ± 3.8 μm). This dual difference in composition and structure results in distinguishable characteristics in terahertz characteristic parameters such as dielectric response and phonon resonance, providing a material basis for the establishment of the non-destructive testing method of the present invention. The differences in the chemical composition and physical structure of tobacco leaves may lead to different performances in terahertz optical parameters. Research shows that the dielectric constant of reducing sugars such as glucose is higher than that of cellulose. Therefore, the propagation speed of terahertz waves slows down in tobacco leaves containing substances with a higher dielectric constant, resulting in a time delay in the signals in the time-domain spectrum. In addition, an increase in the peak intensity usually means that the sample absorbs less terahertz waves and has a higher transmittance.

[0114] Figure 2 and Figure 3 respectively show the comparison of the terahertz time-domain spectra and the terahertz frequency-domain spectra of the three types of tobacco cuttings (cut tobacco, stem cuttings, and reconstituted tobacco) with the reference sample. Through the terahertz spectral characteristic analysis of the "three types of tobacco cuttings", it is found that there are differences in the frequency-domain amplitudes of the three types of tobacco cuttings in different frequency bands: in the low-frequency region (<0.25 THz): the frequency-domain curves of the three types of tobacco cuttings almost coincide, indicating that in this frequency band, their terahertz responses are similar. In the mid-frequency region (0.25 - 0.75 THz): the frequency-domain amplitude shows the trend of reconstituted tobacco > cut tobacco > stem cuttings, suggesting that reconstituted tobacco absorbs the least terahertz waves and has the highest transmittance in this frequency band. In the high-frequency region (>0.75 THz): the frequency-domain amplitude curves of the three types of tobacco cuttings intersect and do not show an obvious regularity. These differences may be due to the different chemical compositions and physical structures of the tobacco cuttings. For example, reconstituted tobacco may contain a higher sugar content and a lower cellulose content, resulting in less absorption of terahertz waves in the mid-frequency region. In addition, differences in physical structures such as fiber length, density, and porosity may also affect the propagation characteristics of terahertz waves, further leading to different performances in the frequency-domain spectrum. Therefore, combining the analysis of chemical composition and physical structure helps to deeply understand the performance differences of tobacco cuttings in the terahertz frequency-domain spectrum.

[0115] Figure 4 、 Figure 5 and Figure 6The terahertz absorption spectrum comparison, dielectric constant spectrum comparison, and refractive index spectrum comparison of the three types of cut tobacco (leaf cut tobacco, stem cut tobacco, and reconstituted tobacco) are respectively shown. By further analyzing the absorption coefficient spectrum, dielectric constant spectrum, and refractive index spectrum of the three types of cut tobacco, it is found that the absorption coefficient and refractive index spectral lines of the three types of cut tobacco have a certain degree of separability. Figure 4 shows the absorption coefficients of the three types of cut tobacco for different frequencies. From Figure 4 it can be seen that the absorption coefficients of the three are not significantly different in the entire THz band, but in the region of 0.7 THz - 1.1 THz, the absorption coefficient of leaf cut tobacco is slightly greater than that of stem cut tobacco and greater than that of reconstituted tobacco. Figure 5 are the dielectric constant data of the three types of cut tobacco. In the frequency range of 0.350 THz - 1.125 THz, the dielectric constant of stem cut tobacco remains basically unchanged at around 3.4 F / m, the dielectric constant of reconstituted tobacco fluctuates slightly between 3.8 F / m - 4.3 F / m, and the dielectric constant of leaf cut tobacco has a small peak of 7.5 F / m at 0.55 THz. From this, it can be seen that in the range of 0.5 THz - 1 THz, the dielectric constants of the three types of cut tobacco are significantly different, and the three can be distinguished. Figure 6 are the refractive index images of the three types of cut tobacco at different frequencies. Figure 6 shows that in the region of 0.3 THz - 0.7 THz, the refractive index of leaf cut tobacco is the largest, which is greater than that of stem cut tobacco and greater than that of reconstituted tobacco. However, in the region of 0.8 THz - 1.3 THz, the refractive index of leaf cut tobacco is the smallest among the three, and the refractive index curves of stem cut tobacco and reconstituted tobacco intersect. The refractive index of leaf cut tobacco is in the range of 1.45 - 1.85, the refractive index range of stem cut tobacco is 1.6 - 1.83, and the refractive index range of reconstituted tobacco is 1.6 - 1.81. Therefore, the differences in the terahertz spectral curves of different types of cut tobacco determine that terahertz spectroscopy can be used to distinguish the types of cut tobacco.

[0116] In order to further analyze the absorption characteristics of the three types of cut tobacco for the THz spectrum, three groups of mixed cut tobacco samples were detected by a terahertz time-domain spectroscopy system. Among them, the weight ratios of the three types of cut tobacco in mixed sample 1 are: leaf cut tobacco: stem cut tobacco: reconstituted tobacco = 1:1:8, the weight ratios of the three types of cut tobacco in mixed sample 2 are: leaf cut tobacco: stem cut tobacco: reconstituted tobacco = 3:2:5, and the weight ratios of the three types of cut tobacco in mixed sample 3 are leaf cut tobacco: stem cut tobacco: reconstituted tobacco = 1:7:2. Figure 7 and Figure 8 are respectively the time-domain spectra and frequency-domain spectra of the three mixed samples and the reference sample obtained by detection. From Figure 7 it can be seen that the time-domain spectra of the three mixed samples are approximately coincident, but the maximum amplitudes of the three are slightly different. And for the frequency-domain spectra Figure 8, which contains a large amount of information. It can be found that there are significant differences in the frequency-domain amplitudes of the three blended tobacco samples in different frequency bands: In the low-frequency region (<0.4 THz): The frequency-domain curves of the three blended tobacco samples almost coincide, indicating that in this frequency band, their terahertz responses are similar. In the medium-frequency region (0.4 THz - 0.8 THz): The frequency-domain amplitudes show that Blend Sample 3 > Blend Sample 2 > Blend Sample 1. In the high-frequency region (>0.8 THz): The frequency-domain amplitude curves of the three blended tobacco samples intersect and do not show obvious regularity. Therefore, the tobacco blends with different proportions have different spectral characteristics, further verifying the feasibility of the method for inferring the proportion content of various tobacco blends through THz information.

[0117] Next, terahertz spectral characteristics analysis was carried out on three blended tobacco samples with different blending ratios. Specifically, based on the time-domain and frequency-domain spectral information of the three blended samples, through the frequency-domain feature extraction algorithm calculation (the specific calculation process will be introduced in step S421), the absorption coefficient spectra, dielectric constant spectra, and refractive index spectra of the three blended samples were obtained as Figure 9 , Figure 10 and Figure 11 . First Figure 9 are the absorption coefficients of the three blended samples for different frequencies. Analyzing Figure 9 it can be found that in the frequency spectrum range of <0.35 THz, the absorption coefficient curves of the three blended tobacco samples basically coincide. For the frequency spectrum range of 0.53 THz - 0.59 THz, their absorption coefficient spectra show that Blend Sample 3 > Blend Sample 2 > Blend Sample 1, while for the frequency spectrum range of 0.86 THz - 1.03 THz, their absorption coefficient spectra show that Blend Sample 3 > Blend Sample 1 > Blend Sample 2. Therefore, the absorption coefficient spectra of samples with different blending ratios have different characteristics and have a certain linear magnitude relationship in some bands. Therefore, the proportion content of various tobacco blends can be analyzed by comparing the absorption coefficient spectra. Figure 10 are the dielectric constant data of the three blended samples. Analyzing Figure 10 it can be found that for the frequency spectrum range of <0.6 THz, the dielectric constant curves of Blend Sample 2 and Blend Sample 3 roughly coincide, and the dielectric constant of Blend Sample 1 is greater than the two. For the frequency spectrum range of 0.67 THz - 1.1 THz, the three blended samples have a certain linear magnitude relationship, and it can be found that Blend Sample 3 > Blend Sample 2 > Blend Sample 1. Therefore, the difference in dielectric constant can also be used as key data to distinguish the content of the three tobacco blends. Figure 11 are the refractive index images of the three blended tobacco samples at different frequencies. Analyzing Figure 11It can be found that in the frequency spectrum range of 0.4 THz to 1.05 THz, the refractive index of the mixed sample 2 is the largest, which is greater than that of the mixed sample 3 and greater than that of the mixed sample 1. After >1.1 THz, the refractive index curves of the three intersect with each other, and their reference value is not great. To sum up, there are obvious differences in the absorption coefficient spectrum, dielectric constant spectrum and refractive index spectrum of the mixed cut tobacco samples with different mixing ratios. Therefore, the mixing ratios of the three kinds of cut tobacco in the samples can be analyzed by comprehensively analyzing the three spectra.

[0118] In an implementation manner of the method for non-destructively detecting the blending ratio of cut tobacco based on terahertz time-domain spectroscopy of the present invention, the step S4 may specifically include:

[0119] Step S41: Preprocess the terahertz time-domain spectroscopy data of each group of mixed cut tobacco samples and single-component cut tobacco samples.

[0120] Through preprocessing, the signal-to-noise ratio and model robustness of the terahertz time-domain spectroscopy data can be improved. In an implementation manner of the method for non-destructively detecting the blending ratio of cut tobacco based on terahertz time-domain spectroscopy of the present invention, the step S41 may specifically include:

[0121] Step S411: Perform 5-layer discrete wavelet decomposition using the Symlet8 wavelet basis function, and combine the soft threshold method to suppress the high-frequency noise in the time-domain waveform.

[0122] Step S412: Perform Savitzky-Golay smoothing and airPLS baseline correction on the frequency-domain signal.

[0123] Through step S412, the low-frequency scattering background and high-frequency spike noise can be effectively eliminated. Among them, the window width of Savitzky-Golay smoothing is 10 points, and a third-degree polynomial fitting is used.

[0124] Step S413: Based on the distribution of characteristic absorption peaks and the system signal-to-noise ratio threshold, intercept the effective frequency band of 0.1-1.5 THz, remove the low-frequency distortion area below 0.1 THz and the high-frequency noise area above 1.5 THz, and retain more than 95% of the effective spectral information.

[0125] The present invention takes the complex chemical system of tobacco as the research object, and the diversity of its metabolites is typical in the plant kingdom. Existing literature has confirmed that more than 5,000 chemical components have been identified in flue-cured tobacco leaves. Among them, the macromolecular backbone vibrations and dipole moment changes of polysaccharides, alkaloids, and phenolic substances show characteristic responses in the terahertz frequency band (0.1 - 10 THz), which provides a molecular vibration theoretical basis for component analysis based on terahertz spectroscopy. In terahertz time-domain spectroscopy analysis, tobacco shreds have a strong absorption of high-frequency terahertz waves, resulting in a large signal noise in the high-frequency region. Although the low-frequency region is rich in information, due to insufficient frequency resolution, the signal fluctuations are more obvious. Therefore, the present invention selects the frequency-domain data in the range of 0.1 THz - 1.50 THz as the analysis basis. This frequency band selection helps to balance the signal quality and frequency resolution, thereby improving the reliability of the analysis results.

[0126] Step S414: Normalize the absorption coefficient and the imaginary part of the dielectric constant through standard normal variate (SNV).

[0127] Through step S414, the dimensional effect caused by the difference in sample thickness can be eliminated.

[0128] Step S42: Obtain the terahertz optical parameters of each group of mixed tobacco shred samples and single-component tobacco shred samples according to the preprocessed terahertz time-domain spectroscopy data of each group of mixed tobacco shred samples and single-component tobacco shred samples.

[0129] In an implementation manner of the method for non-destructively detecting the blending ratio of tobacco shreds based on terahertz time-domain spectroscopy of the present invention, step S42 may specifically include:

[0130] Step S421: Determine the sensitive frequency band in which the tobacco shred thickness and the terahertz optical parameters show a linear response through characteristic band screening, where the terahertz optical parameters include the absorption coefficient α(f), the refractive index n(f), and the dielectric constant ε(f).

[0131] Among them, the terahertz optical parameters of each group of mixed tobacco shred samples and single-component tobacco shred samples are obtained from the corresponding terahertz time-domain spectroscopy data. The following derives the formulas for the terahertz absorption coefficient and refractive index according to the terahertz time-domain spectroscopy mathematical model. By irradiating a glass slide sample with a terahertz laser, the electric field signal after irradiating the sample is collected to obtain the reference signal time-domain electric field signal and the tobacco shred sample time-domain electric field signal . Perform Fourier transform (FFT) on the time-domain signal to obtain the amplitude and phase in the frequency domain:

[0132]

[0133]

[0134] The transmission coefficient (transfer function) is:

[0135]

[0136] According to the Beer-Lambert law, the absorption coefficient (the relationship with the transmission coefficient is:

[0137]

[0138] where represents the sample thickness. Therefore, the absorption coefficient can be calculated as:

[0139]

[0140] If the complex refractive index of the sample is measured

[0141]

[0142] In the formula, represents the complex refractive index, represents the real refractive index, and the real refractive index can describe the dispersion degree of the terahertz wave passing through the sample; is the imaginary unit, represents the extinction coefficient, with the unit of 1, which is used to describe the absorption ability of the sample to the terahertz wave.

[0143] Then, the absorption coefficient can be calculated through the imaginary part of the refractive index Calculate

[0144]

[0145] After calculation, the expression for the real refractive index can be obtained as

[0146]

[0147] The expression for the absorption coefficient is:

[0148]

[0149] In the formula, represents the absorption coefficient, with the unit of ; represents the sample thickness, with the unit of ; represents the electromagnetic wave velocity, with the unit of ; represents the angular frequency, with the unit of ; represents the amplitude ratio of the terahertz wave passing through the sample to the reference signal; represents the phase difference, with the unit of ;

[0150] Assume that the sample is a homogeneous material, and its complex refractive index is:

[0151] ,

[0152] The transmission coefficient can be calculated through the following relationship :

[0153]

[0154] where represents the sample thickness, represents the speed of light, represents the refractive index, represents the extinction coefficient. This equation can be solved by numerical methods (such as iterative solution or Kramers-Kronig relationship) and .

[0155] If the sample is thin enough (single-pass transmission approximation), the logarithmic formula can be used to simplify the calculation:

[0156]

[0157]

[0158] Complex permittivity and the complex refractive index are related as:

[0159]

[0160] Expanding gives

[0161]

[0162]

[0163] represents the real part of the permittivity of the material, which is related to the polarization characteristics of the material, represents the dielectric loss, which describes the absorption of THz waves by the material.

[0164] Step S422: Measure the time-domain spectral signals of each single-component sample in the sensitive frequency band respectively, and extract the standard optical parameters through fast Fourier transform (FFT).

[0165] Step S423: For each group of mixed tobacco samples, synchronously obtain the terahertz time-domain waveform in the sensitive frequency band, and calculate the optical parameter values of each group of mixed tobacco samples.

[0166] Step S43: Based on the spectral response differences of the terahertz optical parameters between the mixed cut tobacco samples in each group and the single-component cut tobacco samples, and based on the principle of linear superposition of optical parameters, construct the following linear spectral mixing model:

[0167]

[0168] Wherein, represents the optical parameter vector of the mixed sample with unknown doping ratio at characteristic frequency points. The optical parameter vector represents the absorption coefficient, refractive index, and dielectric constant. is expressed as

[0169] ,

[0170] represents the standard spectral matrix of the pure components of cut filler, cut stem, and cut lamina, and is expressed as

[0171] ,

[0172] represents the doping ratio matrix vector of cut filler, cut stem, and cut lamina in tobacco leaves, and is expressed as

[0173]

[0174] is the proportional coefficient to be solved. Wherein, , , respectively represent the mass fractions of cut filler, cut stem, and cut lamina, and , represents the measurement noise.

[0175] By finding the linear superposition region of terahertz optical parameters and solving the above linear equation, the doping ratios of the mixed cut tobacco can be obtained , , .

[0176] Step S44: For the linear equations of multiple groups of mixed cut tobacco samples (for example, n = 30 samples (3 groups × 10 repetitions)), solve them by the following formula using the constrained least squares method:

[0177] ,

[0178] Use the active set algorithm to iteratively solve, set the convergence threshold to , and obtain the quantitative inversion result of the cut tobacco blending ratio. The results are shown in Table 1.

[0179] Table 1 Inversion results and verification of the "three cuts" ratio of cut tobacco

[0180]

[0181] In the present invention, by constructing a constrained least squares model, the accurate inversion of the blending ratio of cut tobacco is realized, and the obtained results are shown in Table 1. The ratio inversion results of the three groups of blended samples are highly consistent with the preset values (root mean square error ≤ 5.00%, correlation coefficient ≥ 0.976). Among them, reconstituted tobacco shows the best detection accuracy (relative standard deviation < 1.5%) due to the significant cellulose characteristic peak at 1.0 THz, while the inversion error of cut stem in the spectral overlapping frequency band of leaf tobacco - cut stem (0.9 THz) is slightly higher (up to -1.57%), revealing the key influence of the preferred characteristic band on the model performance. Since this model ensures the physical rationality of the ratio parameters through non - negative constraints and normalization conditions, and combines the matrix batch solution strategy to achieve the efficient analysis of 30 groups of samples (time - consuming < 1.00 s).

[0182] Step S5: Obtain the terahertz time - domain spectral data of the cut tobacco to be measured.

[0183] Specifically, refer to step S12 to prepare the cut tobacco sample to be measured by the simple sandwich encapsulation method, and refer to step S3 for terahertz time - domain spectral detection, which will not be elaborated here.

[0184] Step S6: According to the terahertz time - domain spectral data of the cut tobacco to be measured and the three - wire ratio inversion model, obtain the blending ratio of the cut tobacco to be measured.

[0185] In an implementation manner of the non - destructive detection method for the blending ratio of cut tobacco based on terahertz time - domain spectroscopy of the present invention, step S6 may specifically include:

[0186] Step S61: Obtain the terahertz time - domain spectral data of the cut tobacco to be measured.

[0187] Step S62: Based on the terahertz time - domain spectral data of the cut tobacco to be measured, obtain the terahertz optical parameter vector of the cut tobacco to be measured.

[0188] Step S63: Input the terahertz optical parameter vector of the cut tobacco to be measured into the three - wire ratio inversion model to obtain the blending ratio of the cut tobacco to be measured.

[0189] The non-destructive detection method for the blending ratio of cut tobacco based on terahertz time-domain spectroscopy provided by the embodiments of the present invention is based on terahertz time-domain spectroscopy (THz-TDS). Combining the linear correlation between the absorption coefficient, dielectric constant and refractive index spectrum of cut tobacco in a specific band, a quantitative inversion method for the blending ratio of cut tobacco with multi-parameter fusion is proposed, realizing the rapid non-destructive detection of the blending ratio of three types of cut tobacco. In terms of spectral feature differences: in the frequency band of 0.5 THz - 1.1 THz, cut filler shows significant identification due to its high absorption coefficient (0.7 - 1.1 THz) and the dielectric constant characteristic peak (7.5 F / m) at 0.55 THz; reconstituted tobacco has the best transmittance in the middle frequency band (0.4 - 0.8 THz) due to its high cellulose content, providing a spectral basis for component differentiation. In terms of the correlation of the blending ratio: by preparing three types of mixed samples, it is found that there is a strong sorting relationship between their absorption coefficient and dielectric constant in the characteristic frequency bands (such as 0.53 - 0.59 THz, 0.67 - 1.1 THz) and the blending ratio, verifying the sensitivity of terahertz spectroscopy to the component ratio. In terms of the model performance of the three-component ratio inversion model, the constructed constrained least squares model combines multiple optical parameters, and the root mean square error of the inverted blending ratio ≤ 5% (R 2 ≥ 0.976), among which the detection accuracy of reconstituted tobacco is the best (RSD < 1.5%). The model realizes efficient analysis through non-negative constraints and batch solution strategies (the time-consuming for a single sample < 1 s); the present invention provides a high-precision and high-robustness solution for the non-destructive detection of cut tobacco blending in the tobacco industry.

[0190] So far, the embodiments of the present disclosure have been described in detail. In order to avoid obscuring the concept of the present disclosure, some details well known in the art have not been described. Those skilled in the art can fully understand how to implement the technical solutions disclosed herein based on the above description.

[0191] Although some specific embodiments of the present disclosure have been described in detail by way of examples, those skilled in the art should understand that the above examples are only for the purpose of illustration and not for the purpose of limiting the scope of the present disclosure. Those skilled in the art should understand that the above embodiments can be modified or some technical features can be equivalently replaced without departing from the scope and spirit of the present disclosure. The scope of the present disclosure is defined by the appended claims.

Claims

1. A non-destructive detection method for the blending ratio of cut tobacco based on terahertz time-domain spectroscopy, characterized in that Including: Preparing single-component cut tobacco samples of leaf cut tobacco, stem cut tobacco and reconstituted tobacco respectively; Constructing multiple groups of cut tobacco mixing systems according to a preset mass ratio gradient to obtain multiple groups of mixed cut tobacco samples; Obtaining terahertz time-domain spectroscopy data of single-component cut tobacco samples of leaf cut tobacco, stem cut tobacco and reconstituted tobacco and each group of mixed cut tobacco samples; Constructing a three-wire ratio inversion model based on partial least squares regression according to the spectral response differences between each group of mixed cut tobacco samples and single-component cut tobacco samples; Obtaining terahertz time-domain spectroscopy data of the cut tobacco to be measured; According to the terahertz time-domain spectroscopy data of the cut tobacco to be measured and the three-wire ratio inversion model, obtaining the blending ratio of the cut tobacco to be measured.

2. The non-destructive detection method for the blending ratio of cut tobacco based on terahertz time-domain spectroscopy according to claim 1, wherein The preparing single-component cut tobacco samples of leaf cut tobacco, stem cut tobacco and reconstituted tobacco respectively includes: Performing low-temperature drying pretreatment on leaf cut tobacco, stem cut tobacco and reconstituted tobacco respectively; Preparing cut tobacco test samples of leaf cut tobacco, stem cut tobacco and reconstituted tobacco by using a simple sandwich encapsulation method.

3. The non-destructive detection method for the blending ratio of cut tobacco based on terahertz time-domain spectroscopy according to claim 2, wherein The performing low-temperature drying pretreatment on leaf cut tobacco, stem cut tobacco and reconstituted tobacco respectively includes: Placing leaf cut tobacco, stem cut tobacco and reconstituted tobacco in a drying oven, and performing dry hot air ventilation drying at 40°C - 50°C for 3h - 5h until constant weight for standby; The preparing cut tobacco test samples of leaf cut tobacco, stem cut tobacco and reconstituted tobacco by using a simple sandwich encapsulation method includes: Weighing leaf cut tobacco, stem cut tobacco and reconstituted tobacco respectively, and controlling the total mass to be 4.000 ± 0.005 g; Placing the three kinds of cut tobacco between two clean glass slides respectively; By gently tapping the edge of the glass slide and horizontally rotating and fine-tuning, making the cut tobacco naturally spread flat in the central area; After visually confirming no significant agglomeration or voids, sealing and fixing along the periphery of the glass slide with a low-dielectric polyester film tape to form a sandwich sample with a thickness of 0.3mm - 0.5mm.

4. The non-destructive detection method for the blending ratio of cut tobacco based on terahertz time-domain spectroscopy according to claim 1, characterized in that The constructing multiple groups of cut tobacco mixing systems according to a preset mass ratio gradient to obtain multiple groups of mixed cut tobacco samples includes: Preparing 10 single-component samples of leaf cut tobacco, stem cut tobacco and reconstituted tobacco respectively, and controlling the mass of each sample to be 0.200 ± 0.001 g; Constructing three groups of mixing systems according to a preset mass ratio gradient to obtain slide specimens corresponding to mixed sample 1, mixed sample 2 and mixed sample 3 respectively, wherein the three-wire ratios of leaf cut tobacco:stem cut tobacco:reconstituted tobacco corresponding to mixed sample 1, mixed sample 2 and mixed sample 3 are respectively: 1:1:8, 3:2:5 and 1:7:

2.

5. The non-destructive detection method for the blending ratio of cut tobacco based on terahertz time-domain spectroscopy according to claim 4, wherein The constructing three groups of mixing systems according to a preset mass ratio gradient to obtain slide specimens corresponding to mixed sample 1, mixed sample 2 and mixed sample 3 respectively includes: Preparing 10 parallel samples for each group of the mixing systems according to the corresponding blending ratio; Mixing the 10 parallel samples corresponding to each group of the mixing systems evenly by using an oscillator; Quantitatively clamping cut tobacco mixtures in each group of the mixing systems by using tweezers, spreading them flat between glass slides, regulating to form a uniform distribution area by the tapping-rotation method, and sealing and encapsulating with a low-loss polytetrafluoroethylene tape to make a standardized sandwich detection unit with a thickness of 0.25 ± 0.05 mm.

6. The non-destructive detection method for the blending ratio of cut tobacco based on terahertz time-domain spectroscopy according to claim 1, wherein The obtaining terahertz time-domain spectroscopy data of single-component cut tobacco samples of leaf cut tobacco, stem cut tobacco and reconstituted tobacco and each group of mixed cut tobacco samples includes: The single-component cut tobacco samples of cut filler, cut stem and reconstituted tobacco, as well as the mixed cut tobacco samples of each group, were detected by a terahertz time-domain spectroscopy system to obtain terahertz time-domain spectral data. Among them, the scanning frequency spectrum range was 0 - 3 THz, the spectral resolution was 12.5 GHz, the average number of scans per time was 1024 times, the test environment temperature was 20.5°C - 21.5°C, and the environmental relative humidity was 3% - 5%.

7. The non-destructive detection method for the blending ratio of cut tobacco based on terahertz time-domain spectroscopy according to claim 1, wherein Constructing a three-wire ratio inversion model based on partial least squares regression according to the spectral response differences of the terahertz time-domain spectral data of the mixed cut tobacco samples of each group and the single-component cut tobacco samples, including: Preprocessing the terahertz time-domain spectral data of the mixed cut tobacco samples of each group and the single-component cut tobacco samples; Obtaining the terahertz optical parameters of the mixed cut tobacco samples of each group and the single-component cut tobacco samples according to the preprocessed terahertz time-domain spectral data of the mixed cut tobacco samples of each group and the single-component cut tobacco samples; According to the spectral response differences of the terahertz optical parameters of the mixed cut tobacco samples of each group and the single-component cut tobacco samples, based on the principle of linear superposition of optical parameters, constructing the following linear spectral mixture model: , wherein, represents the optical parameter vectors of the mixed sample with unknown doping ratios at characteristic frequency points, and the optical parameter vectors represent the absorption coefficient, refractive index, and dielectric constant, is expressed as ， The standard spectral matrix representing the pure components of cut tobacco, cut stem and reconstituted tobacco sheet, denoted as ， It represents the doping ratio matrix vector of cut tobacco, stem cuttings and reconstituted tobacco in tobacco leaves, and is expressed as , is the proportionality coefficient to be determined, where , , represent the mass fractions of cut tobacco, cut stem and reconstituted tobacco sheet respectively, and , represents the measurement noise; For the linear equations of multiple groups of mixed cut tobacco samples, solve them by the following formula using the constrained least squares method: ， Iteratively solve using the active set algorithm, and set the convergence threshold to , and obtain the quantitative inversion result of the tobacco blending ratio.

8. The non-destructive detection method for the blending ratio of cut tobacco based on terahertz time-domain spectroscopy according to claim 7, characterized in that The preprocessing of the terahertz time-domain spectral data of the mixed cut tobacco samples of each group and the single-component cut tobacco samples includes: Performing 5-layer discrete wavelet decomposition using the Symlet8 wavelet basis function, and combining the soft threshold method to suppress the high-frequency noise in the time-domain waveform; Performing Savitzky-Golay smoothing and airPLS baseline correction on the frequency-domain signal; Based on the characteristic absorption peak distribution and the system signal-to-noise ratio threshold, intercepting the effective frequency band of 0.1 - 1.5 THz, removing the low-frequency distortion region below 0.1 THz and the high-frequency noise region above 1.5 THz, and retaining more than 95% of the effective spectral information; Normalizing the absorption coefficient and the imaginary part of the dielectric constant through standard normal transformation.

9. The non-destructive detection method for the blending ratio of cut tobacco based on terahertz time-domain spectroscopy according to claim 7, wherein Obtaining the terahertz optical parameters of the mixed cut tobacco samples of each group and the single-component cut tobacco samples according to the preprocessed terahertz time-domain spectral data of the mixed cut tobacco samples of each group and the single-component cut tobacco samples, including: Determine the sensitive frequency band where the thickness of cut tobacco has a linear response to terahertz optical parameters through feature band screening, wherein the terahertz optical parameters include absorption coefficient , refractive index , and dielectric constant ; Measuring the time-domain spectral signals of each single-component sample in the sensitive frequency band respectively, and extracting the standard optical parameters through fast Fourier transform; For each group of mixed cut tobacco samples, synchronously obtaining the terahertz time-domain waveform in the sensitive frequency band, and calculating the optical parameter values of each group of mixed cut tobacco samples, The spectral response differences of the terahertz optical parameters of the mixed cut tobacco samples of each group and the single-component cut tobacco samples include: For the single-component cut tobacco samples in the 0.5 - 1.1 THz frequency band, cut filler shows significant identification due to its high absorption coefficient in the range of 0.7 THz - 1.1 THz and the dielectric constant characteristic peak at 0.55 THz; the transmittance of reconstituted tobacco is optimal in the medium frequency band of 0.4 THz - 0.8 THz; For the mixed cut tobacco samples, the absorption coefficient and the dielectric constant have a strong sorting relationship with the ratio in the characteristic frequency bands of 0.53 THz - 0.59 THz and 0.67 THz - 1.1 THz.

10. The non-destructive detection method for the blending ratio of cut tobacco based on terahertz time-domain spectroscopy according to claim 1, characterized in that Obtaining the blending ratio of the to-be-detected cut tobacco according to the terahertz time-domain spectroscopy data of the to-be-detected cut tobacco and the three-wire ratio inversion model includes: Obtaining the terahertz time-domain spectroscopy data of the to-be-detected cut tobacco; Obtaining the terahertz optical parameter vector of the to-be-detected cut tobacco based on the terahertz time-domain spectroscopy data of the to-be-detected cut tobacco; Inputting the terahertz optical parameter vector of the to-be-detected cut tobacco into the three-wire ratio inversion model to obtain the blending ratio of the to-be-detected cut tobacco.

Images