Method and system for non-destructive testing of total content of unsaturated fatty acids in eucalypt leaves

Abstract

The application discloses a method and system for nondestructive detection of total content of unsaturated fatty acids in eucalyptus leaves, and relates to the technical field of near-infrared spectrum analysis. The method comprises the following steps: collecting a near-infrared diffuse reflection spectrum of an isolated eucalyptus leaf sample; after pretreatment through smoothing and denoising, scattering correction, derivative correction and water interference deduction, screening characteristic wavelengths through a competitive self-adaptive reweighted sampling algorithm combined with a successive projection algorithm, eliminating interference wavelengths within a range of plus or minus 10 nm of a characteristic absorption peak position of 1,8-eucalyptol, and retaining characteristic wavelength points within a 1650-1800 nm and 2200-2400 nm wavelength band with an absolute value of a Pearson correlation coefficient greater than 0.6; and then importing the characteristic wavelength spectrum data into a partial least squares regression quantitative correction model to output a content prediction value. The application also provides a corresponding system. The application does not require chemical pretreatment, a single detection takes less than 3 minutes, the determination coefficient of a prediction set is greater than or equal to 0.85, water and essential oil interference can be eliminated, and fast, green and nondestructive detection is achieved.

Description

Technical Field

[0001] This invention relates to the field of near-infrared spectroscopy analysis technology, and more specifically, to a method and system for non-destructive testing of the total content of unsaturated fatty acids in eucalyptus leaves. Background Technology

[0002] Eucalyptus leaves are rich in polyunsaturated fatty acids, mainly linoleic acid and α-linolenic acid. Accurate determination of the total unsaturated fatty acid content in eucalyptus leaves is of great significance for the evaluation of eucalyptus germplasm resources and genetic breeding assessment.

[0003] Currently, the detection of eucalyptus fatty acids mainly uses gas chromatography-mass spectrometry (GC-MS). This method requires pretreatment of the sample, such as organic solvent extraction and methylation. The sample is consumed during the detection process, making non-destructive testing impossible. The detection cycle for a single sample usually takes several hours, which is difficult to meet the needs of high-throughput screening. Furthermore, the detection process requires the use of toxic and harmful chemical reagents such as chloroform and methanol, posing environmental pollution and health and safety hazards.

[0004] Near-infrared spectroscopy has been used to detect fatty acid content in some plant seeds or oilseeds, and there are also reports of its application in determining the chemical composition of eucalyptus and qualitatively identifying eucalyptus leaves as a traditional Chinese medicine. However, to date, there are no reports of applying near-infrared spectroscopy to the non-destructive detection of total unsaturated fatty acid content in eucalyptus leaves. Eucalyptus leaves are rich in volatile essential oils, and fresh eucalyptus leaves have a high water content. The characteristic absorption bands of essential oils overlap with those of unsaturated fatty acids, and water content causes strong absorption interference in the near-infrared region, making it impossible to directly apply existing detection models.

[0005] In view of the above situation, the present invention provides a method and system for non-destructive testing of the total content of unsaturated fatty acids in eucalyptus leaves. Summary of the Invention

[0006] To achieve the above objectives, the present invention provides the following technical solution: On the one hand, the present invention provides a method for non-destructive testing of the total content of unsaturated fatty acids in eucalyptus leaves, comprising the following steps: S1. Collect near-infrared diffuse reflectance spectra of detached eucalyptus leaf samples; S2. Preprocess the near-infrared diffuse reflectance spectrum. The preprocessing includes smoothing and noise reduction, scattering correction, derivative correction and moisture interference subtraction. S3. Using a competitive adaptive reweighted sampling algorithm combined with a continuous projection algorithm, characteristic wavelengths related to the total content of eucalyptus unsaturated fatty acids are screened from the preprocessed spectrum. The absolute value of the Pearson correlation coefficient between the screened characteristic wavelengths and the total content of eucalyptus unsaturated fatty acids is greater than 0.6. Wavelength variables within ±10 nm of the characteristic absorption peak of 1,8-cineole are removed. The characteristic wavelengths include at least 3 characteristic wavelength points in the 1650-1800 nm band and at least 3 characteristic wavelength points in the 2200-2400 nm band. S4. Import the spectral data of the selected characteristic wavelengths into the pre-established quantitative correction model. The quantitative correction model outputs the predicted value of the total content of eucalyptus unsaturated fatty acids. The quantitative correction model was established using partial least squares regression, with the total content of eucalyptus unsaturated fatty acids determined by gas chromatography-mass spectrometry as a reference value. The characteristic absorption peak position of 1,8-cineole was obtained by pre-determining the near-infrared spectrum of pure 1,8-cineole. During the modeling stage, the wavelength variables to be removed were determined and fixed into the quantitative correction model.

[0007] Preferably, in step S2, the smoothing and denoising process uses the Savitzky-Golay convolutional smoothing method, wherein the window width of the Savitzky-Golay convolutional smoothing method is 15 and the polynomial order is 3.

[0008] Preferably, in step S2, the scattering correction process employs standard normal variable transformation or multivariate scattering correction; The derivative correction process uses either the first or second derivative; Moisture interference subtraction is performed using orthogonal signal correction or external parameter orthogonalization.

[0009] Preferably, in step S4, the quantitative calibration model uses leave-one-out cross-validation to determine the number of principal components, and the number of principal components is 10 to 12.

[0010] Preferably, in step S4, the external prediction set determination coefficient of the quantitative correction model is ≥0.85, the root mean square error of the external prediction is ≤0.06%, and the relative analysis error is ≥2.5.

[0011] Preferably, in step S1, the near-infrared diffuse reflectance spectrum acquisition conditions are: spectral range 1000–2500 nm, resolution 8 cm⁻¹. -1 The number of scans was 64.

[0012] Preferably, the eucalyptus leaf sample is an detached fresh eucalyptus leaf, and the collection part is a mature leaf on the sunny side of the middle of the plant. The total content of unsaturated fatty acids is expressed as a percentage of the fresh weight of the leaf. Alternatively, the moisture content of the eucalyptus leaf sample can be measured simultaneously, and the moisture content can be included as an orthogonal variable in the quantitative correction model. The total content of unsaturated fatty acids can be expressed as a percentage of dry weight.

[0013] On the other hand, the present invention also provides a system for non-destructive detection of the total content of unsaturated fatty acids in eucalyptus leaves, for implementing the above-mentioned method, comprising: The spectral acquisition module is configured to acquire the near-infrared diffuse reflectance spectrum of eucalyptus leaf samples; The data processing module is configured to perform preprocessing on the near-infrared diffuse reflectance spectrum, including smoothing and denoising, scattering correction, derivative correction, and moisture interference subtraction. It is also configured to screen characteristic wavelengths using a competitive adaptive reweighted sampling algorithm combined with a continuous projection algorithm. The screened characteristic wavelengths have an absolute Pearson correlation coefficient with the total content of unsaturated fatty acids in eucalyptus leaves greater than 0.6, and wavelength variables within ±10 nm of the characteristic absorption peak of 1,8-cineole are excluded. Furthermore, the data processing module is configured to import the spectral data corresponding to the screened characteristic wavelengths into a quantitative correction model for content prediction. The output module is configured to output the predicted value of the total unsaturated fatty acid content of eucalyptus leaves; the quantitative correction model is established using partial least squares regression, with the total unsaturated fatty acid content of eucalyptus leaves determined by gas chromatography-mass spectrometry as the reference value.

[0014] Preferably, the spectral acquisition module includes a Fourier transform near-infrared spectrometer, an integrating sphere diffuse reflectance sampling accessory, and a leaf clamping device. The leaf clamping device includes an upper pressure plate, a lower sample stage, a constant pressure spring, and a pressure adjustment knob. The constant pressure spring is configured to apply a constant pressure to the upper pressure plate through the pressure adjustment knob, so that the eucalyptus leaf sample is in constant pressure against the integrating sphere diffuse reflectance sampling window.

[0015] Preferably, the data processing module includes a preprocessing unit, a feature extraction unit, and a model operation unit. The preprocessing unit is configured to perform smoothing and denoising, scattering correction, derivative correction, and moisture interference subtraction. The feature extraction unit is configured to perform feature wavelength selection using a competitive adaptive reweighted sampling algorithm combined with a continuous projection algorithm. The feature extraction unit is also configured to perform a competitive adaptive reweighted sampling algorithm with 500 to 1000 Monte Carlo samplings and use 5-fold cross-validation to determine the optimal subset of variables. The model operation unit is configured to store and run the quantitative correction model.

[0016] The technical effects and advantages of this invention are as follows: 1. The method provided by this invention does not require chemical pretreatment of eucalyptus leaf samples, and the entire process of a single detection takes no more than 3 minutes, enabling rapid detection of the total content of unsaturated fatty acids in eucalyptus leaves; 2. This invention employs a multi-scale combined preprocessing strategy targeting the high water content and rich essential oil content of eucalyptus leaves, and combines a competitive adaptive reweighted sampling algorithm with a continuous projection algorithm to screen feature wavelengths. This effectively eliminates spectral interference from water and essential oil components, resulting in an external prediction set determination coefficient ≥ 0.85 and an external prediction root mean square error ≤ 0.06% for the prediction model, thus ensuring the accuracy of the detection results. 3. The system provided by this invention integrates a spectral acquisition module, a data processing module, and an output module, forming a complete detection chain from spectral acquisition to result output, which is easy to operate; at the same time, the detection process does not require the use of toxic and harmful chemical reagents such as chloroform and methanol, avoiding environmental pollution and health and safety hazards, and realizing green detection. Attached Figure Description

[0017] Figure 1 This is a flowchart of the method of the present invention.

[0018] Figure 2 This is a system diagram of the present invention.

[0019] The attached diagram is labeled as follows: 10, Spectrum acquisition module; 20, Data processing module; 30, Output module. Detailed Implementation

[0020] The technical solutions of the present invention will be clearly and completely described below with reference to the accompanying drawings in the embodiments of the present invention. Example 1

[0021] As attached Figure 1 As shown, this embodiment provides a method for non-destructive testing of the total unsaturated fatty acid content in eucalyptus leaves, including the following steps: S1. Near-infrared diffuse reflectance spectrum of collected eucalyptus leaf samples The specific steps are as follows: Eucalyptus leaf samples were collected from different varieties, growth stages, and ecological conditions. In this embodiment, 120 samples were collected, including 90 calibration set samples and 30 external prediction set (test set) samples. The samples were uniformly collected from mature leaves on the sunny side of the middle of the plant, specifically the third to fifth pairs of leaves, to avoid differences between young and aging leaves. Spectroscopic acquisition was performed immediately after sample collection, or the samples were refrigerated at 4°C and analyzed within 24 hours to minimize the impact of leaf moisture changes and fatty acid oxidation on the spectra. The eucalyptus leaf samples were fresh eucalyptus leaves, and the total unsaturated fatty acid content was expressed as a percentage of the leaf's fresh weight; alternatively, the moisture content of each sample was simultaneously measured and incorporated as an orthogonal variable into the quantitative calibration model, with the total unsaturated fatty acid content expressed as a percentage of the leaf's dry weight. In this embodiment, the percentage is expressed as a percentage of fresh weight.

[0022] A Fourier transform near-infrared spectrometer was used, equipped with an integrating sphere diffuse reflectance sampling accessory. Before spectral acquisition, the instrument was powered on and warmed up for at least 30 minutes to allow it to reach a stable state.

[0023] Freshly picked eucalyptus leaves were placed flat on the integrating sphere diffuse reflectance sampling window, with the underside of the leaf facing upwards, avoiding the midrib area. A specially designed leaf clamping device was used to apply constant pressure, ensuring a tight seal between the leaf and the sampling window. Spectral acquisition was conducted under constant temperature and humidity conditions: temperature controlled at 25±1℃ and relative humidity at 45±5%. Spectral acquisition parameters were: spectral range 1000–2500 nm, resolution 8 cm⁻¹. -1 The scan was performed 64 times at a speed of 0.2 cm / s. Near-infrared diffuse reflectance spectra were collected three times for each sample, and the average spectrum was used as the original spectrum for that sample. Before each spectral acquisition, a dark background spectrum was acquired to subtract instrument dark noise.

[0024] The total unsaturated fatty acid content of each sample was determined by gas chromatography-mass spectrometry (GC-MS) as a reference value. Eucalyptus leaf samples were freeze-dried, pulverized, and then total lipids were extracted using a 2:1 (v / v) mixture of chloroform and methanol. The lipids were then methylated using KOH-CH3OH solution, followed by a secondary methylation treatment with BF3 as a catalyst. Qualitative and quantitative analysis was performed using GC-MS. The GC-MS analytical conditions were as follows: a 30m × 0.25mm × 0.25μm polyethylene glycol capillary column was used; the temperature was programmed, starting at 180℃ and increasing to 240℃ at a rate of 6℃ / min, held for 40 min; the injection port temperature was 250℃; high-purity helium was used as the carrier gas at a flow rate of 1.0 mL / min; an electron impact ionization source with an ionization energy of 70 eV was used as the mass spectrometer; and the mass scan range was m / z 50–550. The unsaturated fatty acids in eucalyptus leaves mainly include linoleic acid, linolenic acid, and oleic acid. The total content of unsaturated fatty acids is calculated as the sum of the contents of each fatty acid, expressed as a percentage of the fresh weight of the leaves. In this example, the total content of unsaturated fatty acids in 120 samples ranged from 0.08% to 0.23%.

[0025] S2. Preprocessing the near-infrared diffuse reflectance spectrum Considering that eucalyptus leaves typically have a moisture content >60% and an essential oil content that can reach 2%–5% of the leaf's dry weight, the near-infrared diffuse reflectance spectra collected in step S1 underwent multi-scale combined preprocessing. The preprocessing included the following steps: S2.1 Smoothing and Denoising Processing: The Savitzky-Golay convolution smoothing method is used to eliminate random noise, with a window width of 15 and a polynomial order of 3. S2.2 Scattering Correction Processing: Standard normal variable transformation or multivariate scattering correction is used to eliminate the scattering effects caused by differences in blade surface morphology and uneven blade thickness. In this embodiment, standard normal variable transformation is used. S2.3 Derivative Correction Processing: First-order or second-order derivative processing is used to eliminate baseline drift and enhance spectral feature information. This embodiment uses first-order derivative processing. S2.4 Moisture Interference Subtraction Processing: Orthogonal signal correction or external parameter orthogonalization is used to subtract the strong absorption interference of moisture near 1400nm and 1900nm. In this embodiment, orthogonal signal correction is used.

[0026] S3. Using a competitive adaptive reweighted sampling algorithm combined with a continuous projection algorithm, characteristic wavelengths related to the total content of unsaturated fatty acids in eucalyptus leaves are screened from the preprocessed spectrum. The specific steps are as follows: The characteristic absorption bands of eucalyptus unsaturated fatty acids are mainly located in the 1650–1800 nm and 2200–2400 nm ranges. Since 1,8-cineole, a component of eucalyptus essential oil, also exhibits characteristic absorption in these bands, it is necessary to screen the characteristic wavelengths that are truly correlated with the total content of eucalyptus unsaturated fatty acids from the pre-processed spectrum to eliminate interference from essential oil components.

[0027] In this step, the specific method for eliminating 1,8-cineole interference is as follows: During the modeling stage, the near-infrared spectrum of pure 1,8-cineole standard is pre-collected, and its characteristic absorption peak positions (mainly located around 1670nm, 1735nm, and 2255nm) are identified. Wavelength variables located within ±10nm of each characteristic absorption peak position are marked as interference variables and eliminated. Once determined during the modeling stage, these wavelength variables are fixed in the quantitative calibration model, eliminating the need to repeatedly collect pure sample spectra when detecting unknown samples.

[0028] This step employs a competitive adaptive reweighted sampling algorithm combined with a continuous projection algorithm to screen characteristic wavelengths. The competitive adaptive reweighted sampling algorithm uses Monte Carlo sampling and an adaptive reweighting strategy to screen wavelength variables with strong correlation to the target variable. The continuous projection algorithm further selects wavelength variables with minimal collinearity and maximum information content from the variables screened by the competitive adaptive reweighted sampling algorithm. The absolute value of the Pearson correlation coefficient between the screened characteristic wavelengths and the total content of eucalyptus unsaturated fatty acids is greater than 0.6. In the competitive adaptive reweighted sampling algorithm, the number of Monte Carlo samplings is set to 500–1000, and 5-fold cross-validation is used to determine the optimal subset of variables. After screening using the combined competitive adaptive reweighted sampling algorithm and the continuous projection algorithm, and removing interfering wavelengths within ±10 nm of the 1,8-cineole characteristic absorption peak, the number of obtained characteristic wavelengths is 10%–15% of the total number of wavelengths. The selected characteristic wavelengths should include at least three characteristic wavelengths in the 1650–1800 nm band and at least three characteristic wavelengths in the 2200–2400 nm band.

[0029] S4. Import the spectral data of the selected characteristic wavelengths into the pre-established quantitative correction model. The quantitative correction model outputs the predicted value of the total content of eucalyptus unsaturated fatty acids. The quantitative correction model is established using partial least squares regression and the total content of eucalyptus unsaturated fatty acids determined by gas chromatography-mass spectrometry is used as the reference value.

[0030] The specific steps are as follows: The spectral data of the characteristic wavelengths selected in step S3 were correlated with the reference values ​​of the total unsaturated fatty acid content determined in step S1, and a quantitative correction model was established using partial least squares regression. Partial least squares regression can effectively handle the multicollinearity problem among spectral variables and is suitable for modeling high-dimensional spectral data.

[0031] The optimal number of principal components is determined using leave-one-out cross-validation, and the selection is based on the principle of minimizing the root mean square error of cross-validation. In this embodiment, the optimal number of principal components is 10 to 12.

[0032] The model evaluation metrics are as follows: coefficient of determination for calibration set ≥ 0.92; coefficient of determination for cross-validation ≥ 0.85; coefficient of determination for external prediction set ≥ 0.85; root mean square error of cross-validation ≤ 0.05%; root mean square error of external prediction ≤ 0.06%; relative analysis error ≥ 2.5.

[0033] External validation was performed using 30 samples from an independent external prediction set, which were not used in the modeling process.

[0034] For the unknown eucalyptus leaf sample to be tested, the sample was collected and its near-infrared diffuse reflectance spectrum was obtained according to the method in step S1. The spectrum was preprocessed according to the method in step S2. The characteristic wavelength spectral data was extracted according to the method in step S3. The preprocessed characteristic spectral data was imported into the quantitative correction model. The quantitative correction model outputs the predicted value of the total unsaturated fatty acid content of the sample to be tested. Example 2

[0035] As attached Figure 2 As shown, this embodiment provides a system for non-destructive testing of the total unsaturated fatty acid content of eucalyptus leaves, used to implement the method of Embodiment 1. The system includes a spectral acquisition module 10, a data processing module 20, an output module 30, and a model maintenance module, the specific details of which are as follows: The spectral acquisition module 10 is configured to acquire near-infrared diffuse reflectance spectra of eucalyptus leaf samples. The spectral acquisition module 10 includes a Fourier transform near-infrared spectrometer, an integrating sphere diffuse reflectance sampling accessory, a leaf clamping device, and an environmental control unit. The Fourier transform near-infrared spectrometer includes a light source, an interferometer, and a detector. The leaf clamping device includes an upper pressure plate, a lower sample stage, a constant pressure spring, and a pressure adjustment knob. In use, fresh eucalyptus leaves are placed on the lower sample stage with the back of the leaf facing upwards. The pressure of the constant pressure spring is adjusted using the pressure adjustment knob, so that the upper pressure plate presses the leaf firmly against the integrating sphere diffuse reflectance sampling window with constant pressure, ensuring a tight fit between the leaf and the sampling window and maintaining constant pressure to avoid spectral differences caused by pressure variations. The environmental control unit includes a temperature controller and a humidity controller to maintain a constant temperature and humidity sampling environment: temperature 25±1℃ and relative humidity 45±5%.

[0036] The data processing module 20 is configured to preprocess the near-infrared diffuse reflectance spectrum, screen characteristic wavelengths, and import them into a quantitative correction model for content prediction. The data processing module 20 includes a preprocessing unit, a feature extraction unit, and a model computation unit. The preprocessing unit is configured to perform spectral preprocessing, which sequentially includes smoothing and denoising, scattering correction, derivative correction, and moisture interference subtraction. Specifically, the smoothing and denoising uses Savitzky-Golay convolution smoothing with a window width of 15 and a polynomial order of 3; the scattering correction uses standard normal variable transformation or multivariate scattering correction; the derivative correction uses first-order or second-order derivatives; and the moisture interference subtraction uses orthogonal signal correction or external parameter orthogonalization. The feature extraction unit is configured to perform feature wavelength screening. Feature wavelengths are obtained through a competitive adaptive reweighted sampling algorithm combined with a continuous projection algorithm. The absolute value of the Pearson correlation coefficient between the screened feature wavelengths and the total content of eucalyptus unsaturated fatty acids is greater than 0.6. Wavelength variables within ±10 nm of the characteristic absorption peak of 1,8-cineole are excluded. The screened feature wavelengths must include at least three characteristic wavelengths located in the 1650–1800 nm and 2200–2400 nm bands. The feature extraction unit is also configured to perform a competitive adaptive reweighted sampling algorithm with 500–1000 Monte Carlo samplings and use 5-fold cross-validation to determine the optimal subset of variables. The model computation unit is configured to store and run a quantitative calibration model established using partial least squares regression, with the total content of eucalyptus unsaturated fatty acids determined by gas chromatography-mass spectrometry as a reference value.

[0037] Output module 30 is configured to output the predicted value of the total unsaturated fatty acid content of eucalyptus leaves. Output module 30 includes a display unit, a storage unit, and a communication unit. The display unit displays the predicted value of the total unsaturated fatty acid content of the sample, the detection date, and the sample number. The storage unit stores historical detection data, supporting data traceability and model updates. The communication unit supports data export and remote transmission.

[0038] The model maintenance module, as the preferred module in this embodiment, is configured to periodically update and optimize the quantitative calibration model. Specifically, every 6 months or every 50 newly added samples, the model is refitted and optimized using the reference values ​​and near-infrared diffuse reflectance spectral data of the new samples. That is, after adding the reference values ​​and spectral data of the new samples, the partial least squares regression method is used again to establish the quantitative calibration model, continuously improving the model's applicability and prediction accuracy. The reference values ​​for the new samples are determined using gas chromatography-mass spectrometry.

[0039] The method and system for non-destructive testing of total unsaturated fatty acid content in eucalyptus leaves provided by this invention do not require any chemical pretreatment of eucalyptus leaf samples. The leaves can be retained for other experiments after testing. The testing time is short and a single test does not exceed 3 minutes. No toxic or harmful chemical reagents are used. This invention can be used for germplasm resource evaluation of detached eucalyptus leaves, rapid screening of eucalyptus leaf varieties with high unsaturated fatty acid content, and rapid assessment of eucalyptus leaf feed quality.

[0040] The above description is merely a preferred embodiment of the present invention and is not intended to limit the present invention. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of the present invention should be included within the protection scope of the present invention.

Claims

1. A method for non-destructive testing of the total unsaturated fatty acid content in eucalyptus leaves, characterized in that: Includes the following steps: S1. Collect near-infrared diffuse reflectance spectra of detached eucalyptus leaf samples; S2. Preprocess the near-infrared diffuse reflectance spectrum. The preprocessing includes smoothing and noise reduction, scattering correction, derivative correction and moisture interference subtraction. S3. Using a competitive adaptive reweighted sampling algorithm combined with a continuous projection algorithm, characteristic wavelengths related to the total content of eucalyptus unsaturated fatty acids are screened from the preprocessed spectrum. The absolute value of the Pearson correlation coefficient between the screened characteristic wavelengths and the total content of eucalyptus unsaturated fatty acids is greater than 0.

6. Wavelength variables within ±10 nm of the characteristic absorption peak of 1,8-cineole are removed. The characteristic wavelengths include at least 3 characteristic wavelength points in the 1650-1800 nm band and at least 3 characteristic wavelength points in the 2200-2400 nm band. S4. Import the spectral data of the selected characteristic wavelengths into the pre-established quantitative correction model. The quantitative correction model outputs the predicted value of the total content of eucalyptus unsaturated fatty acids. The quantitative correction model was established using partial least squares regression, with the total content of eucalyptus unsaturated fatty acids determined by gas chromatography-mass spectrometry as a reference value. The characteristic absorption peak position of 1,8-cineole was obtained by pre-determining the near-infrared spectrum of pure 1,8-cineole. During the modeling stage, the wavelength variables to be removed were determined and fixed into the quantitative correction model.

2. The method according to claim 1, characterized in that: In step S2, the smoothing and denoising process uses the Savitzky-Golay convolutional smoothing method, where the window width of the Savitzky-Golay convolutional smoothing method is 15 and the polynomial order is 3.

3. The method according to claim 1, characterized in that: In step S2, the scattering correction process employs standard normal variable transformation or multivariate scattering correction; The derivative correction process uses either the first or second derivative; Moisture interference subtraction is performed using orthogonal signal correction or external parameter orthogonalization.

4. The method according to claim 1, characterized in that: In step S4, the quantitative calibration model uses leave-one-out cross-validation to determine the number of principal components, which is 10 to 12.

5. The method according to claim 1, characterized in that: In step S4, the external prediction set determination coefficient of the quantitative calibration model is ≥0.85, the root mean square error of the external prediction is ≤0.06%, and the relative analysis error is ≥2.

5.

6. The method according to claim 1, characterized in that: In step S1, the acquisition conditions for the near-infrared diffuse reflectance spectrum are: spectral range 1000–2500 nm, resolution 8 cm⁻¹. -1 The number of scans was 64.

7. The method according to claim 1, characterized in that: Eucalyptus leaf samples are detached fresh eucalyptus leaves, collected from mature leaves on the sunny side of the middle of the plant. The total content of unsaturated fatty acids is expressed as a percentage of the fresh weight of the leaves. Alternatively, the moisture content of the eucalyptus leaf samples can be measured simultaneously and incorporated into the quantitative calibration model as an orthogonal variable, with the total content of unsaturated fatty acids expressed as a percentage of dry weight.

8. A system for non-destructive testing of the total unsaturated fatty acid content of eucalyptus leaves, used to implement the method according to any one of claims 1-7, characterized in that: include: The spectral acquisition module (10) is configured to acquire the near-infrared diffuse reflectance spectrum of eucalyptus leaf samples; The data processing module (20) is configured to perform smoothing and noise reduction, scattering correction, derivative correction and moisture interference subtraction preprocessing on the near-infrared diffuse reflectance spectrum in sequence, and is configured to screen characteristic wavelengths by combining a competitive adaptive reweighted sampling algorithm with a continuous projection algorithm, wherein the absolute value of the Pearson correlation coefficient between the screened characteristic wavelengths and the total content of eucalyptus unsaturated fatty acids is greater than 0.6, and wavelength variables within ±10 nm of the characteristic absorption peak position of 1,8-cineole are removed; the data processing module (20) is also configured to import the spectral data corresponding to the screened characteristic wavelengths into a quantitative correction model for content prediction; The output module (30) is configured to output the predicted value of the total content of unsaturated fatty acids in eucalyptus leaves; the quantitative correction model is established by partial least squares regression, and the total content of unsaturated fatty acids in eucalyptus leaves determined by gas chromatography-mass spectrometry is used as the reference value.

9. The system according to claim 8, characterized in that: The spectral acquisition module (10) includes a Fourier transform near-infrared spectrometer, an integrating sphere diffuse reflectance sampling accessory, and a leaf clamping device. The leaf clamping device includes an upper pressure plate, a lower sample stage, a constant pressure spring, and a pressure adjustment knob. The constant pressure spring is configured to apply constant pressure to the upper pressure plate through the pressure adjustment knob, so that the eucalyptus leaf sample is attached to the integrating sphere diffuse reflectance sampling window with constant pressure.

10. The system according to claim 8, characterized in that: The data processing module (20) includes a preprocessing unit, a feature extraction unit, and a model operation unit. The preprocessing unit is configured to perform smoothing and denoising, scattering correction, derivative correction, and moisture interference subtraction. The feature extraction unit is configured to perform feature wavelength screening using a competitive adaptive reweighted sampling algorithm combined with a continuous projection algorithm. The feature extraction unit is also configured to perform a competitive adaptive reweighted sampling algorithm with 500 to 1000 Monte Carlo samplings and use 5-fold cross-validation to determine the optimal subset of variables. The model operation unit is configured to store and run the quantitative correction model.

Images