Rapid nondestructive detection method for phellinus igniarius polysaccharide content based on near infrared spectrum technology

Abstract

The invention relates to the technical field of trace detection, in particular to a rapid nondestructive detection method for phellinus igniarius polysaccharide content based on a near infrared spectrum technology. The method comprises the following steps: acquiring an original spectrum set of phellinus igniarius samples from various sources under a near infrared spectrum, and dividing the samples from the same source into the same group of original spectrum curves; evaluating the stability and distinguishing capability of light absorption change in different wavelength intervals, and obtaining the light absorption credibility of each target wavelength of the original spectrum curve; analyzing the morphological difference of a single curve in each group of spectrum curves relative to the whole group of curves, and obtaining an initial judgment coefficient for curve smoothing processing; and combining the light absorption credibility with the initial judgment coefficient, performing corresponding smoothing processing on each original spectrum curve according to the determined window length and processing order, and analyzing the actual polysaccharide content of the phellinus igniarius sample of the corresponding type. According to the technical scheme, the accuracy of polysaccharide content detection of phellinus igniarius through near infrared spectroscopy is improved.

Description

Technical Field

[0001] This invention relates to the technical field of trace detection, specifically to a rapid and non-destructive method for detecting the content of Phellinus linteus polysaccharide based on near-infrared spectroscopy. Background Technology

[0002] Phellinus linteus, a precious medicinal and edible fungus, has its main active ingredient being Phellinus linteus polysaccharides, and polysaccharide content is a core indicator for evaluating its quality. Currently, the detection of Phellinus linteus polysaccharide content mainly relies on chemical analysis methods such as the sulfuric acid-anthrone method and the sulfuric acid-phenol method. Although these methods have high detection accuracy, they suffer from drawbacks such as cumbersome operation procedures, long detection cycles, the need for large amounts of chemical reagents, and destructive effects on samples. They cannot meet the needs for rapid batch detection in scenarios such as Phellinus linteus raw material screening, cultivation process monitoring, and processing technology optimization.

[0003] Near-infrared spectroscopy (NIRS) is a rapid, non-destructive, and environmentally friendly analytical technique. By detecting the overtone and combination frequency absorption spectra of hydrogen-containing groups (OH, CH, etc.) in a sample, it can quickly obtain information about the chemical composition of the sample. The core of rapid and non-destructive detection of polysaccharide content based on NIRS lies in acquiring the near-infrared spectrum of the sample and establishing a quantitative model using chemometric algorithms, achieving rapid and non-destructive quantification without complex pretreatment. However, the signal intensity of near-infrared spectra is relatively weak, making it susceptible to interference during the detection of Phellinus linteus, leading to significant errors in the detection of Phellinus linteus polysaccharide content. Summary of the Invention

[0004] To address the technical problem of improving the accuracy of polysaccharide content detection in Phellinus linteus using near-infrared spectroscopy, this invention aims to provide a rapid and non-destructive method for detecting the polysaccharide content of Phellinus linteus based on near-infrared spectroscopy. The specific technical solution adopted is as follows: This invention provides a rapid and non-destructive method for detecting the content of Sanghuang polysaccharides based on near-infrared spectroscopy. The method includes: Obtain the original spectral set of Phellinus linteus samples from diverse sources under near-infrared spectroscopy; each Phellinus linteus sample from the same source in the original spectral set constitutes the same set of original spectral curves; Based on the differences in absorption characteristics of spectral curves within the same group and between different groups in the original spectral set, the absorbance reliability of each target wavelength of the original spectral curve is obtained. Based on the morphological differences between a single curve and the entire set of spectral curves in each group, the initial judgment coefficients for smoothing the original spectral curves in their respective wavelength ranges are obtained. Based on the absorbance confidence level and the initial judgment coefficient, the window length and processing order of each original spectral curve are obtained; The original spectral curves were smoothed according to the window length and processing order, and the actual polysaccharide content of the corresponding Phellinus linteus samples was determined.

[0005] In one optional embodiment, obtaining the original spectral set of Phellinus linteus samples from diverse sources under near-infrared spectroscopy includes: Read the sample labels of Sanghuang samples from diverse sources, which are Sanghuang samples from different growth stages, different origins and different drying methods; Configure the working parameters for near-infrared spectrometer sampling using the integrating sphere diffuse reflectance method based on the sample label; the working parameters include spectral range, resolution, and number of scans. The near-infrared spectrometer was operated under operating parameters to sample and average the spectral curves of the Phellinus linteus samples corresponding to each sample label, and the data were grouped and stored as the original spectral curves of the corresponding sample labels; the original spectral curves are curves of spectral wavelength as a function of absorbance. Based on the set of curves formed by all the original spectral curves under each sample label, the original spectral set of Sanghuang samples from diverse sources is obtained.

[0006] In one optional embodiment, the absorbance confidence of each target wavelength of the original spectral curve is obtained based on the differences in absorbance characteristics of spectral curves within the same group and between groups in the original spectral set, including: Multiple target wavelength ranges are configured based on the wavelengths that are sensitive to polysaccharide detection in each original spectral curve in the original spectral set; Based on the absorbance difference between two original spectral curves of the same group of Sanghuang samples within the target wavelength range, the non-correlation differences of each target wavelength within the corresponding target wavelength range are obtained; Based on the non-correlation difference between the two original spectral curves of the same group of Sanghuang samples for each target wavelength and the Hu invariant moment, spectral noise interference analysis was performed to obtain the unreliability of each target wavelength in the two original spectral curves. Statistical analysis was performed on all the unreliability scores for each target wavelength in each target wavelength range to obtain the absorbance reliability of each target wavelength in the corresponding target wavelength range.

[0007] In one optional embodiment, spectral noise interference analysis is performed based on the non-correlation differences and Hu's invariant moments of the two original spectral curves of the same group of Sanghuang samples for each target wavelength, to obtain the unreliability of each target wavelength in the two original spectral curves, including: Based on the non-correlation difference between the two original spectral curves at each target wavelength and the Hu invariant moment, the noise interference index for the corresponding target wavelength is obtained. The unreliability of each target wavelength in the two original spectral curves is obtained by averaging all noise interference indicators of the two original spectral curves.

[0008] In one optional embodiment, statistical analysis is performed on all the unreliability scores for each target wavelength in each target wavelength range to obtain the absorbance reliability of each target wavelength in the corresponding target wavelength range, including: Based on the maximum and average values ​​of all unreliability scores for each target wavelength in each target wavelength range, the unreliability index for the corresponding target wavelength is obtained. Based on the number of problem groups with a disbelief level greater than the disbelief threshold and the total number of sample groups for each target wavelength range, the proportion of problem groups for each target wavelength range is obtained. The unreliable indicators and problem group proportions for each target wavelength range are normalized and their credibility is converted to obtain the absorbance credibility of each target wavelength in the corresponding target wavelength range.

[0009] In one optional embodiment, based on the morphological differences between a single curve and the entire set of spectral curves, an initial judgment coefficient for smoothing the original spectral curve within its wavelength range is obtained, including: Each set of original spectral curves is normalized to obtain a reference spectral curve for the target wavelength range. Weighted analysis is performed based on the instability characteristics among the reference spectral curves to obtain the standard spectral curves of the corresponding group of original spectral curves. Based on the peak width distribution characteristics of each standard spectral curve, the peak width characteristic value of the corresponding standard spectral curve is obtained; Based on the shape deformation, peak asymmetry, and peak width characteristics of each group of Sanghuang samples in their respective target wavelength ranges, the initial judgment coefficients for the corresponding target wavelength ranges are obtained.

[0010] In one optional embodiment, a weighted analysis is performed based on the instability characteristics among the reference spectral curves to obtain a standard spectral curve for the corresponding group of original spectral curves, including: The instability of the current reference spectral curve is obtained by taking the mean of the Hughes invariant moments of the current reference spectral curve and other reference spectral curves in each target wavelength range of each set of original spectral curves. Based on the overall impact of the instability of the current reference spectral curve on the same group of Sanghuang samples, the stability weight of the current reference spectral curve is obtained. The standard spectral curve for the corresponding group of original spectral curves is obtained by summing the stability weights of all reference spectral curves in each target wavelength range of each group of original spectral curves.

[0011] In one optional embodiment, the peak width characteristic value of the corresponding standard spectral curve is obtained based on the peak width distribution characteristics of each standard spectral curve, including: For each standard spectral curve, the trough distribution is detected and the wavelength difference between adjacent trough points is calculated to obtain the width of several peaks for each standard spectral curve. Based on a preset width threshold, several peak widths are divided into wide peaks and narrow peaks, and the peak width characteristic value of the corresponding standard spectral curve is obtained based on the quantity distribution characteristics of wide peaks and narrow peaks.

[0012] In one optional embodiment, the window length and processing order of each original spectral curve are obtained based on the absorbance confidence level and the initial judgment coefficient, including: Based on the absorbance confidence level and the initial judgment coefficient, the target judgment coefficients for each original spectral curve are obtained after SG smoothing. Based on the current target judgment coefficient and the preset target mapping relationship, the window length and processing order of the original spectral curve corresponding to the current target judgment coefficient are obtained; the target mapping relationship is the preset correspondence between different target judgment coefficients and window length and processing order.

[0013] In one optional embodiment, the target judgment coefficient for SG smoothing of each original spectral curve is obtained based on the absorbance confidence level and the initial judgment coefficient, including: Calculate the weighted Euclidean distance of the absorbance confidence between any two target wavelengths, and obtain the key indicators of the corresponding target wavelength range from all the weighted Euclidean distances in each target wavelength range; We perform weighted statistics based on the key indicators and initial judgment coefficients of each target wavelength range for each Sanghuang sample to obtain the target judgment coefficient of the original spectral curve corresponding to each Sanghuang sample.

[0014] The present invention has the following beneficial effects: The technical solution of this invention obtains the original spectral set of Phellinus linteus samples from diverse sources under near-infrared spectroscopy, and divides samples from the same source into the same group of original spectral curves, so that the spectral differences within the group mainly reflect random noise and measurement disturbances; by comparing the differences in the absorbance characteristics of the original spectral curves within the same group and between groups, the stability and distinguishing ability of absorbance changes in different wavelength ranges are evaluated, and the absorbance reliability of each target wavelength of the original spectral curve is obtained; further, the morphological differences of individual curves in each group of spectral curves relative to the whole group are analyzed to determine the stability of the original spectral curve in the corresponding wavelength range, so as to obtain the results for curve smoothing. The initial judgment coefficient identifies spectral curves with significant morphological deviations as objects with strong noise interference. Combining absorbance confidence with the initial judgment coefficient, and ensuring that polysaccharide characteristic absorption is not weakened, the window length and processing order of each original spectral curve are adaptively determined. This prioritizes information preservation in important wavelength ranges while focusing on smoothing and suppressing noise in areas with strong noise. Based on the determined window length and processing order, each original spectral curve is smoothed accordingly, effectively suppressing random noise and making the absorption of polysaccharide-related characteristics clearer and more stable. Based on this, the actual polysaccharide content of the corresponding *Phellinus linteus* sample is obtained. This technical solution analyzes the key importance of polysaccharide content detection in the wavelength range of *Phellinus linteus* evolution spectrum, as well as the interference from noise. Adaptive curve smoothing parameters improve the spectral filtering effect, resulting in high-quality updated spectra, thereby improving the accuracy of polysaccharide content detection in *Phellinus linteus* using near-infrared spectroscopy. Attached Figure Description

[0015] To more clearly illustrate the technical solutions and advantages in the embodiments of the present invention or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0016] Figure 1 A flowchart illustrating a rapid and non-destructive method for detecting the content of Phellinus linteus polysaccharides based on near-infrared spectroscopy, provided in an embodiment of the present invention; Figure 2 A schematic diagram illustrating the grouping of Phellinus linteus based on different sample sources, provided as an embodiment of the present invention; Figure 3 A flowchart illustrating the calculation of absorbance confidence level according to an embodiment of the present invention; Figure 4 A flowchart illustrating the calculation of the initial judgment coefficients provided in one embodiment of the present invention; Figure 5 This is a flowchart illustrating the calculation of window length and processing order according to an embodiment of the present invention. Detailed Implementation

[0017] To further illustrate the technical means and effects adopted by the present invention to achieve its intended purpose, the following, in conjunction with the accompanying drawings and preferred embodiments, details the specific implementation, structure, features, and effects of a rapid and non-destructive detection method for the content of Phellinus linteus polysaccharides based on near-infrared spectroscopy according to the present invention. In the following description, different "one embodiment" or "another embodiment" do not necessarily refer to the same embodiment. Furthermore, specific features, structures, or characteristics in one or more embodiments can be combined in any suitable form.

[0018] Unless otherwise defined, all technical and scientific terms used herein have the same meaning as commonly understood by one of ordinary skill in the art to which this invention pertains.

[0019] Rapid and non-destructive detection of Sanghuang polysaccharide content based on near-infrared spectroscopy is suitable for screening, processing quality control, and quality grading of Sanghuang raw materials. However, the signal intensity of near-infrared spectroscopy is relatively weak, making it susceptible to instrument noise (e.g., dark current of the instrument, light source fluctuations), environmental interference (e.g., temperature changes, stray light), and intrinsic factors of the Sanghuang sample (e.g., uneven particle size, scattering effects), manifesting as "spiculation" characteristics or irregular fluctuations in the spectral curve. This noise can mask the absorption characteristic peaks generated by the interaction of Sanghuang polysaccharides with near-infrared light (e.g., overtone and combination absorption of CH and OH groups), interfering with the identification and extraction of characteristic peaks and leading to significant errors in the actual polysaccharide content detection.

[0020] The technical solution of this invention analyzes the key wavelength range of polysaccharide content detection in the evolution spectrum of *Sanghuang* (a type of medicinal mushroom), focusing on its key wavelength range and the interference from noise. It then derives parameters for adaptive smoothing processing, improving the filtering effect of the spectrum and obtaining a high-quality updated spectrum, thus ensuring the accuracy of subsequent *Sanghuang* polysaccharide content detection. The specific solution of a rapid and non-destructive detection method for *Sanghuang* polysaccharide content based on near-infrared spectroscopy, provided by this invention, will be described in detail below with reference to the accompanying drawings.

[0021] Please see Figure 1 , Figure 1 This invention provides a flowchart of a rapid, non-destructive detection method for the polysaccharide content of *Sanghuang* (a type of medicinal mushroom) based on near-infrared spectroscopy. This method can be applied to a detection terminal for non-destructive testing of the polysaccharide content of *Sanghuang*. The detection terminal can be a computer or server capable of data processing and running the method; no specific limitations are imposed here. The detection method includes: S11: Obtain the original spectral set of Phellinus linteus samples from various sources under near-infrared spectroscopy; each Phellinus linteus sample from the same source in the original spectral set constitutes the same set of original spectral curves.

[0022] Specifically, the diverse sources of the Sanghuang samples included different growth stages and origins. Non-destructive spectral acquisition of each Sanghuang sample was performed using a near-infrared spectrometer to obtain raw spectral curves reflecting the distribution characteristics of the internal chemical components. Each Sanghuang sample was scanned multiple times, and the scan results were averaged to reduce the impact of random measurement errors. After obtaining the raw spectral curves, they were grouped according to the source attributes of the Sanghuang samples, with samples from the same source grouped together. Since Sanghuang samples from the same source exhibit consistent trends in biological composition and polysaccharide content, the differences between raw spectral curves within the same group mainly stem from instrument noise, sample surface condition, and microstructural differences, and are not affected by origin or growth stage.

[0023] For example, step S11 includes sub-steps S11-1 to S11-4, which are described in detail below: S11-1: Read the sample tags of Sanghuang samples from diverse sources. These diverse sources include Sanghuang samples (fruiting bodies, mycelia, etc.) from different growth stages, origins, and drying methods. Sanghuang samples from different sources exhibit diversity characteristics, with varying polysaccharide content. Therefore, for the obtained Sanghuang sample set, each sample corresponds to a sample tag. The types of sample tags include growth stage, origin, and drying method. At least 10 Sanghuang samples from the same growth stage, origin, and drying method are required to ensure reliable estimation and statistical inference of natural variation under the same conditions. Please refer to [link to relevant documentation]. Figure 2 Samples of Sanghuang with the same sample label are grouped together. These samples can be numbered, for example, A1, A2…A10; B1, B2…B10; C1, C2…C10 (each group in the diagram only shows two samples). Each Sanghuang sample in the set undergoes uniform pretreatment: impurities are removed, it is cut into uniform small pieces, dried at a constant temperature (to avoid damaging the polysaccharide structure of the Sanghuang sample with high temperature), pulverized, and sieved (e.g., 40-60 mesh) to obtain a uniform powder sample.

[0024] S11-2: Configure the working parameters for integrating sphere diffuse reflectance sampling using a near-infrared spectrometer based on the sample label; the working parameters include spectral range, resolution, and number of scans. When using integrating sphere diffuse reflectance sampling, a near-infrared spectrometer can be selected, commonly a Fourier transform near-infrared spectrometer (FT-NIRS); spectral range: 1000-2500 nm (this range contains the combination and overtone absorption peaks of CH, OH, NH groups in polysaccharide molecules, and is the sensitive range for polysaccharide detection); resolution: 4-16 cm⁻¹ -1 (Higher resolution provides richer spectral information, but takes longer to acquire); Number of scans: 32-128 (multiple scans are superimposed to reduce noise interference).

[0025] S11-3: Operate the near-infrared spectrometer under operating parameters to sample and average the spectral curves of the Sanghuang samples corresponding to each sample label a number of times, and group and store them as the original spectral curves of the corresponding sample labels; the original spectral curves are curves of spectral wavelength changing with absorbance. During spectral acquisition, each Sanghuang powder sample from the Sanghuang sample set needs to be placed in a quartz sample cup, compacted to fill it completely, avoiding air bubbles and voids. Then, the sample cup is placed in the integrating sphere sample cell, and near-infrared spectra are acquired. Each Sanghuang powder sample is acquired three times, and the average spectrum is taken as the original spectral curve of that Sanghuang powder sample.

[0026] S11-4: Obtain the original spectral set of Phellinus linteus samples from diverse sources based on the curve set composed of all original spectral curves under each sample label. Each sample label can be configured with corresponding label variables, and the original spectral set can be obtained by storing the data based on the correspondence between the label variables and the original spectral curves.

[0027] If the number of original spectral curves under a certain sample label is less than a preset threshold (e.g., 3), then no further statistical analysis will be performed on that group to ensure the reliability of the statistical results.

[0028] At this point, the original spectral set of the Sanghuang sample under near-infrared spectroscopy has been obtained, and we proceed to step S12.

[0029] S12: Based on the differences in absorption characteristics of spectral curves within the same group and between groups in the original spectral set, obtain the absorption reliability of each target wavelength of the original spectral curve.

[0030] Specifically, when detecting the polysaccharide content of *Phellinus linteus* using near-infrared spectroscopy, the key wavelength ranges (or wavelength intervals) to focus on are very clearly defined and are common knowledge in plant polysaccharide research. The main wavelength ranges of interest include: around 1100 nm (O–H overtones), around 1400 nm (O–H combination overtones), 1500–1600 nm (C–H vibrations), and around 1900 nm (O–H and C–H combination overtones). These bands are sensitive to the hydroxyl groups and sugar ring structures in polysaccharides and can effectively reflect changes in *Phellinus linteus* polysaccharide content. The original spectral curves are divided according to preset or empirically determined wavelength intervals. Within each wavelength interval, the consistency of absorbance changes within the same group of spectral curves and the degree of absorbance differentiation between different groups of spectral curves are statistically analyzed. When the absorbance changes within the same group of spectral curves are relatively concentrated within a certain wavelength interval, while the absorbance differences between different groups are significant, it indicates that the absorbance information within that wavelength interval is stable and has a strong ability to characterize polysaccharide content. Based on the above analysis results, the absorbance reliability is calculated for each wavelength range to quantify the reliability of absorbance information within that range. The absorbance reliability reflects the degree of noise interference in the spectral signal within that wavelength range, as well as its effectiveness in distinguishing the content of *Sanghuang* polysaccharides, thus providing a quantitative basis for subsequent differentiated data processing strategies for different wavelength ranges.

[0031] For example, please refer to Figure 3 Step S12 includes sub-steps S12-1 to S12-4, which are described in detail below: S12-1: Configure multiple target wavelength ranges based on the wavelengths sensitive to polysaccharide detection in each original spectral curve of the original spectral set. Set multiple key target wavelength ranges as [1050, 1150], [1350, 1450], [1500, 1600], and [1850, 1950] (unit: nm), and mark each target wavelength range accordingly to specifically enhance the signal-to-noise ratio in each key target wavelength range.

[0032] S12-2: Based on the absorbance difference between two original spectral curves of the same group of *Sanghuang* samples within the target wavelength range, obtain the non-correlation differences of each target wavelength within the corresponding target wavelength range. In the rapid and non-destructive detection of *Sanghuang* polysaccharide content based on near-infrared spectroscopy, the degree of difference in spectral curves of different *Sanghuang* samples within the same wavelength range can be used to assess the importance of that target wavelength range. The greater the difference in the original spectral curves, the higher the sensitivity of that target wavelength range to the differences in polysaccharide content among *Sanghuang* samples, and the more effective information it contains; therefore, it is more important in establishing quantitative models. Therefore, it is necessary to analyze the importance of spectral data within each key target wavelength range to ensure that subsequent filtering prioritizes enhancing the effective information of more important ranges.

[0033] In the Sanghuang sample set, Sanghuang samples with the same label (same growth stage, same origin, same drying method) are grouped together. Taking any two Sanghuang samples A1 and A2 in any Sanghuang sample group as an example, any wavelength within the target wavelength range is designated as the target wavelength. The absolute value of the difference in absorbance between A1 and A2 at the target wavelength is recorded as the non-correlation difference. This difference mainly reflects random fluctuations in spectral morphology other than concentration changes.

[0034] In practical applications, the absorbance of Phellinus linteus samples at the same wavelength should be similar. Therefore, when the absorbance at the same wavelength is significantly different, it may be due to noise interference. However, due to natural variations under the same conditions, Phellinus linteus samples from the same growth stage, origin, and drying method have basically the same main chemical components (i.e., similar shapes of the original spectral curves). But the content of each component may still have certain differences (differences in absorbance at the same wavelength). Therefore, the greater the difference in the shape of the original spectral curve and the greater the difference in absorbance at the same wavelength, the more likely there is noise interference. Therefore, it is necessary to adjust the non-correlated differences based on the similarity of the original spectral curve shape.

[0035] S12-3: Spectral noise interference analysis is performed based on the non-correlation differences and Hu's invariant moments of the two original spectral curves of the same group of Sanghuang samples at each target wavelength to obtain the unreliability of each target wavelength in the two original spectral curves. Hu's invariant moments are morphological feature parameters calculated based on image or curve geometric moments. They remain unchanged under translation, rotation, and scale changes, and can stably characterize the overall shape characteristics of the curve. In this scheme, they are used to quantify the morphological differences and noise disturbance levels of the two original spectral curves. A preset window length can be set, for example, 21 wavelengths. A target window with a preset length centered on the target wavelength is obtained. The Hu's invariant moments of the original spectral curves A1 and A2 (curve segments in the two-dimensional spectral graph) within the target window are then obtained. The smaller the Hu's invariant moments, the more similar the line segment shapes of the two original spectral curves. The target window can extend beyond the wavelength range of interest. Given a spectral range of 1000-2500 nm, there must be 21 wavelengths within the target window. Unreliability characterizes the degree of confidence in two original spectral curves based on the difference in absorbance. The higher the unreliability, the lower the degree of confidence, and the less emphasis subsequent analysis should focus on; conversely, the lower the unreliability, the higher the degree of confidence.

[0036] It should be noted that Hu's invariant moments are typically used to analyze two-dimensional images. The horizontal axis of the original spectral curve represents wavelength, and the vertical axis represents absorbance. The wavelength and absorbance corresponding to each data point in a segment of the two-dimensional spectrum are normalized to eliminate dimensions, serving as the pixel's coordinates (X, Y). The absorbance is then used as the grayscale value, equivalent to an edge line in the two-dimensional image, from which Hu's invariant moments are calculated.

[0037] Furthermore, sub-step S12-3, based on the above method, derives the unreliability of each target wavelength in the two original spectral curves, specifically including: The first step is to obtain the noise interference index for each target wavelength based on the non-correlation difference between the two original spectral curves and the Hughes invariant moments. The non-correlation difference reflects random perturbations at the amplitude level, while the Hughes invariant moments reflect structural perturbations at the morphological level. The mean of the normalized values ​​of the non-correlation difference and the normalized values ​​of the Hughes invariant moments can be denoted as the noise interference index.

[0038] Specifically, since any two Sanghuang samples in each Sanghuang sample group exhibit uncorrelated differences and Hu's invariant moments at the same target wavelength within the key target wavelength range, the min-max normalization method is used to normalize the uncorrelated differences and Hu's invariant moments between any two Sanghuang samples in all Sanghuang sample groups across all wavelengths.

[0039] It should be noted that the min-max normalization method is a normalization process used to eliminate differences in the numerical scales of different indicators. In the technical solution of this invention, it is necessary to summarize the original indicator values ​​of the same type calculated under all Sanghuang sample groups and all target wavelengths or target wavelength ranges. For example, differences in non-correlation or Hu's invariant moments are considered to form a complete set of corresponding indicators. Subsequently, the minimum and maximum values ​​of each type of indicator are determined in this complete set to characterize the lower and upper limits of the indicator's value within the overall data range. For any specific Sanghuang sample, the indicator value is linearly mapped to a unified numerical range of 0-1 by subtracting the minimum value from the complete set and then dividing by the difference between the maximum and minimum values. After the above processing, the indicator values ​​under different samples, different target wavelengths, or different target wavelength ranges are all at the same directly comparable scale, and the normalized numerical values ​​can intuitively reflect the relative level of the corresponding indicator in the overall data distribution.

[0040] The second step involves obtaining the unreliability of each target wavelength for the two original spectral curves based on the average of all noise interference indices. For any given group of Sanghuang samples, at the target wavelength, the average noise interference between any two Sanghuang samples is obtained and recorded as the unreliability of the difference analysis between the two original spectral curves. The higher the unreliability obtained from the difference analysis, the less reliable the absorbance of this Sanghuang sample group at this wavelength will be in subsequent comparative analyses of Sanghuang samples; conversely, the higher the reliability.

[0041] S12-4: Perform statistical analysis on all unreliability values ​​for each target wavelength in each target wavelength range to obtain the absorbance reliability of each target wavelength in the corresponding target wavelength range. Since unreliability and absorbance reliability are opposite data representations, all unreliability values ​​in each target wavelength range can be analyzed, and the absorbance reliability of each target wavelength can be derived based on the distribution characteristics of all unreliability values.

[0042] Specifically, sub-step S12-4 includes: The first step is to obtain the unreliability index for each target wavelength based on the maximum and average values ​​of all unreliability scores for each target wavelength within each target wavelength range. Each target wavelength range contains multiple target wavelengths, and the maximum value of all unreliability scores for each target wavelength is denoted as... The average value is denoted as The unreliable indicator of the target wavelength is .

[0043] The second step involves determining the proportion of problem groups for each target wavelength within each target wavelength range based on the number of problem groups with unreliability scores exceeding the unreliability threshold and the total number of sample groups. For example, taking a specific target wavelength as an example, with an unreliability threshold of 0.7, the *Sanghuang* sample groups with an unreliability score greater than 0.7 at the target wavelength are designated as problem groups for that wavelength. The number of problem groups, S1, and the total number of sample groups for that target wavelength, S, are then statistically determined. The proportion of problem groups for that target wavelength is then calculated as follows: .

[0044] The third step involves normalizing and converting the unreliable indicators and problem group proportions for each target wavelength range to obtain the absorbance reliability of each target wavelength within the corresponding target wavelength range. This can be based on the formula: The absorbance confidence level of the target wavelength was calculated. , The minimum-maximum normalization function is used to calculate the results for all target wavelengths within the target wavelength range of key interest. Normalization is then performed. As can be seen from the above formula, the more problem groups there are at the target wavelength, and the greater the unreliability of the difference analysis, the less reliable the absorbance at the target wavelength becomes. Similarly, based on the above method, the absorbance reliability for each target wavelength can be obtained.

[0045] At this point, the absorbance reliability of each target wavelength in the original spectral curve has been obtained based on the above method, and we proceed to step S13.

[0046] S13: Based on the morphological differences between a single curve and the entire set of curves in each group of spectral curves, obtain the initial judgment coefficients for smoothing the original spectral curves within their respective wavelength ranges.

[0047] Specifically, it is necessary to analyze the interference of noise on the original spectral curve within each key target wavelength range, and then process the interference accordingly through smoothing. In the technical solution of this invention, an appropriate smoothing method can be selected based on actual needs. For example, SG (Savitzky-Golay) smoothing can be used to smooth the original spectral curve. In SG smoothing of near-infrared spectroscopy, the window size and polynomial order are the core parameters determining the smoothing effect and the degree of spectral feature preservation; the two are complementary and need to be adjusted together. Therefore, based on the absorbance reliability of polysaccharide content in each key target wavelength range, the detection criticality, and the interference of noise, adaptive filtering parameters can be calculated to effectively filter out noise while preserving spectral trends and characteristic peak information, and simultaneously minimize the morphological distortion of the original spectral curve.

[0048] It is understandable that, based on the same set of spectral curves, it is necessary to analyze the morphological differences of individual spectral curves relative to the entire set. By normalizing the spectral curves in the same set, a reference curve structure that reflects the overall spectral characteristics of the group of Sanghuang samples is obtained. Individual spectral curves are compared with the reference spectrum in each wavelength range to analyze deviations in peak position, peak width, curve undulation, and overall trend. When the morphological difference between an individual spectral curve and the entire set of spectra is significant, it indicates that the curve is significantly affected by random noise, sample surface inhomogeneity, or measurement anomalies. Based on the above morphological difference analysis results, initial judgment coefficients are determined for each original spectral curve within its respective wavelength range. These initial judgment coefficients characterize the necessity and intensity of smoothing the spectral curve within the current wavelength range, providing a basis for the adaptive determination of subsequent smoothing parameters.

[0049] For example, please refer to Figure 4 Step S13 includes sub-steps S13-1 to S13-4, which are described in detail below: S13-1: Normalize each set of original spectral curves to obtain a reference spectral curve for the target wavelength range. In this embodiment, to eliminate the influence of differences in absolute absorbance intensity among the original spectral curves of different Sanghuang samples and to highlight the relative morphological characteristics of the spectrum within the target wavelength range, normalize each set of original spectral curves within the target wavelength range K of interest. Within any target wavelength range K of interest, the reference spectral curve is obtained by normalizing the original spectral curve of each Sanghuang sample using the minimum-maximum normalization method.

[0050] S13-2: Weighted analysis is performed based on the instability characteristics among the reference spectral curves to obtain standard spectral curves for the corresponding group of original spectral curves. Since the Sanghuang samples in the same group are chemically consistent, the overall morphology of their reference spectral curves should be highly similar; when individual reference spectral curves are affected by noise interference or measurement anomalies, their morphological stability will significantly decrease. Based on this, a weighted analysis is performed on the instability characteristics among the reference spectral curves, that is, the weight value of each reference spectral curve in the weighting calculation is determined according to the degree of morphological deviation of each reference spectral curve relative to other curves in the same group. Reference spectral curves with higher stability occupy a larger weight in the weighting process, while the weight of curves with higher instability decreases accordingly, and the sum of the overall weights is 1. By weighted superposition of the reference spectral curves, a standard spectral curve that can represent the overall spectral characteristics of the Sanghuang samples in the target wavelength range is obtained.

[0051] For example, the steps to derive a standard spectral curve include: The first step is to obtain the instability of the current reference spectral curve based on the mean of the Hughes invariant moments of the current reference spectral curve and other reference spectral curves in each target wavelength range of each original spectral curve group. For any group of Sanghuang samples U, obtain the reference spectral curve of the i-th Sanghuang sample. The mean of the Hu's invariant moments of the reference spectral curves (curves in the two-dimensional spectral graph) of all other Sanghuang samples is denoted as the reference spectral curve of the i-th Sanghuang sample. instability Since Sanghuang samples from the same origin and using the same drying method have essentially the same main chemical components (similar original spectral curve shapes), the lower the instability, the more likely this reference spectral curve is to be the standard spectral curve for this type of Sanghuang sample.

[0052] The second step is to obtain the stability weight of the current reference spectral curve based on its overall impact on the same group of Sanghuang samples. The instability of the current reference spectral curve is denoted as... Instability to all reference spectra Take the top, and you will get the bottom of it. ,in For a preset minimum positive number (e.g.) This is used to prevent computational overflow. Sum normalization is then applied to the reference spectral curve segments of all Sanghuang samples within sample group U. After normalization, the sum of all normalized values ​​is 1, from which the normalized stability weights are derived. Since the sum of the stability weights of all Sanghuang samples in sample group U is 1, subsequent weighted calculations can be performed based on the stability weights of each Sanghuang sample in the same group U.

[0053] The third step involves summing the stability weights of all reference spectral curves for each target wavelength range of the original spectral curves in each group to obtain the standard spectral curve for that group. This can be done using the formula: Calculate the absorbance at each target wavelength in the standard spectral curve. .in, Let I be the absorbance at a target wavelength of the reference spectral curve corresponding to the i-th Sanghuang sample, and let I be the number of Sanghuang samples in sample group U. Greater stability corresponds to a greater weight. A weighted sum is performed on all reference spectral curves within sample group U to obtain the standard spectral curve. Here, a weighted average of multiple curves is taken to obtain the mean curve. Each curve corresponds to a weight, and the sum of the weights of all curves is 1.

[0054] For example, curves a and b have weights of 0.4 and 0.6 respectively. By weighting 0.4a + 0.6b, the average curve, i.e., the standard spectral curve for this group, can be obtained. When calculating the standard spectral curve based on the reference spectral curve, the weighted data is the absorbance of each target wavelength of the reference spectral curve. The wavelengths corresponding to each absorbance remain unchanged. Thus, the standard spectral curve is obtained based on the weighted calculation method, and the standard spectral curve has the same wavelengths as the reference spectral curve. It should be noted that due to natural variations under the same conditions, the standard spectral curve cannot be directly used as the standard curve for analysis of each Sanghuang sample within the Sanghuang sample group. This will be further explained below.

[0055] S13-3: Based on the peak width distribution characteristics of each standard spectral curve, obtain the peak width characteristic value of the corresponding standard spectral curve. By detecting the waveform structure of each standard spectral curve within the target wavelength range, identify the positions of adjacent troughs in the curve, and determine the actual width of each peak based on the wavelength difference corresponding to adjacent troughs. This yields multiple peak width parameters of the standard spectral curve within the target wavelength range, and further classification based on these multiple peak width parameters results in the peak width characteristic value.

[0056] Furthermore, sub-step S13-3 includes: The first step is to detect the trough distribution of each standard spectral curve and calculate the difference in wavelength between adjacent trough points to obtain the peak widths of each standard spectral curve. A trough detection algorithm can be used to sequentially obtain the standard spectral curves. The troughs in the wave pattern form a trough sequence. In the trough sequence, the absolute value of the difference in wavelength between adjacent troughs is obtained as a peak width, thus yielding several peak widths.

[0057] The second step involves dividing the peak widths into broad and narrow peaks based on a preset width threshold, and obtaining the peak width characteristic values ​​of the corresponding standard spectral curves based on the distribution characteristics of the number of broad and narrow peaks. The width threshold is set to 0.6; peaks with a normalized width greater than 0.6 are classified as broad peaks, otherwise as narrow peaks. The min-max normalization method is used to normalize the peak widths of all peaks corresponding to the standard spectral curves of any given *Phellinus linteus* sample group across all key target wavelength ranges.

[0058] Before detecting the peak width distribution characteristics, the standard spectral curve is first processed by second derivative or envelope removal to highlight the peak and trough characteristics of overlapping peaks.

[0059] For example, if the standard spectral curve If only broad peaks exist within the range, then let the peak width eigenvalue be... =1; if the standard spectral curve If only narrow peaks exist within the peak, then let the peak width eigenvalue be... =0; if the standard spectral curve If a region contains both wide and narrow peaks, then the peak width eigenvalue is... for: S5 is the standard spectral curve. The total number of broad and narrow peaks, S6 is the standard spectral curve. The number of broad peaks on the surface Standard spectral curve The average peak width of all broad peaks. In other words, the larger the proportion of broad peaks and the larger the average peak width, the higher the peak width characteristic value. The larger the value, the more pronounced the peak width. This can be understood as the average peak width... It can be normalized to make its value range between 0 and 1; the proportion of wide peaks Also between 0 and 1, based on the peak width characteristic value obtained from the above formula It is between 0 and 1.

[0060] It should be noted that during SG smoothing, the window size must be matched to the width of the spectral characteristic peaks. The wider the characteristic peak, the larger the window can be; the narrower the characteristic peak, the smaller the window should be to avoid peak shape distortion. Therefore, the peak width characteristic value of the standard spectral curve is used as the true peak width performance of each Sanghuang sample within the Sanghuang sample group under noise-free conditions within the target wavelength range of focus. Therefore, regarding the peak width characteristic value... , where is the peak width characteristic value of each Sanghuang sample within the sample group U in the key target wavelength range K. In other words, the larger the value of G, the larger the window can be selected, providing a better noise reduction effect; while the smaller the value of G, the smaller the window can be selected, preserving the true narrow peak information.

[0061] S13-4: Based on the shape deformation, peak asymmetry, and peak width characteristics of each group of Sanghuang samples within their respective target wavelength ranges, initial judgment coefficients for the corresponding target wavelength ranges are obtained. Reference and standard spectral curves for each Sanghuang sample within sample group U can be obtained. The Euclidean distance is used as the shape deformation representation of each Sanghuang sample within the Sanghuang sample group U in the target wavelength range K. The greater the shape deformation, the closer the reference spectral curve is to the standard spectral curve. The greater the shape difference, the greater the noise interference.

[0062] Within the target wavelength range K, for each Sanghuang sample in sample group U, a peak-valley detection algorithm is used to sequentially obtain peak and valley points. Following this sequence, the spectral curve is divided into several peak curves, progressing from valley to peak and back to valley. For any given peak curve, it is further divided into left and right peak curves using a downward vertical division. The areas S3 and S4 of the left and right peak curves are then obtained using definite integrals. As a measure of the asymmetry of the spectral peak curves, the mean of the asymmetry of all spectral peak curves is obtained, denoted as the peak asymmetry of each Sanghuang sample within the Sanghuang sample group U within the target wavelength range K of interest. .

[0063] Since the characteristic absorption peaks of polysaccharides usually exhibit a certain degree of symmetry, asymmetrical absorption peaks are likely due to noise or scattering interference. Using the minimum-maximum normalization method, the shape deformation and peak asymmetry of all *Phellinus linteus* samples were normalized across all key target wavelength ranges, yielding the following results: and .

[0064] Therefore, the initial parameter judgment coefficients for each Sanghuang sample within the sample group U in the key target wavelength range K are... for: Among the three scenarios analyzed above, the peak width characteristic value G takes the values ​​of 0, 0-1, or 1. A larger peak width characteristic value G indicates that a larger window can be selected for filtering within the target wavelength range of the *Phellinus linteus* sample. and The larger the value, the greater the noise interference the Sanghuang sample experiences within the target wavelength range. Therefore, a larger window needs to be selected so that more data points participate in the fitting, resulting in a stronger suppression effect on random noise and a smoother spectral curve.

[0065] At this point, the initial judgment coefficients for each target wavelength range have been calculated based on the above, and we proceed to step S14.

[0066] S14: Based on the absorbance confidence level and the initial judgment coefficient, obtain the window length and processing order of each original spectral curve.

[0067] Specifically, absorbance reliability reflects the reliability of information at the wavelength range level, while the initial judgment coefficient reflects the noise interference level of a single spectral curve within that wavelength range. The obtained absorbance reliability and initial judgment coefficient are analyzed together to determine the window length and processing order of each original spectral curve. By jointly evaluating these two types of parameters, when the absorbance reliability of a certain wavelength range is high and the initial judgment coefficient of the corresponding spectral curve is small, it indicates that the spectral information in that range is stable and the noise is low. In this case, a smaller window length and a lower processing order are used for smoothing to preserve polysaccharide characteristic information as much as possible. Conversely, when the absorbance reliability is low or the initial judgment coefficient is large, a larger window length and a higher processing order are used to enhance the noise suppression capability. This achieves adaptive configuration of smoothing parameters for different spectral curves and different wavelength ranges.

[0068] For example, please refer to Figure 5 Step S14 includes sub-steps S14-1 to S14-2, which are described in detail below: S14-1: Based on the absorbance confidence level and the initial judgment coefficient, obtain the target judgment coefficients for SG smoothing of each original spectral curve. Absorbance confidence level characterizes the reliability of spectral absorbance information within the target wavelength range, reflecting the overall level of noise interference in that range. The initial judgment coefficient characterizes the morphological stability and noise perturbation level of a single original spectral curve within the target wavelength range. The target judgment coefficients for each original spectral curve can be obtained by weighted fusion of the absorbance confidence level and the initial judgment coefficients, allowing the target judgment coefficients to simultaneously reflect the information confidence level at the range level and the morphological stability at the curve level. The target judgment coefficients serve as a comprehensive criterion for configuring subsequent SG smoothing parameters, measuring the intensity requirement for smoothing the original spectral curves within different target wavelength ranges.

[0069] Furthermore, sub-step S14-1 includes: The first step is to calculate the weighted Euclidean distance between the absorbance confidence levels of any two target wavelengths, and then obtain the key index for each target wavelength interval by calculating all weighted Euclidean distances for that interval. Within any target wavelength interval, using the absorbance confidence level of each target wavelength as a weight, calculate the weighted Euclidean distance between the original spectral curves of any two Sanghuang samples. Obtain the mean of the weighted Euclidean distances between all any two original spectral curves in the Sanghuang sample set, and denote this as the key index for polysaccharide content detection in that target wavelength interval. Therefore, wavelengths with higher absorbance reliability are given greater weight, ensuring the accuracy of spectral difference analysis among different Sanghuang samples, thus obtaining the key to accurate polysaccharide content detection. The minimum-maximum normalization method is used to determine the key to polysaccharide content detection for all Sanghuang samples across all target wavelength ranges. Normalization was performed to obtain key indicators. .

[0070] The second step involves weighted statistical analysis based on the key indicators and initial judgment coefficients for each target wavelength range of each Sanghuang sample to obtain the target judgment coefficient for the corresponding original spectral curve of each Sanghuang sample. The SG-smoothed target judgment coefficient of the original spectral curve of each Sanghuang sample is denoted as... Through the formula: Calculate the target judgment coefficient Where N is the number of target wavelength ranges. and These are the initial judgment coefficient and the critical normalized value for polysaccharide content detection for each Phellinus linteus sample in the nth target wavelength interval, respectively, where n is a natural number greater than 1. The larger the value, the greater the noise interference in the nth target wavelength range for the *Phellinus linteus* sample, requiring a larger window to ensure more data points participate in the fitting and improve the suppression of random noise; while key indicators The larger the value, the more important the spectral data of the *Phellinus linteus* sample is in the nth target wavelength interval. Therefore, a smaller window is needed to ensure stronger locality of the fit and better preservation of the original characteristic peak shapes and details of the spectrum. After completing the above calculations, the min-max normalization method can be used to smooth the target judgment coefficients of the SG smoothing of the original spectral curves of all *Phellinus linteus* samples. After normalization, we get .

[0071] S14-2: Based on the current target judgment coefficient and the preset target mapping relationship, obtain the window length and processing order of the original spectral curve corresponding to the current target judgment coefficient; the target mapping relationship is the preset correspondence between different target judgment coefficients and window length and processing order. It can be understood that a larger known window allows for more data points to participate in the fitting, resulting in stronger suppression of random noise and a smoother spectral curve, but may mask narrow peaks or subtle spectral variations. Conversely, a smaller window results in stronger locality of the fit, better preserving the original characteristic peak shapes and details of the spectrum, but with weaker noise filtering. A higher processing order provides greater flexibility in the fitted curve, better matching the complex variations of the original spectral curve, reducing systematic errors caused by underfitting, and is more suitable for preserving subtle spectral features; however, excessively high processing orders increase the complexity of the fitting. A lower processing order makes the fitted curve more linear or constant, with a more significant smoothing effect; however, excessively low processing orders cannot match the nonlinear variations of the spectrum, leading to severe distortion of characteristic peaks.

[0072] Therefore, a large window needs to be paired with a low-order polynomial to avoid overfitting local noise and ensure a smooth effect. A small window needs to be paired with a high-order polynomial to enhance the fit of the fitted curve to the spectral features and avoid feature loss. The larger the value, the larger the window required for the original spectral curve of the Phellinus linteus sample, and it needs to be paired with a lower-order polynomial for filtering.

[0073] In this scheme, the target mapping relationship is set with three sets of parameters: window length and processing order are (21, 2), (11, 2), and (5, 3) respectively. Thresholds are then set to 0.3 and 0.7. Take (21, 2); when Take (11, 2); when The value is (5, 3). A window length of 21 represents 21 consecutive wavelengths. Therefore, SG smoothing, combined with adaptive parameters, is used to filter the original spectrum of each Sanghuang sample to obtain an updated target spectral curve, thereby eliminating irrelevant signals and highlighting characteristic peaks related to polysaccharide content.

[0074] At this point, the window length and processing order of each original spectral curve have been obtained based on the above processing method, and we proceed to step S15.

[0075] S15: Smooth each original spectral curve according to the window length and processing order, and analyze the actual polysaccharide content of the corresponding type of Phellinus linteus sample.

[0076] Specifically, after determining the window length and processing order corresponding to each original spectral curve, the original spectral curves are smoothed according to the aforementioned parameters to reduce the impact of random noise and abnormal fluctuations on spectral characteristics. Differential smoothing enhances the overall signal-to-noise ratio and data stability while preserving the polysaccharide-related absorption characteristics. After spectral smoothing, a correlation model between the polysaccharide content of *Sanghuang* and its near-infrared spectrum is established based on the processed spectral data, and the polysaccharide content of the *Sanghuang* sample to be tested is analytically calculated. Since the smoothing parameters are adaptively determined by combining absorbance reliability and curve morphology differences, it effectively avoids feature loss due to over-smoothing or noise residue due to insufficient smoothing, thereby significantly improving the accuracy and stability of rapid, non-destructive detection of *Sanghuang* polysaccharide content via near-infrared spectroscopy.

[0077] When analyzing the actual polysaccharide content, it is necessary to determine a polysaccharide reference value. A classical chemical method can be used to determine the actual polysaccharide content of each *Sanghuang* sample in the sample set, which will serve as the polysaccharide reference value. The specific method is the phenol-sulfuric acid method (polysaccharides are hydrolyzed into monosaccharides, which react with phenol-sulfuric acid to produce a colored substance; the absorbance is measured using a UV-Vis spectrophotometer, and the content is calculated). Each *Sanghuang* sample should be measured in triplicate, and the average value should be taken as the final polysaccharide reference value to ensure data accuracy.

[0078] Further screening of spectral data was conducted. For each Sanghuang sample in the sample set, the correlation coefficient method was used to select spectral data within wavelength ranges highly correlated with polysaccharide reference values. Near-infrared spectral data, in particular, had high dimensionality (approximately 1500 wavelength points in the 1000-2500nm range) and contained redundant information, requiring dimensionality reduction through further screening.

[0079] To further develop a quantitative model, the spectral data of each Sanghuang sample was used as the independent variable, and the reference value of polysaccharide for each Sanghuang sample was used as the dependent variable. A chemometric model was then established, specifically using the well-known partial least squares regression (PLSR), which is the most commonly used near-infrared quantitative model. It is suitable for situations with high dimensionality of independent variables and multicollinearity, and can effectively extract the correlation information between the spectrum and the polysaccharide content.

[0080] Model validation and optimization were implemented by dividing the Phellinus linteus sample set into a calibration set (80% of the samples, used for model training) and a validation set (20% of the samples, used for model testing). The calibration and validation sets were randomly partitioned and covered the same polysaccharide content gradient. The model performance was then evaluated using metrics. If the model performance was unsatisfactory, the model parameters (e.g., the number of principal components in PLS) could be adjusted and optimized until the ideal accuracy was achieved. Further details are omitted here.

[0081] Among them, the well-known metric for evaluating model performance is the correction determination coefficient (Rc). 2 ), verification coefficient of determination (Rp) 2 The closer the value is to 1, the better the model fit; Corrected root mean square error (RMSEC) and Validation root mean square error (RMSEP): the smaller the value, the higher the model prediction accuracy; Relative analysis error (RPD): RPD>3, the model can be used for actual detection, RPD>2.5, the model can be used for preliminary screening.

[0082] Thus, a quantitative model is obtained. For unknown Sanghuang samples, the above steps are followed to filter the original spectral curve, resulting in an updated target spectral curve. This allows the acquisition of screening spectral data for unknown Sanghuang samples, which is then input into the quantitative model to output the actual polysaccharide content of the corresponding Sanghuang samples.

[0083] It should be noted that the order of the above embodiments of the present invention is merely for descriptive purposes and does not represent the superiority or inferiority of the embodiments. The processes depicted in the accompanying drawings do not necessarily require a specific or sequential order to achieve the desired result. In some embodiments, multitasking and parallel processing are also possible or may be advantageous.

[0084] The various embodiments in this specification are described in a progressive manner. The same or similar parts between the various embodiments can be referred to each other. Each embodiment focuses on describing the differences from other embodiments.

Claims

1. A rapid and non-destructive method for detecting the content of *Sanghuang* polysaccharides based on near-infrared spectroscopy, characterized in that, The method includes: Obtain the original spectral set of Phellinus linteus samples from various sources under near-infrared spectroscopy; each Phellinus linteus sample from the same source in the original spectral set constitutes the same set of original spectral curves; Based on the differences in absorption characteristics of the spectral curves in the original spectral set within the same group and between different groups, the absorption reliability of each target wavelength of the original spectral curve is obtained. Based on the morphological differences between a single curve and the entire set of spectral curves in each group, the initial judgment coefficients for smoothing the original spectral curves in their respective wavelength ranges are obtained. Based on the absorbance confidence level and the initial judgment coefficient, the window length and processing order of each original spectral curve are obtained; Based on the window length and the processing order, the original spectral curves are smoothed accordingly, and the actual polysaccharide content of the corresponding type of Phellinus linteus sample is analyzed.

2. The rapid and non-destructive method for detecting the content of Phellinus linteus polysaccharides according to claim 1, characterized in that, Obtain the original spectral set of near-infrared spectra of Phellinus linteus samples from diverse sources, including: Read the sample labels of Sanghuang samples from diverse sources, which are Sanghuang samples from different growth stages, different origins and different drying methods; The operating parameters for sampling using the integrating sphere diffuse reflectance method on the near-infrared spectrometer are configured according to the sample label; the operating parameters include spectral range, resolution, and number of scans. The near-infrared spectrometer is operated under the specified operating parameters to sample and average the spectral curves of the Phellinus linteus samples corresponding to each sample label, and the data is then grouped and stored as the original spectral curves of the corresponding sample labels; the original spectral curves are curves of spectral wavelength as a function of absorbance. Based on the set of curves formed by all the original spectral curves under each sample label, the original spectral set of Sanghuang samples from diverse sources is obtained.

3. The rapid and non-destructive method for detecting the content of Phellinus linteus polysaccharides according to claim 1, characterized in that, Based on the differences in absorption characteristics of spectral curves within the same group and between groups in the original spectral set, the absorbance reliability of each target wavelength of the original spectral curve is obtained, including: Multiple target wavelength ranges are configured based on the wavelengths that are sensitive to polysaccharide detection in each original spectral curve in the original spectral set; Based on the absorbance difference between two original spectral curves of the same group of Sanghuang samples within the target wavelength range, the non-correlation differences of each target wavelength within the corresponding target wavelength range are obtained; Based on the non-correlation difference between the two original spectral curves of the same group of Sanghuang samples for each target wavelength and the Hu invariant moment, spectral noise interference analysis was performed to obtain the unreliability of each target wavelength in the two original spectral curves. Statistical analysis was performed on all the unreliability scores for each target wavelength in each target wavelength range to obtain the absorbance reliability of each target wavelength in the corresponding target wavelength range.

4. The rapid and non-destructive method for detecting the content of Phellinus linteus polysaccharides according to claim 3, characterized in that, Based on the non-correlation differences between the two original spectral curves of the same group of Sanghuang samples at each target wavelength, and the Hu invariant moments, spectral noise interference analysis was performed to obtain the unreliability of each target wavelength in the two original spectral curves, including: Based on the non-correlation difference between the two original spectral curves at each target wavelength and the Hu invariant moment, the noise interference index for the corresponding target wavelength is obtained. The unreliability of each target wavelength in the two original spectral curves is obtained by averaging all noise interference indicators of the two original spectral curves.

5. The rapid and non-destructive method for detecting the content of Phellinus linteus polysaccharides according to claim 3, characterized in that, Statistical analysis was performed on all the unreliability scores for each target wavelength in each target wavelength range to obtain the absorbance reliability for each target wavelength in the corresponding target wavelength range, including: Based on the maximum and average values ​​of all unreliability scores for each target wavelength in each target wavelength range, the unreliability index for the corresponding target wavelength is obtained. Based on the number of problem groups with a disbelief level greater than the disbelief threshold and the total number of sample groups in each target wavelength range, the proportion of problem groups in each target wavelength range is obtained. The unreliable indicators and problem group proportions for each target wavelength range are normalized and their credibility is converted to obtain the absorbance credibility of each target wavelength in the corresponding target wavelength range.

6. The rapid and non-destructive method for detecting the content of Phellinus linteus polysaccharides according to claim 1, characterized in that, Based on the morphological differences between individual curves and the entire set of curves in each group of spectra, initial judgment coefficients are obtained for smoothing the original spectra within their respective wavelength ranges, including: Each set of original spectral curves is normalized to obtain a reference spectral curve for the target wavelength range. Weighted analysis is performed based on the instability characteristics among the reference spectral curves to obtain the standard spectral curves of the corresponding group of original spectral curves. Based on the peak width distribution characteristics of each standard spectral curve, the peak width characteristic value of the corresponding standard spectral curve is obtained; Based on the shape deformation, peak asymmetry, and peak width characteristic value of each group of Sanghuang samples in their respective target wavelength ranges, the initial judgment coefficients for the corresponding target wavelength ranges are obtained.

7. The rapid and non-destructive method for detecting the content of Phellinus linteus polysaccharides according to claim 6, characterized in that, A weighted analysis is performed based on the instability characteristics among the reference spectral curves to obtain standard spectral curves for the corresponding group of original spectral curves, including: The instability of the current reference spectral curve is obtained by taking the mean of the Hughes invariant moments of the current reference spectral curve and other reference spectral curves in each target wavelength range of each set of original spectral curves. Based on the overall impact of the instability of the current reference spectral curve on the same group of Sanghuang samples, the stability weight of the current reference spectral curve is obtained. The standard spectral curve for the corresponding group of original spectral curves is obtained by summing the stability weights of all reference spectral curves in each target wavelength range of each group of original spectral curves.

8. The rapid and non-destructive method for detecting the content of Phellinus linteus polysaccharides according to claim 6, characterized in that, Based on the peak width distribution characteristics of each standard spectral curve, the peak width characteristic values ​​of the corresponding standard spectral curves are obtained, including: For each standard spectral curve, the trough distribution is detected and the wavelength difference between adjacent trough points is calculated to obtain the width of several peaks for each standard spectral curve. Based on a preset width threshold, several peak widths are divided into wide peaks and narrow peaks, and the peak width characteristic value of the corresponding standard spectral curve is obtained based on the quantity distribution characteristics of the wide peaks and the narrow peaks.

9. The rapid and non-destructive method for detecting the content of Phellinus linteus polysaccharides according to claim 1, characterized in that, Based on the absorbance confidence level and the initial judgment coefficient, the window length and processing order of each original spectral curve are obtained, including: Based on the absorbance confidence level and the initial judgment coefficient, the target judgment coefficients for SG smoothing of each original spectral curve are obtained; Based on the current target judgment coefficient and the preset target mapping relationship, the window length and processing order of the original spectral curve corresponding to the current target judgment coefficient are obtained; the target mapping relationship is a preset correspondence between different target judgment coefficients and window length and processing order.

10. The rapid and non-destructive method for detecting the content of Phellinus linteus polysaccharides according to claim 9, characterized in that, Based on the absorbance confidence level and the initial judgment coefficient, the target judgment coefficients for SG smoothing of each original spectral curve are obtained, including: Calculate the weighted Euclidean distance of the absorbance confidence between any two target wavelengths, and obtain the key indicators of the corresponding target wavelength range from all the weighted Euclidean distances in each target wavelength range; We perform weighted statistics based on the key indicators and initial judgment coefficients of each target wavelength range for each Sanghuang sample to obtain the target judgment coefficient of the original spectral curve corresponding to each Sanghuang sample.

Images