Poria cocos medicinal material quality evaluation method based on intelligent sensory technology

Abstract

The invention belongs to the technical field of quality evaluation of traditional Chinese medicinal materials, and provides a poria cocos medicinal material quality evaluation method based on an intelligent sensory technology, which comprises the following steps: S1, data acquisition: respectively acquiring tactile, visual, olfactory and gustation data of different poria cocos samples by using a texture analyzer, a color photometer, an electronic nose and an electronic tongue; s2, model establishment: carrying out data fusion on the tactile, visual, olfactory and taste data of the different poria cocos samples, and establishing a production place identification model or an authenticity identification model or a grade discrimination model by adopting PLS-DA; s3, sample detection: inputting tactile, visual, olfactory and taste data of a to-be-detected poria cocos sample into the production place identification model, the authenticity identification model or the grade identification model, and judging the type, authenticity or grade of the production place according to a model prediction result. The method can distinguish poria cocos from different producing areas, and provides a scientific basis for quality control and producing area identification of poria cocos.

Description

Technical Field

[0001] This invention belongs to the field of quality evaluation technology of Chinese medicinal materials, and relates to a method for quality evaluation of Poria cocos based on intelligent sensory technology. Background Technology

[0002] Poria cocos was first recorded in the *Shennong Bencao Jing* (Shennong's Classic of Materia Medica) and listed as a "superior" herb. It also has other names such as Fuling, Futu, Songyu, Busimian, and Fushen. Its source is the dried sclerotium of the fungus *Poria cocos* (Schw.) Wolf., belonging to the genus *Poria* of the Polyporaceae family. It has the effects of promoting diuresis and removing dampness, strengthening the spleen, and calming the mind, and is a commonly used medicinal material in traditional Chinese medicine. It is mainly produced in Hebei, Henan, Yunnan, Anhui, Hubei, and Sichuan provinces. Poria cocos has been used as a traditional Chinese medicine and dietary supplement for over two thousand years, with the saying "nine out of ten prescriptions contain Poria cocos." Classic prescriptions include Linggui Zhugan Decoction from *Jinkui Yaolue*, Huopu Xialing Decoction from *Yiyuan*, and Chushi Weiling Decoction from *Yizong Jinjian*. The main chemical components of Poria cocos include polysaccharides and triterpenoids, as well as trace components such as sterols and trace elements such as calcium, iron, zinc, selenium, potassium, sodium, and phosphorus. Among them, the main pharmacologically active components are Poria cocos polysaccharides and triterpenoids.

[0003] The 2025 edition of the Chinese Pharmacopoeia specifies that the main specifications of Poria cocos are "whole Poria cocos, slices, and blocks." Currently, Poria cocos is mainly circulated in the market in the form of blocks and slices. The thickness of the slices and the size of the blocks vary. Grading methods are mainly based on their color, uniformity, and texture, with "white, firm, and regular" being the best. Based on the traditional concept of judging quality by appearance, texture, appearance, odor, and taste are widely used as important characteristic attributes in the quality evaluation of Chinese medicine. Currently, the quality evaluation of Poria cocos mainly relies on manual experience for identification, grading based on its appearance characteristics such as color, texture, and odor. This method is highly subjective and has poor repeatability. Although modern analytical methods such as HPLC and GC-MS have high precision, their pretreatment is complex, time-consuming, and costly, making rapid on-site testing difficult.

[0004] Electronic sensory technologies (such as electronic eyes, electronic noses, electronic tongues, and texture analyzers) can simulate human vision, smell, taste, and touch, enabling the objective and digital representation of the appearance, aroma, taste, and texture of traditional Chinese medicine (TCM). These technologies are gradually being applied to the quality evaluation of TCM. However, the information provided by a single sensory device is limited and cannot comprehensively reflect the overall quality characteristics of medicinal materials. For example, Chinese patent CN 110307871 A discloses a method for rapid detection of the quality of Chinese herbal medicine slices using electronic sensory fusion. The system used in this method consists of a data acquisition system, a data fusion system, and a pattern recognition system. The data acquisition system includes an electronic sensory tactile system, an electronic sensory visual system, an electronic sensory olfactory system, and an electronic sensory gustatory system. The electronic sensory tactile system and the electronic sensory visual system measure the physical properties of the slices, such as hardness, brittleness, powderiness, and fibrousness. The electronic sensory visual system measures the appearance and optical signal data of the slices. The electronic sensory olfactory system and the electronic sensory gustatory system measure the gas and taste data of the sample, respectively. The measured data are transmitted to the data fusion system for data conversion and fusion, and then enter the pattern recognition system to match the corresponding sub-model. Finally, the quality detection results of the slices are output, effectively solving the detection problems of authenticity, origin, place of origin, and grade of slices. This method claims to be applicable to any Chinese herbal medicine slices, but it does not provide a scientific and reliable digital evaluation and place of origin differentiation for specific medicinal materials (such as Poria cocos). Summary of the Invention

[0005] This invention proposes a quality evaluation method for Poria cocos based on intelligent sensory technology. This method can establish a rapid, objective, and quantifiable quality evaluation system for Poria cocos, replacing the traditional method that relies on subjective human judgment (such as "identifying the appearance and quality"), and providing a scientific basis for the quality control and origin identification of Poria cocos.

[0006] The technical solution of this invention is implemented as follows: A method for evaluating the quality of Poria cocos medicinal materials based on intelligent sensory technology includes the following steps: S1. Data Acquisition: Tactile, visual, olfactory, and gustatory data of different Poria cocos samples were collected using a texture analyzer, colorimeter, electronic nose, and electronic tongue. S2. Model establishment: The tactile, visual, olfactory and gustatory data of the different Poria cocos samples are fused, and a place of origin identification model, authenticity identification model or grade discrimination model is established using PLS-DA. S3. Sample testing: Input the tactile, visual, olfactory and gustatory data of the Poria cocos sample to be tested into the origin identification model, authenticity identification model or grade discrimination model, and determine its origin category, authenticity or grade based on the model prediction results.

[0007] Preferably, it includes the following steps: S1. Data Acquisition: Tactile, visual, olfactory, and gustatory data of Poria cocos samples were collected using a texture analyzer, colorimeter, electronic nose, and electronic tongue, respectively, to form a multi-dimensional feature vector; S2. Labeling system construction: Combine the origin and grade information of known Poria cocos samples and define them as a unified combined category label; S3. Establishing a fusion model: Associate the multi-dimensional feature vector obtained in step S1 with the combined category label defined in step S2, and train a multi-class discriminant model using the partial least squares discriminant analysis (PLS-DA) algorithm; S4. Sample testing: Collect tactile, visual, olfactory and gustatory data of the Poria cocos sample to be tested, form a multi-dimensional feature vector, and input it into the multi-classification discrimination model established in step S3. Determine its place of origin and grade category based on the model prediction results.

[0008] Preferably, the tactile data includes firmness and extrusion.

[0009] Preferably, the visual data includes L, a, and b values ​​and calculates the total color difference Eab.

[0010] Preferably, the olfactory data includes data on volatile odor components.

[0011] Preferably, the taste data is obtained by measuring the original taste and aftertaste, wherein the original taste includes sourness, saltiness, bitterness, astringency, umami, and sweetness; the aftertaste includes bitter aftertaste, astringent aftertaste, and umami richness. After the electronic tongue measurement, the data is converted into quantifiable data using the workstation software analysis software, i.e., taste analysis application software.

[0012] Preferably, the texture analyzer detection method includes: a trigger force of 3±0.6N, a detection speed of 15±3mm / min, a deformation of 15±3%, a return speed of 30±6mm / min, and a return distance of 15±3mm. If the trigger force is too large or the detection speed is too fast, it cannot accurately reflect the texture characteristics; if it is too small, it may not be able to trigger effective detection or the signal may be too weak. Parameters such as a trigger force of 3±0.6N and a detection speed of 15±3mm / min can obtain tactile data with high signal-to-noise ratio and good repeatability without damaging the structure of the Poria cocos sample, which is the foundation for the successful establishment of subsequent models. If the data fluctuates greatly, a reliable model cannot be established.

[0013] Preferably, the texture analyzer detection method includes: a trigger force of 3N, a detection speed of 15mm / min, a deformation of 15%, a return speed of 30mm / min, and a return distance of 15mm.

[0014] Preferably, the colorimeter detection method includes: a standard light source D65, an observer angle of 10°, a reflective aperture of 25.4 mm, a measurement mode SCI, recording the L, a, and b values, and calculating the total color difference Eab. These detection conditions ensure the comparability and consistency of color data (L, a, b*) measured from different batches, by different operators, and at different times. If different light sources or modes are used, the measurement results will deviate significantly and cannot be used to establish a reliable color evaluation model.

[0015] Preferably, the electronic nose detection method includes: using headspace inhalation, an injection flow rate of 300 mL / min, a washing time of 70-90 s, a sampling interval of 1 s, and a data acquisition time of 140-160 s, taking the average value of data from the 147th to the 149th s. This specific time point selection greatly improves the stability and representativeness of the odor data, providing a unified and reliable benchmark for comparisons between different samples, thus enabling the electronic nose data to be effectively used for origin identification.

[0016] Preferably, the electronic nose detection method includes: using headspace aspiration, with an injection flow rate of 300 mL / min, a cleaning time of 80 s, a sampling interval of 1 s, and a collection time of 150 s, taking the average value of the data from the 147th to the 149th s. The test room temperature is 25±2℃ and the humidity is 50±5%.

[0017] Preferably, the electronic tongue detection method includes: using the supernatant of an aqueous extract, measuring for 120 seconds, cycling the five-taste sensor 4 times, and cycling the sweetness sensor 5 times. The test room temperature is 25±2℃, and the humidity is 50±5%.

[0018] The beneficial effects of the present invention using the above technical solution are as follows: 1. The method provided by this invention is convenient to operate and accurate in testing. By objectively and quantitatively analyzing the color, odor, taste and texture of Poria cocos samples from different origins, it achieves a scientific and digital evaluation of Poria cocos quality. Furthermore, it can establish a fingerprint spectrum to achieve origin identification and quality differentiation to ensure clinical drug use. This method does not require complicated pretreatment, is rapid in detection, has strong practical application value, and has significant economic and social benefits.

[0019] 2. This invention, through radar charts and model analysis of electronic nose and electronic tongue, can not only distinguish the place of origin, but also predict the potential differences in volatile components (such as nitrogen oxides and sulfides) and flavor substances (such as sourness) of Poria cocos from different places of origin, providing more dimensions of basis for quality evaluation. Attached Figure Description

[0020] The present invention will now be described in further detail with reference to the accompanying drawings and specific embodiments.

[0021] Figure 1This is a PCA score graph of the texture analyzer of this invention.

[0022] Figure 2 This is a PLS-DA score chart of the texture analyzer of this invention.

[0023] Figure 3 This is a graph showing the displacement test results of the PLS-DA model of the texture analyzer of this invention.

[0024] Figure 4 This is a radar diagram showing the colorimetric response of the colorimeter of this invention.

[0025] Figure 5 This is the PCA score chart of the colorimeter of this invention.

[0026] Figure 6 This is the PLS-DA score chart of the colorimeter of this invention.

[0027] Figure 7 This is a diagram showing the replacement test results of the PLS-DA model of the colorimeter of this invention.

[0028] Figure 8 This is a radar diagram showing the response of the electronic nose sensor of the present invention.

[0029] Figure 9 This is the PCA score chart for the electronic nose of this invention.

[0030] Figure 10 This is the PLS-DA score chart of the electronic nose of the present invention.

[0031] Figure 11 This is a diagram showing the replacement test results of the PLS-DA electronic nose model of the present invention.

[0032] Figure 12 This is a radar diagram showing the response of the electronic tongue sensor of the present invention.

[0033] Figure 13 This is the PCA score diagram of the electronic tongue of the present invention.

[0034] Figure 14 This is the PLS-DA score chart of the electronic tongue of the present invention.

[0035] Figure 15 This is a permutation test result diagram of the PLS-DA model of the electronic tongue of the present invention. Detailed Implementation

[0036] The technical solutions of the present invention will be clearly and completely described below with reference to the embodiments of the present invention. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those of ordinary skill in the art without creative effort are within the scope of protection of the present invention.

[0037] 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. The terminology used herein in the specification is for the purpose of describing particular embodiments only and is not intended to limit the invention. Unless otherwise specified, the experimental or testing methods involved in the embodiments of this invention are conventional methods in the prior art, and their names and / or abbreviations are conventional names in the art, clearly defined in their respective fields of application. Those skilled in the art can understand the conventional process steps based on these names and apply the corresponding equipment, implementing them under conventional conditions or conditions recommended by the manufacturer. The various instruments, equipment, raw materials, or reagents used in the embodiments of this invention are not subject to any special restrictions on their source; they are all conventional products that can be purchased through legitimate commercial channels and can be prepared according to conventional methods well known to those skilled in the art.

[0038] The following examples selected 20 batches of Poria cocos medicinal materials from four production areas. The specific instruments and materials used are as follows: 1. Instruments and Materials 1.1 Instruments TMS-pro texture analyzer (FTC, USA); YS6010 benchtop spectrophotometer (Guangdong Sanenshi Intelligent Technology Co., Ltd.), colorimetric values ​​are described in Table 1; The sensor performance description for the PEN3 electronic nose system (AIRSENSE GmbH, Germany) is shown in Table 2. The sensor performance description for the SA402B electronic tongue system (Insent Corporation, Japan) is shown in Table 3.

[0039] Table 1 Colorimetric values ​​of the colorimeter Table 2 Electronic nose sensor performance Table 3 Performance of Electronic Tongue Sensor 1.2 Materials The Poria cocos samples came from 20 batches from four production areas: Yuexi in Anhui Province, Dabie Mountains in Anhui Province, Shangluo in Shaanxi Province, and Qujing in Yunnan Province. Experts with extensive experience in traditional Chinese medicine identification and research identified them as Poria cocos, a fungus belonging to the genus *Polyporus* of the family Polyporaceae. Poria cocos Dried sclerotia of (Schw.)Wolf. Information on Poria cocos samples from different origins is shown in Table 4.

[0040] Table 4 Information on Poria cocos samples from different origins The method for evaluating the quality of Poria cocos medicinal materials based on intelligent sensory technology provided by this invention includes the following steps: S1. Data Acquisition: Tactile, visual, olfactory, and gustatory data of Poria cocos samples from different origins and of different quality batches were collected using a texture analyzer, colorimeter, electronic nose, and electronic tongue. S2. Model Establishment: The tactile, visual, olfactory, and gustatory data of the Poria cocos samples from different origins are fused to form a multi-source fusion feature dataset, and a partial least squares discriminant analysis (PLS-DA) is used to establish an origin identification model. Alternatively, the tactile, visual, olfactory, and gustatory data of the Poria cocos samples from different quality batches are fused to form a multi-source fusion feature dataset, and a partial least squares discriminant analysis (PLS-DA) is used to establish a authenticity identification model or a grade identification model. Specifically, the method for establishing the origin identification model is as follows: Known Poria cocos samples are used as the training set, with their multi-source fusion feature data as the independent variable (X) and the origin category as the dependent variable (Y) for model training. The effectiveness of the model is verified through permutation testing to prevent overfitting, thereby achieving the identification of the origin of Poria cocos.

[0041] In this process, all or more sets of detection data obtained in step S1 are standardized and preprocessed to form a multi-source fusion feature dataset. This requires S1 to collect as many and as comprehensive known origin samples as possible. S3. Sample Testing: Tactile, visual, olfactory, and gustatory data of the Poria cocos sample to be tested are input into the origin identification model, authenticity identification model, or grade discrimination model. Based on the model's prediction results, its origin category, authenticity, or grade quality is determined. Specifically, The multi-source intelligent sensory data of Poria cocos samples from unknown origins were detected according to the method in step S1. The detection data is input into the validated PLS-DA model established in step S2, and the judgment is made based on the model prediction results.

[0042] At the same time, the radar chart or key parameter values ​​of the sample can be compared with the standard charts or parameter ranges of known high-quality production areas (such as Qujing, Yunnan) to achieve rapid evaluation of its quality.

[0043] Example 1 A method for evaluating the quality of Poria cocos medicinal materials based on intelligent sensory technology includes the following steps: S1. Data Acquisition: Tactile, visual, olfactory, and gustatory data of Poria cocos samples were collected using a texture analyzer, colorimeter, electronic nose, and electronic tongue, respectively; details are as follows: S101. The method for collecting tactile data of Poria cocos samples using a texture analyzer includes the following steps: Using a test range of 500N, and selecting the test procedures for firmness and extrusion degree, a conical probe was used to measure 20 batches of Poria cocos samples.

[0044] Texture analyzer measurement parameters: trigger force 3N, experimental detection speed 15mm / min, deformation percentage 15%, experimental return speed 30mm / min, experimental return distance 15mm. Each batch was tested 7 times, and the average value was taken.

[0045] S102. The method for collecting visual data of Poria cocos samples using a colorimeter includes the following steps: (1) Sample preparation Twenty batches of Poria cocos samples were pulverized, passed through a No. 2 sieve, and used as samples for testing.

[0046] (2) Colorimetric analysis of the appearance of Poria cocos After the machine is powered on and stabilized, adjust the lens exposure and focal length, and take a sample for measurement.

[0047] Colorimeter measurement parameters: Standard light source: D65; Observer angle: 10°; Sample type: Reflection; Reflection aperture: 25.4mm; Measurement mode: SCI. Record the sample's L* (luminance), a* (red / green hue), and b* (yellow / blue hue) values, and calculate the total color value E. * ab. E * ab=[(L * ) 2 +(a * ) 2 +(b * ) 2 ] 1 / 2 E * The higher the ab value, the lighter the color.

[0048] S103. The method for collecting olfactory data from Poria cocos samples using an electronic nose includes the following steps: Accurately weigh 2 g of each sample, place them in a 100 mL beaker, seal with double-layer sealing film, and let stand for 20 min.

[0049] Electronic nose measurement parameters: Measurements were performed using the headspace inhalation method, with a stable response value as the indicator. The injection volume was 300 mL / min. -1 The cleaning time was 80 s, the sampling interval was 1 s, the zero-point fine-tuning time was 10 s, the pre-sampling time was 5 s, and the acquisition time was 150 s / group. To reduce measurement error, data were selected at 147 s, 148 s, and 149 s for each group. Each batch of samples was measured three times, and the average of the three measurements was taken.

[0050] S104. The method for collecting taste data from Poria cocos samples using an electronic tongue includes the following steps: Accurately weigh 1g of each sample and place them in an Erlenmeyer flask. Add 100mL of purified water, extract by sonication for 30 min, and then transfer to a 100mL centrifuge tube. Centrifuge at 4000 r·min. -1 Centrifuge for 10 minutes, and take the supernatant into the electronic tongue's special sample cup for testing.

[0051] Electronic tongue measurement parameters: Measurement time 120 s, five-taste sensor cycle test 4 times, sweetness sensor cycle test 5 times, and the average of the last 3 data for either five-taste or sweetness is taken as the test result. Taste analysis application software is used to convert the electronic tongue detection results into taste values.

[0052] The specific parameters collected and measured are shown in Table 5-7 below.

[0053] Table 5 Table 6 Table 7 S2. Data Processing and Model Building: The collected tactile, visual, olfactory, and gustatory data are processed to build an origin identification model; specifically, Origin 2024 software (OriginLab, USA) is used to draw radar charts, and SIMCA14.1 (Umetrics AB, Sweden) is used to draw PCA and PLS-DA charts. S201, Tactile Data Processing Data on the firmness and extrusion degree of 20 batches of Poria cocos samples were analyzed. PCA (PC1: 0.982, PC2: 0.0183, Ellipse: Hotelling's T2 (95%)) analysis was performed on various color values. Figure 1 Poria cocos from Qujing, Yunnan, is concentrated in the first and fourth quadrants, while Poria cocos from the other three producing areas is concentrated in the second and third quadrants.

[0054] To verify the differences in firmness of Poria cocos from different origins, PLS-DA analysis was performed using firmness and extrusion degree of 20 batches of Poria cocos samples from different origins as dependent variables and different origins as independent variables. Figure 2 The independent variable in the analysis is the fit index (R²). 2 X) is 1, and the dependent variable fit index (R) is 1. 2 Y) is 0.482, and the model prediction index (Q) is 0.482. 2 The value is 0.418, indicating model stability. After 200 permutation tests, as shown... Figure 3 As shown, Q 2An intercept less than 0 indicates that the model is not overfitting and the model validation is effective. Poria cocos from Qujing, Yunnan, clusters at the boundary between the second and third quadrants, while Poria cocos from the other three producing areas are located in the first and fourth quadrants. The PLS-DA and PCA results are similar.

[0055] S202, Visual Data Processing L*, a*, b*, and E values ​​of 20 batches of Poria cocos samples * Analyze the data of each color value of ab, such as Figure 4 As shown, 20 batches of Poria cocos samples showed differences in L* and E. * The high ab response indicates that the colors of the Poria cocos samples are all bright white and light. PCA (PC1: 0.801, PC2: 0.193, Ellipse: Hotelling's T2 (95%)) analysis was performed on each color value. Figure 5 As shown, Poria cocos from Yuexi, Anhui Province, clustered in the first quadrant, while Poria cocos from Qu County, Yunnan Province, clustered in the third quadrant. Poria cocos from Dabie Mountain, Anhui Province, and Shangluo, Shaanxi Province, overlapped and clustered in the first and fourth quadrants, indicating that there were differences in color among the 20 batches of Poria cocos from different production areas.

[0056] To verify the differences in appearance color of Poria cocos from different origins, PLS-DA analysis was performed using the color values ​​of 20 batches of Poria cocos samples from different origins as the dependent variable and the origin of the samples as the independent variable. Figure 6 As shown. The independent variable in the analysis is the fit index (R²). 2 X) is 1, and the dependent variable fit index (R) is 1. 2 Y) is 0.753, and the model prediction index (Q) is 0.753. 2 The value is 0.725, indicating model stability. After 200 permutation tests, as shown... Figure 7 As shown, Q 2 An intercept less than 0 indicates that the model is not overfitting and the model validation is effective. Poria cocos from Yuexi, Anhui, is located in the fourth quadrant; Poria cocos from Qujing, Yunnan, clusters at the boundary between the second and third quadrants; and Poria cocos from Dabie Mountains, Anhui, and Shangluo, Shaanxi, are located in the first quadrant. The results of PLS-DA and PCA are similar.

[0057] S203, Olfactory Data Processing Data analysis was performed on the average response values ​​of the electronic nose sensor for 20 batches of Poria cocos samples. Figure 8 It can be seen that sensor W5S has a significant response, indicating that the Poria cocos samples all contain a relatively high amount of nitrogen oxides. Furthermore, the Poria cocos samples from Qujing, Yunnan also show significant responses to sensors W6S, W1S, and W1W, suggesting that Poria cocos from Qujing, Yunnan contains more hydrogen, short-chain alkanes, and sulfides than Poria cocos from other producing areas. Figure 9As shown, PCA (PC1: 0.681, PC2: 0.187, Ellipse: Hotelling's T2 (95%)) analysis revealed that Poria cocos from Yuexi, Anhui Province, was concentrated in the second quadrant, while Poria cocos from Qujing, Yunnan Province, was distributed in the third quadrant. Poria cocos from Dabie Mountain, Anhui Province, and Shangluo, Shaanxi Province, were mainly concentrated in the first and fourth quadrants, indicating that the aroma of Poria cocos from Yuexi, Anhui Province, Qujing, Yunnan Province, Dabie Mountain, Anhui Province, and Shangluo, Shaanxi Province, are all different.

[0058] To clarify the differences in aroma of Poria cocos from different origins, PLS-DA analysis was performed using the average response value of the electronic nose sensor of 20 Poria cocos samples from different origins as the dependent variable and the samples from different origins as the independent variable. Figure 10 As shown. Yuexi, Anhui is located in the fourth quadrant; Qujing, Yunnan is at the junction of the first and second quadrants; Dabie Mountains, Anhui and Shangluo, Shaanxi are in the third and fourth quadrants, respectively. The independent variable in the analysis is the fit index (R²). 2 X) is 0.984, and the dependent variable fit index (R) is 0.984. 2 Y) is 0.905, and the model prediction index (Q) is 0.905. 2 The value is 0.823, R 2 and Q 2 A value greater than 0.5 indicates an acceptable model fit, after 200 permutation tests. Figure 11 As shown, Q 2 An intercept less than 0 indicates that the model is not overfitting and the model validation is effective. PLS-DA and PCA have similar discrimination results.

[0059] S204, Taste Data Processing An electronic tongue sensor was used to detect and analyze the taste quality of 20 batches of Poria cocos samples, focusing on five inherent flavors (sweet, sour, bitter, astringent, umami, and salty) and three aftertastes (bitter aftertaste, astringent aftertaste, and umami richness). Based on taste values, such as... Figure 12 As shown, Poria cocos from the four origins exhibited relatively high responses to sweetness, umami, and bitterness, with Poria cocos from Qujing, Yunnan showing a higher response to sourness. To determine whether there were statistically significant differences in the taste quality among Poria cocos from different origins, as... Figure 13As shown, PCA (PC1: 0.789, PC2: 0.188, Ellipse: Hotelling's T2 (95%)) analysis was performed on the taste values ​​of Poria cocos samples from different origins. Poria cocos from different origins showed relative clustering. Specifically, Poria cocos from Yuexi, Anhui clustered at the boundary between the first and second quadrants; Poria cocos from Qujing, Yunnan concentrated in the fourth quadrant; Poria cocos from Dabie Mountain, Anhui was in the third quadrant; and Poria cocos from Shangluo, Shaanxi was at the boundary between the second and third quadrants. This indicates that Poria cocos from the four origins clustered independently, with significant taste differences. To further clarify the taste differences among Poria cocos from different origins, PLS-DA analysis was performed using the taste values ​​of 5 primary flavors and 3 aftertastes from 20 Poria cocos slices from different origins as dependent variables and samples from different origins as independent variables. Figure 14 As shown in the figure. The results are close to those of the PCA analysis. Poria cocos from Yuexi, Anhui, is located in the first quadrant; from Qujing, Yunnan, in the third quadrant; from the Dabie Mountains of Anhui, in the fourth quadrant; and from Shangluo, Shaanxi, at the boundary between the first and fourth quadrants. The independent variable in the analysis is the fit index (R²). 2 X) is 0.998, and the dependent variable fit index (R) is 0.998. 2 Y) is 0.955, and the model prediction index (Q) is 0.955. 2 The value is 0.933, R 2 and Q 2 A value greater than 0.5 indicates an acceptable model fit, after 200 permutation tests. Figure 15 As shown, Q 2 An intercept less than 0 indicates that the model is not overfitting, the model is valid, and the taste of Poria cocos varies from place to place.

[0060] As can be seen, the results of the determination of 20 batches of Poria cocos samples from four production areas by intelligent sensory (texture analyzer, colorimeter, electronic nose, electronic tongue) show that the quality evaluation method of Poria cocos based on intelligent sensory technology provided by this invention can distinguish Poria cocos from different production areas.

[0061] PLS-DA models were established for the four types of tactile, visual, olfactory, and gustatory data. Inputting the detection data of the Poria cocos sample into each PLS-DA model yielded a probability or score vector reflecting the likelihood of the sample belonging to each origin. Several rules can be used to summarize these results and output the final origin, such as simple majority voting, where the origin with the most votes is the final output. Alternatively, considering the different discriminative abilities of each model (e.g., the accuracy of the odor model may be much higher than that of the color model), a weight can be assigned to each model, using a weighted average voting. However, these two methods cannot utilize the effects between features. Therefore, a further improvement is proposed: merging the overall data to establish the model.

[0062] The collected tactile, visual, olfactory, and gustatory data were processed and fused to form a multi-source fusion feature dataset. Partial least squares-discriminant analysis (PLS-DA) was then used to establish an origin identification model.

[0063] Then, sample testing was performed: the tactile, visual, olfactory, and gustatory data of the Poria cocos samples to be tested were input into the origin identification model, and its origin category was output. In this embodiment, a new batch of known origin samples (10 batches) from the above-mentioned origins were used as the validation set. The results showed that the evaluation method was reliable and effective, and the accuracy of the output results reached 100%.

[0064] Example 2 The method for evaluating the quality of Poria cocos medicinal materials based on intelligent sensory technology includes the following steps: S1. Data Acquisition: Same as in Example 1, tactile, visual, olfactory and gustatory data of different quality batches of Poria cocos samples were collected using a texture analyzer, colorimeter, electronic nose and electronic tongue; S2. Model Establishment: The tactile, visual, olfactory, and gustatory data of the different quality batches of Poria cocos samples are fused to form a multi-source fusion feature dataset, and a genuine / counterfeit identification model is established using partial least squares discriminant analysis (PLS-DA). In this process, all or more sets of detection data obtained in step S1 are standardized and preprocessed to form a multi-source fusion feature dataset. S1 requires collecting as many and as comprehensive samples from known origins as possible. S3. Sample Testing: The tactile, visual, olfactory, and gustatory data of the Poria cocos sample to be tested are input into the authenticity identification model. The model's prediction results determine its authenticity. Specifically, The multi-source intelligent sensory data of Poria cocos samples from unknown origins were detected according to the method in step S1. The detection data is input into the validated PLS-DA model established in step S2, and its authenticity is determined based on the model's prediction results.

[0065] Furthermore, this invention can also compare the radar chart or key parameter values ​​of a sample with the standard charts or parameter ranges of known high-quality production areas (such as Qujing, Yunnan), thereby enabling rapid evaluation of its quality.

[0066] Example 3 1. Tag System Construction 1.1 The "Place of Origin" and "Grade" information levels are merged into a single first-level combined label, forming 12 categories: For example, Anhui-Yuexi-Excellent, Anhui-Yuexi-Medium, Anhui-Yuexi-Inferior, Anhui-Dabieshan-Excellent, Anhui-Dabieshan-Medium, Anhui-Dabieshan-Inferior, Shaanxi-Shangluo-Excellent, Shaanxi-Shangluo-Medium, Shaanxi-Shangluo-Inferior, Yunnan-Qujing-Excellent, Yunnan-Qujing-Medium, Yunnan-Qujing-Inferior.

[0067] 1.2 The grading is based on the "differentiation of appearance and quality" principle under the Poria cocos section of the 2025 edition of the Chinese Pharmacopoeia, and has been verified by expert consensus.

[0068] 2. Data Preparation 2.1 The same method as in Example 1 was used to collect four-modal vectors (tactile 2D, color 4D, electronic nose 10D, electronic tongue 8D, for a total of 24 dimensions) from 20 batches of known samples.

[0069] 2.2 Standardize all feature data and associate them with 12 categories of combined labels.

[0070] 3. Establish a multi-objective discrimination model: The partial least squares discriminant analysis (PLS-DA) algorithm is adopted and extended to a multi-class classification model that can handle 12 categories simultaneously.

[0071] During model training, different weights are assigned to "origin" and "grade", with an origin weight of 1.0 and a grade weight of 0.8, to balance the problem of imbalance in the number of samples of different grades.

[0072] 4. Prediction 4.1 Input the 24-dimensional data of the unknown sample into the trained model. The model will output a vector containing 12 scores. Use the Softmax function to convert these scores into probabilities of belonging to each category. Select the category with the highest probability as the result, and then decompose it to obtain both "origin" and "grade" information.

[0073] The above are merely preferred embodiments of the present invention and are 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 evaluating the quality of Poria cocos medicinal material based on intelligent sensory technology, characterized in that, Includes the following steps: S1. Data Acquisition: Tactile, visual, olfactory, and gustatory data of different Poria cocos samples were collected using a texture analyzer, colorimeter, electronic nose, and electronic tongue. S2. Model establishment: The tactile, visual, olfactory and gustatory data of the different Poria cocos samples are fused, and a place of origin identification model, authenticity identification model or grade discrimination model is established using PLS-DA. S3. Sample testing: Input the tactile, visual, olfactory and gustatory data of the Poria cocos sample to be tested into the origin identification model, authenticity identification model or grade discrimination model, and determine its origin category, authenticity or grade based on the model prediction results.

2. The method for evaluating the quality of Poria cocos medicinal materials based on intelligent sensory technology according to claim 1, characterized in that, The texture analyzer detection method includes: trigger force 3±0.6N, detection speed 15±3mm / min, deformation 15±3%, return speed 30±6mm / min, and return distance 15±3mm.

3. The method for evaluating the quality of Poria cocos medicinal materials based on intelligent sensory technology according to claim 1, characterized in that, The colorimeter detection method includes: a standard light source D65, an observer angle of 10°, a reflective aperture of 25.4mm, a measurement mode SCI, recording L, a, and b values, and calculating the total color difference Eab.

4. The method for evaluating the quality of Poria cocos medicinal materials based on intelligent sensory technology according to claim 1, characterized in that, The electronic nose detection method includes: using headspace aspiration, with an injection flow rate of 300 mL / min, a washing time of 70-90 s, a sampling interval of 1 s, an acquisition time of 140-160 s, and taking the average value of the data from the 147th to the 149th s.

5. The method for evaluating the quality of Poria cocos medicinal materials based on intelligent sensory technology according to claim 1, characterized in that, The electronic tongue detection method includes: using the supernatant of an aqueous extract, measuring for 120 seconds, cycling the five-taste sensor 4 times, and cycling the sweetness sensor 5 times.

Images