Traditional Chinese medicine quality detection method and system

Abstract

The invention provides a traditional Chinese medicine quality detection method and system, and belongs to the technical field of quality detection.The method comprises the steps that to-be-detected traditional Chinese medicine is scanned, surface features of the to-be-detected traditional Chinese medicine are obtained, and the surface features are compared with genuine standard features to determine first quality; determining a volatile component response curve by adopting an electronic smell capture array, identifying a characteristic peak to determine a second quality, and performing component analysis based on the characteristic peak to determine a third quality; performing conflict data arbitration based on the first quality, the second quality and the third quality to obtain a fourth quality, performing environment dynamic compensation based on the fourth quality to obtain a fifth quality, marking an upstream planting end, pushing an optimization suggestion upwards, and identifying execution conditions of a historical mark and the optimization suggestion to obtain a current mark of the upstream planting end; and performing compensation adjustment on the fifth quality in combination with the current mark and a three-dimensional compensation system to obtain the final quality. Comprehensive detection and optimization of traditional Chinese medicine quality are realized, and accuracy and efficiency of traditional Chinese medicine quality detection are improved.

Description

Technical Field

[0001] The present invention relates to the technical field of quality inspection, and particularly to a method and system for inspecting the quality of traditional Chinese medicine. Background Art

[0002] As an important part of traditional Chinese medicine in China, the corresponding quality inspection of traditional Chinese medicine includes various aspects. The traditional sensory evaluation by visual inspection and manual touch lacks objective standards, while instrumental analysis such as HPLC and GC-MS is accurate but time-consuming and requires sample destruction, making it difficult to be applied on a large scale. In addition, due to the lack of quantitative standards for the evaluation of the authenticity of traditional Chinese medicine, there are large differences in the determination between production areas, and it is impossible to track the quality changes after harvesting.

[0003] Therefore, the present invention provides a method and system for inspecting the quality of traditional Chinese medicine. Summary of the Invention

[0004] A method and system for inspecting the quality of traditional Chinese medicine provided by the present invention realizes the comprehensive inspection and optimization of the quality of traditional Chinese medicine by performing multi-modal data fusion on the surface characteristics, volatile components and component analysis of traditional Chinese medicine, using conflict data arbitration and environmental dynamic compensation, combining blockchain traceability and three-dimensional compensation system, improving the accuracy and efficiency of traditional Chinese medicine quality inspection, realizing data sharing and collaboration in the whole chain of traditional Chinese medicine planting, processing and circulation, and promoting the sustainable development of the traditional Chinese medicine industry.

[0005] The present invention provides a method for inspecting the quality of traditional Chinese medicine, including: Step 1: Scanning the traditional Chinese medicine to be detected to obtain the surface characteristics of the traditional Chinese medicine to be detected, comparing the surface characteristics with the standard characteristics of the authentic medicine, and determining the first quality; Step 2: Using an electronic odor capture array to record the response curve of the volatile components of the traditional Chinese medicine to be detected, identifying the characteristic peaks, determining the second quality, and performing component analysis on the traditional Chinese medicine to be detected based on the characteristic peaks, and further determining the third quality; Step 3: Performing conflict data arbitration based on the first quality, the second quality and the third quality to obtain the fourth quality, and performing environmental dynamic compensation based on the fourth quality to obtain the fifth quality; Step 4: Marking the upstream planting end according to the fifth quality and pushing optimization suggestions upward, and at the same time identifying the implementation situation of the historical marks and optimization suggestions to obtain the current marks of the corresponding upstream planting end, and compensating and adjusting the fifth quality in combination with the current marks and the three-dimensional compensation system to obtain the final quality.

[0006] The present invention provides a method for inspecting the quality of traditional Chinese medicine, which scans the traditional Chinese medicine to be detected to obtain the surface characteristics of the traditional Chinese medicine to be detected, compares the surface characteristics with the standard characteristics of the authentic medicine, and determines the first quality, including: Use a hyperspectral camera to scan the traditional Chinese medicine to be detected to obtain the hyperspectral pixel coordinates of the traditional Chinese medicine to be detected, and use structured light to scan the traditional Chinese medicine to be detected to obtain the 3D point cloud coordinates of the traditional Chinese medicine to be detected; Perform spatio-temporal alignment on the hyperspectral pixel coordinates and the 3D point cloud coordinates through the SIFT algorithm, map the hyperspectral data obtained from the hyperspectral pixel coordinates after alignment to the 3D point cloud coordinates, perform spectral-geometry correlation analysis and defect feature quantification to determine the surface features of the traditional Chinese medicine to be detected; Based on the comparison between the surface features and the genuine standard features, trigger the preset knowledge graph rules according to the comparison results to determine the abnormal results, retrieve and match the associated nodes corresponding to the abnormal results in the preset knowledge graph, and determine the first quality based on the associated nodes and the corresponding surface features.

[0007] The present invention provides a method for detecting the quality of traditional Chinese medicine. An electronic odor capture array is used to record the response curve of the volatile components of the traditional Chinese medicine to be detected, identify the characteristic peaks, and determine the second quality, including: Set a bifurcated odor sampling head, combine the bifurcated odor sampling head with a multimodal sensor to form an electronic odor capture array, and determine the scanning mode of the electronic odor capture array according to the surface features of the traditional Chinese medicine to be detected; Generate a time-response matrix according to the electronic odor capture array, and inhale pure air every preset time to reset the sensor baseline, thereby determining the response curve of the volatile components of the traditional Chinese medicine to be detected; Calculate the first derivative of the response curve, set a sliding window according to the scanning mode, determine the characteristic peaks of the response curve, determine the main peak intensity based on the characteristic peaks, use the sliding window centered on the main peak intensity as the post-peak interval to calculate the platform stability, and then obtain the peak platform ratio. Based on the main peak intensity, platform stability, and peak platform ratio, determine the discrimination conditions, and obtain the second quality of the traditional Chinese medicine to be detected based on the discrimination conditions.

[0008] The present invention provides a method for detecting the quality of traditional Chinese medicine. Set a bifurcated odor sampling head, combine the bifurcated odor sampling head with a multimodal sensor to form an electronic odor capture array, and determine the scanning mode of the electronic odor capture array according to the surface features of the traditional Chinese medicine to be detected, including: Adopt a dual-channel bifurcated structure. The dual-channel bifurcated structure adopts bilateral independent air extraction, and the bilateral integrated independent micro air pumps can control different flow rates. Multimodal sensors of the same model are arranged at the end of each channel to form an electronic odor capture array; Perform regional division on the traditional Chinese medicine to be detected according to the surface features, determine the dynamic configuration of the multimodal sensor according to the regional division results, and at the same time, set corresponding capture algorithms for each column in the electronic odor capture array; Determine the scanning mode based on the dynamic configuration and the capture algorithm.

[0009] The present invention provides a method for detecting the quality of traditional Chinese medicine, which performs component analysis on the traditional Chinese medicine to be detected based on the characteristic peaks, and then determines the third quality, comprising: Quantum dots with different fluorescence wavelengths are modified on the sensor surface of the electronic odor capture array to form a fluorescence-electrochemical dual signal channel. The fluorescence intensity ratio of the quantum dots and the electrochemical signal are detected based on the fluorescence-electrochemical dual signal channel, and the three-dimensional data of the traditional Chinese medicine to be detected is obtained by combining the time information; The three-dimensional data is marked with time anchor points based on characteristic peaks combined with convolutional neural networks, and the marked results are matched with the traditional Chinese medicine component database to generate a dynamic standard curve; The content of the active ingredient in the Chinese medicine to be tested is calculated based on the dynamic standard curve, and the impurities are semi-quantitatively analyzed to obtain the content of the impurity component. The third quality of the Chinese medicine to be tested is obtained based on the content of the active ingredient and the content of the impurity component.

[0010] The present invention provides a method for detecting the quality of traditional Chinese medicine, which performs conflicting data arbitration based on the first quality, the second quality, and the third quality to obtain a fourth quality, and performs environmental dynamic compensation based on the fourth quality to obtain a fifth quality, comprising: Comparing the first quality, the second quality, and the third quality, identifying numerical contradictions, determining the first conflict type, identifying safety threshold conflicts, determining the second conflict type, identifying authenticity disagreements, determining the third conflict type; constructing a three-level arbitration logic framework based on the first conflict type, the second conflict type, and the third conflict type, and determining a fourth quality according to the three-level arbitration logic framework; Acquire environmental data, determine environmental-quality impact parameters of the Chinese medicine to be tested from an environmental-quality impact library based on the environmental data and surface features corresponding to the Chinese medicine to be tested, set an environmental dynamic compensation formula based on the environmental-quality impact parameters, and derive the fifth quality by combining the fourth quality.

[0011] The present invention provides a method for detecting the quality of traditional Chinese medicine. The method marks the upstream planting end according to the fifth quality and pushes optimization suggestions upward. The method also identifies the historical markings and the execution status of the optimization suggestions to obtain the current marking of the corresponding upstream planting end. The method combines the current markings with a three-dimensional compensation system to compensate and adjust the fifth quality to obtain the final quality. The method includes: The fifth quality is anchored to the planting batch ID through blockchain, and an NFT traceability certificate is generated to record the quality grade, key defect parameters and environmental compensation records of the corresponding planting batch ID, and the upstream planting end is marked according to the NFT traceability certificate; Directly push optimization suggestions based on the key defect parameters and environmental compensation records, obtain the execution status of the optimization suggestions by the upstream planting end, and determine the current mark based on the execution status and historical marks; Adjust the compensation factors for the three dimensions in the preset three-dimensional compensation system based on the current mark, and perform compensation adjustment on the fifth quality according to the compensation factors to obtain the final quality.

[0012] The present invention provides a traditional Chinese medicine quality detection system, including: The first quality determination module: scan the traditional Chinese medicine to be detected, obtain the surface characteristics of the traditional Chinese medicine to be detected, compare the surface characteristics with the genuine standard characteristics, and determine the first quality; The second and third quality determination modules: use an electronic odor capture array to record the response curve of the volatile components of the traditional Chinese medicine to be detected, identify the characteristic peaks, determine the second quality, and perform component analysis on the traditional Chinese medicine to be detected based on the characteristic peaks, and then determine the third quality; The fourth and fifth quality determination modules: perform arbitration on conflicting data based on the first quality, the second quality, and the third quality to obtain the fourth quality, and perform environmental dynamic compensation based on the fourth quality to obtain the fifth quality; The final quality determination module: mark the upstream planting end according to the fifth quality and push optimization suggestions upward. At the same time, identify the execution situation of the historical mark and the optimization suggestions to obtain the current mark corresponding to the upstream planting end, and perform compensation adjustment on the fifth quality in combination with the current mark and the three-dimensional compensation system to obtain the final quality.

[0013] Compared with the prior art, the beneficial effects of the present application are as follows: By performing multi-modal data fusion on the surface characteristics, volatile components, and component analysis of traditional Chinese medicine, using conflict data arbitration and environmental dynamic compensation, combined with blockchain traceability and a three-dimensional compensation system, the comprehensive detection and optimization of traditional Chinese medicine quality are realized, the accuracy and efficiency of traditional Chinese medicine quality detection are improved, data sharing and collaboration in the whole chain of traditional Chinese medicine planting, processing, and circulation are realized, and the sustainable development of the traditional Chinese medicine industry is promoted.

[0014] Other features and advantages of the present invention will be described in the following description, and part of them will become obvious from the description, or be understood by implementing the present invention. The objectives and other advantages of the present invention can be achieved and obtained by the structure specifically pointed out in the written description and the drawings.

[0015] The technical solutions of the present invention will be further described in detail below with reference to the drawings and embodiments. Description of the Drawings

[0016] The drawings are used to provide a further understanding of the present invention, and constitute a part of the description. They are used to explain the present invention together with the embodiments of the present invention, and do not constitute a limitation to the present invention. In the drawings: Figure 1 is a schematic flowchart of a traditional Chinese medicine quality detection method provided by an embodiment of the present invention; Figure 2 It is a schematic structural diagram of a traditional Chinese medicine quality detection system provided by an embodiment of the present invention. Specific embodiments

[0017] The following is a description of the preferred embodiments of the present invention in conjunction with the accompanying drawings. It should be understood that the preferred embodiments described herein are only for the purpose of illustrating and explaining the present invention, and are not used to limit the present invention. Embodiment 1:

[0018] An embodiment of the present invention provides a method for detecting the quality of traditional Chinese medicine, as Figure 1 shown, including: Step 1: Scan the traditional Chinese medicine to be detected, obtain the surface characteristics of the traditional Chinese medicine to be detected, compare the surface characteristics with the standard characteristics of genuine regional origin, and determine the first quality; Step 2: Use an electronic odor capture array to record the response curve of the volatile components of the traditional Chinese medicine to be detected, identify the characteristic peaks, determine the second quality, and perform component analysis on the traditional Chinese medicine to be detected based on the characteristic peaks, and then determine the third quality; Step 3: Arbitrate conflicting data based on the first quality, the second quality, and the third quality to obtain the fourth quality, and perform environmental dynamic compensation based on the fourth quality to obtain the fifth quality; Step 4: Mark the upstream planting end according to the fifth quality, and push optimization suggestions upward. At the same time, identify the implementation status of the historical marks and optimization suggestions to obtain the current marks of the corresponding upstream planting end. Combine the current marks with the three-dimensional compensation system to compensate and adjust the fifth quality to obtain the final quality.

[0019] In this embodiment, the first quality is obtained by comparing the surface characteristics with the standard of genuine regional origin, including excellent, 80 - 100 points: fully meeting the standards of genuine regional origin herbs, such as texture, color, size, etc., without any defects, such as mildew spots, insect damage; qualified, 60 - 79 points: the main characteristics meet the standards, allowing slight deviations, such as slightly lighter color, but the active ingredients are not affected; unqualified, 0 - 59 points: significantly deviating from the genuine regional origin standard, such as severe deformation, pollution, or having prohibited defects, such as excessive sulfur fumigation. For example, excellent ginseng: the root bowl is clear, the main root is thick, and there are no scars; qualified astragalus: the color is slightly lighter but there is no mildew; unqualified angelica: the area of mildew spots on the surface > 5%.

[0020] In this embodiment, the second quality is obtained based on the volatile component response curve, including excellent: high characteristic peak intensity, such as the main peak ≥ 250 mV, peak plateau ratio > 3, and volatile oil content exceeding the standard by more than 20%; qualified: the main peak intensity meets the standard, such as 150 - 250 mV, peak plateau ratio > 2, meeting the minimum medicinal requirements; unqualified: the main peak is missing or the intensity is insufficient, such as < 100 mV, or abnormal peaks are present, such as sulfide peaks. For example, excellent Pogostemon cablin: the peak intensity of patchouli alcohol is 280 mV, without miscellaneous peaks; qualified Citrus reticulata Blanco: the peak intensity of limonene is 180 mV, with trace volatiles; unqualified Atractylodes macrocephala: a pesticide residue peak is detected.

[0021] In this embodiment, the third quality is obtained based on the quantification of active ingredients and impurities, including excellent: the content of active ingredients ≥ 120% of the pharmacopoeia standard, such as saponins ≥ 1.2%, and no impurities are detected; qualified: the active ingredients reach 80% - 120% of the standard, such as 0.8% - 1.2%, and trace impurities, such as ash content < 5%; unqualified: the active ingredients < 80% or harmful impurities exceed the standard, such as heavy metals and aflatoxins. For example, excellent Salvia miltiorrhiza: the content of tanshinone IIA is 1.5%, and the standard ≥ 1.0%; qualified Glycyrrhiza uralensis Fisch: the content of glycyrrhizic acid is 0.9%, the standard ≥ 1.0%, and the ash content is 3%; unqualified Poria cocos: the starch adulteration > 10%.

[0022] In this embodiment, the fourth quality is to comprehensively arbitrate the first three types of ratings according to the priority, such as safety > ingredients > appearance. For example: excellent in the first quality + unqualified in the second quality → downgraded to unqualified, and abnormal odor may cause pollution; the fifth quality is adjusted in combination with environmental factors. For example, the color requirement is relaxed in dry areas; the final quality: compensated by upstream planting data. For example, organic planting can improve the rating and output the conclusion. Arbitration case: The appearance of ginseng is excellent but the volatile oil is insufficient, downgraded to qualified; Compensation case: The surface was misjudged due to excessive humidity during detection, and it was corrected to excellent through data traceability.

[0023] In this embodiment, the surface features are digital descriptions that comprehensively consider geometric and spectral characteristics, including: average roughness, principal component spectra, characteristic absorption peak positions, etc., with more than 20 - dimensional feature vectors. For example, the standard features of genuine Fritillaria cirrhosa D. Don: Ra = 2.5 ± 0.3 μm, the peak value of PC1 load is at 1380 ± 5 nm, and the peak area = 120 ± 15 au.

[0024] In this embodiment, the standard features of genuine regional products are the digital standard library of medicinal materials from the authentic production areas, including the allowable ranges of surface topography, spectral characteristics, and component contents. A confidence interval of mean ± 3σ is established through big data analysis to support dynamic updates. For example, the standard of Panax notoginseng from Wenshan: the epidermal roughness is 1.8 - 2.2 μm, and the half - height width of the saponin characteristic peak at 1220 nm ≤ 15 nm.

[0025] In this embodiment, the electronic odor capture array consists of a bifurcated odor sampling head and a multimodal sensor, simulating the biological olfactory system to collect and analyze the volatile gases of traditional Chinese medicine in real time. Its core functions include dynamic baseline calibration, multi-channel signal synchronous acquisition, data noise reduction, etc. Through array detection, a fingerprint spectrum of the odor of traditional Chinese medicine can be constructed for quality grading and authenticity identification.

[0026] In this embodiment, the characteristic peak is a significant peak in the response curve, corresponding to a specific volatile component. For example, the peak of atractylodin is located at 120 seconds. The intensity corresponding to its height and the duration corresponding to its width can be used to quantify the component content. For example, a high main peak intensity usually indicates a high volatile oil content.

[0027] In this embodiment, the process of component analysis is to modify quantum dots with different fluorescence wavelengths on the surface of the sensors of the electronic odor capture array to form a fluorescence-electrochemical dual-signal channel, detect the fluorescence intensity ratio and electrochemical signals, and combine time information to obtain three-dimensional data; use characteristic peaks and time anchor points to mark the three-dimensional data, and combine with the traditional Chinese medicine component database to generate a dynamic standard curve; calculate the content of active ingredients, perform semi-quantitative analysis of impurities to obtain the content of impurity components, and determine the third quality by comprehensively considering the content of active ingredients and impurities.

[0028] In this embodiment, the data arbitration process is divided into three stages: conflict identification and classification. Compare the first appearance, second odor, and third component quality data to identify three types of conflicts: numerical contradictions, such as excellent appearance but qualified components; safety threshold conflicts, such as direct rejection if heavy metals exceed the standard; provenance disagreements, such as authentic appearance but inconsistent components. The three-level arbitration logic processes according to the priority: safety conflicts, highest > provenance > numerical contradictions; output the fourth quality. For example, if there is a safety conflict, it is directly judged as unqualified, and numerical contradictions are downgraded to commercial grade. Environmental compensation and correction. According to environmental data such as temperature and humidity, retrieve parameters from the environment-quality impact library. For example, the humidity compensation coefficient is 0.003 / %; apply the dynamic compensation formula to adjust the fourth quality to generate the fifth quality. For example, after compensation, it upgrades from 78 points to 85 points and becomes excellent. For example, for a batch of ginseng, due to the high detection temperature resulting in the attenuation of the odor signal, after arbitration and compensation, the final quality is corrected from qualified to excellent.

[0029] In this embodiment, the environmental dynamic compensation is executed by obtaining environmental parameters to construct an environmental dynamic compensation formula.

[0030] In this embodiment, marking the upstream planting end is to label the growers according to NFT data, such as high-quality suppliers or sulfur fumigation risk households. The marking affects the subsequent procurement weight. For example, a certain cooperative has obtained the AAA label for three consecutive batches of excellent products and is given priority to receive orders.

[0031] In this embodiment, the optimization suggestions are improvement plans pushed for defects, such as reducing sulfur fumigation: switching to hot air drying. The suggestions need to be specific and executable, and bound to the batch ID. Example: Push to farmers to control water for 10 days before harvesting in September to increase the saponin content.

[0032] In this embodiment, the implementation situation is to record the implementation of suggestions by the upstream, such as having purchased hot air equipment or not having rectified. Verified through Internet of Things devices, such as soil sensors or manual audits. For example: The sensor shows that the humidity compliance rate has increased from 60% to 90%.

[0033] In this embodiment, the quality grade is anchored to the planting batch through the blockchain to generate an NFT certificate containing environmental compensation parameters. Intelligent push planting optimization suggestions based on defect analysis and track the execution feedback. Dynamically adjust the three-dimensional compensation factors of time - geography - process in combination with historical markers.

[0034] In this embodiment, the final quality is the final rating after three-dimensional compensation. For example: The original fifth quality is 78 points; planting technology compensation +4 points, environmental compensation +2 points → final quality 84 points, upgraded to excellent.

[0035] The working principle and beneficial effects of the above technical solution are: Through multi-modal data fusion of the surface characteristics, volatile components and component analysis of traditional Chinese medicine, using conflict data arbitration and environmental dynamic compensation, combined with blockchain traceability and three-dimensional compensation system, realize the comprehensive detection and optimization of the quality of traditional Chinese medicine, improve the accuracy and efficiency of traditional Chinese medicine quality detection, realize data sharing and collaboration in the whole chain of traditional Chinese medicine planting, processing and circulation, and promote the sustainable development of the traditional Chinese medicine industry. Example 2:

[0036] An embodiment of the present invention provides a method for detecting the quality of traditional Chinese medicine, scanning the traditional Chinese medicine to be detected to obtain the surface characteristics of the traditional Chinese medicine to be detected, comparing the surface characteristics with the standard characteristics of the genuine region, and determining the first quality, including: Use a hyperspectral camera to scan the traditional Chinese medicine to be detected to obtain the hyperspectral pixel coordinates of the traditional Chinese medicine to be detected, and use structured light to scan the traditional Chinese medicine to be detected to obtain the 3D point cloud coordinates of the traditional Chinese medicine to be detected; Perform spatio-temporal alignment on the hyperspectral pixel coordinates and the 3D point cloud coordinates through the SIFT algorithm, map the hyperspectral data obtained from the hyperspectral pixel coordinates after alignment to the 3D point cloud coordinates, perform spectral-geometry correlation analysis and defect feature quantification to determine the surface characteristics of the traditional Chinese medicine to be detected; Based on the comparison between the surface characteristics and the standard characteristics of the genuine region, trigger the preset knowledge graph rules according to the comparison results to determine the abnormal results, retrieve and match the associated nodes corresponding to the abnormal results in the preset knowledge graph, and determine the first quality based on the associated nodes and the corresponding surface characteristics.

[0037] In this embodiment, the hyperspectral camera is a device that can capture hundreds of consecutive narrow-band spectral images. Each pixel contains a complete spectral curve from visible light to near-infrared. Non-destructive detection of chemical components is achieved through the unique spectra of substances. The spectral resolution can reach 5nm, and the spatial resolution depends on the lens configuration. For example, when detecting Astragalus membranaceus slices, an abnormally increased absorbance is found at 1680nm, indicating that the polysaccharide content exceeds the standard. The normal value should be <0.25, and the measured value is 0.32.

[0038] In this embodiment, the hyperspectral pixel coordinates are a three-dimensional matrix that records the spatial position of each pixel point and the corresponding spectral data. Each coordinate point contains position information and reflectance / absorbance values of hundreds of bands, constituting hypercube data. For example, the reflectance of a certain pixel point drops suddenly by 30% at 1450nm, and the corresponding area is confirmed to be a mildewed depression through 3D scanning.

[0039] In this embodiment, structured light is a technique for calculating the three-dimensional shape of an object by projecting a specifically encoded light pattern, such as blue light stripes, and calculating based on the deformation of the pattern. The phase-shift method can achieve a depth resolution of 0.01mm and is applicable to scanning complex surfaces. For example, when projecting a sinusoidal stripe light onto the surface of ginseng, a 0.2mm deep insect-eaten channel is found through deformation reconstruction, and the normal surface undulation is <0.05mm.

[0040] In this embodiment, the 3D point cloud coordinates are a set of three-dimensional coordinates of the geometric features on the object surface. Each point contains spatial position and normal vector information. Parameters such as curvature and roughness can be calculated through a dense point cloud, and the point spacing can reach an accuracy of 50μm. For example, the point cloud of Angelica sinensis slices shows that the edge thickness difference >15%, and the normal is <8%, indicating uneven drying.

[0041] In this embodiment, the SIFT algorithm is a scale-invariant feature transform algorithm that realizes robust matching of multi-source data by detecting key points in the image, such as corner points and edges, and generating 128-dimensional descriptors. It is invariant to rotation, scaling, and illumination changes.

[0042] In this embodiment, the hyperspectral data is a multi-band spectral reflectance / absorbance matrix at each spatial position, usually containing 200 - 300 bands. The composition of substances can be quantified through features such as derivative spectra and absorption peak area. For example, the reflectance curve of Salvia miltiorrhiza shows a double-peak feature at 550nm, while the normal is a single peak, indicating the doping of other plant rhizomes.

[0043] In this embodiment, spectral-geometric correlation analysis is a method of jointly modeling spectral features with three-dimensional shape parameters, such as curvature and roughness. A mapping relationship of spectrum-morphology-defect is established through machine learning to achieve multi-dimensional feature fusion. For example, the area that simultaneously satisfies the reflectance at 650nm <0.3 and the local curvature > is determined to be insect-eaten, with an accuracy of 98.7%.

[0044] In this embodiment, the quantification of defect features is to convert the abnormal area into computable metrics: area ratio, average depth, spectral shift, intensity deviation, etc. The defect level is determined by a threshold. For example, for a batch of pseudo-ginseng, the area ratio of mildew is 8.3%, the standard is <3%, and the average spectral shift Δ = 12.5 nm, which is determined to be severely polluted.

[0045] In this embodiment, the comparison result is a quantification report of the differences between the sample to be tested and the standard features, including: Euclidean distance, Mahalanobis distance, similarity score, etc. The comprehensive deviation degree is obtained through multi-index weighted fusion. For example, the morphological similarity of a certain astragalus sample is 92%, but the spectral Mahalanobis distance reaches 4.8, and the threshold is <3, which is determined to be suspected of doping.

[0046] In this embodiment, the preset knowledge graph rules are semantic networks containing the association relationships of feature - cause - measure, and are stored using RDF triples. Probabilistic reasoning is supported, such as: IF hygroscopicity defect, confidence 85% THEN associate with the warehouse humidity node. For example, detecting an increase in the peak width at 1450 nm + surface porosity triggers the rule: associate with the node of excessive sulfur fumigation, probability 78%.

[0047] In this embodiment, the abnormal result is a feature deviation event exceeding the preset threshold, recording: abnormal type, deviation degree, spatial distribution. Hierarchical early warning is supported, mild / moderate / severe. For example, 3 local reflectivity abnormalities are found in a certain angelica batch, Δ>20%, and the system marks it as moderately chemically polluted.

[0048] In this embodiment, the associated node is an entity in the knowledge graph that has a causal relationship with the abnormal feature, including: processing technology, storage conditions, supplier information, etc. Potential roots are mined through graph traversal algorithms. For example, the surface dark spot node is associated with: too high drying temperature, weight 0.6, warehouse mildew, weight 0.3, germplasm degradation, weight 0.1.

[0049] The working principle and beneficial effects of the above technical solution are: the spectral features and three-dimensional geometric data of traditional Chinese medicine are obtained through a hyperspectral camera and structured light scanning respectively, and the accurate spatio-temporal alignment of multi-source data is realized using the SIFT algorithm. The surface defect features are quantified through spectral - geometric correlation analysis, and the authentic standard features are intelligently matched in combination with the knowledge graph to determine the first quality of the traditional Chinese medicine to be detected, realizing the objective evaluation of the surface quality of traditional Chinese medicine, breaking through the subjective limitations of traditional sensory evaluation, improving the detection accuracy, and adapting to complex working conditions. Embodiment 3:

[0050] The embodiment of the present invention provides a method for detecting the quality of traditional Chinese medicine, which uses an electronic odor capture array to record the response curve of the volatile components of the traditional Chinese medicine to be detected, identify the characteristic peaks, and determine the second quality, including: Set a bifurcated odor sampling head, combine the bifurcated odor sampling head with a multimodal sensor to form an electronic odor capture array, and determine the scanning mode of the electronic odor capture array according to the surface characteristics of the traditional Chinese medicine to be detected; Generate a time-response matrix based on the electronic odor capture array, and inhale pure air every preset time to reset the sensor baseline, so as to determine the response curve of the volatile components of the traditional Chinese medicine to be detected; Calculate the first derivative of the response curve, set a sliding window according to the scanning mode, determine the characteristic peaks of the response curve, determine the main peak intensity based on the characteristic peaks, use the sliding window centered on the main peak intensity as the post-peak interval to calculate the platform stability, and then obtain the peak platform ratio. Comprehensively determine the identification conditions based on the main peak intensity, platform stability, and peak platform ratio, and obtain the second quality of the traditional Chinese medicine to be detected based on the identification conditions.

[0051] In this embodiment, the bifurcated odor sampling head is a multi-channel gas collection device with a bifurcated structure design. It can simultaneously collect volatile components from different parts of traditional Chinese medicine, such as roots, stems, and leaves, avoiding errors caused by single sampling. Its bifurcated end is equipped with a micro air pump to control the gas flow rate and ensure sampling uniformity. It is suitable for traditional Chinese medicine with complex surfaces, such as wrinkles and cracks, improving the coverage rate and accuracy of odor detection.

[0052] In this embodiment, the multimodal sensor is a composite detection module integrating multiple sensing technologies, such as metal oxide semiconductors, electrochemistry, and photoionization detectors. It can simultaneously detect different categories of volatile organic compounds, VOCs. For example, metal oxide sensors are sensitive to alcohols and aldehydes, and electrochemical sensors respond strongly to sulfides. Multimodal data fusion can improve the specificity of odor recognition and reduce misjudgment.

[0053] In this embodiment, the scanning mode refers to the sampling strategy of the electronic odor capture array. According to the surface characteristics of traditional Chinese medicine, such as humidity and roughness, the sampling path, gas flow rate, and sensor activation sequence are dynamically adjusted. For example, high-frequency sampling is used for dry medicinal materials, and the flow rate is reduced for wet medicinal materials to prevent condensation interference. Optimizing the scanning mode can improve the detection efficiency and reduce invalid data.

[0054] In this embodiment, the time-response matrix is a data table output by the electronic odor capture array, recording the signal intensities of each sensor at different time points. The horizontal axis is time, and the vertical axis is the sensor channel. The matrix data is used to plot the response curve and analyze the dynamic release law of volatile components.

[0055] In this embodiment, the preset time is the interval for the sensor to automatically reset the baseline, such as every 30 seconds. Residual gas is removed by inhaling pure air to avoid signal drift. For example, when detecting highly volatile mint, frequent resetting, such as every 20 seconds, can prevent sensor saturation and ensure data accuracy.

[0056] In this embodiment, the response curve is a line graph showing the change of a single sensor signal over time, reflecting the release dynamics of specific components. For example, the volatile components of Astragalus membranaceus, such as β-pinene, will exhibit a peak-shaped curve with a rapid rise and a slow decline, while moldy medicinal materials may show miscellaneous peaks or irregular fluctuations.

[0057] In this embodiment, the sliding window is a time interval set during data analysis, such as a width of 5 seconds, for local feature extraction. For example, by sliding and calculating the first derivative on the response curve, the starting point and peak position of the characteristic peak can be accurately identified, avoiding noise interference.

[0058] In this embodiment, the main peak intensity is the highest signal value of the characteristic peak, directly reflecting the concentration of the target component. For example, for high-quality Pogostemon cablin, the main peak intensity of the patchouli alcohol peak should be ≥250 mV, and values below the threshold may indicate inferior products.

[0059] In this embodiment, the post-peak interval is a section of data after the end of the main peak, such as 20 seconds after the peak, for evaluating signal stability. If the volatility in this interval >5%, such as for moldy medicinal materials, it indicates the presence of interfering components; a stable platform indicates pure component release.

[0060] In this embodiment, the platform stability is an index of the smoothness of the signal in the post-peak interval, calculated as the standard deviation or volatility. For example, the platform fluctuation of qualified medicinal materials should be <3%, and if >10%, it may be doped with other impurities, such as sulfur fumigation residues.

[0061] In this embodiment, the peak-to-platform ratio is the ratio of the main peak intensity to the average signal in the post-peak interval, reflecting the component purity. For example, the peak-to-platform ratio of genuine Aquilaria sinensis >5, indicating a high-purity resin component, while adulterated products may be <2.

[0062] In this embodiment, the discrimination conditions are a combination of rules for determining the second quality, for example: main peak intensity >200 mV, peak-to-platform ratio >3, platform stability <5%. Traditional Chinese medicine that meets the conditions simultaneously can be classified as the second quality, such as commercial-grade medicinal materials that meet the standards but are not the best.

[0063] The working principle and beneficial effects of the above technical solution are: setting up a forked odor sampling head combined with a multi-modal sensor to form an electronic odor capture array, determining the scanning mode according to the surface characteristics of the traditional Chinese medicine to be detected, generating a time-response matrix and regularly resetting the sensor baseline, calculating the first derivative of the response curve, determining the characteristic peak and main peak intensity, calculating the platform stability to obtain the peak-to-platform ratio, and comprehensively determining the discrimination conditions based on the main peak intensity, platform stability, and peak-to-platform ratio to obtain the second quality of the traditional Chinese medicine to be detected, improving the accuracy and efficiency of detecting the volatile components of traditional Chinese medicine. Example 4:

[0064] An embodiment of the present invention provides a method for detecting the quality of traditional Chinese medicine. A bifurcated odor sampling head is set, and the bifurcated odor sampling head is combined with a multimodal sensor to form an electronic odor capture array. The scanning mode of the electronic odor capture array is determined according to the surface characteristics of the traditional Chinese medicine to be detected, including: Adopt a dual-channel bifurcated structure. The dual-channel bifurcated structure adopts bilateral independent air extraction, and bilateral integrated independent micro air pumps can control different flow rates. Multimodal sensors of the same model are arranged at the end of each channel to form an electronic odor capture array; The traditional Chinese medicine to be detected is divided into regions according to the surface characteristics, and the dynamic configuration of the multimodal sensor is determined according to the region division result. At the same time, a corresponding capture algorithm is set for each column in the electronic odor capture array; The scanning mode is determined based on the dynamic configuration and the capture algorithm.

[0065] In this embodiment, the dual-channel bifurcated structure is a gas sampling device. It adopts a symmetric bifurcated design and has two independent gas collection channels. Each channel can synchronously collect volatile gases from different parts of traditional Chinese medicine, such as rhizomes and leaves, avoiding the limitations of single-point sampling. The structure uses anti-adsorption materials, such as polytetrafluoroethylene, to reduce residues and is suitable for detecting traditional Chinese medicine with complex surfaces, such as wrinkled and cracked ones, improving the spatial resolution of odor data.

[0066] In this embodiment, bilateral independent air extraction means that each of the two bifurcated channels is equipped with a micro air pump, and the air extraction rate can be independently adjusted. For example, the left channel is 0.5 L / min and the right channel is 1 L / min. For example, when detecting moist medicinal materials, the air extraction speed in the high-humidity area is reduced to prevent condensation; in the dry area, the flow rate is increased to enhance the signal. This design can adapt to traditional Chinese medicine with different materials and volatile characteristics and optimize the gas collection efficiency.

[0067] In this embodiment, region division is to divide traditional Chinese medicine into multiple detection regions according to its surface characteristics, such as color, texture, and humidity, such as region A for rhizomes and region B for leaves. The physical differences of each region are automatically identified through hyperspectral or 3D imaging technology, providing a basis for subsequent dynamic configuration of sensors. For example, the mildewed area may show dark spots and requires key sampling.

[0068] In this embodiment, dynamic configuration refers to adjusting the activation strategy of the sensor according to the region division result. For example: in the high-volatility area, such as fresh leaves: activate the electrochemical sensor to detect aldehydes; in the low-volatility area, such as dry rhizomes: activate the metal oxide sensor to detect terpenes.

[0069] In this embodiment, the capture algorithm is a data analysis method corresponding to each column of sensors. For example: for fast-volatile components: use short-time window Fourier transform to extract transient features; for slow-release components: use moving average filtering to smooth the signal.

[0070] In this embodiment, for ginseng detection, the area is divided as follows: the rhizome is area A, and the main root is area B; for dynamic configuration: high-sensitivity sensors are enabled in area A to detect saponin volatiles, and moisture-resistant sensors are enabled in area B; for the capture algorithm: transient peak detection is used in area A, and steady-state analysis is used in area B.

[0071] The working principle and beneficial effects of the above technical solution are as follows: A dual-channel bifurcated structure is adopted, and different flow rates are controlled through bilateral independent air extraction and a micro air pump. A multi-modal sensor array for electronic odor capture is arranged at the end; the area is divided according to the surface characteristics of traditional Chinese medicine, multi-modal sensors are dynamically configured and a capture algorithm is set to determine the scanning mode, so as to more accurately capture and identify the odor information of different areas of traditional Chinese medicine, and provide more comprehensive data support for the quality control of traditional Chinese medicine. Example 5:

[0072] The embodiment of the present invention provides a method for detecting the quality of traditional Chinese medicine. Based on the characteristic peaks, the components of the traditional Chinese medicine to be detected are analyzed, and then the third quality is determined, including: Quantum dots with different fluorescence wavelengths are modified on the surface of the sensors of the electronic odor capture array to form a fluorescence-electrochemical dual-signal channel. Based on the fluorescence-electrochemical dual-signal channel, the fluorescence intensity ratio of the quantum dots and the electrochemical signal are detected, and three-dimensional data of the traditional Chinese medicine to be detected are obtained by combining time information; Based on the characteristic peaks, the three-dimensional data are marked with time anchors by combining a convolutional neural network, and a dynamic standard curve is generated according to the matching result with the traditional Chinese medicine ingredient database; Based on the dynamic standard curve, the content of the active ingredients in the traditional Chinese medicine to be detected is calculated, and at the same time, a semi-quantitative analysis of the impurities is carried out to obtain the content of the impurity components. According to the content of the active ingredients and the content of the impurity components, the third quality of the traditional Chinese medicine to be detected is obtained.

[0073] In this embodiment, quantum dots QDs are a kind of nanoscale semiconductor material with adjustable fluorescence characteristics. Quantum dots with different fluorescence wavelengths are modified on the surface of the sensors, such as CdSe / ZnS QDs emitting 520nm and 620nm, so that they produce specific fluorescence responses to different chemical components. For example, 520nm QDs are sensitive to flavonoids, and 620nm QDs respond stronger to alkaloids. This modification can enhance the selectivity of the sensor and the ability to detect multiple components.

[0074] In this embodiment, the dual-signal channel refers to simultaneously collecting the fluorescence signal, optical and electrochemical signals, current / impedance of the quantum dots. The fluorescence signal reflects the concentration of specific components, and the electrochemical signal provides redox characteristics. For example, when detecting Astragalus membranaceus, the fluorescence channel captures astragaloside IV at 520nm, and the electrochemical channel monitors phenolic substances. The dual-channel data are complementary to improve the detection reliability.

[0075] In this embodiment, the fluorescence intensity ratio is the signal ratio of QDs with different wavelengths, such as I520 / I620, which is used to eliminate environmental interference; the electrochemical signal, such as the peak current, quantifies the content of the active ingredient. For example, when I520 / I620 > 1.5 and the current > 10 μA, it is determined as the active ingredient, such as the enrichment of saponins. The combination of the two can distinguish compounds with similar structures.

[0076] In this embodiment, the three-dimensional data consists of time on the X-axis, fluorescence / electrochemical signal intensity on the Y-axis, and wavelength or potential on the Z-axis. For example, a certain component shows a fluorescence peak at 520 nm with an intensity of 2000 a.u. and an oxidation peak at 0.5 V with a current of 15 μA at t = 30 s. Such data can comprehensively characterize the dynamic release process of traditional Chinese medicine components.

[0077] In this embodiment, the time anchor marking is to identify the characteristic peaks in the three-dimensional data, such as fluorescence peaks and oxidation peaks, through a convolutional neural network (CNN), and mark their occurrence times. For example, the peak of astragaloside IV is located at t = 25 s. The anchor points are used to align data from different batches to ensure the temporal consistency when matching the database.

[0078] In this embodiment, the dynamic standard curve is generated by matching the traditional Chinese medicine component database according to the time anchor points, and it is a calibration curve that changes with time, such as the fluorescence intensity-concentration relationship curve. For example, the standard curve of baicalin at t = 40 s is y = 50x + 100, with R² > 0.99, which is used to calculate the content in real time.

[0079] In this embodiment, the content of the active ingredient is quantified by the dynamic standard curve for the target component, such as ginsenoside Rg1. For example, when the measured fluorescence intensity of a sample is 800 a.u., the corresponding content is 2.3 mg / g. The content needs to meet the pharmacopoeia standards, such as ≥ 0.8%, otherwise, the quality is determined to be insufficient.

[0080] In this embodiment, semi-quantitative analysis estimates the content of unknown impurities, such as sulfides, through the signal intensity range, such as low / medium / high. For example, if a sulfur peak is detected at 850 nm with an intensity of 300 a.u., but there is no standard curve, the content is determined to be medium based on historical data.

[0081] The working principle and beneficial effects of the above technical solutions are as follows: Quantum dots with different fluorescence wavelengths are modified on the surface of the sensors of the electronic odor capture array to form a fluorescence-electrochemical dual-signal channel. The fluorescence intensity ratio and electrochemical signal are detected, and three-dimensional data is obtained by combining time information. The three-dimensional data is marked with characteristic peaks and time anchor points, and a dynamic standard curve is generated by combining the traditional Chinese medicine component database. The content of the active ingredient is calculated, and semi-quantitative analysis of impurities is carried out to obtain the content of impurity components. The third quality is determined by comprehensively considering the content of the active ingredient and impurities, realizing the synchronous quantitative analysis of the active ingredient and impurities, and completing the accurate evaluation of the quality of traditional Chinese medicine. Example 6:

[0082] An embodiment of the present invention provides a method for detecting the quality of traditional Chinese medicine. Based on the first quality, the second quality, and the third quality, conflict data arbitration is performed to obtain the fourth quality. Based on the fourth quality, environmental dynamic compensation is performed to obtain the fifth quality, including: Compare the first quality, the second quality, and the third quality, identify numerical contradictions, determine the first conflict type, identify safety threshold conflicts to determine the second conflict type, identify differences in genuineness, and determine the third conflict type; Based on the first conflict type, the second conflict type, and the third conflict type, construct a three-level arbitration logic framework, and determine the fourth quality according to the three-level arbitration logic framework; Obtain environmental data, determine the environmental-quality impact parameters for the traditional Chinese medicine to be detected from the environmental-quality impact library according to the environmental data and the surface characteristics corresponding to the traditional Chinese medicine to be detected, set an environmental dynamic compensation formula according to the environmental-quality impact parameters, and combine the fourth quality to obtain the fifth quality.

[0083] In this embodiment, the numerical contradiction refers to the inconsistent rating results of the first quality, appearance, the second quality, odor, and the third quality, composition. For example: excellent appearance, first quality, but unqualified odor detection, second quality, or insufficient component content, third quality. Such contradictions need to be arbitrated through conflict type analysis to ensure the scientific and reasonable final rating.

[0084] In this embodiment, the first conflict type is the rating disagreement caused by differences in detection methods. For example: ginseng has excellent appearance and no defects, but the volatile oil content is only qualified, second quality, or the saponin content is slightly low, third quality. When arbitrating, the component safety should be considered first, and then the weights of appearance and odor should be coordinated.

[0085] In this embodiment, the safety threshold conflict means that the detection result touches the safety red line, such as excessive heavy metals or pesticide residues, which conflicts with other quality ratings. For example: angelica has excellent appearance, but aflatoxin is detected and the third quality is unqualified. Such conflicts directly veto other ratings and are judged as unqualified.

[0086] In this embodiment, the second conflict type is the conflict between safety indicators and other qualities, and it has the highest priority during arbitration. For example: astragalus has excellent odor, but the sulfur residue exceeds the standard and the third quality is unqualified. Even if the appearance is qualified, it is finally judged as unqualified.

[0087] In this embodiment, the difference in genuineness refers to that the appearance of the medicinal material conforms to the genuine standard, the first quality is excellent, but the composition or odor does not conform to the genuine characteristics. For example: sichuan fritillary has qualified appearance, but the alkaloid content is lower than the genuine standard and the third quality is qualified. It is necessary to further verify by combining the origin data.

[0088] In this embodiment, the third type of conflict is the contradiction between the authentic features and the ingredients / odor. For example: The texture of Zhejiang Atractylodes macrocephala Koidz. conforms to the authentic standard, and the first-class quality is excellent, but the volatile oil content is insufficient, and the second-class quality is qualified. During arbitration, it can be downgraded to the commercial grade, not the authentic excellent grade.

[0089] In this embodiment, the three-level arbitration logic processes conflicts according to the priority: safety conflicts, with the highest priority, directly vetoed; authenticity conflicts, which need to be combined with origin data; numerical contradictions, coordinating the weights of appearance, odor, and ingredients. For example: Excellent appearance but qualified ingredients → downgraded to the commercial grade.

[0090] In this embodiment, the fourth-class quality is the preliminary conclusion after three-level arbitration. For example: Arbitration result 1: Excellent appearance + qualified odor + qualified ingredients → downgraded to qualified; Arbitration result 2: Safety conflict → directly unqualified.

[0091] In this embodiment, the environmental data includes temperature, humidity, light, storage conditions, etc. during detection. For example: High humidity may cause the surface to be misjudged as mildew, and the first-class quality is downgraded, which needs to be corrected through environmental compensation.

[0092] In this embodiment, the environment-quality impact library stores the impact rules of environmental factors on quality. For example: When the humidity > 80%, the misjudgment rate of appearance detection +15%; the volatile oil signal decays by 20% under high temperature.

[0093] In this embodiment, the environment-quality impact parameter is a parameter that quantifies the impact of the environment on the rating. For example: Humidity compensation coefficient: If the humidity exceeds the standard, the appearance score × 1.1; Temperature compensation formula: Volatile oil signal = original signal / (1 + 0.02 × (T - 25)). 12. Environmental dynamic compensation formula In this embodiment, the environmental dynamic compensation formula is where, represents the fifth-class quality; represents the appearance compensation function; represents the original appearance score; represents the odor compensation function; represents the original odor signal intensity; represents the ingredient compensation function; represents the original ingredient content; represents the environmental coupling factor; E represents the set of environmental parameters; represents the weight coefficient of the appearance compensation function; represents the weight coefficient of the odor compensation function; represents the weight coefficient of the ingredient compensation function; represents the humidity impact coefficient; represents the light impact coefficient; H represents the real-time environmental humidity; represents the standard humidity; L represents the real-time light intensity; represents the standard light intensity; α represents the temperature attenuation coefficient; T represents the real-time temperature; Indicates standard temperature; β indicates airflow correction factor; V indicates real-time airflow velocity; represents the standard air flow velocity; γ represents the pressure sensitivity coefficient; P represents the real-time air pressure; Indicates standard atmospheric pressure; represents the global sensitivity coefficient of the i-th environmental parameter; Represents the real-time value of the i-th environmental parameter; Indicates the standard value of the i-th environmental parameter.

[0094] In this embodiment, the fifth quality is the final rating after environmental compensation. For example, in a compensation case, the original fourth quality is qualified, but because the humidity is 85% during the test, it is upgraded to excellent after compensation. In a maintenance case, the safety conflict cannot be compensated and remains unqualified.

[0095] The working principle and beneficial effects of the above technical solution are: comparing the three quality results, identifying numerical contradictions, safety threshold conflicts and authenticity differences, building a three-level arbitration logic framework to resolve conflicts, and determining the fourth quality; obtaining environmental data, combining the environment-quality impact library and the environmental dynamic compensation formula, performing environmental dynamic compensation on the fourth quality, and deriving the fifth quality, thereby improving the detection stability in complex environments and achieving full transparency of quality judgment. Example 7:

[0096] An embodiment of the present invention provides a method for detecting the quality of traditional Chinese medicine. The method marks an upstream planting end according to the fifth quality and pushes optimization suggestions upward. The method also identifies historical markings and the execution status of the optimization suggestions to obtain a current marking of the corresponding upstream planting end. The method combines the current marking with a three-dimensional compensation system to compensate and adjust the fifth quality to obtain the final quality. The method includes: The fifth quality is anchored to the planting batch ID through blockchain, and an NFT traceability certificate is generated to record the quality grade, key defect parameters and environmental compensation records of the corresponding planting batch ID, and the upstream planting end is marked according to the NFT traceability certificate; Directly push optimization suggestions based on the key defect parameters and environmental compensation records, obtain the execution status of the optimization suggestions by the upstream planting end, and determine the current mark based on the execution status and historical marks; Based on the current mark, compensation factors of three dimensions in a preset three-dimensional compensation system are adjusted, and compensation adjustment is performed on the fifth quality according to the compensation factors to obtain a final quality.

[0097] In this embodiment, the planting batch ID is a code that uniquely identifies the production batch of medicinal materials, such as 2023-CQ-GS-001, which contains information such as the place of origin, variety, and serial number. Through blockchain anchoring, it ensures that the data cannot be tampered with and is associated with the entire process of detection, including planting, processing, and detection. For example, certain metadata such as temperature, humidity, and operator information during detection are bound to a certain batch ID.

[0098] In this embodiment, the NFT traceability certificate is a digital certificate based on blockchain, which records the quality grade of the batch ID, such as the fifth quality rating, key defect parameters, mildew spot area of 0.2%, and environmental compensation records, with a humidity compensation of +5%. The NFT hash value is publicly available for verification by downstream manufacturers. For example, the Panax notoginseng NFT certificate shows top-grade quality and no detectable sulfur residue.

[0099] In this embodiment, the key defect parameters refer to the core issues that lead to downgrading, such as sulfur dioxide fumigation exceeding the standard by 0.03% and 3 insect holes. The type, value, and impact level of the defects need to be clearly recorded in the NFT. For example, the Astragalus membranaceus NFT indicates 2 main root fractures and is downgraded to qualified.

[0100] In this embodiment, the environmental compensation record stores the details of environmental interference and compensation during detection. For example, a temperature of 30°C causes a 12% attenuation of the volatile oil signal, and the rating is corrected after compensation. Example: For Lycium barbarum, the detection humidity of 80% triggers the compensation formula, and the original score of 85 → 78 after compensation.

[0101] In this embodiment, the current mark is a dynamic rating that combines historical marks and the latest execution situation, such as upgrading from grade B to grade A. Example: A certain grower's mark changes from risk observation to a trusted supplier due to implementing optimization suggestions.

[0102] In this embodiment, the preset three-dimensional compensation system includes planting techniques, such as organic fertilization +0.5 points; environmental control, such as the greenhouse temperature and humidity meeting the standards +0.3 points; historical reputation, consecutive top-grade batches +0.2 points.

[0103] In this embodiment, the compensation factor adjustment is based on the current mark to correct the three-dimensional weights. For example, the weight of the technology dimension changes from 0.4 → 0.5 because the farmer purchased new equipment; the weight of the reputation dimension changes from 0.3 → 0.2 due to fluctuations in a recent batch of detections.

[0104] The working principle and beneficial effects of the above technical solution are as follows: By anchoring the quality grade to the planting batch through blockchain, an NFT certificate containing environmental compensation parameters is generated. Based on defect analysis, intelligent optimization suggestions for planting are pushed and the execution feedback is tracked. Combining historical marks, the time-geography-technology three-dimensional compensation factors are dynamically adjusted to achieve precise correction of quality and full-chain optimization. Example 8:

[0105] An embodiment of the present invention provides a traditional Chinese medicine quality detection system, such asFigure 2 As shown in the figure, it includes: The first quality determination module: Scanning the traditional Chinese medicine to be detected, obtaining the surface characteristics of the traditional Chinese medicine to be detected, comparing the surface characteristics with the standard characteristics of the genuine region, and determining the first quality; The second and third quality determination modules: Using an electronic odor capture array to record the response curve of the volatile components of the traditional Chinese medicine to be detected, identifying characteristic peaks, determining the second quality, and performing component analysis on the traditional Chinese medicine to be detected based on the characteristic peaks, and then determining the third quality; The fourth and fifth quality determination modules: Arbitrating conflict data based on the first quality, the second quality, and the third quality to obtain the fourth quality, and performing environmental dynamic compensation based on the fourth quality to obtain the fifth quality; The final quality determination module: Marking the upstream planting end according to the fifth quality and pushing optimization suggestions upward. At the same time, identifying the implementation situation of the historical marks and the optimization suggestions to obtain the current marks of the corresponding upstream planting end, and compensating and adjusting the fifth quality in combination with the current marks and the three-dimensional compensation system to obtain the final quality.

[0106] The working principle and beneficial effects of the above technical solution are: Through multi-modal data fusion of the surface characteristics, volatile components, and component analysis of traditional Chinese medicine, using conflict data arbitration and environmental dynamic compensation, combined with blockchain traceability and a three-dimensional compensation system, the comprehensive detection and optimization of the quality of traditional Chinese medicine are realized, the accuracy and efficiency of traditional Chinese medicine quality detection are improved, data sharing and collaboration in the whole chain of traditional Chinese medicine planting, processing, and circulation are realized, and the sustainable development of the traditional Chinese medicine industry is promoted.

[0107] Finally, it should be noted that: The above embodiments are only used to illustrate the technical solutions of the present invention, rather than to limit them; Although the present invention has been described in detail with reference to the foregoing embodiments, those of ordinary skill in the art should understand that: They can still modify the technical solutions recorded in the foregoing embodiments, or perform equivalent replacements on some of the technical features; And these modifications or replacements do not make the essence of the corresponding technical solutions deviate from the spirit and scope of the technical solutions of the embodiments of the present invention.

Claims

1. A method for detecting the quality of traditional Chinese medicine, characterized in that, Including: Step 1: Scan the traditional Chinese medicine to be detected, obtain the surface features of the traditional Chinese medicine to be detected, compare the surface features with the authentic standard features, and determine the first quality; Step 2: Use an electronic odor capture array to record the response curve of the volatile components of the traditional Chinese medicine to be detected, identify the characteristic peaks, determine the second quality, perform component analysis on the traditional Chinese medicine to be detected based on the characteristic peaks, and further determine the third quality; Step 3: Arbitrate the conflict data based on the first quality, the second quality, and the third quality to obtain the fourth quality, and perform environmental dynamic compensation based on the fourth quality to obtain the fifth quality; Step 4: Mark the upstream planting end according to the fifth quality and push the optimization suggestions upward. At the same time, identify the execution situation of the historical marks and the optimization suggestions to obtain the current marks of the corresponding upstream planting end. Combine the current marks with the three-dimensional compensation system to compensate and adjust the fifth quality to obtain the final quality.

2. The method for detecting the quality of traditional Chinese medicine according to claim 1, wherein, Scanning the traditional Chinese medicine to be detected, obtaining the surface features of the traditional Chinese medicine to be detected, comparing the surface features with the authentic standard features, and determining the first quality, including: Scanning the traditional Chinese medicine to be detected with a hyperspectral camera to obtain the hyperspectral pixel coordinates of the traditional Chinese medicine to be detected, and scanning the traditional Chinese medicine to be detected with structured light to obtain the 3D point cloud coordinates of the traditional Chinese medicine to be detected; Perform spatio-temporal alignment on the hyperspectral pixel coordinates and the 3D point cloud coordinates through the SIFT algorithm, map the hyperspectral data obtained from the hyperspectral pixel coordinates after alignment to the 3D point cloud coordinates, perform spectral-geometric correlation analysis and defect feature quantification, and determine the surface features of the traditional Chinese medicine to be detected; Based on the comparison between the surface features and the authentic standard features, trigger the preset knowledge graph rules according to the comparison results to determine the abnormal results, retrieve and match the associated nodes corresponding to the abnormal results in the preset knowledge graph, and determine the first quality based on the associated nodes and the corresponding surface features.

3. The method for detecting the quality of traditional Chinese medicine according to claim 1, characterized in that Using an electronic odor capture array to record the response curve of the volatile components of the traditional Chinese medicine to be detected, identifying the characteristic peaks, and determining the second quality, including: Set a forked odor sampling head, combine the forked odor sampling head with a multimodal sensor to form an electronic odor capture array, and determine the scanning mode of the electronic odor capture array according to the surface features of the traditional Chinese medicine to be detected; Generate a time-response matrix according to the electronic odor capture array, and inhale pure air to reset the sensor baseline every preset time, thereby determining the response curve of the volatile components of the traditional Chinese medicine to be detected; Calculate the first derivative of the response curve, set a sliding window according to the scanning mode, determine the characteristic peaks of the response curve, determine the main peak intensity based on the characteristic peaks, use the sliding window centered on the main peak intensity as the post-peak interval to calculate the platform stability, and then obtain the peak platform ratio. Combine the main peak intensity, platform stability, and peak platform ratio to determine the discrimination conditions, and obtain the second quality of the traditional Chinese medicine to be detected based on the discrimination conditions.

4. The quality inspection method of traditional Chinese medicine according to claim 3, wherein, Set a forked odor sampling head, combine the forked odor sampling head with a multimodal sensor to form an electronic odor capture array, and determine the scanning mode of the electronic odor capture array according to the surface features of the traditional Chinese medicine to be detected, including: Adopt a dual-channel bifurcation structure. The dual-channel bifurcation structure adopts bilateral independent air extraction, and bilateral integrated independent micro air pumps can control different flow rates. Multimodal sensors of the same model are arranged at the end of each channel to form an electronic odor capture array; Divide the traditional Chinese medicine to be detected according to surface characteristics, determine the dynamic configuration of multimodal sensors according to the division results. At the same time, set corresponding capture algorithms for each column in the electronic odor capture array; Determine the scanning mode based on the dynamic configuration and capture algorithm.

5. The Chinese medicine quality detection method according to claim 2, characterized in that, Conduct component analysis on the traditional Chinese medicine to be detected based on the characteristic peaks, and then determine the third quality, including: Modify quantum dots with different fluorescence wavelengths on the surface of the sensors in the electronic odor capture array to form a fluorescence-electrochemical dual-signal channel. Based on the fluorescence-electrochemical dual-signal channel, detect the fluorescence intensity ratio of quantum dots and electrochemical signals, and combine time information to obtain three-dimensional data of the traditional Chinese medicine to be detected; Mark time anchor points for the three-dimensional data based on the characteristic peaks combined with a convolutional neural network, and match according to the marked results with the traditional Chinese medicine component database to generate a dynamic standard curve; Calculate the content of active ingredients in the traditional Chinese medicine to be detected based on the dynamic standard curve, and conduct semi-quantitative analysis on impurities to obtain the impurity component content. According to the active ingredient content and the impurity component content, obtain the third quality of the traditional Chinese medicine to be detected.

6. The method for detecting the quality of traditional Chinese medicine according to claim 1, characterized in that, Conduct arbitration on conflicting data based on the first quality, second quality, and third quality to obtain the fourth quality. Conduct environmental dynamic compensation based on the fourth quality to obtain the fifth quality, including: Compare the first quality, second quality, and third quality, identify numerical contradictions to determine the first conflict type, identify safety threshold conflicts to determine the second conflict type, and identify differences in authenticity to determine the third conflict type; Construct a three-level arbitration logic framework based on the first conflict type, second conflict type, and third conflict type, and determine the fourth quality according to the three-level arbitration logic framework; Obtain environmental data, determine the environmental-quality influence parameters for the traditional Chinese medicine to be detected from the environmental-quality influence library according to the environmental data and the corresponding surface characteristics of the traditional Chinese medicine to be detected. Set an environmental dynamic compensation formula according to the environmental-quality influence parameters, and combine the fourth quality to obtain the fifth quality.

7. A method for detecting the quality of traditional Chinese medicine according to claim 1, characterized in that, Mark the upstream planting end according to the fifth quality and push optimization suggestions upward. At the same time, identify the execution situation of historical marks and optimization suggestions to obtain the current mark of the corresponding upstream planting end. Combine the current mark with the three-dimensional compensation system to compensate and adjust the fifth quality to obtain the final quality, including: Anchor the fifth quality with the planting batch ID through the blockchain to generate an NFT traceability certificate, record the quality grade, key defect parameters, and environmental compensation records of the corresponding planting batch ID, and mark the upstream planting end according to the NFT traceability certificate; Directly push optimization suggestions according to the key defect parameters and environmental compensation records, and obtain the execution situation of the upstream planting end for the optimization suggestions. Determine the current mark according to the execution situation and historical marks; Adjust the compensation factors for the three dimensions in the preset three-dimensional compensation system based on the current markings, and perform compensation adjustment on the fifth quality according to the compensation factors to obtain the final quality.

8. A traditional Chinese medicine quality detection system, characterized in that, It includes: The first quality determination module: Scan the traditional Chinese medicine to be detected, obtain the surface characteristics of the traditional Chinese medicine to be detected, compare the surface characteristics with the genuine standard characteristics, and determine the first quality. The second and third quality determination modules: Use an electronic odor capture array to record the response curve of the volatile components of the traditional Chinese medicine to be detected, identify the characteristic peaks, determine the second quality, and perform component analysis on the traditional Chinese medicine to be detected based on the characteristic peaks, and then determine the third quality. The fourth and fifth quality determination modules: Perform conflict data arbitration based on the first quality, the second quality, and the third quality to obtain the fourth quality, and perform environmental dynamic compensation based on the fourth quality to obtain the fifth quality. The final quality determination module: Mark the upstream planting end according to the fifth quality, and push optimization suggestions upward. At the same time, identify the execution status of the historical markings and optimization suggestions to obtain the current markings of the corresponding upstream planting end, and combine the current markings with the three-dimensional compensation system to perform compensation adjustment on the fifth quality to obtain the final quality.

Images