A rapid detection system for traditional Chinese medicine based on an artificial intelligence model

Abstract

The present invention relates to the technical field of Chinese herbal medicine detection, and specifically discloses a rapid detection system for Chinese herbal medicine based on an artificial intelligence model. The system includes: a conveying module for conveying and transporting Chinese herbal medicine; an image acquisition module for acquiring image information of the Chinese herbal medicine on the transmission module; an odor acquisition module for acquiring odor information of the Chinese herbal medicine on the conveying module; a sorting module for sorting and removing and sorting and preparing the Chinese herbal medicine on the conveying module, and sorting and removing and sorting and returning the Chinese herbal medicine on the sorting and temporary storage module. The present invention rapidly screens Chinese herbal medicine according to the characteristic similarity coefficient of Chinese herbal medicine, and then further fully detects the Chinese herbal medicine with a medium similarity coefficient by a near-infrared spectrometer, so as to achieve the purpose of rapidly detecting Chinese herbal medicine, effectively reducing the detection amount of the near-infrared spectrometer detection, and also ensuring the accuracy of the detection of Chinese herbal medicine.

Description

Technical Field

[0001] The present invention relates to the technical field of traditional Chinese medicine detection, and particularly to a rapid detection system for traditional Chinese medicine based on an artificial intelligence model. Background Art

[0002] For a long time, most of the traditional Chinese medicine production areas in China have decentralized processing. After the raw medicinal materials are initially processed at the production areas, they are further processed by various regions and enterprises. Due to differences in production implementation conditions and the lack of standardized production in various places, especially the repeated occurrence of sulfur fumigation in some places, the quality of processed Chinese herbal pieces is uneven, and adulteration in the market is also common, which affects the clinical efficacy of traditional Chinese medicine. To ensure the quality of traditional Chinese medicine, many enterprises need to detect the quality of traditional Chinese medicine during the process of purchasing or processing traditional Chinese medicine.

[0003] The traditional detection methods for traditional Chinese medicine mostly detect the components of traditional Chinese medicine through a liquid chromatograph. Although this detection method can effectively detect traditional Chinese medicine, its detection process is complex, including crushing, drying, physical and chemical analysis of the medicinal materials and detection by a liquid chromatograph; the data generation cycle is long, which may take several days or even longer. In the application process, mostly only sampling detection of purchased traditional Chinese medicine can be carried out, and it is difficult to achieve the purpose of fully detecting the quality of traditional Chinese medicine. With the development of technology, there is also a method of detecting traditional Chinese medicine by a near-infrared spectrometer. The detection time of a single sample of traditional Chinese medicine by the near-infrared spectrometer can be completed within 1 minute. Compared with the liquid chromatograph, the detection efficiency is greatly improved (saving more than 99.9% of the time). Compared with the detection results of the liquid chromatograph, the correlation coefficient reaches more than 0.95, which can ensure the reliability of the detection results and effectively improve the detection efficiency of traditional Chinese medicine. However, when it is actually applied to the detection of traditional Chinese medicine in the process of traditional Chinese medicine processing, its detection speed still cannot meet the purpose of quickly and fully detecting traditional Chinese medicine. Summary of the Invention

[0004] The purpose of the present invention is to provide a rapid detection system for traditional Chinese medicine based on an artificial intelligence model, and solve the following technical problems:

[0005] How to achieve the purpose of rapid detection of traditional Chinese medicine based on artificial intelligence.

[0006] The purpose of the present invention can be achieved through the following technical solutions:

[0007] A rapid detection system for traditional Chinese medicine based on an artificial intelligence model, the system includes:

[0008] A conveying module for conveying and transporting traditional Chinese medicine;

[0009] An image acquisition module for acquiring image information of traditional Chinese medicine on the transmission module;

[0010] An odor collection module for collecting the odor information of traditional Chinese medicinal materials on the conveying module;

[0011] A sorting module for sorting and removing and sorting and preparing the traditional Chinese medicinal materials on the conveying module, and sorting and removing and sorting and returning the traditional Chinese medicinal materials on the sorting temporary storage module;

[0012] A sorting temporary storage module for temporarily storing the traditional Chinese medicinal materials to be detected by the near-infrared spectrometer and the remaining prepared medicinal materials after sorting and preparation;

[0013] A near-infrared spectrometer for performing infrared spectrum analysis on the prepared traditional Chinese medicinal materials to generate a similarity coefficient of the components of the traditional Chinese medicinal materials;

[0014] An analysis and control module for generating a signal for the sorting module to sort and prepare the traditional Chinese medicinal materials at a preset frequency; processing and analyzing the image information and odor information to obtain a similarity coefficient of the characteristics of the traditional Chinese medicinal materials, and generating a signal for the sorting module to sort and remove and sort and prepare the traditional Chinese medicinal materials; generating a signal for the sorting module to sort and remove and sort and return the traditional Chinese medicinal materials according to the similarity coefficient of the components of the traditional Chinese medicinal materials.

[0015] Further, the process of obtaining the similarity coefficient of the characteristics of the traditional Chinese medicinal materials includes:

[0016] S1. Performing preprocessing such as denoising, enhancing, and resizing on the image information to extract the color, texture, and shape features of the traditional Chinese medicinal materials;

[0017] S2. Performing preprocessing such as filtering and amplifying on the odor information to extract the odor features of the traditional Chinese medicinal materials;

[0018] S3. Generating a similarity coefficient of the characteristics of the traditional Chinese medicinal materials by comparing the color, texture, shape, and gas characteristics of the traditional Chinese medicinal materials with the data in the traditional Chinese medicinal materials database.

[0019] Further, in step S3, the process of generating the similarity coefficient of the characteristics of the traditional Chinese medicinal materials includes:

[0020]

[0021] Calculating and analyzing to obtain the similarity coefficient Q of the characteristics of the traditional Chinese medicinal materials through formulas (1)-(2) s ;

[0022] where E i is the i-th characteristic similarity evaluation parameter, E i,stdis the standard evaluation parameter for the i-th feature. When i = 1, E1 is the comprehensive similarity evaluation parameter for the texture features of traditional Chinese medicinal materials. When i = 2, E2 is the similarity evaluation parameter for the color features of traditional Chinese medicinal materials. When i = 3, E3 is the similarity evaluation parameter for the shape features of traditional Chinese medicinal materials. When i = 4, E4 is the similarity evaluation parameter for the odor features of traditional Chinese medicinal materials. When i = 5, E5 is the similarity evaluation parameter for other features of traditional Chinese medicinal materials, ω i is the influence coefficient of the i-th feature on the detection of traditional Chinese medicinal materials. R1 is the number of textures on the traditional Chinese medicinal materials whose texture similarity reaches the first preset coefficient, and R2 is the number of textures on the traditional Chinese medicinal materials whose texture similarity does not reach the second preset coefficient. is the weight coefficient corresponding to different numbers of textures.

[0023] Further, the process of the sorting and preparation module generating signals for sorting and removing and sorting and preparing traditional Chinese medicinal materials includes:

[0024] Compare the similarity coefficient Q of traditional Chinese medicinal materials s with the preset threshold range [Q1, Q2] of traditional Chinese medicinal material features;

[0025] If Q s > Q2, no sorting signal is generated, and the analysis and control module generates signals for the sorting and preparation module to sort and prepare traditional Chinese medicinal materials at a preset frequency;

[0026] If Q s ∈[Q1, Q2], the analysis and control module marks the detected traditional Chinese medicinal materials and generates corresponding sorting and preparation signals;

[0027] If Q s < Q1, the analysis and control module marks the detected traditional Chinese medicinal materials and generates corresponding sorting and removal signals.

[0028] Further, the process of generating the similarity coefficient of traditional Chinese medicinal material components includes:

[0029] Collect the spectrum of the prepared traditional Chinese medicinal materials through a near-infrared spectrometer to obtain the spectrum data of the corresponding traditional Chinese medicinal materials;

[0030] Preprocess the spectrum data and input it into the constructed artificial intelligence big data model for comparison with the chemical component content in the preset traditional Chinese medicinal material samples;

[0031] Calculate the similarity coefficient of the components of the corresponding traditional Chinese medicinal materials according to the comparison results.

[0032] Further, the process of the sorting and preparation module generating signals for sorting and removing and sorting and returning traditional Chinese medicinal materials includes:

[0033] Compare the similarity coefficient of traditional Chinese medicinal material components with the preset standard similarity coefficient of traditional Chinese medicinal material components;

[0034] If the similarity coefficient of traditional Chinese medicine components is greater than or equal to the preset similarity coefficient of traditional Chinese medicine material components, the analysis and control module marks the remaining traditional Chinese medicine materials after sorting and preparation and generates a corresponding sorting and homing signal;

[0035] If the similarity coefficient of traditional Chinese medicine components is less than the preset similarity coefficient of traditional Chinese medicine material components, the analysis and control module marks the remaining traditional Chinese medicine materials after sorting and preparation and generates a corresponding sorting and rejection signal.

[0036] Furthermore, the system further includes:

[0037] A statistical adjustment module that performs statistical analysis on the result data of the traditional Chinese medicine materials detected by the near-infrared spectrometer and adjusts the preset range of traditional Chinese medicine material characteristic thresholds; performs statistical analysis on the data of the traditional Chinese medicine materials sorted, prepared, and unsorted by the sorting module on the transmission module and adjusts the preset frequency.

[0038] Furthermore, the process of adjusting the preset range of traditional Chinese medicine material characteristic thresholds includes:

[0039] Perform statistical analysis on the number of signals generated after the near-infrared spectrometer detects a preset number of traditional Chinese medicine materials recently;

[0040]

[0041] Calculate and analyze to obtain the defective rate Z1 of the preset number of traditional Chinese medicine materials through formula (3);

[0042] Compare the defective rate Z1 of the preset number of traditional Chinese medicine materials with the preset defective rate threshold range [Z a , Z b ;

[0043] If Z1 > Z b , then update the left endpoint Q1 of the threshold range [Q1, Q2] to Q 1，re , Q 1，re = Q1 + ΔQ;

[0044] If Z1 < Z a , then update the left endpoint Q1 of the threshold range [Q1, Q2] to Q 1，re , Q 1，re = Q1 - ΔQ;

[0045] Among them, Q 1，re is the lowest threshold of the updated preset range of traditional Chinese medicine material characteristic thresholds, ΔQ is the adjustment coefficient of the traditional Chinese medicine material characteristic threshold range, A risk is the number of sorting and rejection signals generated by the near-infrared spectrometer, and A all is the total number of signals generated by the near-infrared spectrometer;

[0046] After the near-infrared spectrometer recently detects a preset number of Chinese herbal medicines, statistical analysis is performed on the number of generated signals and the corresponding number of signals generated after sorting and preparing a certain number of Chinese herbal medicines at a preset frequency;

[0047]

[0048] The abnormality rate Z2 of the preset number of Chinese herbal medicines is obtained through timely analysis by formula (4);

[0049] Compare the abnormality rate Z2 of the preset number of Chinese herbal medicines with the preset abnormality rate threshold range [Z c , Z d ;

[0050] If Z2 > Z d , then update the right endpoint Q2 of the threshold range [Q1, Q2] to Q 2，re , Q 2，re = Q2 + ΔQ;

[0051] If Z2 < Z c , then update the right endpoint Q2 of the threshold range [Q1, Q2] to Q 2，re , Q 2，re = Q2 - ΔQ;

[0052] Among them, Q 2，re is the highest threshold of the updated preset Chinese herbal medicine characteristic threshold range, A no is the number of Chinese herbal medicines sorted and removed among a certain number of Chinese herbal medicines sorted and prepared at a preset frequency, and τ no is the Chinese herbal medicine abnormality rate adjustment coefficient.

[0053] Furthermore, the process of adjusting the preset frequency includes:

[0054]

[0055] Calculate and analyze to obtain the sorting rate F of the sorting module through formula (5) s ;

[0056] Among them, D f is the number of Chinese herbal medicines sorted by the sorting module within a preset positive integer multiple of the sorting cycle, D all is the number of Chinese herbal medicines determined by the analysis and control module according to the Chinese herbal medicine characteristic similarity coefficient within a preset positive integer multiple of the sorting cycle, and the sorting cycle is the number of Chinese herbal medicines detected during the process of generating adjacent sorting and preparation signals at a preset frequency;

[0057] Compare the sorting rate F of the sorting module s with the preset sorting rate F std ;

[0058] If Fs ≥F std , then P s = P std *(1 + F s );

[0059] If F s < F std , then P s = P std ;

[0060] Among them, P s is the preset frequency, and P std is the base frequency.

[0061] Advantages of the present invention:

[0062] (1) The present invention uses the trained artificial intelligence model to quickly identify and analyze all traditional Chinese medicines, quickly screen the traditional Chinese medicines according to the similarity coefficient of the characteristics of the traditional Chinese medicines, and then further fully detect the traditional Chinese medicines with medium similarity coefficients by using a near-infrared spectrometer, so as to achieve the purpose of quickly detecting traditional Chinese medicines. Compared with the method of only detecting by a near-infrared spectrometer in the prior art, the detection amount of the near-infrared spectrometer is effectively reduced, the purpose of quickly detecting traditional Chinese medicines is achieved, and the accuracy of detecting traditional Chinese medicines is also ensured.

[0063] (2) The present invention intelligently adjusts the detection standard and sampling frequency in the traditional Chinese medicine detection process according to the feedback of the detection results during the detection of traditional Chinese medicines. This intelligent adjustment method can be adaptively adjusted for traditional Chinese medicines of different qualities, further improving the efficiency in the traditional Chinese medicine detection process and the accuracy of traditional Chinese medicine detection. Description of the drawings

[0064] The following further describes the present invention with reference to the accompanying drawings.

[0065] Figure 1 is a schematic block diagram of a rapid detection system for traditional Chinese medicines based on an artificial intelligence model proposed by the present invention;

[0066] Figure 2 is a flowchart of the steps of the process of obtaining the similarity coefficient of the characteristics of traditional Chinese medicines proposed by the present invention. Specific embodiments

[0067] The following will clearly and completely describe the technical solutions in the embodiments of the present invention with reference to the accompanying drawings in the embodiments of the present invention. Obviously, the described embodiments are only a part of the embodiments of the present invention, rather than all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those of ordinary skill in the art without creative efforts belong to the scope of protection of the present invention.

[0068] Please refer to Figure 1 - Figure 2 As shown, in one embodiment, a rapid detection system for traditional Chinese medicinal materials based on an artificial intelligence model is provided. The system includes:

[0069] A conveying module for transporting traditional Chinese medicinal materials.

[0070] An image acquisition module for acquiring image information of traditional Chinese medicinal materials on the transmission module.

[0071] An odor acquisition module for acquiring odor information of traditional Chinese medicinal materials on the conveying module.

[0072] A sorting module for sorting and removing traditional Chinese medicinal materials on the conveying module and sorting and preparing them, and sorting and removing and sorting and returning traditional Chinese medicinal materials on the sorting temporary storage module.

[0073] A sorting temporary storage module for temporarily storing traditional Chinese medicinal materials to be detected by a near-infrared spectrometer after sorting preparation and the remaining medicinal materials after preparation.

[0074] A near-infrared spectrometer for performing infrared spectral analysis on the prepared traditional Chinese medicinal materials to generate a similarity coefficient of the components of the traditional Chinese medicinal materials.

[0075] An analysis and control module that generates signals for the sorting module to sort and prepare traditional Chinese medicinal materials at a preset frequency; processes and analyzes the image information and odor information to obtain a similarity coefficient of the characteristics of the traditional Chinese medicinal materials, and generates signals for the sorting module to sort and remove and sort and prepare traditional Chinese medicinal materials; generates signals for the sorting module to sort and remove and sort and return traditional Chinese medicinal materials according to the similarity coefficient of the components of the traditional Chinese medicinal materials.

[0076] Through the above technical solution, this embodiment provides a rapid detection system for traditional Chinese medicines based on an artificial intelligence model. During the process of an enterprise detecting traditional Chinese medicines, the transmission module is used to convey and transport the traditional Chinese medicines. The conveying module can be a conveyor belt. On one side of the conveying module, an image acquisition module, an odor acquisition module, and a sorting module are sequentially installed. The image acquisition module can be a high-definition camera, which is used to take pictures of the traditional Chinese medicines during the conveying and transportation process and collect the image information of the traditional Chinese medicines. The odor acquisition module can be a traditional Chinese medicine odor analyzer, which is used to collect the odor information emitted by the traditional Chinese medicines during the conveying and transportation process. The sorting module can be a robotic arm and a preparation platform. The preparation platform is mainly used to take a small part of the traditional Chinese medicines for sampling and perform preparation processes such as grinding, mixing, and tablet pressing, so as to facilitate subsequent component detection of the samples by a near-infrared spectrometer. The preparation platform can be an automatic sample processor or can also be prepared manually. The robotic arm is used to pick up the traditional Chinese medicines on the conveying module and place them in the unqualified area for rejection, or can also pick up the traditional Chinese medicines on the conveying module and place them on the preparation platform for preparation, as well as sort and reject and sort and return the traditional Chinese medicines on the sorting and temporary storage module. During the detection process, the analysis and control module can process and analyze the image information and odor information based on the artificial intelligence model to obtain the characteristic similarity coefficient of the traditional Chinese medicines, and judge the possibility of each traditional Chinese medicine being the target traditional Chinese medicine according to the characteristic similarity coefficient of the traditional Chinese medicines. If the characteristic similarity coefficient of the detected traditional Chinese medicine is relatively high, then this traditional Chinese medicine is not sorted. If the characteristic similarity coefficient of the detected traditional Chinese medicine is medium, then a signal for sorting and preparing this traditional Chinese medicine is generated, and this traditional Chinese medicine is sorted and prepared through the sorting module. If the characteristic similarity coefficient of the detected traditional Chinese medicine is relatively low, then a signal for sorting and rejecting this traditional Chinese medicine is generated, and this traditional Chinese medicine is sorted and rejected through the sorting module. The samples after sorting and preparation and the remaining traditional Chinese medicines after preparation are all placed on the sorting and temporary storage module for sequential arrangement and placement, waiting for sequential analysis by the near-infrared spectrometer. Then, the near-infrared spectrometer performs infrared spectrum analysis on the prepared traditional Chinese medicine samples to generate the component similarity coefficient of the traditional Chinese medicines. Finally, the analysis and control module judges the possibility of each prepared traditional Chinese medicine being the target traditional Chinese medicine according to the component similarity coefficient of the traditional Chinese medicines. If the component similarity coefficient of the detected traditional Chinese medicine is relatively high, then a signal for sorting and returning this traditional Chinese medicine is generated, and the remaining part of this traditional Chinese medicine placed in the sorting and temporary storage module is picked up by the sorting module and put back onto the transmission module. If the component similarity coefficient of the detected traditional Chinese medicine is relatively low, then a signal for sorting and rejecting this traditional Chinese medicine is generated, and the remaining part of this traditional Chinese medicine placed in the sorting and temporary storage module is picked up by the sorting module and placed in the unqualified area for rejection. In this way, the process of rapidly detecting traditional Chinese medicines can be completed.

[0077] It should be noted that the training process of the artificial intelligence model for identifying traditional Chinese medicinal materials can be obtained through multiple steps such as data collection, data cleaning, data annotation, and model training. This is prior art and will not be elaborated here. To improve the accuracy of the detection of traditional Chinese medicinal materials, the analysis and control module will generate signals for the sorting module to sort and prepare traditional Chinese medicinal materials at a preset frequency, so as to achieve the purpose of randomly inspecting the traditional Chinese medicinal materials that do not need to be sorted on the conveying module. According to the results of the random inspection, the standard for generating sorting signals can be intelligently adjusted.

[0078] During the detection of traditional Chinese medicinal materials by this system, a database for identifying the characteristics of medicinal materials is established by using machine vision and odor collection methods. Through the completed database, a preliminary identification and analysis of all traditional Chinese medicinal materials is carried out. The identification features include but are not limited to the shape of the medicinal materials, odor, slice diameter size, thickness, color degree, etc. According to the identification features, the characteristic similarity coefficient of traditional Chinese medicinal materials is calculated and analyzed. According to the characteristic similarity coefficient of traditional Chinese medicinal materials, the traditional Chinese medicinal materials are quickly screened, and then the traditional Chinese medicinal materials with a characteristic similarity coefficient within a certain threshold are further fully detected by a near-infrared spectrometer, so as to achieve the purpose of quickly detecting traditional Chinese medicinal materials. Compared with the method of only detecting by a near-infrared spectrometer in the prior art, the detection amount of the near-infrared spectrometer is effectively reduced, the purpose of quickly detecting traditional Chinese medicinal materials is achieved, and the accuracy of the detection of traditional Chinese medicinal materials is ensured. The operation process is simple and standardized, and the operator does not need to have a professional background or receive special training; although the modeling process takes a long time, once it is completed, efficient and accurate detection can be achieved.

[0079] In one embodiment, the process of obtaining the characteristic similarity coefficient of traditional Chinese medicinal materials includes:

[0080] S1. Perform preprocessing such as denoising, enhancement, and resizing on the image information, and extract features such as the color, texture, and shape of traditional Chinese medicinal materials;

[0081] S2. Perform preprocessing such as filtering and amplification on the odor information, and extract the odor characteristics of traditional Chinese medicinal materials;

[0082] S3. By comparing the color, texture, shape, odor, and other characteristics of traditional Chinese medicinal materials with the data in the traditional Chinese medicinal material database, generate the characteristic similarity coefficient of traditional Chinese medicinal materials.

[0083] In step S3, the process of generating the characteristic similarity coefficient of traditional Chinese medicinal materials includes:

[0084]

[0085] The characteristic similarity coefficient Q of traditional Chinese medicinal materials is calculated and analyzed through formulas (1)-(2) s ;

[0086] where E i is the i-th characteristic similarity evaluation parameter, Ei,std is the standard evaluation parameter for the i-th feature, which can be preset for each feature data of the target Chinese medicinal material in the artificial intelligence model. When i = 1, E1 is the comprehensive similarity evaluation parameter of the texture feature of the Chinese medicinal material, and E 1，std is the standard evaluation parameter of the texture feature of the target Chinese medicinal material, which can be set according to multiple factors such as the texture distribution, shape, and type on the target Chinese medicinal material. When i = 2, E2 is the similarity evaluation parameter of the color feature of the Chinese medicinal material, which can be obtained by identifying and evaluating through the artificial intelligence model, and E 2,std is the standard evaluation parameter of the color feature of the target Chinese medicinal material, which can be preset according to multiple factors such as the color distribution, chromaticity, and corresponding wavelength of the Chinese medicinal material. For example, it can be set according to the wavelength corresponding to the color feature. If the color of the target Chinese medicinal material is red corresponding to a wavelength of 700 nanometers, the standard evaluation parameter of the color feature of the target Chinese medicinal material can be set to 700, and the similarity evaluation parameter of the color feature of the Chinese medicinal material is the detected wavelength value corresponding to the color feature minus the 700 wavelength value. When i = 3, E3 is the similarity evaluation parameter of the shape feature of the Chinese medicinal material, which can be obtained by identifying and evaluating through the artificial intelligence model, and E 3,std is the standard evaluation parameter of the shape feature of the target Chinese medicinal material, which can be set according to multiple factors such as the contour similarity, partial contour similarity, and shape similarity of the same part of the Chinese medicinal material. When i = 4, E4 is the similarity evaluation parameter of the odor feature of the Chinese medicinal material, which can be obtained by identifying and evaluating through the artificial intelligence model, and E 4,std is the standard evaluation parameter of the odor feature of the target Chinese medicinal material, which can be set according to multiple factors such as the volatile oil, odor sensation, and concentration of the target Chinese medicinal material. When i = 5, E5 is the similarity evaluation parameter of other features of the Chinese medicinal material, which can be obtained by identifying and evaluating through the artificial intelligence model, and E 5,std is the standard evaluation parameter of other features of the target Chinese medicinal material, which can be set according to multiple features such as the cross-section feature, luster, and special mark of the target Chinese medicinal material, ω i is the influence coefficient of the i-th feature on the detection of Chinese medicinal materials, which is obtained by setting according to experience. R1 is the number of textures on the Chinese medicinal material with a texture similarity reaching the first preset coefficient, and R2 is the number of textures on the Chinese medicinal material with a texture similarity not reaching the second preset coefficient. Both the first preset coefficient and the second preset coefficient are preset according to experience. For example, the first preset coefficient can be set to 80% similarity, and the second preset coefficient can be preset to 20% similarity. The texture similarity can be obtained by identifying and evaluating through the artificial intelligence model, is the weight coefficient corresponding to different numbers of textures, which is preset according to experience.

[0087] The process of the generation and sorting module for sorting and removing Chinese medicinal materials and generating signals for sorting and preparation includes:

[0088] Making the Chinese medicinal material feature similarity coefficient Q sCompare with the preset traditional Chinese medicine characteristic threshold range [Q1, Q2], and the preset traditional Chinese medicine characteristic threshold range [Q1, Q2] can be initially preset according to experience;

[0089] If Q s > Q2, it indicates that the corresponding traditional Chinese medicine is a qualified traditional Chinese medicine, and no sorting signal is generated. The analysis and control module generates a signal for the sorting module to sort and prepare the traditional Chinese medicine at a preset frequency, and controls the sorting module to conduct spot checks on the qualified traditional Chinese medicine at a preset frequency;

[0090] If Q s ∈ [Q1, Q2], it indicates that the corresponding traditional Chinese medicine is a traditional Chinese medicine that needs further detection. The analysis and control module marks the detected traditional Chinese medicine and generates a corresponding sorting and preparation signal;

[0091] If Q s < Q1, it indicates that the corresponding traditional Chinese medicine is an unqualified traditional Chinese medicine. The analysis and control module marks the detected traditional Chinese medicine and generates a corresponding sorting and rejection signal.

[0092] The process of generating the similarity coefficient of traditional Chinese medicine components includes:

[0093] Collect the spectrum of the prepared traditional Chinese medicine through a near-infrared spectrometer to obtain the spectrum data of the corresponding traditional Chinese medicine;

[0094] Preprocess the spectrum data and input it into the constructed artificial intelligence big data model for comparison with the chemical component content in the preset traditional Chinese medicine sample;

[0095] Calculate the similarity coefficient of the corresponding traditional Chinese medicine components according to the comparison results. The similarity coefficient of traditional Chinese medicine components can be calculated by various methods, such as Euclidean distance, cosine similarity, Pearson correlation coefficient, etc. According to the generated similarity coefficient of traditional Chinese medicine components, analyze the component similarity between the detected traditional Chinese medicine and the traditional Chinese medicine sample. The higher the similarity coefficient, the more similar the components between the detected traditional Chinese medicine and the traditional Chinese medicine sample; the lower the similarity coefficient, the greater the component difference between the detected traditional Chinese medicine and the traditional Chinese medicine sample.

[0096] The process of generating signals for the sorting module to sort and reject and sort and return the traditional Chinese medicine includes:

[0097] Compare the similarity coefficient of traditional Chinese medicine components with the preset standard similarity coefficient of traditional Chinese medicine components, and the preset standard similarity coefficient of traditional Chinese medicine components can be preset according to the enterprise's traditional Chinese medicine demand standard;

[0098] If the similarity coefficient of traditional Chinese medicine components is greater than or equal to the preset standard similarity coefficient of traditional Chinese medicine components, it indicates that the corresponding traditional Chinese medicine is a qualified traditional Chinese medicine. The analysis and control module marks the remaining traditional Chinese medicine after sorting and preparation and generates a corresponding sorting and return signal;

[0099] If the similarity coefficient of traditional Chinese medicine components is less than the preset standard similarity coefficient of traditional Chinese medicine components, it indicates that the corresponding traditional Chinese medicine is unqualified. The analysis and control module marks the remaining traditional Chinese medicine after sorting and preparation and generates a corresponding sorting and rejection signal.

[0100] Through the above technical solution, this embodiment provides a method for fully detecting and identifying traditional Chinese medicine based on an artificial intelligence model. First, the characteristics of traditional Chinese medicine are identified, analyzed and processed by the artificial intelligence model to obtain the characteristic similarity coefficient of traditional Chinese medicine. According to the characteristic similarity coefficient of traditional Chinese medicine, three classification methods of sorting and rejection, sorting and preparation, and non-sorting are carried out for traditional Chinese medicine. The traditional Chinese medicine judged as sorting and rejection is unqualified, and the traditional Chinese medicine judged as non-sorting is qualified. The components of the traditional Chinese medicine after sorting and preparation are further analyzed by a near-infrared spectrometer to generate the corresponding similarity coefficient of traditional Chinese medicine components. According to the similarity coefficient of traditional Chinese medicine components, the traditional Chinese medicine after sorting and preparation is further sorted and returned and sorted and rejected. The traditional Chinese medicine judged as sorting and rejection is unqualified, and the traditional Chinese medicine judged as sorting and returned is qualified. In this way, the artificial intelligence model and the near-infrared spectrometer can cooperate with each other to quickly and fully detect traditional Chinese medicine, which not only improves the detection speed of traditional Chinese medicine, but also ensures the accuracy in the detection process of traditional Chinese medicine.

[0101] The application of this traditional Chinese medicine detection system can effectively reduce the labor cost, reducing the traditional Chinese medicine detection that originally required at least 5 to 10 people to participate to only 1 person to operate, and the labor cost is reduced by more than 80%; the detection cycle is greatly shortened, shortening the traditional detection cycle from dozens of hours to several days to minutes, and the detection efficiency is increased by more than 99.9%; the sample volume coverage is comprehensive, and more than 1000 common traditional Chinese medicines can be effectively detected, covering various types such as rhizome types and fruit types; the model adaptability is high, and the established big data model has an accuracy rate of more than 90% in the detection of traditional Chinese medicines from different origins and batches. In practical applications, it effectively helps enterprises improve their product quality control capabilities and production efficiency. At the same time, the detection results are not interfered by human factors and have high repeatability, providing a new method and new way for the detection of the traditional Chinese medicine industry that is efficient, accurate and stable; this invention is expected to produce multiple benefits in the field of traditional Chinese medicine decoction pieces detection; economically, it significantly reduces the detection cost and improves the enterprise benefits; socially, it alleviates the shortage of skilled inspectors, improves the quality of traditional Chinese medicine decoction pieces, ensures the safety of public medication, and enhances the industry reputation; ecologically, efficient detection can promote the standardized development of the traditional Chinese medicine industry, rationally utilize traditional Chinese medicine resources, reduce waste, and promote the modernization process of the traditional Chinese medicine industry.

[0102] Furthermore, the system further includes:

[0103] The statistical adjustment module performs statistical analysis on the result data of the Chinese medicinal materials detected by the near-infrared spectrometer, and adjusts the preset range of Chinese medicinal material characteristic thresholds; it also performs statistical analysis on the data of the Chinese medicinal materials sorted, removed, prepared, and unsorted by the sorting module on the transmission module, and adjusts the preset frequency.

[0104] The process of adjusting the preset range of Chinese medicinal material characteristic thresholds includes:

[0105] After the near-infrared spectrometer detects a preset number of Chinese medicinal materials recently (the preset number can be preset according to experience), statistical analysis is performed on the number of generated signals;

[0106]

[0107] The defective rate Z1 of the preset number of Chinese medicinal materials is obtained through calculation and analysis by formula (3);

[0108] The defective rate Z1 of the preset number of Chinese medicinal materials is compared with the preset defective rate threshold range [Z a , Z b , and the preset defective rate threshold range [Z a , Z b is preset according to experience;

[0109] If Z1 > Z b , it indicates that the defective rate of the Chinese medicinal materials detected by the near-infrared spectrometer is too high, and the detection volume by the near-infrared spectrometer is large, affecting the detection efficiency. It is necessary to increase the minimum threshold of the unqualified Chinese medicinal materials determined according to the Chinese medicinal material characteristic similarity coefficient in the analysis and control module. Then, the left endpoint Q1 of the threshold range [Q1, Q2] is updated to Q 1，re , Q 1，re = Q1 + ΔQ;

[0110] If Z1 < Z a , it indicates that the defective rate of the Chinese medicinal materials detected by the near-infrared spectrometer is relatively low. There may be some qualified Chinese medicinal materials among the unqualified Chinese medicinal materials determined by the analysis and control module according to the Chinese medicinal material characteristic similarity coefficient. It is necessary to reduce the minimum threshold of the unqualified Chinese medicinal materials determined by the analysis and control module. Then, the left endpoint Q1 of the threshold range [Q1, Q2] is updated to Q 1，re , Q 1，re = Q1 - ΔQ;

[0111] Among them, Q 1，re is the minimum threshold of the updated preset range of Chinese medicinal material characteristic thresholds, ΔQ is the adjustment coefficient of the Chinese medicinal material characteristic threshold range, which is preset according to experience, ΔQ << Q2 - Q1, A risk is the number of sorting and removing signals generated by the near-infrared spectrometer, which can be obtained by statistically analyzing the number of sorting and removing signals generated by the near-infrared spectrometer, A allThe total number of signals generated for the near-infrared spectrometer is obtained by counting the number of signals generated by the near-infrared spectrometer;

[0112] After the near-infrared spectrometer recently detects a preset number of traditional Chinese medicines, the number of generated signals and the corresponding number of sorting and rejection signals generated after sorting a certain number of traditional Chinese medicines at a preset frequency are statistically analyzed;

[0113]

[0114] The abnormality rate Z2 of the preset number of traditional Chinese medicines is obtained by timely analysis through formula (4);

[0115] Compare the abnormality rate Z2 of the preset number of traditional Chinese medicines with the preset abnormality rate threshold range [Z c , Z d , and the preset abnormality rate threshold range [Z c , Z d is obtained by preset according to experience;

[0116] If Z2 > Z d , it indicates that the defective rate of the traditional Chinese medicines detected by the near-infrared spectrometer is too high or the probability of unqualified traditional Chinese medicines in the sampled traditional Chinese medicines is too high. Among the traditional Chinese medicines determined to be qualified according to the similarity coefficient of traditional Chinese medicine characteristics in the analysis and control module, there may be some unqualified traditional Chinese medicines. It is necessary to increase the highest threshold of the traditional Chinese medicines determined to be qualified according to the similarity coefficient of traditional Chinese medicine characteristics in the analysis and control module. Then, the right endpoint Q2 of the threshold range [Q1, Q2] is updated to Q 2.re , Q 2，re = Q2 + ΔQ;

[0117] If Z2 < Z c , it indicates that the defective rate of the traditional Chinese medicines detected by the near-infrared spectrometer is relatively low or the probability of unqualified traditional Chinese medicines in the sampled traditional Chinese medicines is relatively low. The analysis and control module determines that there are more qualified traditional Chinese medicines among the traditional Chinese medicines to be further detected according to the similarity coefficient of traditional Chinese medicine characteristics, and the qualification rate of the traditional Chinese medicines determined to be qualified according to the similarity coefficient of traditional Chinese medicine characteristics meets the standard. The detection volume by the near-infrared spectrometer is relatively large, which affects the detection efficiency. It is necessary to reduce the highest threshold of the traditional Chinese medicines determined to be qualified according to the similarity coefficient of traditional Chinese medicine characteristics in the analysis and control module. Then, the right endpoint Q2 of the threshold range [Q1, Q2] is updated to Q 2，re , Q 2，re = Q2 - ΔQ;

[0118] Among them, Q 2，re is the highest threshold of the updated preset traditional Chinese medicine characteristic threshold range, A noTo prepare the quantity of Chinese medicinal materials to be sorted out from a certain quantity of Chinese medicinal materials at a preset frequency, it is obtained by counting the quantity of Chinese medicinal materials to be sorted out in the randomly selected Chinese medicinal materials analyzed by a near-infrared spectrometer. The certain quantity of Chinese medicinal materials is the preset quantity of recently randomly selected Chinese medicinal materials, which is preset according to experience. For example, the certain quantity is preset to 100, A no That is, it is the quantity of unqualified Chinese medicinal materials detected in the recently 100 randomly selected Chinese medicinal materials detected by the near-infrared spectrometer, τ no is the adjustment coefficient of the abnormality rate of Chinese medicinal materials, which is preset according to experience.

[0119] The process of adjusting the preset frequency includes:

[0120]

[0121] The sorting rate F of the sorting module is obtained by calculation and analysis through formula (5) s ;

[0122] where D f is the quantity of Chinese medicinal materials sorted by the sorting module within the sorting cycle of a preset positive integer multiple. The quantity of sorted Chinese medicinal materials includes the quantity of Chinese medicinal materials sorted out and prepared, and can be obtained by the statistical adjustment module counting the sorting signals generated by the analysis and control module according to the similarity coefficient of Chinese medicinal material characteristics. D all is the quantity of Chinese medicinal materials determined by the analysis and control module according to the similarity coefficient of Chinese medicinal material characteristics within the sorting cycle of a preset positive integer multiple. The sorting cycle is the quantity of Chinese medicinal materials detected during the process of generating adjacent sorting and preparation signals at the preset frequency. The preset positive integer is preset according to experience. For example, if the preset positive integer is 1 and the preset frequency is to prepare one Chinese medicinal material from 100 randomly selected Chinese medicinal materials, these 100 Chinese medicinal materials are the Chinese medicinal materials determined to be qualified by the analysis and control module according to the similarity coefficient of Chinese medicinal material characteristics. During the process of the analysis and control module judging 100 Chinese medicinal materials, 10 unqualified Chinese medicinal materials and 20 Chinese medicinal materials that need further detection may be determined. So at this time, D f is 30, D all is 130, D f and D all can both be obtained by the statistical adjustment module;

[0123] Compare the sorting rate F s of the sorting module with the preset sorting rate F std . The preset sorting rate F std can be preset according to experience;

[0124] If F s ≥F std , then P s =P std *(1 + F s));

[0125] If F s < F std , then P s = P std ;

[0126] wherein, P s is a preset frequency, and P std is a base frequency, which can be obtained by presetting according to experience.

[0127] Through the above technical solution, this embodiment provides a method for intelligently adjusting the preset traditional Chinese medicine characteristic threshold range and the preset frequency. Through the above analysis and adjustment process, the detection standard and sampling frequency in the traditional Chinese medicine detection process can be intelligently adjusted according to the feedback of the detection results during the traditional Chinese medicine detection process. This intelligent adjustment method can perform adaptive adjustment for traditional Chinese medicines of different qualities, further improving the efficiency and accuracy of the traditional Chinese medicine detection process.

[0128] The above has described in detail an embodiment of the present invention, but the content described is only a preferred embodiment of the present invention and cannot be considered as limiting the implementation scope of the present invention. All equal changes and improvements made according to the scope of the present invention application shall still fall within the scope covered by the patent of the present invention.

Claims

1. A rapid detection system for Chinese herbal medicine based on an artificial intelligence model, characterized in that: The system comprises: A conveying module, used for conveying and transporting Chinese medicinal materials; An image acquisition module, used for acquiring image information of the Chinese medicinal materials on the transmission module; An odor collection module is used to collect odor information of the Chinese medicinal materials on the transmission module; The sorting module is used to sort, remove and prepare the Chinese medicinal materials on the conveying module, and to sort, remove and return the Chinese medicinal materials on the sorting temporary storage module; A sorting and temporary storage module is used to temporarily store the sorted and prepared Chinese medicinal materials to be tested by near-infrared spectrometer and the remaining medicinal materials; Near infrared spectrometer, used to perform infrared spectrum analysis on prepared Chinese medicinal materials and generate similarity coefficients of Chinese medicinal materials components; The analysis control module generates a signal for the sorting module to sort and prepare the Chinese medicinal materials at a preset frequency; processes and analyzes the image information and the odor information to obtain the similarity coefficient of the Chinese medicinal materials characteristics, and generates a signal for the sorting module to sort, reject and prepare the Chinese medicinal materials; generates a signal for the sorting module to sort, reject and sort and return the Chinese medicinal materials according to the similarity coefficient of the Chinese medicinal materials components; The process of generating a signal for the sorting module to sort, reject and prepare the Chinese medicinal materials includes: The similarity coefficient of Chinese medicinal materials characteristics Q s Compare with the preset threshold range of Chinese herbal medicine characteristics [Q1, Q2; If Q s >Q2, no sorting signal is generated, and the analysis and control module generates a signal for the sorting module to sort and prepare the Chinese medicinal materials at a preset frequency; If Q s ∈[Q1,Q2], the analysis and control module marks the detected Chinese medicinal materials and generates corresponding sorting and preparation signals; If Q s <Q1, the analysis and control module marks the detected Chinese medicinal materials and generates corresponding sorting and rejection signals; The statistical adjustment module performs statistical analysis on the result data of Chinese medicinal materials detected by the near-infrared spectrometer and adjusts the preset threshold range of Chinese medicinal materials characteristics; the sorting module performs statistical analysis on the data of sorting, rejection, sorting preparation and unsorted Chinese medicinal materials on the transmission module and adjusts the preset frequency.

2. A rapid detection system for Chinese medicinal materials based on an artificial intelligence model according to claim 1, characterized in that: The process of obtaining the similarity coefficient of Chinese medicinal material characteristics includes: S1, pre-process the image information by denoising, enhancing, and resizing, and extract the color, texture, and shape features of Chinese medicinal materials; S2, filtering and amplifying the odor information to extract the odor characteristics of Chinese medicinal materials; S3. Generate a similarity coefficient of Chinese medicinal material characteristics by comparing the color, texture, shape and gas characteristics of Chinese medicinal materials with the data in the Chinese medicinal material database.

3. A rapid detection system for Chinese medicinal materials based on an artificial intelligence model according to claim 2, characterized in that: In step S3, the process of generating the similarity coefficient of Chinese medicinal material characteristics includes: E1=R1*φ1-exp(R2)*φ2 (2) The similarity coefficient Q of Chinese medicinal materials characteristics is obtained by calculation and analysis through formula (1)-(2) s ; Among them, E i is the similarity evaluation parameter of the i-th feature, E i,std is the i-th feature standard evaluation parameter. When i=1, E1 is the comprehensive similarity evaluation parameter of the texture feature of Chinese medicinal materials. When i=2, E2 is the similarity evaluation parameter of the color feature of Chinese medicinal materials. When i=3, E3 is the similarity evaluation parameter of the shape feature of Chinese medicinal materials. When i=4, E4 is the similarity evaluation parameter of the smell feature of Chinese medicinal materials. When i=5, E5 is the similarity evaluation parameter of other features of Chinese medicinal materials. ω i is the influence coefficient of the i-th feature on the detection of Chinese medicinal materials, R1 is the number of textures on the Chinese medicinal materials whose texture similarity reaches the first preset coefficient, R2 is the number of textures on the Chinese medicinal materials whose texture similarity does not reach the second preset coefficient, φ1 and φ2 are the weight coefficients corresponding to different texture numbers.

4. A rapid detection system for Chinese medicinal materials based on an artificial intelligence model according to claim 3, characterized in that: The process of generating the similarity coefficient of Chinese medicinal material components includes: The spectrum of the prepared Chinese medicinal materials is collected by a near-infrared spectrometer to obtain the spectrum data of the corresponding Chinese medicinal materials; Preprocess the spectral data and input it into the constructed artificial intelligence big data model to compare it with the chemical component content in the preset Chinese herbal medicine samples; The similarity coefficient of the Chinese medicinal materials components of the corresponding Chinese medicinal materials is calculated based on the comparison results.

5. A rapid detection system for Chinese medicinal materials based on an artificial intelligence model according to claim 4, characterized in that: The process of generating a signal for the sorting module to sort and reject the Chinese medicinal materials and sort and return them to their original positions includes: Compare the similarity coefficient of Chinese medicinal materials ingredients with the preset standard similarity coefficient of Chinese medicinal materials ingredients; If the similarity coefficient of the Chinese medicinal material components is greater than or equal to the preset similarity coefficient of the Chinese medicinal material components standard, the analysis and control module marks the remaining Chinese medicinal materials after sorting and preparation and generates a corresponding sorting homing signal; If the similarity coefficient of the Chinese medicinal material ingredients is less than the preset standard similarity coefficient of the Chinese medicinal material ingredients, the analysis and control module marks the remaining Chinese medicinal materials after sorting and preparation and generates a corresponding sorting and rejection signal.

Images