Intelligent control method and system of medicine winnowing machine

Abstract

The application discloses a kind of medicine material winnowing machine intelligent control method and system, the method includes obtaining the mass distribution data of medicinal material, density characteristic value, profile data and texture feature, obtain medicinal material characteristic data set;According to characteristic data set, classification identification medicinal material, generate physical characteristic label, obtain classified medicinal material;According to classified medicinal material, query preset medicinal material winnowing parameter mapping database, obtain the equipment control parameter scheme required by winnowing;According to classified medicinal material, adjust fan speed, obtain the fan speed matched with medicinal material;According to density characteristic value, adjust air duct angle and airflow distribution plate, obtain actual air duct angle;Monitoring medicinal material separation effect, if it is detected that medicinal material separation trajectory deviates from preset range, then correct air duct angle and fan speed, update optimal air duct angle and fan speed to winnowing parameter mapping database, obtain final equipment control parameter scheme.The method can accurately identify medicinal material characteristics and dynamically match winnowing parameters.

Description

Technical Field

[0001] This invention relates to the field of intelligent control technology for medicinal herb processing equipment, and in particular to an intelligent control method and system for a medicinal herb air separator. Background Technology

[0002] Currently, medicinal herb processing is a crucial foundational link in the traditional Chinese medicine (TCM) industry, directly impacting the quality, efficacy, and clinical safety of TCM products. Air separation technology, a key process for separating and purifying medicinal materials, utilizes airflow to exert differentiated forces on materials of varying masses and densities. This enables the separation of medicinal materials from impurities and the grading and screening of high-quality herbs. It plays an indispensable role in the pre-processing of TCM herbs and is widely applied to the processing of various medicinal materials, including roots, leaves, and flowers.

[0003] Currently, most existing air separation equipment relies on preset parameters based on human experience, or only supports limited adjustments to fan speed and duct angle. For example, operators roughly set the airflow based on the type of medicinal material, but the same setting is insufficient to handle density fluctuations, shape differences, and surface characteristic variations between batches of medicinal materials. Furthermore, some equipment attempts to introduce weight detection for preliminary classification, such as dividing medicinal materials into light, medium, and heavy categories based on weight and matching corresponding air separation parameters. However, weight is only one factor influencing air separation behavior; the surface texture and shape of the medicinal materials also directly affect their aerodynamic characteristics. Air separation parameters obtained solely based on weight cannot distinguish between rough, easily adhered light impurities and smooth, high-quality light medicinal materials, nor can they handle irregularly shaped, high-quality heavy medicinal materials that easily cause airflow disturbances.

[0004] In summary, existing technologies lack the ability to accurately perceive the multi-dimensional physical properties of medicinal materials (especially roughness and uniformity), resulting in the inability of the set air separation parameters to be adapted to high-quality processing scenarios for medicinal materials. Summary of the Invention

[0005] This invention provides an intelligent control method and system for a medicinal herb air separator, addressing the problem of inaccurate identification of medicinal herb characteristics and dynamic matching of air separator parameters.

[0006] Firstly, in order to solve the above-mentioned technical problems, the present invention provides an intelligent control method for a medicinal herb air separator, comprising: Acquire the mass distribution data and density feature values ​​of medicinal materials, scan in real time to obtain the outline data and texture features of medicinal materials, analyze the mass distribution data and density feature values ​​to obtain a medicinal material characteristic dataset; Based on the medicinal material characteristic dataset, the medicinal materials are classified and identified to obtain the classified medicinal materials, and corresponding physical characteristic labels are generated; wherein, the physical characteristic labels include uniformity and roughness; Based on the classified medicinal materials and the physical property labels, the pre-established medicinal material air separation parameter mapping database is queried to obtain the equipment control parameter scheme required for medicinal material air separation; Based on the classified medicinal materials, the fan speed is dynamically adjusted to obtain a fan speed that matches the category of medicinal materials; Based on the medicinal material characteristic dataset, the duct angle and the position of the airflow distribution plate are adjusted using a servo motor to obtain the actual duct angle. The separation trajectory and separation effect data of medicinal materials are collected in real time. If the separation trajectory of medicinal materials deviates from the preset range, the actual air duct angle and the actual fan speed are corrected to obtain the optimal air duct angle and the optimal fan speed. The optimal air duct angle and the optimal fan speed are then updated to the pre-established medicinal material air separation parameter mapping database to obtain the final equipment control parameter scheme.

[0007] Preferably, the step of acquiring the mass distribution data and density feature values ​​of the medicinal materials, scanning the medicinal materials in real time for contour extraction and texture analysis, and analyzing the mass distribution data and density feature values ​​to obtain a medicinal material characteristic dataset includes: The first dataset is obtained by acquiring the mass distribution data and density characteristic values ​​of medicinal materials. The second dataset is obtained by scanning the medicinal materials in real time to extract contours and perform texture analysis. By merging the first data set and the second data set, and integrating multi-dimensional physical parameters, a third data set is generated; Compare the parameters in the third data set with a preset parameter threshold. If the parameter values ​​in the third data set meet the preset parameter threshold, then record the third data set to obtain the medicinal material characteristic dataset. If the parameter values ​​in the third data set do not meet the preset parameter threshold, the medicinal materials are re-collected, the first data set and the second data set are regenerated, and then merged back into the third data set. If the third data set still does not meet the preset parameter threshold after multiple re-collections, the medicinal materials are determined to be temporarily unsuitable for the air separation process.

[0008] Preferably, the step of classifying and identifying medicinal materials based on the medicinal material characteristic dataset to obtain classified medicinal materials and generating corresponding physical characteristic labels includes: Extract the acquired density feature values ​​from the medicinal material characteristic dataset. The density feature values ​​are density parameters directly obtained through the detection device. The density feature values ​​are quantized, and the mean and standard deviation of the density feature values ​​are calculated to obtain the quantized density feature values. If the quantified density characteristic value is less than the preset lightness threshold, the medicinal material is initially classified as a light medicinal material category. If the quantified density feature value is greater than the preset heavy threshold, the medicinal material is initially classified as a heavy medicinal material category. If the quantified density feature value is between the preset light threshold and the preset heavy threshold, the medicinal material is initially classified as a medium-quality medicinal material, and the classified medicinal material is obtained. By combining machine learning models, multidimensional features are extracted from the classified medicinal materials to obtain corresponding physical property labels.

[0009] Preferably, the step of querying a pre-established medicinal herb air-sorting parameter mapping database based on the classified medicinal materials and the physical property labels to obtain the equipment control parameter scheme required for medicinal herb air-sorting includes: Based on the classified medicinal materials and the physical property labels, a pre-established medicinal material air separation parameter mapping database is queried to obtain a preliminary set of parameters for the fan speed and duct angle corresponding to the classified medicinal materials and the physical property labels. Based on the classified medicinal materials and the preliminary parameter set, the range of fan speed parameters and the range of air duct angle are determined, and the range of fan speed parameters and the range of air duct angle are combined to obtain the equipment control parameter scheme required for the air separation of medicinal materials.

[0010] Preferably, the step of dynamically adjusting the fan speed based on the medicinal material classification results and the real-time monitored airflow intensity to match the fan speed with the medicinal material category, thereby obtaining the actual fan speed, includes: Based on the classified medicinal materials, query the pre-established medicinal material air separation parameter mapping database to obtain the target rotation speed range and air separation airflow intensity parameters corresponding to the medicinal material classification results; Based on the classified medicinal materials, the airflow intensity parameters of the air sorting are dynamically adjusted to obtain the baseline airflow intensity; The matching degree between the actual airflow intensity and the benchmark airflow intensity during the air separation process is monitored in real time, and the fan speed is dynamically adjusted to correct the deviation, so as to obtain the actual fan speed that is accurately matched with the type of medicinal material.

[0011] Preferably, the step of adjusting the air duct angle and the position of the airflow distribution plate using a servo motor based on the medicinal material characteristic dataset to obtain the actual air duct angle includes: The density feature value is obtained from the medicinal material characteristic dataset, and the air duct tilt angle is calculated based on the density feature value to obtain the optimal air duct tilt angle. If the optimal duct tilt angle exceeds the preset duct tilt angle range, the duct is adjusted to the preset duct tilt angle range by a servo motor to obtain the adjusted duct angle. Based on the adjusted air duct angle, the position of the airflow distribution plate is adjusted by the servo motor to make the airflow form a flow field distribution pattern that conforms to the characteristics of the medicinal materials, thus obtaining the actual air duct angle.

[0012] Preferably, the real-time acquisition of the separation trajectory and separation effect data of the medicinal materials, if the separation trajectory deviates from a preset range, corrects the actual air duct angle and the actual fan speed to obtain the optimal air duct angle and the optimal fan speed. The optimal air duct angle and the optimal fan speed are then updated to the pre-established medicinal material air separation parameter mapping database to obtain the final equipment control parameter scheme, including: Real-time acquisition of separation trajectory data and separation effect data of medicinal materials, extraction of the position coordinates of medicinal materials, and obtaining the actual separation trajectory; The actual separation trajectory is compared with the preset trajectory range, and the trajectory deviation value is calculated; If the trajectory deviation value exceeds the preset trajectory deviation threshold, a feedback adjustment signal is generated to calculate the adjustment amount of the duct angle and the fan speed. The actual duct angle and the actual fan speed are corrected based on the adjustment amount to obtain the optimal duct angle and the optimal fan speed. The optimal parameter combination is obtained based on the optimal duct angle and the optimal fan speed. The optimal parameter combination is then updated to the pre-established medicinal herb air separation parameter mapping database to obtain the final equipment control parameter scheme.

[0013] Secondly, the present invention provides an intelligent control system for a medicinal herb air separator, comprising: The data acquisition module acquires the mass distribution data and density characteristic values ​​of medicinal materials, scans in real time to acquire the outline data and texture features of medicinal materials, and analyzes the mass distribution data and density characteristic values ​​to obtain a medicinal material characteristic dataset. The classification and recognition module classifies and recognizes the medicinal materials based on the medicinal material characteristic dataset, obtains the classified medicinal materials, and generates corresponding physical characteristic labels; wherein, the physical characteristic labels include uniformity and roughness; The parameter matching module queries a pre-established medicinal herb air separation parameter mapping database based on the classified medicinal herbs and the physical property labels to obtain the equipment control parameter scheme required for medicinal herb air separation. The speed control module dynamically adjusts the fan speed according to the classified medicinal materials to obtain a fan speed that matches the category of medicinal materials; The flow field adjustment module uses a servo motor to adjust the air duct angle and the position of the airflow distribution plate based on the medicinal material characteristic dataset to obtain the actual air duct angle. The closed-loop monitoring module collects the separation trajectory and separation effect data of the medicinal materials in real time. If the separation trajectory of the medicinal materials deviates from the preset range, the actual air duct angle and the actual fan speed are corrected to obtain the optimal air duct angle and the optimal fan speed. The optimal air duct angle and the optimal fan speed are then updated to the pre-established medicinal material air separation parameter mapping database to obtain the final equipment control parameter scheme.

[0014] Thirdly, the present invention also provides an electronic device, including a processor, a memory, and a computer program stored in the memory and configured to be executed by the processor, wherein the processor executes the computer program to implement the intelligent control method for a medicinal herb wind separator as described in any one of the above.

[0015] Fourthly, the present invention also provides a computer-readable storage medium comprising a stored computer program, wherein, when the computer program is executed, it controls the device containing the computer-readable storage medium to perform any of the above-described intelligent control methods for a medicinal herb wind separator.

[0016] Compared with the prior art, the present invention has the following beneficial effects: (1) This invention stores the range of suitable parameters for different categories of medicinal materials by pre-establishing a database of wind-sorting parameters for medicinal materials. Based on the classification results and physical property labels, the corresponding range of fan speed and air duct angle can be quickly queried, avoiding the waste of high-quality medicinal materials caused by manual trial and error. It solves the problem that the existing technology relies solely on quality for simple classification, which cannot cope with the wind-sorting deviation and limited classification accuracy caused by surface roughness, irregular shape and other characteristics in the high-quality processing scenario of medicinal materials.

[0017] (2) This invention selects the optimal parameter combination by evaluating the separation effect, updates the medicinal material air separation parameter mapping database, and forms a continuous optimization learning mechanism. By recording the separation performance under different parameters, the optimal parameter combination is continuously accumulated and the database is updated, enabling the system to continuously improve the processing accuracy of similar medicinal materials through data accumulation, while enhancing its adaptability to new medicinal materials and complex working conditions. This achieves continuous updating and long-term efficient operation of the air separation system in high-quality medicinal material processing scenarios.

[0018] (3) By introducing physical property labels, especially uniformity and roughness, this invention improves the wind separation parameter setting mechanism, further increases the success rate of wind separation of high-quality medicinal materials, adapts to the refined needs of high-quality medicinal materials for high-quality processing, dynamically generates wind separation parameter combinations exclusive to medicinal materials, and realizes intelligent control. Attached Figure Description

[0019] Figure 1This is a schematic diagram of the intelligent control method for a medicinal herb air separator provided in the first embodiment of the present invention; Figure 2 This is a schematic diagram of the intelligent control system for a medicinal herb air separator provided in the second embodiment of the present invention. Detailed Implementation

[0020] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0021] This embodiment is based on a high-quality processing scenario for medicinal materials. First, batches of medicinal materials are inspected and air-separation parameters are assigned, followed by air-separation processing. During air-separation, operations are performed directly according to the determined air-separation parameters. Furthermore, the operation data and air-separation results can be recorded as historical data for updating air-separation parameters and optimizing the SVM model.

[0022] Reference Figure 1 The first embodiment of the present invention provides an intelligent control method for a medicinal herb air separator, comprising the following steps: S11, acquire the mass distribution data and density feature value of the medicinal material, scan in real time to acquire the outline data and texture features of the medicinal material, analyze the mass distribution data and density feature value, and obtain the medicinal material characteristic dataset; S12, Based on the medicinal material characteristic dataset, classify and identify the medicinal materials to obtain the classified medicinal materials and generate corresponding physical characteristic labels; S13. Based on the classified medicinal materials and the physical property labels, query the pre-established medicinal material air separation parameter mapping database to obtain the equipment control parameter scheme required for medicinal material air separation. S14. Based on the classified medicinal materials, dynamically adjust the fan speed to obtain a fan speed that matches the category of medicinal materials; S15, Based on the medicinal material characteristic dataset, use a servo motor to adjust the air duct angle and the position of the airflow distribution plate to obtain the actual air duct angle; S16. Real-time acquisition of the separation trajectory and separation effect data of medicinal materials. If the separation trajectory of medicinal materials is detected to deviate from the preset range, the actual air duct angle and the actual fan speed are corrected to obtain the optimal air duct angle and the optimal fan speed. The optimal air duct angle and the optimal fan speed are updated to the pre-established medicinal material air separation parameter mapping database to obtain the final equipment control parameter scheme.

[0023] In step S11, the mass distribution data and density feature values ​​of the medicinal materials are obtained, the contour data and texture features of the medicinal materials are obtained by real-time scanning, and the mass distribution data and density feature values ​​are analyzed to obtain a medicinal material characteristic dataset.

[0024] In one feasible approach, the acquisition of the mass distribution data and density feature values ​​of the medicinal materials, real-time scanning of the medicinal materials for contour extraction and texture analysis, and analysis of the mass distribution data and density feature values ​​to obtain a medicinal material characteristic dataset include: The first dataset is obtained by acquiring the mass distribution data and density characteristic values ​​of medicinal materials. The second dataset is obtained by scanning the medicinal materials in real time to extract contours and perform texture analysis. By merging the first data set and the second data set, and integrating multi-dimensional physical parameters, a third data set is generated; Compare the parameters in the third data set with a preset parameter threshold. If the parameter values ​​in the third data set meet the preset parameter threshold, then record the third data set to obtain the medicinal material characteristic dataset. If the parameter values ​​in the third data set do not meet the preset parameter threshold, the medicinal materials are re-collected, the first data set and the second data set are regenerated, and then merged back into the third data set. If the third data set still does not meet the preset parameter threshold after multiple re-collections, the medicinal materials are determined to be temporarily unsuitable for the air separation process.

[0025] It should be noted that real-time scanning of the medicinal materials using a mass sensor acquires their mass distribution data, while a high-precision X-ray density scanner obtains their density characteristic values. The acquired mass distribution data and density characteristic values ​​are integrated to form a first dataset, which comprehensively reflects the physical characteristics of the medicinal materials in terms of mass and density. High-resolution 3D laser scanning technology is used to scan the medicinal materials from all angles, acquiring their precise contour data and extracting their length, width, height, surface area, volume, and geometric features such as contour regularity (defined as the ratio of the actual projected area of ​​the medicinal material's contour to the area of ​​its circumscribed rectangle). Texture feature extraction (including texture roughness) utilizes a texture analysis algorithm based on Gray-Level Co-occurrence Matrix (GLCM) and Local Binary Pattern (LBP). Specifically, the GLCM-based algorithm converts the acquired surface image of the medicinal materials into a grayscale image; to balance computational complexity and feature discrimination, the image grayscale levels are compressed to 16 levels. The pixel-to-pixel distance d = 1 (pixel unit) was set, and the gray-level co-occurrence matrices were calculated in four directions: 0°, 45°, 90°, and 135°. A set of texture features, including contrast and correlation, was calculated from the matrix in each direction. The average of these four features was then used as the GLCM texture feature vector for the medicinal material. Simultaneously, to capture the microscopic texture patterns on the surface of the medicinal material, a uniform mode LBP algorithm was employed. The algorithm parameters were set as follows: neighborhood radius R = 1 (pixel) and number of neighborhood points P = 8. The LBP feature map of the entire image was calculated, and then the histogram was used as a texture descriptor. The LBP parameters reflect the roughness of the medicinal material's surface. The extracted contour data and texture features were organized to form a second dataset. This dataset characterizes the appearance of the medicinal material from both geometric and surface structure perspectives.

[0026] The first and second datasets are merged, and the multidimensional physical parameters from both datasets are organized and combined to form a third dataset. This fusion process is a systematic data preprocessing and feature integration workflow. By using unified timestamps and medicinal material batch IDs, the correct association of data from the medicinal materials is ensured. The Z-score normalization method is used to transform all feature parameters (density, quality, contour regularity, texture contrast) to a unified quantitative scale, eliminating the influence of dimensions. The calculation formula is as follows: Where x is the original value, The mean, The standard deviation is used. All processed feature values ​​are combined into a multi-dimensional feature vector in the order of density, quality, contour regularity, and texture contrast, generating a third dataset. Each parameter in the third dataset is compared with a preset parameter threshold. The preset parameter threshold range is a numerical value constructed based on multi-dimensional prior knowledge; initial training is required before the system is first deployed or before introducing new medicinal herb varieties. A massive amount (usually 30,000 samples) of medicinal herb samples of known varieties and known quality grades (suitable for and unsuitable for air sorting) is collected to construct an initial knowledge base, establishing an independent benchmark model for each specific medicinal herb variety. For high-quality samples of each variety, the statistical distribution of its standardized feature parameters (density Z-score, quality Z-score, contour regularity Z-score, and texture contrast regularity Z-score) is calculated, and its mean ± 3σ (covering 99.7% of the data) is used as the initial preset threshold range suitable for the variety to enter the air sorting process. If all parameter values ​​are within the preset threshold range, it indicates that the data set accurately reflects the normal physical characteristics of the medicinal materials. At this point, the third dataset is recorded as the medicinal material characteristic dataset for subsequent processes. If some parameter values ​​do not meet the preset threshold, the system will automatically trigger a re-collection mechanism to ensure data accuracy and reliability. Data is collected again from the same batch of medicinal materials, generating a new first and second dataset using the same method, and then merging them again to obtain a new third dataset. The parameter threshold is then checked again. If, after three re-collections and checks, the parameter values ​​in the third dataset still fail to meet the preset threshold, it is determined that the batch of medicinal materials may have quality abnormalities and is temporarily unsuitable for the air-separation process.

[0027] In step S12, the medicinal materials are classified and identified according to the medicinal material characteristic dataset to obtain the classified medicinal materials and generate corresponding physical characteristic labels.

[0028] In one feasible approach, the step of classifying and identifying medicinal materials based on the medicinal material characteristic dataset to obtain classified medicinal materials and generating corresponding physical characteristic labels includes: Extract the acquired density feature values ​​from the medicinal material characteristic dataset. The density feature values ​​are density parameters directly obtained through the detection device. The density feature values ​​are quantized, and the mean and standard deviation of the density feature values ​​are calculated to obtain the quantized density feature values. If the quantified density characteristic value is less than the preset lightness threshold, the medicinal material is initially classified as a light medicinal material category. If the quantified density feature value is greater than the preset heavy threshold, the medicinal material is initially classified as a heavy medicinal material category. If the quantified density feature value is between the preset light threshold and the preset heavy threshold, the medicinal material is initially classified as a medium-quality medicinal material, and the classified medicinal material is obtained. By combining machine learning models, multidimensional features are extracted from the classified medicinal materials to obtain corresponding physical property labels.

[0029] It should be noted that the density characteristic values ​​of medicinal materials obtained using a high-precision X-ray density scanner are quantified. First, the density units output by the monitoring equipment are uniformly converted to the preset dimension g / cm³ to eliminate the influence of dimensional differences on classification. Then, the mean and standard deviation of the density of the batch of medicinal materials are calculated through statistical analysis. The mean density is used as the core, and is compared numerically with preset light and heavy thresholds. The preset light and heavy thresholds are determined based on a large amount of experimental data, referring to the density distribution range of common medicinal materials. Combining industry data, the light threshold is set at 0.5 g / cm³, and the heavy threshold at 1.0 g / cm³. If the mean density of the medicinal material is less than the light threshold, it is initially classified as a light medicinal material; if the mean density is greater than the heavy threshold, it is initially classified as a heavy medicinal material; if the mean density is between the light and heavy thresholds, it is initially classified as a medium-weight medicinal material.

[0030] The process of optimizing classified medicinal materials using the Support Vector Machine (SVM) algorithm integrates multi-dimensional features to achieve accurate classification and labeling. Sample data from the classified medicinal materials are extracted for three categories: light, medium, and heavy. Standardized feature vectors are constructed for each category. Each sample's feature vector contains four core features: density mean (μ), which represents the uniformity of density, the arithmetic mean of sample density (unit: g / cm³), reflecting the overall density level; density standard deviation (σ), reflecting the uniformity of sample density distribution, with a larger value indicating more significant density fluctuations; contour data, quantified using contour regularity (calculated as: actual contour area of ​​the medicinal material / area of ​​the circumscribed rectangle of the medicinal material × 100%), with a higher value indicating a more regular shape; and texture features, i.e., texture roughness, extracted using the Local Binary Pattern (LBP) algorithm. The neighborhood radius is set to R = 1 pixel, and the number of neighborhood points is P = 8. The LBP map is calculated using uniform mode, and its histogram is statistically analyzed. The LBP feature values ​​obtained after histogram normalization are mapped to the interval [0,1], with a higher value indicating a rougher surface. The original values ​​of the four core features were linearly scaled and uniformly mapped to the [0,1] interval to eliminate the influence of dimensional differences on the SVM model. The standardized feature vectors of the three classes of samples (light, medium, and heavy protons) were merged into a total dataset and randomly divided into a training set (80%) and a test set (20%) in an 8:2 ratio. The training set was used for parameter learning and training of the SVM model; the test set was used to evaluate the model's classification accuracy. The radial basis function (RBF) was selected as the SVM kernel function to handle the nonlinear relationships between features. A hierarchical training strategy was adopted, training three sub-models (light proton model, medium proton model, and heavy proton model) for the three initially classified medicinal materials to achieve refined subdivision of samples within the same class. For example, the light proton model further divides light medicinal materials into four subclasses based on the distribution characteristics of texture roughness (LBP value) and density uniformity (σ): Subclass 1, low uniformity (σ≥0.15) + high roughness (LBP≥0.6); Subclass 2, low uniformity (σ≥0.15) + low roughness (LBP<0.6); Subclass 3, high uniformity (σ<0.15) + high roughness (LBP≥0.6); Subclass 4, high uniformity (σ<0.15) + low roughness (LBP<0.6). To address the strong discriminative power of roughness and uniformity, recursive feature elimination (RFE) is used during model training to increase their weighting. An initial Linear SVM model is trained using a training set containing all initial features. Features are then ranked based on the absolute value of the coefficients generated by the current SVM model; the smaller the absolute value of the coefficient, the lower the contribution of the feature to the classification decision. The feature with the lowest contribution in the current ranking is removed. This process is repeated on the remaining feature subset, retraining the model and removing the least important feature again. This process is repeated until the final feature subset contains only roughness and uniformity, which is defined as the optimal feature subset.Subsequently, the optimal feature subset is used to retrain the final SVM classification model, thereby achieving refined classification and obtaining the classification results of medicinal materials and their corresponding physical property labels.

[0031] It's worth noting that the Linear SVM model is a specific form of the SVM (Support Vector Machine) model; the two have a containment relationship. SVM is a generalized model framework, while Linear SVM is a specific implementation of SVM under certain conditions. The core idea of ​​SVM is to achieve regression by finding the optimal hyperplane. Its goal is to find a hyperplane in the feature space that clearly separates samples of different classes, and maximizes the distance between the hyperplane and its nearest neighbors. When samples are linearly inseparable in the original feature space, SVM can map the samples to a higher-dimensional space using a kernel function, making them linearly separable in the new space. Linear SVM, on the other hand, is a special case of SVM. It uses only a linear kernel function, directly searching for a linear hyperplane in the original feature space for classification. The expression for the linear kernel function is: (i.e., the inner product of the feature vectors of two samples), at which point the decision boundary of the model is linear.

[0032] This invention employs a hybrid model combining Linear Support Vector Machine (Linear SVM) and Recursive Feature Elimination (RFE). The core principle is to use RFE to filter key features and then leverage Linear SVM for accurate classification. The initial input features are multidimensional physical parameters from the medicinal herb characteristic dataset, including density features (mean density (μ, unit g / cm³), density standard deviation (σ, reflecting uniformity); contour features (contour regularity, ratio of actual contour area to circumscribed rectangle area, %); and texture features: roughness extracted based on Local Binary Pattern (LBP) (mapped to the [0,1] interval, higher values ​​indicate rougher surfaces). These features are standardized using Z-scores to form an initial feature vector (4 dimensions). The RFE and Linear SVM structure is then combined. The core of Linear SVM is the use of a linear kernel function. The algorithm separates different categories of medicinal materials by finding the optimal hyperplane, and the output feature coefficients (the weights of each feature corresponding to the hyperplane) are used to measure feature importance. The RFE (Resource Estimation and Filtering) screening mechanism uses Linear SVM as the base model and iteratively removes features with low contribution while retaining key features. Specifically: a Linear SVM is trained with the current feature subset to obtain the absolute value of the coefficients of each feature; the coefficients are sorted in ascending order of absolute value (the smaller the value, the lower the contribution of the feature to the classification decision); the last feature in the sort is removed; the above steps are repeated until only roughness and uniformity are retained in the feature subset, forming the optimal feature subset (dimensional 2).

[0033] Model training: Model training is a supervised learning process, divided into two stages: initial feature training and RFE screening, followed by retraining of the optimal feature subset. The specific steps are as follows: First, data preparation is performed. 30,000 samples of medicinal materials of known categories (light / medium / heavy) and quality grades are collected and divided into a training set (80%, used for model learning) and a test set (20%, used for accuracy evaluation) at an 8:2 ratio. All initial features (uniformity, density standard deviation, contour regularity, roughness) are Z-score standardized to eliminate dimensional differences. Then, initial Linear SVM training is performed using all initial features (4-dimensional) from the training set, with a penalty parameter C=1.0 and a hinge loss function. To optimize the objective, the optimal hyperplane parameters are solved using the Lagrange duality method, outputting the coefficient vector of each feature. The absolute value of the feature coefficients is used as an indicator to quantify the contribution of each feature to the classification decision; the larger the absolute value of the coefficient, the higher the feature importance. Then, a recursive feature elimination (RFE) process is performed. In the first round, features are sorted by absolute value from smallest to largest, and the last feature is removed, leaving three dimensions. In the second round, the model is retrained and sorted based on the three-dimensional features, and the last feature is removed, ultimately retaining the uniformity and roughness two-dimensional features to form the optimal feature subset. The trained model can receive uniformity and roughness features from new samples, outputting refined classification results and corresponding physical characteristic labels, providing a basis for parameter adjustment of the wind sorting equipment.

[0034] In step S13, based on the classified medicinal materials and the physical property labels, the pre-established medicinal material air separation parameter mapping database is queried to obtain the equipment control parameter scheme required for medicinal material air separation.

[0035] In one feasible approach, the step of querying a pre-established medicinal herb air-sorting parameter mapping database based on the classified medicinal herbs and the physical property labels to obtain the equipment control parameter scheme required for medicinal herb air-sorting includes: Based on the classified medicinal materials and the physical property labels, a pre-established medicinal material air separation parameter mapping database is queried to obtain a preliminary set of parameters for the fan speed and duct angle corresponding to the classified medicinal materials and the physical property labels. Based on the classified medicinal materials and the preliminary parameter set, the range of fan speed parameters and the range of air duct angle are determined, and the range of fan speed parameters and the range of air duct angle are combined to obtain the equipment control parameter scheme required for the air separation of medicinal materials.

[0036] It should be noted that the pre-established medicinal herb air-separation parameter mapping database is a structured data storage system built based on physical principles and parameter correlation analysis. By combining theoretical modeling (suspension velocity in fluid mechanics) and simulation optimization (CFD flow field simulation), the mapping relationship between the physical properties of medicinal herbs (density, mass, profile, texture) and air-separation parameters (fan speed, duct angle) is established. Its core function is to achieve accurate mapping between medicinal herb properties and air-separation parameters. Specifically, density directly affects the suspension velocity of medicinal herbs in the airflow. According to the formula (in For the quality of medicinal materials, It is the acceleration due to gravity. air density, For the windward surface area of ​​the medicinal herbs, The density of medicinal materials (representing drag coefficient) requires higher suspension velocities, thus necessitating higher fan speeds and steeper duct angles to ensure effective separation. Mass distribution affects the inertia and acceleration of the medicinal materials in the airflow; heavier materials require stronger airflow forces (achieved by increasing fan speed) and optimized duct angles to overcome inertia, ensuring they are fully aerated and move along the intended trajectory. Profile data (such as regularity and projected area) determines the windward area and drag characteristics of the medicinal materials. Irregularly shaped materials or those with larger projected areas are more significantly affected by airflow, requiring appropriate reductions in fan speed and adjustments to duct angles to avoid excessive tumbling or deviation from the trajectory. Surface texture affects the development of the airflow boundary layer and frictional resistance; rougher textures result in higher surface friction coefficients, requiring higher fan speeds to generate stronger airflow to overcome resistance, while simultaneously adjusting duct angles to optimize airflow impact angles and ensure effective separation. Through these mapping relationships, the database presets optimal fan speed and duct angle parameter combinations for medicinal materials with different characteristics, providing data support for intelligent control.

[0037] The database employs a multidimensional index structure. The main table contains fields such as "Medicinal Material Classification Result," "Physical Property Label," "Initial Range of Fan Speed," and "Initial Range of Duct Angle." The "Physical Property Label" field must precisely match sub-labels such as "Lightweight - High / Low Uniformity - High / Low Roughness," "Medium-High / Low Uniformity - High / Low Roughness," and "Heavyweight - High / Low Uniformity - High / Low Roughness," ensuring that the granularity of the parameter query matches the refinement of the classification results. During the initial parameter set query, the system uses a two-dimensional matching algorithm. First, it locates the primary parameter range based on the medicinal material classification results (lightweight / medium-heavyweight). Lightweight medicinal materials correspond to a low-speed base range (500-800 rpm) and a small-angle base range (5-20 degrees), while heavyweight medicinal materials correspond to a high-speed base range (1200-1500 rpm) and a large-angle base range (25-40 degrees). Then, it performs a secondary calibration of the primary range using sub-features (uniformity, roughness) in the physical property labels. For example, for the label "lightweight - low uniformity - high roughness", the system will increase the upper limit of the fan speed by 10% (to 880 rpm) within the basic range of lightweight medicinal materials to cope with density fluctuations, while lowering the lower limit of the air duct angle by 2 degrees (to 3 degrees) to enhance the force of airflow on rough surfaces, ultimately generating a targeted preliminary parameter set. When determining the range of fan speed parameters and air duct angle, feature correlation analysis is required. If the physical characteristic label indicates high roughness, the rotation speed is increased by 50 rpm and the angle is decreased by 2 degrees to enhance airflow penetration and reduce trajectory divergence caused by surface friction. If the label indicates low roughness, the rotation speed is decreased by 40 rpm and the angle is increased by 3 degrees to reduce the impact of airflow on smooth surfaces and stabilize the trajectory using gravity by increasing the tilt angle. If the label indicates low uniformity, the rotation speed fluctuation range is reduced by 50 rpm and the angle fluctuation range is increased by 5 degrees to adapt to local density differences in the medicinal materials through more flexible angle changes. If the label indicates high uniformity, the rotation speed fluctuation range is reduced by 30 rpm and the angle fluctuation range is reduced by 3 degrees. By directly pairing the fan speed parameter range and the duct angle range, the equipment control parameter scheme required for the air separation of medicinal materials is obtained.

[0038] In step S14, based on the medicinal material classification results and combined with the real-time monitored airflow intensity, the fan speed is dynamically adjusted to match the fan speed with the medicinal material category, thus obtaining the actual fan speed.

[0039] In one feasible approach, the step of dynamically adjusting the fan speed based on the medicinal material classification results and in conjunction with the real-time monitored airflow intensity to match the fan speed with the medicinal material category, thereby obtaining the actual fan speed, includes: Based on the classified medicinal materials, query the pre-established medicinal material air separation parameter mapping database to obtain the target rotation speed range and air separation airflow intensity parameters corresponding to the medicinal material classification results; Based on the classified medicinal materials, the airflow intensity parameters of the air sorting are dynamically adjusted to obtain the baseline airflow intensity; The matching degree between the actual airflow intensity and the benchmark airflow intensity during the air separation process is monitored in real time, and the fan speed is dynamically adjusted to correct the deviation, so as to obtain the actual fan speed that is accurately matched with the type of medicinal material.

[0040] It should be noted that the pre-established database of medicinal herb air separation parameters includes the target rotation speed range and airflow intensity parameters required for herb air separation. The matching relationship between the target rotation speed range and the airflow intensity parameters is established based on the correlation between the physical characteristics of the herb type and the airflow force. For light medicinal herbs (such as peppermint leaves), the target rotation speed range is set to 500-800 rpm, and the matching airflow intensity parameter is 0.8-1.2 Pa (suitable for generating a weak airflow to avoid excessive separation of the herbs); for heavy medicinal herbs (such as astragalus root), the target rotation speed range is 1200-1500 rpm, and the airflow intensity parameter is 2.5-3.0 Pa (requiring a stronger airflow to ensure separation effect); for medium-weight medicinal herbs, the corresponding rotation speed range is 800-1200 rpm and the airflow intensity is 1.2-2.5 Pa, forming a stepped parameter system. When dynamically adjusting the airflow intensity parameters of the air classifier, a secondary calibration is required, taking into account the physical property labels of the medicinal materials. The core of this secondary calibration is to refine the initial baseline airflow intensity value obtained from the query based on the combination of the medicinal material's density category and its physical property labels. The calibrated baseline airflow intensity = initial baseline airflow intensity × (1 + baseline airflow intensity adjustment rate). For light medicinal materials, if the physical property label is "high uniformity - high roughness," the baseline airflow intensity adjustment rate increases by 8%; if it is "high uniformity - low roughness," the adjustment rate decreases by 7%; if it is "low uniformity - high roughness," the adjustment rate increases by 12%; and if it is "low uniformity - low roughness," the adjustment rate remains unchanged. For medium-weight medicinal materials, if the physical property label is "high uniformity - high roughness," the adjustment rate increases by 5%; and if it is "high uniformity - low roughness," the adjustment rate decreases by 7%. For heavy medicinal materials, if the physical characteristic label is "low uniformity - high roughness", the reference airflow intensity adjustment rate increases by 10%; if the physical characteristic label is "low uniformity - low roughness", the reference airflow intensity adjustment rate increases by 2%. For heavy medicinal materials, if the physical characteristic label is "high uniformity - high roughness", the reference airflow intensity adjustment rate increases by 4%; if the physical characteristic label is "high uniformity - low roughness", the reference airflow intensity adjustment rate remains unchanged; if the physical characteristic label is "low uniformity - high roughness", the reference airflow intensity adjustment rate increases by 6%; if the physical characteristic label is "low uniformity - low roughness", the reference airflow intensity adjustment rate increases by 3%. The reference airflow intensity is obtained according to the calibration rules for different types of medicinal materials.

[0041] In the real-time monitoring stage, a high-precision hot-wire anemometer is used to collect the actual air flow intensity in the separation chamber at a frequency of 10 times per second. The data sampling points cover three key areas: the air duct outlet, the middle of the separation chamber, and the discharge outlet, ensuring the spatial representativeness of the monitoring. The matching degree between the actual air flow intensity and the reference air flow intensity is calculated by the deviation rate: deviation rate = (actual value - reference value) / reference value × 100%. When the absolute value of the deviation rate ≤ 5%, the air flow matching is judged to be qualified, and the current fan speed is maintained; if the deviation rate > 5% or < -5%, the PID control algorithm is triggered for dynamic adjustment. The parameter settings of the PID control algorithm need to be configured differently according to the types of medicinal materials: for light medicinal materials, a smaller proportional coefficient (Kp = 0.3) and a larger integral time (Ti = 2.0 s) are adopted to avoid sudden changes in air flow caused by excessive fan speed adjustment; for heavy medicinal materials, a larger proportional coefficient (Kp = 0.6) and a smaller integral time (Ti = 1.0 s) are adopted to ensure a rapid response to air flow deviation. For example, when the actual air flow intensity of heavy medicinal materials is 2.3 Pa (reference value 3.0 Pa, deviation rate -23.3%), the fan speed adjustment amount calculated by the PID algorithm is +150 revolutions per minute, and the fan speed is increased from 1200 revolutions per minute to 1350 revolutions per minute, so that the actual air flow intensity rises to 2.8 Pa (deviation rate -6.7%), until it stabilizes within the range of ±5% of the reference value, and finally the actual fan speed is determined.

[0042] In step S15, according to the medicinal material characteristic data set, a servo motor is used to adjust the air duct angle and the position of the air flow distribution plate to obtain the actual air duct angle.

[0043] In an implementable manner, the using a servo motor to adjust the air duct angle and the position of the air flow distribution plate according to the medicinal material characteristic data set to obtain the actual air duct angle includes: Obtaining the density characteristic value from the medicinal material characteristic data set, calculating the air duct inclination angle according to the density characteristic value to obtain the optimal air duct inclination angle; If the optimal air duct inclination angle exceeds the preset air duct inclination angle range, the air duct is adjusted to the preset air duct inclination angle range by a servo motor to obtain the adjusted air duct angle; According to the adjusted air duct angle, the servo motor is adjusted to adjust the position of the air flow distribution plate so that the air flow forms a flow field distribution pattern conforming to the medicinal material characteristics to obtain the actual air duct angle.

[0044] It should be noted that the calculation relationship between the density characteristic value and the air duct inclination angle is constructed based on the principle of fluid mechanics. Through a linear regression model, an empirical formula is obtained by fitting the principle of fluid mechanics and a large amount of experimental data: optimal air duct inclination angle , where is the density characteristic value (unit: g / cm³), and Here are the model parameters, where =25、 =5, meaning that for every 0.1 g / cm³ increase in density, the angle increases by 2.5 degrees. In the specific derivation process, through force analysis of the medicinal material in the air duct, the equation of motion and equilibrium conditions are introduced, and a steady-state assumption is made. After simplification, the relationship between density and mass is introduced, leading to the following formula: in, air density; This is the drag coefficient; The windward surface area of ​​the medicinal herb; These are density eigenvalues; Wind speed; The coefficient of contact friction; This refers to the volume of the medicinal material. Let be the acceleration due to gravity. Further simplifying, this can be represented as a mapping relationship: After sorting, we get: .

[0045] The preset duct tilt angle range needs to be determined in conjunction with the equipment's mechanical limits and safety thresholds, typically set between 5 and 40 degrees. If the calculated optimal angle exceeds this range, such as 42 degrees for heavy medicinal materials, the system triggers the servo motor to adjust the limit, locking the duct angle to 40 degrees; if the calculated optimal angle is below this range, the system triggers the servo motor to adjust the limit, locking the duct angle to 5 degrees. The position adjustment of the airflow distribution plate forms a linkage mechanism with the duct angle. The distribution plate is equipped with independently controllable guide vanes, and their opening degree (0-100%) determines the diffusion range and intensity of the airflow. First, the opening degree of the basic distribution plate vanes is determined based on the type of medicinal material and the duct angle. Different types of medicinal materials (light, medium, and heavy) correspond to different duct angle ranges, and the opening degree must match the duct angle to ensure that the airflow direction and diffusion range are adapted to the density characteristics of the medicinal material. For lightweight medicinal materials, the corresponding airflow angle is 5-20 degrees, with a basic opening and closing degree of 20-40%. At this angle, the airflow needs to be concentrated (narrow flow field) to prevent excessive dispersion of the lightweight materials. For medium-weight medicinal materials, the corresponding airflow angle is 20-30 degrees, with a basic opening and closing degree of 40-60%. The airflow needs to be evenly diffused to balance separation efficiency and stability. For heavy medicinal materials, the corresponding airflow angle is 30-40 degrees, with a basic opening and closing degree of 60-80%. The airflow needs to cover a wide area to enhance the driving force on the heavy medicinal materials. Secondly, dynamic adjustments are made based on the physical property labels (roughness, uniformity) of the medicinal materials to adapt to their microscopic characteristics. Regarding roughness, if the label indicates high roughness, the opening and closing degree needs to be increased by 5-10% to enhance airflow penetration (counteracting the obstruction of airflow by surface friction); if the label indicates low roughness, the opening and closing degree needs to be decreased by 3-8% to prevent excessive airflow impact leading to trajectory deviation. Regarding uniformity, if the label has low uniformity, the opening and closing fluctuation range should be reduced to ±3%; if the label has high uniformity, the opening and closing fluctuation range can be expanded to ±5% to maintain airflow consistency.

[0046] The effectiveness of the flow field distribution pattern is verified in real time using particle image velocimetry (PIV) technology to ensure the uniformity of the airflow velocity field within the separation chamber (requiring a coefficient of variation of velocity at each monitoring point ≤10%). If local airflow turbulence, uneven velocity distribution, or eddies caused by improper distribution plate positioning are detected, the system will initiate a feedback-based flow field optimization algorithm. This algorithm aims to minimize the coefficient of variation of velocity, employing a gradient descent search strategy to make small, incremental adjustments to the angles of each guide vane (typically with a step size of 0.5°-1°), and evaluates the PIV feedback data in real time. The system iteratively performs the fine-tuning of the vane angles until the flow field uniformity index meets the preset model requirements (coefficient of variation of velocity at each monitoring point ≤10%), ultimately locking in the current airflow distribution plate configuration and duct angle, and determining it as the actual duct angle.

[0047] In step S16, the separation trajectory and separation effect data of the medicinal materials are collected in real time. If the separation trajectory of the medicinal materials is detected to deviate from the preset range, the actual air duct angle and the actual fan speed are corrected to obtain the optimal air duct angle and the optimal fan speed. The optimal air duct angle and the optimal fan speed are then updated to the pre-established medicinal material air separation parameter mapping database to obtain the final equipment control parameter scheme.

[0048] In one feasible approach, the real-time acquisition of the separation trajectory and separation effect data of the medicinal materials, if the separation trajectory deviates from a preset range, corrects the actual air duct angle and the actual fan speed to obtain the optimal air duct angle and the optimal fan speed. The optimal air duct angle and the optimal fan speed are then updated to the pre-established medicinal material air separation parameter mapping database to obtain the final equipment control parameter scheme, including: Real-time acquisition of separation trajectory data and separation effect data of medicinal materials, extraction of the position coordinates of medicinal materials, and obtaining the actual separation trajectory; The actual separation trajectory is compared with the preset trajectory range, and the trajectory deviation value is calculated; If the trajectory deviation value exceeds the preset trajectory deviation threshold, a feedback adjustment signal is generated to calculate the adjustment amount of the duct angle and the fan speed. The actual duct angle and the actual fan speed are corrected based on the adjustment amount to obtain the optimal duct angle and the optimal fan speed. The optimal parameter combination is obtained based on the optimal duct angle and the optimal fan speed. The optimal parameter combination is then updated to the pre-established medicinal herb air separation parameter mapping database to obtain the final equipment control parameter scheme.

[0049] It should be noted that the real-time acquisition process employs a visual monitoring system consisting of a high-resolution industrial camera (resolution no less than 1920×1080) and infrared supplementary lighting, capturing the movement of the medicinal materials within the separation chamber at a frequency of 30 frames per second. Using a deep learning-based target tracking algorithm, YOLOv5 combined with Kalman filtering, the centroid coordinates (x, y) of individual medicinal materials are extracted from the image, forming a continuous sequence of trajectory points, which constitutes the actual separation trajectory. The preset trajectory range is a dynamically defined area pre-calibrated for different types of medicinal materials. The trajectory deviation value is calculated using dynamically weighted Euclidean distance: a higher weight (0.7) is assigned to the initial segment of the trajectory (just entering the separation chamber), and a lower weight (0.3) is assigned to the later segment (near the outlet), to focus on correcting deviations in the initial separation stage. The deviation threshold (D0) is determined based on the type and size of the medicinal material and the required separation accuracy. The base value is obtained through... The principle is based on the coordinate standard deviation of standard trajectory samples, combined with dynamic correction for equipment errors (3-6mm), and stored according to medicinal material category-characteristic label. When the real-time collected deviation value of the medicinal material separation trajectory does not exceed the preset trajectory deviation threshold, it indicates that the current air separation parameters can achieve the expected separation effect, and the system does not need to adjust the parameters. When the deviation value exceeds the deviation threshold, the system will automatically generate a feedback adjustment signal to trigger the parameter adjustment command. The adjustment amount is not randomly set, but is calculated based on the trajectory deviation to adjust the air duct angle and fan speed. Let the current air duct angle be... (Unit: degrees), current fan speed is (Unit: revolutions per minute); trajectory deviation value is (Calculated using dynamically weighted Euclidean distance); the deviation threshold is... The trajectory fluctuation amplitude is (Maximum displacement difference of trajectory points over 10 consecutive frames, unit: mm). Trajectory deviation value. The calculation, derived through dynamic weighted Euclidean distance, quantifies the deviation between the actual separation trajectory of the medicinal material and a preset standard trajectory, assigning different weights based on the importance of each stage of the trajectory. The core coordinate sequence of the medicinal material is acquired in real-time using an industrial camera (30 frames / second). Where n is the number of trajectory points, and each trajectory segment contains 50 frames of data, corresponding to 50 points), the trajectory range is preset for specific medicinal material categories and physical characteristic labels, i.e. The trajectory deviation value is calculated based on the actual trajectory points and preset trajectory points at the same time and position stage, that is, for each group... Calculate the deviation. For the first... i Frame trajectory points, calculate the straight-line distance between the actual point and the preset point. The unit is mm, reflecting the degree of deviation in a single frame. The first segment of the trajectory, i.e., the first 30% of the trajectory points, is... i =1 to i =0.3n, assign weight w =0.7; the latter 70% of the trajectory points, i.e. i =0.3n+1 to i =n, assign weights w =0.3. The weighted distance of all trajectory points is averaged. When the trajectory is generally biased towards the inside of the air duct. At that time, the duct angle adjustment amount is (Slightly adjust outwards, with each adjustment increment fixed at 0.5 degrees), the fan speed adjustment amount is... The speed is reduced in 5% increments from the current speed, and the adjusted speed is: All other parameters remain unchanged; when the trajectory diverges... The airflow distribution plate opening and closing degree adjustment amount is: (+5% if the trajectory deviates to the left, -5% if it deviates to the right), air duct angle auxiliary adjustment amount: The direction is opposite to the trajectory offset direction, and the adjusted angle is... After adjustment, the trajectory is re-acquired until the deviation values ​​of three consecutive trajectory segments are all less than the preset deviation threshold. The parameters at this point are the optimal duct angle and the optimal fan speed. The optimal duct angle and the optimal fan speed are combined to obtain the optimal parameter combination. The optimal parameters are then bound to the medicinal material classification and physical characteristic labels, and stored in the database through a hierarchical index structure (stored according to "medicinal material category - characteristic label"). The final equipment control parameter scheme is then obtained by updating the database.

[0050] To facilitate understanding of the present invention, some preferred embodiments of the present invention will be described in further detail below.

[0051] In this embodiment, taking the air separation process of honeysuckle as an example, the specific process of the intelligent control method and system for a medicinal material air separation machine based on the present invention is as follows: A mass sensor array was used to scan honeysuckle samples to obtain single-grain mass distribution data (range 0.12-0.48g). Density characteristic values ​​(mean 0.42g / cm³, standard deviation 0.05g / cm³) were obtained using a high-precision X-ray density scanner. Contour data were extracted using a 3D laser scanner. The mean regularity of the honeysuckle flower contour (actual contour area / circumscribed rectangle area) was 72%, and the mean regularity of the flower stalk contour was 45%. Texture was analyzed using Gray-Level Co-occurrence Matrix (GLCM) and Local Binary Pattern (LBP). The mean surface roughness (LBP value) of the flower was 0.3, and the mean roughness of dust impurities was 0.8. A third dataset was formed by fusing mass, density, contour, and texture data. This dataset was compared with preset thresholds (mass 0.1-0.5g, density 0.3-0.6g / cm³, contour regularity ≥50%, roughness ≤0.6). All parameters met the thresholds, and the dataset was identified as the honeysuckle characteristic dataset.

[0052] The extracted density feature value was 0.42 g / cm³, and compared with the preset light threshold (0.5 g / cm³) and heavy threshold (1.0 g / cm³), since 0.42 < 0.5, it was initially classified as a light medicinal material. A Support Vector Machine (SVM) algorithm was used to fuse multidimensional features, with the input features being a density mean of 0.42 g / cm³, contour regularity of 72% (high), and roughness of 0.3 (low). The secondary classification results distinguished between high-quality honeysuckle flowers (approximately 85%) and light impurities (dust) (approximately 15%). A physical characteristic label "light-high uniformity-low roughness" was generated (corresponding to high-quality flowers).

[0053] Based on the tags "lightweight, high uniformity, low roughness," a preliminary parameter set was matched from the preset air separation parameter mapping database, with a fan speed range of 500-700 rpm and an air duct angle range of 8-15 degrees. Considering the characteristic that honeysuckle flowers are easily dispersed by strong airflow, the parameter range was refined to a fan speed of 550-650 rpm and an air duct angle of 10-12 degrees, forming the equipment control parameter scheme.

[0054] The baseline target speed range was obtained as 550-650 rpm, corresponding to a baseline airflow intensity of 0.85 Pa. The actual airflow intensity was monitored using a hot-wire anemometer and found to be 0.78 Pa (deviation rate -8.2%, exceeding the ±5% threshold). PID control was triggered with a proportional gain Kp = 0.3 (suitable for lightweight medicinal materials, avoiding sudden speed changes) and an integral time Ti = 2.0 s. The adjustment was calculated, increasing the speed by 30 rpm. The actual speed stabilized at 580 rpm, and the airflow intensity reached 0.83 Pa (deviation rate -2.3%, meeting the requirements). The actual fan speed was determined to be 580 rpm.

[0055] Based on the average density of 0.42 g / cm³, substitute into the formula. The optimal airflow angle was determined to be 15.5 degrees (within the preset range of 5-40 degrees). Combined with the "high uniformity" setting, the opening and closing degree of the airflow distribution plate was adjusted to 30% to create a narrow and stable airflow field (avoiding impact on the flowers), ultimately determining the actual airflow angle to be 15.5 degrees.

[0056] The separation trajectory of honeysuckle flowers was captured using an industrial camera (30 frames / second), and the centroid coordinates (x, y) were extracted. The preset trajectory range for high-quality flowers was x=80-120mm and y=200-250mm. The actual trajectory was x=75mm and y=220mm (deviation -6.25%, exceeding the 5% threshold). A feedback signal was generated, and adjustments were calculated: the air duct angle was increased by 0.5 degrees (to 16 degrees), and the fan speed was decreased by 20 rpm (to 560 rpm). The corrected trajectory was x=82mm (deviation +2.5%), which met the preset range. The optimal parameters (air duct angle 16 degrees, fan speed 560 rpm) were updated to the database as the default parameter scheme for subsequent honeysuckle wind separation.

[0057] Reference Figure 2 The second embodiment of the present invention provides an intelligent control system for a medicinal herb air separator, comprising: The data acquisition module acquires the mass distribution data and density characteristic values ​​of medicinal materials, scans in real time to acquire the outline data and texture features of medicinal materials, and analyzes the mass distribution data and density characteristic values ​​to obtain a medicinal material characteristic dataset. The classification and recognition module classifies and recognizes the medicinal materials based on the medicinal material characteristic dataset, obtains the classified medicinal materials, and generates corresponding physical characteristic labels; wherein, the physical characteristic labels include uniformity and roughness; The parameter matching module queries a pre-established medicinal herb air separation parameter mapping database based on the classified medicinal herbs and the physical property labels to obtain the range of fan speed and duct angle required for medicinal herb air separation. The speed control module dynamically adjusts the fan speed according to the classified medicinal materials to obtain a fan speed that matches the category of medicinal materials; The flow field adjustment module uses a servo motor to adjust the air duct angle and the position of the airflow distribution plate based on the medicinal material characteristic dataset to obtain the actual air duct angle. The closed-loop monitoring module collects the separation trajectory and separation effect data of the medicinal materials in real time. If the separation trajectory of the medicinal materials deviates from the preset range, the actual air duct angle and the actual fan speed are corrected to obtain the optimal air duct angle and the optimal fan speed. The optimal air duct angle and the optimal fan speed are then updated to the pre-established medicinal material air separation parameter mapping database to obtain the final equipment control parameter scheme.

[0058] It should be noted that the intelligent control device for a medicinal herb air separator provided in this embodiment of the invention is used to execute all the process steps of the intelligent control method for a medicinal herb air separator in the above embodiment. The working principles and beneficial effects of the two are one-to-one, so they will not be described again.

[0059] This invention also provides an electronic device. The electronic device includes a processor, a memory, and a computer program stored in the memory and executable on the processor. When the processor executes the computer program, it implements the steps described in the embodiments of the intelligent control method for a medicinal herb wind separator, for example... Figure 1 The step S11 shown. Alternatively, when the processor executes the computer program, it implements the functions of each module / unit in the above-described device embodiments, such as the data acquisition module.

[0060] For example, the computer program may be divided into one or more modules / units, which are stored in the memory and executed by the processor to complete the present invention. The one or more modules / units may be a series of computer program instruction segments capable of performing a specific function, which describe the execution process of the computer program in the electronic device.

[0061] The electronic device may be a desktop computer, laptop, handheld computer, or smart tablet, etc. The electronic device may include, but is not limited to, a processor and memory. Those skilled in the art will understand that the above components are merely examples of electronic devices and do not constitute a limitation on the electronic device. It may include more or fewer components than described above, or combine certain components, or different components. For example, the electronic device may also include input / output devices, network access devices, buses, etc.

[0062] The processor can be a Central Processing Unit (CPU), or other general-purpose processors, digital signal processors (DSPs), application-specific integrated circuits (ASICs), field-programmable gate arrays (FPGAs), or other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components, etc. A general-purpose processor can be a microprocessor or any conventional processor. The processor is the control center of the electronic device, connecting all parts of the electronic device via various interfaces and lines.

[0063] The memory can be used to store the computer programs and / or modules. The processor implements various functions of the electronic device by running or executing the computer programs and / or modules stored in the memory and by calling data stored in the memory. The memory may mainly include a program storage area and a data storage area. The program storage area may store the operating system, at least one application program required for a function (such as sound playback function, image playback function, etc.), etc.; the data storage area may store data created according to the use of the mobile phone (such as audio data, phonebook, etc.). In addition, the memory may include high-speed random access memory, and may also include non-volatile memory, such as hard disk, memory, plug-in hard disk, smart media card (SMC), secure digital (SD) card, flash card, at least one disk storage device, flash memory device, or other volatile solid-state storage device.

[0064] Wherein, if the modules / units integrated in the electronic device are implemented as software functional units and sold or used as independent products, they can be stored in a computer-readable storage medium. Based on this understanding, all or part of the processes in the methods of the above embodiments of the present invention can also be implemented by a computer program instructing related hardware. The computer program can be stored in a computer-readable storage medium, and when executed by a processor, it can implement the steps of the various method embodiments described above. The computer program includes computer program code, which can be in the form of source code, object code, executable files, or certain intermediate forms. The computer-readable medium can include: any entity or device capable of carrying the computer program code, recording media, USB flash drives, portable hard drives, magnetic disks, optical disks, computer memory, read-only memory (ROM), random access memory (RAM), electrical carrier signals, telecommunication signals, and software distribution media, etc. It should be noted that the content included in the computer-readable medium can be appropriately added or removed according to the requirements of legislation and patent practice in the jurisdiction. For example, in some jurisdictions, according to legislation and patent practice, computer-readable media do not include electrical carrier signals and telecommunication signals.

[0065] It should be noted that the device embodiments described above are merely illustrative. The units described as separate components may or may not be physically separate, and the components shown as units may or may not be physical units; that is, they may be located in one place or distributed across multiple network units. Some or all of the modules can be selected to achieve the purpose of this embodiment according to actual needs. Furthermore, in the accompanying drawings of the device embodiments provided by this invention, the connection relationships between modules indicate that they have communication connections, which can be specifically implemented as one or more communication buses or signal lines. Those skilled in the art can understand and implement this without any creative effort.

[0066] The specific embodiments described above further illustrate the purpose, technical solution, and beneficial effects of the present invention. It should be understood that the above descriptions are merely specific embodiments of the present invention and are not intended to limit the scope of protection of the present invention. In particular, it should be noted that any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of the present invention should be included within the scope of protection of the present invention for those skilled in the art.

Claims

1. A method for intelligent control of a medicinal herb air separator, characterized in that, include: Acquire the mass distribution data and density feature values ​​of medicinal materials, scan in real time to obtain the outline data and texture features of medicinal materials, analyze the mass distribution data and density feature values ​​to obtain a medicinal material characteristic dataset; Based on the medicinal material characteristic dataset, the medicinal materials are classified and identified to obtain the classified medicinal materials, and corresponding physical characteristic labels are generated; wherein, the physical characteristic labels include uniformity and roughness; Based on the classified medicinal materials and the physical property labels, the pre-established medicinal material air separation parameter mapping database is queried to obtain the equipment control parameter scheme required for medicinal material air separation; Based on the classified medicinal materials, the fan speed is dynamically adjusted to obtain a fan speed that matches the category of medicinal materials; Based on the medicinal material characteristic dataset, the duct angle and the position of the airflow distribution plate are adjusted using a servo motor to obtain the actual duct angle. The separation trajectory and separation effect data of medicinal materials are collected in real time. If the separation trajectory of medicinal materials deviates from the preset range, the actual air duct angle and the actual fan speed are corrected to obtain the optimal air duct angle and the optimal fan speed. The optimal air duct angle and the optimal fan speed are then updated to the pre-established medicinal material air separation parameter mapping database to obtain the final equipment control parameter scheme.

2. The intelligent control method for a medicinal herb air separator according to claim 1, characterized in that, The process involves acquiring the mass distribution data and density feature values ​​of medicinal materials, scanning the medicinal materials in real time for contour extraction and texture analysis, and analyzing the mass distribution data and density feature values ​​to obtain a medicinal material characteristic dataset, including: The first dataset is obtained by acquiring the mass distribution data and density characteristic values ​​of medicinal materials. The second dataset is obtained by scanning the medicinal materials in real time to extract contours and perform texture analysis. By merging the first data set and the second data set, and integrating multi-dimensional physical parameters, a third data set is generated; Compare the parameters in the third data set with a preset parameter threshold. If the parameter values ​​in the third data set meet the preset parameter threshold, then record the third data set to obtain the medicinal material characteristic dataset. If the parameter values ​​in the third data set do not meet the preset parameter threshold, the medicinal materials are re-collected, the first data set and the second data set are regenerated, and then merged back into the third data set. If the third data set still does not meet the preset parameter threshold after multiple re-collections, the medicinal materials are determined to be temporarily unsuitable for the air separation process.

3. The intelligent control method for a medicinal herb air separator according to claim 1, characterized in that, The step of classifying and identifying medicinal materials based on the medicinal material characteristic dataset to obtain classified medicinal materials and generating corresponding physical characteristic labels includes: Extract the acquired density feature values ​​from the medicinal material characteristic dataset. The density feature values ​​are density parameters directly obtained through the detection device. The density feature values ​​are quantized, and the mean and standard deviation of the density feature values ​​are calculated to obtain the quantized density feature values. If the quantified density characteristic value is less than the preset lightness threshold, the medicinal material is initially classified as a light medicinal material category. If the quantified density feature value is greater than the preset heavy threshold, the medicinal material is initially classified as a heavy medicinal material category. If the quantified density feature value is between the preset light threshold and the preset heavy threshold, the medicinal material is initially classified as a medium-quality medicinal material, and the classified medicinal material is obtained. By combining machine learning models, multidimensional features are extracted from the classified medicinal materials to obtain corresponding physical property labels.

4. The intelligent control method for a medicinal herb air separator according to claim 1, characterized in that, The step of querying a pre-established medicinal herb air-sorting parameter mapping database based on the classified medicinal herbs and the physical property labels to obtain the equipment control parameter scheme required for medicinal herb air-sorting includes: Based on the classified medicinal materials and the physical property labels, a pre-established medicinal material air separation parameter mapping database is queried to obtain a preliminary set of parameters for the fan speed and duct angle corresponding to the classified medicinal materials and the physical property labels. Based on the classified medicinal materials and the preliminary parameter set, the range of fan speed parameters and the range of air duct angle are determined, and the range of fan speed parameters and the range of air duct angle are combined to obtain the equipment control parameter scheme required for the air separation of medicinal materials.

5. The intelligent control method for a medicinal herb air separator according to claim 1, characterized in that, The step of dynamically adjusting the fan speed according to the classified medicinal materials to obtain a fan speed that matches the category of medicinal materials includes: Based on the classified medicinal materials, query the pre-established medicinal material air separation parameter mapping database to obtain the target rotation speed range and air separation airflow intensity parameters corresponding to the medicinal material classification results; Based on the classified medicinal materials, the airflow intensity parameters of the air sorting are dynamically adjusted to obtain the baseline airflow intensity; The matching degree between the actual airflow intensity and the benchmark airflow intensity during the air separation process is monitored in real time, and the fan speed is dynamically adjusted to correct the deviation, so as to obtain the actual fan speed that is accurately matched with the type of medicinal material.

6. The intelligent control method for a medicinal herb air separator according to claim 1, characterized in that, The step of adjusting the air duct angle and the position of the airflow distribution plate using a servo motor based on the medicinal material characteristic dataset to obtain the actual air duct angle includes: The density feature value is obtained from the medicinal material characteristic dataset, and the air duct tilt angle is calculated based on the density feature value to obtain the optimal air duct tilt angle. If the optimal duct tilt angle exceeds the preset duct tilt angle range, the duct is adjusted to the preset duct tilt angle range by a servo motor to obtain the adjusted duct angle. Based on the adjusted air duct angle, the position of the airflow distribution plate is adjusted by the servo motor to make the airflow form a flow field distribution pattern that conforms to the characteristics of the medicinal materials, thus obtaining the actual air duct angle.

7. The intelligent control method for a medicinal herb air separator according to claim 1, characterized in that, The system collects real-time data on the separation trajectory and effect of medicinal materials. If the separation trajectory deviates from a preset range, the actual air duct angle and actual fan speed are corrected to obtain the optimal air duct angle and optimal fan speed. The optimal air duct angle and optimal fan speed are then updated to the pre-established medicinal material air separation parameter mapping database to obtain the final equipment control parameter scheme, including: Real-time acquisition of separation trajectory data and separation effect data of medicinal materials, extraction of the position coordinates of medicinal materials, and obtaining the actual separation trajectory; The actual separation trajectory is compared with the preset trajectory range, and the trajectory deviation value is calculated; If the trajectory deviation value exceeds the preset trajectory deviation threshold, a feedback adjustment signal is generated to calculate the adjustment amount of the duct angle and the fan speed. The actual duct angle and the actual fan speed are corrected based on the adjustment amount to obtain the optimal duct angle and the optimal fan speed. The optimal parameter combination is obtained based on the optimal duct angle and the optimal fan speed. The optimal parameter combination is then updated to the pre-established medicinal herb air separation parameter mapping database to obtain the final equipment control parameter scheme.

8. An intelligent control system for a medicinal herb air separator, characterized in that, include: The data acquisition module acquires the mass distribution data and density characteristic values ​​of medicinal materials, scans in real time to acquire the outline data and texture features of medicinal materials, and analyzes the mass distribution data and density characteristic values ​​to obtain a medicinal material characteristic dataset. The classification and recognition module classifies and recognizes the medicinal materials based on the medicinal material characteristic dataset, obtains the classified medicinal materials, and generates corresponding physical characteristic labels; wherein, the physical characteristic labels include uniformity and roughness; The parameter matching module queries a pre-established medicinal herb air separation parameter mapping database based on the classified medicinal herbs and the physical property labels to obtain the equipment control parameter scheme required for medicinal herb air separation. The speed control module dynamically adjusts the fan speed according to the classified medicinal materials to obtain a fan speed that matches the category of medicinal materials; The flow field adjustment module uses a servo motor to adjust the air duct angle and the position of the airflow distribution plate based on the medicinal material characteristic dataset to obtain the actual air duct angle. The closed-loop monitoring module collects the separation trajectory and separation effect data of the medicinal materials in real time. If the separation trajectory of the medicinal materials deviates from the preset range, the actual air duct angle and the actual fan speed are corrected to obtain the optimal air duct angle and the optimal fan speed. The optimal air duct angle and the optimal fan speed are then updated to the pre-established medicinal material air separation parameter mapping database to obtain the final equipment control parameter scheme.

9. An electronic device, characterized in that, The device includes a processor, a memory, and a computer program stored in the memory and configured to be executed by the processor, wherein the processor executes the computer program to implement an intelligent control method for a medicinal herb wind separator as described in any one of claims 1 to 7.

10. A computer-readable storage medium, characterized in that, The computer-readable storage medium includes a stored computer program, wherein, when the computer program is executed, it controls the device containing the computer-readable storage medium to perform an intelligent control method for a medicinal herb wind separator as described in any one of claims 1 to 7.

Images