Screening method of macroporous resin for purifying ephedra root tannin

Abstract

The invention discloses a screening method of macroporous resin for purifying tannin in the root of ephedra sinica stapf, which comprises the following steps: loading the macroporous resin to be screened into a chromatographic column, and enabling an extracting solution of the root of ephedra sinica stapf to pass through the pretreated chromatographic column; washing the chromatographic column with deionized water until the absorbance value of effluent is stable, desorbing with an ethanol solution, and collecting a desorption solution; measuring the wet stacking density of the adsorbed resin; inputting the ephedra tannin adsorption quantity in unit time in an adsorption stage, the ephedra tannin recovery rate in a desorption stage and the wet stacking density rho into a first sub-model based on a linear regression algorithm, a second sub-model based on a support vector regression algorithm and a third sub-model based on a random forest algorithm; and adding the scores of the sub-models according to the dynamic weight to calculate a comprehensive score, and selecting the macroporous resin with the highest comprehensive score. According to the macroporous resin screening device, accurate screening of macroporous resin can be achieved.

Description

Technical Field

[0001] The present invention relates to the technical field related to the purification of tannins from ephedra roots, and more specifically, to a method for screening macroporous resins for purifying tannins from ephedra roots. Background Art

[0002] Macroporous resin chromatography technology is widely used in the separation of plant tannin components due to its advantages such as high adsorption selectivity and simple regeneration. Tannin compounds in Ephedra root have specific pharmacological activities, and their purification process places high demands on the performance of macroporous resins. However, the current macroporous resin screening method for the purification of Ephedra root tannins has many areas that need to be improved, resulting in the difficulty of the screened resins to balance adsorption efficiency, desorption stability and process adaptability in practical applications.

[0003] Traditional macroporous resin screening methods usually only focus on a single or a few performance indicators, such as static adsorption capacity and simple desorption recovery rate, but ignore the key parameters of resin physical properties and dynamic adsorption process. For example, the wet bulk density of the resin directly affects the fluid distribution and mass transfer efficiency in the chromatographic column. If the bulk density is unreasonable, it may lead to excessive pressure drop or uneven flow channels in the chromatographic column, thereby affecting the adsorption kinetics process; the specific surface area of ​​the resin is an important representation of the richness of adsorption sites, and its binding ability with tannin molecules lacks quantitative evaluation, which makes it difficult to accurately match the molecular characteristics of the target components during screening. In addition, parameters that reflect process stability, such as the stability of the desorption solution concentration (such as the coefficient of variation) and the dynamic adsorption mass transfer coefficient, are often ignored in traditional screening, resulting in large fluctuations in the composition of the desorption solution in actual production, increasing the complexity of subsequent purification steps. In terms of evaluation model construction, existing methods mostly use a single linear regression model, which can only handle simple linear relationships and is difficult to accurately describe the nonlinear coupling between the adsorption amount, desorption recovery rate and the physical and chemical properties of the resin, resulting in deviations between the screening results and the actual process performance.

[0004] Therefore, it is necessary to design a technical solution that can overcome the above-mentioned defects. Summary of the invention

[0005] One object of the present invention is to provide a method for screening macroporous resins for purifying ephedra root tannins, which can achieve accurate screening of macroporous resins.

[0006] To achieve these objects and other advantages of the present invention, according to one aspect of the present invention, the following steps are provided: S1: After loading the macroporous resin to be screened into a chromatography column, it is rinsed with an ethanol solution with a concentration of 60%-70% at a flow rate of 1.5-2.5 times the column volume per hour for 2.5-3.5 hours, and then rinsed with deionized water until the conductivity of the effluent is less than 15-25 μS / cm; the extract of Ephedra sinica roots is passed through the pretreated chromatography column at a flow rate of 1.0-2.0 times the column volume per hour, and the pH value of the sample loading solution is controlled to be 3.0-4.0. Adsorption is stopped when the concentration of Ephedra sinica root tannin in the effluent reaches 10%-20% of the initial concentration of the sample loading solution; S2: After rinsing the chromatography column with deionized water until the absorbance value at a wavelength of 280 nm of the effluent is stable at 0.03-0.08, desorption is carried out with an ethanol solution with a concentration of 70%-85% at a flow rate of 0.8-1.2 times the column volume per hour, and the volume of the desorbed solution collected is 2-4 times the column volume; S3: Measure the wet bulk density of the resin after adsorption. Transfer the resin to a graduated cylinder with a known mass, let it stand and settle for 20-40 minutes, record the volume of the resin and weigh the total mass. The wet bulk density is calculated by the formula (total mass - graduated cylinder mass) / resin volume; S4: Input the adsorption amount of Ephedra sinica root tannin per unit time in the adsorption stage, the recovery rate of Ephedra sinica root tannin in the desorption stage, and the measured value of the wet bulk density into the evaluation model, calculate the score, and select the macroporous resin with the highest score.

[0007] Furthermore, it also includes: measuring the specific surface area of the resin. After vacuum degassing the pretreated resin, the specific surface area is measured by the nitrogen adsorption method; measuring the coefficient of variation of the desorbed solution concentration. The tannin concentration of the desorbed solution is detected by a liquid chromatograph, and the coefficient of variation is calculated as the percentage of the standard deviation to the average value; measuring the dynamic adsorption mass transfer coefficient. According to the adsorption breakthrough curve data, the mass transfer coefficient is obtained by fitting calculation using the adsorption kinetics model; input the adsorption amount of Ephedra sinica root tannin per unit time in the adsorption stage, the recovery rate of Ephedra sinica root tannin in the desorption stage, the measured value of the wet bulk density, the specific surface area of the resin, the coefficient of variation of the desorbed solution concentration, and the dynamic adsorption mass transfer coefficient into the evaluation model, and calculate the said score.

[0008] Furthermore, the first sub-model is based on the linear regression algorithm. The input parameters are the adsorption amount of Ephedra root tannin per unit time in the adsorption stage, the recovery rate of Ephedra root tannin in the desorption stage, and the wet bulk density of the resin. The first sub-model score is output. The second sub-model is based on the support vector regression algorithm. The input parameters include the adsorption amount of Ephedra root tannin per unit time in the adsorption stage, the coefficient of variation of the desorption liquid concentration, and the dynamic adsorption mass transfer coefficient. After mapping the non-linear relationship through the kernel function, the second sub-model score is output. The third sub-model is based on the random forest algorithm. The input parameters include the adsorption amount of Ephedra root tannin per unit time in the adsorption stage, the recovery rate of Ephedra root tannin in the desorption stage, the wet bulk density of the resin, the specific surface area of the resin, and the coefficient of variation of the desorption liquid concentration. After parallel training by multiple decision trees, the third sub-model score is output. According to the historical prediction errors of each sub-model on the validation data set, the weight ratios of the first sub-model, the second sub-model, and the third sub-model are dynamically allocated, and the scores of each sub-model are added according to the dynamic weights to generate the final comprehensive score.

[0009] Furthermore, the kernel function of the second sub-model is the Gaussian kernel, the kernel width σ ∈ [0.25, 0.35], the regularization parameter C ∈ [1.5, 2.5], and the insensitive loss parameter ε ≤ 0.03. The number of decision trees of the third sub-model n_trees ∈ [80, 120], the maximum depth of a single tree max_depth ∈ [12, 18], and the minimum number of samples for node splitting min_samples_split ≥ 15.

[0010] Furthermore, when the dynamic adsorption mass transfer coefficient K ≥ 0.45 cm / s, the weight W2 of the second sub-model is increased to 50% - 60%. When the coefficient of variation of the desorption liquid concentration CV ≤ 12%, the weight W3 of the third sub-model is decreased to 15% - 25%. When the specific surface area of the resin S ≥ 400 m² / g and the wet bulk density ρ ≤ 1.15 g / cm³, the weight W1 of the first sub-model is fixed at 20%, and the weights W2 of the second sub-model and W3 of the third sub-model are allocated the remaining weights in a ratio of 2:3. The adjustment step of the weights of each sub-model does not exceed ±5% / minute, and the total weight allocation error is controlled within ±1.5%.

[0011] Furthermore, when the temperature T of the chromatography column ∈ [25°C, 35°C], the weight W2 of the second sub-model is corrected according to the formula: and the total weight is kept at 100% through normalization processing.

[0012] Furthermore, when it is detected that the pH value fluctuation ΔpH > 0.3 in the adsorption stage, the fault tolerance processing is activated: the output value of the first sub-model is multiplied by the attenuation factor the kernel function of the second sub-model is switched to the polynomial kernel, and the decision tree depth of the third sub-model is temporarily reduced to 10 layers.

[0013] Further, it also includes: selecting the two macroporous resins with the highest scores for secondary verification, repeating the adsorption-desorption process three times under the same operating conditions, calculating the standard deviation of the tannin purity of Ephedra sinica roots in the three experiments, and retaining the resins with a standard deviation less than 1.0% - 2.0% as the final screening results.

[0014] Further, each cycle needs to meet the following requirements: the temperature control accuracy of the chromatography column is ±1°C, and the fluctuation range of the desorbent flow rate is ≤ ±5% of the set value; detecting the tannin purity of Ephedra sinica roots obtained in each cycle, calculating the standard deviation σ of the three purities. If σ ∈ [1.0%, 2.0%], it is determined that the resin is qualified; when both resins are qualified, select the one with a smaller σ value as the final screening result; if only one is qualified, directly select it; if both are unqualified, re-evaluate and exclude the top two resins with the highest original scores.

[0015] Further, the selected resins are treated under accelerated aging conditions, and the conditions are: temperature 40 ± 2°C, relative humidity 75 ± 5%, lasting for 72 - 120 hours; repeating the adsorption-desorption process after aging treatment, and calculating the attenuation rate ΔQ of the tannin adsorption amount, ; where Q0 is the adsorption amount before aging and Q1 is the adsorption amount after aging; retain the resins with ΔQ ≤ 5.0%.

[0016] The present invention has at least the following beneficial effects: The present invention incorporates the wet bulk density, specific surface area, coefficient of variation of desorbent concentration, and dynamic adsorption mass transfer coefficient into the screening indicators, accurately reflecting the fluid distribution ability, richness of adsorption sites, desorption uniformity, and mass transfer efficiency of the resin in the actual chromatography process, avoiding screening deviations caused by single indicators, enabling the resin performance to be deeply matched with the requirements of the tannin purification process, and greatly improving the accuracy of macroporous resin screening. Using linear regression, support vector regression, and random forest algorithms to construct three-level sub-models to separately handle linear relationships, non-linear kinetic characteristics, and multi-variable complex coupling effects, and through a dynamic weight allocation mechanism to respond to changes in working conditions such as temperature and pH in real time, significantly improving the fitting accuracy of the model for complex adsorption-desorption processes. The secondary verification link screens resins with qualified standard deviations through repetitive experiments, combined with accelerated aging tests to control the attenuation rate of adsorption amount, ensuring that the selected resins maintain stable performance during long-term use and reducing the process fluctuation risk in industrial production.

[0017] Other advantages, objectives, and features of the present invention will be partially reflected by the following description and partially understood by those skilled in the art through research and practice of the present invention. Detailed Embodiments

[0018] The following further elaborates on the present invention in conjunction with embodiments, so that those skilled in the art can implement it according to the description in the specification.

[0019] It should be understood that terms such as "having", "including", and "comprising" used in the embodiments of the present application do not exclude the presence or addition of one or more other elements or combinations thereof. All directional indications (such as up, down, left, right, front, back...) in the embodiments of the present application are only used to explain the relative positional relationship, movement conditions, etc. between components in a specific posture. If the specific posture changes, the directional indications will also change accordingly. When an element is referred to as being "fixed to" or "disposed on" another element, it can be directly on the other element or there may be an intermediate element present at the same time. When an element is referred to as being "connected" to another element, it can be directly connected to the other element or it can also be indirectly connected to the other element through an intermediate element. The descriptions in the embodiments of the present application involving "first", "second", etc. are only for descriptive purposes and cannot be understood as indicating or implying their relative importance or implicitly specifying the quantity of the indicated technical features. Thus, features defined with "first", "second" may explicitly or implicitly include at least one of such features.

[0020] It should be noted that the technical solutions between the various embodiments of the present application can be combined with each other, but it must be based on the fact that those of ordinary skill in the art can implement them. When the combination of technical solutions results in contradictions or cannot be implemented, it should be considered that such a combination of technical solutions does not exist and is not within the scope of protection required by the present application.

[0021] The embodiments of the present application provide a screening method for macroporous resins for the purification of Ephedra sinica roots tannins, including: S1: After loading the macroporous resin to be screened into the chromatography column, rinse it with an ethanol solution with a concentration of 60%-70% at a flow rate of 1.5-2.5 times the column volume per hour for 2.5-3.5 hours, and then rinse it with deionized water until the conductivity of the effluent is less than 15-25 μS / cm; pass the Ephedra sinica root extract through the pretreated chromatography column at a flow rate of 1.0-2.0 times the column volume per hour, control the pH value of the sample loading solution to be 3.0-4.0, and stop adsorption when the concentration of Ephedra sinica root tannin in the effluent reaches 10%-20% of the initial concentration of the sample loading solution; S2: After rinsing the chromatography column with deionized water until the absorbance value of the effluent at a wavelength of 280 nm stabilizes at 0.03-0.08, perform desorption with an ethanol solution with a concentration of 70%-85% at a flow rate of 0.8-1.2 times the column volume per hour, and collect the desorbed solution with a volume of 2-4 times the column volume; S3: Measure the wet bulk density of the resin after adsorption. Transfer the resin to a graduated cylinder with a known mass, let it stand and settle for 20-40 minutes, record the volume of the resin and weigh the total mass. The wet bulk density is calculated by the formula (total mass - graduated cylinder mass) / resin volume; S4: Input the Ephedra sinica root tannin adsorption amount per unit time in the adsorption stage, the Ephedra sinica root tannin recovery rate in the desorption stage, and the measured value of the wet bulk density into the evaluation model, calculate the score, and select the macroporous resin with the highest score.

[0022] Exemplarily, the ethanol solution concentration may be selected as 60%, 65%, 70%, the rinse flow rate may be selected as 1.5 times, 2.0 times, 2.5 times the column volume per hour, the rinse time may be selected as 2.5 hours, 3.0 hours, 3.5 hours, the effluent conductivity control target may be selected as 15 μS / cm, 20 μS / cm, 25 μS / cm, the sample loading flow rate may be selected as 1.0 times, 1.5 times, 2.0 times the column volume per hour, the pH value of the sample loading solution may be selected as 3.0, 3.5, 4.0, the effluent concentration threshold for stopping adsorption may be selected as 10%, 15%, 20%, the ethanol concentration for desorption may be selected as 70%, 75%, 80%, 85%, the desorption flow rate may be selected as 0.8 times, 1.0 times, 1.2 times the column volume per hour, the collected volume of the desorbed solution may be selected as 2 times, 3 times, 4 times the column volume, and the standing settlement time may be selected as 20 minutes, 30 minutes, 40 minutes. The equipment that can be used includes glass chromatography columns, conductivity meters (such as DDS-307 type), ultraviolet-visible spectrophotometers (such as UV-2550 type), electronic balances, and graduated cylinders. In terms of materials, macroporous resins can be selected from commercially available products such as AB-8 type, D101 type, X-5 type, etc. When calculating the wet bulk density, first weigh the graduated cylinder with a known mass, add the resin and weigh again to get the total mass. After standing and settling, read the volume of the resin, and calculate by the formula (total mass - graduated cylinder mass) / resin volume.

[0023] The following data were obtained through experiments on three macroporous resins A, B, and C: Resin A: The tannin adsorption amount per unit time in the adsorption stage is 80 mg / h, the tannin recovery rate in the desorption stage is 85%, and the wet bulk density is 1.15 g / cm³; Resin B: The tannin adsorption amount per unit time in the adsorption stage is 75 mg / h, the tannin recovery rate in the desorption stage is 90%, and the wet bulk density is 1.20 g / cm³. If the evaluation model uses the linear weighted method, and the weights of the adsorption amount, recovery rate, and bulk density are set to 0.4, 0.3, and 0.3 respectively, and each parameter is transformed into 0 - 100 points in a normalized manner (for example, the full score of the adsorption amount is 100 mg / h, 0 points for 50 mg / h; the full score of the recovery rate is 100%, 0 points for 70%; the bulk density takes 1.2 g / cm³ as the central value, and the score is lower the farther it deviates), then the calculation process is as follows: For Resin A: The score of the adsorption amount is 80 points, the score of the recovery rate is 85 points, the score of the bulk density is 90 points, and the total score is 80×0.4 + 85×0.3 + 90×0.3 = 84.5 points; For Resin B: The score of the adsorption amount is 75 points, the score of the recovery rate is 90 points, the score of the bulk density is 80 points, and the total score is 75×0.4 + 90×0.3 + 80×0.3 = 81 points. During use, first load the macroporous resin into the chromatography column, pre-treat it with an ethanol solution of a specific concentration and deionized water, then let the extract of Ephedra sinica roots pass through the chromatography column at a specified flow rate, control the pH value and stop adsorption at an appropriate concentration, then rinse with deionized water and perform desorption, collect the desorption solution, then measure the wet bulk density of the resin, and finally input the relevant data into the evaluation model to calculate the score for resin screening. This method can provide a quantitative basis for the screening of macroporous resins through standardized pre-treatment, adsorption, desorption, and measurement steps, which helps to improve the scientificity and accuracy of the screening process.

[0024] In another embodiment, it further includes: measuring the specific surface area of the resin. After vacuum degassing the pre-treated resin, the specific surface area is measured by the nitrogen adsorption method; measuring the coefficient of variation of the desorption solution concentration. The tannin concentration of the desorption solution is detected by a liquid chromatograph, and the coefficient of variation is calculated as the percentage of the standard deviation to the average value; measuring the dynamic adsorption mass transfer coefficient. According to the adsorption breakthrough curve data, the mass transfer coefficient is obtained by fitting calculation using the adsorption kinetic model; inputting the tannin adsorption amount of Ephedra sinica roots per unit time in the adsorption stage, the tannin recovery rate of Ephedra sinica roots in the desorption stage, the measured value of the wet bulk density, the specific surface area of the resin, the coefficient of variation of the desorption solution concentration, and the dynamic adsorption mass transfer coefficient into the evaluation model to calculate the said score.

[0025] Exemplarily, for the determination of the specific surface area of the resin, the pretreated resin needs to be vacuum degassed, and the nitrogen adsorption method can be used for determination with a specific surface area analyzer (such as Micromeritics ASAP 2020). The concentration of the desorbing solution can be detected using a high-performance liquid chromatograph (such as Agilent 1260 Infinity), and the coefficient of variation is calculated as the percentage of the standard deviation to the average value. The determination of the dynamic adsorption mass transfer coefficient is based on the adsorption breakthrough curve, and the adsorption kinetic model used can be the pseudo-first-order kinetic model or the pseudo-second-order kinetic model. Specifically, during the adsorption process, the change in the tannin concentration of the effluent needs to be recorded synchronously to form an "adsorption breakthrough curve" (i.e., the curve of the effluent concentration changing with time or the effluent volume). During the specific operation, the extract of Ephedra sinica roots is passed through the chromatography column at a fixed flow rate. Starting from the beginning of adsorption, the tannin concentration of the effluent is detected at regular intervals until the concentration approaches the initial concentration of the sample solution and then stopped, obtaining a curve that first rises slowly and then rapidly approaches the initial concentration. To calculate the mass transfer coefficient, a suitable adsorption kinetic model (such as a model describing the relationship between the diffusion rate of substances and the adsorption process) needs to be selected to fit the breakthrough curve. In actual operation, the concentration data measured in the experiment is input into the model using data analysis software (such as Origin, MATLAB), and the software will automatically adjust the model parameters to make the model curve as close as possible to the experimental data. The key parameter directly related to the mass transfer rate obtained after adjustment is the dynamic adsorption mass transfer coefficient.

[0026] When used as a whole, on the basis of the previous embodiment, the determination of the specific surface area of the resin, the coefficient of variation of the desorbing solution concentration, and the dynamic adsorption mass transfer coefficient are added, and these parameters are input into the evaluation model together with the previous adsorption capacity, recovery rate, and wet bulk density to calculate the score. By introducing more parameters characterizing the resin performance, the applicability of the macroporous resin in the purification of tannins from Ephedra sinica roots can be evaluated more comprehensively, providing a richer basis for screening and helping to select a resin with better comprehensive performance.

[0027] In another embodiment, the first sub-model is based on a linear regression algorithm. The input parameters are the adsorption amount of Ephedra sinica root tannin per unit time in the adsorption stage, the recovery rate of Ephedra sinica root tannin in the desorption stage, and the wet bulk density of the resin, and the first sub-model score is output. The second sub-model is based on a support vector regression algorithm. The input parameters include the adsorption amount of Ephedra sinica root tannin per unit time in the adsorption stage, the coefficient of variation of the desorbing solution concentration, and the dynamic adsorption mass transfer coefficient. After mapping the non-linear relationship through a kernel function, the second sub-model score is output. The third sub-model is based on a random forest algorithm. The input parameters include the adsorption amount of Ephedra sinica root tannin per unit time in the adsorption stage, the recovery rate of Ephedra sinica root tannin in the desorption stage, the wet bulk density of the resin, the specific surface area of the resin, and the coefficient of variation of the desorbing solution concentration. After parallel training by multiple decision trees, the third sub-model score is output. According to the historical prediction errors of each sub-model on the validation data set, the weight ratios of the first sub-model, the second sub-model, and the third sub-model are dynamically allocated, and the scores of each sub-model are added according to the dynamic weights to generate the final comprehensive score.

[0028] That is, the first sub-model uses a linear regression algorithm to process the three parameters of adsorption amount, recovery rate, and wet bulk density. The second sub-model is based on a support vector regression algorithm, and the kernel function can be selected as a Gaussian kernel, etc. The input parameters are the adsorption amount, the coefficient of variation, and the mass transfer coefficient. The third sub-model is based on a random forest algorithm, and the input parameters include the adsorption amount, the recovery rate, the wet bulk density, the specific surface area, and the coefficient of variation, and is trained in parallel by multiple decision trees. Data processing can be implemented on a computer using programming languages such as Python, with the help of machine learning libraries such as Scikit-learn.

[0029] Three sub-models with different algorithms are constructed, and the corresponding parameters are input for scoring respectively. Then, according to the historical prediction errors, the weights are dynamically allocated, and the scores of each sub-model are weighted to obtain the final comprehensive score. This way of combining multiple models can make full use of the advantages of different algorithms, consider the linear and non-linear relationships among multiple parameters, improve the accuracy and robustness of the evaluation model, and make the screening results more reliable.

[0030] In another embodiment, the kernel function of the second sub-model is a Gaussian kernel, the kernel width σ ∈ [0.25, 0.35], the regularization parameter C ∈ [1.5, 2.5], and the insensitive loss parameter ε ≤ 0.03; the number of decision trees of the third sub-model n_trees ∈ [80, 120], the maximum depth of a single tree max_depth ∈ [12, 18], and the minimum number of samples for node splitting min_samples_split ≥ 15.

[0031] Exemplarily, the Gaussian kernel width σ of the second sub-model may be selected as 0.25, 0.30, 0.35, the regularization parameter C may be selected as 1.5, 2.0, 2.5, and the insensitive loss parameter ε may take values of 0.01, 0.02, 0.03. The number of decision trees n_trees of the third sub-model may be selected as 80, 100, 120, the maximum depth max_depth of a single tree may be selected as 12, 15, 18, and the minimum number of samples min_samples_split for node splitting may be a value ≥15 such as 15, 20, 25, etc. These parameter settings can be configured in a machine learning library, such as the SVR and RandomForestRegressor modules in Scikit-learn.

[0032] The Gaussian kernel support vector regression algorithm with specific parameters is adopted in the second sub-model, and appropriate random forest parameters are set in the third sub-model. Through the optimization of these parameters, the sub-models can better handle the non-linear relationships and complex features of the input data, improve the prediction accuracy of the sub-models, and thus enhance the accuracy of the final comprehensive score, which helps to more accurately screen the macroporous resin.

[0033] In another embodiment, when the dynamic adsorption mass transfer coefficient K ≥ 0.45 cm / s, the weight W2 of the second sub-model is increased to 50% - 60%; when the coefficient of variation CV of the desorbing liquid concentration ≤ 12%, the weight W3 of the third sub-model is decreased to 15% - 25%; when the resin specific surface area S ≥ 400 m² / g and the wet bulk density ρ ≤ 1.15 g / cm³, the weight W1 of the first sub-model is fixed at 20%, and the remaining weights of the second sub-model's weight W2 and the third sub-model's weight W3 are distributed in a ratio of 2:3; the adjustment step size of each sub-model's weight does not exceed ±5% / minute, and the total weight distribution error is controlled within ±1.5%.

[0034] Exemplarily, the threshold of the dynamic adsorption mass transfer coefficient K is 0.45 cm / s. When this value is reached, the weight W2 of the second sub-model is between 50% - 60%, and may be 50%, 55%, 60%. The threshold of the coefficient of variation CV of the desorbing liquid concentration is 12%. When it is lower than this value, the weight W3 of the third sub-model is between 15% - 25%, and may be 15%, 20%, 25%. When the resin specific surface area S ≥ 400 m² / g and the wet bulk density ρ ≤ 1.15 g / cm³, the weight W1 of the first sub-model is fixed at 20%, and the remaining 80% is distributed according to W2:W3 = 2:3, that is, W2 is 32%, W3 is 48%, etc. The weight adjustment step size does not exceed ±5% / minute, and the total weight error is controlled within ±1.5%. These condition judgments and weight adjustments can be implemented through programming in a data processing system.

[0035] According to the actual measured values of different parameters, the weights of each sub-model are dynamically adjusted according to preset rules, so that the weight distribution is more in line with the performance characteristics of the resin. For example, when the mass transfer coefficient is high, the second sub-model is emphasized; when the coefficient of variation is low, the weight of the third sub-model is reduced; when the specific surface area and bulk density are specific, the weights are allocated in a fixed ratio, ensuring that the comprehensive score can better highlight the current key performance indicators and improve the pertinence and rationality of screening.

[0036] In another embodiment, when the chromatography column temperature T ∈ [25°C, 35°C], the weight W2 of the second sub-model is corrected according to the formula: And the total weight is kept at 100% through normalization.

[0037] When the chromatography column temperature T is between 25°C and 35°C, the reference temperature in the correction formula is 30°C. For every 1°C deviation from the reference temperature, the weight of the second sub-model is adjusted by a coefficient of 0.02. For example, when T = 25°C, W2′ = W2 × (1 - 0.1) = 0.9W2; when T = 35°C, W2′ = W2 × (1 + 0.1) = 1.1W2. After the weight adjustment, normalization is performed to ensure that the total weight of the three sub-models is always 100% and weight imbalance is avoided. Temperature monitoring can use a thermocouple or a temperature sensor (such as PT100 type), which is connected to the data processing system to obtain temperature data in real time and perform weight correction calculations.

[0038] When the chromatography column temperature changes, the weight of the second sub-model is linearly corrected according to the temperature. Considering that the temperature may affect the adsorption and desorption processes, and thus affect related parameters such as the mass transfer coefficient, through this correction, the weight distribution is more in line with the actual working conditions, improving the adaptability of the evaluation model to temperature changes and ensuring the reliability of the screening results under different temperature conditions.

[0039] In another embodiment, when it is detected that the pH value fluctuation ΔpH > 0.3 during the adsorption stage, the fault tolerance processing is activated: the output value of the first sub-model is multiplied by the attenuation factor The kernel function of the second sub-model is switched to a polynomial kernel, and the decision tree depth of the third sub-model is temporarily reduced to 10 layers.

[0040] The threshold value of the pH value fluctuation ΔpH is 0.3, and when it exceeds this value, the fault tolerance processing is activated. The attenuation factor is For example, when ΔpH = 0.4, the attenuation factor is approximately . The kernel function of the second sub-model is switched from a Gaussian kernel to a polynomial kernel, and the degree of the polynomial kernel can remain default or be set according to the situation. The decision tree depth of the third sub-model is temporarily reduced from the original 12 - 18 layers to 10 layers to reduce the model complexity. The pH value can be detected using a pH meter (such as S40 type) to monitor the pH value during the adsorption stage in real time. When the fluctuation exceeds the threshold value, the data processing system triggers the corresponding fault tolerance mechanism.

[0041] When the pH value fluctuates greatly during the adsorption stage, it may affect the stability of the adsorption effect. By attenuating the output of the first sub-model, switching the kernel function of the second sub-model, and reducing the decision tree depth of the third sub-model, the evaluation model can maintain a certain robustness under abnormal conditions, reduce the adverse impact of pH fluctuations on the screening results, and improve the fault tolerance and reliability of the method.

[0042] In another embodiment, it further includes: selecting the two macroporous resins with the highest scores for secondary verification, repeating the adsorption-desorption process three times under the same operating conditions, calculating the standard deviation of the tannin purity of Ephedra sinica roots in the three experiments, and retaining the resins with a standard deviation less than 1.0% - 2.0% as the final screening results.

[0043] For the secondary verification, select the two resins with the highest scores, repeat the experiment three times under the same conditions, calculate the standard deviation of the purity, and the standard deviation threshold is less than 1.0%, 1.5%, 2.0%, etc. The experimental conditions need to be the same as before, including column parameters, solution concentration, flow rate, pH value, etc. Purity detection can use a high-performance liquid chromatograph or other suitable analytical methods to ensure the accuracy of the detection results.

[0044] After initially screening out the two resins with the highest scores, repeat the experiment through secondary verification, calculate the standard deviation of the purity, and retain the resins with good stability (small standard deviation). This step can further test the repeatability and stability of the resins in actual operation, avoid screening errors caused by accidental factors, ensure the reliable performance of the finally selected resins in actual applications, and improve the practicality of the screening results.

[0045] In another embodiment, each cycle needs to meet: the temperature control accuracy of the chromatography column is ±1°C, and the fluctuation range of the desorbing solution flow rate is ≤ ±5% of the set value; detect the tannin purity of Ephedra sinica roots obtained in each cycle, calculate the standard deviation σ of the three purities. If σ ∈ [1.0%, 2.0%], then determine that the resin is qualified; when both resins are qualified, select the one with a smaller σ value as the final screening result; if only one is qualified, directly select it; if both are unqualified, re-score and exclude the top two resins with the highest original scores.

[0046] The temperature control accuracy of the chromatography column is ±1°C. A chromatography column with a temperature control device (such as a constant temperature chromatography column) can be used. The fluctuation range of the desorbing solution flow rate is ≤ ±5% of the set value, and the flow rate is controlled by equipment such as a peristaltic pump (such as BT100 - 2J type). After purity detection, calculate the standard deviation σ of the three times, and the qualified range is from 1.0% to 2.0%. When both resins are qualified, select the one with a smaller σ. If only one is qualified, select that one. If both are unqualified, re-score and exclude the top two.

[0047] In the secondary verification, strictly control the experimental conditions to ensure the stability of temperature and flow rate, and judge the qualification and quality of the resin by calculating the standard deviation of purity. Such strict verification and screening rules can further ensure that the finally selected resin has a stable purification effect under the same conditions, improve the reliability and effectiveness of the screening results, and provide a more stable resin selection for actual production applications.

[0048] In another embodiment, the screened resin is treated under accelerated aging conditions, and the conditions are: temperature 40 ± 2 °C, relative humidity 75 ± 5%, for 72 - 120 hours; after the aging treatment, repeat the adsorption - desorption process, and calculate the attenuation rate ΔQ of the tannin adsorption amount before and after aging. ; where Q0 is the adsorption amount before aging and Q1 is the adsorption amount after aging; retain the resin with ΔQ ≤ 5.0%.

[0049] The accelerated aging conditions are temperature 40 ± 2 °C, relative humidity 75 ± 5%, and the duration is 72 hours, 96 hours, 120 hours, etc. The aging treatment can be carried out in a constant temperature and humidity chamber (such as LHS - 100CH type). After aging, repeat the adsorption - desorption process, calculate the attenuation rate ΔQ, and the threshold is ≤ 5.0%. When calculating, first measure the adsorption amount Q0 before aging and the adsorption amount Q1 after aging, and calculate the attenuation rate through the formula.

[0050] Carry out accelerated aging treatment on the screened resin, simulate the aging situation in actual use, detect the attenuation rate of the adsorption amount before and after aging, and retain the resin with a low attenuation rate. This step can evaluate the durability and stability of the resin, ensure that the selected resin has good performance during long - term use, reduce the problem of the decline in purification effect caused by resin aging, and improve the durability and reliability of the screening results in actual applications.

[0051] In another embodiment, the screening method includes: Select commercially available macroporous resins of types X, Y, and Z, and respectively load them into glass chromatography columns with an inner diameter of 2.5 cm and a height of 50 cm (the column volume is about 245 mL). First, rinse with a 65% ethanol solution at a flow rate of 2.0 times the column volume per hour (490 mL / h) for 3.0 hours, and then rinse with deionized water until the conductivity of the effluent is 18 μS / cm. Load the extract of Ephedra sinica roots with a pH of 3.5 at a flow rate of 1.5 times the column volume per hour (367.5 mL / h), and stop adsorption when the tannin concentration in the effluent reaches 15% of the initial concentration of the loading solution. Rinse with deionized water until the absorbance at 280 nm is stable at 0.05, and then desorb with an 80% ethanol solution at a flow rate of 1.0 times the column volume per hour (245 mL / h), and collect 3 times the column volume (735 mL) of the desorbing solution.

[0052] Transfer the adsorbed X-type resin to a 100 mL graduated cylinder with a known mass (50 g), let it stand for 30 minutes, record the resin volume of 80 mL, weigh the total mass of 135 g, and calculate the density = (135 - 50) / 80 = 1.06 g / cm³.

[0053] The pretreated X-type resin is vacuum degassed and measured with a Micromeritics ASAP 2020 specific surface area analyzer. The specific surface area is 450 m 2 / g. Use an Agilent 1260 high performance liquid chromatography to detect the tannin concentration in the desorption solution of the X-type resin. The three repeated measurement values are 25.1 mg / mL, 24.5 mg / mL, and 25.8 mg / mL, with an average value of 25.13 mg / mL, a standard deviation of 0.65, and a coefficient of variation = 0.65 / 25.13×100% = 2.59%. Based on the breakthrough curve data of the X-type resin adsorption, use Origin software to fit the liquid film diffusion model and obtain a mass transfer coefficient K = 0.48 cm / s. The first sub-model (linear regression): Input the adsorption capacity (85 mg / h), recovery rate (88%), and bulk density (1.06 g / cm³) of the X-type resin. The first sub-model can assume the model form: Score = a×adsorption capacity + b×recovery rate + c × bulk density + d, where a, b, c, d are coefficients obtained by least squares fitting. After inputting the specific values, calculate the score of the first sub-model according to the fitting equation (which needs to be combined with normalization or standardization processing, such as converting the parameters to 0 - 100 points and then weighting).

[0054] The second sub-model (support vector regression, Gaussian kernel σ = 0.3, C = 2.0, ε = 0.02): Input the adsorption capacity, coefficient of variation (2.59%), and mass transfer coefficient (0.48 cm / s). Kernel function and parameters: Use the Gaussian kernel (RBF kernel), and the parameters are: kernel width (sigma = 0.3), regularization parameter (C = 2.0), and insensitive loss parameter (varepsilon = 0.02). Map the input data to a high-dimensional space through the kernel function to construct a non-linear regression model. After the input parameters are mapped by the kernel function, output the score of the second sub-model through the trained SVR model.

[0055] The third sub-model (random forest, n_trees = 100, max_depth = 15, min_samples_split = 20): Input all the above 6 parameters (adsorption capacity, recovery rate, bulk density, specific surface area, coefficient of variation, mass transfer coefficient). The algorithm initializes and generates a specified number of decision trees (controlled by n_trees). For example, setting it to 100 means generating 100 different trees. The training data for each tree is sampled with replacement (Bootstrap) from the original dataset to form a differentiated sub-dataset. The samples not drawn are called out-of-bag data (OOB), which can be used for subsequent verification. Next, for the construction process of each tree, the algorithm recursively splits starting from the root node: At each node, check whether the number of samples at the current node meets the requirement of min_samples_split (for example, setting it to 20 means that at least 20 samples are required in the node to allow splitting). If not, stop splitting, directly generate a leaf node and give the prediction result; if so, randomly select a subset of all features (feature randomness), traverse these features and their possible thresholds, and select the feature and threshold that can most reduce the impurity (such as Gini index or information gain) for splitting. The split sub-nodes repeat this process until the maximum depth set by max_depth (for example, a depth of 5 means that there are at most 5 layers of splitting from the root node to the leaf node) or no further splitting is possible. This splitting mechanism makes the complexity of each tree limited by the depth (max_depth to prevent overfitting) and avoids overfitting to extremely small groups through the sample quantity threshold (min_samples_split). When all the trees are constructed, the random forest outputs the final result through an ensemble strategy: taking the average of the prediction values of all the trees. After inputting the 6 parameters, the score of the third sub-model is output through the random forest model.

[0056] The calculation processes of X-type resins W1, W2, and W3 are as follows: 1. Initial weight assignment conditions When the dynamic adsorption mass transfer coefficient (K≥0.45 cm / s), the weight W2 of the second sub-model is increased to 50% - 60%. For X-type resin, K = 0.48 which is greater than 0.45, so the initial W2 = 55% (within the range of 50% - 60%).

[0057] When the specific surface area of the resin (S≥400 m² / g) and the wet bulk density ρ≤1.15 g / cm³, the weight W1 of the first sub-model is fixed at 20%, and the remaining weights are allocated according to W2:W3 = 2:3.

[0058] The X-type resin has S = 450 m² / g and ρ = 1.06 g / cm³, meeting the conditions. Therefore, W1 = 20%, and the remaining 80% originally needed to be distributed as W2 = 32% and W3 = 48% according to a ratio of 2:3. However, since K ≥ 0.45 cm / s is triggered first, W2 is increased to 55%.

[0059] 2. Temperature correction formula When the chromatography column temperature T ∈ [25°C, 35°C], the weight of the second sub-model is corrected to ; For the X-type resin, T = 28°C. Substituting it in, we get W2' = 55% × [1 + 0.02 × (28 - 30)] = 52.8%.

[0060] 3. Weight normalization processing Since W1 + W2' + W3 = 100% Combined with the above text, so W1 = 20%, W2' = 52.8%, and W3 = 27.2%. The original W1 = 22% and W3 = 25.2% were fine-tuned by considering historical prediction errors simultaneously.

[0061] Select the top two scoring X-type and Y-type resins, and repeat the adsorption-desorption three times under the same conditions (temperature 30 ± 1°C, flow rate fluctuation ≤ 5%) to measure the tannin purity: The purities of the X-type resin for three times are 92.1%, 91.8%, and 92.4%, with a standard deviation of 0.26%; The purities of the Y-type resin for three times are 89.5%, 90.2%, and 88.9%, with a standard deviation of 0.65%.

[0062] Since the standard deviations of both are ≤ 2.0%, select the X-type resin with a smaller standard deviation for the next step. Place the X-type resin in a constant temperature and humidity box at 40 ± 2°C and relative humidity 75 ± 5% for 96 hours. After aging, repeat the screening process. Measure the adsorption capacity Q0 = 90 mg / g before aging and Q1 = 86 mg / g after aging. The attenuation rate ΔQ = (90 - 86) / 90 ≈ 4.4% ≤ 5.0%. Finally, determine the X-type resin as the screening result.

[0063] Although the embodiments of the present invention have been disclosed above, it is not limited to only the applications listed in the specification and embodiments. It can be fully applied to various fields suitable for the present invention. For those familiar with the field, additional modifications can be easily achieved. Therefore, without departing from the general concept defined by the claims and the equivalent scope, the present invention is not limited to specific details and the embodiments shown and described herein.

Claims

1. A method for screening macroporous resins for purifying tannins from Ephedra sinica roots, characterized in that: include: S1: The macroporous resin to be screened is loaded into a chromatography column, and washed with an ethanol solution having a concentration of 60%-70% at a flow rate of 1.5-2.5 column volumes per hour for 2.5-3.5 hours, and then washed with deionized water until the conductivity of the effluent is less than 15-25 μS / cm; the ephedra root extract is passed through the chromatography column at a flow rate of 1.0-2.0 column volumes per hour, and the pH value of the loading solution is controlled to be 3.0-4.0, and the adsorption is stopped when the ephedra root tannin concentration of the effluent reaches 10%-20% of the initial concentration of the loading solution; S2: rinse the column with deionized water until the absorbance value of the effluent at a wavelength of 280 nm is stabilized at 0.03-0.08, then desorb with a 70%-85% ethanol solution at a flow rate of 0.8-1.2 column volumes per hour, and collect a desorbed liquid with a volume of 2-4 column volumes; S3: Determine the wet bulk density ρ of the macroporous resin after adsorption. Transfer the macroporous resin to a measuring cylinder and let it settle for 20-40 minutes. Record the volume of the resin and weigh the total mass. The wet bulk density ρ is calculated by the formula (total mass - measuring cylinder mass) / resin volume; S4: The adsorption amount of ephedra root tannins per unit time in the adsorption stage, the recovery rate of ephedra root tannins in the desorption stage, and the wet bulk density ρ are input into the first sub-model based on the linear regression algorithm, the second sub-model based on the support vector regression algorithm, and the third sub-model based on the random forest algorithm. The scores of each sub-model are added according to the dynamic weights to calculate the comprehensive score, and the macroporous resin with the highest comprehensive score is selected.

2. The method for screening macroporous resin for purifying tannin from Ephedrae root according to claim 1, characterized in that: Also includes: Determination of the specific surface area S of the macroporous resin: After vacuum degassing the pretreated resin, the specific surface area S of the macroporous resin was determined by nitrogen adsorption method; Desorption solution concentration coefficient of variation CV determination, the concentration of tannin in the desorption solution was detected by liquid chromatography, and the coefficient of variation CV of the desorption solution concentration was calculated according to the percentage of the standard deviation to the average value; Dynamic adsorption mass transfer coefficient K was determined by fitting the adsorption kinetic model based on the adsorption breakthrough curve data to obtain the dynamic adsorption mass transfer coefficient K. The input parameters of the first sub-model are the adsorption amount of ephedra root tannin per unit time in the adsorption stage, the recovery rate of ephedra root tannin in the desorption stage and the wet bulk density ρ, and the first sub-model score is output; The input parameters of the second sub-model include the adsorption amount of tannin from Ephedra sinica root per unit time in the adsorption stage, the coefficient of variation CV of the desorption solution concentration and the dynamic adsorption mass transfer coefficient K, and the second sub-model score is output after the nonlinear relationship is mapped by the kernel function; The input parameters of the third sub-model include the adsorption amount of ephedra root tannin per unit time in the adsorption stage, the recovery rate of ephedra root tannin in the desorption stage, the wet state bulk density ρ, the specific surface area S of the macroporous resin, and the coefficient of variation CV of the desorption solution concentration. The third sub-model score is output after parallel training of multiple decision trees. According to the historical prediction errors of each sub-model on the validation data set, the weight ratios of the first sub-model, the second sub-model, and the third sub-model are dynamically allocated, and the scores of each sub-model are added according to the dynamic weights to generate the final comprehensive score.

3. The method for screening macroporous resin for purifying tannin from Ephedra root according to claim 2, characterized in that: The kernel function of the second sub-model is a Gaussian kernel, with a kernel width σ∈[0.25,0.35], a regularization parameter C∈[1.5,2.5], and an insensitive loss parameter ε≤0.03; The number of decision trees of the third sub-model is n_trees∈[80,120], the maximum depth of a single tree is max_depth∈[12,18], and the minimum number of samples for node splitting is min_samples_split≥15.

4. The method for screening macroporous resin for purifying Ephedra root tannins according to claim 3, characterized in that: When the dynamic adsorption mass transfer coefficient K ≥ 0.45 cm / s, the weight of the second sub-model W2 is increased to 50%-60%; When the coefficient of variation of the desorption solution concentration CV≤12%, the weight W3 of the third sub-model is reduced to 15%-25%; When the specific surface area S of the macroporous resin is ≥ 400 m² / g and the wet bulk density ρ is ≤ 1.15 g / cm³, the weight W1 of the first submodel is fixed at 20%, and the weight W2 of the second submodel and the weight W3 of the third submodel are allocated the remaining weight in a ratio of 2:3; The adjustment step of each sub-model weight does not exceed ±5% / minute, and the total weight distribution error is controlled within ±1.5%.

5. The method for screening macroporous resin for purifying Ephedra root tannins according to claim 4, characterized in that: When the column temperature T∈[25℃,35℃], the weight W2 of the second sub-model is modified according to the formula: , and the sum of weights is kept at 100% through normalization.

6. The method for screening macroporous resin for purifying Ephedra root tannins according to claim 4, characterized in that: When the pH fluctuation ΔpH>0.3 is detected during the adsorption phase, the fault tolerance process is activated: The output value of the first sub-model is multiplied by the attenuation factor , the kernel function of the second sub-model is switched to a polynomial kernel, and the depth of the decision tree of the third sub-model is temporarily reduced to 10 layers.

7. The method for screening macroporous resin for purifying tannin from Ephedra root according to claim 2, characterized in that: Also includes: The two macroporous resins with the highest scores were selected for secondary verification. The adsorption-desorption process was repeated three times under the same operating conditions. The standard deviation of the purity of ephedra root tannins in the three experiments was calculated, and the macroporous resins with a standard deviation less than 1.0%-2.0% were retained as the final screening results.

8. The method for screening macroporous resin for purifying tannin from Ephedrae root according to claim 7, characterized in that: Each cycle must meet the following requirements: column temperature T control accuracy ±1°C, desorption liquid flow rate fluctuation range ≤±5% of the set value; Detect the purity of the Ephedra root tannin obtained in each cycle, calculate the standard deviation σ of the three purities, and if σ∈[1.0%, 2.0%], the macroporous resin is judged to be qualified; When both macroporous resins are qualified, the one with the smaller σ value is selected as the final screening result; if only one is qualified, it is directly selected; if both are unqualified, they are re-scored and the first two macroporous resins with the highest original scores are excluded.

9. The method for screening macroporous resin for purifying tannin from Ephedra root according to claim 8, characterized in that: Treat the selected macroporous resin under accelerated aging conditions, wherein the conditions are: temperature 40±2°C, relative humidity 75±5%, for 72-120 hours; After the aging treatment, the adsorption-desorption process of claim 1 is repeated to calculate the attenuation rate ΔQ of the adsorption amount of the ephedra root tannin before and after the aging; , where Q0 is the adsorption amount before aging, and Q1 is the adsorption amount after aging; Retain macroporous resin with ΔQ ≤ 5.0%.