Method for extracting and purifying daidzein

Abstract

The invention provides an extraction and purification method of daidzein, which comprises the following steps: optimizing multi-stage separation by applying a simulated annealing algorithm according to a thermosensitive degradation threshold value and a temperature boundary condition, and generating an optimized temperature set value set; the method comprises the following steps: classifying and marking glucoside impurities and structural similarity impurities by adopting a high performance liquid chromatography-mass spectrometry technology, measuring impurity response characteristic data, and generating an impurity removal temperature window; monitoring the stability of the compound through an online infrared spectrum and a mass spectrum, adjusting a temperature parameter, and generating a stability adjustment temperature parameter; configuring a multi-stage temperature optimization strategy, and generating impurity removal effect data; applying a gradient descent algorithm to finely adjust the temperature curve to generate a final temperature control parameter set; and constructing a correlation model of the temperature gradient and the separation effect through regression analysis, and generating a self-adaptive temperature control strategy database.

Description

Technical Field

[0001] This invention relates to the field of daidzein extraction technology, and in particular to a method for the extraction and purification of daidzein. Background Technology

[0002] Daidzein, as an important phytoestrogen compound, plays a crucial role in the development of functional foods, health products, and pharmaceuticals. These compounds have attracted significant attention due to their unique biological activities, and their high-purity preparation technology directly impacts the quality and efficacy of related products.

[0003] Current methods for extracting daidzein generally suffer from significant drawbacks. Traditional extraction processes often employ a fixed-temperature operation mode, failing to address the specific requirements of different separation stages. Existing purification techniques lack specificity when handling complex components, particularly when dealing with mixtures of structurally similar compounds, where the separation precision is insufficient to meet the demands of high-purity preparation. These methods demonstrate significant inadequacy in practical applications and struggle to address the multiple technical challenges involved in daidzein purification.

[0004] The core challenges in daidzein purification lie primarily in the technical contradiction between temperature control strategies and the selectivity of isomer separation. The temperature parameters directly determine the stability and separation efficiency of the compound, but these two requirements often conflict. While low temperatures can maintain compound structural stability and prevent thermosensitive degradation, they also significantly reduce mass transfer efficiency and separation precision. This contradiction is particularly pronounced when dealing with mixtures of isomers, as these compounds have extremely similar physicochemical properties, requiring precise separation while ensuring stability. The complexity of temperature control is further demonstrated in the removal of glycoside impurities. Different types of impurities respond significantly differently to temperature changes, and a single temperature condition cannot simultaneously meet the dual requirements of protecting the target compound and removing impurities. For example, higher temperatures are needed in certain purification stages to improve separation efficiency, but excessively high temperatures may lead to isomerization or degradation of daidzein, thus affecting the purity and activity of the final product.

[0005] Establishing a scientific and reasonable temperature gradient control system to achieve highly selective separation and effectively remove various impurities while ensuring the structural stability of daidzein has become a key issue in the development of efficient daidzein extraction technology. Summary of the Invention

[0006] This invention provides a method for extracting and purifying daidzein, mainly comprising: Physicochemical property data of a mixture of daidzein compounds were obtained. Thermogravimetric analysis (TGA) and differential scanning calorimetry (DSC) were used to determine the thermosensitive degradation threshold and isomerization reaction critical point, generating an initial temperature gradient range and temperature boundary conditions for each purification stage. Based on the thermosensitive degradation threshold and temperature boundary conditions, a simulated annealing algorithm was applied to optimize multi-stage separation, generating a set of optimized temperature setpoints. High-performance liquid chromatography-mass spectrometry (HPLC-MS / MS) was used to classify and label glycoside impurities and structurally similar impurities, measuring impurity response characteristic data and generating impurity removal temperature windows. A temperature gradient control system was configured, dynamically adjusting temperature parameters and gradient slope based on isomer concentration changes and separation status of structurally similar compounds monitored by online HPLC, generating real-time temperature adjustment parameters. Compound stability was monitored by online infrared spectroscopy and mass spectrometry, adjusting temperature parameters to generate stability adjustment temperature parameters. A multi-stage temperature optimization strategy was configured, generating impurity removal effect data. A gradient descent algorithm was applied to fine-tune the temperature curve, generating a final set of temperature control parameters. A regression analysis was used to construct a correlation model between temperature gradient and separation effect, generating an adaptive temperature control strategy database.

[0007] Furthermore, the acquisition of physicochemical property data of the daidzein mixture includes: determining the molecular weight, polarity, and solubility of the daidzein mixture using high-resolution mass spectrometry and nuclear magnetic resonance to obtain a physicochemical property dataset; extracting features from the molecular weight, polarity, and solubility in the physicochemical property dataset using principal component analysis to obtain the distribution characteristics of structural similarity parameters; determining the mass loss of each component at different temperatures using thermogravimetric analysis to obtain a thermosensitive degradation threshold; if the thermosensitive degradation threshold is lower than a preset temperature threshold, determining the isomerization reaction critical point by measuring the thermal absorption peak using differential scanning calorimetry; generating an initial temperature gradient range using linear interpolation based on the thermosensitive degradation threshold and the isomerization reaction critical point to obtain a temperature gradient distribution; dividing each purification stage using a k-means clustering algorithm to obtain the temperature boundary conditions for each stage; and optimizing the temperature boundary conditions using a simulated annealing algorithm to obtain optimized temperature control parameters for the purification stage.

[0008] Furthermore, the application of simulated annealing algorithm to optimize multi-stage separation includes: obtaining an initial temperature range from the thermosensitive degradation threshold and the temperature boundary conditions; initializing the temperature setpoints for multi-stage separation using simulated annealing algorithm to obtain a preliminary temperature set; configuring an online monitoring system based on the preliminary temperature set to acquire mass transfer efficiency data for each stage in real time to obtain a mass transfer efficiency dataset; increasing the temperature setpoint for the corresponding stage if the mass transfer efficiency is lower than a preset threshold to obtain an adjusted temperature setpoint; detecting reaction signals for each stage in real time to determine if isomerization reaction signals exist; decreasing the temperature setpoint for the corresponding stage if isomerization reaction signals are detected to obtain an updated temperature setpoint; iteratively applying the simulated annealing algorithm to optimize the multi-stage separation process using the updated temperature setpoints to obtain an optimized temperature setpoint set; and generating the final temperature control parameters for each stage using the optimized temperature setpoint set.

[0009] Furthermore, the classification and labeling of glycoside impurities and structurally similar impurities using high-performance liquid chromatography-mass spectrometry (HPLC-MS / MS) includes: classifying and labeling glycoside impurities and structurally similar impurities using HPLC-MS / MS to obtain classification and labeling data; if the signal intensity difference between glycoside impurities and structurally similar impurities in the classification and labeling data is greater than a preset threshold, then performing preliminary separation of the impurities using HPLC to obtain preliminary separation data; performing dimensionality reduction processing on the preliminary separation data using principal component analysis (PCA) to obtain a response feature vector; if the principal component variance contribution rate of the response feature vector is greater than a preset threshold, then performing secondary analysis on the response feature vector using mass spectrometry (MS / MS) to obtain molecular weight distribution data of the impurities; grouping the molecular weight distribution data using k-means clustering to determine the temperature response characteristics of each group of impurities to obtain temperature response characteristic data; and adjusting the separation conditions based on the temperature response characteristic data to determine the impurity removal temperature window data.

[0010] Furthermore, the configured temperature gradient control system includes: obtaining the temperature gradient range and initial separation conditions from the impurity removal temperature window data, initializing the temperature parameters for each stage, and obtaining initial temperature configuration data; acquiring real-time monitoring data by using online high-performance liquid chromatography (HPLC) to collect concentration change data of isomers and separation status data of structurally similar compounds; if the rate of change of isomer concentration in the real-time monitoring data exceeds a preset threshold, then using a support vector machine algorithm to classify the concentration change data, determine the temperature range of abnormal concentration, and obtain abnormal temperature range data; dynamically adjusting the parameters and gradient slope of the corresponding temperature range based on the abnormal temperature range data to obtain optimized temperature regulation data; if the peak resolution of the separation status data in the optimized temperature regulation data is lower than a preset threshold, then iteratively optimizing the gradient slope using a gradient descent algorithm to obtain optimized gradient slope data; and generating real-time temperature regulation parameters based on the optimized gradient slope data.

[0011] Furthermore, the monitoring of compound stability via online infrared spectroscopy and mass spectrometry includes: monitoring compound stability from the real-time temperature adjustment parameters using online infrared spectroscopy and mass spectrometry to obtain thermosensitive degradation product concentration data; if the degradation rate of the target compound in the thermosensitive degradation product concentration data exceeds a preset threshold, then reducing the temperature parameter of the corresponding stage to obtain adjusted temperature parameters; if the separation efficiency in the thermosensitive degradation product concentration data is lower than a preset threshold, then increasing the temperature parameter of the corresponding stage within the thermosensitive degradation threshold range to obtain updated temperature parameters; generating stability adjustment temperature parameters based on the updated temperature parameters; optimizing the multi-stage separation process using the stability adjustment temperature parameters to obtain optimized separation effect data; and adjusting the temperature control parameters based on the optimized separation effect data to generate a set of stability adjustment temperature parameters.

[0012] Furthermore, the configuration of the multi-stage temperature optimization strategy includes: obtaining temperature response characteristic data of glycoside impurities and structurally similar impurities from the impurity response characteristic data and the stability adjustment temperature parameters; configuring temperature control parameters using a medium temperature gradient mode for the glycoside impurities to generate glycoside impurity removal effect data; configuring temperature control parameters using a fine temperature control mode for the structurally similar impurities to generate structurally similar impurity removal effect data; grouping the impurity removal effects using a k-means clustering algorithm based on the glycoside impurity removal effect data and the structurally similar impurity removal effect data to obtain removal efficiency data for each group of impurities; optimizing the multi-stage temperature control parameters based on the removal efficiency data to generate an impurity removal effect data set; and determining the temperature optimization strategy for each stage using the impurity removal effect data set.

[0013] Furthermore, the application of the gradient descent algorithm to fine-tune the temperature curve includes: obtaining temperature response data and purity data at each stage from the impurity removal effect data and isomer purity data monitored by online high-performance liquid chromatography; iteratively optimizing the temperature response data and purity data using the gradient descent algorithm to obtain optimized temperature curve data; adjusting the temperature parameters and gradient slope at each stage based on the optimized temperature curve data to obtain optimized temperature control parameters; verifying the separation effect of the optimized temperature control parameters using online high-performance liquid chromatography to obtain verified separation effect data; if the isomer purity in the verified separation effect data is lower than a preset threshold, iteratively adjusting the temperature parameters to generate a final temperature control parameter set; and determining the temperature configuration for the high-purity daidzein product based on the final temperature control parameter set.

[0014] The technical solutions provided by the embodiments of the present invention may include the following beneficial effects: 1. This invention discloses an efficient separation and purification method for a mixture of daidzein aglycones. The method determines the thermosensitive degradation threshold through thermal analysis, optimizes the multi-stage separation temperature using a simulated annealing algorithm, and classifies and labels impurities using high-performance liquid chromatography-mass spectrometry to achieve precise control of the temperature gradient.

[0015] 2. This invention employs an online monitoring and dynamic adjustment strategy, adjusting temperature parameters in real time based on changes in isomer concentration and the separation state of structurally similar compounds, and monitoring compound stability through infrared spectroscopy and mass spectrometry to ensure high purity and high yield of the target product.

[0016] 3. For glycoside impurities and structurally similar impurities, medium and fine temperature gradient modes were adopted respectively. The temperature curve was fine-tuned by gradient descent algorithm. Finally, a correlation model between temperature gradient and separation effect was constructed to achieve adaptive temperature control for different batches of raw materials, which significantly improved the separation and purification efficiency of daidzein and product quality. Attached Figure Description

[0017] Figure 1 This is a flowchart of a method for extracting and purifying daidzein according to the present invention.

[0018] Figure 2 This is a schematic diagram of a method for extracting and purifying daidzein according to the present invention. Detailed Implementation

[0019] To enable those skilled in the art to better understand the technical solutions in this specification, the technical solutions in the embodiments of this specification will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of this specification, and not all embodiments. Based on the embodiments in this specification, all other embodiments obtained by those skilled in the art without creative effort should fall within the scope of protection of this specification.

[0020] Please see Figures 1-2 As shown, the extraction and purification method for daidzein in this embodiment may specifically include: S101. Obtain physicochemical property data from the daidzein mixture, including molecular weight, polarity, solubility and structural similarity parameters. Determine the thermosensitive degradation threshold and isomerization reaction critical point of each component by thermogravimetric analysis and differential scanning calorimetry, and generate the initial temperature gradient range and temperature boundary conditions for each purification stage.

[0021] The molecular weight of each component in a mixture of daidzein was determined by high-resolution mass spectrometry, and its polarity and solubility were determined by nuclear magnetic resonance (NMR) to obtain a physicochemical property dataset containing molecular weight, polarity, and solubility. Principal component analysis (PCA) was used to extract features from the physicochemical property dataset to generate a distribution feature dataset of structural similarity parameters. Based on the distribution feature dataset of structural similarity parameters, thermogravimetric analysis (TGA) was used to determine the mass loss of each component at different temperatures to obtain a thermosensitive degradation threshold dataset. If the thermosensitive degradation threshold was lower than a preset temperature threshold, differential scanning calorimetry (DSC) was used to determine the thermal absorption peaks of each component to generate an isomerization reaction critical point dataset. Based on the thermosensitive degradation threshold dataset and the isomerization reaction critical point dataset, a linear interpolation method was used to calculate the initial temperature gradient range to generate temperature gradient distribution data. For the temperature gradient distribution data, k-means clustering was applied to divide the purification stages to generate a temperature boundary condition dataset for each purification stage. Based on the temperature boundary condition datasets for each purification stage, a simulated annealing algorithm was used to optimize the temperature control parameters to generate an optimized purification stage temperature control parameter dataset.

[0022] In one possible implementation, when obtaining physicochemical property data from a mixture of daidzeins, high-resolution mass spectrometry is used to determine the molecular weight. For example, the sample is ionized by an electrospray ionization source and then introduced into the mass spectrometer to separate ion peaks with different mass-charge ratios, thereby obtaining accurate molecular weight values. This method helps to distinguish structurally similar components because molecular weight differences directly affect the choice of separation strategy, thereby improving purification efficiency.

[0023] For example, when determining polarity and solubility using nuclear magnetic resonance (NMR) technology, chemical shifts and coupling constants are analyzed using proton or carbon spectra to calculate polarity parameters such as dipole moments. Solubility is then determined by combining dissolution experiments with the spectra. This combination method can accurately capture the hydrophilic or lipophilic properties of components, which is beneficial for the precision of subsequent temperature control.

[0024] When using principal component analysis (PCA) to extract features from a dataset, the algorithm calculates the covariance matrix and extracts the principal component vectors, reducing the dimensionality of molecular weight, polarity, and solubility to a low-dimensional feature space and generating a distribution of structural similarity parameters. For example, similarity can be quantified into a herringbone distribution map, which helps to identify clustering patterns between components, thereby providing targeted input for thermal analysis and avoiding the time wastage caused by blind testing.

[0025] In one possible implementation, when performing thermogravimetric analysis based on a dataset of distribution characteristics of structural similarity parameters, the technique involves placing the sample in a temperature-controlled furnace and gradually increasing the temperature, recording the mass loss curve. For example, when the temperature rises to a certain point, the mass drops sharply, indicating a thermosensitive degradation threshold. This threshold dataset can reveal the stability limits of the components and is beneficial for preventing degradation losses during the purification process.

[0026] If the threshold is lower than the preset temperature, differential scanning calorimetry identifies the thermal absorption peak by measuring the heat flow difference between the sample and the reference. For example, the peak corresponds to the starting point of the isomerization reaction. This method can accurately locate the critical point, thereby ensuring that the temperature gradient design avoids unnecessary isomerization and improves product purity.

[0027] In one possible implementation, when using a linear interpolation method based on a dataset of thermosensitive degradation thresholds and isomerization reaction critical points, this method calculates a continuous temperature gradient range by inserting a linear function between threshold points, such as a smooth curve from a low threshold to a high critical point, to generate distribution data. This helps to form a uniform temperature transition, which is beneficial for the smooth progress of the purification stage.

[0028] When applying the k-means clustering algorithm to temperature gradient distribution data, the algorithm initializes k centroids and iteratively assigns data points to the nearest centroid and updates the centroids, dividing the distribution into purification stages, such as low temperature, medium temperature and high temperature stages, generating a temperature boundary condition dataset. This division can optimize resource allocation and ensure that each stage is tailored to the specific characteristics of the components.

[0029] In one possible implementation, when optimizing the temperature boundary condition dataset using a simulated annealing algorithm, which simulates the metal annealing process, gradually cools from an initial high temperature state, randomly perturbs the boundary parameters and accepts improved solutions, such as gradually adjusting the boundary values ​​to minimize the separation error represented by the energy function, an optimized temperature control parameter dataset is generated. This optimization can significantly improve purification efficiency, avoid local optimum traps, and thus achieve stable acquisition of high-purity daidzein products.

[0030] S102. From the thermosensitive degradation threshold and temperature boundary conditions in step 1, apply the simulated annealing algorithm to optimize the multi-stage separation, calculate the temperature setpoint for each purification stage, increase the temperature by 5 degrees Celsius when the mass transfer efficiency is lower than the preset threshold, and decrease the temperature by 5 degrees Celsius when an isomerization reaction signal is detected, thereby generating an optimized temperature setpoint set.

[0031] From the physicochemical properties data of the soybean aglycone mixture, molecular weight, polarity, and solubility parameters were obtained. The mass loss of each component was determined by thermogravimetric analysis to identify the thermosensitive degradation threshold. The thermal absorption peak was determined by differential scanning calorimetry to obtain the critical point of the isomerization reaction. An initial temperature gradient range was generated based on linear interpolation, and the temperature boundary conditions for each purification stage were defined. Based on the initial temperature gradient range and temperature boundary conditions, the temperature setpoints for multi-stage separation were initialized using a simulated annealing algorithm to generate a preliminary temperature set. An online monitoring system was configured to collect mass transfer efficiency data for each stage in real time, resulting in a mass transfer efficiency dataset.

[0032] From the mass transfer efficiency dataset, it is determined whether the mass transfer efficiency of each stage is lower than the preset threshold. If it is lower than the threshold, the temperature of the corresponding stage is increased by 5 degrees Celsius through the control module to obtain the adjusted temperature set value. The reaction signals of each stage are detected in real time to determine whether there are isomerization reaction signals.

[0033] If an isomerization reaction signal is detected, the temperature of the corresponding stage is reduced by 5 degrees Celsius through the control module to obtain an updated temperature setpoint. The simulated annealing algorithm is then applied iteratively using the updated temperature setpoint to optimize the multi-stage separation process and generate an optimized temperature setpoint set.

[0034] In one possible implementation, molecular weight, polarity, and solubility parameters are obtained from the physicochemical properties of the daidzein mixture. These parameters help to understand the separation behavior of the components. For example, components with larger molecular weights may require higher temperatures to promote mass transfer, while polarity parameters can guide solvent selection to enhance solubility. Thermogravimetric analysis (TGA) is used to measure the mass loss curve of each component as a function of temperature under controlled conditions, thereby determining the thermosensitive degradation threshold. This threshold represents the temperature at which the substance begins to decompose, which helps prevent component degradation during purification. Differential scanning calorimetry (DSC) is then used, which identifies heat absorption peaks by detecting changes in heat flow during heating, to obtain the isomerization reaction critical point, i.e., the temperature at which the component structure changes. Linear interpolation is used to smooth the initial temperature gradient range by interpolating values ​​between known temperature points. This ensures uniform temperature distribution, avoids local overheating, and delineates the temperature boundary conditions for each purification stage, which helps to ensure precise temperature control at each stage and improves overall separation stability.

[0035] For example, based on the initial temperature gradient range and temperature boundary conditions, a simulated annealing algorithm is applied. This algorithm simulates the metal annealing process to optimize the global optimum through random search and probabilistic acceptance of suboptimal solutions. The temperature setpoints for multi-stage separation are initialized to generate a preliminary temperature set. This set provides a starting point to avoid local optima. An online monitoring system is configured, i.e., sensors are installed to collect data in real time to obtain a mass transfer efficiency dataset. This allows for dynamic tracking of process changes and timely adjustments to improve separation efficiency.

[0036] In one possible implementation, the mass transfer efficiency of each stage is determined from the mass transfer efficiency dataset to see if it is lower than a preset threshold. This threshold is an efficiency lower limit set based on experience. If it is lower than the threshold, the temperature of the corresponding stage is increased by 5 degrees Celsius by the control module to obtain the adjusted temperature setting value. For example, increasing the temperature in a low-efficiency stage can accelerate molecular diffusion, increase the mass transfer rate, and help optimize resource utilization. The reaction signals of each stage are detected in real time, and the presence of isomerization reaction signals is determined by spectroscopy or sensors. This signal indicates an undesired structural change.

[0037] For example, if an isomerization reaction signal is detected, the temperature of the corresponding stage is reduced by 5 degrees Celsius through the control module to obtain an updated temperature setpoint. This can suppress the reaction and protect the integrity of the components. The simulated annealing algorithm is applied iteratively using the updated temperature setpoint, that is, the algorithm is repeated to refine and optimize the multi-stage separation process, generating an optimized temperature setpoint set. This set integrates dynamic adjustments, which is beneficial to achieving efficient purification and reducing energy consumption.

[0038] S103. From the optimized temperature setpoints in step 2, high performance liquid chromatography-mass spectrometry is used to classify and label glycoside impurities and structurally similar impurities, and the response characteristic data of each type of impurity at different temperatures are measured to generate impurity removal temperature windows and separation accuracy requirements.

[0039] From the optimized temperature setpoints, high-performance liquid chromatography-mass spectrometry (HPLC-MS) was used to classify and label glycoside impurities and structurally similar impurities, and signal intensity data of each type of impurity were obtained.

[0040] Based on the signal intensity data, the response characteristics of various impurities at different temperatures were measured to generate response characteristic data. Based on the response characteristic data, high performance liquid chromatography was used to perform preliminary separation of glycoside impurities and structurally similar impurities to obtain preliminary separation data.

[0041] For the initial separation data, the response features are reduced in dimensionality using principal component analysis to generate response feature vectors. Then, mass spectrometry is used to perform secondary analysis on the response feature vectors to obtain the molecular weight distribution data of the impurities.

[0042] Based on molecular weight distribution data, the response peak characteristics of various impurities at different temperatures are determined, generating temperature response characteristic data. Based on the temperature response characteristic data, the separation conditions in the temperature setpoint set are adjusted and optimized to determine the impurity removal temperature window data. For the impurity removal temperature window data, high performance liquid chromatography is used for iterative optimization to generate separation accuracy data.

[0043] In one possible implementation, a process is employed to classify and label glycoside impurities and structurally similar impurities from an optimized set of temperature setpoints using high-performance liquid chromatography-mass spectrometry.

[0044] Specifically, the sample is injected into a chromatographic column, and different impurities are separated by elution with the mobile phase. Then, mass spectrometry detects the ionized molecular signals to obtain signal intensity data. This technique combines the high separation capability of chromatography with the high sensitivity of mass spectrometry, which is beneficial for accurately distinguishing impurity types. For example, in pharmaceutical samples, glycoside impurities may show specific glycosyl fragment ions, while structurally similar impurities have similar parent ion peaks. This can improve classification accuracy because it allows for real-time monitoring of signal differences and avoids the problem of impurity confusion in traditional methods.

[0045] When measuring the response characteristics of signal intensity data, the curve of peak area changing with temperature is recorded. For example, glycoside impurities respond more strongly at lower temperatures, thus generating response characteristic data, which helps to optimize separation conditions to improve purity.

[0046] In one possible implementation, the process of preliminary separation using high-performance liquid chromatography (HPLC) based on response characteristic data involves adjusting column temperature and flow rate to separate impurities and obtain preliminary separation data. For example, extending the retention time for impurity components with strong responses can help reduce peak overlap and improve separation efficiency.

[0047] The process of dimensionality reduction of initially separated data using principal component analysis (PCA) is described. PCA is a statistical method that reduces dimensionality by calculating the data covariance matrix and extracting principal components.

[0048] Specifically, by using response features such as peak height and width as variables, principal components with high contribution rates are calculated to generate response feature vectors. This process simplifies data complexity and makes subsequent analysis more efficient. For example, in samples with diverse impurities, it can highlight key variations and avoid information redundancy.

[0049] In one possible implementation, the process of secondary analysis using mass spectrometry on the response feature vector involves further ionization and fragmentation of molecules based on the initial separation to obtain molecular weight distribution data. For example, it can be detected that the molecular weight of glycoside impurities is concentrated in a specific range, while structurally similar impurities are more widely distributed. This is beneficial for the precise identification of impurity structures. When determining the response peak characteristics based on the molecular weight distribution data, it is mapped onto the temperature variable to generate temperature response characteristic data. This can provide guidance for optimizing temperature selection, such as identifying the peak decay of certain impurities at high temperatures, thereby providing a basis for removal strategies.

[0050] In one possible implementation, the process of adjusting the separation conditions based on temperature response characteristic data involves iteratively modifying the set of temperature setpoints to determine the impurity removal temperature window data, for example, narrowing the window to the temperature range with the lowest peak value, which is beneficial to maximizing the impurity removal rate.

[0051] The process of iteratively optimizing the impurity removal temperature window data using high-performance liquid chromatography involves repeatedly testing separation parameters until the accuracy meets the requirements and generating separation accuracy data. This iteration can improve the overall separation accuracy, for example, by gradually refining the conditions during continuous operation to ensure that impurity residues are minimized.

[0052] S104. From the impurity removal temperature window in step 3, configure a temperature gradient control system to adjust the separation process in real time. Based on the changes in isomer concentration and the separation status of structurally similar compounds monitored by online high-performance liquid chromatography, dynamically adjust the temperature parameters and gradient slope of each stage to generate real-time temperature adjustment parameters.

[0053] The temperature gradient range and initial separation conditions are obtained from the impurity removal temperature window data. The temperature parameters of each stage are initialized through the temperature gradient control system to obtain the initial temperature configuration data. The online high performance liquid chromatography technology collects the isomer concentration change data and the separation status data of structurally similar compounds in real time to obtain real-time monitoring data. The isomer concentration change rate is extracted from the real-time monitoring data to determine whether the concentration change rate exceeds the preset threshold T4.

[0054] If the rate of concentration change exceeds the threshold T4, the concentration change data is classified using a support vector machine algorithm to determine the temperature range of concentration anomalies, thereby obtaining abnormal temperature range data. From the abnormal temperature range data, the parameters of the abnormal temperature range are obtained, and the temperature parameters and gradient slope of the corresponding range are adjusted by a temperature gradient control system to obtain optimized temperature regulation data. Based on the optimized temperature regulation data, online high-performance liquid chromatography technology re-monitors the separation status of structurally similar compounds to obtain updated separation status data.

[0055] From the updated separation state data, peak resolution data is extracted, and the temperature gradient control system adjusts the temperature parameters and gradient slope at each stage according to the peak resolution data to generate real-time temperature regulation parameters.

[0056] For example, the process of obtaining the temperature gradient range and initial separation conditions from the impurity removal temperature window data involves analyzing a preset temperature range. For instance, when separating daidzein, the temperature gradient range may gradually increase from room temperature to a higher temperature to optimize solubility. The temperature parameters at each stage are initialized by a temperature gradient control system, which is an automated device that sets the starting temperature, heating rate, and holding time based on the input range data to obtain the initial temperature configuration data. This initialization helps ensure the stability of the separation process and avoids premature precipitation of impurities or degradation of the target compound.

[0057] Online high-performance liquid chromatography (HPLC) technology acquires real-time data on isomer concentration changes and separation status of structurally similar compounds, providing real-time monitoring data. This technology captures concentration fluctuations by continuously injecting samples and monitoring chromatographic peaks, which is beneficial for timely identification of separation anomalies and thus improves overall purification efficiency.

[0058] In one possible implementation, the step of extracting the isomer concentration change rate from real-time monitoring data includes calculating the percentage increase or decrease in concentration per unit time, determining whether it exceeds a preset threshold T4, and if so, classifying the concentration change data using a support vector machine (SVM) algorithm. SVM is a machine learning method that constructs a hyperplane to classify data into normal and abnormal categories, determines the temperature range of abnormal concentrations, and obtains abnormal temperature range data. This classification helps to accurately locate problem areas. For example, when dealing with structurally similar compounds, a sharp drop in concentration may indicate that the temperature is too high, leading to degradation. The application of the SVM algorithm can quickly isolate these ranges, thereby reducing losses during the separation process and improving product purity.

[0059] For example, after obtaining the parameters of the abnormal temperature range from the abnormal temperature range data, the process of adjusting the temperature parameters and gradient slope of the corresponding range through the temperature gradient control system involves modifying the heating curve, such as reducing the slope of the abnormal range to mitigate the change, and obtaining optimized temperature regulation data. This adjustment is beneficial to restoring the separation equilibrium and avoiding peak overlap. The online high-performance liquid chromatography technology re-monitors the separation status of structurally similar compounds based on the optimized temperature regulation data to obtain updated separation status data. This re-monitoring can verify the adjustment effect. For example, if the original state shows that the peaks are not clearly distinguishable, the peaks are more clearly separated after adjustment, thereby ensuring the reliability of dynamic regulation and optimizing impurity removal.

[0060] In one possible implementation, the step of extracting peak resolution data from the updated separation state data quantifies the resolution by calculating the distance and width between chromatographic peaks. The temperature gradient control system adjusts the temperature parameters and gradient slope at each stage based on the peak resolution data to generate real-time temperature adjustment parameters. This resolution-based adjustment is beneficial for achieving fine control, such as further fine-tuning the slope to maximize separation efficiency when the isomer concentration is stable, thereby generating highly adaptable real-time temperature adjustment parameters that support the optimization of the entire separation process.

[0061] S105. From the real-time temperature adjustment parameters in step 4, online infrared spectroscopy and mass spectrometry are used to monitor the stability of the compound and determine the concentration of thermosensitive degradation products. If the degradation rate of the target compound exceeds the preset threshold, the temperature is reduced by 5 degrees Celsius. If the separation efficiency is lower than the preset threshold, the temperature is increased by 3 degrees Celsius within the thermosensitive degradation threshold range of step 1, thereby generating stability adjustment temperature parameters.

[0062] From the real-time temperature adjustment parameters in step 4, the characteristic absorption peaks of the target compound are monitored by online infrared spectroscopy, and the molecular weight distribution of the thermosensitive degradation products is analyzed by mass spectrometry to obtain the degradation rate and concentration of the target compound. If the degradation rate of the target compound exceeds the preset threshold, the temperature is reduced by 5 degrees Celsius to generate the first temperature adjustment parameter.

[0063] Based on the first temperature adjustment parameter, it is determined whether the separation efficiency is lower than the preset threshold. If it is lower than the threshold, the temperature is increased by 3 degrees Celsius within the range of the thermosensitive degradation threshold in step 1 to generate the second temperature adjustment parameter.

[0064] Based on the second temperature adjustment parameter, the characteristic absorption peak changes of the target compound are continuously monitored using online infrared spectroscopy, and the concentration changes of the thermosensitive degradation products are verified by mass spectrometry to obtain updated degradation rate data. Based on the updated degradation rate data, the parameters of the temperature control system are adjusted to generate stability adjustment temperature parameters.

[0065] In one embodiment, starting from real-time temperature control parameters, online infrared spectroscopy is used to monitor the characteristic absorption peaks of the target compound. This spectroscopic technique assesses the structural integrity of the compound by capturing the infrared absorption signal generated by the vibration of the compound molecules. For example, in the pharmaceutical purification process, scanning the hydroxyl absorption peak of a heat-sensitive drug such as vitamin C can reflect the changes in the stability of molecular bonds in real time, thereby calculating the degradation rate. This helps to detect thermally induced breakage early and avoid product loss.

[0066] By combining mass spectrometry analysis with the molecular weight distribution of thermosensitive degradation products, mass spectrometry identifies fragments based on the mass-to-charge ratio by ionizing the sample and separating ions. For example, when separating protein derivatives, the detection of a low molecular weight peak indicates that degradation has occurred. If the degradation rate data obtained exceeds a preset threshold, the temperature is lowered to protect the integrity of the compound. This adjustment can improve the overall purification yield and reduce the accumulation of byproducts.

[0067] For example, when judging separation efficiency, the first temperature adjustment parameter is used for evaluation. This efficiency is determined by monitoring the ratio of product purity to effluent rate. If it is below the threshold, the temperature is increased within the range of the thermosensitive degradation threshold. For example, when separating aromatic hydrocarbons in a distillation column, if the efficiency decreases, it indicates insufficient mass transfer. Increasing the temperature can enhance the molecular diffusion rate, thereby improving the separation effect. This method ensures a balance between temperature optimization and stability, which is beneficial for maintaining high purity output rather than blindly heating and causing degradation.

[0068] In one embodiment, online infrared spectroscopy is used to continuously monitor changes in characteristic absorption peaks. This continuous monitoring involves continuous spectral acquisition and peak intensity comparison. For example, for easily oxidized compounds such as polyphenols, the decay of carbon-oxygen bond peaks is observed to track real-time stability. Combined with mass spectrometry to verify concentration changes, the reduction of degradation products is confirmed by comparing ion abundance peaks. This verification process can provide accurate and updated degradation rate data, which helps to refine temperature control and bring higher process reliability and product quality stability.

[0069] For example, the parameters of the temperature control system can be adjusted based on updated degradation rate data. Such systems typically include feedback loops and actuators to modify heating element settings. For instance, in a continuous flow reactor, if a decrease in degradation rate indicates that the adjustment is effective, the parameters are locked to generate stability-adjusted temperature parameters. This generation method ensures the dynamic adaptability of the parameters, which is beneficial for achieving efficient separation while minimizing thermal damage, thereby maintaining the long-term stability of the target compound in multi-stage purification.

[0070] S106. From the impurity response characteristic data in step 3 and the stability adjustment temperature parameters in step 5, configure a multi-stage temperature optimization strategy. Use a medium temperature gradient mode for glycoside impurities and a fine temperature control mode for structurally similar impurities to generate impurity removal effect data for each stage.

[0071] S107. From the impurity removal effect data in step 6 and the isomer purity data monitored by online high performance liquid chromatography, the gradient descent algorithm is applied to fine-tune the temperature curve, optimize the temperature parameters and gradient slope at each stage, and generate the final temperature control parameter set for high-purity daidzein products.

[0072] S108. From the final temperature control parameter set in step 7 and the temperature response data at each stage, a correlation model between temperature gradient and separation effect is constructed through regression analysis, generating an adaptive temperature control strategy database for different batches of raw materials.

[0073] The above embodiments are only used to illustrate the technical solutions of the present invention and are not intended to limit it. The present invention has been described in detail with reference to preferred embodiments. Those skilled in the art should understand that modifications or equivalent substitutions can be made to the technical solutions of the present invention without departing from the spirit and scope of the present invention, and all such modifications and substitutions should be covered within the scope of the claims of the present invention.

Claims

1. A method for extracting and purifying daidzein, characterized in that, include: Physicochemical property data of daidzein mixtures were obtained, and thermosensitive degradation thresholds and isomerization reaction critical points were determined by thermogravimetric analysis and differential scanning calorimetry to generate the initial temperature gradient range and temperature boundary conditions for each purification stage. Based on the thermosensitive degradation threshold and the temperature boundary conditions, a simulated annealing algorithm is applied to optimize the multi-stage separation and generate a set of optimized temperature setpoints. High performance liquid chromatography-mass spectrometry was used to classify and label glycoside impurities and structurally similar impurities, measure impurity response characteristic data, and generate impurity removal temperature windows. Configure a temperature gradient control system to dynamically adjust temperature parameters and gradient slope based on the changes in isomer concentration and the separation status of structurally similar compounds monitored by online high-performance liquid chromatography, and generate real-time temperature adjustment parameters. By monitoring the stability of compounds using online infrared spectroscopy and mass spectrometry, temperature parameters are adjusted, and stability adjustment temperature parameters are generated. A multi-stage temperature optimization strategy is configured to generate impurity removal effect data. The temperature curve is fine-tuned by applying the gradient descent algorithm to generate the final set of temperature control parameters. A correlation model between temperature gradient and separation effect is constructed through regression analysis to generate an adaptive temperature control strategy database.

2. The method for extracting and purifying daidzein as described in claim 1, characterized in that, The acquisition of physicochemical property data of the daidzein mixture includes: determining the molecular weight, polarity, and solubility of the daidzein mixture by high-resolution mass spectrometry and nuclear magnetic resonance to obtain a physicochemical property dataset; Principal component analysis was used to extract features of molecular weight, polarity, and solubility from the physicochemical property dataset to obtain the distribution characteristics of structural similarity parameters. Thermogravimetric analysis was used to determine the mass loss of each component at different temperatures, and the thermosensitive degradation threshold was obtained. If the thermosensitive degradation threshold is lower than the preset temperature threshold, the thermal absorption peak is determined by differential scanning calorimetry to determine the critical point of the isomerization reaction. Based on the thermosensitive degradation threshold and the isomerization reaction critical point, an initial temperature gradient range is generated using a linear interpolation method to obtain the temperature gradient distribution; The k-means clustering algorithm was used to divide the purification process into different stages, and the temperature boundary conditions for each stage were obtained. The temperature boundary conditions were then optimized using a simulated annealing algorithm to obtain the optimized temperature control parameters for each purification stage.

3. The method for extracting and purifying daidzein as described in claim 1, characterized in that, The application of simulated annealing algorithm to optimize multi-stage separation includes: obtaining an initial temperature range from the thermosensitive degradation threshold and the temperature boundary conditions, initializing the temperature setpoints for multi-stage separation using simulated annealing algorithm, and obtaining a preliminary temperature set. Based on the initial temperature set, an online monitoring system is configured to acquire mass transfer efficiency data at each stage in real time, thereby obtaining a mass transfer efficiency dataset. If the mass transfer efficiency is lower than a preset threshold, the temperature setpoint for the corresponding stage is increased to obtain the adjusted temperature setpoint. The reaction signals at each stage are monitored in real time to determine whether isomerization reaction signals exist. If the isomerization reaction signal is detected, the temperature setpoint of the corresponding stage is reduced to obtain an updated temperature setpoint. Using the updated temperature setpoints, the simulated annealing algorithm is iteratively applied to optimize the multi-stage separation process, resulting in an optimized temperature setpoint set. The final temperature control parameters for each stage are then generated using the optimized temperature setpoint set.

4. The method for extracting and purifying daidzein as described in claim 1, characterized in that, The method of classifying and labeling glycoside impurities and structurally similar impurities using high-performance liquid chromatography-mass spectrometry includes: classifying and labeling glycoside impurities and structurally similar impurities using high-performance liquid chromatography-mass spectrometry, and obtaining classification and labeling data; If the signal intensity difference between glycoside impurities and structurally similar impurities in the classification label data is greater than a preset threshold, the impurities are initially separated by high performance liquid chromatography to obtain preliminary separation data. Principal component analysis (PCA) is used to reduce the dimensionality of the preliminary separated data to obtain the response feature vector. If the principal component variance contribution rate of the response feature vector is greater than a preset threshold, the response feature vector is analyzed a second time using mass spectrometry to obtain the molecular weight distribution data of the impurities; the molecular weight distribution data is grouped using a k-means clustering algorithm to determine the temperature response characteristics of each group of impurities, thereby obtaining temperature response characteristic data; based on the temperature response characteristic data, the separation conditions are adjusted to determine the impurity removal temperature window data.

5. The method for extracting and purifying daidzein as described in claim 1, characterized in that, The configured temperature gradient control system includes: obtaining the temperature gradient range and initial separation conditions from the impurity removal temperature window data, initializing the temperature parameters of each stage, and obtaining initial temperature configuration data; Real-time monitoring data is obtained by collecting concentration change data of isomers and separation status data of structurally similar compounds in real time using online high performance liquid chromatography. If the concentration change rate of isomers in the real-time monitoring data exceeds a preset threshold, the concentration change data is classified by support vector machine algorithm to determine the temperature range of concentration anomalies and obtain abnormal temperature range data. Based on the abnormal temperature range data, the parameters and gradient slope of the corresponding temperature range are dynamically adjusted to obtain optimized temperature regulation data. If the peak resolution of the separated state data in the optimized temperature regulation data is lower than a preset threshold, the gradient slope is iteratively optimized using a gradient descent algorithm to obtain optimized gradient slope data. Based on the optimized gradient slope data, real-time temperature regulation parameters are generated.

6. The method for extracting and purifying daidzein as described in claim 1, characterized in that, The method of monitoring compound stability through online infrared spectroscopy and mass spectrometry includes: monitoring compound stability from the real-time temperature adjustment parameters using online infrared spectroscopy and mass spectrometry to obtain thermosensitive degradation product concentration data; if the degradation rate of the target compound in the thermosensitive degradation product concentration data exceeds a preset threshold, then reducing the temperature parameter of the corresponding stage to obtain the adjusted temperature parameter. If the separation efficiency in the thermosensitive degradation product concentration data is lower than a preset threshold, the temperature parameter of the corresponding stage is increased within the thermosensitive degradation threshold range to obtain updated temperature parameters. Based on the updated temperature parameters, stability adjustment temperature parameters are generated. Through the stability adjustment temperature parameters, the multi-stage separation process is optimized to obtain optimized separation effect data. Based on the optimized separation effect data, the temperature control parameters are adjusted to generate a set of stability adjustment temperature parameters.

7. The method for extracting and purifying daidzein as described in claim 1, characterized in that, The configuration of the multi-stage temperature optimization strategy includes: obtaining temperature response characteristic data of glycoside impurities and structurally similar impurities from the impurity response characteristic data and the stability adjustment temperature parameters; For the glycoside impurities, a medium temperature gradient mode is used to configure temperature control parameters to generate data on the removal effect of glycoside impurities; for the structurally similar impurities, a fine temperature control mode is used to configure temperature control parameters to generate data on the removal effect of structurally similar impurities. Based on the glycoside impurity removal effect data and the structural similarity impurity removal effect data, the impurity removal effect is grouped using the k-means clustering algorithm to obtain the removal efficiency data of each group of impurities. Based on the removal efficiency data, the multi-stage temperature control parameters are optimized to generate an impurity removal effect data set. Through the impurity removal effect data set, the temperature optimization strategy for each stage is determined.

8. The method for extracting and purifying daidzein as described in claim 1, characterized in that, The application of the gradient descent algorithm to fine-tune the temperature curve includes: obtaining temperature response data and purity data at each stage from the impurity removal effect data and the isomer purity data monitored by online high performance liquid chromatography; and iteratively optimizing the temperature response data and the purity data using the gradient descent algorithm to obtain optimized temperature curve data. Based on the optimized temperature curve data, the temperature parameters and gradient slope of each stage are adjusted to obtain the optimized temperature control parameters; the separation effect of the optimized temperature control parameters is verified by online high-performance liquid chromatography to obtain the verified separation effect data. If the purity of isomers in the verified separation effect data is lower than a preset threshold, the temperature parameters are iteratively adjusted to generate a final set of temperature control parameters. Based on the final set of temperature control parameters, the temperature configuration of the high-purity daidzein product is determined.

Images