Real-time monitoring method and system for flow rate of at-211 column separation process

Abstract

The present application relates to the technical field of flow monitoring, and discloses a method and system for real-time monitoring of flow in an At-211 column separation process, the method comprising: collecting the conductivity of eluent in the outlet pipeline of the At-211 chromatographic column in real time, continuously analyzing the conductivity, and generating a state flag; wherein when it is detected that the eluent is switched to anhydrous ethanol phase, the state flag is valid; otherwise, the state flag is invalid; by collecting the conductivity and generating the state flag, the timing of anhydrous ethanol phase switching is accurately determined, the collected pressure time series data is analyzed to generate a steady-state pressure value, then secondary analysis is performed to obtain a pressure memory value, the stable state of pressure and the continuous influence of flow are comprehensively captured, and finally a flow stability index is generated, which can evaluate the flow state of the whole alcohol phase elution process. The present application does not need to rely on the continuity of the conductivity signal, effectively filters transient interference, and improves the accuracy of flow monitoring.

Description

Technical Field

[0001] This invention relates to the field of flow monitoring technology, specifically to a method and system for real-time flow monitoring during the At-211 column separation process. Background Technology

[0002] In the separation process of At-211 extraction chromatography column, especially in the anhydrous ethanol elution stage, conductivity monitoring is currently the most common method. The conductivity monitoring threshold is set based on the aqueous acid system. The continuity of the conductivity signal enables real-time monitoring of the eluent flow rate, thereby indirectly judging the stability of the eluent flow rate.

[0003] However, the above monitoring method still has the following defects: the core eluent for separation in the At-211 extraction chromatography column is anhydrous ethanol, whose conductivity is much lower than that of the aqueous acid solution. After the alcohol phase enters the detection flow path, the conductivity signal will drop sharply to below the monitoring threshold, resulting in the inability to capture the effective conductivity signal, causing the instantaneous flow rate calculation to fail, and it is difficult to evaluate the state of the eluent flow rate through the continuity of the conductivity signal, which affects the accuracy of flow rate monitoring during the elution process. Summary of the Invention

[0004] To address the shortcomings of existing technologies, this invention provides a method and system for real-time flow monitoring in the At-211 column separation process, thus solving the aforementioned problems.

[0005] The above-mentioned technical objective of the present invention is achieved through the following technical solution:

[0006] The real-time flow monitoring method for the At-211 column separation process includes:

[0007] Step S1: Real-time acquisition of the conductivity of the eluent in the outlet pipeline of the At-211 chromatographic column, continuous analysis of the conductivity, and generation of status indicators;

[0008] Specifically, when the eluent is detected to have switched to anhydrous ethanol, the status flag is valid; otherwise, the status flag is invalid.

[0009] Step S2: When the status flag is valid, the pressure of the eluent in the outlet pipeline is continuously collected in real time to obtain pressure time series data. Steady-state analysis is performed on the pressure time series data to generate a steady-state pressure value representing the pressure level under the current stable flow state.

[0010] Step S3: When the status flag is valid, perform a second analysis on the pressure time series data to obtain the pressure memory value that represents the continuous impact of flow on the current pressure.

[0011] Step S4: Analyze the steady-state pressure value and pressure memory value to generate a flow stability index that evaluates the flow state throughout the ethanol phase elution process.

[0012] Furthermore, continuous analysis of conductivity is performed to generate state flags, including:

[0013] Conductivity analysis is performed to generate a baseline conductivity and a control upper limit value;

[0014] The deviation of the conductivity from the reference conductivity is calculated, and the first derivative of the deviation over time is calculated. The deviation and the first derivative are then combined to generate a phase transition response factor that reflects the magnitude and rate of change of the deviation.

[0015] Furthermore, continuous analysis of conductivity is performed to generate state flags, which also includes:

[0016] The phase transition response factor is analyzed, its concentration is calculated, and the drift stability index is obtained.

[0017] A dynamic phase transition threshold is generated by jointly calculating the reference conductivity and the upper control limit value.

[0018] If the drift stability index remains below the dynamic phase transition threshold, the eluent is determined to have switched to the anhydrous ethanol phase, and the generated status flag is valid; otherwise, it is invalid.

[0019] Furthermore, steady-state analysis is performed on the pressure time-series data to generate steady-state pressure values ​​representing the pressure level under the current stable flow state, including:

[0020] The continuously collected pressure data is arranged according to the collection time to form pressure time series data;

[0021] Phase space reconstruction is performed on the pressure time series data, attractor structures are identified in the phase space, and the pressure micro-cluster distribution entropy of the volume evolution of each attractor structure over time is calculated.

[0022] The gradient change of the distribution entropy of pressure microclusters is analyzed, pressure segments are divided, and the similarity of all attractor structures within each pressure segment is calculated to generate an evolution consistency factor.

[0023] Furthermore, steady-state analysis is performed on the pressure time-series data to generate steady-state pressure values ​​representing the pressure level under the current stable flow state, which also includes:

[0024] Based on the evolution consistency factor, convergence pressure segments are screened and analyzed to generate pressure backbone sequences.

[0025] Based on the pressure baseline sequence, the convergence is calculated, and the pressure baseline sequence is calibrated using the convergence to generate a steady-state pressure value representing the pressure level under the current steady flow state.

[0026] Furthermore, when the status flag is valid, a secondary analysis is performed on the pressure time series data to obtain pressure memory values ​​representing the continuous impact of flow on the current pressure, including:

[0027] During the valid time period of the status flag, analyze the pressure fluctuations in the pressure time series data and generate a pressure imprint vector representing the dynamic fluctuations of pressure.

[0028] Analyze the pressure imprint vector to find the differences between different pressure values ​​and generate the afterimage attenuation value of the pressure residual intensity.

[0029] Furthermore, when the status flag is valid, a secondary analysis is performed on the pressure time series data to obtain a pressure memory value representing the continuous impact of flow on the current pressure, which also includes:

[0030] The pressure imprint vector and the afterimage attenuation value are fused to generate a pressure memory value that represents the continuous impact of flow on the current pressure.

[0031] Furthermore, based on steady-state pressure values ​​and pressure memory values, a flow stability index is generated to evaluate the flow state throughout the entire ethanol phase elution process, including:

[0032] The degree of fit between steady-state pressure value and pressure memory value is analyzed, and a pressure memory fit factor representing the degree of synergy between steady-state pressure and flow rate is generated.

[0033] Furthermore, the analysis, based on steady-state pressure values ​​and pressure memory values, generates a flow stability index to evaluate the flow state throughout the entire ethanol phase elution process, which also includes:

[0034] Based on the pressure-memory adaptation factor, the dynamic influence weight of flow rate on pressure state is calculated to obtain the flow-pressure correlation weight.

[0035] The steady-state pressure value, pressure memory value, flow-pressure correlation weight, and pressure memory adaptation factor are calculated in a coordinated manner to generate a flow stability index that evaluates the flow state throughout the entire ethanol phase elution process.

[0036] Furthermore, the real-time flow monitoring system for the At-211 column separation process, applied to the aforementioned monitoring method, includes:

[0037] The status analysis unit is used to collect the conductivity of the eluent in the outlet line of the At-211 chromatographic column in real time, continuously analyze the conductivity, and generate status indicators.

[0038] Specifically, when the eluent is detected to have switched to anhydrous ethanol, the status flag is valid; otherwise, the status flag is invalid.

[0039] The pressure analysis unit is used to continuously collect the pressure of the eluent in the outlet pipeline in real time when the status flag is valid, obtain pressure time series data, perform steady-state analysis on the pressure time series data, and generate a steady-state pressure value that represents the pressure level under the current stable flow state.

[0040] The impact analysis unit is used to perform secondary analysis on the pressure time series data when the status flag is valid, and obtain the pressure memory value representing the continuous impact of the flow rate on the current pressure.

[0041] The flow rate determination unit is used to analyze the steady-state pressure value and the pressure memory value to generate a flow rate stability index that evaluates the flow rate status throughout the ethanol phase elution process.

[0042] In summary, the present invention has the following main beneficial effects:

[0043] The state analysis unit collects conductivity data in real time and, through the collaborative calculation of baseline conductivity, control upper limit, phase change response factor, drift stability index, and dynamic phase change threshold, accurately determines whether the eluent has switched to the anhydrous ethanol phase, generating a precise and effective state flag to avoid invalid monitoring and false triggering. When the state flag is valid, the pressure analysis unit collects pressure time-series data. Through phase space reconstruction, attractor structure identification, pressure micro-particle distribution entropy calculation, pressure segment division, and evolution consistency factor analysis, it screens convergent pressure segments and calibrates to generate steady-state pressure values, effectively filtering transient pressure fluctuations and accurately representing the pressure level under the current stable flow state, while also influencing analysis. The unit generates a pressure imprint vector and a residual attenuation value through secondary analysis, and obtains a pressure memory value after fusion. This comprehensively captures the continuous impact and hysteresis effect of flow on the current pressure. Finally, the flow judgment unit calculates the pressure memory adaptation factor and flow-pressure correlation weight, and generates a flow stability index by coordinating the steady-state pressure value and the pressure memory value. This index is then used to evaluate the flow status of the entire ethanol phase elution process. This scheme does not rely on the continuity of conductivity signals, has strong anti-interference ability, improves the accuracy of flow monitoring, effectively avoids the failure of instantaneous flow calculation, accurately identifies the flow fluctuation pattern, ensures the smooth and controllable separation and elution process of the At-211 column, and avoids the adverse effects of flow fluctuations on the separation effect. Attached Figure Description

[0044] Figure 1 This is a flowchart illustrating the real-time flow monitoring method for the At-211 column separation process of the present invention.

[0045] Figure 2 This is a schematic diagram of the real-time flow monitoring system for the At-211 column separation process of the present invention. Detailed Implementation

[0046] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0047] refer to Figure 1 and Figure 2 A method for real-time flow monitoring during the At-211 column separation process includes:

[0048] Step S1: Real-time acquisition of the conductivity of the eluent in the outlet pipeline of the At-211 chromatographic column, continuous analysis of the conductivity, and generation of status indicators;

[0049] Specifically, when the eluent is detected to have switched to anhydrous ethanol, the status flag is valid; otherwise, the status flag is invalid.

[0050] Step S2: When the status flag is valid, the pressure of the eluent in the outlet pipeline is continuously collected in real time to obtain pressure time series data. Steady-state analysis is performed on the pressure time series data to generate a steady-state pressure value representing the pressure level under the current stable flow state.

[0051] Step S3: When the status flag is valid, perform a second analysis on the pressure time series data to obtain the pressure memory value that represents the continuous impact of flow on the current pressure.

[0052] Step S4: Analyze the steady-state pressure value and pressure memory value to generate a flow stability index that evaluates the flow state throughout the ethanol phase elution process.

[0053] In one embodiment, continuous analysis of conductivity is performed to generate a state flag, including:

[0054] Conductivity analysis is performed to generate a baseline conductivity and a control upper limit value. Specifically, this includes: sampling conductivity at a frequency of twice per second, setting the initial length of the adaptive adjustment window to 10 seconds, and adjusting the length of the adaptive adjustment window according to the following rules: calculating the absolute value of the difference between the conductivity of the two latest sampling points; if the absolute value of the difference is less than 0.1 μS / cm, increasing the length of the adaptive adjustment window by 0.5 seconds, but the total length shall not exceed 30 seconds; if the absolute value of the difference is greater than 0.5 μS / cm, decreasing the length of the adaptive adjustment window by 1 second, but the total length shall not be less than 5 seconds; otherwise, ensuring that the length of the adaptive adjustment window remains unchanged.

[0055] For each data point in the adaptive adjustment window, a local window is formed by taking data for 2 seconds before and after the data point. The ratio of the standard deviation to the mean of all data in the local window is calculated and is called the local coefficient of variation. The weight of the data point is 0.5 divided by the sum of the local coefficient of variation and 0.01 to ensure that the data points with higher stability have greater weights.

[0056] Sort all data points according to their conductivity from smallest to largest. For each sorted data point, calculate the sum of the weights of all data points up to the data point with the smallest conductivity. Then divide this sum of cumulative weights by the sum of the weights of all data points to get the cumulative weight percentage of that data point.

[0057] Among all data points, find the last data point with a cumulative weight percentage ≤ 35%, and the next data point immediately following it; for the conductivity and cumulative weight percentage of two adjacent data points, the conductivity corresponding to the cumulative weight percentage of 35% is the reference conductivity. The reference conductivity is used as the basic reference value for determining the conductivity of the ethanol phase, which is adapted to the extremely low conductivity of the ethanol phase.

[0058] Calculate the absolute value of the residual between the conductivity of each data point within the adaptive adjustment window and the reference conductivity value. Then, calculate the difference between the upper and lower quartiles of these residual absolute value sequences to obtain the interquartile range. Add the reference conductivity of the data point at the current moment to the interquartile range to obtain the control upper limit value. The control upper limit value is used to define the normal fluctuation range of the conductivity of the ethanol phase, distinguish the normal low conductivity fluctuation of the alcohol phase from the substantial phase transition from the water phase to the alcohol phase, and avoid misjudging the normal conductivity fluctuation of the alcohol phase as no signal.

[0059] The deviation of conductivity from the reference conductivity is calculated, and the first derivative of the deviation over time is calculated. The deviation and the first derivative are then fused to generate a phase transition response factor that reflects the magnitude and rate of change of the deviation. Specifically, this includes: calculating the absolute value of the difference between the current conductivity and the reference conductivity, then dividing it by the difference between the upper control limit and the reference conductivity, and normalizing the calculation result to the 0-1 interval to obtain the standard deviation at each sampling point at the corresponding time.

[0060] The sampling time interval is 0.5 seconds, based on the standard deviation of the previous and next sampling points of the current sampling point.

[0061] Subtract the previous standard deviation from the next standard deviation, divide by twice the sampling time interval, and then normalize the result to the 0-1 interval to obtain the first derivative of the standard deviation over time.

[0062] Multiply the standard deviation by a weight of 0.65 and the first derivative by a weight of 0.35, and then normalize the calculation result to the 0-1 range to generate a phase transition response factor that reflects the magnitude and rate of change of the deviation. The phase transition response factor is used to comprehensively judge the magnitude and rate of decrease of conductivity, to distinguish between a true phase transition and ordinary noise fluctuations, to ensure the accuracy of state switching judgment, and to avoid false triggering.

[0063] The standard deviation is mainly used to determine whether a substantial phase transition has occurred, such as switching from the aqueous phase to the ethanol phase. Therefore, it is given a high weight of 0.65 to ensure that the phase transition response factor is mainly anchored to the significant shift in signal amplitude, thereby effectively resisting the interference of instantaneous fluctuations caused by random noise. The first derivative captures the trend and speed of the deviation change and is given a relatively low weight of 0.35. The purpose is to suppress the excessive influence of its own potentially large fluctuations on the phase transition response factor, so that the phase transition response factor can respond quickly to the initial change of the signal without being falsely triggered by the short-term drastic fluctuations in the rate of change.

[0064] In one embodiment, the continuous analysis of conductivity and generation of state flags further includes:

[0065] The phase transition response factors are analyzed to calculate their concentration and obtain the drift stability index. Specifically, this involves: acquiring multiple phase transition response factors within a 15-second analysis window and forming a phase transition response factor sequence; calculating the 90th and 10th quantiles of the phase transition response factor sequence, subtracting the 10th quantile from the 90th quantile to obtain the quantile interval; calculating the average value of all phase transition response factors in the sequence, dividing the quantile interval by the average value, and normalizing to the 0-1 range to obtain the drift stability index. The drift stability index is mainly used to evaluate the stability of conductivity. The closer the drift stability index is to 1, the higher the concentration of the phase transition response factors, the more stable the signal, and the less drift.

[0066] The reference conductivity and the upper limit of control are calculated together to generate a dynamic phase transition threshold. Specifically, this includes: calculating the difference between the current upper limit of control and the reference conductivity to obtain the basic fluctuation amplitude; multiplying the basic fluctuation amplitude by the initial proportional coefficient 1.5 to obtain the initial threshold; and using the sum of the initial threshold and the reference conductivity as the initial candidate value of the dynamic phase transition threshold.

[0067] If the current phase transition response factor is <0.2, the output adjustment coefficient is 1.2; if the current phase transition response factor is ≥0.2, the output adjustment coefficient is 0.8. Multiply the initial candidate value by this dynamic adjustment coefficient and normalize the calculation result to the 0-1 interval to obtain the dynamic phase transition threshold. When the phase transition response factor is high, it indicates that the signal is changing drastically. At this time, the dynamic phase transition threshold will be lowered, and the judgment will be more sensitive. When the phase transition response factor is low, it indicates that the signal is stable. At this time, the dynamic phase transition threshold will be increased to enhance the anti-interference ability and thus adapt to different noises during judgment.

[0068] If the drift stability index remains below the dynamic phase transition threshold, the eluent is determined to have been switched to anhydrous ethanol phase, and the generated status flag is valid; otherwise, it is invalid. Specifically, this includes setting an 8-second observation window and obtaining 16 consecutive drift stability index values ​​and corresponding dynamic phase transition thresholds at a sampling frequency of 2 times per second within the observation window.

[0069] For each sampling moment within the observation window, the instantaneous deviation can be obtained by subtracting the drift stability index from the dynamic phase transition threshold at that moment and then dividing by the dynamic phase transition threshold.

[0070] The judgment threshold is set at 75% of the total data points in the judgment observation window, i.e., 12 sampling points. If the number of sampling points with positive instantaneous deviation in the judgment observation window is greater than or equal to the judgment threshold, it is considered that the drift stability index is continuously lower than the dynamic phase transition threshold. In this case, the eluent is determined to have been switched to the anhydrous ethanol phase, and the generated status flag is valid. Otherwise, the status flag remains invalid.

[0071] The baseline conductivity, adapted to the low conductivity characteristics of the alcohol phase, is calculated by adaptively adjusting the window. The upper limit of control is determined by combining the interquartile range, which accurately defines the normal fluctuation range of the alcohol phase conductivity, avoiding false judgments of no signal. Furthermore, the standard deviation and first derivative are integrated based on the phase change response factor, taking into account both the magnitude of deviation and the speed of change, effectively resisting the instantaneous fluctuation interference of random noise. Among them, the drift stability index evaluates the conductivity stability, and the dynamic phase change threshold is adaptively adjusted according to the signal state. Combined with the quantitative judgment rules of the judgment observation window, it can accurately identify the phase switching of anhydrous ethanol, continuously capture effective conductivity signals, and ensure the accuracy of real-time monitoring of eluent flow rate.

[0072] In one embodiment, steady-state analysis is performed on pressure time-series data to generate a steady-state pressure value representing the pressure level under the current stable flow state, including:

[0073] Pressure values ​​are collected for 15 seconds using the same acquisition frequency with an adaptive adjustment window. The continuously collected pressure values ​​are then arranged according to the acquisition time to form pressure time series data.

[0074] Phase space reconstruction is performed on the pressure time series data to identify attractor structures in the phase space and calculate the pressure micro-cluster distribution entropy of the volume evolution of each attractor structure over time. Specifically, this includes: reconstructing the phase space of the pressure time series data, setting the embedding dimension to 5 and the time delay to 3 sampling intervals. Based on this, the one-dimensional pressure time series data is mapped to a sequence of trajectory points in a five-dimensional phase space: the coordinates of each trajectory point are composed of five pressure values, which are taken from the positions of every 3 sampling points in the pressure time series data, thus forming a continuously reconstructed high-dimensional phase space trajectory. Here, the coordinates are the five-dimensional vector composed of five pressure values.

[0075] In phase space, the average distance between each phase space point and its 15 nearest neighbors is calculated to obtain the local density index. Phase space points whose local density index is greater than 1.2 times the average local density index of all phase space points are marked as candidate core points. Temporary clusters are formed with each candidate core point as the center and 0.3 times its local density index value as the radius. If the distance between two temporary clusters is less than the sum of the radii of the two temporary clusters, the two temporary clusters are merged. The independent clusters formed after merging are an attractor structure.

[0076] The evolution of the current attractor structure is tracked over five consecutive analysis cycles: each analysis cycle is 2 seconds. At the end of each analysis cycle, the number of pressure particles falling into the attractor structure is counted. The pressure particle is a five-dimensional hypercube in phase space with a side length of 0.1 times the standard deviation of the pressure time series data.

[0077] At the end of each analysis cycle, the geometric center of all pressure cluster centers within the current attractor structure is calculated: the arithmetic mean of the coordinates of the pressure cluster centers in each dimension of the five-dimensional phase space is calculated, and the point determined by these five average values ​​is the geometric center.

[0078] Calculate the Euclidean distance from the center point of each pressure micro-element to the geometric center, and calculate the arithmetic mean of all Euclidean distances to obtain the average radial distance;

[0079] The farthest Euclidean distance from the geometric center to the center of all pressure clusters is taken as the characteristic radius of the current attractor structure. Then, the average radial distance is divided by the characteristic radius to obtain the distribution concentration of the analysis period. The standard deviation of the distribution concentration of five analysis periods is then calculated to obtain the pressure cluster distribution entropy of the volume of each attractor structure over time. The pressure cluster distribution entropy is used to reflect the fluctuation of the compactness of the attractor structure. The lower the pressure cluster distribution entropy, the more stable the pressure dynamics, and also the more stable the flow rate.

[0080] The gradient change of the distribution entropy of pressure microclusters is analyzed, pressure segments are divided, and the similarity of all attractor structures within each pressure segment is calculated to generate an evolution consistency factor. Specifically, based on the distribution entropy sequence of pressure microclusters over ten consecutive analysis periods, the difference between the distribution entropy of pressure microclusters in two adjacent analysis periods is calculated, and this difference is divided by the analysis period duration of 2 seconds to obtain the instantaneous gradient at each time point. The arithmetic mean of the absolute values ​​of all instantaneous gradients is calculated, and one-quarter of this arithmetic mean is set as the gradient threshold. All time points where the absolute value of the instantaneous gradient is less than this gradient threshold are identified, and these points are marked as stable inflection points.

[0081] Starting from the beginning of the pressure time series data, the data between the beginning and the first stable inflection point is divided into the first pressure segment. Then, the data between two adjacent stable inflection points is divided into an independent pressure segment. The data between the last stable inflection point and the end time is divided into the last pressure segment.

[0082] Each pressure segment contains all its corresponding attractor structures. For any two attractor structures belonging to the same pressure segment and existing within the same analysis period, their centroid movement velocity vectors at the same time are calculated. Here, the centroid is the geometric center of the attractor structure. The centroid movement velocity vector is the change in coordinates of the centroid in each dimension between the current analysis period and the previous analysis period, divided by the analysis period duration of 2 seconds. The cosine similarity of these two centroid movement velocity vectors in five-dimensional space is also calculated. All attractor structure pairs and times within the pressure segment are traversed, and the arithmetic mean of these cosine similarities is calculated. The arithmetic mean is normalized to the 0-1 interval, which is the evolution consistency factor of the pressure segment. The closer the evolution consistency factor is to 1, the more consistent the evolution of the attractor structure and the more stable the pressure segment. This indicates that the flow rate in the entire flow path is uniform and stable, without local abrupt changes or blockages.

[0083] In one embodiment, performing steady-state analysis on pressure time-series data to generate a steady-state pressure value representing the pressure level under the current stable flow state further includes:

[0084] Based on the evolution consistency factor, convergence pressure segments are screened and analyzed to generate a pressure backbone sequence. Specifically, this includes: calculating the mean of the evolution consistency factor of all pressure segments, using the mean as the convergence threshold, and marking pressure segments with evolution consistency factors greater than the convergence threshold as convergence pressure segments; for each convergence pressure segment, the 60th percentile of all pressure values ​​within it is used as the representative pressure.

[0085] Arrange all convergent pressure segments in chronological order, and connect their corresponding representative pressures sequentially to form a representative pressure sequence. This representative pressure sequence is the pressure baseline sequence. The pressure baseline sequence is mainly used to filter out transient fluctuations and unstable segments, retaining only the representative values ​​of those stable and consistent pressure segments.

[0086] Based on the pressure baseline sequence, the convergence is calculated, and the pressure baseline sequence is calibrated using the convergence to generate a steady-state pressure value representing the pressure level under the current stable flow state. Specifically, for each convergent pressure segment representing the pressure in the pressure baseline sequence, the coefficient of variation of all pressure values ​​within that convergent pressure segment is subtracted from 1 to obtain the pressure tightness; the convergence corresponding to that pressure is obtained by multiplying the pressure tightness by the evolution consistency factor and taking the arithmetic square root.

[0087] By multiplying each representative pressure in the pressure baseline sequence by its corresponding convergence, summing all the product results, dividing by the sum of all convergences, and normalizing the result to the 0-1 interval, a steady-state pressure value representing the pressure level under the current stable flow state can be generated.

[0088] By reconstructing the phase space to identify the attractor structure, the calculation of the pressure cluster distribution entropy can accurately reflect the pressure dynamics and flow stability. The pressure segments are divided according to the gradient change of the pressure cluster distribution entropy, and the calculation of the evolution consistency factor can effectively identify the pressure stable segments, eliminate interference caused by local abrupt changes and blockages in the flow path, and at the same time, the convergent pressure segments are screened to generate a pressure baseline sequence. The final generated steady-state pressure value can effectively filter transient fluctuations, accurately characterize the pressure level under the current stable flow state, realize the reliable determination of flow stability in the alcohol phase stage, and improve the accuracy of flow monitoring throughout the separation process.

[0089] In one embodiment, when the status flag is valid, a secondary analysis is performed on the pressure time series data to obtain a pressure memory value representing the continuous impact of flow rate on the current pressure, including:

[0090] During the valid period of the status flag, analyze the pressure fluctuations in the pressure time series data and generate a pressure imprint vector representing the dynamic pressure fluctuations. Specifically, during the duration of the valid status flag, extract N consecutive pressure values ​​from the pressure time series data from back to front, where N≥4.

[0091] The difference between each pair of adjacent pressure values ​​is calculated sequentially, i.e., the pressure value of the later one minus the pressure value of the earlier one, to obtain N-1 instantaneous changes. The absolute values ​​of the instantaneous changes that are greater than zero are added together to obtain the positive energy. The absolute values ​​of the instantaneous changes that are less than zero are added together to obtain the negative energy. The sum of the positive energy and the negative energy is divided by the time length of the analysis window to obtain the fluctuation trend energy.

[0092] For each instantaneous change, if it is ≥0, then use the symbol 1 to represent an increase; otherwise, use the symbol 0 to represent a decrease, thus forming a symbol sequence of length N-1. For each symbol sequence, perform sliding truncation by grouping three consecutive symbols, with a truncation step size of 1, resulting in a total of N-3 groups. Each group of symbols represents a fluctuation pattern. In all groups, count the number of occurrences of each fluctuation pattern. Subtract the ratio of the minimum number of occurrences of a fluctuation pattern to the maximum number of occurrences of a fluctuation pattern from 1 to obtain the uniqueness factor. If the minimum number of occurrences of a fluctuation pattern is 0, then the uniqueness factor is directly 1. If all the occurrences of fluctuation patterns are exactly the same, then the ratio of the minimum number of occurrences to the maximum number of occurrences is 1, and the uniqueness factor is 0.

[0093] After normalizing the fluctuation trend energy and the pattern uniqueness factor to the 0-1 interval respectively, they are combined to form a pressure imprint vector representing the dynamic fluctuation of pressure.

[0094] The pressure imprint vector is analyzed to generate a residual attenuation value representing the residual pressure intensity. Specifically, this involves: multiplying the fluctuation trend energy in the pressure imprint vector by a uniqueness factor to obtain the cross energy density; calculating the ratio of the standard deviation of the pressure time series data within the most recent three seconds to the standard deviation of the pressure time series data within the entire effective period, and normalizing the calculation results to the 0-1 interval to obtain the recent pressure fluctuation intensity; and multiplying the recent pressure fluctuation intensity by the cross energy density to generate a residual attenuation value representing the residual pressure intensity.

[0095] In one embodiment, when the status flag is valid, a secondary analysis is performed on the pressure time series data to obtain a pressure memory value representing the continuous impact of flow rate on the current pressure, which further includes:

[0096] The pressure imprint vector and the afterimage attenuation value are fused to generate a pressure memory value that represents the continuous impact of flow on the current pressure. Specifically, the product of the fluctuation trend energy and the uniqueness factor in the pressure imprint vector is calculated to obtain the imprint synthesis intensity.

[0097] Divide the afterimage attenuation value by the imprint synthesis intensity, and input the calculation result into a hyperbolic sine function for nonlinear amplification to obtain the attenuation modulation factor; then calculate the skewness of the pressure time series data in the last five seconds of the effective period of the status flag as the fluctuation asymmetry factor: first calculate the average value of the five-second pressure data, then calculate the cube of the difference between each data point and the average value, sum these cube values ​​and divide by the number of data points to obtain the third central moment, at the same time calculate the standard deviation of the pressure data within five seconds, divide the third central moment by the cube of the standard deviation, and normalize the calculation result to the 0-1 interval to obtain the fluctuation asymmetry factor;

[0098] By multiplying the imprint synthesis intensity, attenuation modulation factor, and fluctuation asymmetry factor, and normalizing the calculation result to the 0-1 interval, the pressure memory value representing the continuous impact of flow on the current pressure can be obtained.

[0099] By analyzing pressure time-series data in a secondary manner, the continuous impact of flow rate on pressure is accurately captured. The constructed pressure imprint vector can accurately characterize the dynamic fluctuation characteristics of pressure. Combined with cross-energy density and recent pressure fluctuation intensity, a residual attenuation value is generated, which can effectively reflect the residual pressure intensity. The generated pressure memory value can accurately represent the continuous effect of flow rate on current pressure, capture the lag effect of flow rate fluctuation, filter out invalid transient fluctuations, avoid the failure of instantaneous flow rate calculation, improve the continuity and accuracy of flow rate monitoring in the alcohol phase, strengthen the accuracy of flow rate and pressure correlation determination, and ensure the stability of flow rate monitoring in the elution process.

[0100] In one embodiment, the analysis, based on the steady-state pressure value and the pressure memory value, generates a flow stability index to evaluate the flow state throughout the ethanol phase elution process, including:

[0101] The degree of fit between steady-state pressure values ​​and pressure memory values ​​is analyzed to generate a pressure-memory fit factor that represents the degree of synergy between the continuous influence of steady-state pressure and flow. Specifically, within a synchronization window of ten seconds, the steady-state pressure value sequence and the pressure memory value sequence are acquired respectively. The data in the steady-state pressure value sequence and the pressure memory value sequence are normalized to the 0-1 interval and then first-order difference operation is performed to obtain the pressure change sequence and the pressure memory change sequence respectively.

[0102] The pressure change sequence and the pressure memory change sequence are convolved to generate an envelope sequence reflecting the matching degree of their change rhythms. The pressure memory change sequence is fixed, and its time order is reversed. Then, the reversed pressure change sequence is gradually slid and aligned from the beginning of the pressure memory change sequence. For each sliding position, the sum of the products of corresponding data points in the overlapping portion of the two sequences is calculated. This process generates a new sequence where the value of each point represents the degree of overlap between the reversed pressure change sequence and the pressure memory change sequence at that offset. This new sequence is the envelope sequence reflecting the matching degree of change rhythms. The sum of squares of all values ​​in this envelope sequence is calculated to obtain the envelope energy.

[0103] Calculate the absolute value of the difference between the steady-state pressure value sequence and the pressure memory value sequence at the same time point, and calculate the average of these absolute values ​​of difference to obtain the instantaneous mismatch degree;

[0104] Divide the envelope energy by the instantaneous mismatch to obtain the fit ratio. Simultaneously analyze the proportion of identical signs in the pressure change sequence and the pressure memory change sequence to obtain the symbol synchronization rate: examine each data point at the same position in the pressure change sequence and the pressure memory change sequence respectively. If the values ​​of the two data points are both greater than zero or both are less than zero, it is determined that the points have the same sign. Count the number of positions with identical signs and divide the number by the total number of positions to obtain the symbol synchronization rate.

[0105] Multiply the fit ratio by the symbol synchronization rate and normalize the result to the 0-1 range to obtain the compressibility factor.

[0106] In one embodiment, the analysis of a flow stability index, which evaluates the flow rate status throughout the ethanol phase elution process, based on steady-state pressure and pressure memory values, further includes:

[0107] Based on the pressure memory adaptation factor, the dynamic influence weight of flow on pressure state is calculated to obtain the flow-pressure correlation weight. Specifically, it includes: taking the most recent 10 seconds as the correlation period, identifying the moment when the steady-state pressure value first appears to be a local maximum within the correlation period, recording the pressure memory adaptation factor corresponding to this moment, and marking it as the inflection point adaptation factor. The inflection point adaptation factor is used to capture the synergistic state of flow memory and steady-state pressure during the critical period of pressure inflection.

[0108] For a sequence of pressure memory values ​​within a related time period, the pressure memory value data before the local maximum value is taken as the boundary, and the pressure memory value data at and after the time is taken as the second half. The difference between the mean of all data in the second half and the mean of all data in the first half is calculated, and then divided by the standard deviation of the data in the second half to obtain the backward drift of the memory. The backward drift of the memory mainly reflects the cumulative direction and intensity trend of the flow influence.

[0109] The number of monotonic events is calculated by counting the number of consecutive increases or decreases in pressure time series data that exceed three sampling points within the correlation period. If the number of monotonic events is 0, the numerator is set to the product of the inflection point adaptation factor and the memory backward drift, multiplied by 0.5, and the denominator is set to 2. The initial weight is obtained by dividing the numerator by the denominator. Otherwise, the initial weight is obtained by multiplying the inflection point adaptation factor, the memory backward drift, and the number of monotonic events, and then dividing by the sum of the number of monotonic events and 2. The initial weight is then multiplied by the reciprocal of the current pressure memory adaptation factor, and the result is normalized to the 0-1 interval to obtain the flow-pressure correlation weight.

[0110] The steady-state pressure value, pressure memory value, flow-pressure correlation weight, and pressure-memory adaptation factor are synergistically calculated to generate a flow stability index that evaluates the flow state throughout the ethanol phase elution process. Specifically, this includes: calculating the ratio of the pressure memory value to the steady-state pressure value to obtain the memory load factor; calculating the natural logarithm of the reciprocal of the pressure-memory adaptation factor and using the exponent of the natural logarithm as the adaptation modulation coefficient; multiplying the memory load factor, the adaptation modulation coefficient, and the difference between 1 and the flow-pressure correlation weight to obtain the dynamic instability; subtracting the number 1 from the dynamic instability and normalizing the calculation result to the 0-1 range to obtain the flow stability index. The closer the flow stability index is to 1, the more stable the flow rate remains under the influence of fluctuations during the ethanol phase elution process.

[0111] By calculating the pressure-memory adaptation factor, the degree of synergy between stable pressure and continuous flow rate influence is accurately reflected. Combined with parameters of the relevant time period, the flow-pressure correlation weight is calculated to capture the dynamic influence and trend of flow rate on pressure. Finally, a flow stability index is generated by coordinating multiple parameters, which can evaluate the flow rate status throughout the alcohol phase elution process, effectively filter transient interference, accurately identify the pattern of flow rate fluctuations, avoid the failure of instantaneous flow rate calculation, improve the accuracy and continuity of flow rate monitoring, and ensure the stability and controllability of the separation and elution process of the extraction chromatography column and the separation effect.

[0112] In one embodiment, the real-time flow monitoring system for the At-211 column separation process is applied to the above-described monitoring method, including:

[0113] The status analysis unit is used to collect the conductivity of the eluent in the outlet line of the At-211 chromatographic column in real time, continuously analyze the conductivity, and generate status indicators.

[0114] Specifically, when the eluent is detected to have switched to anhydrous ethanol, the status flag is valid; otherwise, the status flag is invalid.

[0115] The pressure analysis unit is used to continuously collect the pressure of the eluent in the outlet pipeline in real time when the status flag is valid, obtain pressure time series data, perform steady-state analysis on the pressure time series data, and generate a steady-state pressure value that represents the pressure level under the current stable flow state.

[0116] The impact analysis unit is used to perform secondary analysis on the pressure time series data when the status flag is valid, and obtain the pressure memory value representing the continuous impact of the flow rate on the current pressure.

[0117] The flow rate determination unit is used to analyze the steady-state pressure value and the pressure memory value to generate a flow rate stability index that evaluates the flow rate status throughout the ethanol phase elution process.

[0118] Although embodiments of the invention have been shown and described, it will be understood by those skilled in the art that various changes, modifications, substitutions and alterations can be made to these embodiments without departing from the principles and spirit of the invention, the scope of which is defined by the appended claims and their equivalents.

Claims

1. A method for real-time monitoring of flow rate in At-211 column separation process, characterized by, include: Step S1: Real-time acquisition of the conductivity of the eluent in the outlet pipeline of the At-211 chromatographic column, continuous analysis of the conductivity, and generation of status flags, including: analysis of conductivity, generation of baseline conductivity and control upper limit value; The deviation of the conductivity from the reference conductivity is calculated, and the first derivative of the deviation over time is calculated. The deviation and the first derivative are then combined to generate a phase transition response factor that reflects the magnitude and rate of change of the deviation. The phase transition response factor is analyzed, its concentration is calculated, and the drift stability index is obtained. A dynamic phase transition threshold is generated by jointly calculating the reference conductivity and the upper control limit value. If the drift stability index remains below the dynamic phase transition threshold, the eluent is determined to have been switched to the anhydrous ethanol phase, and the generated status flag is valid; otherwise, it is invalid. Specifically, when the eluent is detected to have switched to anhydrous ethanol, the status flag is valid; otherwise, the status flag is invalid. Step S2: When the status flag is valid, the pressure of the eluent in the outlet pipeline is continuously collected in real time to obtain pressure time series data. Steady-state analysis is performed on the pressure time series data to generate a steady-state pressure value representing the pressure level under the current stable flow state. This includes arranging the continuously collected pressure according to the collection time to form pressure time series data. Phase space reconstruction is performed on the pressure time series data, attractor structures are identified in the phase space, and the pressure micro-cluster distribution entropy of the volume evolution of each attractor structure over time is calculated. The gradient change of the distribution entropy of pressure microclusters is analyzed, pressure segments are divided, and the similarity of all attractor structures in each pressure segment is calculated to generate an evolution consistency factor. Step S3: When the status flag is valid, perform a second analysis on the pressure time series data to obtain the pressure memory value that represents the continuous impact of flow on the current pressure. Step S4: Analyze the steady-state pressure value and pressure memory value to generate a flow stability index that evaluates the flow state throughout the ethanol phase elution process.

2. The method for real-time monitoring of flow rate in At-211 column separation process as claimed in claim 1 wherein, Steady-state analysis of pressure time-series data is performed to generate steady-state pressure values ​​representing the pressure level under the current steady flow state. This also includes: Based on the evolution consistency factor, convergence pressure segments are screened and analyzed to generate pressure backbone sequences. Based on the pressure baseline sequence, the convergence is calculated, and the pressure baseline sequence is calibrated using the convergence to generate a steady-state pressure value representing the pressure level under the current steady flow state.

3. The method for real-time monitoring of flow rate in At-211 column separation process as claimed in claim 2 wherein, When the status flag is valid, a secondary analysis is performed on the pressure time series data to obtain pressure memory values ​​representing the continuous impact of flow on the current pressure, including: Within the valid time period of the status flag, analyze the pressure fluctuations in the pressure time series data and generate a pressure imprint vector representing the dynamic fluctuations of pressure. Analyze the pressure imprint vector to find the differences between different pressure values ​​and generate the afterimage attenuation value of the pressure residual intensity.

4. The method for real-time monitoring of flow rate in At-211 column separation process as claimed in claim 3 wherein, When the status flag is valid, a secondary analysis is performed on the pressure time series data to obtain the pressure memory value representing the continuous impact of flow on the current pressure, which also includes: The pressure imprint vector and the afterimage attenuation value are fused to generate a pressure memory value that represents the continuous impact of flow on the current pressure.

5. The method for real-time monitoring of flow rate in At-211 column separation process as claimed in claim 4 wherein, Based on steady-state pressure and pressure memory values, an analysis is performed to generate a flow stability index that evaluates the flow state throughout the ethanol phase elution process, including: The degree of fit between steady-state pressure value and pressure memory value is analyzed to generate a pressure memory fit factor that represents the degree of synergy between steady-state pressure and flow rate.

6. The method for real-time monitoring of flow rate in At-211 column separation process as claimed in claim 5 wherein, The analysis, based on steady-state pressure and pressure memory values, generates a flow stability index to evaluate the flow state throughout the ethanol phase elution process. This also includes: Based on the pressure-memory adaptation factor, the dynamic influence weight of flow rate on pressure state is calculated to obtain the flow-pressure correlation weight. The steady-state pressure value, pressure memory value, flow-pressure correlation weight, and pressure memory adaptation factor are calculated in a coordinated manner to generate a flow stability index that evaluates the flow state throughout the entire ethanol phase elution process.

7. A system for real-time monitoring of flow rates in an At-211 column separation process, for use in a monitoring method as claimed in any one of claims 2 to 6, characterised in that, include: The status analysis unit is used to collect the conductivity of the eluent in the outlet pipeline of the At-211 chromatographic column in real time, continuously analyze the conductivity, and generate status indicators. Specifically, when the eluent is detected to have switched to anhydrous ethanol, the status flag is valid; otherwise, the status flag is invalid. The pressure analysis unit is used to continuously collect the pressure of the eluent in the outlet pipeline in real time when the status flag is valid, obtain pressure time series data, perform steady-state analysis on the pressure time series data, and generate a steady-state pressure value that represents the pressure level under the current stable flow state. The impact analysis unit is used to perform secondary analysis on the pressure time series data when the status flag is valid, and obtain the pressure memory value representing the continuous impact of the flow rate on the current pressure. The flow rate determination unit is used to analyze the steady-state pressure value and the pressure memory value to generate a flow rate stability index that evaluates the flow rate status throughout the ethanol phase elution process.

Images