Real-time pH control and stabilization method and system for online column separation

Abstract

This invention discloses a method and system for real-time pH control and stabilization in online column separation, belonging to the field of ion exchange chromatography separation technology. The method includes: applying a proton perturbation signal to the chromatographic column and detecting its elution response signal to obtain a parameter spectrum characterizing the dynamic buffer capacity of the chromatographic column; inputting the parameter spectrum and the target pH gradient into a physical transport model based on convection, diffusion, and adsorption coupling equations; the physical transport model outputs a pre-distorted pH gradient program to compensate for the dynamic buffer capacity characteristics; and based on the fluid ratio requirements defined by the pre-distorted pH gradient program and varying over time. This allows for the acquisition of dynamic buffer capacity characteristics through proton perturbation testing and the construction of a pre-distorted pH gradient program. Combined with closed-loop iterative optimization, this effectively counteracts the interference of the chromatographic column's dynamic buffer capacity on proton transport, ensuring that the pH curve of the post-column elution accurately matches the target linear gradient and eliminating pH nonlinear distortion and plateau problems.

Description

Technical Field

[0001] This invention relates to the field of ion exchange chromatography separation technology, and more specifically, to a method and system for real-time pH control and stabilization for online column separation. Background Technology

[0002] Ion exchange chromatography is widely used in biomedicine and other fields. In this process, pH gradient elution achieves multi-component separation by linearly adjusting the pH of the buffer solution to change the charge interaction strength between the stationary phase and the sample components. However, existing chromatographic columns have buffering capacity in their stationary phases, which can interfere with the input linear pH gradient. This causes the pH response of the effluent to deviate from the preset gradient, exhibiting nonlinear characteristics or the appearance of a pH plateau. This directly increases the difficulty of predicting elution time, reduces the resolution of adjacent component peaks, and worsens process repeatability. This is due to the non-correlation between the input gradient program and the output pH of the effluent. The reason is that there is no correspondence between the input gradient program and the actual output pH of the effluent. This problem is exacerbated when using weak ion exchangers as the stationary phase or when separating high-concentration protein samples, as the protein's own charge interacts with the protons in the buffer solution, further disrupting the pH change pattern and increasing the complexity of the pH environment within the column.

[0003] Currently, when using linear pH buffers for elution, the pH change curve of the effluent is severely distorted, directly causing problems such as unpredictable elution time, decreased separation, and poor process repeatability, failing to meet the requirements for high-precision separation. The reason for this phenomenon is that the dynamic buffer capacity of the stationary phase, sample, or stagnant liquid can produce unpredictable adsorption and release effects on protons. These components together constitute a system similar to a proton buffer pool, dynamically adsorbing or releasing protons, thereby distorting the preset pH change waveform. Moreover, this effect changes dynamically with the elution process and the chemical environment inside the column. In the scenario of separating high-concentration, multi-charged proteins with weak ion exchangers, the proton buffer pool effect makes the local desorption pH environment uncontrollable, seriously deteriorating the separation efficiency and restricting the development of technology towards high precision and high stability. Summary of the Invention

[0004] To address the problems existing in the prior art, the present invention aims to provide a method and system for real-time pH control and stabilization in online column separation. This method can obtain dynamic buffer capacity characteristics through proton perturbation testing and construct a pre-distorted pH gradient program. Combined with closed-loop iterative optimization, it can effectively counteract the interference of the dynamic buffer capacity of the chromatographic column on proton transfer, so that the pH curve of the post-column effluent accurately matches the target linear gradient, and eliminate pH nonlinear distortion and plateau problems.

[0005] To solve the above problems, the present invention adopts the following technical solution:

[0006] The first aspect is a method for real-time pH control and stabilization in online column separation, which includes:

[0007] Step 1: Apply a proton perturbation signal to the chromatographic column and detect its elution response signal to obtain a parameter spectrum characterizing the dynamic buffer capacity of the chromatographic column;

[0008] Step 2: Input the parameter spectrum and target pH gradient into the physical transport model based on the convection, diffusion and adsorption coupling equations. The physical transport model outputs a pre-distorted pH gradient program to compensate for dynamic buffer capacity characteristics.

[0009] Step 3: Based on the fluid ratio requirements that vary over time as defined by the pre-distorted pH gradient program, adjust the mixing ratio of the two limiting buffer solutions at a frequency higher than the preset frequency to generate the input fluid flow.

[0010] Step 4: The input liquid stream is introduced into the chromatographic column, and the dynamic buffer capacity characteristic of the chromatographic column acts on the input liquid stream;

[0011] Step 5: Obtain the actual pH curve of the column output solution. Based on the deviation between the actual pH curve and the target pH gradient, adjust the diffusion coefficient parameter in the physical transport model.

[0012] Step 6: Repeat steps 1 through 5 in sequence.

[0013] Further, step 1 includes:

[0014] Step 11: Continuously inject bipolar step pulses consisting of acidic and basic slugs into the chromatographic column to obtain a composite elution profile.

[0015] Step 12: Convert the real-time pH and conductivity values ​​in the composite effluent curve into instantaneous proton concentrations, calculate the proton effluent rate in combination with the flow rate, continuously subtract the cumulative amount of effluent protons from the known total amount of protons in the bipolar step pulse to obtain the net proton inventory in the column, differentiate the net proton inventory in the column to obtain the net proton adsorption-desorption rate curve, deconvolve the net proton adsorption-desorption rate curve to separate the forward bound proton flow and the reverse released proton flow;

[0016] Step 13: Based on the forward-bound proton flow and the reverse-release proton flow under different pulse intensities, construct a three-dimensional relationship surface between proton perturbation flux, instantaneous proton retention in the column, and instantaneous net proton exchange rate.

[0017] Further, step 1 includes:

[0018] Step 101: Apply a proton perturbation signal containing a continuous proton concentration increase segment and a concentration decrease segment to the chromatographic column to obtain a comprehensive response signal;

[0019] Step 102: Based on the known input flux of the proton perturbation signal, the comprehensive response signal is synchronously reverse-calculated based on the material conservation principle to obtain the real-time adsorption proton flux and the real-time desorption proton flux.

[0020] Step 103: Map the real-time adsorption proton flux and the real-time desorption proton flux onto a two-dimensional relationship plane with the current proton loading state of the chromatographic column and the net proton exchange rate as coordinates to generate a two-dimensional dynamic rate map.

[0021] Furthermore, the parameter maps and target pH gradient are input into a physical transport model based on the convection, diffusion, and adsorption coupling equations, including:

[0022] Step 21: The dynamic rate relationship in the two-dimensional dynamic rate spectrum is transformed into a dynamic function distributed along the chromatographic column axis. The dynamic function defines the relationship between the local net proton exchange rate and the proton concentration and proton saturation of the liquid flow.

[0023] Step 22: Construct a physical transport model with an inverse solution structure based on dynamic functions. Use the target column post-column linear pH gradient as the endpoint boundary condition of the physical transport model. Solve the convection, diffusion and adsorption coupling equations in reverse to obtain the trajectory of the column inlet proton concentration changing with time as a pre-distorted pH gradient program.

[0024] Furthermore, the physical transfer model outputs a pre-distorted pH gradient program to compensate for dynamic buffer capacity characteristics, including:

[0025] Step 23: The relationship represented by the two-dimensional dynamic rate map is analyzed and reconstructed into a bivariate nonlinear relationship mapping with the first variable being the current proton loading state of the chromatographic column and the second variable being the proton concentration of the liquid flow through the micro-element of the chromatographic column. The instantaneous net proton exchange rate of the micro-element is then mapped out.

[0026] Step 24: The bivariate nonlinear relationship mapping is embedded into the spatiotemporally discretized chromatographic column axial transport grid. The column outlet proton concentration corresponding to the target linear pH gradient is used as the end boundary condition. The synchronous backtracking is performed in reverse along the column axis and time axis. The instantaneous net proton exchange rate at each point is obtained by querying the mapping, and the discretized convection, diffusion and adsorption coupling equations are solved. The sequence of column inlet proton concentration changes with time is derived as the pre-distorted pH gradient program.

[0027] Furthermore, based on the time-varying fluid ratio requirements defined by the pre-distorted pH gradient program, the mixing ratio of the two limiting buffer solutions is adjusted at a frequency higher than a preset frequency to generate the input fluid flow, including:

[0028] Step 31: Analyze the proton concentration change sequence over time in the pre-distorted pH gradient program. For the change in proton concentration between adjacent time points in the sequence, calculate the forward action correction amount based on the preset fluid dynamic delay characteristics of the system flow path, and add the forward action correction amount to the control command sequence before the corresponding time point to form a mixed control command.

[0029] Step 32: Execute the hybrid control command at a frequency higher than the preset frequency to generate the input fluid flow, obtain the real-time feedback signal of the input fluid flow, compare the actual state represented by the real-time feedback signal with the expected state of the hybrid control command to generate a real-time compensation signal, and superimpose the real-time compensation signal onto the hybrid control command to form the final execution command.

[0030] Furthermore, the input liquid stream is introduced into the chromatographic column, and the dynamic buffering capacity characteristics of the chromatographic column act on the input liquid stream, including:

[0031] Step 41: Using bivariate nonlinear relationship mapping, based on the proton concentration and real-time flow rate of the input liquid flow at the column inlet over time, the instantaneous change rate of proton saturation of each axial micro-element of the chromatographic column is calculated simultaneously, and the spatial distribution state of the proton loading wave is dynamically generated.

[0032] Step 42: Capture the secondary physical signal at the column outlet that is sensitive to the spatial distribution of the proton load wave, and compare and interlock the actual change trajectory of the secondary physical signal with the expected change trajectory predicted based on the spatial distribution of the proton load wave in real time.

[0033] Furthermore, obtain the actual pH profile of the column output solution, including:

[0034] Step 51: Determine the confidence level of the original pH measurement signal based on the real-time comparison and interlock verification results. According to the confidence level, when it is determined to be low confidence, the soft measurement algorithm of proton concentration based on bivariate nonlinear relationship mapping and real-time flow rate is activated to reconstruct the low confidence section of the original pH measurement signal to generate the corrected pH curve.

[0035] Step 52: The corrected pH curve is inverted and correlated with the spatial distribution of the proton load wave. By adjusting the local transmission resistance parameter, the deviation between the simulated pH curve at the column outlet output by the model and the corrected pH curve is minimized, and the model error spectrum characterizing the systematic deviation of the model is obtained.

[0036] Furthermore, the diffusion coefficient parameters in the physical transport model are adjusted based on the deviation between the actual pH curve and the target pH gradient, including:

[0037] Step 53: Utilize the distribution characteristics of the model error spectrum along the chromatographic column axis, and according to the mapping rule between the preset error distribution pattern and the diffusion coefficient correction amount, allocate diffusion coefficient increments or decrements to different axial segments to generate diffusion coefficient correction distributions.

[0038] Step 54: Store the diffusion coefficient correction distribution, model error spectrum, and process conditions in the historical database. Based on the characteristics of the diffusion coefficient correction distribution and combined with similar data in the historical database, dynamically optimize the initial preset value of the diffusion coefficient parameter of the physical transfer model in future similar separation tasks through progressive weighted fusion rules.

[0039] In a second aspect, the present invention also provides a real-time pH control and stabilization system for online column separation, comprising: a column characterization module, which applies a proton perturbation signal to the chromatographic column and detects its elution response signal to obtain a parameter spectrum characterizing the dynamic buffer capacity of the chromatographic column;

[0040] The pre-distortion generation module inputs the parameter spectrum and the target pH gradient into a physical transport model based on the convection, diffusion and adsorption coupling equations. The physical transport model outputs a pre-distortion pH gradient program to compensate for dynamic buffer capacity characteristics.

[0041] The gradient generation module adjusts the mixing ratio of the two limiting buffer solutions at a higher frequency than preset, based on the fluid ratio requirements that change over time as defined by the pre-distorted pH gradient program, in order to generate the input fluid flow.

[0042] The column injection module introduces the input liquid stream into the chromatographic column, and the dynamic buffer capacity of the chromatographic column acts on the input liquid stream.

[0043] The calibration module acquires the actual pH curve of the column output liquid and adjusts the diffusion coefficient parameter in the physical transport model based on the deviation between the actual pH curve and the target pH gradient.

[0044] The iterative module sequentially executes the operations in the column characterization module, pre-distortion generation module, gradient generation module, column injection module, and correction module.

[0045] Compared with the prior art, the beneficial effects of the present invention are as follows:

[0046] (1) This scheme obtains the dynamic buffer capacity characteristics through proton perturbation test and constructs a pre-distorted pH gradient program. Combined with closed-loop iterative optimization, it effectively counteracts the interference of the dynamic buffer capacity of the chromatographic column on proton transfer, so that the pH curve of the post-column effluent accurately matches the target linear gradient and eliminates the problems of pH nonlinear distortion and plateau region.

[0047] (2) This scheme optimizes the physical transport model parameters through closed-loop iteration, and combines proton concentration soft measurement correction and secondary physical signal interlock verification to achieve precise real-time control of pH value, which can improve the accuracy of elution time prediction, improve the separation degree of adjacent component peaks, and meet the needs of high-precision separation in fields such as biomedicine.

[0048] (3) This scheme dynamically adapts to the changing characteristics of the dynamic buffer capacity of the chromatographic column by cyclically executing the entire process of proton perturbation test, gradient generation, liquid flow control, signal correction and model parameter adjustment, solves the problem of uncontrollable local desorption pH environment when separating high-concentration multi-charged proteins with weak ion exchangers, and improves the separation efficiency of complex samples.

[0049] (4) This scheme continuously reduces the deviation between the model simulation and the actual proton transfer process by using a buffer mixing control mechanism of forward-looking action correction and real-time feedback compensation, combined with the progressive weighted optimization of the diffusion coefficient parameter, which greatly improves the repeatability and stability of the online column separation process and reduces the impact of process fluctuations on the separation effect. Attached Figure Description

[0050] To more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the accompanying drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are merely some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without any creative effort.

[0051] Figure 1 This is a flowchart of the method for real-time pH control and stabilization in online column separation according to the present invention;

[0052] Figure 2 This is a flowchart illustrating the various modules in the real-time pH control and stabilization system for online column separation according to the present invention. Detailed Implementation

[0053] The technical solutions of the present invention will be clearly and completely described below with reference to the accompanying drawings of the embodiments of the present invention. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. All other embodiments obtained by those skilled in the art based on the embodiments of the present invention without creative effort are within the scope of protection of the present invention.

[0054] Example 1

[0055] Please see Figure 1 A method for real-time pH control and stabilization in online column separation, comprising:

[0056] Step 1: Apply a proton perturbation signal to the chromatographic column and detect its elution response signal to obtain a parameter spectrum characterizing the dynamic buffer capacity of the chromatographic column. The specific operation is as follows:

[0057] In the online column separation process, a preset proton perturbation signal is first continuously applied to the chromatographic column in the state of separation or pretreatment. This signal needs to be able to excite the proton adsorption and desorption behavior of components such as the stationary phase and stagnant liquid in the chromatographic column. At the same time, a dedicated detection device at the chromatographic column outlet captures the response signal of the effluent after passing through the chromatographic column in real time. This response signal contains characteristic information related to proton exchange, such as changes in proton concentration and conductivity. By synchronously acquiring and correlating the input data of the perturbation signal and the output data of the effluent response signal, characteristic parameters that reflect the changes in the buffering capacity of the chromatographic column at different times and under different proton loading conditions are extracted. These characteristic parameters are systematically organized and presented in a chromatographic format, ultimately forming a parameter spectrum that can accurately characterize the dynamic buffering capacity of the chromatographic column.

[0058] Obtaining the parameter chromatogram characterizing the dynamic buffer capacity of the chromatographic column also includes the following steps:

[0059] Step 11: Apply a proton perturbation signal containing a continuous proton concentration increase segment and a concentration decrease segment to the chromatographic column to obtain a comprehensive response signal. The specific operation is as follows:

[0060] To ensure that the acquired response signal fully covers the dynamic characteristics of the column during the entire process of proton adsorption and desorption, a specific form of proton perturbation signal needs to be applied to the column. This proton perturbation signal includes a continuous proton concentration increment segment and a proton concentration decrement segment. In the increment segment, the proton concentration gradually increases from the initial reference value to a set peak value at a preset rate, while in the decrement segment, the proton concentration gradually decreases from the peak value back to the initial reference value at a preset rate. The two signal segments are continuously connected without abrupt changes to avoid causing severe impact on the chemical environment inside the column. Throughout the application of this perturbation signal, other process parameters such as the flow rate and temperature of the column are kept stable. At the same time, a high-sensitivity detection device deployed at the column outlet continuously collects the comprehensive response signal of the effluent. This signal is a comprehensive reflection of the interaction between the stationary phase, stagnant liquid, and any possible sample residues and protons after the proton perturbation acts on the column. It includes characteristic information of multiple physicochemical processes such as proton adsorption, desorption, and diffusion.

[0061] Step 12: Based on the known input flux of the proton perturbation signal, perform synchronous reverse calculation on the comprehensive response signal based on material conservation to obtain the real-time adsorption proton flux and the real-time desorption proton flux. The specific operation is as follows:

[0062] Based on the comprehensive response signal obtained in step 11, combined with the known proton perturbation signal input flux, synchronous reverse calculation is performed according to the law of conservation of mass to obtain the real-time adsorption proton flux and the real-time desorption proton flux. The application logic of the law of conservation of mass in this process is that the total proton flux input to the chromatographic column per unit time is equal to the sum of the proton flux adsorbed by the components in the chromatographic column per unit time, the proton flux desorbed from the components in the chromatographic column per unit time, and the proton flux discharged with the effluent per unit time. Based on this, the correlation formula can be derived as follows:

[0063] ;

[0064] The derivation of this formula establishes a material balance relationship for the input, output, and internal transformation process of protons within the chromatographic column, using time t as the node. It eliminates the influence of irrelevant factors such as system leakage, considering only the adsorption, desorption, and flow transfer of protons, thereby abstracting a quantitative relationship between proton fluxes. In the formula, This represents the input flux of the proton perturbation signal at time t, in units of... ; This represents the real-time adsorption proton flux of the component within the chromatographic column at time t, in units of... ; This represents the real-time proton flux of the component in the chromatographic column at time t, in units of... ; This represents the proton flux discharged with the effluent at time t, in units of 1. This value can be obtained by converting the comprehensive response signal. During the calculation process, time is used as the synchronization axis, and the known values ​​at each moment are... and Substituting into the above formula, we can solve it in reverse to obtain the corresponding time. and This enables continuous acquisition of real-time proton adsorption flux and real-time proton desorption flux, among which... It can be obtained by converting the comprehensive response signal.

[0065] Step 13: Map the real-time adsorption proton flux and the real-time desorption proton flux onto a two-dimensional relationship plane with the current proton loading state of the column and the net proton exchange rate as coordinates to generate a two-dimensional dynamic rate map. The specific operation is as follows:

[0066] To transform the real-time adsorption and desorption proton fluxes into parameters that directly characterize the dynamic buffer capacity of the chromatographic column, a two-dimensional mapping process is required. First, the coordinate system of the two-dimensional plane is determined. The horizontal axis represents the current proton loading state of the column, obtained by integrating the real-time adsorption and desorption proton fluxes over time, reflecting the total amount of protons adsorbed by the components within the column at a given moment. The vertical axis represents the net proton exchange rate, calculated by the difference between the real-time adsorption and desorption proton fluxes. That is, the net proton exchange rate equals the real-time adsorption proton flux minus the real-time desorption proton flux, reflecting the proton adsorption and desorption process within the column at a given moment. The dynamic equilibrium state is determined. During the mapping process, with time as the correlation dimension, the real-time adsorption proton flux and real-time desorption proton flux at each moment are converted into the current proton loading state value and net proton exchange rate value of the column at that moment, respectively. The two parameter values ​​at each moment correspond to a feature point on the two-dimensional relational plane. The feature points at all moments during the entire disturbance and response test are continuously collected and connected to form a trajectory curve that reflects the change law of net proton exchange rate under different proton loading states. At the same time, combined with the density distribution, slope of change and other information of each feature point, a two-dimensional dynamic rate spectrum is finally generated. This spectrum can clearly show the change characteristics of the column's dynamic buffer capacity with proton loading state and exchange rate.

[0067] In a preferred embodiment of the present invention, step 2 is further included, in which the parameter spectrum and the target pH gradient are input into a physical transport model based on the convection, diffusion and adsorption coupling equations. The physical transport model outputs a pre-distorted pH gradient program for compensating for dynamic buffer capacity characteristics. The specific operation is as follows:

[0068] After obtaining the dynamic buffer capacity characteristic parameter spectrum of the chromatographic column, a pre-distorted pH gradient program that can compensate for the influence of the dynamic buffer capacity needs to be generated by combining it with the preset target pH gradient through a dedicated physical transport model. This physical transport model is based on the convection, diffusion and adsorption coupling equations and can characterize the transport law of protons in the chromatographic column and the interaction process with the components in the column. In the operation, the parameter spectrum characterizing the dynamic buffer capacity of the chromatographic column obtained in step 1 and the target pH gradient required for the separation task, which is usually a linear pH gradient, are first input into the physical transport model. The parameter spectrum provides the model with the basis for the dynamic characteristics of proton adsorption and desorption in the column, and the target pH gradient sets the final output control target of the model. The model quantifies and analyzes the degree of interference of the dynamic buffer capacity on proton transport through the coupling operation of the input information, and then reversely derives the column inlet pH gradient adjustment law that can offset the interference, and finally outputs the pre-distorted pH gradient program. This program can make the actual pH curve of the column outlet effluent accurately match the target pH gradient after the input liquid flows through the dynamic buffer capacity of the chromatographic column.

[0069] The process of inputting the parameter spectrum and the target pH gradient into a physical transport model based on the coupling equations of convection, diffusion, and adsorption specifically includes the following steps:

[0070] Step 21: Transform the dynamic rate relationships in the two-dimensional dynamic rate chromatogram into a dynamic function distributed along the column axis. The dynamic function defines the relationship between the local net proton exchange rate and the proton concentration and proton saturation of the liquid stream. The specific operation is as follows:

[0071] To adapt the characteristic information in the two-dimensional dynamic rate spectrum to the computational requirements of the physical transport model, the dynamic rate relationship contained in the spectrum needs to be transformed into a dynamic function distributed along the column axis. The two-dimensional dynamic rate spectrum reflects the correlation between the overall net proton exchange rate of the column and the proton loading state and the proton concentration of the liquid stream. However, the stationary phase distribution and proton loading state differ at different positions along the column axis, and a single overall relationship cannot accurately characterize the local proton transport characteristics. Therefore, the two-dimensional dynamic rate spectrum needs to be discretized along the axis to extract the correlation data between the net proton exchange rate and the proton concentration and proton saturation of the liquid stream at different axial positions. Through data fitting and pattern summarization, a dynamic function distributed point-by-point along the column axis is constructed. This dynamic function clearly defines the quantitative relationship between the local net proton exchange rate at a certain axial position and the proton concentration of the liquid stream flowing through that position and the proton saturation of the stationary phase at that position.

[0072] Step 22: Construct a physical transport model with an inverse solution structure based on dynamic functions. Use the target post-column linear pH gradient as the endpoint boundary condition of the physical transport model. Solve the convection, diffusion, and adsorption coupling equations in reverse to obtain the trajectory of the column inlet proton concentration over time as a pre-distorted pH gradient program. The specific operations are as follows:

[0073] Based on the axial distribution dynamic function obtained in step 21, a physical transport model with a reverse solution structure is constructed. The pre-distorted pH gradient program is obtained by solving the coupled equations in reverse. First, based on the convection, diffusion, and adsorption coupled equations, the dynamic function constructed in step 21 is embedded into the core calculation of the model, enabling the model to accurately quantify the impact of local proton exchange on proton transport. Considering that conventional forward solution models cannot directly obtain inlet conditions that can offset the interference of dynamic buffer capacity, this model adopts a reverse solution structure. The column outlet proton concentration corresponding to the linear pH gradient after the target column is used as the model's endpoint boundary condition, i.e., the proton concentration at a fixed point in the outlet is taken as the target value. During the solution process, the model reverses its evolution from the column outlet to the column inlet. By solving the convection, diffusion, and adsorption coupled equations, the required proton concentration at the column inlet at each moment is determined, ultimately forming the trajectory of the column inlet proton concentration changing over time. This trajectory is the pre-distorted pH gradient program. The simplified form of the coupled equations is:

[0074] ;

[0075] The derivation of the formula is based on the proton transport process within the chromatographic column, considering both convective migration and molecular diffusion, as well as the adsorption and desorption reactions of protons with components within the column, expressed as the proton consumption or release rate. A differential equation is established based on the law of conservation of mass, abstracting the spatiotemporal variation of proton concentration. In the formula, c is the proton concentration in the liquid stream (mol / L); t is time (s); u is the average flow rate (m / s); z is the axial coordinate of the chromatographic column (m); and D is the proton diffusion coefficient (m² / s). The local net proton exchange rate, in units of , defined by the dynamic function in step 21, where For proton saturation, θ = actual proton adsorption / maximum proton adsorption, dimensionless.

[0076] The physical transfer model outputs a pre-distorted pH gradient program to compensate for dynamic buffer capacity characteristics, which includes the following steps:

[0077] Step 23: The relationship represented by the two-dimensional dynamic rate map is analyzed and reconstructed into a bivariate nonlinear relationship mapping with the first variable being the current proton loading state of the chromatographic column and the second variable being the proton concentration of the liquid flowing through the micro-element of the chromatographic column. The mapping outputs the instantaneous net proton exchange rate of the micro-element. The specific operation is as follows:

[0078] To achieve accurate simulation of the intracolumn proton exchange process using a physical transport model, the relationships represented by the two-dimensional dynamic rate spectrum need to be analyzed and reconstructed to form a bivariate nonlinear relationship mapping. First, the characteristic trajectories and data points in the two-dimensional dynamic rate spectrum are systematically analyzed to extract the net proton exchange rate data corresponding to different column proton loading states and different liquid flow proton concentrations, clarifying the nonlinear correlation among the three. Considering that the subsequent model requires precise calculations at the micro-element level, this correlation is reconstructed into a bivariate input, univariate output nonlinear mapping, where the first variable is set as the current mass transfer rate of the column. The first variable is the proton loading state, which reflects the total amount of protons adsorbed by the stationary phase within the micro-element and is obtained by integrating the historical proton exchange flux. The second variable is set as the proton concentration of the liquid flowing through the micro-element of the chromatographic column, which reflects the proton supply intensity inside and outside the micro-element. During the reconstruction process, discrete data points are transformed into continuous mapping relationships through conventional mathematical methods such as data interpolation and nonlinear fitting. This ensures that any set of proton loading state and liquid proton concentration parameters can be input, and the corresponding instantaneous net proton exchange rate of the micro-element can be output through this mapping. This mapping will serve as the data for subsequent micro-element-level proton transfer calculations and directly determine the accuracy of the model simulation.

[0079] Step 24: Embed the bivariate nonlinear relationship mapping into the spatiotemporally discretized chromatographic column axial transport grid. Using the column outlet proton concentration corresponding to the target linear pH gradient as the terminal boundary condition, iterate back synchronously along the column axis and time axis in reverse. By querying the mapping, obtain the instantaneous net proton exchange rate at each point and solve the discretized convection, diffusion, and adsorption coupling equations. The sequence of column inlet proton concentration changes with time is derived as the pre-distorted pH gradient program. The specific operation is as follows:

[0080] The reconstructed bivariate nonlinear relationship is embedded into the spatiotemporally discretized axial transport grid of the chromatographic column. Through reverse synchronous iterative backtracking, the sequence of column inlet proton concentration changes over time is derived. First, the chromatographic column is axially discretized, dividing the entire column into several equal-volume micro-elements to form the axial transport grid. Simultaneously, the entire elution time period is divided into several consecutive time steps, achieving time dimension discretization. Spatiotemporal discretization transforms the complex continuous proton transport process into a step-by-step, micro-element-level problem. Subsequently, the column outlet proton concentration corresponding to the target linear pH gradient is used as the terminal boundary condition; that is, the outlet proton concentration of the last axial micro-element is fixed to the target value at each time step, and the process is iteratively backtracked. During the process, the process progresses gradually from the column outlet micro-element to the column inlet micro-element, while simultaneously backtracking along the time axis from the end time to the initial time. At each spatiotemporal node, such as a certain micro-element or a certain time step, the instantaneous net proton exchange rate corresponding to that node is obtained by querying the bivariate nonlinear relation mapping. This rate is then substituted into the discretized convection-diffusion equation for solution, yielding the proton concentration at that node. The convection-diffusion equation can be discretized from the coupled equation set. Through synchronous iterative calculations at each micro-element and time step, the proton concentration values ​​of the column inlet micro-element at all time steps are finally obtained. These values ​​are then arranged in chronological order to form a sequence of column inlet proton concentration changes over time. This sequence is the pre-distorted pH gradient program used to compensate for the dynamic buffer capacity characteristics.

[0081] In a preferred embodiment of the present invention, step 3 is further included: adjusting the mixing ratio of the two limiting buffer solutions at a higher frequency than a preset frequency according to the fluid ratio requirements that vary over time as defined by the pre-distorted pH gradient program, in order to generate the input fluid flow. The specific operation is as follows:

[0082] After obtaining the pre-distorted pH gradient program, the mixing ratio of the two limiting buffers needs to be precisely adjusted to generate an input flow that meets the program requirements. The two limiting buffers refer to buffers with proton concentrations within a preset extreme range. Their mixing ratio directly determines the proton concentration of the output flow, thereby matching the proton concentration requirements that change over time in the pre-distorted pH gradient program. During the operation, the pre-distorted pH gradient program is first analyzed to extract the fluid ratio requirements that change over time. These ratio requirements correspond to the ratio of the two limiting buffers required to generate the target proton concentration flow at different times. Subsequently, the mixing ratio of the two limiting buffers is dynamically adjusted at a frequency higher than the preset frequency. The preset frequency must be higher than the characteristic frequency corresponding to the system flow path response frequency and the concentration change rate to ensure that the adjustment action can respond to the program requirements in a timely manner and avoid concentration fluctuations. Through continuous, high-frequency ratio adjustment, the two limiting buffers are fully mixed, ultimately generating an input flow whose proton concentration changes over time meets the requirements of the pre-distorted pH gradient program.

[0083] Step 3 also includes the following sub-steps:

[0084] Step 31: Analyze the proton concentration change sequence over time in the pre-distorted pH gradient program. For the magnitude and direction of the proton concentration change between adjacent time points in the sequence, calculate the forward-looking action correction based on the preset fluid dynamic delay characteristics of the system flow path. Then, add the forward-looking action correction to the control command sequence before the corresponding time point to form a mixed control command. The specific operation is as follows:

[0085] To offset the impact of flow path hydrodynamic delay on mixing regulation, the pre-distorted pH gradient program needs to be analyzed to generate mixing control instructions with forward-looking corrections. First, the proton concentration change sequence over time in the pre-distorted pH gradient program is systematically analyzed, extracting the change in proton concentration between adjacent time points, i.e., the magnitude and direction of the change (e.g., increase or decrease). Then, the dynamic change pattern of proton concentration at each time point is clarified. Due to the inherent hydrodynamic delay in the flow path—that is, the buffer solution takes time to travel from the mixing device to the column inlet—adjusting the mixing ratio according to the immediate concentration requirement will result in a time difference between the actual liquid concentration entering the column and the target concentration. Therefore, based on the pre-set hydrodynamic delay characteristics of the system flow path, the forward-looking action correction amount corresponding to each time point needs to be calculated in advance using experimentally calibrated delay times. The logic of this correction amount is to predict the required concentration adjustment range within the delay time in advance, ensuring that the liquid concentration entering the column after the delay accurately matches the target requirement at the corresponding time. The forward-looking action correction amount can be calculated using the formula: The derivation of this formula is based on the hydrodynamic delay time τ, combined with the concentration change rate at time t+τ. The magnitude of the advance adjustment is quantified by a proportionality coefficient k, where k is a preset calibration coefficient, which is dimensionless. Finally, the quantitative relationship between the advance action correction amount and the delay characteristics and concentration change law is abstracted and obtained. The forward action correction at time t is dimensionless and corresponds to the adjustment range of the buffer mixing ratio. This represents the change in proton concentration between time t+τ and time t, expressed in mol / L. The time interval is in seconds. After calculating the forward action correction amount at each time point, it is added to the control command sequence τ time in advance before the corresponding time point. The transmission delay is offset by the advance issuance of the command, and finally a hybrid control command is formed.

[0086] Step 32: Execute the hybrid control command at a frequency higher than the preset frequency to generate the input fluid flow, obtain the real-time feedback signal of the input fluid flow, compare the actual state represented by the real-time feedback signal with the expected state of the hybrid control command to generate a real-time compensation signal, and superimpose the real-time compensation signal onto the hybrid control command to form the final execution command. The specific operation is as follows:

[0087] To ensure the input fluid flow matches the pre-distorted pH gradient program requirements, control commands need to be optimized through high-frequency execution and real-time feedback compensation mechanisms. First, mixing control commands are executed at a higher frequency than preset. This frequency must meet the concentration response speed requirements after mixing the two limiting buffer solutions, ensuring that each adjustment action is quickly translated into a change in fluid flow concentration, avoiding concentration deviations caused by adjustment lag. During command execution, feedback signals from the input fluid flow are collected in real-time through detection devices deployed at the mixing device outlet or flow path nodes. These feedback signals contain core parameters such as real-time proton concentration and flow rate, providing a direct representation of the actual state of the input fluid flow. Subsequently, the real-time feedback signals are... The actual state of the characterization is compared with the desired state preset in the mixing control command, corresponding to the concentration requirements of the pre-distorted pH gradient program. Then, the two are compared time-by-time to calculate the deviation value. Based on the deviation value, a real-time compensation signal is generated. The magnitude and direction of the compensation signal are determined according to the deviation characteristics. It is used to offset the concentration deviation caused by instantaneous small fluctuations, such as fluctuations in buffer supply pressure and mixing device adjustment errors. Finally, the real-time compensation signal is superimposed on the original mixing control command to form the final execution command. Through dynamic correction of the command, it is ensured that the actual state of the input liquid flow continuously matches the desired state, providing a stable and accurate injection liquid flow for the chromatographic column and ensuring the stability and accuracy of pH control.

[0088] In a preferred embodiment of the present invention, step 4 is further included, in which the input liquid stream is introduced into the chromatographic column, and the dynamic buffering capacity characteristic of the chromatographic column acts on the input liquid stream. The specific operation is as follows:

[0089] The input liquid stream generated after step 3 needs to be continuously introduced into the chromatographic column for online separation. During this process, the dynamic buffer capacity of the chromatographic column will interact with the input liquid stream, directly affecting the transfer and distribution of protons in the liquid stream. The input liquid stream carries a proton concentration distribution characteristic that conforms to the pre-distorted pH gradient program. After entering the chromatographic column, the protons in the liquid stream will undergo adsorption or desorption with components such as the stationary phase and stagnant liquid in the chromatographic column. The intensity of this interaction is determined by the dynamic buffer capacity of the chromatographic column, which has been quantified and stored in the parameter spectrum through the disturbance and response test in step 1. As the liquid stream advances along the axial direction of the chromatographic column, the proton concentration at each axial position in the column will change regularly due to the effect of the dynamic buffer capacity. This change is not a simple linear decrease or increase, but is dynamically correlated with parameters such as the proton loading state and liquid flow rate at each position, ultimately resulting in a specific proton concentration distribution at the column outlet. Throughout the process, it is necessary to keep the process parameters such as the flow rate and temperature of the input liquid stream stable to avoid non-target factors interfering with the interaction effect between the dynamic buffer capacity and the liquid stream.

[0090] Step 4 also includes the following sub-steps:

[0091] Step 41: Using bivariate nonlinear mapping, based on the proton concentration and real-time flow rate of the input liquid flow at the column inlet over time, the instantaneous change rate of proton saturation of each axial element of the chromatographic column is calculated simultaneously, dynamically generating the spatial distribution state of the proton loading wave. The specific operation is as follows:

[0092] To monitor the dynamic changes in proton distribution within the chromatographic column in real time, the bivariate nonlinear relationship mapping reconstructed in step 23 is used, combined with the parameters of the input liquid flow, to calculate the instantaneous rate of change of proton saturation in each axial micro-element and generate the spatial distribution state of the proton loading wave. First, the chromatographic column is divided into several continuous micro-elements along the axial direction. The volume, stationary phase content, and other parameters of each micro-element are known and can be determined through the column structure parameters and discretization processing. The micro-elements are interconnected without gaps, which can accurately characterize the local characteristics of the axial direction of the chromatographic column. Subsequently, the proton concentration at the column inlet of the input liquid flow is collected in real time as a function of time, which can be obtained through the feedback detection in step 3. The real-time flow rate can also be collected by a flow meter. These two parameters are synchronously input into the bivariate nonlinear relationship mapping. Based on the previously obtained dynamic buffer capacity characteristics, this mapping can output the instantaneous net proton exchange rate corresponding to the micro-element at the injection port. Then, the instantaneous net proton exchange rate of each subsequent micro-element is deduced through the axial transmission law.

[0093] The instantaneous rate of change of proton saturation is expressed by the formula:

[0094] ;

[0095] The derivation logic of this formula is as follows: based on the conservation of proton mass within a micro-element, the instantaneous change in proton saturation within the micro-element originates from the net amount of proton exchange. The net rate of proton exchange per unit volume is converted into the rate of change of proton saturation, and this quantitative relationship is abstracted. The meanings of the symbols in the formula are as follows: The instantaneous rate of change of proton saturation, in units of ; The instantaneous net proton exchange rate of the infinitesimal element, in units of . , It can be obtained by transforming the formula in step 22; A is the cross-sectional area of ​​the chromatographic column, in units of... ; The volume of a single axial infinitesimal element, in units of ; The maximum proton adsorption capacity of the stationary phase, in units of By calculating the instantaneous change rate of proton saturation for each axial micro-element at each time step, and combining the spatial position information of each micro-element, the spatial distribution of the proton loading wave in the chromatographic column can be dynamically generated, intuitively presenting the propagation and change law of protons in the column.

[0096] Step 42: Capture the secondary physical signal at the column outlet that is sensitive to the spatial distribution of the proton loading wave. Perform real-time comparison and interlocking verification between the actual trajectory of the secondary physical signal and the predicted trajectory based on the spatial distribution of the proton loading wave. The specific operations are as follows:

[0097] To verify the authenticity and accuracy of the spatial distribution of the proton-loaded wave, it is necessary to capture secondary physical signals and perform real-time comparison and interlocking verification. First, a dedicated detection device is deployed at the column outlet to capture secondary physical signals sensitive to the spatial distribution of the proton-loaded wave. These signals are usually derived signals related to proton concentration, such as conductivity signals and ultraviolet absorption signals. Their variation patterns can indirectly reflect the morphology and intensity of the proton-loaded wave at the outlet, and their response sensitivity is higher than in some scenarios of direct pH detection. Subsequently, based on the spatial distribution of the proton-loaded wave dynamically generated in step 41, combined with the boundary conditions at the column outlet, such as flow rate and diffusion coefficient, the expected changes in the secondary physical signals are predicted using a physical transfer model. The trajectory, which is the theoretically expected signal characteristic of the proton load wave after reaching the exit, is compared in real time with the actual trajectory of the secondary physical signal captured by the detection equipment. The deviation between the two is calculated and it is determined whether the deviation is within the preset allowable range. At the same time, interlocking verification logic is executed. That is, when the deviation exceeds the allowable range, it is determined that there is an anomaly in the calculation of the spatial distribution state of the proton load wave or the signal detection, and the subsequent signal correction or parameter verification process is triggered in time. When the deviation is within the allowable range, the accuracy of the spatial distribution state of the proton load wave is confirmed. This real-time comparison and interlocking verification mechanism can effectively avoid the error of a single detection method and ensure accurate monitoring of the proton transport process in the column.

[0098] In a preferred embodiment of the present invention, step 5 is further included: obtaining the actual pH curve of the chromatographic column output solution; and adjusting the diffusion coefficient parameter in the physical transport model based on the deviation between the actual pH curve and the target pH gradient. The specific operation is as follows:

[0099] By acquiring the actual pH curve of the column output liquid, the deviation from the target pH gradient is quantified, and the diffusion coefficient parameter in the physical transport model is adjusted to correct the deviation between the model and the actual proton transport process in the column, thereby improving the generation accuracy of the pre-distorted pH gradient program. First, the corrected actual pH curve is obtained through signal reliability judgment and soft measurement reconstruction. Then, the systematic deviation of the model is analyzed by inversion correlation calculation to generate a model error spectrum. Subsequently, the diffusion coefficient correction amount is allocated according to the axial distribution characteristics of the error spectrum. Finally, the initial value of the diffusion coefficient of the future model is optimized by combining historical data.

[0100] Obtaining the actual pH profile of the column output solution includes the following steps:

[0101] Step 51: Determine the confidence level of the original pH measurement signal based on the real-time comparison and interlock verification results. According to the confidence level, when it is determined to be low confidence, the proton concentration soft measurement algorithm based on bivariate nonlinear relationship mapping and real-time flow rate is activated to reconstruct the low confidence segment of the original pH measurement signal to generate a corrected pH curve. The specific operation is as follows:

[0102] Based on the real-time comparison and interlock verification results of step 42, the credibility level of the original pH measurement signal is determined, and the signal is then corrected according to the determination results. The real-time comparison and interlock verification results are based on the deviation between the actual trajectory and the expected trajectory of the secondary physical signal. A preset deviation threshold is used as the credibility level classification standard. When the deviation value is less than or equal to the threshold, the original pH measurement signal is determined to be of high credibility and can be directly used as the basis for subsequent analysis. When the deviation value is greater than the threshold, it is determined to be of low credibility, and the proton concentration soft measurement algorithm needs to be used for signal reconstruction. The soft measurement algorithm takes the bivariate nonlinear relationship mapping reconstructed in step 23, the real-time collected liquid flow velocity, and the spatial distribution state of the proton load wave generated in step 41 as inputs. It simulates and calculates the proton concentration in the low credibility segment through the physical transfer model, and then converts the proton concentration into the corresponding pH value, replaces the low credibility segment in the original pH measurement signal, and finally integrates the high credibility segment and the reconstructed segment to form the corrected pH curve.

[0103] Step 52: Perform inversion correlation calculation between the corrected pH curve and the spatial distribution of the proton load wave. By adjusting the local transport resistance parameter, minimize the deviation between the simulated column outlet pH curve output by the model and the corrected pH curve, and obtain the model error spectrum characterizing the systematic bias of the model. The specific operation is as follows:

[0104] The corrected pH curve is inverted and correlated with the spatial distribution of the proton loading wave generated in step 41. The proton transport process within the column is inferred from the spatial distribution characteristics of the proton loading wave, thereby simulating the column outlet pH curve. This simulation is then compared with the corrected pH curve to optimize the model parameters. The spatial distribution of the proton loading wave directly determines the axial transport and adsorption / desorption behavior of protons within the chromatographic column. Based on this, the local transport resistance parameter in the physical transport model is used as the variable to be optimized. The objective is to minimize the deviation between the simulated column outlet pH curve output by the model and the corrected pH curve. The optimal local transport resistance parameter is solved using numerical optimization methods. The objective function of this optimization process is:

[0105] ;

[0106] The derivation process is as follows: using the sum of squared deviations of proton concentration over time as the deviation quantification index, the objective function is obtained by minimizing this index value to achieve the optimal fit between the model simulation results and the actual calibration data. In the formula: J is the deviation optimization objective function, which is dimensionless; N is the number of time sampling points of the pH curve, which is dimensionless. The column outlet proton concentration at time t is the output of the model, in units of ; The proton concentration corresponding to the corrected pH curve at time t, in units of After optimization, the deviations between the simulated and corrected values ​​of the model at each axial micro-element are compiled to form a model error spectrum that characterizes the systematic deviation of the physical transfer model. This spectrum uses the axial coordinates of the chromatographic column as the horizontal axis and the deviation value as the vertical axis, clearly showing the deviation distribution characteristics of the model at different axial positions.

[0107] The adjustment of the diffusion coefficient parameter in the physical transport model based on the deviation between the actual pH curve and the target pH gradient includes the following steps:

[0108] Step 53: Utilizing the distribution characteristics of the model error chromatogram along the column axis, and based on the preset mapping rule between the error distribution pattern and the diffusion coefficient correction amount, allocate diffusion coefficient increments or decrements to different axial segments to generate a diffusion coefficient correction distribution. The specific operation is as follows:

[0109] By utilizing the axial distribution characteristics of the model error spectrum and combining the pre-defined mapping rules between the error distribution pattern and the diffusion coefficient correction amount, the diffusion coefficient increment or decrement is assigned to different axial segments of the chromatographic column, ultimately generating a diffusion coefficient correction distribution. The axial distribution characteristics of the model error spectrum include the magnitude and direction of the deviation, i.e., positive or negative deviation; and the distribution pattern, such as uniform distribution, local peaks, gradient distribution, etc. The pre-defined mapping rules are determined through preliminary experimental calibration and theoretical analysis, clarifying the magnitude and direction of the diffusion coefficient correction amount corresponding to different error distribution patterns. For example, when a positive deviation occurs in a certain axial segment, if the model simulates a proton concentration higher than the actual correction value, a diffusion coefficient decrement is assigned to reduce the proton diffusion rate; when a negative deviation occurs, an increment is assigned. The larger the absolute value of the deviation, the larger the absolute value of the corresponding correction amount. Based on this mapping rule, all axial micro-elements of the model error spectrum are traversed, and the corresponding diffusion coefficient correction amount is calculated for each micro-element. After integrating the correction amounts of all micro-elements, a diffusion coefficient correction distribution distributed along the chromatographic column axis is formed. This distribution will be directly used to update the diffusion coefficient parameters in the physical transport model and correct the proton transport simulation deviation of the model.

[0110] Step 54: Store the diffusion coefficient correction distribution, model error spectrum, and process conditions in the historical database. Based on the characteristics of the diffusion coefficient correction distribution and combined with similar data in the historical database, dynamically optimize the initial preset values ​​of the diffusion coefficient parameters of the physical transfer model in future similar separation tasks through progressive weighted fusion rules. The specific operations are as follows:

[0111] The generated diffusion coefficient correction distribution, model error spectrum, and process conditions for this separation task, including liquid flow rate, column temperature, buffer type and concentration, and sample characteristics, are simultaneously stored in a historical database to form a structured historical data record. This provides data support for optimizing model parameters in subsequent similar separation tasks. Based on the characteristics of the diffusion coefficient correction distribution, such as the axial distribution pattern and correction magnitude, similar data records with similar process conditions are selected from the historical database. A progressive weighted fusion rule is used to fuse these data for calculation, which can dynamically optimize the initial preset value of diffusion coefficient parameters for physical transport models in future similar separation tasks. The core logic of the progressive weighted fusion rule is to assign higher weights to recent data and data with more significant correction magnitudes, and lower weights to historical data and data with smaller correction magnitudes. The optimized initial preset value of diffusion coefficient is calculated by weighted averaging. This initial value is closer to the actual proton transport characteristics in the column than the original preset value, which can effectively reduce the adjustment range of model parameters in subsequent cycles and improve the initial simulation accuracy of the physical transport model.

[0112] In a preferred embodiment of the present invention, step 6 is further included, which involves sequentially executing steps 1 to 5, as follows:

[0113] After the initial completion of steps 1 to 5, an iterative loop is initiated. During the loop, the established operational logic and technical parameter requirements of steps 1 to 5 are strictly followed. The entire process—proton perturbation testing and parameter spectrum acquisition, pre-distortion pH gradient program generation, buffer mixing and input flow generation, flow introduction and proton loading wave monitoring, actual pH curve correction, and model parameter adjustment—is repeated sequentially. The optimization logic for each loop is that the adjusted physical transfer model diffusion coefficient parameters, the generated model error spectrum, and the diffusion coefficient correction distribution from the previous step (step 5) are used as the initial input parameters for steps 1 to 5 in this loop. In step 1, when reapplying the proton perturbation signal, the application parameters can be optimized based on the spatial distribution of the proton loading wave from the previous loop to obtain a parameter spectrum that better reflects the actual dynamic buffer capacity characteristics of the current column. Step 2 utilizes the updated parameter spectrum and the optimized model parameters… Step 1 generates a pre-distorted pH gradient program with better compensation effect; Steps 3 and 4 perform regulation and liquid flow introduction based on a more precise gradient program, further improving the proton concentration matching degree of the input liquid flow; Step 5, based on the actual pH curve that is closer to the target in this round, makes more refined adjustments to parameters such as the diffusion coefficient of the physical transport model, continuously reducing the deviation between the model simulation and the actual proton transport process in the column; The termination condition of the cycle iteration can be preset according to the actual separation requirements, including the deviation value between the actual pH curve and the target pH gradient is less than the preset allowable threshold, the parameter adjustment range of multiple consecutive cycles is less than the preset minimum value, or the preset maximum number of cycles is reached. When any termination condition is met, the cycle iteration stops. At this time, the pre-distorted pH gradient program and the adjusted physical transport model parameters can ensure that the pH gradient of the effluent after the column accurately matches the target requirements, realizing high-precision and high-stability online column separation pH control.

[0114] Example 2

[0115] Please see Figure 2 Based on Example 1, this example provides a real-time pH control and stabilization system for online column separation, including: a pre-distortion generation module that inputs parameter spectra and target pH gradients into a physical transport model based on convection, diffusion and adsorption coupling equations, and the physical transport model outputs a pre-distortion pH gradient program to compensate for dynamic buffer capacity characteristics.

[0116] The gradient generation module adjusts the mixing ratio of the two limiting buffer solutions at a higher frequency than preset, based on the fluid ratio requirements that change over time as defined by the pre-distorted pH gradient program, in order to generate the input fluid flow.

[0117] The column injection module introduces the input liquid stream into the chromatographic column, and the dynamic buffer capacity of the chromatographic column acts on the input liquid stream.

[0118] The calibration module acquires the actual pH curve of the column output liquid and adjusts the diffusion coefficient parameter in the physical transport model based on the deviation between the actual pH curve and the target pH gradient.

[0119] The iterative module sequentially executes the operations in the column characterization module, pre-distortion generation module, gradient generation module, column injection module, and correction module.

[0120] The above description is merely a preferred embodiment of the present invention; however, the scope of protection of the present invention is not limited thereto. Any equivalent substitutions or modifications made by those skilled in the art within the scope of the technology disclosed in the present invention, based on the technical solution and its improved concepts, should be covered within the scope of protection of the present invention.

Claims

1. A method for real-time pH control and stabilization in online column separation, characterized in that, include: Step 1: Apply a proton perturbation signal to the chromatographic column and detect its elution response signal to obtain a parameter spectrum characterizing the dynamic buffer capacity of the chromatographic column; Step 2: Input the parameter spectrum and target pH gradient into the physical transport model based on the convection, diffusion and adsorption coupling equations. The physical transport model outputs a pre-distorted pH gradient program to compensate for dynamic buffer capacity characteristics. Step 3: Based on the fluid ratio requirements that vary over time as defined by the pre-distorted pH gradient program, adjust the mixing ratio of the two limiting buffer solutions at a frequency higher than the preset frequency to generate the input fluid flow. Step 4: The input liquid stream is introduced into the chromatographic column, and the dynamic buffer capacity characteristic of the chromatographic column acts on the input liquid stream; Step 5: Obtain the actual pH curve of the column output solution. Based on the deviation between the actual pH curve and the target pH gradient, adjust the diffusion coefficient parameter in the physical transport model, including: The credibility level of the original pH measurement signal is determined based on the real-time comparison and interlock verification results. According to the credibility level, when it is determined to be low credibility, the proton concentration soft measurement algorithm based on bivariate nonlinear relationship mapping and real-time flow rate is activated to reconstruct the low credibility segment of the original pH measurement signal to generate the corrected pH curve. The corrected pH curve is inverted and correlated with the spatial distribution of the proton load wave. By adjusting the local transmission resistance parameter, the deviation between the simulated pH curve at the column outlet output by the model and the corrected pH curve is minimized, and the model error spectrum characterizing the systematic deviation of the model is obtained. By utilizing the distribution characteristics of the model error spectrum along the chromatographic column axis, and based on the mapping rule between the preset error distribution pattern and the diffusion coefficient correction amount, diffusion coefficient increments or decrements are assigned to different axial segments to generate a diffusion coefficient correction distribution. The diffusion coefficient correction distribution, model error spectrum, and process conditions are stored in a historical database. Based on the characteristics of the diffusion coefficient correction distribution and combined with similar data in the historical database, the initial preset values ​​of the diffusion coefficient parameters of the physical transfer model in future similar separation tasks are dynamically optimized through progressive weighted fusion rules. Step 6: Repeat steps 1 through 5 in sequence.

2. The method for real-time pH control and stabilization for online column separation according to claim 1, characterized in that, Obtain parameter chromatograms characterizing the dynamic buffer capacity of the chromatographic column, including: Step 11: Apply a proton perturbation signal containing a continuous proton concentration increase segment and a concentration decrease segment to the chromatographic column to obtain a comprehensive response signal; Step 12: Based on the known input flux of the proton perturbation signal, the comprehensive response signal is synchronously reverse-calculated based on the material conservation principle to obtain the real-time adsorption proton flux and the real-time desorption proton flux.

3. The method for real-time pH control and stabilization for online column separation according to claim 2, characterized in that, Obtaining a parameter spectrum characterizing the dynamic buffer capacity of the chromatographic column further includes: step 13, mapping the real-time adsorption proton flux and the real-time desorption proton flux to a two-dimensional relationship plane with the current proton loading state of the chromatographic column and the net proton exchange rate as coordinates, and generating a two-dimensional dynamic rate spectrum.

4. The method for real-time pH control and stabilization for online column separation according to claim 3, characterized in that, The parametric profiles and target pH gradients are input into a physical transport model based on the coupling equations of convection, diffusion, and adsorption, including: Step 21: The dynamic rate relationship in the two-dimensional dynamic rate spectrum is transformed into a dynamic function distributed along the chromatographic column axis. The dynamic function defines the relationship between the local net proton exchange rate and the proton concentration and proton saturation of the liquid flow. Step 22: Construct a physical transport model with a reverse solution structure based on dynamic functions. Use the target column post-column linear pH gradient as the endpoint boundary condition of the physical transport model. Solve the convection, diffusion and adsorption coupling equations in reverse to obtain the trajectory of the column inlet proton concentration changing with time as a pre-distorted pH gradient program.

5. The method for real-time pH control and stabilization for online column separation according to claim 4, characterized in that, The physical transport model outputs a pre-distorted pH gradient program to compensate for dynamic buffer capacity characteristics, including: Step 23: The relationship represented by the two-dimensional dynamic rate map is analyzed and reconstructed into a bivariate nonlinear relationship mapping with the first variable being the current proton loading state of the chromatographic column and the second variable being the proton concentration of the liquid flow through the micro-element of the chromatographic column. The instantaneous net proton exchange rate of the micro-element is then mapped out. Step 24: The bivariate nonlinear relationship mapping is embedded into the spatiotemporally discretized chromatographic column axial transport grid. The column outlet proton concentration corresponding to the target linear pH gradient is used as the end boundary condition. The synchronous backtracking is performed in reverse along the column axis and time axis. The instantaneous net proton exchange rate at each point is obtained by querying the mapping, and the discretized convection, diffusion and adsorption coupling equations are solved. The sequence of column inlet proton concentration changes with time is derived as the pre-distorted pH gradient program.

6. The method for real-time pH control and stabilization for online column separation according to claim 5, characterized in that, Based on the time-varying fluid ratio requirements defined by the pre-distorted pH gradient program, the mixing ratio of the two limiting buffer solutions is adjusted at a frequency higher than a preset frequency to generate the input fluid flow, including: Step 31: Analyze the proton concentration change sequence over time in the pre-distorted pH gradient program. For the change in proton concentration between adjacent time points in the sequence, calculate the forward action correction amount based on the preset fluid dynamic delay characteristics of the system flow path, and add the forward action correction amount to the control command sequence before the corresponding time point to form a mixed control command. Step 32: Execute the hybrid control command at a frequency higher than the preset frequency to generate the input fluid flow, obtain the real-time feedback signal of the input fluid flow, compare the actual state represented by the real-time feedback signal with the expected state of the hybrid control command to generate a real-time compensation signal, and superimpose the real-time compensation signal onto the hybrid control command to form the final execution command.

7. The method for real-time pH control and stabilization for online column separation according to claim 6, characterized in that, The input liquid stream is introduced into the chromatographic column, and the dynamic buffering capacity characteristics of the column affect the input liquid stream, including: Step 41: Using bivariate nonlinear relationship mapping, based on the proton concentration and real-time flow rate of the input liquid flow at the column inlet over time, the instantaneous change rate of proton saturation of each axial micro-element of the chromatographic column is calculated simultaneously, and the spatial distribution state of the proton loading wave is dynamically generated. Step 42: Capture the secondary physical signal at the column outlet that is sensitive to the spatial distribution of the proton load wave, and compare and interlock the actual change trajectory of the secondary physical signal with the expected change trajectory predicted based on the spatial distribution of the proton load wave in real time.

8. A real-time pH control and stabilization system for online column separation, applied to the real-time pH control and stabilization method for online column separation as described in any one of claims 1-7, characterized in that, include: The column characterization module applies a proton perturbation signal to the chromatographic column and detects its elution response signal to obtain a parameter spectrum characterizing the dynamic buffer capacity of the chromatographic column. The pre-distortion generation module inputs the parameter spectrum and the target pH gradient into a physical transport model based on the convection, diffusion and adsorption coupling equations. The physical transport model outputs a pre-distortion pH gradient program to compensate for dynamic buffer capacity characteristics. The gradient generation module adjusts the mixing ratio of the two limiting buffer solutions at a higher frequency than preset, based on the fluid ratio requirements that change over time as defined by the pre-distorted pH gradient program, in order to generate the input fluid flow. The column injection module introduces the input liquid stream into the chromatographic column, and the dynamic buffer capacity of the chromatographic column acts on the input liquid stream. The calibration module acquires the actual pH curve of the column output liquid and adjusts the diffusion coefficient parameter in the physical transport model based on the deviation between the actual pH curve and the target pH gradient. The iterative module sequentially executes the operations in the column characterization module, pre-distortion generation module, gradient generation module, column injection module, and correction module.

Images