Medical intermediate synthesis real-time monitoring method and system

Abstract

The invention provides a real-time monitoring method and system for medical intermediate synthesis, which are applied to the technical field of medical production monitoring, whether a co-elution peak exists or not is judged by acquiring spectral data of an elution peak and comparing the spectral data with a reference spectrum, and process control is adjusted according to a judgment result, so that the medical intermediate synthesis quality is improved. Therefore, the problems of erroneous judgment and erroneous control caused by the fact that single-wavelength detection cannot recognize the co-outflow peak in the prior art are solved, and the method has the advantages that the co-outflow peak can be recognized, erroneous judgment is avoided, and more accurate process control is achieved.

Description

Technical Field

[0001] The present application relates to the technical field of pharmaceutical production monitoring, and in particular to a method and system for real-time monitoring of pharmaceutical intermediate synthesis. Background Art

[0002] In the production of pharmaceutical intermediates, real-time, meticulous process monitoring of core synthesis reactions is crucial to ensure high product purity and improve batch stability. Traditional offline sampling and analysis methods, such as sending samples to the laboratory for high-performance liquid chromatography (HPLC) analysis, have significant time lags. The long time between sampling and obtaining analytical results makes it difficult for production control to achieve real-time dynamic response to reaction processes.

[0003] To overcome the lag inherent in offline analysis, modern pharmaceutical production facilities have introduced online HPLC monitoring systems. These systems automatically extract samples from the reaction system, pre-treat them online, and then inject them into a chromatographic system for separation and detection. Using pre-set chromatographic conditions, the system can monitor concentrations of the target product and known key impurities. For example, the concentration of a key impurity must be strictly controlled at extremely low levels. If it shows signs of exceeding control, immediate action is required to adjust process parameters to suppress its formation.

[0004] However, existing online chromatography monitoring systems are typically equipped with a single-wavelength UV-visible detector. This detector operates at a specific wavelength and identifies and quantifies the target component by analyzing the retention time and peak area of ​​the chromatographic peak. This identification method based on retention time and single-wavelength signals has limitations when dealing with complex reaction systems. In actual production processes, new, unknown impurities may be generated due to differences in raw material batches or unexpected reaction pathways. If these unknown impurities have similar polarity to known key impurities under the existing optimized chromatographic conditions, they may not be effectively separated, resulting in chromatographic co-elution.

[0005] When an unknown impurity co-elutes with a known key impurity, the signal received by the single-wavelength detector is the superposition of the two at a single wavelength. The system's data processing software cannot distinguish between the two components and can only integrate the superimposed peak as a whole, and mistakenly attribute the total peak area to the known key impurity. This leads to a misjudgment of the concentration of the known key impurity. Based on this misjudged concentration data, the automated process control system may trigger an erroneous control response, such as mistakenly executing an emergency interruption reaction or drastically adjusting process parameters when the known key impurity does not actually exceed the standard. This erroneous decision may not only lead to a decrease in the yield of the target product and economic losses, but may also introduce new process instability factors.

[0006] In view of the above problems, the existing technology is in urgent need of improvement. Summary of the Invention

[0007] In view of the above-mentioned deficiencies in the prior art, the present application provides a real-time monitoring method and system for the synthesis of pharmaceutical intermediates, which has the advantages of being able to identify co-elution peaks, avoid misjudgment, and thus achieve more accurate process control.

[0008] Furthermore, a method for real-time monitoring of the synthesis of a pharmaceutical intermediate is provided, the method comprising the steps of:

[0009] S1: When an elution peak of a target analyte is detected within a preset retention time window, acquiring spectral data of the elution peak;

[0010] S2: comparing the spectral data with preset reference spectral data of the target analyte to obtain a comparison result;

[0011] S3: judging whether the elution peak is a co-elution peak containing an unknown component according to the comparison result, and obtaining a judgment result;

[0012] S4: When the judgment result is that the elution peak is a co-elution peak, adjusting the automated process control response triggered based on the elution peak.

[0013] This application proposes a real-time monitoring method for pharmaceutical intermediate synthesis, using spectral analysis to identify coelution and adjust automated control. Its core approach is to utilize the spectral characteristics of substances to compensate for the shortcomings of chromatographic separation, thereby avoiding misjudgment and miscontrol caused by coelution.

[0014] Furthermore, step S1 includes:

[0015] S11: when the elution peak is detected within the preset retention time window, collecting spectral data corresponding to a plurality of time points within the duration of the elution peak to form a spectral data set;

[0016] S12: determining the signal intensity corresponding to each spectral data in the spectral data set;

[0017] S13: performing weighted processing on the spectral data set based on the signal intensity corresponding to each spectral data to generate a target spectral data;

[0018] S14: Using the target spectrum data as the acquired spectrum data of the elution peak.

[0019] The present application proposes a real-time monitoring method for the synthesis of pharmaceutical intermediates, which improves the representativeness and accuracy of spectral data by weighted processing of spectral data.

[0020] Furthermore, step S13 includes:

[0021] S131: Acquire baseline spectral data in a time region before and after the duration of the elution peak;

[0022] S132: determining, based on the baseline spectral data, a signal change amount caused by baseline drift during the duration of the elution peak;

[0023] S133: subtracting the signal variation from each spectral data in the spectral data set to obtain the corrected spectral data set;

[0024] S134: performing weighted processing on the corrected spectral data set based on the signal intensity corresponding to each spectral data to generate the target spectral data.

[0025] The present application proposes a real-time monitoring method for the synthesis of pharmaceutical intermediates, which eliminates the influence of baseline drift on spectral data through baseline correction, thereby further improving the accuracy of spectral data.

[0026] Furthermore, step S2 includes:

[0027] S21: generating corresponding derivative spectra according to the spectral data and the reference spectral data respectively;

[0028] S22: Comparing the generated derivative spectra to generate the comparison result.

[0029] The present application proposes a real-time monitoring method for the synthesis of pharmaceutical intermediates. By comparing derivative spectra, it is possible to more sensitively detect minute differences in spectra and improve the ability to identify co-eluting peaks.

[0030] Furthermore, step S21 includes:

[0031] S211: Obtain corresponding spectra according to the spectrum data and the reference spectrum data respectively;

[0032] S212: selecting a data window including each wavelength point in the spectrum;

[0033] S213: fitting the data in the data window using a polynomial function to determine a fitting function;

[0034] S214: Calculate the derivative value at the wavelength point based on the fitting function, and construct the derivative spectrum with the derivative values ​​corresponding to all wavelength points.

[0035] Furthermore, step S3 includes:

[0036] S31: Compare the comparison result with a preset judgment threshold;

[0037] S32: When the spectral difference indicated by the comparison result exceeds the judgment threshold, determining that the judgment result is that the elution peak is a co-elution peak.

[0038] Furthermore, step S4 includes:

[0039] S41: determining the number of consecutive analysis cycles in which the judgment result is a co-elution peak before the current analysis cycle, and obtaining a number of consecutive cycles;

[0040] S42: Execute a preset adjustment response corresponding to the number of consecutive cycles according to the number of consecutive cycles.

[0041] Furthermore, step S42 includes:

[0042] S421: when the determination result changes from a co-elution peak to a non-co-elution peak, starting a fault recovery period;

[0043] S422: Determining a target response level for executing the adjustment response, wherein, when the judgment result is that the current analysis cycle of the co-elution peak is within the fault recovery period, determining an increased response level as the target response level based on the number of consecutive cycles;

[0044] S423: When the current analysis period is not within the fault recovery period, determining a basic response level corresponding to the number of consecutive periods as the target response level;

[0045] S424: Execute an adjustment instruction according to the target response level to adjust the automated process control response.

[0046] Furthermore, step S423 includes:

[0047] S4231: Determine an initial adjustment instruction for a first process parameter based on the target response level;

[0048] S4232: Determine one or more second process parameters affected by the adjustment of the first process parameter according to a preset process parameter association relationship;

[0049] S4233: determining, for one or more of the second process parameters, a compensation adjustment instruction for offsetting a disturbance caused by the initial adjustment instruction;

[0050] S4234: Coordinately execute the initial adjustment instruction and the compensation adjustment instruction to adjust the automated process control response.

[0051] In a second aspect, a real-time monitoring system for the synthesis of pharmaceutical intermediates is provided, for implementing any of the above methods, the system comprising:

[0052] Acquisition module: when detecting an elution peak of the target analyte within a preset retention time window, acquiring spectral data of the elution peak;

[0053] Comparison module: compares the spectral data with the preset reference spectral data of the target analyte to obtain a comparison result;

[0054] A judgment module: judging whether the elution peak is a co-elution peak containing an unknown component according to the comparison result, and obtaining a judgment result;

[0055] Adjustment module: When the judgment result is that the elution peak is a co-elution peak, adjust the automated process control response triggered by the elution peak.

[0056] Beneficial effects: The present application proposes a method and system for real-time monitoring of the synthesis of pharmaceutical intermediates. By acquiring the spectral data of the elution peak and comparing it with the reference spectrum, it is determined whether there is a co-elution peak, and the process control is adjusted according to the judgment result, thereby solving the problem in the prior art that single-wavelength detection cannot identify the co-elution peak, leading to misjudgment and erroneous control. It has the advantage of being able to identify the co-elution peak, avoid misjudgment, and thus achieve more accurate process control. BRIEF DESCRIPTION OF THE DRAWINGS

[0057] Figure 1 This is a flow chart of a real-time monitoring method for the synthesis of pharmaceutical intermediates proposed in this application.

[0058] Figure 2 This is a structural diagram of a real-time monitoring system for the synthesis of pharmaceutical intermediates proposed in this application.

[0059] Figure 3 This is an architectural diagram of a real-time monitoring system for pharmaceutical intermediate synthesis proposed in this application.

[0060] Description of reference numerals: 201, acquisition module; 202, comparison module; 203, judgment module; 204, adjustment module. DETAILED DESCRIPTION

[0061] The technical solutions in the embodiments of the present application will be clearly and completely described below in conjunction with the drawings in the embodiments of the present application. Obviously, the described embodiments are only a part of the embodiments of the present application, rather than all of the embodiments. The components of the embodiments of the present application generally described and marked in the drawings here can be arranged and designed in various different configurations. Therefore, the following detailed description of the embodiments of the present application provided in the drawings is not intended to limit the scope of the application for protection, but merely represents the selected embodiments of the present application. Based on the embodiments of the present application, all other embodiments obtained by those skilled in the art without making creative work fall within the scope of protection of the present application.

[0062] It should be noted that similar reference numerals and letters represent similar items in the following figures. Therefore, once an item is defined in one figure, it does not need to be further defined or explained in subsequent figures. At the same time, in the description of this application, the terms "first", "second", etc. are only used to distinguish the description and should not be understood as indicating or implying relative importance.

[0063] Furthermore, a method for real-time monitoring of the synthesis of pharmaceutical intermediates comprises the steps of:

[0064] S1: When an elution peak of a target analyte is detected within a preset retention time window, spectral data of the elution peak is acquired;

[0065] S2: comparing the spectral data with preset reference spectral data of the target analyte to obtain a comparison result;

[0066] S3: Based on the comparison result, determine whether the elution peak is a co-elution peak containing the unknown component, and obtain a determination result;

[0067] S4: When the judgment result is that the elution peak is a co-elution peak, adjust the automated process control response based on the elution peak trigger.

[0068] Step S1 involves collecting absorption information at multiple wavelengths when a signal peak is detected within the expected retention time range of the target analyte during chromatographic analysis. This can be achieved using a multi-wavelength detector, such as a diode array detector, to obtain the spectral characteristics of the elution peak for distinguishing different components.

[0069] Step S2 involves comparing the real-time elution peak spectrum with a pre-established standard spectrum representing the pure target analyte. This comparison method calculates a spectral similarity coefficient to determine the similarity between the real-time spectrum and the target spectrum. This quantifies the degree of difference between the real-time spectrum and the pure target spectrum.

[0070] Step S3 involves determining whether the elution peak is formed by the superposition of multiple components based on the quantitative differences obtained from the spectral comparison and according to preset judgment criteria. The purpose is to identify the purity status of the elution peak, especially whether there are unknown interfering components.

[0071] Step S4 involves modifying or intervening in automated process control operations that would normally be triggered by the elution peak if it is determined to be a coelution peak. These adjustments can include pausing control instructions, issuing an abnormality warning, or switching to manual intervention mode. The goal is to avoid erroneous automated control decisions caused by misjudgment of concentration due to coelution.

[0072] Specifically, the operating logic of this solution is: in the online chromatographic monitoring of the pharmaceutical intermediate synthesis process, when the system detects the appearance of an elution peak of the target analyte within the preset retention time window, it no longer only obtains the signal intensity at a single wavelength, but instead uses a multi-wavelength detector to collect the absorption spectrum data of the elution peak at multiple wavelengths in real time.

[0073] The collected real-time spectral data is then compared with pre-stored reference spectral data representing the pure target analyte, and the similarity or difference between the two is calculated to obtain a comparison result. Based on the comparison result, the system makes a logical judgment. If there is a significant difference between the real-time spectrum and the reference spectrum, the elution peak is determined to be a co-elution peak containing the unknown component.

[0074] Once a coelution peak is identified, the system no longer uses the peak's total area or height to calculate the target analyte concentration and directly trigger the pre-set automated process control response as per the conventional process. Instead, the system adjusts or intervenes in the original automated control response based on the coelution result, such as pausing the execution of the control instruction or issuing an abnormality warning, thereby avoiding incorrect process control operations caused by misjudgment of concentration due to coelution.

[0075] As a preferred embodiment, the solution of the present application is specifically implemented as follows: in the online chromatography monitoring system for the synthesis of pharmaceutical intermediates, the single-wavelength ultraviolet detector is replaced with a diode array detector. When monitoring the key impurity Imp-1, a standard spectrum of pure Imp-1 is pre-established as a reference spectrum. When the online chromatograph detects that Imp-1 has a peak within a preset retention time window, the diode array detector collects the spectral data of the elution peak in real time. The system calculates the spectral similarity coefficient between the real-time spectrum and the Imp-1 reference spectrum. If the similarity is lower than the preset threshold, the system determines that an unknown component co-elutes with the elution peak. Based on this judgment, the system does not trigger the automated temperature adjustment instruction that may be caused by the Imp-1 concentration originally calculated based on the peak area, but instead issues a warning of "Imp-1 peak suspected of co-elution" and recommends that the operator perform offline confirmation analysis while suspending related automatic control.

[0076] Through the above scheme, the present application can identify unknown components that co-elute with known key impurities in real time, solving the problem that the existing single-wavelength system cannot distinguish signals when facing such interference, resulting in concentration misjudgment. By linking the spectral analysis results with the process control logic, the system can avoid triggering erroneous automated process control instructions based on superposition signals, and prevent unnecessary reaction quenching due to misjudgment of exceeding the standard. This improves the robustness and accuracy of the online monitoring system in complex reaction systems, reduces the risk of batch scrapping or yield loss due to analytical misjudgment, and improves the reliability and economy of the production process.

[0077] Furthermore, step S1 includes:

[0078] S11: When an elution peak is detected within a preset retention time window, spectral data corresponding to multiple time points are collected within the duration of the elution peak to form a spectral data set;

[0079] S12: determining the signal intensity corresponding to each spectral data in the spectral data set;

[0080] S13: performing weighted processing on the spectral data set based on the signal intensity corresponding to each spectral data to generate a target spectral data;

[0081] S14: The target spectrum data is used as the spectrum data of the acquired elution peak.

[0082] The spectral data set refers to the collection of spectral data collected at multiple time points during the duration of the elution peak. Signal intensity refers to the intensity of the detector response at a specific time point, such as the absorbance value at a specific wavelength or multiple wavelengths, or the peak height on a chromatogram. The target spectral data refers to the single spectral data obtained after weighted processing, which is used to represent the spectral characteristics of the entire elution peak.

[0083] To more clearly illustrate the method for acquiring elution peak spectral data proposed in this application, an example is provided below. A diode array detector can be used to acquire spectral data at a fixed frequency, for example, once per second, for the duration of the elution peak. Simultaneously, the absorbance value at the wavelength of maximum absorption of the target analyte at each time point can be recorded as the signal intensity.

[0084] Weighted processing can be performed using a weighted average method: the spectral data collected at each time point is multiplied by the signal intensity corresponding to that time point as a weight. All weighted spectral data are then added together and divided by the sum of all signal intensities to obtain the target spectral data. For example, if spectral data S1, S2, ..., Sn are collected at time points t1, t2, ..., tn, and the corresponding signal intensities are I1, I2, ..., In, the target spectral data S_target = (S1*I1+S2*I2+...+Sn*In) / (I1+I2+...+In). This target spectral data S_target will be used for subsequent comparison with the reference spectrum.

[0085] Through the above method, the present application can achieve the following technical effects. By weighting multiple spectral data within the elution peak duration based on signal intensity, the generated spectral data can more accurately reflect the true spectral characteristics of the elution peak, effectively reducing the influence of peak edges or noise. This provides more reliable data input for subsequent spectral comparison and co-elution judgment, thereby improving the accuracy of the entire monitoring method in judging co-elution.

[0086] Furthermore, step S13 includes:

[0087] S131: acquiring baseline spectrum data in a time region before and after the duration of the elution peak;

[0088] S132: determining, based on the baseline spectral data, an amount of signal change caused by baseline drift during the duration of the elution peak;

[0089] S133: subtracting the signal variation from each spectral data in the spectral data set to obtain a corrected spectral data set;

[0090] S134: performing weighted processing on the corrected spectral data set based on the signal intensity corresponding to each spectral data to generate target spectral data.

[0091] Among them, baseline spectral data refers to the spectral signal collected by the detector when there is no target analyte or impurity elution, which mainly reflects background signals such as mobile phase, solvent, and detector noise itself.

[0092] Baseline drift refers to the unexpected, slow change in the baseline signal over time during a chromatographic run; the signal change refers to the amount by which the baseline signal deviates from the ideal value (such as zero or a stable value) during the duration of the elution peak due to baseline drift.

[0093] The corrected spectral dataset refers to the dataset obtained by subtracting the signal variation caused by baseline drift from the original spectral dataset.

[0094] When an elution peak is detected, the detector can continuously collect spectral data within a time window before the onset of the elution peak and a time window after the elution peak has completely eluted. This collected data can be used as baseline spectral data. For example, the time window between a certain time point before the peak retention time and the peak onset time, and the time window between the peak end time and a certain time point after the peak retention time, can be selected. Using this baseline spectral data, methods such as linear interpolation or polynomial fitting can be used to estimate the baseline value corresponding to each time point within the duration of the elution peak.

[0095] Then, the estimated baseline value of the corresponding time point is subtracted from the raw spectral data collected at each time point during the duration of the elution peak, thereby obtaining a baseline-corrected spectral data set. When weighting the corrected spectral data set, the absorbance of the corrected spectrum at a specific wavelength (e.g., the maximum absorption wavelength of the component) at each time point can be calculated as the signal intensity at that time point. Subsequently, using these signal intensities as weights, the corrected spectral data set is weighted averaged or weighted summed to generate the final target spectral data. For example, the higher the signal intensity of the time point, the greater the proportion of the corresponding corrected spectrum in the weighted average.

[0096] Furthermore, step S2 includes:

[0097] S21: Generate corresponding derivative spectra according to the spectral data and the reference spectral data respectively;

[0098] S22: Comparing the generated derivative spectra to generate a comparison result.

[0099] The derivative spectrum refers to the curve that reflects the rate of change of the spectral shape obtained after the original spectral data is mathematically transformed.

[0100] Generating corresponding derivative spectra refers to mathematically processing the original spectral data to obtain its derivative spectra. Comparing the generated derivative spectra refers to performing similarity or difference analysis on the two derivative spectra, such as calculating a correlation coefficient or Euclidean distance between them or comparing the differences in derivative values ​​at specific wavelengths to generate a comparison result.

[0101] In one embodiment, the derivative spectrum can be generated using the Savitzky-Golay smooth differentiation method. For example, the Savitzky-Golay algorithm can be applied to the spectral data and the reference spectral data, respectively, and the appropriate window size and polynomial order can be selected to calculate the corresponding derivative spectrum. When the generated derivative spectra are compared, the correlation coefficient between the two derivative spectra can be calculated. For example, the Pearson correlation coefficient between the derivative spectrum of the real-time spectrum and the derivative spectrum of the reference spectrum is calculated. The calculated correlation coefficient is used as the comparison result. If the correlation coefficient is lower than a preset threshold, it is considered that there is a difference between the two spectra.

[0102] Specifically, assuming that the real-time spectral data is S real , the reference spectrum data is S ref , they have the same wavelength range and the number of data points is n.

[0103] Select the window size m (usually an odd number, such as 5 or 7), the polynomial order p (such as 2 or 3). The coefficients obtained by the Savitzky-Golay algorithm are a j, where j ranges from -k to k,

[0104] For each wavelength point λ i , the calculation formula of the derivative spectrum is:

[0105] Wherein, S represents spectral data, including real-time spectral data and reference spectral data.

[0106] Derivative spectrum corresponding to real-time spectral data: D real =[D real (λ1),D real (λ2),…,D real (λ n )],in,

[0107] Derivative spectrum corresponding to reference spectrum data: D ref =[D ref (λ1),D ref (λ2),…,D ref (λ n )],in,

[0108] By calculating the Pearson correlation coefficient between two derivative spectra, the correlation coefficient of the derivative spectra can be compared. Assuming that the derivative spectrum D corresponding to the real-time spectrum data is real The derivative spectrum D corresponding to the reference spectrum data ref The data points are x i and y i , where i = 1, 2,…, n.

[0109] Calculate the corresponding mean

[0110] Calculate the numerator:

[0111] Calculate the denominator:

[0112] The Pearson correlation coefficient is:

[0113] The comparison result is: if the Pearson correlation coefficient is close to 1 (for example, greater than a preset threshold value of 0.95), it is considered that the real-time spectrum is very similar to the reference spectrum and there is no co-elution phenomenon.

[0114] If the Pearson correlation coefficient is lower than a preset threshold (eg, less than 0.95), it is considered that there is a significant difference between the real-time spectrum and the reference spectrum, and there may be a co-elution phenomenon.

[0115] Furthermore, step S21 includes:

[0116] S211: Obtain corresponding spectra according to the spectral data and the reference spectral data respectively;

[0117] S212: According to each wavelength point in the spectrum, a data window including the wavelength point is selected;

[0118] S213: fitting the data in the data window using a polynomial function to determine a fitting function;

[0119] S214: Based on the fitting function, the derivative value at the wavelength point is calculated, and the derivative values ​​corresponding to all wavelength points are used to form a derivative spectrum.

[0120] The data window refers to a local range selected in the wavelength dimension of the spectrum, which contains a series of spectral data points around a specific wavelength point; the polynomial function refers to a mathematical expression that contains one or more terms, each of which consists of a constant multiplied by a non-negative integer power of the independent variable, and is used to describe the local shape of the curve.

[0121] Based on the above technical features, this solution generates derivative spectra using the following operating principle. First, the spectrum to be analyzed and the reference spectrum are acquired separately, forming the basis for spectral comparison. Next, for each wavelength point in the spectrum, a data window is selected that encompasses that wavelength point. This selection of a data window allows the derivative calculation to consider the spectral information surrounding that wavelength point, rather than just the isolated point, thereby introducing a local smoothing effect.

[0122] A polynomial function is fitted to the discrete spectral data points within the data window, yielding a fitting function that describes the spectral shape of that local region. Polynomial fitting effectively smooths random noise in the original data and captures local spectral trends. Based on this fitting function, the derivative value at that wavelength is calculated. By taking the derivative of the smoothed fitting function, a more accurate and smoother derivative value can be obtained than by directly differencing the original discrete data.

[0123] This process is repeated, and derivatives are calculated for all wavelength points in the spectrum, ultimately yielding a complete derivative spectrum. This derivative calculation method, based on local polynomial fitting, effectively suppresses the effects of noise and baseline drift in the original spectral data on the derivative calculation, while highlighting subtle spectral features, such as shoulders or inflection points of overlapping peaks, which may not be obvious in the original spectrum but are amplified in the derivative spectrum.

[0124] Specifically, suppose a p-order polynomial function is selected to fit the spectral data in the data window. The form of the polynomial function is: f(λ) = a0 + a1λ + a2λ 2 +...+a p λp , where a0, a1, a2…, a p are the coefficients of the polynomial.

[0125] By least squares fitting, these coefficients can be determined so that the fitting function f(λ) best fits the spectral data points within the data window. Specifically, the following residual sum of squares needs to be minimized:

[0126] After determining the fitting function f(λ), the wavelength λ can be calculated by taking the derivative of the fitting function. i The derivative value at . The derivative function is:

[0127] f'(λ)=a1+2a 21 λ+3a3λ 2 +...+pa p λ p-1 .

[0128] At wavelength λ i The derivative value at is: D(λ i )=f'(λ i )=a1+2a 21 λ+3a3λ 2 +...+pa p λ p-1 .

[0129] For each wavelength point λ i Select data window: λ i-k ,λ i-k+1 , ..., λ i , ..., λ i+k-1 ,λ i+k .

[0130] Using the polynomial function f(λ)=a0+a1λ+a2λ 2 +...+a p λ p Fit the spectral data in the data window and determine the coefficients a0, a1, a2...a p .

[0131] Calculate the derivative value: D(λ i )=f'(λ i )=a1+2a 21 λ+3a3λ 2 +...+pa p λ p-1 .

[0132] Repeat this process to calculate the derivatives for all wavelength points, and finally obtain the complete derivative spectrum:

[0133] DS=[D(λ1),D(λ2),...,D(λ n )].

[0134] The derivative spectrum generated in this way more accurately reflects the differences between the analyte and the reference spectrum, particularly subtle spectral shape changes caused by coelution. Using this derivative spectrum for subsequent spectral comparison improves the sensitivity and accuracy of detecting coeluting components, providing a more reliable data foundation for subsequent coelution judgment.

[0135] Furthermore, step S3 includes:

[0136] S31: Compare the comparison result with a preset judgment threshold;

[0137] S32: When the spectrum difference indicated by the comparison result exceeds the judgment threshold, determining that the judgment result is that the elution peak is a co-elution peak.

[0138] Based on the above technical means, the working principle of this solution is that after obtaining the spectral data of the elution peak and comparing it with the reference spectral data to obtain a quantitative comparison result, in order to convert this quantitative result into a clear co-elution judgment, this solution introduces a preset judgment threshold. This threshold is determined based on the understanding of the spectral characteristics of known pure components and a large amount of experimental data. It represents the maximum difference or minimum similarity allowed between the real-time acquired spectrum and the reference spectrum under normal circumstances. Specifically, the comparison result obtained in step S2 is compared with this preset judgment threshold. The comparison result can indicate the similarity or difference of the spectra. If the spectral difference indicated by the comparison result exceeds this judgment threshold (or the similarity is lower than the corresponding threshold), it is considered that the real-time acquired spectrum is significantly different from the reference spectrum of the pure target analyte. Under the premise of stable chromatographic conditions, this significant difference strongly indicates that in addition to the target analyte, there are other components with different spectral characteristics in the elution peak, that is, co-elution has occurred. Therefore, when the spectral difference indicated by the comparison result exceeds the judgment threshold, the system determines that the elution peak is a co-elution peak containing an unknown component. This threshold-based judgment method provides an objective and repeatable criterion for abstract spectral comparison results, avoiding the uncertainty of subjective judgment. By setting an appropriate threshold, it can effectively distinguish between minor spectral differences caused by normal fluctuations and significant spectral changes caused by coelution, thereby improving the accuracy of coelution judgment. This accurate judgment provides a reliable basis for subsequent automated process control responses, avoiding the triggering of erroneous control operations due to misjudgment of coelution, thus resolving the problem of misjudgment caused by unclear or inaccurate judgment methods in existing technologies.

[0139] Furthermore, step S4 includes:

[0140] S41: determining the number of consecutive analysis cycles in which the judgment result is a co-elution peak before the current analysis cycle, and obtaining a consecutive cycle number;

[0141] S42: According to the number of consecutive cycles, executing a preset adjustment response corresponding to the number of consecutive cycles.

[0142] The analysis cycle refers to the time interval required for the online monitoring system to complete a complete process from sampling, injection, chromatographic separation to detection and data processing. It can be determined by a fixed time interval or based on a specific event trigger.

[0143] The number of consecutive analysis cycles refers to the number of consecutive analysis cycles before the currently ongoing analysis cycle in which the system has obtained the judgment result that the elution peak is a co-elution peak in multiple preceding analysis cycles, which can be determined by a counter or historical record storage.

[0144] Preset adjustment responses refer to a series of predefined operating instructions or control strategies for automated process control systems based on the number of consecutive cycles. These responses can vary depending on the number of consecutive cycles, for example, including issuing different levels of warnings, performing different degrees of process parameter adjustments, suspending specific control logic, or recommending manual intervention.

[0145] The solution of this application does not immediately execute a fixed adjustment action after determining that an elution peak is a coelution peak. Instead, it first traces back and counts the number of consecutive analysis cycles in which the coelution determination result appeared before the current analysis cycle. The system maintains a historical record of the coelution determination status for several past analysis cycles. At the end of each new analysis cycle and the coelution determination result, the system queries this historical record and counts the number of consecutive coelution determination results, thereby obtaining a continuous cycle count. The system then searches a pre-defined adjustment response strategy table based on this continuous cycle count. This strategy table defines specific adjustment actions or control strategies corresponding to different continuous cycle counts. Based on the found strategy, the system executes the preset adjustment response that matches the current continuous cycle count. For example, when the number of continuous cycles is small, it may only trigger a low-level warning or log event. As the number of continuous cycles increases, it may trigger a higher-level warning, recommend manual intervention, or perform minor process parameter adjustments. When the number of continuous cycles reaches a certain threshold, it may trigger more significant process parameter adjustments, suspend specific automated control logic, or even initiate an emergency shutdown process. In this way, this application upgrades the response to co-elution issues from a simple "yes / no" judgment to a persistence-based graded process, allowing the automated process control response to more accurately reflect the actual duration or potential severity of the co-elution problem. This persistence-based graded adjustment strategy, combined with a technical solution that only performs co-elution judgment, can avoid overreaction to occasional or minor co-elution situations, thereby improving the robustness and cost-effectiveness of automated control and reducing unnecessary production interruptions or losses.

[0146] For example, in an online chromatography monitoring application for a pharmaceutical intermediate synthesis process, after completing an analysis cycle, the system determines that the current elution peak has coelution. The system queries the historical records and finds that this is the third consecutive analysis cycle in which the coelution determination occurs. Therefore, the number of consecutive cycles is determined to be three.

[0147] The system's preset adjustment response strategy stipulates that when co-elution occurs once in a row, a low-level warning is issued; when it occurs twice in a row, a medium-level warning is issued and the reaction temperature is fine-tuned (for example, by lowering it by 2°C); when it occurs three or more times in a row, a high-level warning is issued and the reaction temperature is significantly adjusted or automatic feeding is suspended. Since the current number of consecutive cycles is three, the system executes the preset adjustment response corresponding to "three or more times in a row", for example, a high-level warning is issued, and the reaction temperature is significantly lowered (for example, by 20°C), while automatic feeding is suspended. This graded response avoids taking extreme adjustment measures the first or second time co-elution occurs. Instead, the response level is upgraded only when the problem persists and may worsen, making the control strategy more reasonable.

[0148] Furthermore, step S42 includes:

[0149] S421: when the judgment result changes from a co-elution peak to a non-co-elution peak, starting a fault recovery period;

[0150] S422: determining a target response level for performing the adjustment response, wherein, when the result of the determination is that the current analysis cycle of the co-elution peak is within the fault recovery period, determining an increased response level as the target response level based on the number of consecutive cycles;

[0151] S423: When the current analysis period is not within the fault recovery period, determining a basic response level corresponding to the number of consecutive periods as a target response level;

[0152] S424: Execute an adjustment instruction according to the target response level to adjust the automated process control response.

[0153] The fault recovery period refers to a specific time period or state mark that the system enters after the state of detecting the co-elution peak ends, which can be implemented by using a timer.

[0154] The target response level refers to a control intensity or strategy level determined according to the current system state and the number of consecutive cycles and used to guide the execution of adjustment instructions, which can be implemented by using an index in the set response strategy set.

[0155] Among them, the raised response level refers to a response level determined based on the number of consecutive cycles under a specific state, which is higher than the corresponding level under normal conditions. It can be achieved by adding a fixed value to the basic response level or multiplying it by a coefficient.

[0156] The basic response level refers to a response level directly determined according to the number of consecutive cycles under normal operating conditions, which can be implemented by using a preset mapping table between the number of consecutive cycles and the response level.

[0157] In a specific implementation scenario, assume an automated process control system monitors the synthesis of a pharmaceutical intermediate and adjusts the reaction temperature based on online chromatographic monitoring results. The system sets a response level based on the number of consecutive coelution peaks detected: 1-2 consecutive detections correspond to Level 1 (e.g., a minor temperature adjustment), 3-5 consecutive detections correspond to Level 2 (e.g., a moderate temperature adjustment), and more than 5 consecutive detections correspond to Level 3 (e.g., a significant temperature adjustment or feed suspension).

[0158] At a certain point, the system detected coelution peaks for five consecutive analysis cycles and executed a basic level 3 adjustment command in the fifth analysis cycle. In the subsequent sixth analysis cycle, the system determined that the peaks were not coeluting. At this point, the system initiated a fault recovery period, for example, setting the recovery period to last for three analysis cycles. In the seventh analysis cycle (within the fault recovery period), the system detected coelution peaks again. At this point, the system determined that the peaks were coeluting, and the current analysis cycle was within the fault recovery period.

[0159] The system determines the number of cycles in which the co-elution peak currently appears continuously (for example, the continuous count since the start of the recovery period, here taking the continuous count since the start of the recovery period as an example, that is, 1 time). Based on this number of consecutive cycles (1 time), the system determines an increased response level as the target response level. For example, if the basic level 1 corresponds to 1 time, the increased level can be set to basic level 2. According to the determined target response level (increased level 2), the system executes the corresponding adjustment instructions, such as making a medium-scale temperature adjustment. If the 7th analysis cycle is not within the fault recovery period and a co-elution peak is detected (1 time in a row), the system will determine the basic response level (basic level 1) corresponding to the number of consecutive cycles (1 time) as the target response level, and execute the adjustment instruction of basic level 1.

[0160] In this way, during the sensitive phase when the system is recovering from an abnormal state, even if a brief co-elution peak appears again, the system will take more active or cautious control measures (upgrade the level) than in the normal state to better cope with process fluctuations and avoid poor control effects caused by simply repeating routine responses.

[0161] In some of the above-mentioned embodiments of the present application, it is proposed to execute adjustment instructions according to the target response level to adjust the automated process control response. The execution of adjustment instructions according to the target response level can be specifically by determining a main process parameter that needs to be adjusted, and generating an adjustment instruction for the main process parameter based on the target response level. For example, when an increase in the key impurity content is monitored, the system determines that the target response level is a moderate abnormality, and generates an instruction to lower the reaction temperature according to the level. In this way, corresponding control measures can be taken according to the degree of process abnormality. However, in its implementation process, adjusting a process parameter only according to a target response level may not fully consider the complex correlation between process parameters, resulting in adverse changes in other related parameters when adjusting one parameter, thereby affecting the stability and efficiency of the entire process, and even generating new process problems.

[0162] In order to solve the above situation where adjusting only a single process parameter may cause secondary problems, step S423 further includes:

[0163] S4231: Determine an initial adjustment instruction for the first process parameter based on the target response level;

[0164] S4232: Determine one or more second process parameters affected by the adjustment of the first process parameter according to a preset process parameter association relationship;

[0165] S4233: Determine, for one or more second process parameters, a compensation adjustment instruction for offsetting a disturbance caused by the initial adjustment instruction;

[0166] S4234: Coordinately execute the initial adjustment instruction and the compensation adjustment instruction to adjust the automated process control response.

[0167] Among them, the preset process parameter correlation relationship refers to the mutual influence and dependence between different process parameters in the synthesis process of pharmaceutical intermediates.

[0168] This solution provides a more sophisticated and comprehensive automated process control response adjustment strategy. Its core principle is to recognize that process parameters in pharmaceutical intermediate synthesis do not exist in isolation but rather as a complex system of interconnected and mutually influential processes. Therefore, when adjusting a key process parameter based on monitoring results, the potential ripple effects on other related parameters must be considered, and appropriate compensatory measures must be implemented to maintain the stability and optimal operation of the entire process system.

[0169] Specifically, the solution first identifies the key process parameter that needs to be adjusted first based on the degree of abnormality determined by the system (reflected by the target response level) and generates preliminary adjustment instructions for that parameter. For example, if impurities are detected to be increasing, the reaction temperature may need to be lowered. In this case, the reaction temperature is the first process parameter, and the instruction to lower the temperature is the initial adjustment instruction.

[0170] Based on this, the solution intelligently identifies which other process parameters will be affected when the first process parameter is adjusted according to the initial instructions, using pre-established or learned relationships between process parameters. This process demonstrates an understanding and application of the inherent complexity of process systems, enabling the prediction of potential secondary effects of adjustments. For example, lowering the reaction temperature may affect the reaction rate, material viscosity, and even the selectivity of side reactions. These affected parameters are referred to as secondary process parameters.

[0171] The solution then calculates and determines compensatory adjustments for one or more identified secondary process parameters to offset or mitigate the adverse perturbations caused by the initial adjustment. These compensation adjustments are intended to maintain or restore the secondary process parameters within the desired range, thereby preventing adjustments to the primary parameters from causing new problems or degrading overall process performance. For example, if lowering the reaction temperature results in a decrease in reaction rate, the compensatory adjustment might be to appropriately extend the reaction time or fine-tune the catalyst dosage to maintain overall conversion.

[0172] Ultimately, the solution coordinates the execution of the initial adjustment command and all defined compensating adjustments. This means the system doesn't simply execute an adjustment for the first process parameter. Instead, the primary and compensating adjustments are implemented simultaneously or in a coordinated manner as a single, integrated control action. This coordinated execution ensures comprehensive process adjustments, resolving the primary issue while minimizing the negative impact on other related parameters, resulting in a smoother and more efficient automated process control response.

[0173] Specifically, assuming the target response level is R, find the first process parameter P1 and adjustment instruction ΔP1 corresponding to R from the preset mapping table, and output the first process parameter P1 and the initial adjustment instruction (the same as the corresponding adjustment instruction in the preset mapping table) ΔP1.

[0174] In a specific implementation, the preset mapping table may be:

[0175] Response Level Main process parameters Adjust direction Adjustment range 1 Reaction temperature reduce 2℃ 2 Reaction temperature reduce 5℃ 3 Reaction temperature reduce 10℃ 4 Feed rate pause - 5 Reaction temperature reduce 15℃

[0176] Through this preset mapping table, the system can quickly determine the process parameters that need to be adjusted and their adjustment range according to different response levels, thereby achieving graded response and automated control of co-flow problems.

[0177] In the synthesis of pharmaceutical intermediates, there are usually complex interactions and dependencies between process parameters. Therefore, a correlation table of process parameters can be pre-set, specifically:

[0178]

[0179]

[0180] This process parameter association table describes the relationships between key process parameters, allowing for the prediction and compensation of chain reactions to second process parameters when adjusting a first process parameter. For example, appropriately extending the reaction time or increasing the catalyst dosage can extend the reaction time by 10 minutes, or increasing the inhibitor dosage by 5%. Compensation amounts for other adjustments are set in advance by technical personnel based on specific circumstances.

[0181] By applying this collaborative adjustment strategy that takes parameter associations into account to the basic scheme of executing adjustment instructions according to the target response level, this scheme can respond to process anomalies more comprehensively, avoid the negative chain reactions that may be caused by a single adjustment, improve the accuracy and robustness of automated control, and effectively solve the limitations brought about by only adjusting a single parameter based on the anomaly level.

[0182] Please refer to Figure 2 、 Figure 3 A real-time monitoring system for the synthesis of pharmaceutical intermediates is provided, for implementing any of the above methods, the system comprising:

[0183] Acquisition module 201: when an elution peak of a target analyte is detected within a preset retention time window, spectral data of the elution peak is acquired;

[0184] Comparison module 202: compares the spectral data with preset reference spectral data of the target analyte to obtain a comparison result;

[0185] Determination module 203: Determine whether the elution peak is a co-elution peak containing an unknown component based on the comparison result, and obtain a determination result;

[0186] Adjustment module 204: When the determination result is that the elution peak is a co-elution peak, adjust the automated process control response triggered based on the elution peak.

[0187] The acquisition module 201 is a unit for acquiring spectral data of an elution peak of a target analyte when an elution peak within a preset retention time window is detected, and can be implemented by a combination of a multi-wavelength detector and a data acquisition unit.

[0188] The comparison module 202 is a unit for comparing the spectral data with the reference spectral data preset for the target analyte to obtain a comparison result. It can be implemented using a software algorithm module running on a processor, which performs spectral similarity calculation or characteristic wavelength ratio comparison.

[0189] The judgment module 203 is a unit for determining whether the elution peak is a co-elution peak containing an unknown component based on the comparison result, thereby obtaining a judgment result. This module can be implemented using a software logic judgment module running on a processor, which outputs a judgment result based on the relationship between the comparison result and a preset threshold.

[0190] Adjustment module 204 is a unit for adjusting the automated process control response triggered by the elution peak when the elution peak is determined to be a co-elution peak. This module can be implemented using an interface unit and a control logic module that communicate with the automated control system. The module outputs adjustment instructions based on the judgment result.

[0191] As a whole, the system begins by acquiring spectral data from the target analyte elution peak detected by acquisition module 201. This spectral data contains information about the absorption of the elution peak at different wavelengths. This acquired spectral data is then transmitted to comparison module 202, where it is compared with a preset reference spectrum of the pure target analyte. This comparison quantifies the difference between the real-time elution peak spectrum and the spectrum of the pure component. The comparison results are then sent to determination module 203.

[0192] The judgment module 203 determines whether the elution peak is a co-elution peak containing an unknown component based on the comparison results, such as whether the degree of spectral difference exceeds a preset threshold. If the judgment module 203 determines that the elution peak is a co-elution peak, the adjustment module 204 receives this judgment result. The adjustment module 204 no longer executes the conventional automated process control instructions that calculate the concentration based on the total area of ​​the elution peak and triggers it. Instead, it executes a preset adjustment response based on the co-elution judgment result. This adjustment response may include issuing a warning, suspending automatic control, or adopting other conservative strategies.

[0193] In this way, the system directly links spectral analysis results to automated process control, ensuring that control decisions reflect the true composition of elution peaks and avoiding operational errors caused by misjudgment due to single-wavelength detection. The system provides a physical or logical basis for implementing this method, enabling online, real-time, and automated execution. This addresses the technical challenge of applying spectral analysis-based coelution determination methods to actual online monitoring systems.

[0194] In this document, relational terms such as first and second, etc. are used merely to distinguish one entity or operation from another entity or operation, but do not necessarily require or imply any actual relationship or order between these entities or operations.

[0195] The foregoing is merely an embodiment of the present application and is not intended to limit the scope of protection of the present application. Persons skilled in the art will readily appreciate that the present application may be modified and altered in various ways. Any modifications, equivalent substitutions, or improvements made within the spirit and principles of the present application shall be included within the scope of protection of the present application.

Claims

1. A method for real-time monitoring of pharmaceutical intermediate synthesis, characterized in that: The method comprises the steps of: S1: When an elution peak of a target analyte is detected within a preset retention time window, acquiring spectral data of the elution peak; S2: comparing the spectral data with preset reference spectral data of the target analyte to obtain a comparison result; S3: judging whether the elution peak is a co-elution peak containing an unknown component according to the comparison result, and obtaining a judgment result; S4: When the judgment result is that the elution peak is a co-elution peak, adjusting the automated process control response triggered based on the elution peak.

2. A method for real-time monitoring of pharmaceutical intermediate synthesis according to claim 1, characterized in that, Step S1 includes: S11: when the elution peak is detected within the preset retention time window, collecting spectral data corresponding to a plurality of time points within the duration of the elution peak to form a spectral data set; S12: determining the signal intensity corresponding to each spectral data in the spectral data set; S13: performing weighted processing on the spectral data set based on the signal intensity corresponding to each spectral data to generate a target spectral data; S14: Using the target spectrum data as the acquired spectrum data of the elution peak.

3. A method for real-time monitoring of pharmaceutical intermediate synthesis according to claim 2, characterized in that, Step S13 includes: S131: Acquire baseline spectral data in a time region before and after the duration of the elution peak; S132: determining, based on the baseline spectral data, a signal change amount caused by baseline drift during the duration of the elution peak; S133: subtracting the signal variation from each spectral data in the spectral data set to obtain the corrected spectral data set; S134: performing weighted processing on the corrected spectral data set based on the signal intensity corresponding to each spectral data to generate the target spectral data.

4. A method for real-time monitoring of pharmaceutical intermediate synthesis according to claim 1, characterized in that, Step S2 includes: S21: generating corresponding derivative spectra according to the spectral data and the reference spectral data respectively; S22: Comparing the generated derivative spectra to generate the comparison result.

5. A method for real-time monitoring of pharmaceutical intermediate synthesis according to claim 2, characterized in that, Step S21 includes: S211: Obtain corresponding spectra according to the spectrum data and the reference spectrum data respectively; S212: selecting a data window including each wavelength point in the spectrum; S213: fitting the data in the data window using a polynomial function to determine a fitting function; S214: Calculate the derivative value at the wavelength point based on the fitting function, and construct the derivative spectrum with the derivative values ​​corresponding to all wavelength points.

6. A method for real-time monitoring of pharmaceutical intermediate synthesis according to claim 1, characterized in that, Step S3 includes: S31: Compare the comparison result with a preset judgment threshold; S32: When the spectral difference indicated by the comparison result exceeds the judgment threshold, determining that the judgment result is that the elution peak is a co-elution peak.

7. A method for real-time monitoring of pharmaceutical intermediate synthesis according to claim 1, characterized in that: Step S4 includes: S41: determining the number of consecutive analysis cycles in which the judgment result is a co-elution peak before the current analysis cycle, and obtaining a number of consecutive cycles; S42: Execute a preset adjustment response corresponding to the number of consecutive cycles according to the number of consecutive cycles.

8. A method for real-time monitoring of pharmaceutical intermediate synthesis according to claim 7, characterized in that, Step S42 includes: S421: when the determination result changes from a co-elution peak to a non-co-elution peak, starting a fault recovery period; S422: Determining a target response level for executing the adjustment response, wherein, when the judgment result is that the current analysis cycle of the co-elution peak is within the fault recovery period, determining an increased response level as the target response level based on the number of consecutive cycles; S423: When the current analysis period is not within the fault recovery period, determining a basic response level corresponding to the number of consecutive periods as the target response level; S424: Execute an adjustment instruction according to the target response level to adjust the automated process control response.

9. A method for real-time monitoring of pharmaceutical intermediate synthesis according to claim 8, characterized in that: Step S423 includes: S4231: Determine an initial adjustment instruction for a first process parameter based on the target response level; S4232: Determine one or more second process parameters affected by the adjustment of the first process parameter according to a preset process parameter association relationship; S4233: determining, for one or more of the second process parameters, a compensation adjustment instruction for offsetting a disturbance caused by the initial adjustment instruction; S4234: Coordinately execute the initial adjustment instruction and the compensation adjustment instruction to adjust the automated process control response.

10. A real-time monitoring system for the synthesis of pharmaceutical intermediates, characterized in that: For implementing the method according to any one of claims 1 to 9, the system comprises: Acquisition module: when detecting an elution peak of the target analyte within a preset retention time window, acquiring spectral data of the elution peak; Comparison module: compares the spectral data with the preset reference spectral data of the target analyte to obtain a comparison result; A judgment module: judging whether the elution peak is a co-elution peak containing an unknown component according to the comparison result, and obtaining a judgment result; Adjustment module: When the judgment result is that the elution peak is a co-elution peak, adjust the automated process control response triggered by the elution peak.

Images