A method and system for detecting the conversion rate of vanillin based on chromatographic data
By adaptively adjusting the chromatographic slope threshold, the problem of vanillin peak tailing in complex fermentation broths was solved, enabling accurate calculation of conversion rate, ensuring precise determination of fermentation endpoint and optimization of feeding rate.
Patent Information
- Authority / Receiving Office
- CN · China
- Patent Type
- Patents(China)
- Current Assignee / Owner
- KUNSHAN YAXIANG SPICEL CO LTD
- Filing Date
- 2026-02-05
- Publication Date
- 2026-05-15
AI Technical Summary
Existing chromatographic integration methods are easily affected by matrix interference in complex fermentation broths, leading to vanillin peak tailing, making it impossible to accurately calculate conversion rate, and affecting the determination of fermentation endpoint and feeding rate.
By introducing instantaneous elution resistance factors and concentration loading factors, the slope threshold is dynamically adjusted, the chromatographic peak tailing boundary is adaptively identified, an adaptive slope threshold is generated, the peak areas of vanillin and ferulic acid are accurately integrated, and the conversion rate is calculated in combination with molar mass parameters.
Under complex fermentation conditions, the non-Gaussian tail portion of the vanillin peak can be accurately identified, noise interference can be blocked, the quantitative accuracy of conversion rate can be improved, and the fermentation endpoint can be accurately located.
Smart Images
Figure CN121703341B_ABST
Abstract
Description
Technical Field
[0001] This application relates to the field of instrumental analysis and biochemical testing technology, specifically to a method and system for detecting the conversion rate of ferulic acid vanillin based on chromatographic data. Background Technology
[0002] The bio-fermentation production process of natural vanillin is a key technology in the food additive industry. Its core process involves using specific engineered bacteria in a fermenter to bioconvert the substrate natural ferulic acid into natural vanillin. In order to accurately control the fermentation process, it is usually necessary to take multiple samples of the fermentation broth and use high performance liquid chromatography to determine the amount of substrate consumed and the amount of product generated, and then calculate the conversion rate. The conversion rate is the core basis for determining the fermentation endpoint, adjusting the feeding rate, and evaluating the metabolic activity of the strain.
[0003] The commonly used chromatographic quantitative algorithm in the current technology is the chromatographic integration method. The process of this method is as follows: acquire the chromatographic elution curve signal, calculate the slope of each sampling point of the curve signal, set a fixed slope threshold, when the slope of a certain sampling point drops below the slope threshold, determine that the chromatographic peak ends, stop integration, obtain the area of the chromatographic peak and convert it into concentration, and then calculate the conversion rate.
[0004] However, during the preparation of vanillin, as fermentation time progresses, a large number of metabolic byproducts (such as guaiacol and dimers) and macromolecules (such as proteins and polysaccharides) produced by cell lysis accumulate in the fermentation broth system. These complex matrix components competitively occupy the active sites of the stationary phase of the chromatographic column, causing vanillin molecules to be subjected to additional physical adsorption resistance during elution, resulting in severe non-Gaussian kinetic tailing. The vanillin peak signal cannot quickly fall back to the baseline, but instead exhibits a long tail shape with slow decay.
[0005] At this point, the fixed slope threshold of traditional chromatographic integration methods cannot detect this special waveform change caused by physical adsorption. Often, the integration is cut off prematurely because the absolute value of the slope decreases before the tailing ends, or the integration endpoint drifts because it cannot distinguish between tailing and baseline noise. Ultimately, this leads to inaccurate conversion rates. Moreover, this error will dynamically increase as the fermentation substrate deteriorates and cannot be corrected by a simple linear coefficient. This makes it difficult for operators to accurately determine the fermentation endpoint or adjust the feeding rate based on the conversion rate, which can easily lead to over-fermentation or waste of raw materials.
[0006] Therefore, there is an urgent need for a method to detect the conversion rate of ferulic acid vanillin that can sense the complexity of the fermentation broth matrix and adaptively identify the kinetic tailing boundary, in order to solve the problem of inaccurate quantification caused by the traditional chromatographic integration method under the interference of complex matrices. Summary of the Invention
[0007] To address the problem that traditional chromatographic integration methods are easily affected by complex fermentation broth matrices and peak tailing caused by high concentration overload, leading to inaccurate conversion rate calculations, this invention proposes a method and system for detecting the conversion rate of ferulic acid vanillin based on chromatographic data.
[0008] In a first aspect, the present invention provides a method for detecting the conversion rate of ferulic acid vanillin based on chromatographic data, comprising:
[0009] The chromatographic elution curves of vanillin preparation were obtained using a chromatographic system. Based on the chromatographic elution curves, the vanillin peak and ferulic acid peak, as well as the tailing interval between the vanillin peak and the ferulic acid peak, were extracted.
[0010] For each sampling point in the tailing interval, based on the signal intensity, signal change slope, and time deviation of the sampling point relative to the peak of the vanillin peak, an instantaneous elution resistance factor characterizing the ease with which the eluted material overcomes the adsorption force of the stationary phase is calculated. The instantaneous elution resistance factors of all sampling points in the tailing interval are then subjected to energy-weighted aggregation based on signal intensity to generate a matrix hysteresis cumulative index characterizing the overall contamination degree of the fermentation broth matrix on the chromatographic column.
[0011] By analyzing the concentration loading effect, a concentration loading factor characterizing the degree of column overload is determined. Based on the concentration loading factor and the matrix hysteresis accumulation index, the slope threshold of the chromatographic integration method is dynamically adjusted to generate an adaptive slope threshold that is suitable for the current preparation process.
[0012] The effective integration endpoint of the chromatographic elution curve is determined using the adaptive slope threshold. Based on the effective integration endpoint, the integrated areas of the vanillin peak and the ferulic acid peak are determined, and the conversion rate is determined in combination with the molar mass parameter.
[0013] This technical solution introduces a physical-level instantaneous elution resistance factor, moving beyond a purely geometric view of peak tailing to a kinetic process of matrix contamination hindering the eluted substance within the column. By calculating the matrix hysteresis accumulation index, it quantifies the matrix complexity of the fermentation broth and the degree of impurity occupancy in the column. Based on this index, it dynamically adjusts the integration cutoff slope threshold, effectively establishing an adaptive mechanism: when matrix contamination is heavy and tailing is essentially caused by resistance, the algorithm automatically lowers the slope threshold, waiting for signal attenuation to complete and capture the full product peak area. Conversely, when the matrix is relatively clean, it quickly truncates the peak to avoid noise interference. Simultaneously, by introducing a concentration loading factor, it incorporates physical overload tailing caused by high concentrations, ensuring accurate integration under both high contamination and high overload conditions. This overcomes the shortcomings of traditional chromatographic integration methods based on fixed slope thresholds, improving the quantitative accuracy of vanillin and ferulic acid under complex and variable fermentation conditions, thus achieving accurate and reliable conversion rates.
[0014] Preferably, the chromatographic elution curve of vanillin preparation is obtained by a chromatographic system, including: injecting the pretreated fermentation broth into a high-performance liquid chromatography system, extracting the inherent physical parameters of the high-performance liquid chromatography system, including the flow rate of the mobile phase and the dead volume of the chromatographic column; acquiring absorbance signals at a preset sampling frequency at a preset specific wavelength, and generating a discrete time series as the chromatographic elution curve.
[0015] Preferably, the method for extracting vanillin peak and ferulic acid peak, and the tailing interval between vanillin peak and ferulic acid peak based on chromatographic elution curve is as follows: obtain the first retention time window of vanillin and the second retention time window of ferulic acid in the chromatographic elution curve; use the maximum search method to identify the peak of vanillin peak in the first retention time window and the peak of ferulic acid peak in the second retention time window respectively; search for the sampling point with the smallest signal intensity after the peak of vanillin peak and before the peak of ferulic acid peak, and determine it as the trough point; define the continuous time period from the peak of vanillin peak to the trough point as the tailing interval.
[0016] Preferably, the instantaneous elution resistance factor is determined as follows: a time constant term reflecting the signal attenuation characteristics at each sampling point is constructed based on the ratio of the absolute value of the signal intensity to the absolute value of the signal change slope; a time normalization term for eliminating the influence of natural diffusion is constructed based on the time deviation of each sampling point relative to the peak of the vanillin peak, wherein the time normalization term is negatively correlated with the time deviation; the time constant term and the time normalization term are multiplied and fused to obtain the instantaneous elution resistance factor for that sampling point.
[0017] This technical solution delves into the microscopic kinetics of chromatographic separation. It distinguishes two mechanisms that lead to peak broadening: one is normal physical diffusion, which naturally broadens over time, and the other is matrix adsorption caused by fermentation. By introducing a time normalization term that is negatively correlated with time deviation, the influence of natural diffusion can be offset, so that the calculated instantaneous elution resistance factor can accurately reflect the degree of matrix contamination.
[0018] Preferably, the matrix hysteresis accumulation index is determined as follows: The instantaneous elution resistance factor at each sampling point in the tailing interval is nonlinearly amplified to reflect the matrix deterioration effect; a signal validity gating mechanism is constructed, using the ratio of the signal intensity at the sampling point to the signal intensity at the peak of the vanillin peak as a weight to weight the nonlinearly amplified instantaneous elution resistance factor; a sampling time interval is introduced, and the weighted instantaneous elution resistance factor at all sampling points in the tailing interval is discretely integrated and accumulated; the result of the discrete integration accumulation is dimensionlessly processed using a pre-acquired system reference time, and boundary condition corrections are performed; finally, a logarithmic transformation is performed to obtain the matrix hysteresis accumulation index for the tailing interval.
[0019] This technical solution uses weighted calculations to numerically amplify the difference between anomalous adsorption and random fluctuations, accurately capturing the nonlinear contamination characteristics of the chromatographic column. It introduces a signal effectiveness gating mechanism based on relative intensity and implements soft threshold filtering to suppress mathematical singularities caused by the slope approaching zero at the tailing end, ensuring that the accumulated results originate only from genuine chemical elution behavior rather than baseline artifacts in low signal-to-noise ratio regions. Logarithmic transformation compresses the dynamic range of the accumulated effect spanning several orders of magnitude, mapping the exponentially increasing physical resistance to a linear numerical control range, enabling the subsequent threshold adjustment model to respond stably to nonlinear matrix deterioration with linear sensitivity.
[0020] Preferably, the concentration loading factor, which characterizes the degree of column overload, is determined by analyzing the concentration loading effect, including: obtaining a reference saturation concentration value set in the high-performance liquid chromatography system to represent the maximum capacity of the column; determining the estimated concentration of vanillin by the signal intensity at the peak of the vanillin peak; and determining the concentration loading factor based on the ratio of the estimated concentration of vanillin to the reference saturation concentration value, wherein the concentration loading factor is positively correlated with the ratio.
[0021] This technical solution introduces the dimension of the physical capacity of the chromatographic column. In the later stage of fermentation, high concentrations of products will reach the adsorption saturation limit of the chromatographic column, resulting in physical tailing even without impurity interference. This overload risk is assessed by calculating the ratio of the current concentration to the saturation concentration, providing another key dimension for subsequent threshold adjustment. This avoids the problem of detection inaccuracy in high-yield fermenters caused by ignoring overload due to only considering impurities.
[0022] Preferably, the slope threshold of the chromatographic integration method is dynamically adjusted based on the concentration loading factor and the matrix hysteresis cumulative index to generate an adaptive slope threshold that is suitable for the current preparation process. This includes: obtaining the slope threshold set by the chromatographic integration method under the standard operating conditions of the chromatographic system and using this slope threshold as a reference slope threshold; constructing a dynamic adjustment model that satisfies the exponential decay law based on the concentration loading factor and the matrix hysteresis cumulative index; using the dynamic adjustment model to perform negative feedback adjustment on the reference slope threshold to generate an adaptive slope threshold that is suitable for the current preparation process.
[0023] This technical solution establishes a dynamic feedback adjustment mechanism driven by two factors. It considers not only the influence of the external environment but also the influence of internal factors. It uses the exponential decay law for negative feedback adjustment, which means that the more severe the pollution or the higher the concentration, the lower the slope threshold will be. This forces the integral algorithm to maintain the integration state for a longer time when facing the tail signal, until the signal change rate is extremely low before stopping. This automatically delays the integration endpoint and obtains the accurate integration area.
[0024] Preferably, the method for determining the effective integration endpoint of the chromatographic elution curve using the adaptive slope threshold is as follows: starting from the peak of vanillin, the sampling points are traversed sequentially according to their time order, and the absolute value of the instantaneous slope of each sampling point is calculated in real time. When the absolute value of the instantaneous slope of a certain sampling point is not greater than the adaptive slope threshold, the process stops, and that sampling point is determined as the effective integration endpoint.
[0025] This technical solution reduces complex kinetic analysis to specific execution criteria. By comparing real-time slope with adaptive threshold point by point, it accurately locks the physical moment when the integration action stops. This moment is adapted to the specific state of the current fermentation broth, which not only ensures the integrity of the vanillin peak integration, but also forms a clear boundary with the ferulic acid peak.
[0026] Preferably, the integral areas of the vanillin and ferulic acid peaks are determined based on the effective integration endpoint, and the conversion rate is determined in conjunction with the molar mass parameter, including:
[0027] Using chromatographic integration, with the effective integration endpoint as the dividing line between the vanillin peak and the ferulic acid peak, area integration was performed to obtain the areas of the vanillin peak and the ferulic acid peak, respectively. Based on the areas of the vanillin peak and the ferulic acid peak, the conversion rate of ferulic acid to vanillin was obtained by combining external standard quantification with molar conversion.
[0028] Secondly, the present invention provides a ferulic vanillin conversion rate detection system based on chromatographic data. The ferulic vanillin conversion rate detection system includes a memory and a processor. The memory stores a computer program, and the processor executes the computer program to implement the steps of any of the ferulic vanillin conversion rate detection methods described in the present invention.
[0029] The present invention has the following effects:
[0030] This invention introduces an instantaneous elution resistance factor and a concentration loading factor, and constructs an adaptive integral model that reflects the degree of matrix contamination and product overload by adaptively and dynamically determining the slope threshold. This achieves dynamic physical locking of the integral boundary, accurately identifies the peak area of the non-Gaussian tail portion of the vanillin peak, blocks the overlapping interference of the non-Gaussian tail portion on the ferulic acid peak, and restores the accurate conversion rate under complex fermentation conditions. This enables process engineers to accurately lock the optimal fermentation endpoint. Attached Figure Description
[0031] Figure 1 This is a schematic diagram of the method flow of the present invention;
[0032] Figure 2 This is a schematic diagram of the detection results of the vanillin tailing region during the entire fermentation process. Detailed Implementation
[0033] The technical solutions of the present invention will be clearly and completely described below with reference to the accompanying drawings in the embodiments of the present invention.
[0034] refer to Figure 1 This invention provides a method for detecting the conversion rate of ferulic acid vanillin based on chromatographic data, comprising:
[0035] S1: Obtain the chromatographic elution curve during the vanillin preparation process using a chromatographic system, and extract the tailing region based on the chromatographic elution curve.
[0036] In the bio-fermentation production process of natural vanillin, in addition to the target product vanillin and the substrate ferulic acid, there are also metabolic byproducts and macromolecules produced by cell lysis. The interference of these matrix components causes the adsorption and desorption kinetics of vanillin molecules to deviate from the ideal Gaussian distribution, resulting in tailing. Chromatography workstations often smooth or filter the signal at the data acquisition front end. Although this mathematical modification beautifies the peak shape, it smooths out the key microscopic kinetic characteristics reflecting matrix interference.
[0037] Therefore, this step aims to establish a rigorous standard for acquiring physical signals, to acquire raw physical signals without mathematical smoothing according to standardized industrial processes, and to define an analysis domain containing complete dynamic information, thereby constructing a unified physical analysis benchmark for the entire process.
[0038] First, a sample of the fermentation broth was aseptically extracted from the bio-fermenter using a 0.22... Microporous membrane filtration is used to remove bacteria and large particulate impurities. The filtrate is diluted 50 times with methanol or mobile phase and injected into a high-performance liquid chromatography system equipped with a C18 reversed-phase column.
[0039] Secondly, the inherent physical parameters of the high-performance liquid chromatography system were extracted, including: the mobile phase flow rate set by the chromatography system was 1.0 mL / min and the dead volume of the column was 2.5 mL. These two parameters will serve as the physical benchmark for subsequent elimination of hardware differences.
[0040] Next, absorbance signals were continuously acquired at a sampling frequency of 10 Hz at the characteristic absorption wavelength (280 nm) of vanillin and ferulic acid. In order to preserve the noise and small fluctuations that reflect the microscopic characteristics of matrix interference, the original discrete time series of the detector output was directly obtained as the chromatographic elution curve. The horizontal axis of the curve represents each sampling point (each time), and the total axis is the signal intensity of that sampling point.
[0041] Finally, the vanillin peak and ferulic acid peak, as well as the tailing region, were extracted, including:
[0042] Because vanillin is relatively polar and has a weaker interaction with the nonpolar stationary phase under commonly used reversed-phase chromatography conditions (such as C18 column), its retention time is shorter and it usually elutes first. Ferulic acid contains acrylic acid side chains in its molecular structure, making it relatively hydrophobic, and it retains more strongly on the C18 column, thus elutes later. Ideally, both peaks are perfectly symmetrical Gaussian peaks with flat baselines. However, in actual fermentation production, a tailing region will appear, especially to the right of the vanillin peak.
[0043] Therefore, based on historical experience data of preparing vanillin from ferulic acid using a high-performance liquid chromatography (HPLC) system under standard operating conditions, and combining the elution order and elution time of the ferulic acid and vanillin peaks, the retention time windows for the ferulic acid and vanillin peaks were determined. The retention time window is a pre-defined time interval based on the retention time and symmetrical distribution of standards under specific chromatographic conditions. It is typically used to limit the appearance of target component peaks within this time period, enabling preliminary identification and localization of the vanillin and ferulic acid peaks.
[0044] Within the first and second retention time windows, the peaks of vanillin and ferulic acid are identified using the maximum search method. After the peak of vanillin and before the peak of ferulic acid, the sampling point with the lowest signal intensity is searched and identified as the trough point. The continuous time period from the peak of vanillin to the trough point is defined as the tailing interval, which fully includes the hysteresis falling edge signal affected by matrix adsorption.
[0045] S2: For each sampling point in the tailing interval, calculate the instantaneous elution resistance factor based on the signal strength, signal change slope, and time deviation of the sampling point relative to the peak of the vanillin peak.
[0046] After obtaining the tailing region, considering that traditional chromatographic integration methods typically use a fixed slope threshold, which assumes that signal attenuation is controlled only by longitudinal diffusion, it is impossible to distinguish between signal attenuation caused by normal diffusion and signal attenuation caused by adsorption hysteresis. In the later stages of fermentation, site competition and mass transfer resistance hysteresis caused by matrix effects mean that vanillin molecules require higher activation energies to desorb from deep micropores. This physical resistance manifests as an abnormal decrease in the absolute value of the slope in the macroscopic signal. To accurately identify this nonlinear hysteresis phenomenon caused by physical adsorption kinetics, this step constructs a microscopic index to assess whether the signal attenuation at each sampling point is subject to additional physical resistance.
[0047] Specifically, it includes:
[0048] For each sampling point in the tailing interval, a time constant term reflecting the signal attenuation characteristics at that sampling point is constructed based on the ratio of the absolute value of the signal strength to the absolute value of the signal change slope at that sampling point; a time normalization term to eliminate the influence of natural diffusion is constructed based on the time deviation of each sampling point relative to the peak of the vanillin peak, and the time normalization term is negatively correlated with the time deviation; the time constant term and the time normalization term are multiplied and fused to obtain the instantaneous elution resistance factor of that sampling point.
[0049] For the tail section For each sampling point, the instantaneous elution resistance factor satisfies the following relationship:
[0050]
[0051] in, The first of the trailing intervals The instantaneous elution resistance factor at each sampling point; the larger the value, the higher the instantaneous elution resistance factor at that sampling point. The greater the resistance to signal attenuation, the better. For the first Signal strength at each sampling point For the first The first derivative of the signal strength at each sampling point It is the absolute value symbol. This reflects the slope of the signal change at that sampling point. To prevent the denominator from being zero, the parameter is usually set to a very small positive number. , This corresponds to the moment of the peak of Vanillin Peak, and .
[0052] In this relation, As the time constant term, in pure diffusion-dominated signal attenuation, the signal decreases exponentially with time. The larger the slope of the signal change at a sampling point, the faster the signal is attenuating, meaning that molecules in the fermentation broth are more likely to leave the chromatographic column quickly without obstruction. In this case, since the slope of the signal change is in the denominator, the time constant term is small and stable. Conversely, the smaller the slope of the signal change at a sampling point, the flatter the signal becomes, and the attenuation is extremely slow. This indicates that the tailing region shows a gentle downward trend, meaning that molecules in the fermentation broth are more likely to be adsorbed onto the chromatographic column, resulting in adsorption and desorption obstruction. In this case, since the slope of the signal change is in the denominator, the time constant term increases, thus keenly capturing this physical obstruction characteristic.
[0053] In this relation, This is a time normalization term designed to eliminate the interference of natural physical diffusion on the assessment of kinetic resistance. Based on the van der Munther equation in chromatographic theory, the longitudinal diffusion effect of solute molecules within the column accumulates with increasing residence time, leading to a natural broadening of the chromatographic peak. This means that even without matrix adsorption interference, the natural decay rate of the signal will physically slow down as the sampling point moves away from the peak, resulting in a natural increase in the calculated time constant. Introducing a normalization term that is negatively correlated with time deviation essentially constructs a dynamic sensitivity weighting mechanism. When a sampling point is in the near-peak region... The smaller the time normalization term, the larger the detection sensitivity. This is because in the early stages of elution, the longitudinal diffusion effect is not yet significant. If signal attenuation is delayed at this time, it is highly likely due to the adsorption of molecules by highly active, strongly adsorbed sites, rather than natural diffusion. Therefore, high weight is given to the resistance characteristics at this point to keenly capture early abnormal tailing. When a sampling point is in the far peak region, The relatively large time normalization term, coupled with its smaller value, helps suppress misjudgments. Since natural diffusion has caused signal attenuation to become extremely slow at this point, the simple time constant term would become very large. By dividing by the large time deviation, this physical increment is penalized, giving low weight to the resistance characteristic at this point and preventing normal physical diffusion from being misjudged as chemisorption resistance. Thus, the time normalization term decouples the time-varying natural physical diffusion from the chemisorption resistance caused by matrix effects, ensuring that the instantaneous elution resistance factor only characterizes the latter.
[0054] It should be noted that, It's the signal strength. If the first derivative of the signal strength (containing the dimension of time) is... It is a numerical value with the dimension of time, because The dimension is time. and Multiplying them cancels out their dimensions, resulting in an instantaneous elution resistance factor that is a dimensionless value.
[0055] Thus, by multiplying and fusing the time constant term and the time normalization term, the time constant term reflects how slowly the signal decays at the sampling point, and the time normalization term reflects how abnormal the signal decay is. After multiplication and fusion, when the signal at a sampling point is in a state of slow decay, and this decay is not a natural diffusion caused by the passage of time, the sampling point is more likely to generate additional adsorption resistance, and the instantaneous elution resistance factor is larger.
[0056] S3: Perform energy-weighted aggregation of the instantaneous elution resistance factors of all sampling points in the tailing interval to generate the matrix hysteresis cumulative index of the tailing interval.
[0057] Considering that the transient elution resistance factor is an instantaneous micro-value and is prone to fluctuation, a high resistance at a single point does not necessarily mean that the entire column is contaminated. It is necessary to examine the cumulative effect of this resistance over time in the entire tailing region. The more complex the fermentation broth matrix, the more severely the active sites of the column are occupied, which is reflected in the higher resistance factor of the entire tailing region.
[0058] To adjust the integration strategy for the entire peak, this step aims to establish a macroscopic, energy-weighted model based on signal strength. By nonlinearly integrating the resistance factor within the tail region and assigning it a physical meaning consistent with the signal energy weights, the microscopic fluctuations are aggregated into a macroscopic comprehensive index characterizing the system state.
[0059] Specifically, a nonlinear amplification process is performed on the instantaneous elution resistance factor at each sampling point to reflect the matrix deterioration effect. A signal validity gating mechanism is constructed, using the ratio of the signal intensity at the sampling point to the signal intensity at the peak of the vanillin peak as a weight to weight the nonlinearly amplified instantaneous elution resistance factor. A sampling time interval is introduced, and the weighted instantaneous elution resistance factor of all sampling points in the tailing interval is discretely integrated and accumulated. The result of the discrete integration accumulation is dimensionless using the pre-acquired system reference time, and boundary conditions are corrected. Finally, a logarithmic transformation is performed to obtain the matrix hysteresis accumulation index in the tailing interval.
[0060] For the trailing region, the matrix hysteresis cumulative index satisfies the following relationship:
[0061]
[0062] in, This is the matrix lag cumulative index for the tailing interval. It is the natural logarithm function. This serves as the system reference time constant (taken as the dead time of the chromatographic column) to eliminate the time dimension of the discrete integral accumulation term, ensuring that the input to the logarithmic function is a dimensionless value. This is the trailing section. The first of the trailing intervals Instantaneous elution resistance factor at each sampling point For the first Signal strength at each sampling point The signal strength at the peak of the vanillin peak. This represents the sampling time interval.
[0063] This relationship is constructed through a three-dimensional model, enabling accurate assessment of matrix interference effects. A nonlinear amplification mechanism was constructed to enhance anomalous adsorption characteristics. Column contamination does not accumulate linearly; a small number of strong adsorption sites can cause an exponential increase in local fluid resistance. By calculating the square of the resistance factor and utilizing the convexity of the power function, a nonlinear amplifier was constructed. When it increases, It exhibits quadratic growth, assigning a much higher computational weight to the high-resistance region (severe tailing segment) than to the low-resistance region. This treatment numerically widens the difference between chemisorption anomalies and random noise fluctuations, making the cumulative index highly sensitive to the actual matrix deterioration effect, which is consistent with the physical fact that the matrix effect deteriorates nonlinearly.
[0064] In this relation, A signal validity gating mechanism was constructed, which acts as an energy weighting factor. At the tail end, the signal tends to flatten, and the slope of change approaches zero, which may lead to… When mathematical singularities or large oscillations occur, weighting is performed using relative signal strength as the weight. Near the weak signal baseline, this ratio approaches 0, forcibly compressing the drag contribution in the low signal-to-noise ratio region to zero. In the real effective tail region, the drag factor is large and the signal strength is high, so the weight is large and it is identified as an effective chemical elution component and included in the total amount. This ensures that the cumulative result comes only from the real chemical elution behavior, thereby achieving noise reduction at the physical level.
[0065] In this relation, This enables the integration and aggregation of discrete point sets into continuous physical quantities. Simple numerical accumulation is limited by the difference in sampling frequency; the higher the sampling rate and the more points, the larger the accumulated value. However, by introducing a sampling time interval... Performing product summation is essentially calculating the integral area of the drag energy in the time domain. This operation gives the algorithm sampling rate independence, ensuring that no matter how the chromatograph's sampling frequency is set, the calculated matrix hysteresis accumulation always represents the total drag load at the physical level, rather than the accumulation of points at the data level. The introduction of this method is to ensure dimensional consistency in mathematical operations. The physical result of the integral term has a time dimension, while the input of the logarithmic function must be a dimensionless pure numerical value. The system reference time constant serves as the physical benchmark, eliminating the time unit of the integral result and converting it into a dimensionless ratio relative to the system's inherent scale. This not only ensures the rationality of the mathematical operation but also transforms the absolute time accumulation into a relative system load rate, giving the index a universal physical meaning across different chromatographic columns. A linearized dynamic range compression mechanism was jointly constructed. Since the deterioration of chromatographic behavior by fermentation impurities (such as column efficiency decay and increased tailing) typically follows an exponential growth pattern, the total resistance value after direct integration can span several orders of magnitude. By using the natural logarithm function to compress the dynamic range, the exponentially increasing physical resistance is mapped to a linear numerical control interval. This allows the subsequent slope threshold adjustment model to respond stably to nonlinear matrix deterioration with constant sensitivity, avoiding control system instability caused by excessive input value fluctuations and ensuring the robustness of the control system. The operation is a boundary condition correction term, ensuring that when there is no tail or an ideal Gaussian peak, the integral term approaches 0, and the logarithmic function is 1 internally, so that the final calculated matrix hysteresis cumulative exponent is... This aligns with the physical principle that there is no hysteresis effect without pollution, ensuring the accuracy of the zero point of the detection indicators.
[0066] Overall, the terms within this relation construct a closed loop of physical perception for complex substrates through layered linkages of micro-level discrimination, macro-level aggregation, and control adaptation. This constitutes a mass-energy dual confirmation mechanism. It is responsible for nonlinearly amplifying the adsorption resistance characteristics at the physical level. The signal energy is used as a benchmark for gating. The two work together to ensure that the algorithm maintains extremely high sensitivity to weak chemical tails and can filter out baseline noise without energy, thus achieving discrimination at the micro level. This involves aggregating transient microscopic resistance fluctuations into a total macroscopic system load that is independent of the sampling rate and decoupled from hardware specifications; ultimately... This completes the mapping from nonlinear physical space to linear control space, compressing the exponentially deteriorating substrate contamination state during fermentation into a linearly changing numerical index.
[0067] In this way, the chaotic and nonlinear chromatographic adsorption kinetics process can be characterized as a stable linear control signal, thereby driving the subsequent slope threshold to be adaptively adjusted by negative feedback, which solves the quantitative problem under complex matrix interference in a physical sense.
[0068] S4: The concentration loading effect is analyzed to determine the concentration loading factor. Based on the concentration loading factor and the matrix hysteresis cumulative index, the slope threshold of the chromatographic integral method is dynamically adjusted to generate an adaptive slope threshold.
[0069] After obtaining the matrix hysteresis cumulative index, which characterizes the overall contamination level of the system, this step aims to reshape the decision logic of the standard algorithm by using the matrix hysteresis cumulative index as the main feedback variable and the concentration loading effect as a co-correction term to dynamically calculate the optimal slope threshold.
[0070] First, the concentration loading effect is analyzed to determine the concentration loading factor characterizing the degree of column overload. This includes: obtaining a reference saturation concentration value set in the high-performance liquid chromatography system to represent the maximum capacity of the column; determining the estimated concentration of vanillin by the signal intensity at the peak of the vanillin peak; and determining the concentration loading factor based on the ratio of the estimated concentration of vanillin to the reference saturation concentration value. The concentration loading factor is positively correlated with this ratio.
[0071] Specifically, the adaptive slope threshold is determined based on the following relationship:
[0072]
[0073] in, As an adaptive slope threshold, this parameter will directly replace the fixed threshold in the standard algorithm. The preset slope threshold under standard operating conditions. This is the sensitivity adjustment coefficient, a dimensionless constant, set to 1.0, used to adjust the algorithm's response gain to changes in matrix effects. The vanillin concentration is estimated based on the signal intensity at the peak of the vanillin peak. Specifically, it is estimated using a pre-constructed linear regression equation between vanillin concentration and signal intensity, based on the Lambert-Beer law. This is the reference saturation concentration of the chromatographic column, i.e., the saturation point of the Langmuir isotherm. This is the matrix lag cumulative index for the tailing interval.
[0074] In this relation, It is a concentration loading factor, expressed as a square root rather than a direct linear multiplication. This is based on the physical properties of the Langmuir adsorption isotherm. When the column is near overload, the increase in tailing degree has a non-linear (sublinear) relationship with the increase in concentration. The square root function effectively simulates this soft saturation characteristic, preventing excessive correction at ultra-high concentrations from leading to an excessively low slope threshold and introducing baseline noise. The concentration loading factor ensures compensation for physical overload tailing caused by high-concentration solute occupying active sites on the column when sample concentrations are too high. Even if the matrix effect exponent remains unchanged, this correction term increases the overall value of the exponential part to lower the slope threshold and prevent area loss due to high concentration overload. The increment operation ensures the physical consistency of the boundary conditions, especially when sample concentrations are extremely low. much smaller At this point, the concentration loading factor approaches 1. Controlled only by the matrix hysteresis cumulative index, once overload occurs, the concentration loading factor is greater than 1. The concentration loading factor and the matrix hysteresis cumulative index control work together to suppress the threshold, preventing the tailing area caused by physical overload from being missed due to simply considering the matrix effect.
[0075] In this relation, The function internally constructs an exponential decay model, implementing a negative feedback regulation mechanism. The more complex the fermentation broth substrate and the higher the vanillin concentration, the longer the tail will be, requiring a smoother trend to capture the accurate integral area. In this case, the exponential term increases. From A downward decay and a lower slope threshold mean that the integrator will wait until the slope becomes gentler before stopping, thus extending the integration window and capturing the effective area of the vanillin peak obscured by the tailing region. Conversely, when the fermentation broth matrix is simpler and no tailing occurs, the preset slope threshold under standard operating conditions will be applied. Define the integration window.
[0076] S5: The effective integration endpoint of the chromatographic elution curve is determined by using an adaptive slope threshold, and the vanillin peak and ferulic acid peak are integrated, and the conversion rate is calculated in combination with the molar mass parameter.
[0077] After calculating the optimal adaptive slope threshold at the current moment, the application of the adaptive threshold not only identifies the tailing area of vanillin but also cuts off its interference with the subsequent ferulic acid peak, thereby achieving bidirectional quantitative correction. The molar conversion yield is calculated by using the area of the corrected vanillin peak and the area of the ferulic acid peak.
[0078] Molar conversion yield characterizes the selectivity of biological metabolism. It reflects what proportion of the substrate consumed by the microorganism actually flows to the target product, rather than being used for the growth and reproduction of the microorganism itself or the generation of metabolic byproducts. Compared with the traditional conversion rate (which only looks at how much raw material is reduced), molar conversion yield can more sensitively reflect the health status of the fermentation broth. If the molar conversion yield decreases, it indicates that the microorganism may be aging or the fermentation environment may be deteriorating, leading to an increase in side reactions. It can help process engineers more accurately determine whether to add feed, adjust the aeration rate, or terminate the fermentation, avoiding the ineffective consumption of raw materials.
[0079] Specifically, it includes:
[0080] Starting from the peak of vanillin, the absolute value of the instantaneous slope of the chromatographic curve is calculated point by point. When the absolute value of the instantaneous slope at a certain point is detected to be less than or equal to the adaptive slope threshold for the first time, the search is stopped and that moment is locked as the effective integration endpoint.
[0081] Using chromatographic integration, the effective integration endpoint is used as the physical marker:
[0082] For the vanillin peak: the integration interval is extended to the effective integration endpoint, thus recovering the tail area truncated by the traditional fixed threshold method, and obtaining the corrected vanillin peak area. This operation corrected the underestimation of product concentration caused by matrix adsorption.
[0083] For the ferulic acid peak: Starting from the effective integration endpoint, remove the overlapping interference from the vanillin tail extending into the ferulic acid region to obtain the pure ferulic acid peak area. This operation corrects for overestimation or underestimation of substrate concentration caused by peak overlap.
[0084] Calculate the conversion rate: Using a pre-constructed external standard curve (characterizing the linear relationship between peak area and concentration), the conversion rate is calculated. and Convert to mass concentration in fermentation broth respectively and Using the molar mass of vanillin molar mass of ferulic acid This converts mass concentration into molar concentration.
[0085] Calculate conversion rate:
[0086]
[0087] in, It is the molar conversion yield of ferulic acid to vanillin. The initial substrate concentration and the initial ferulic acid concentration in the fermentation broth are given. The molar concentration of the actual product (vanillin) is represented. It characterizes the amount of substrate actually consumed by microorganisms in metabolism. This represents the molar concentration of the substrate (ferulic acid) actually consumed.
[0088] This formula is the standard formula for calculating molar conversion yield. Molar conversion yield eliminates the interference of fermentation byproducts and cell adsorption on mass balance, and can accurately reflect the core metabolic efficiency of engineered bacteria in converting substrates into target products. When an abnormal decline occurs, it indicates to the process personnel that the fermentation environment has deteriorated or the strain has aged, thus guiding them to accurately pinpoint the optimal fermentation endpoint or replenish the material in a timely manner to avoid waste of raw materials.
[0089] like Figure 2 As shown, during the complete vanillin biotransformation cycle (0-72 hours), the matrix complexity of the fermentation broth continuously increases due to the accumulation of metabolic byproducts as fermentation time progresses, leading to severe nonlinear kinetic tailing of the chromatographic peaks. Existing chromatographic integration methods, due to their fixed slope threshold, cannot detect the dynamic changes in elution resistance, resulting in erroneous segmentation before the tailing interval has fully ended. This leads to the systematic underestimation of the product integration area, resulting in inaccurate conversion calculations, and the error increases nonlinearly with the fermentation process. In contrast, this invention introduces an instantaneous elution resistance factor and a matrix hysteresis accumulation index to achieve adaptive physical compensation for complex and variable operating conditions. It dynamically lowers the integration threshold to accurately capture the complete tailing interval, thereby improving the accuracy of the conversion rate.
[0090] The above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of the present invention should be included within the protection scope of the present invention.
Claims
1. A method for detecting the conversion rate of ferulic acid vanillin based on chromatographic data, characterized in that, include: The chromatographic elution curves of vanillin preparation were obtained using a chromatographic system. Based on the chromatographic elution curves, the vanillin peak and ferulic acid peak, as well as the tailing interval between the vanillin peak and the ferulic acid peak, were extracted. For each sampling point in the tailing interval, based on the signal intensity, signal change slope, and time deviation of the sampling point relative to the peak of the vanillin peak, an instantaneous elution resistance factor characterizing the ease with which the eluted material overcomes the adsorption force of the stationary phase is calculated. The instantaneous elution resistance factors of all sampling points in the tailing interval are then subjected to energy-weighted aggregation based on signal intensity to generate a matrix hysteresis cumulative index characterizing the overall contamination degree of the fermentation broth matrix on the chromatographic column. By analyzing the concentration loading effect, a concentration loading factor characterizing the degree of column overload is determined. Based on the concentration loading factor and the matrix hysteresis accumulation index, the slope threshold of the chromatographic integration method is dynamically adjusted to generate an adaptive slope threshold that is suitable for the current preparation process. The effective integration endpoint of the chromatographic elution curve is determined using the adaptive slope threshold. Based on the effective integration endpoint, the integrated areas of the vanillin peak and the ferulic acid peak are determined, and the conversion rate is determined in combination with the molar mass parameter.
2. The method for detecting the conversion rate of ferulic acid vanillin according to claim 1, characterized in that, The chromatographic elution curves of vanillin preparation were obtained using a chromatographic system, including: The pretreated fermentation broth was injected into a high-performance liquid chromatography (HPLC) system to extract the inherent physical parameters of the HPLC system, including the flow rate of the mobile phase and the dead volume of the chromatographic column. The absorbance signal was acquired at a preset sampling frequency at a preset wavelength to generate a discrete time series as the chromatographic elution curve.
3. The method for detecting the conversion rate of ferulic acid vanillin according to claim 1, characterized in that, The method for extracting vanillin and ferulic acid peaks, as well as the tailing interval between the vanillin and ferulic acid peaks, based on chromatographic elution curves is as follows: The first retention time window of vanillin and the second retention time window of ferulic acid in the chromatographic elution curve are obtained. The peak of vanillin peak is identified in the first retention time window and the peak of ferulic acid peak is identified in the second retention time window, respectively, using the maximum search method. The sampling point with the minimum signal intensity after the peak of vanillin peak and before the peak of ferulic acid peak is searched and determined as the trough point. The continuous time period from the peak of vanillin peak to the trough point is defined as the tailing interval.
4. The method for detecting the conversion rate of ferulic acid vanillin according to claim 1, characterized in that, The instantaneous elution resistance factor is determined based on the following method: A time constant term reflecting the signal attenuation characteristics at each sampling point is constructed based on the ratio of the absolute value of the signal intensity to the absolute value of the signal change slope. A time normalization term to eliminate the influence of natural diffusion is constructed based on the time deviation of each sampling point relative to the peak of the vanillin peak. The time normalization term is negatively correlated with the time deviation. The instantaneous elution resistance factor at the sampling point is obtained by multiplying and fusing the time constant term and the time normalization term.
5. The method for detecting the conversion rate of ferulic acid vanillin according to claim 1, characterized in that, The matrix hysteresis accumulation index is determined as follows: The instantaneous elution resistance factor at each sampling point in the tailing interval is nonlinearly amplified to reflect the matrix deterioration effect; a signal validity gating mechanism is constructed, using the ratio of the signal intensity at the sampling point to the signal intensity at the peak of the vanillin peak as a weight to weight the nonlinearly amplified instantaneous elution resistance factor; a sampling time interval is introduced, and the weighted instantaneous elution resistance factor at all sampling points in the tailing interval is discretely integrated and accumulated; the result of the discrete integration accumulation is dimensionless using a pre-acquired system reference time, and boundary conditions are corrected; finally, a logarithmic transformation is performed to obtain the matrix hysteresis accumulation index for the tailing interval.
6. The method for detecting the conversion rate of ferulic acid vanillin according to claim 1, characterized in that, The concentration loading effect was analyzed to determine the concentration loading factor characterizing the degree of column overload, including: The reference saturation concentration value set in the high performance liquid chromatography system to represent the maximum capacity of the chromatographic column is obtained. The estimated concentration of vanillin is determined by the signal intensity at the peak of the vanillin peak. The concentration loading factor is determined based on the ratio of the estimated concentration of vanillin to the reference saturation concentration value. The concentration loading factor is positively correlated with this ratio.
7. The method for detecting the conversion rate of ferulic acid vanillin according to claim 1, characterized in that, The slope threshold of the chromatographic integration method is dynamically adjusted based on the concentration loading factor and the matrix hysteresis accumulation index to generate an adaptive slope threshold that fits the current preparation process, including: Obtain the slope threshold set under standard operating conditions of the chromatographic system using the chromatographic integration method, and use this slope threshold as the reference slope threshold; Based on the concentration loading factor and the matrix hysteresis accumulation index, a dynamic adjustment model that satisfies the exponential decay law is constructed. The dynamic adjustment model is used to perform negative feedback adjustment on the benchmark slope threshold to generate an adaptive slope threshold that is adapted to the current preparation process.
8. The method for detecting the conversion rate of ferulic acid vanillin according to claim 1, characterized in that, The method for determining the effective integration endpoint of the chromatographic elution curve using the adaptive slope threshold is as follows: Starting from the peak of vanillin, the sampling points are traversed sequentially according to their time order. The absolute value of the instantaneous slope of each sampling point is calculated in real time. The process stops when the absolute value of the instantaneous slope of a certain sampling point is not greater than the adaptive slope threshold, and that sampling point is determined as the valid integration endpoint.
9. The method for detecting the conversion rate of ferulic acid vanillin according to claim 1, characterized in that, The integral areas of the vanillin and ferulic acid peaks are determined based on the effective integration endpoint, and the conversion rate is determined in conjunction with the molar mass parameter, including: Using chromatographic integration, with the effective integration endpoint as the dividing line between the vanillin peak and the ferulic acid peak, area integration was performed to obtain the areas of the vanillin peak and the ferulic acid peak, respectively. Based on the areas of the vanillin peak and the ferulic acid peak, the conversion rate of ferulic acid to vanillin was obtained by combining external standard quantification with molar conversion.
10. A system for detecting the conversion rate of ferulic acid vanillin based on chromatographic data, characterized in that, The system includes a memory and a processor, the memory storing a computer program, and the processor executing the computer program to implement the steps of the method for detecting the conversion rate of ferulic acid vanillin based on chromatographic data as described in any one of claims 1-9.