A quality control method and system for erythromycin primary body fermentation production

Abstract

This application provides a quality control method and system for the primary fermentation production of erythromycin, relating to the field of bio-fermentation control technology. By performing time-delay correlation analysis on sensor variables in the primary fermentation equipment, the response lag time of the sensor variables to the specific yield of erythromycin is obtained. A modulation coefficient is generated based on the activity level of the bacterial strain, and the amplitude of the sensor variables is adjusted using the modulation coefficient to obtain a state descriptor for the primary fermentation equipment. Then, based on the state descriptor and metabolic parameters, a metabolic offset is determined. Based on the metabolic offset and the response lag time, the action category and execution time of quality control are determined. When the action category is metabolic remodeling, closed-loop control is performed on the temperature, dissolved oxygen concentration, and sugar supplementation rate in the primary fermentation equipment based on the metabolic offset at the execution time. This application can improve the adaptive adjustment capability of the primary fermentation process of erythromycin to abnormal metabolic changes.

Description

Technical Field

[0001] This application relates to the field of bio-fermentation control technology, and more specifically, to a quality control method and system for the production of erythromycin primary fermentation. Background Technology

[0002] Biological fermentation control refers to the regulation of environmental parameters such as temperature, dissolved oxygen concentration, feeding rate, and pH to provide suitable reaction conditions for microbial growth and product synthesis. It is a core technical means to ensure the stability of the fermentation process and the yield of the target product. Erythromycin primary fermentation is a critical stage in erythromycin production. This stage is characterized by mycelial proliferation and biomass accumulation of the producing strain, and its metabolic state directly determines the efficiency of erythromycin macrolide ring synthesis in subsequent secondary metabolic stages.

[0003] Currently, quality control of erythromycin primary fermentation typically employs a combination of offline sampling and manual adjustment. This involves periodically collecting fermentation broth samples for potency testing and manually adjusting parameters such as temperature, dissolved oxygen concentration, and glucose supplementation rate based on the results. However, existing methods rely on offline sampling and analysis, making it difficult to dynamically track changes in metabolic state during the control process and adjust control parameters in real time. Furthermore, the varying response lags of different sensor variables to erythromycin specific yield during fermentation lead to a misalignment between the timing of control actions and the peaks and troughs of the strain's metabolic response. Additionally, differences in strain activity between different batches result in a mismatch between the intensity of control and actual metabolic demands. Therefore, achieving a closed-loop regulation of the fermentation state that integrates strain activity correction, response lag compensation, and metabolic offset determination to enhance the adaptive adjustment capability of the erythromycin primary fermentation process to abnormal metabolic changes remains a significant challenge for the industry. Summary of the Invention

[0004] This application provides a quality control method and system for the fermentation production of erythromycin primary, which can realize closed-loop regulation of fermentation state by modifying the activity of fusion strains, compensating for response lag, and determining metabolic offset, so as to improve the adaptive adjustment ability of the erythromycin primary fermentation process to abnormal metabolic changes.

[0005] In a first aspect, this application provides a quality control method for the fermentation production of erythromycin primary stock, the quality control method comprising the following steps:

[0006] The sensor variables in the erythromycin primary fermentation equipment were monitored, and time-delay correlation analysis was performed on the sensor variables to obtain the response lag time of the sensor variables to the specific yield of erythromycin.

[0007] A modulation coefficient is generated based on the activity level of the strain, and the amplitude of the sensing variable is adjusted by the modulation coefficient to obtain the state descriptor in the erythromycin primary fermentation device.

[0008] Obtain the metabolic parameters of the strain, determine the metabolic offset based on the state descriptor and the metabolic parameters, and determine the action category and execution time of quality regulation based on the metabolic offset and the response lag time;

[0009] When the action category is metabolic remodeling, the temperature, dissolved oxygen concentration, and sugar replenishment rate in the erythromycin primary fermentation equipment are controlled in a closed loop based on the metabolic offset at the execution time.

[0010] In this embodiment, the sensing variables in the erythromycin primary fermentation equipment are monitored by Raman spectrometer and tail gas spectrometer; the sensing variables include feed rate, dissolved oxygen concentration, carbon dioxide concentration and Raman spectral data.

[0011] In this embodiment, time-lag correlation analysis is performed on the sensed variable to obtain the response lag time of the sensed variable to the specific yield of erythromycin, specifically including:

[0012] A time series analysis of the sensed variables for a fixed duration was performed to obtain multiple translation step sizes;

[0013] Determine the correlation coefficient between the sensor variable corresponding to each translation step size and the specific yield of erythromycin;

[0014] Then, the response lag time of the sensing variable to the specific yield of erythromycin was obtained by screening all correlation coefficients.

[0015] In this embodiment, generating the modulation coefficient based on the activity level of the strain specifically includes:

[0016] The activity level is determined based on the ratio of the strain's current specific growth rate to the average specific growth rate of historical batches.

[0017] The upper and lower limits of the modulation coefficient range were determined based on historical batch data from the primary fermentation of erythromycin.

[0018] The modulation coefficient is obtained by linear interpolation between the upper and lower limits of the modulation coefficient range based on the activity level.

[0019] In this embodiment, the amplitude adjustment of the sensing variable using the modulation coefficient to obtain the state descriptor in the erythromycin primary fermentation device specifically includes:

[0020] The weighted amplitude of the sensed variable is determined by the modulation coefficient;

[0021] The state of the weighted amplitude is monitored to obtain the state change curve;

[0022] Linear fitting was performed on the state change curve to obtain the state descriptor of the erythromycin primary fermentation device.

[0023] In this embodiment, the metabolic parameters of the strain are obtained based on the scattering signal from a Raman spectrometer.

[0024] In this embodiment, determining the metabolic offset based on the state descriptor and the metabolic parameters specifically includes:

[0025] The carbon source consumption rate and dissolved oxygen consumption rate are extracted from the state descriptor, and the cumulative concentration of precursor substances is extracted from the metabolic parameters.

[0026] The cumulative concentration, the carbon source consumption rate, and the dissolved oxygen consumption rate are set as baseline values ​​respectively, thereby determining the precursor deviation, carbon source deviation, and dissolved oxygen deviation.

[0027] The metabolic offset is obtained by weighted summation of the precursor bias, the carbon source bias, and the dissolved oxygen bias.

[0028] In this embodiment, determining the action category and execution time of quality regulation based on the metabolic offset and the response lag time specifically includes:

[0029] Obtain historical batch data and control cycles for the primary fermentation of erythromycin;

[0030] The first and second thresholds for quality control actions are determined using the historical batch data.

[0031] When the metabolic offset is below the first threshold, the action type is to maintain the current operating condition; when the metabolic offset is within the closed interval between the first threshold and the second threshold, the action type is to replenish feed for correction; when the metabolic offset exceeds the second threshold, the action type is to refactor metabolism.

[0032] When the response lag time is less than the control period, the execution time is the current time; when the response lag time is greater than or equal to the control period, the execution time is delayed to the time point corresponding to the sum of the current time and the response lag time.

[0033] In this embodiment, when the action category is metabolic remodeling, the closed-loop control of temperature, dissolved oxygen concentration, and sugar supplementation rate in the erythromycin primary fermentation equipment based on the metabolic offset at the execution time specifically includes:

[0034] At the execution time, the temperature reduction range, dissolved oxygen increase range, and glucose replenishment rate are set according to the metabolic offset.

[0035] The temperature setpoint is lowered according to the temperature reduction range, the dissolved oxygen concentration setpoint is increased according to the dissolved oxygen increase range, the sugar replenishment rate is proportionally adjusted, and real-time sensor variables are collected during the adjustment process.

[0036] The metabolic offset is updated based on the real-time sensor variables, and the temperature reduction, dissolved oxygen up-reduction, and glucose replenishment rate are re-determined based on the updated metabolic offset until the metabolic offset is lower than a first threshold.

[0037] Secondly, this application provides a quality control system for the primary fermentation production of erythromycin, used to execute a quality control method for the primary fermentation production of erythromycin, the quality control system comprising:

[0038] The monitoring and analysis module is used to monitor the sensor variables in the erythromycin primary fermentation equipment, perform time-delay correlation analysis on the sensor variables, and obtain the response lag time of the sensor variables to the specific yield of erythromycin.

[0039] The state generation module is used to generate modulation coefficients based on the activity level of the strain, and adjust the amplitude of the sensing variable through the modulation coefficients to obtain the state descriptor in the erythromycin primary fermentation device.

[0040] The metabolic decision module is used to acquire the metabolic parameters of the strain, determine the metabolic offset based on the state descriptor and the metabolic parameters, and determine the action category and execution time of quality regulation based on the metabolic offset and the response lag time.

[0041] The control execution module is used to perform closed-loop control of the temperature, dissolved oxygen concentration and sugar supplementation rate in the erythromycin primary fermentation equipment based on the metabolic offset at the execution time when the action category is metabolic remodeling.

[0042] The technical solutions provided by the embodiments disclosed in this application have the following beneficial effects:

[0043] The technical solution provided in this application enables closed-loop regulation of fermentation state by integrating strain activity correction, response lag compensation, and metabolic offset determination. First, by monitoring sensor variables and performing time-lag correlation analysis, the response lag time of the sensor variables to erythromycin specific yield is determined, providing a quantitative basis in the time dimension for determining the timing of regulation. Then, modulation coefficients are generated based on strain activity levels, and the amplitude of the sensor variables is adjusted to form a state descriptor, achieving adaptive calibration of sensor signals under different activity batches, thereby providing characteristic inputs reflecting the actual metabolic state of the strain for determining metabolic offset. Finally, the state descriptor is fused with strain metabolic parameters to determine the metabolic offset, and the results are combined with... By combining response lag time to jointly determine the type and timing of regulatory actions, the triggering timing and intervention intensity of regulatory actions are subject to synergistic constraints of time and amplitude dimensions, respectively. This can solve the problems of misalignment between regulatory timing and metabolic response peaks and valleys, and mismatch between regulatory intensity and strain activity state. Finally, when the action type is metabolic remodeling, closed-loop regulation of temperature, dissolved oxygen concentration, and sugar supplementation rate is performed based on metabolic offset at the execution time. During the regulation process, the metabolic offset is continuously updated based on real-time sensor variables to correct the regulatory parameters. This enables synergistic decision-making between the timing of regulatory execution and the amplitude of metabolic offset, which is beneficial to improving the matching accuracy between quality regulation actions and strain metabolic state during erythromycin primary fermentation.

[0044] In summary, the technical solution adopted in this application can realize closed-loop regulation of fermentation state by modifying the activity of fusion strains, compensating for response lag, and determining metabolic offset, so as to improve the adaptive adjustment ability of the erythromycin primary fermentation process to abnormal metabolic changes. Attached Figure Description

[0045] To more clearly illustrate the technical solutions in the embodiments of this application or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are only embodiments of this application. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0046] Figure 1 This is a flowchart of a quality control method for the fermentation production of erythromycin primary body according to the present application;

[0047] Figure 2 This is a structural schematic diagram of the erythromycin primary fermentation equipment provided in this application;

[0048] Figure 3 This is an exemplary flowchart for determining metabolic offsets provided in this application;

[0049] Figure 4This is a modular structure diagram of a quality control system for the primary fermentation production of erythromycin, provided in this application. Detailed Implementation

[0050] The technical solutions of the embodiments of this application will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of this application, and not all embodiments. Based on the embodiments of this application, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of this application.

[0051] Example 1: To better understand the technical solution of this application, the above technical solution will be described in detail below with reference to the accompanying drawings and specific embodiments. Figure 1 As shown in the figure, this is a flowchart of a quality control method for the fermentation production of erythromycin primary body according to this embodiment of the present application. The quality control method includes the following steps:

[0052] In step S1, the sensor variables in the erythromycin primary fermentation equipment are monitored, and time-delay correlation analysis is performed on the sensor variables to obtain the response lag time of the sensor variables to the specific yield of erythromycin.

[0053] In this specific implementation, Raman spectroscopy and tail gas spectroscopy are used to monitor the sensor variables in the erythromycin primary fermentation equipment; preferably, reference can be made to Figure 2 As shown, this figure is a schematic diagram of the structure of the erythromycin primary fermentation device provided in this application; wherein, the Raman spectrometer is an online Raman spectrometer equipped with an immersion fiber optic probe, which is directly inserted into the fermentation broth inside the erythromycin primary fermentation device to collect the Raman spectral scattering signals corresponding to each metabolic component in the fermentation broth in real time; the tail gas spectrometer is an online mass spectrometer configured at the exhaust port of the erythromycin primary fermentation device to monitor the volume concentration of carbon dioxide and oxygen in the exhaust gas of the erythromycin primary fermentation device in real time; the erythromycin primary fermentation device is a bioreactor that performs the erythromycin primary fermentation process, such as... Figure 2 As shown, the erythromycin primary fermentation equipment is equipped with a temperature sensor, dissolved oxygen electrode, feed pump, stirring motor, optical density sensor, Raman spectrometer and tail gas spectrometer. Optionally, the erythromycin primary fermentation equipment can be a glass fermenter to provide a suitable growth and metabolic environment for the production strain.

[0054] It should be noted that the sensing variables include feed rate, dissolved oxygen concentration, carbon dioxide concentration, and Raman spectral data. Feed rate refers to the volume of carbon source solution added to the erythromycin primary fermentation unit per unit time via the feed pump, used to characterize the intensity of carbon source supply. Dissolved oxygen concentration refers to the percentage of dissolved oxygen saturation in the fermentation broth of the erythromycin primary fermentation unit, used to characterize the degree of oxygen solubility in the fermentation broth and the oxygen consumption state of the strain. Carbon dioxide concentration refers to the percentage of carbon dioxide volume concentration in the exhaust gas of the erythromycin primary fermentation unit, used to characterize the intensity of respiratory metabolism of the strain. Raman spectral data refers to the set of scattered signal intensities at characteristic wavenumbers of the fermentation broth collected by a Raman spectrometer, used to characterize the relative concentration changes of precursors, byproducts, and substrates in the erythromycin synthesis pathway.

[0055] In this embodiment, time-lag correlation analysis is performed on the sensed variable to obtain the response lag time of the sensed variable to the specific yield of erythromycin. This can be achieved in the following manner:

[0056] A time series analysis of the sensed variables for a fixed duration was performed to obtain multiple translation step sizes;

[0057] Determine the correlation coefficient between the sensor variable corresponding to each translation step size and the specific yield of erythromycin;

[0058] Then, the response lag time of the sensing variable to the specific yield of erythromycin was obtained by screening all correlation coefficients.

[0059] It should be noted that the response lag time in this application refers to the time delay required for a change in the sensing variable to produce an observable response in the specific yield of erythromycin. It is used to characterize the speed at which different sensing variables affect the specific yield of erythromycin. The response lag time of the sensing variable can constrain the timing of the sensing variable's participation in the regulatory decision.

[0060] In practice, a fixed duration of 48 hours can be set, which is also determined based on the complete fermentation cycle of the erythromycin progeny to cover the entire time range in which the sensing variable may have an impact. Using 1-hour increments, the historical time-series data of each sensing variable is shifted forward by multiple steps to obtain the corresponding shift steps for that sensing variable. For example, for the feeding rate, the time-series data of the feeding rate continuously collected within the historical period is used as the baseline sequence, and each shift is shifted forward by 48 steps, with each shift amount corresponding to one shift step. The 48 shift steps corresponding to each sensing variable can be obtained in this way. Then, for each translation step, the translated time-series data of the sensed variable is compared point-by-point with the erythromycin specific yield time-series data within the same time period. The correlation coefficient between the sensed variable and the erythromycin specific yield at that translation step is calculated. Specifically, when calculating the correlation coefficient, the average value of the translated time-series data of the sensed variable and the average value of the erythromycin specific yield time-series data are calculated separately. Then, the difference between the sensed variable value at each sampling time and its average value is multiplied by the difference between the erythromycin specific yield value at the same sampling time and its average value. The products of all sampling times are summed to obtain the numerator. The square root of the sum of the squares of the deviations of the sensed variable at each time and the sum of the squares of the deviations of the erythromycin specific yield at each time is taken as the denominator. The ratio obtained by dividing the numerator by the denominator is used as the correlation coefficient at that translation step. The correlation coefficient for each translation step is calculated sequentially using the above method to obtain the correlation coefficient corresponding to each translation step. Finally, the correlation coefficients corresponding to all the calculated translation step sizes are compared, and the translation step size corresponding to the maximum correlation coefficient is selected. This translation step size is used as the response lag time of the sensing variable to the specific yield of erythromycin.

[0061] In step S2, a modulation coefficient is generated based on the activity level of the strain, and the amplitude of the sensing variable is adjusted using the modulation coefficient to obtain the state descriptor in the erythromycin primary fermentation device.

[0062] In this embodiment, the modulation coefficient is generated based on the activity level of the strain in the following manner:

[0063] The activity level is determined based on the ratio of the strain's current specific growth rate to the average specific growth rate of historical batches.

[0064] The upper and lower limits of the modulation coefficient range were determined based on historical batch data from the primary fermentation of erythromycin.

[0065] The modulation coefficient is obtained by linear interpolation between the upper and lower limits of the modulation coefficient range based on the activity level.

[0066] In practice, firstly, the optical density value of the fermentation broth is collected in real time by an optical density sensor installed in the erythromycin primary fermentation equipment. The rate of change of optical density value over time is taken as the current specific growth rate of the strain. Simultaneously, the specific growth rates of all historical batches of the same strain at the same fermentation time are extracted from the historical batch database of erythromycin primary fermentation. The average specific growth rate of all historical batches is calculated as the average specific growth rate of historical batches. The current specific growth rate is divided by the average specific growth rate of historical batches, and the ratio obtained by the division is taken as the activity level of the strain. Then, fermentation data of multiple historical batches are extracted from the historical batch database of erythromycin primary fermentation. Each historical batch records the activity level of the strain in that batch and the sensor variable scaling ratio that achieves the optimal specific yield of erythromycin at that activity level. All historical batches are sorted from smallest to largest activity level. The average sensor variable scaling ratio corresponding to the top 10% of historical batches after sorting is taken as the lower limit of the modulation coefficient range, and the average sensor variable scaling ratio corresponding to the bottom 10% of historical batches after sorting is taken as the upper limit of the modulation coefficient range. Finally, using the activity level as the interpolation position and the lower and upper limits as the two endpoints of the interpolation interval, the modulation coefficient is calculated according to a linear proportional relationship. Specifically, the difference between the activity level and the lowest activity level in the historical batch is calculated, and the difference between the highest and lowest activity levels in the historical batch is calculated. The ratio of these two differences is used as a proportionality coefficient. Then, the difference between the upper and lower limits is calculated. This difference is multiplied by the proportionality coefficient and added to the lower limit. The sum is used as the modulation coefficient. The modulation coefficient is multiplied by the current reading of the sensor variable to complete the amplitude adjustment of the sensor variable.

[0067] It should be noted that the activity level in this application refers to the degree of deviation of the strain's current specific growth rate from the historical batch average specific growth rate. It is used to characterize the strength of the current strain's metabolic activity. An activity level greater than 1 indicates that the strain's metabolic activity is higher than the historical average, while an activity level less than 1 indicates that the strain's metabolic activity is lower than the historical average. The activity level is used as the positioning basis for linear interpolation in subsequent steps. The modulation coefficient refers to the scaling factor used to adjust the sensor variable reading based on the strain's activity level. It is used to match the amplitude of the sensor variable with the actual metabolic state of the current strain. A modulation coefficient greater than 1 indicates amplification of the sensor variable's response amplitude, while a modulation coefficient less than 1 indicates reduction of the sensor variable's response amplitude. The modulation coefficient is used to adjust the amplitude of the sensor variable to obtain a state descriptor.

[0068] In this embodiment, the state descriptor in the erythromycin primary fermentation device is obtained by adjusting the amplitude of the sensing variable using the modulation coefficient in the following manner:

[0069] The weighted amplitude of the sensed variable is determined by the modulation coefficient;

[0070] The state of the weighted amplitude is monitored to obtain the state change curve;

[0071] Linear fitting was performed on the state change curve to obtain the state descriptor of the erythromycin primary fermentation device.

[0072] In specific implementation, firstly, the current readings of the sensed variables are multiplied by the modulation coefficients to obtain the weighted amplitudes of the sensed variables. Specifically: the current reading of the feeding rate is multiplied by the modulation coefficient, and the resulting value is used as the weighted amplitude of the feeding rate; the current reading of the dissolved oxygen concentration is multiplied by the modulation coefficient, and the resulting value is used as the weighted amplitude of the dissolved oxygen concentration; the current reading of the carbon dioxide concentration is multiplied by the modulation coefficient, and the resulting value is used as the weighted amplitude of the carbon dioxide concentration; the scattered signal intensity at each characteristic wavenumber of the Raman spectral data is multiplied by the modulation coefficient, and the set of scattered signal intensities obtained is used as the weighted amplitude of the Raman spectral data. Then, the weighted amplitudes of the sensed variables are arranged sequentially according to the acquisition time sequence. The arranged weighted amplitude sequence is then smoothed using a sliding window to remove random noise introduced during the acquisition process. The continuous curve obtained after the sliding window smoothing is used as the state change curve. Finally, the state change curve is subjected to least squares linear fitting. During fitting, the acquisition time is used as the independent variable and the weighted amplitude is used as the dependent variable. The slope and intercept of the fitting line that minimizes the sum of squared residuals between the weighted amplitude of all sampling points and their predicted values ​​on the fitting line are calculated. The slope and intercept are organized and arranged according to the sensor variable dimension. The set of organized slopes and intercepts is used as the state descriptor in the erythromycin primary fermentation device.

[0073] It should be noted that the state descriptor in this application refers to a multidimensional parameter set containing the weighted amplitude fitting slope and intercept of the sensing variables, used to characterize the comprehensive operating state of the erythromycin primary fermentation equipment after strain activity modulation at the current moment. Each dimension in the state descriptor corresponds to a pair of fitting parameters for a sensing variable. The state descriptor is used to determine the metabolic offset together with the metabolic parameters in subsequent steps.

[0074] In step S3, the metabolic parameters of the strain are obtained, the metabolic offset is determined based on the state descriptor and the metabolic parameters, and the action category and execution time of quality regulation are determined based on the metabolic offset and the response lag time.

[0075] In specific implementation, the metabolic parameters of the strain are obtained based on the scattering signal from a Raman spectrometer. Specifically, the Raman scattering signal of the fermentation broth in the erythromycin primary fermentation device is acquired using a Raman spectrometer. Background subtraction and smoothing filtering are performed on the Raman scattering signal. The scattering peak intensities at the characteristic wavenumbers corresponding to propionyl-CoA and methylmalonyl-CoA in the erythromycin synthesis pathway are extracted from the processed Raman scattering signal. The extracted propionyl-CoA scattering peak intensities are substituted into a pre-established first standard curve to obtain the cumulative concentration of propionyl-CoA in the fermentation broth. The extracted methylmalonyl-CoA scattering peak intensities are substituted into a pre-established second standard curve to obtain the cumulative concentration of methylmalonyl-CoA in the fermentation broth. Then, the propionyl-CoA... The cumulative concentrations of A and methylmalonyl-CoA are used as metabolic parameters of the strain. The first standard curve is established as follows: multiple standard solutions of propionyl-CoA with known concentrations are prepared, and the Raman scattering peak intensities of each standard solution at the characteristic wavenumber of propionyl-CoA are collected. A linear fit is performed with concentration as the abscissa and peak intensity as the ordinate, and the fitted regression equation is used as the standard curve of propionyl-CoA. The second standard curve is established in the same way, except that the characteristic wavenumber corresponding to methylmalonyl-CoA is selected. The characteristic wavenumber of propionyl-CoA is around 1710 cm⁻¹, which corresponds to the carbonyl stretching vibration of propionyl-CoA. The characteristic wavenumber of methylmalonyl-CoA is around 1685 cm⁻¹, which corresponds to the ester bond stretching vibration of methylmalonyl-CoA.

[0076] It should be noted that the metabolic parameters in this application refer to the cumulative concentration set of key precursor substances in the erythromycin synthesis pathway in the fermentation broth, including the cumulative concentration of propionyl-CoA and methylmalonyl-CoA, which are used to characterize the sufficiency of the supply of precursors in the erythromycin synthesis pathway. The metabolic parameters can provide a quantitative basis at the metabolic level for quality control decisions.

[0077] Preferred, Reference Figure 3 As shown, this figure is an exemplary flowchart for determining metabolic offsets according to the present application. In this embodiment, determining metabolic offsets based on the state descriptor and the metabolic parameters can be achieved through the following steps:

[0078] In step S31, the carbon source consumption rate and dissolved oxygen consumption rate are extracted from the state descriptor, and the cumulative concentration of precursor substances is extracted from the metabolic parameters.

[0079] In step S32, baseline values ​​for the cumulative concentration, the carbon source consumption rate, and the dissolved oxygen consumption rate are set respectively, thereby determining the precursor deviation, carbon source deviation, and dissolved oxygen deviation.

[0080] In step S33, the precursor bias, the carbon source bias, and the dissolved oxygen bias are weighted and summed to obtain the metabolic offset.

[0081] In specific implementation, firstly, the fitting slope of the feeding rate and the fitting slope of the dissolved oxygen concentration are extracted from the state descriptor. The fitting slope of the feeding rate is taken as the carbon source consumption rate, and the fitting slope of the dissolved oxygen concentration is taken as the dissolved oxygen consumption rate. Then, the cumulative concentrations of propionyl-CoA and methylmalonyl-CoA are read from the metabolic parameters. The cumulative concentrations of propionyl-CoA and methylmalonyl-CoA are added together, and the sum is taken as the total cumulative concentration of the precursor substance. Then, multiple historical batches at which the specific yield of erythromycin was at its optimal state were extracted from the historical batch database of erythromycin primary fermentation. The total cumulative concentration of precursor substances, carbon source consumption rate, and dissolved oxygen consumption rate of each optimal batch were calculated. The average total cumulative concentration of precursor substances of all optimal batches was used as the cumulative concentration benchmark value. The average carbon source consumption rate of all optimal batches was used as the carbon source consumption rate benchmark value. The average dissolved oxygen consumption rate of all optimal batches was used as the dissolved oxygen consumption rate benchmark value. The difference between the total cumulative concentration of precursor substances and the cumulative concentration benchmark value was used as the precursor deviation. The difference between the carbon source consumption rate and the carbon source consumption rate benchmark value was used as the carbon source deviation. The difference between the dissolved oxygen consumption rate and the dissolved oxygen consumption rate benchmark value was used as the dissolved oxygen deviation. Finally, the precursor bias, carbon source bias, and dissolved oxygen bias are weighted and summed, and the sum is used as the metabolic offset. The weighting coefficients of precursor bias, carbon source bias, and dissolved oxygen bias are determined as follows: metabolic parameters and sensor variable data of each historical batch are extracted from the historical batch data of erythromycin primary fermentation. The correlation coefficients between precursor bias and erythromycin specific yield, carbon source bias and erythromycin specific yield, and dissolved oxygen bias and erythromycin specific yield are calculated respectively. All correlation coefficients are normalized so that the sum of the three correlation coefficients equals one. The three normalized correlation coefficients are used as the weighting coefficients of precursor bias, carbon source bias, and dissolved oxygen bias, respectively.

[0082] It should be noted that the metabolic offset in this application is a comprehensive quantitative index used to characterize the degree to which the actual metabolic state of the erythromycin primary fermentation device deviates from the ideal metabolic state at the current moment. This comprehensive quantitative index integrates three metabolic dimensions: the sufficiency of precursor supply, the carbon source consumption rate, and the dissolved oxygen consumption rate. The larger the value of the metabolic offset, the further the actual metabolic state deviates from the ideal state. The metabolic offset is used together with the response lag time to determine the action type and execution time of quality control.

[0083] In this embodiment, determining the action category and execution time of quality control based on the metabolic offset and the response lag time can be achieved in the following manner:

[0084] Obtain historical batch data and control cycles for the primary fermentation of erythromycin;

[0085] The first and second thresholds for quality control actions are determined using the historical batch data.

[0086] When the metabolic offset is below the first threshold, the action type is to maintain the current operating condition; when the metabolic offset is within the closed interval between the first threshold and the second threshold, the action type is to replenish feed for correction; when the metabolic offset exceeds the second threshold, the action type is to refactor metabolism.

[0087] When the response lag time is less than the control period, the execution time is the current time; when the response lag time is greater than or equal to the control period, the execution time is delayed to the time point corresponding to the sum of the current time and the response lag time.

[0088] In practice, firstly, complete fermentation data for all historical batches of erythromycin primary fermentation are extracted from the historical batch database. This complete fermentation data includes metabolic offset records for each historical batch at various sampling times and the final erythromycin specific yield for that batch. Simultaneously, the control cycle is set based on the physical response speed of the feed pump and temperature control system in the erythromycin primary fermentation equipment. Secondly, batches with optimal erythromycin specific yields are selected from all historical batches as normal operating condition batches. Metabolic offsets at various times are extracted from all normal operating condition batches. These metabolic offsets are then used to form a statistical sample set. The arithmetic mean and standard deviation of this statistical sample set are calculated. The arithmetic mean is added to one standard deviation to obtain a first threshold, and the arithmetic mean is added to two standard deviations to obtain a second threshold. Then, the metabolic offset at the current moment is compared with the first threshold and the second threshold, respectively. When the metabolic offset is lower than the first threshold, maintaining the current operating condition is taken as the action category for quality control. When the metabolic offset is greater than or equal to the first threshold and less than or equal to the second threshold, feeding correction is taken as the action category for quality control. When the metabolic offset exceeds the second threshold, metabolic reconstruction is taken as the action category for quality control. Finally, the minimum value of the response lag time of the sensor variable is obtained and compared with the control period. When the minimum value of the response lag time is less than the control period, the current moment is taken as the execution moment of quality control. When the minimum value of the response lag time is greater than or equal to the control period, the sum of the current moment and the minimum value of the response lag time is calculated, and the calculated moment is taken as the execution moment of quality control.

[0089] It should be noted that the first threshold in this application refers to the upper boundary value of the metabolic state within the normal fluctuation range, used to distinguish whether the metabolic state has deviated from the expected state requiring intervention. Values ​​below the first threshold indicate that the metabolic state is within the normal fluctuation range and no adjustment is needed. The second threshold refers to the extreme boundary value of the metabolic state within the acceptable deviation range, used to distinguish between mild and severe deviations. Values ​​exceeding the second threshold indicate that the metabolic state has severely deviated from the ideal state and metabolic remodeling is required. Values ​​between the first and second thresholds indicate a mild deviation in the metabolic state, requiring only a small correction to the glucose supplementation rate. Both the first and second thresholds are determined based on the statistical distribution of metabolic deviations under historical normal operating conditions. The first threshold is the sum of the mean and one standard deviation in the distribution of metabolic deviations under normal operating conditions, covering 68% of the metabolic fluctuation range under normal operating conditions; the second threshold is the sum of the mean and two standard deviations, covering 95% of the metabolic fluctuation range under normal operating conditions. The control period is the sum of the response delay time of the feed pump and the thermal inertia time of the temperature control system in the erythromycin primary fermentation equipment. It is used to compare with the response lag time to determine whether the control action should be executed immediately at the current moment or delayed until the time point corresponding to the response lag time. The action categories include three types: maintain current operating conditions, feed correction, and metabolic remodeling. Maintaining current operating conditions means not performing any control operation, feed correction means only adjusting the sugar supply rate, and metabolic remodeling means simultaneously adjusting the temperature setpoint, dissolved oxygen concentration setpoint, and sugar supply rate.

[0090] In step S4, when the action category is metabolic remodeling, the temperature, dissolved oxygen concentration, and sugar replenishment rate in the erythromycin primary fermentation equipment are controlled in a closed loop based on the metabolic offset at the execution time.

[0091] In this embodiment, when the action category is metabolic remodeling, the closed-loop control of temperature, dissolved oxygen concentration, and sugar supplementation rate in the erythromycin primary fermentation equipment based on the metabolic offset at the execution time can be specifically carried out in the following manner:

[0092] At the execution time, the temperature reduction range, dissolved oxygen increase range, and glucose replenishment rate are set according to the metabolic offset.

[0093] The temperature setpoint is lowered according to the temperature reduction range, the dissolved oxygen concentration setpoint is increased according to the dissolved oxygen increase range, the sugar replenishment rate is proportionally adjusted, and real-time sensor variables are collected during the adjustment process.

[0094] The metabolic offset is updated based on the real-time sensor variables, and the temperature reduction, dissolved oxygen up-reduction, and glucose replenishment rate are re-determined based on the updated metabolic offset until the metabolic offset is lower than a first threshold.

[0095] In specific implementation, firstly, when the execution time arrives, the carbon source deviation is extracted from the metabolic offset. The absolute value of the carbon source deviation is multiplied by the temperature adjustment coefficient to obtain the temperature reduction magnitude. The absolute value of the carbon source deviation is multiplied by the dissolved oxygen adjustment coefficient to obtain the dissolved oxygen increase magnitude. Simultaneously, the adjustment direction of the sugar replenishment rate is determined according to the sign of the carbon source deviation. When the carbon source deviation is positive, the adjustment direction is to decrease the sugar replenishment rate; when the carbon source deviation is negative, the adjustment direction is to increase the sugar replenishment rate. The absolute value of the carbon source deviation is multiplied by the sugar replenishment rate adjustment coefficient to obtain the sugar replenishment rate adjustment step size. The temperature adjustment coefficient, dissolved oxygen adjustment coefficient, and sugar replenishment rate adjustment coefficient are determined as follows: from the historical batch data of erythromycin primary fermentation, all historical batches that have undergone closed-loop regulation are screened, and the initial carbon source deviation and the final stable regulation value of each batch are extracted. The actual temperature adjustment, dissolved oxygen adjustment, and sugar replenishment rate adjustment were determined. Linear regression was performed with carbon source deviation as the independent variable and actual temperature adjustment as the dependent variable, and the regression coefficients obtained were used as the temperature adjustment coefficients. Linear regression was also performed with carbon source deviation as the independent variable and actual dissolved oxygen adjustment as the dependent variable, and the regression coefficients obtained were used as the dissolved oxygen adjustment coefficients. Similarly, linear regression was performed with carbon source deviation as the independent variable and actual sugar replenishment rate adjustment as the dependent variable, and the regression coefficients obtained were used as the sugar replenishment rate adjustment coefficients. The linear regression involved plotting all data points on a two-dimensional coordinate system with carbon source deviation as the abscissa and the corresponding adjustment amount as the ordinate, finding a straight line that minimizes the sum of the squares of the vertical distances from all data points to that line, and using the slope of that line as the adjustment coefficient.

[0096] Then, the temperature setpoint of the erythromycin primary fermentation equipment is lowered from its current value according to the stated temperature reduction range, and the dissolved oxygen concentration setpoint is increased from its current value according to the stated dissolved oxygen increase range. The sugar replenishment rate is proportionally adjusted according to the stated adjustment direction and adjustment step size. During the adjustment process, real-time sensor variables are continuously collected using a Raman spectrometer and a tail gas spectrometer. Finally, the carbon source consumption rate and dissolved oxygen consumption rate are recalculated based on the real-time sensor variables. The updated carbon source consumption rate is subtracted from the carbon source consumption rate baseline to obtain the updated carbon source deviation. The adjustment step size of the temperature reduction range, dissolved oxygen increase range, and sugar replenishment rate is re-determined based on the updated carbon source deviation. The temperature setpoint, dissolved oxygen concentration setpoint, and sugar replenishment rate are adjusted again according to the re-determined range and step size. The above recalculation and readjustment process is repeated until the recalculated metabolic offset is lower than the first threshold, at which point closed-loop control is stopped.

[0097] It should be noted that the temperature adjustment range in this application refers to the amount by which the temperature setpoint is reduced based on the carbon source deviation, used to slow down the respiratory metabolic intensity of the strain and reduce excessive carbon source consumption to carbon dioxide; the dissolved oxygen adjustment range refers to the amount by which the dissolved oxygen concentration setpoint is increased based on the carbon source deviation, used to improve the redox balance in the precursor synthesis process by increasing oxygen supply; the direction of glucose replenishment rate adjustment refers to the direction of glucose replenishment rate change determined by the positive or negative carbon source deviation. When carbon source consumption is too fast, the glucose replenishment rate is reduced to avoid metabolic overflow caused by carbon source excess; when carbon source consumption is too slow, the glucose replenishment rate is increased to supplement carbon source supply; the adjustment step size of the glucose replenishment rate refers to the amount of change in the glucose replenishment rate in each adjustment, used to control the fineness of the adjustment and avoid drastic fluctuations in metabolic state caused by excessive single adjustment. Closed-loop regulation refers to an iterative control process that continuously monitors sensor variables and repeatedly corrects temperature, dissolved oxygen concentration, and glucose replenishment rate accordingly. This process continues to run after the regulation action is triggered until the metabolic deviation returns to the normal range, used to ensure that the intensity of the regulation action always matches the actual metabolic demand and avoids over-regulation or under-regulation.

[0098] In summary, the technical solution adopted in this application can realize closed-loop regulation of fermentation state by modifying the activity of fusion strains, compensating for response lag, and determining metabolic offset, so as to improve the adaptive adjustment ability of the erythromycin primary fermentation process to abnormal metabolic changes.

[0099] Example 2: This application provides a quality control system for the fermentation production of erythromycin primary stock, referencing... Figure 4 As shown, this figure is a modular structure diagram of a quality control system for the fermentation production of erythromycin primary body according to this application. The quality control system includes:

[0100] The monitoring and analysis module 100 is used to monitor the sensing variables in the erythromycin primary fermentation equipment, perform time-delay correlation analysis on the sensing variables, and obtain the response lag time of the sensing variables to the specific yield of erythromycin.

[0101] The state generation module 200 is used to generate a modulation coefficient based on the activity level of the strain, and adjust the amplitude of the sensing variable through the modulation coefficient to obtain the state descriptor in the erythromycin primary fermentation device.

[0102] The metabolic decision module 300 is used to acquire the metabolic parameters of the strain, determine the metabolic offset based on the state descriptor and the metabolic parameters, and determine the action category and execution time of quality regulation based on the metabolic offset and the response lag time.

[0103] The regulation execution module 400 is used to perform closed-loop regulation of the temperature, dissolved oxygen concentration and sugar supplementation rate in the erythromycin primary fermentation equipment based on the metabolic offset at the execution time when the action category is metabolic remodeling.

[0104] This application is described with reference to flowchart illustrations and / or block diagrams of methods, apparatus (systems), and computer program products according to embodiments of this application. It will be understood that each block of the flowchart illustrations and / or block diagrams, as well as combinations of blocks in the flowchart illustrations and / or block diagrams, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, special-purpose computer, embedded processor, or other programmable data processing apparatus to produce a machine, such that the instructions, which execute via the processor of the computer or other programmable data processing apparatus, generate instructions for implementing the flowchart. Figure 1 One or more processes and / or boxes Figure 1 A device that provides the functions specified in one or more boxes.

Claims

1. A quality control method for the fermentation production of erythromycin primary stock, characterized in that, The quality control method includes the following steps: The sensor variables in the erythromycin primary fermentation equipment were monitored, and time-delay correlation analysis was performed on the sensor variables to obtain the response lag time of the sensor variables to the specific yield of erythromycin. A modulation coefficient is generated based on the activity level of the strain, and the amplitude of the sensing variable is adjusted by the modulation coefficient to obtain the state descriptor in the erythromycin primary fermentation device. Obtain the metabolic parameters of the strain, determine the metabolic offset based on the state descriptor and the metabolic parameters, and determine the action category and execution time of quality regulation based on the metabolic offset and the response lag time; When the action category is metabolic remodeling, the temperature, dissolved oxygen concentration, and sugar replenishment rate in the erythromycin primary fermentation equipment are controlled in a closed loop based on the metabolic offset at the execution time.

2. The quality control method for the fermentation production of erythromycin primary culture as described in claim 1, characterized in that, Sensing variables in the erythromycin primary fermentation equipment were monitored using Raman spectroscopy and tail gas spectroscopy; the sensing variables included feed rate, dissolved oxygen concentration, carbon dioxide concentration, and Raman spectral data.

3. The quality control method for the fermentation production of erythromycin primary culture as described in claim 1, characterized in that, A time-lag correlation analysis was performed on the sensed variables to obtain the specific lag time of the sensed variables in response to the specific yield of erythromycin: A time series analysis of the sensed variables for a fixed duration was performed to obtain multiple translation step sizes; Determine the correlation coefficient between the sensor variable corresponding to each translation step size and the specific yield of erythromycin; Then, the response lag time of the sensing variable to the specific yield of erythromycin was obtained by screening all correlation coefficients.

4. The quality control method for the fermentation production of erythromycin primary culture as described in claim 1, characterized in that, The modulation coefficient generated based on the activity level of the strain specifically includes: The activity level is determined based on the ratio of the strain's current specific growth rate to the average specific growth rate of historical batches. The upper and lower limits of the modulation coefficient range were determined based on historical batch data from the primary fermentation of erythromycin. The modulation coefficient is obtained by linear interpolation between the upper and lower limits of the modulation coefficient range based on the activity level.

5. The quality control method for the primary fermentation production of erythromycin as described in claim 1, characterized in that, The state descriptor obtained by adjusting the amplitude of the sensed variable using the modulation coefficient specifically includes: The weighted amplitude of the sensed variable is determined by the modulation coefficient; The state of the weighted amplitude is monitored to obtain the state change curve; Linear fitting was performed on the state change curve to obtain the state descriptor of the erythromycin primary fermentation device.

6. The quality control method for the fermentation production of erythromycin primary culture as described in claim 1, characterized in that, Metabolic parameters of the strain were obtained based on the scattering signals from a Raman spectrometer.

7. The quality control method for the primary fermentation production of erythromycin as described in claim 1, characterized in that, Determining the metabolic offset based on the state descriptor and the metabolic parameters specifically includes: The carbon source consumption rate and dissolved oxygen consumption rate are extracted from the state descriptor, and the cumulative concentration of precursor substances is extracted from the metabolic parameters. The cumulative concentration, the carbon source consumption rate, and the dissolved oxygen consumption rate are set as baseline values ​​respectively, thereby determining the precursor deviation, carbon source deviation, and dissolved oxygen deviation. The metabolic offset is obtained by weighted summation of the precursor bias, the carbon source bias, and the dissolved oxygen bias.

8. The quality control method for the fermentation production of erythromycin primary culture as described in claim 1, characterized in that, Determining the action category and execution time of quality regulation based on the metabolic offset and the response lag time specifically includes: Obtain historical batch data and control cycles for the primary fermentation of erythromycin; The first and second thresholds for quality control actions are determined using the historical batch data. When the metabolic offset is below the first threshold, the action type is to maintain the current operating condition; when the metabolic offset is within the closed interval between the first threshold and the second threshold, the action type is to replenish feed for correction; when the metabolic offset exceeds the second threshold, the action type is to refactor metabolism. When the response lag time is less than the control period, the execution time is the current time; when the response lag time is greater than or equal to the control period, the execution time is delayed to the time point corresponding to the sum of the current time and the response lag time.

9. The quality control method for the primary fermentation production of erythromycin as described in claim 1, characterized in that, When the action category is metabolic remodeling, the closed-loop control of temperature, dissolved oxygen concentration, and sugar supplementation rate in the erythromycin primary fermentation equipment based on the metabolic offset at the execution time specifically includes: At the execution time, the temperature reduction range, dissolved oxygen increase range, and glucose replenishment rate are set according to the metabolic offset. The temperature setpoint is lowered according to the temperature reduction range, the dissolved oxygen concentration setpoint is increased according to the dissolved oxygen increase range, the sugar replenishment rate is proportionally adjusted, and real-time sensor variables are collected during the adjustment process. The metabolic offset is updated based on the real-time sensor variables, and the temperature reduction, dissolved oxygen up-reduction, and glucose replenishment rate are re-determined based on the updated metabolic offset until the metabolic offset is lower than a first threshold.

10. A quality control system for the fermentation production of erythromycin primary molecules, characterized in that, A quality control system for performing a quality control method for the primary fermentation production of erythromycin as described in any one of claims 1 to 9, the quality control system comprising: The monitoring and analysis module is used to monitor the sensor variables in the erythromycin primary fermentation equipment, perform time-delay correlation analysis on the sensor variables, and obtain the response lag time of the sensor variables to the specific yield of erythromycin. The state generation module is used to generate modulation coefficients based on the activity level of the strain, and adjust the amplitude of the sensing variable through the modulation coefficients to obtain the state descriptor in the erythromycin primary fermentation device. The metabolic decision module is used to acquire the metabolic parameters of the strain, determine the metabolic offset based on the state descriptor and the metabolic parameters, and determine the action category and execution time of quality regulation based on the metabolic offset and the response lag time. The control execution module is used to perform closed-loop control of the temperature, dissolved oxygen concentration and sugar supplementation rate in the erythromycin primary fermentation equipment based on the metabolic offset at the execution time when the action category is metabolic remodeling.

Images