CN106701846B

CN106701846B - Method for producing sodium gluconate by on-line monitoring and optimizing aspergillus niger fermentation

Info

Publication number: CN106701846B
Application number: CN201710039712.8A
Authority: CN
Inventors: 杭海峰; 史梦菲; 庄英萍; 储炬; 张明; 李英英; 王伟飞; 苏立宇
Original assignee: East China University of Science and Technology
Current assignee: East China University of Science and Technology
Priority date: 2017-01-18
Filing date: 2017-01-18
Publication date: 2020-09-18
Anticipated expiration: 2037-01-18
Also published as: CN106701846A

Abstract

The invention relates to a method for producing sodium gluconate by on-line monitoring and optimization of Aspergillus niger fermentation. In order to know the change dynamics of raw materials and products in the aspergillus niger fermentation process in real time and high efficiency, the inventor applies a mid-infrared spectrometer to real-time detection and analyzes the contents of glucose and sodium gluconate in fermentation liquor by optimizing modeling. And based on the real-time detection result, the inventor also optimizes the production process of the sodium gluconate.

Description

Method for producing sodium gluconate by on-line monitoring and optimizing aspergillus niger fermentation

Technical Field

The invention belongs to the field of monitoring and optimizing a biological reaction process, and particularly relates to a method for monitoring and optimizing the production of sodium gluconate by fermenting Aspergillus niger on line.

Background

The sodium gluconate is an organic acid salt with good stability, no toxicity and no corrosiveness, and is widely applied to medicines, foods, chemical industry, light industry and the like. Sodium gluconate is obtained industrially mainly by fermentation of aspergillus niger. Aspergillus niger converts glucose as a substrate into gluconic acid-lactone through Glucose Oxidase (GOD), and then hydrolyzes the gluconic acid into gluconic acid, and the gluconic acid reacts with a neutralizer NaOH to generate sodium gluconate. Aspergillus niger synthesizes GOD and becomes a carrier of the GOD, and the reaction has the characteristic of high-efficiency catalytic reaction of enzyme. Therefore, the concentration of the substrate glucose and the product sodium gluconate change rapidly in the fermentation process, the existing off-line measurement method is time-consuming and labor-consuming, the change cannot be observed in real time, and the judgment of the fermentation process is influenced. The rapid and real-time online monitoring of the concentration changes of the substrate and the product becomes the key for improving the reaction rate and further improving the yield.

Mid-Infrared (MIR) spectroscopy is a novel fast, pollution-free, non-destructive, and green analytical technique that does not require sample pretreatment. The region of the map (4000- & lt400 & gt cm)^-1) Particularly in a fingerprint area, the functional group of a substance molecule can generate a special absorption peak, so that not only can the complex structure of the substance be identified, but also the substance can be quantitatively analyzed. The use of an Attenuated Total Reflection (ATR) probe can offset the strong absorption of MIR to water and solve the problem of weak penetration. In order to obtain more useful information from maps, quantitative applications of MIR require the use of chemometric methods, the most common of which is Partial Least Squares (PLS). And (3) establishing a quantitative model by using a chemometric method, and predicting the substrate and the product in the fermentation liquid. In the current stage, off-line sampling monitoring is mainly adopted, and the accuracy of model prediction also hinders the development of the MIR technology. The invention optimizes the modeling method and improvesThe accuracy of model prediction is realized, and the MIR online and in-situ monitoring fermentation process is realized. And (3) guiding and optimizing the fermentation process of producing sodium gluconate by Aspergillus niger by combining with online MIR. However, the use of MIR is essentially off-line for the quantification or characterization of certain substances and is not widely used for monitoring fermentation processes. For monitoring the fermentation process, offline sampling is mostly used for measurement, online MIR is less, and the method is applied to a yeast or escherichia coli system. The invention monitors glucose and sodium gluconate in the fermentation process of producing sodium gluconate by aspergillus niger by using online MIR for the first time. The main difficulty is the problem of prediction accuracy, so that a proper method needs to be found out to realize online and in-situ detection.

Disclosure of Invention

The invention aims to provide a method for producing sodium gluconate by on-line monitoring and optimizing aspergillus niger fermentation.

In a first aspect of the present invention, there is provided a method for monitoring a fermentation process of producing sodium gluconate by fermentation of aspergillus niger on line, the method comprising: and collecting fermentation liquor on line, obtaining a fermentation liquor map by using a mid-infrared spectrometer, and analyzing the contents of glucose and sodium gluconate in the fermentation liquor.

In a preferred embodiment, the online collected fermentation liquor map is introduced into an optimally designed PLS model, and the contents of glucose and sodium gluconate in the fermentation liquor are analyzed.

In another preferred embodiment, the optimally designed PLS model is established as follows:

(1) designing standard solutions of glucose and sodium gluconate in a gradient manner, and obtaining a correction set glucose and sodium gluconate binary-culture medium standard solution and a verification set glucose and sodium gluconate binary-culture medium standard solution through full factor design; importing the maps of the correction set and the verification set into spectral analysis software, and obtaining a model with relatively highest performance index by changing modeling parameters (including a modeling method, a modeling area and a preprocessing method); in the standard solution, a culture medium for removing glucose is used as a background;

(2) and (3) on the basis of the model in the step (1), adding the online collected fermentation liquor atlas in the correction set to optimize the model, and obtaining the PLS model with optimized design.

In another preferred example, in the step (2), the method further includes: for the fermentation liquor collected on line, detecting by a chemical method to obtain the content data of glucose and sodium gluconate in the fermentation liquor sample; correlating the acquired map with the detection data of the chemical method, and analyzing the model by using a multiple regression analysis method, thereby obtaining the model with the relatively highest performance index.

In another preferred embodiment, the spectrum analysis software is TQ analysis software.

In another preferred embodiment, the glucose-removing medium comprises: corn steep liquor, MgSO₄·7H₂O、KH₂PO₄、(NH₄)₂HPO₄(ii) a Preferably, it comprises: corn steep liquor 2 +/-0.5 g/L, MgSO₄·7H₂O 0.2±0.05g/L、KH₂PO₄0.17±0.05g/L、(NH₄)₂HPO₄0.25±0.05g/L。

In another preferred embodiment, the fermentation liquor is collected at 38 +/-0.5 ℃ to obtain a fermentation liquor spectrum.

In another aspect of the present invention, there is provided a method for increasing the reaction rate of sodium gluconate produced by fermentation of aspergillus niger, the method comprising: adding oxygen with oxygen amount higher than 8% in the reaction process; more preferably, oxygen is added with the oxygen amount of 8-15 percent; more preferably, oxygen is added at an oxygen flux of 9-11%.

In a preferred embodiment, the fermentation medium is: glucose 300 + -40 g/L, corn steep liquor 2 + -0.5 g/L, MgSO₄·7H₂O 0.2±0.05 g/L、KH₂PO₄0.17±0.05g/L、(NH₄)₂HPO₄0.25±0.05 g/L。

In another preferred embodiment, in the early stage of the method, the glucose concentration in the seed culture medium is 300 +/-15 g/L; more preferably 300 + -10 g/L; still more preferably 300. + -.5 g/L.

Other aspects of the invention will be apparent to those skilled in the art in view of the disclosure herein.

Drawings

FIG. 1 shows the mid-infrared spectra of the calibration set and the validation set characteristic regions of glucose and sodium gluconate binary-medium standard samples.

FIG. 2 shows a concentration distribution diagram of a series of glucose and sodium gluconate concentration gradients randomly selected to prepare a standard mixture using the principle of scatter point distribution.

FIG. 3 is an off-line collected mid-infrared spectrum of a fermentation broth.

FIG. 4 is a mid-infrared spectrum of an online collected fermentation broth.

Fig. 5 shows the glucose and sodium gluconate binary-medium solutions obtained from the calibration set and the validation set by the full factor design (full factor design) method.

FIG. 6 is a standard sample map of glucose and sodium gluconate binary-medium for the improved model.

FIG. 7 is a graph showing a concentration distribution after supplementing an online collected fermentation broth spectrum. The diamond marks are concentration distribution of a modeling set, the square marks are verification sets, and the two are components of an original model; the triangular mark is the concentration of the newly supplemented fermentation broth spectrum.

Fig. 8, predicted values of glucose (a) and sodium gluconate (B) are compared with reference values.

FIG. 9, MIR on-line monitoring fermentation process diagram.

FIG. 10, an infrared spectrum of the fermentation broth is acquired on line by MIR.

FIG. 11, MIR on-line monitoring Aspergillus niger fermentation process.

FIG. 12, comparison of glucose changes in control and experimental groups.

Fig. 13, comparison of sodium gluconate changes in control and experimental groups.

FIG. 14, concentration change of glucose (A) and sodium gluconate (B) during fermentation with different initial sugar concentrations.

Detailed Description

The inventor is dedicated to the optimization of the process of producing sodium gluconate by aspergillus niger fermentation, and carries out deep research in order to know the change dynamics of raw materials and products in the aspergillus niger fermentation production process in real time and high efficiency. Based on the real-time detection result, the inventor also optimizes the production process of the sodium gluconate.

The invention utilizes FT-MIR-ATR combined with PLS method to establish a glucose and sodium gluconate prediction model, and simultaneously monitors the concentration of substrate glucose and product sodium gluconate in the fermentation process, thereby realizing the on-line control of the fermentation process of producing sodium gluconate by Aspergillus niger. Under the condition of changing aeration conditions, the model can be used for simply and quickly predicting the concentration change of the substrate and the product, checking the accuracy of the model and guiding the optimization of the fermentation process.

In the prior art, in the process of producing sodium gluconate by fermenting aspergillus niger, a high performance liquid method is generally used for measurement, the method consumes time, and the method is also used for pretreating a sample, filtering out thalli, diluting and the like, so that the sample is damaged. Therefore, the inventor changes the detection method and applies the mid-infrared spectrometer to real-time detection, but the inventor also finds that the fermentation liquor of the aspergillus niger is different from the commonly prepared standard liquor, the components are complex, the influence of the peak generated by impurities is easy to be influenced during modeling, and the factors can influence the quantification, so that the model to be established by using the mid-infrared measurement is complex, and the influencing factors in the modeling process need to be searched, determined and eliminated to establish an optimized model.

In the modeling process, the inventor uses the prepared standard solution to establish a model and measure the glucose and the sodium gluconate in the fermentation liquid, and the obtained measurement result has large difference with the actual concentration of the glucono-sodium gluconate. Through analysis and repeated experiments, the inventor changes the establishment mode of the standard solution, and the prediction accuracy is improved by taking the culture medium (containing other components except glucose in the culture medium) without glucose as the background.

On the basis of the above optimization, although the measurement result is less different from the true value, the present inventors expected that the measurement result is more accurate by further optimization. Therefore, the inventor applies the online acquired spectrum of the fermentation liquor to establish a model on the basis of the original model established by the standard liquor, and the accuracy of prediction is obviously improved by applying the model.

Therefore, the invention provides a method for establishing an on-line monitoring method for the concentrations of glucose and sodium gluconate in the fermentation process of Aspergillus niger by using an analysis technology of Fourier transform mid-infrared spectroscopy combined with attenuated total reflectance (FT-MIR-ATR). As a preferred embodiment of the invention, the method comprises the following steps: (1) and collecting the standard mixed liquid spectrum designed by the full factor by using a mid-infrared spectrometer. (2) And the spectrum of the fermentation liquor is acquired on line by the intermediate infrared spectrometer, and the contents of glucose and sodium gluconate in the fermentation liquor sample are obtained by chemical detection. (3) And correlating the acquired spectrum with chemical method detection data, and establishing a model by using a multiple regression analysis method. (4) And (4) introducing the fermentation liquor spectrum with unknown glucose and sodium gluconate contents acquired on line into a prediction model to predict the content of the target substance. In a specific example, in the aspergillus niger fermentation process of on-line monitoring, the predicted glucose RMSEP is 3.24g/L, the relative error is 2.21%, and the sodium gluconate RMSEP is 4.03g/L, the relative error is 2.77%.

The invention mainly has the outstanding technical improvement and innovation points that:

(1) uses on-line mid-infrared spectroscopy

The mid-infrared spectroscopy is a rapid and efficient detection method. A common method for measuring sodium gluconate is high performance liquid. However, this method requires complicated pretreatment such as preparation of a mobile phase, filtration of a sample, dilution, and the like, and it takes more than 10 minutes to measure one sample. The medium infrared spectroscopy does not need mobile phase preparation, can directly measure fermentation liquor, does not need sample pretreatment, does not damage samples, and only needs tens of seconds for measurement. The advantage in terms of rapidity is extremely evident. The mid-infrared spectrum can be used for monitoring the fermentation process on line, and the real-time performance and the timeliness are more highlighted.

(2) Technical improvement of model

Aiming at the problem of the prediction accuracy of the components of the fermentation liquor, the inventor improves a model, utilizes an experimental design-full factor design method for modeling, uses a culture medium except glucose as a background, and adds an online collected map in a correction set to realize the improvement of the model and improve the prediction accuracy; the MIR technology of the present invention enables on-line, in-situ detection of substrates and products, without off-line sampling, fast and simple) to prove that this patent first proposes such an application, thereby improving the novelty or creativity of this patent).

In the optimized method of the invention, the influence of other substances in the fermentation medium on the quantification of the target substance can be filtered against the background of the medium components from which glucose is removed. The conventional process is against a background of water.

In the optimization method, the spectrum of the standard mixed liquor which is randomly selected and distributed in scattered points is combined with the spectrum of the fermentation liquor which is collected on line from the spectrum of the standard mixed liquor which is designed by full factors (an experimental design method). And the method is not limited to applying conventional standard solution for modeling, so that the detection accuracy is obviously improved.

(3) Improvements in fermentation processes

The inventor utilizes the mid-infrared spectroscopy to detect the fermentation process of the aspergillus niger on line, finds key factors in the fermentation process, selects the optimal oxygen flux and initial sugar concentration, and optimizes the fermentation process.

The invention will be further illustrated with reference to the following specific examples. It should be understood that these examples are for illustrative purposes only and are not intended to limit the scope of the present invention. The experimental procedures, for which specific conditions are not noted in the following examples, are generally performed according to conventional conditions such as those described in J. SammBruk et al, molecular cloning protocols, third edition, scientific Press, 2002, or according to the manufacturer's recommendations.

Materials and methods

(1) Bacterial strains

The strain used in the invention is Aspergillus nigerAspergillus niger) Obtained from Shandong Fuyang Biotech, Inc.

Culture medium

Slant culture medium: 60g/L glucose (monohydrate), 1g/L, MgSO corn steep liquor₄·7H₂O 0.02g/L、KH₂PO₄0.13g/L of urea and 0.2g/L, CaCO of urea₃5g/L agar and 20g/L agar.

Seed culture medium: 250g/L glucose (monohydrate) and 8g/L, MgSO g corn steep liquor₄·7H₂O 0.17g/L、KH₂PO₄0.6g/L、(NH₄)₂HPO₄0.25g/L。

Fermentation medium: 330g/L glucose (monohydrate), 2g/L, MgSO corn steep liquor₄·7H₂O 0.2g/L、KH₂PO₄0.17g/L、(NH₄)₂HPO₄0.25g/L。

Culture conditions

Slant culture: picking the single spore colony on the plate to an eggplant bottle, and culturing for 60-72h at 35 ℃.

Culturing in a fermentation tank: in the following examples, a 5L stirred bioreactor (Shanghai enhanced Biochemical engineering Equipment Co., Ltd.) was used for the batch fermentation of Aspergillus niger. The working volume in the tank is 3L, the temperature is 38 ℃, the ventilation volume is 4vvm, and the initial stirring speed is 500 r/min. During the fermentation process, 250g/L NaOH solution is added to adjust the pH value to 5.5. The seed tank was inoculated with 50mL of fresh Aspergillus niger spore suspension, and then 0.5L of the seed tank fermentation broth was transferred to the fermentor.

Off-line measuring method

a. Glucose

Taking a fermentation liquid sample in the fermentation process, centrifuging to take the supernatant, diluting to a proper concentration, and measuring the glucose concentration in the fermentation liquid by using a glucose kit.

Sodium gluconate

Agilent 1100 High Performance Liquid Chromatography (HPLC) from Agilent ltd measures the concentration of sodium gluconate. Measurement conditions were as follows: C18-H column (4.6mm × 250mm, 5 μm), column temperature 26 deg.C, mobile phase V (10% methanol):V (2.4% phosphoric acid) = 1: 1, flow rate 1mL/min, retention time 8min, sample injection amount 20 μ L, detector wavelength 210 nm. The data obtained serve as reference values for the on-line measurement.

Spectrometer and collection conditions

The on-line mid-infrared spectrogram was acquired by a U.S. Thermo NICOLET iS10 type fourier transform mid-infrared spectrometer equipped with a diamond ATR probe.

During collection, the temperature in the tank was kept constant at 38 ℃ against the background of the components of the fermentation medium excluding glucose. Wave number range is 4000-650cm^-1The resolution is 4cm < -1 >, the scanning times are 16 times, and liquid nitrogen is added in time to cool the detector. The self-contained software Result 3.0 collects the atlas on line, completes the on-line prediction and transmits the atlas to the Biostar software. PLS model creation is done by TQ analysis 9.0 software.

Example 1 Infrared Spectroscopy

Glucose and sodium gluconate binary-culture medium standard samples are prepared by a full factor design method, and infrared standard maps are collected. Because the chemical structures of the glucose and the sodium gluconate are similar, the absorption peak areas are similar and are respectively 950-^-1、1177-1505cm^-1、1505-1650cm^-1Three regions, as in fig. 1. With varying glucose and sodium gluconate concentrations, the peak areas are also changing. Sodium gluconate has one more carboxyl structure than glucose, and thus can distinguish two substances according to the characteristics. 1505 and 1650cm^-1The region forms a single strong peak, which is caused by the stretching vibration of the C = O double bond in the carboxyl structure of the sodium gluconate and is a peak specific to the sodium gluconate, but because the peak is also the position of the water peak, even if the background is removed, the noise is not smooth. 1177-1505cm^-1The double peaks of the regions are caused by COO-stretching vibration of sodium gluconate and C-H bending vibration of glucose and sodium gluconate. 950-^-1The double peak is mainly caused by C-O stretching vibration of the two. The peaks of the latter two regions both contribute, and these three regions are selected as modeled quantitative regions. As can be seen from the figure, the original patterns of the two substances are very similar, and the content of the target substance and the peak value of the absorption peak are not simply linear relationship, and need to be carried out by means of chemometricsAnd (5) carrying out quantitative analysis.

The modeling method comprises the following steps: and intercepting a proper modeling area according to the map features acquired by the infrared spectrometer, then properly changing parameters, and performing calculation modeling by using software.

Example 2 PLS modeling

A series of concentration gradients of glucose and sodium gluconate are randomly selected by using the principle of scatter point distribution to prepare a standard mixed solution, and the concentration distribution is shown in figure 2.

And (3) collecting the infrared spectrum of the series of standard mixed solutions by taking water as a collection background and taking the conditions in the step (5) of the spectrometer and the collection condition as collection conditions. It is imported into the TQ Analyst 9.0 software by selecting appropriate modeling parameters such as: a modeling method, a preprocessing method, a modeling area, and the like, as shown in table 1, to obtain a model with the highest performance index. The modeling method of the model is a PLS method.

TABLE 1 Performance index and correlation coefficient of the model under each parameter

Note: of the four-digit numbers in column 1 of the table, the first number from left to right represents the modeled region: 2 is two characteristic peaks, and 3 is three characteristic peaks; the second digital representation map preprocessing method: 1 is the original spectrum, 2 is the first derivative; the third figure is the smoothing method: 1 is not smooth, 2 is Savitzky-Golay filter; the fourth number is the baseline correction method: 1 is uncorrected, 2 is linear correction, and 3 is curvilinear correction.

It can be seen from the above table that 3222, namely the modeling area, is three characteristic peaks, the performance index of the preprocessing method of the first derivative, Savitzky-Golayfilter and linear baseline correction is the highest and is 98.0, the RMSEP of glucose is the lowest and the RMSEP of sodium gluconate is the lowest and is 1.57, and both similarity coefficients are 0.9999, so the parameter is selected for modeling.

Based on the model, the concentrations of glucose and sodium gluconate in the fermentation broth were predicted offline, and the results are shown in table 2.

TABLE 2 model prediction of glucose and sodium gluconate concentrations in fermentation broths

From the above table data, it can be seen that: the error percentage of the low concentration is large, and the high concentration is relatively small. The percent mean error for glucose was 6.51% and the root mean square error for prediction, RMSEP, was 8.73 g/L. The average error percentage of the sodium gluconate is 5.27 percent, and the RMSEP is 5.06 g/L. This result indicates that the error between the predicted value and the actual value is large, and it is necessary to find the cause of the error.

Based on the above problems, the present inventors have analytically studied the causes that may cause such errors. Aspects of the analysis include: whether the influence on the atlas collection is large or not is caused by vibration, bubbles, thalli and the like caused by stirring in a fermentation tank; influence of binary mixed liquor preparation; the composition difference with the fermentation liquor is large by taking water as background. Comparing the offline (fig. 3) and online (fig. 4) maps, the offline map is smoother, the characteristic peak is obvious, and the quantitative analysis is easy to be performed, while the online collected map is influenced by the complex situation in the tank, and the noise is larger, so that the quantitative accuracy is influenced, which indicates that the difficulty of online prediction is high.

After repeated research and experiments, the inventor determines that the difference of physicochemical properties of the fermentation liquor and the binary mixed standard sample is the main reason for the larger error. Therefore, the inventors reformulated the standards and rebuild the model.

Example 3 PLS model improvement (full factor design, control of acquisition temperature, and change of acquisition background)

In this example, a design of experiment (DOE) was performed on a mixed standard based on the experience of the model in example 2. In the fermentation process of producing sodium gluconate by Aspergillus niger, the concentrations of glucose and sodium gluconate are generally 1-400g/L, so that the concentrations of the standard solution of glucose and sodium gluconate are designed in seven gradients of 1-400g/L, namely 1g/L, 10g/L, 50g/L, 100g/L, 200g/L, 300g/L and 400 g/L. 49 correction sets of glucose and sodium gluconate binary-medium solutions were obtained by full factor design (full factor design), and 12 verification sets were obtained, as shown in fig. 5.

The standard mixed sample is added with inorganic salt component KH in the culture medium in addition to glucose and sodium gluconate₂PO₄0.17g/L，(NH4)₂HPO₄0.25g/L，MgSO₄0.2g/L, the inorganic salts may influence the quantitative problem, so the fermentation liquor environment is simulated as much as possible, and influencing factors are eliminated. When the standard sample atlas is collected, the fermentation medium without glucose is taken as the background, and the collection temperature is ensured to be 38 ℃.

To remove glucose from the fermentation medium (formulation: 2g/L, MgSO corn steep liquor)₄·7H₂O 0.2g/L、KH₂PO₄0.17g/L、(NH₄)₂HPO₄0.25g/L) is taken as the background of the standard mixed sample, and a standard mixed sample map at the constant temperature of 38 ℃ is collected, as shown in figure 6. The standard mixed sample atlas shows that the absorbance values of all peaks are in gradient distribution, the distribution is uniform, and the standard sample proportion and collection are good. It is then imported into the TQ Analyst 9.0 software by changing modeling parameters such as: a modeling method, a preprocessing method, a modeling area, and the like, as shown in table 3, the model with the highest performance index is obtained.

Table 3, performance index and correlation coefficient of DOE model under each parameter

It can be seen from the above table that 3223, namely the modeling area, is three characteristic peaks, the first derivative, Savitzky-Golayfilter, and the preprocessing method for curve baseline correction have the highest performance index of 98.4, glucose RMSEP of 1.73, sodium gluconate RMSEP of 1.69, which are the lowest, and the similarity coefficients of the two are 1.0000 and 0.9999, so the parameter is selected for modeling.

Based on the model, the concentrations of glucose and sodium gluconate in the fermentation broth were predicted offline, and the results are shown in table 4.

TABLE 4 DOE model prediction of glucose and sodium gluconate concentrations in fermentation broth

From the above table data, it can be seen that: the error percentage of the low concentration is large, and the high concentration is relatively small. The percent mean error for glucose was 5.93% and the root mean square error for prediction, RMSEP, was 4.95 g/L. The average error percentage of the sodium gluconate is 5.13 percent, and the RMSEP is 4.60 g/L. Slightly better than the model without DOE. However, because the online acquired map is influenced by complex factors in the fermentation environment, the noise is large, and the difference from the offline acquired correction set map is obvious, and further improvement is needed in comparison with the just-added online map and offline map.

Example 4 optimization of the improved PLS model

Because the improved PLS model is composed of the mid-infrared spectrum of the offline glucose-sodium gluconate binary mixed solution, certain errors still exist although the PLS model is subjected to experimental design-full factor design. Moreover, the fermentation tank has a complex environment, and factors such as stirring and bubbles can influence the collection of the atlas. In the fermentation process, 1 molecule of glucose is converted into 1 molecule of sodium gluconate, so the concentration relation of the two is almost linear relation and is used for modeling without dispersity. Therefore, on the basis of the original model, the fermentation liquor atlas collected on line is added into the standard mixed sample atlas with the full factor design of the dispersibility (during modeling, the atlas required by modeling and the corresponding standard values of the concentrations of glucose and sodium gluconate need to be introduced, and the standard values are measured by a chemical method off line to form a prediction model. FIG. 7 is a graph showing the concentration distribution after supplementing the spectrum of the fermentation broth collected on-line. The diamond markers are the concentration distributions of the modeling set, the square markers are the validation set, and both are the components of the original model. The triangular mark is the concentration of the newly supplemented fermentation broth spectrum.

Taking a culture medium except glucose as a background, and collecting a map at a constant temperature of 38 ℃. It is then imported into the TQAnalyst 9.0 software by changing modeling parameters such as: a modeling method, a modeling area, a preprocessing method, and the like, as shown in table 5, the model with the highest performance index is obtained.

TABLE 5 Performance index and correlation coefficient of DOE model under each parameter

As seen from table 5, the difference between the performance coefficients of the groups is not obvious, the highest performance coefficient 95.2 corresponds to three groups of parameters, glucose RMSEP of 3221 and 3223 is better, and sodium gluconate RMSEP of 3223 is better, so 3223 is selected as the best model, that is, the modeling region is three characteristic peaks, the performance index of the preprocessing method of first-order derivative, Savitzky-Golay filter and linear baseline correction is the highest, 95.2, glucose RMSEP is 4.03, sodium gluconate RMSEP is 5.44, and the similarity coefficients of the two are 0.9997 and 0.9991. Fig. 8 shows the difference between the predicted and actual values for glucose and sodium gluconate, where the data points are around y = x, demonstrating a similarity coefficient of 0.999.

Based on the model, the concentrations of glucose and sodium gluconate in the fermentation broth were predicted offline, and the results are shown in table 6.

TABLE 6 optimization model prediction of glucose and sodium gluconate concentrations in fermentation broth

From the data in table 6, it can be seen that: the error percentage of the low concentration is large, and the high concentration is relatively small. The percent mean error for glucose was 5.64% and the root mean square error for prediction, RMSEP, was 4.64 g/L. The concentration is slightly better than that of the original model, namely 5.93 percent and 4.95 g/L. The average error percentage of the sodium gluconate is 4.65 percent, the RMSEP is 4.12g/L, and the error percentage is better than that of the original model, namely 5.13 percent and 4.60 g/L. However, the fermentation liquid atlas acquired offline is different from the model atlas adapted to online acquisition, so that the online acquired atlas is predicted by using each model below, and whether the improvement and optimization of the model are effective or not is verified, as shown in table 7.

TABLE 7 comparison of predicted values for each model

As is evident from Table 7, the optimized model added with the online acquired map has more accurate prediction capability than the original two models. Before correction, glucose and sodium gluconate RMSEP predicted by an optimization model are about 10g/L, the error is 5-9%, and the original model is far larger than the values. After correction, the glucose and sodium gluconate RMSEP predicted by the optimization model is within 5g/L, and the error is about 2%. It can be seen that the percent error in glucose is gradually improved from 2.74% to 2.27% to 2.16%. The error percentage of the sodium gluconate is from 4.14% to 3.14% and then to 2.08%, and the improvement of the prediction capability is very obvious. Probably because the online collected atlas has more noise peaks at the peak with more contribution of sodium gluconate, the optimization of the model reduces the influence of the noise peaks on the quantification. This shows that the existing optimized model has higher performance, and can more accurately perform on-line monitoring, timely react and guide the Aspergillus niger fermentation process.

Example 5 on-line monitoring of fermentation Process

After a plurality of problems of background acquisition, model optimization and the like are overcome, the on-line monitoring of the fermentation process of the aspergillus niger by the intermediate infrared spectrometer is realized, and a schematic diagram is shown in fig. 9. The ATR optical fiber probe is inserted into the reactor through an adaptive interface after being sterilized, the spectrum of fermentation liquor in the tank can be acquired on line by the intermediate infrared spectrometer and transmitted to the computer, the Result 3.0 software processes the spectrum, the concentrations of target glucose and sodium gluconate are predicted, the target glucose and sodium gluconate are transmitted to the bios software, and the biostar software is combined with other online data to reflect and guide the fermentation process in time.

In the process of producing sodium gluconate by aspergillus niger fermentation, aspergillus niger consumes glucose and converts the glucose into gluconic acid, and the gluconic acid reacts with NaOH to generate the sodium gluconate. The mid-infrared spectrometer can acquire the infrared spectrum of the fermentation liquid in the fermentation tank in real time, as shown in fig. 10, and realize the on-line monitoring of the concentration changes of the substrate glucose and the product sodium gluconate in the fermentation process, as shown in fig. 11. Compared with the infrared map of the modeling, the infrared map of the fermentation liquid collected on line is very close to the peak position and the peak type, and the characteristic peak is obvious. The characteristic peak-to-peak value also changes regularly with the passage of time. 1505 and 1650cm^-1The single intensity peak in the region (X axis) is a peak specific to sodium gluconate, and the absorbance (Y axis) gradually increases with increasing fermentation time (Z axis), indicating that the concentration of sodium gluconate gradually increases. 1177-1505cm^-1The change of the two peaks is not obvious and slowly increases, and the sodium gluconate contributes to the peaks greatly because the concentration of the sodium gluconate gradually increases and the concentration of glucose gradually decreases. 950-^-1The absorbance of the two peaks of the region gradually decreases with time, i.e., the contribution of glucose to the region is greater.

At the beginning of the fermentation, glucose is consumed at a smaller rate. After about 6h of fermentation, the rotation speed increased, DO increased, and the increase of oxygen increased the reaction rate, and further the OUR increased, and the consumption of glucose increased, and thus the slope of the curve increased, as shown in FIG. 11A. And fermenting for about 18h, wherein the glucose concentration is close to zero and does not decrease any more, and the fermentation is finished. Comparing the glucose concentration measured by off-line sampling with the concentration predicted on-line, fitting a curve, and obtaining a similarity coefficient R²0.9988, as shown in FIG. 11B. The sodium gluconate is prepared from glucoseThe conversion is related to glucose, the synthesis is carried out at a low speed in the early fermentation stage, the difference between the offline measurement value and the online detection value is large after 3.5h, and the offline measurement is possibly abnormal according to the reflection trend. After about 6 hours of fermentation, the reaction rate is increased due to the increase of the rotation speed, and further the rate of synthesizing sodium gluconate is increased, the fermentation lasts for about 18 hours, the concentration of sodium is not increased any more, and the fermentation is finished, corresponding to the change of glucose, as shown in fig. 11C. The concentration of the sodium gluconate sampled and measured off-line is compared with the concentration detected on-line, and the similarity coefficient R of the fitting curve²0.9971, as shown in FIG. 11D.

The RMSEP value of the glucose predicted in real time is 3.24g/L, the relative error is 2.21 percent, the RMSEP value of the sodium gluconate is 4.03g/L, and the relative error is 2.77 percent. Because the online monitoring is influenced by multiple factors such as air bubbles, stirring, vibration and the like, a certain error exists between an offline measured value and an online measured value, and the predicted concentration range is 1-400g/L, the RMSEP value is slightly higher. After simple fitting correction, the online monitoring value is very close to the offline measurement value, the concentration change condition of the target object can be correctly reflected, and the reliability of online monitoring of the intermediate infrared spectrometer is proved.

Example 6 mid-infrared Spectrum guidance for optimization of Aspergillus niger fermentation Process 1

In the reaction of producing sodium gluconate from Aspergillus niger, oxygen is necessary as one of the reactants. The inventors have found that in the middle of the reaction, the dissolved oxygen is usually close to 0 and the lack of oxygen results in a limited reaction. Therefore, an experiment is designed, pure oxygen with gradient concentration of 5% (v/v), 7.5% and 10% is introduced, and the sugar consumption rate and the product synthesis rate are observed to judge the limiting factors of the reaction, so that the fermentation process is optimized. The 5L fermenter culture method was as described above.

The addition of the intermediate infrared spectrometer can reflect the rapidity of real-time monitoring. Although the mass flow meter is used to control the oxygen supply amount, the oxygen supply amount is sometimes unstable due to unstable cylinder pressure, and oxygen is a dangerous combustion improver, so that the problem and the consequence are not obvious. The dissolved oxygen is still in a state close to zero, and the condition of introducing oxygen cannot be reflected, so that the fermentation process is influenced. The on-line mid-infrared spectroscopy can timely detect the contents of glucose and sodium gluconate, calculate the sugar consumption rate rs and the product synthesis rate rp, and can timely reflect the fermentation condition according to the parameters, thereby effectively guiding and optimizing the fermentation process.

Although the initial sugar concentration of each group (10% pure oxygen group; 7.5% pure oxygen group; 5% pure oxygen group)) was slightly different (due to the inconsistent degree of sugar consumption in the seed tank), the slope still reflects the reaction condition of each group, as shown in FIG. 12. At the initial stage of fermentation, the sugar consumption rate of each group is low, but the slope of the experimental group with 10% pure oxygen is obviously higher than that of other groups, and the sugar consumption rate of other groups is relatively close. After fermentation for 6h, the rotating speed is increased, so that the dissolved oxygen is slightly increased, the slope is increased, and the reaction rate is accelerated. In the later fermentation stage, the slope of the sugar concentration of the control group is obviously smaller than that of each experimental group, and the sugar consumption rate is smaller than that of the experimental group, so that the introduction of oxygen is proved, and the reaction is accelerated to a certain extent. In the three gradient oxygen experimental groups and the 10% pure oxygen experimental group, the oxygen is introduced most, the slope is the largest, and the reaction rate is the highest. The experimental group of 7.5% pure oxygen was second, and the experimental group of 5% pure oxygen was relatively slowest. It is shown that the more oxygen is introduced in a certain range, the faster the reaction rate is, and the enzyme is not limited.

The initial Sodium Gluconate (SG) concentrations varied for each set of seed tanks, but the slopes still reflected the product synthesis for each set, as shown in fig. 13. At the initial stage of fermentation, the synthesis rate of sodium gluconate of each group is low, and the experimental synthesis rate is similar to that of the control group and is slightly higher than that of the control group. After fermentation for 6h, the rotating speed is increased, the dissolved oxygen is increased, the slope is increased, and the reaction rate is accelerated. The oxygen of the experimental group with 10% pure oxygen is introduced most, the slope is obviously higher than that of each group, the SG synthesis rate is fastest, and the experimental group with 7.5% pure oxygen is inferior. The slope difference between the 5% experimental group and the control group is small and is not obvious in the figure. However, it can be still shown that the more oxygen is introduced in a certain range, the faster the reaction rate is, and the enzyme is not limited. Consistent with the consumption of glucose.

TABLE 8 comparison of prediction error under different ventilation conditions

The predicted conditions of glucose and sodium gluconate are shown in Table 8, wherein the RMSEP values of glucose and sodium gluconate are both between 2 and 4g/L, and the concentration deviation is slightly larger due to the larger predicted concentration range. And the relative errors are all between 1 and 3 percent, R²The concentration of residual sugar and products in the fermentation tank can be accurately reflected in real time at about 0.998. Furthermore, the medium infrared can still accurately reflect the change of the concentration of glucose and sodium gluconate under the condition of introducing oxygen, is hardly influenced by the oxygen, and can reflect the fermentation condition in time to guide the fermentation process.

TABLE 9 comparison of fermentation in different batches

Batches of	Unit of	Control group	5% pure oxygen	7.5% pure oxygen	10% pure oxygen
						Fermentation period h	h		18	17.5	17	15
Average rate of sugar consumption	g/ L·h	16.50	18.85	19.29	20.53
						On-line monitoring of sugar consumption rate	g/ L·h	16.20	18.78	18.94	20.29
Average product Synthesis Rate	g/ L·h	15.62	16.79	17.24	19.59
						On-line monitoring of product synthesis rate	g/ L·h	15.45	16.81	17.01	19.37
Yield of the product	mol/mol	0.86	0.84	0.86	0.91
						On-line monitoring yield	mol/mol	0.86	0.84	0.86	0.91
Concentration of fungus	g/L	1.33	1.68	1.89	2.05

In combination with the growth and reaction conditions of several batches of fermentations (Table 9), it can be seen that, although the concentration (average concentration during fermentation) is different, the reaction rate is influenced by the oxygen of the reactant, and the amount of enzyme is not limited enough, so that the concentration is not a factor influencing the reaction rate. With the increase of the oxygen input, the fermentation period is gradually shortened, the sugar consumption rate rs and the product synthesis rate rp are also increased, and the yield is not lost.

The sugar consumption rate, the product synthesis rate and the yield of the mid-infrared measurement are close to reference values, the fermentation conditions can be reflected in time in the reaction, and the rapidness, timeliness and accuracy are realized. After the fermentation is finished, the sugar consumption rate rs and the product synthesis rate rp can be highest under the condition of 10 percent oxygen introduction through online data judgment, and the yield is not influenced. Under the condition that the ventilation condition is changed, the on-line mid-infrared prediction capability is almost not influenced, and the concentrations of glucose and sodium gluconate in the fermentation liquor can be accurately predicted. Oxygen is a dangerous combustion-supporting product, can not be supplemented too much, is easy to cause dangers such as explosion and the like, and tail oxygen is close to 28 percent and quickly exceeds the limit of accurate measurement of tail gas mass spectrum monitoring, namely, the oxygen uptake rate OTR is limited, the supplemented oxygen can not be dissolved in fermentation liquor in time to participate in reaction, but is directly discharged out of a fermentation tank, and waste is also caused. Therefore, the pure oxygen introduction amount of 10% is the optimal oxygen supply amount at this time.

Example 7 mid-infrared Spectrum guidance optimization of Aspergillus niger fermentation Process 2

The present inventors have also discovered that different initial sugar (glucose) concentrations in the seed tank may have an effect on the reaction using mid-infrared spectroscopy directed on-line detection methods. Therefore, the present example designs fermentation with different initial sugar concentrations of 200, 300, 400, 500 g/L in the seed culture medium, and selects the optimum initial sugar concentration of the seed tank to optimize the fermentation process by observing the conditions of sugar consumption rate and product synthesis rate in real time by using MIR online monitoring technology.

The fermentation process of producing sodium gluconate by aspergillus niger is mainly a primary metabolism process, has relatively short period and high reaction rate, and has the characteristics of high efficiency, high speed, high permeability and high oxygen consumption. Because the concentration of residual sugar in the lag phase is not changed greatly and is easy to be infected by bacteria, MIR is used for online detection from the production phase. Fig. 14 clearly reflects that the concentration of glucose and sodium gluconate changes with fermentation time on-line monitored by mid-infrared, and the slope of the curve can reflect the consumption rate of glucose and the synthesis rate of sodium gluconate. The fermentation time is 0 h, the rotating speed of each batch is set to be 800 rpm, and after the fermentation time is 4h, the slope of the batch curve of 300g/L of initial sugar is obviously greater than that of other batches, which shows that under the initial sugar concentration of 300g/L of Aspergillus niger, the glucose consumption rate and the sodium gluconate synthesis rate are fastest, namely the reaction rate is fastest. When the initial sugar concentration is 500 g/L, the osmotic pressure is higher, so that the lag phase of the strain is increased, the strain cannot react and synthesize the sodium gluconate immediately, and the strain is fermented for about 9 hours to begin to consume the glucose to synthesize the sodium gluconate. The batch rate of 200g/L of initial sugar was relatively lowest. The change conditions of substrates and products can be clearly, quickly and real-timely monitored by monitoring the fermentation of the aspergillus niger through mid-infrared on line, so that the fermentation process can be better known, and abnormal fermentation phenomena can be timely fed back. As a result, the initial sugar concentration was 300g/L, and the sugar consumption rate and the product synthesis rate were the fastest during the production period.

Under the condition of different initial sugar concentrations, except for the batches with 500 g/L initial sugar, the Root Mean Square (RMS) errors of the glucose and the sodium gluconate in different batches(RMSEP) is between 2 and 4g/L, the error is about 3 percent, R²Close to 0.999 as shown in table 10. The Robustness (Robustness, meaning stability under the influence of external factors) of the model established by the method is better, and the concentrations of residual sugar and products in the fermentation tank can be accurately reflected in real time.

TABLE 10 comparison of prediction error for each batch

The fermentation process of producing sodium gluconate by aspergillus niger is carried out under the condition of high sugar. As seen in Table 11, the in vitro culture conditions of each batch of the initial sugars were consistent, the yields and the concentrations of the bacteria were not very different, and the sugar consumption rate and the product synthesis rate of 300g/L of the initial sugars were much higher than those of the other initial sugars. Consistent with the fermentation situation monitored on-line in fig. 14. Therefore, the initial sugar concentration is too high or too low, which inhibits the sodium gluconate-producing reaction of the strain. The calculated glucose consumption rate, product synthesis rate and yield are slightly different from the reference value of off-line measurement by utilizing the glucose and sodium gluconate concentrations predicted on line by the mid-infrared spectrometer, and the reference values except for the batch of 500 g/L of initial sugar possibly exceed the range predicted by the model; on the other hand, it is also suggested that the metabolic properties of cells are affected by a higher osmotic concentration. The method explains that the middle infrared on-line monitoring of the Aspergillus niger fermentation process can accurately predict the concentrations of glucose and sodium gluconate, accurately calculate the sugar consumption rate, the product synthesis rate and the yield, and better guide and optimize the fermentation process. By integrating the offline measured value and the on-line monitoring data of the mid-infrared, the initial sugar concentration of 300g/L is the optimal initial sugar condition for the growth period of Aspergillus niger, and the product synthesis rate at the moment is 20.8 g/L.h.

TABLE 11 comparison of fermentation conditions of the respective batches

3. Discussion of the related Art

The invention applies FT-MIR-ATR combinationThe small two-step multiplication successfully realizes the on-line monitoring of the fermentation process of the aspergillus niger. The concentration of glucose and sodium gluconate is 1650cm at 950-^-1The area has obvious infrared absorption peak, a standard solution prediction model is established by a PLS method, the glucose RMSEC and RMSEV of the model are 0.96 and 1.71g/L, and the similarity coefficient R²1.000 and 0.999 respectively, the RMSEC and RMSEV of the sodium gluconate are 2.67 and 1.59g/L, R²The prediction performance is better when the difference is 0.999 and 0.999 respectively. And adding an online map optimization model into the standard liquid map model due to the large difference of physicochemical properties of the standard liquid and the fermentation liquid. The optimized model predicts the fermentation liquor atlas collected on line, the glucose RMSEP is 3.68g/L, the relative error is 2.16%, the sodium gluconate RMSEP is 4.58g/L, and the relative error is 2.08%, so that the model is better than the original model and is more suitable for the on-line monitoring process. MIR simultaneously detects the concentration of substrate glucose and sodium gluconate, the RMSEP of the glucose is 3.24g/L, and the relative error is 2.21%. The RMSEP value of the sodium gluconate is 4.03g/L, the relative error is 2.77%, the concentration change conditions of a substrate and a product can be rapidly and timely predicted, the fermentation condition of Aspergillus niger is effectively monitored, and the metabolic regulation and control of the synthesis of the sodium gluconate are guided. Under different aeration conditions and different initial sugar concentrations, the fermentation process can still be accurately presented, and the optimal fermentation state at the stage can be optimized.

All documents referred to herein are incorporated by reference into this application as if each were individually incorporated by reference. Furthermore, it should be understood that various changes and modifications of the present invention can be made by those skilled in the art after reading the above teachings of the present invention, and these equivalents also fall within the scope of the present invention as defined by the appended claims.

Claims

1. A method for monitoring the fermentation process of producing sodium gluconate by Aspergillus niger fermentation on line is characterized by comprising the following steps: collecting fermentation liquor on line, obtaining a fermentation liquor map by using a mid-infrared spectrometer, and analyzing the contents of glucose and sodium gluconate in the fermentation liquor; the method comprises the steps of (1) introducing an online collected fermentation liquor map into an optimally designed PLS model, and analyzing the contents of glucose and sodium gluconate in the fermentation liquor; wherein, the PLS model of the optimized design is established as follows:

(1) designing standard solutions of glucose and sodium gluconate in a gradient manner, and obtaining a correction set glucose and sodium gluconate binary-culture medium standard solution and a verification set glucose and sodium gluconate binary-culture medium standard solution through full factor design; importing the maps of the correction set and the verification set into spectral analysis software, and obtaining a model with relatively highest performance index by changing modeling parameters; in the standard solution, a culture medium for removing glucose is used as a background;

(2) on the basis of the model in the step (1), adding the online collected fermentation liquor atlas in the correction set to optimize the model, and obtaining an optimally designed PLS model; the method comprises the following steps of (1) detecting fermentation liquor acquired on line by using a chemical method to obtain content data of glucose and sodium gluconate in a fermentation liquor sample; correlating the acquired map with the detection data of the chemical method, and analyzing the model by using a multiple regression analysis method so as to obtain a model with relatively highest performance index;

wherein the performance index is determined according to the Root Mean Square (RMSEP) of the prediction error.

2. The method as claimed in claim 1, wherein the spectral analysis software is TQ Analyst software.

3. The method of claim 1, wherein the glucose-depleted medium comprises: corn steep liquor 2 +/-0.5 g/L, MgSO₄·7H₂O 0.2±0.05g/L、KH₂PO₄0.17±0.05g/L、(NH₄)₂HPO₄0.25±0.05g/L。

4. The method of claim 1, wherein the fermentation broth is collected at 38 ± 0.5 ℃ to obtain a fermentation broth map.