L-threonine fermentation post-extraction and purification method based on temperature gradient control

Abstract

The present application belongs to the technical field of separation engineering, and provides a L-threonine fermentation post-extraction purification method based on temperature gradient control, which comprises the following steps: collecting turbidity and particle count data under different cooling rates, stirring intensities and impurity concentrations, establishing a metastable zone model and outputting a sensitivity ranking; after confirming that the cooling rate is the highest influencing parameter, respectively calibrating the fine crystal storm critical rate and the phantom seed critical rate, and taking the smaller value of the two as the safety upper limit; combining the metastable zone model to back-calculate the maximum cooling rate allowed in each temperature interval, and outputting a segmented rate control curve; and dynamically adjusting the cooling rate according to the deviation of the real-time supersaturation and the critical value; the present application solves the problems of fine crystal storm and phantom seed through accurate control of the cooling rate, and improves the particle size uniformity, purity and yield of the L-threonine crystalline product.

Description

Technical Field

[0001] This invention belongs to the field of separation engineering technology, specifically a method for the extraction and purification of L-threonine after fermentation based on temperature gradient control. Background Technology

[0002] L-Threonine is primarily produced industrially through microbial fermentation. The general extraction and purification process after fermentation is as follows: fermentation broth → cell separation (membrane filtration or centrifugation) → decolorization → ion exchange → concentration → crystallization → separation → drying → finished product. Among these, the crystallization process is the core step that determines the product's particle size, purity, and yield.

[0003] Currently, cooling crystallization is the mainstream process for the extraction and purification of L-threonine. This process is based on the characteristic that the solubility of L-threonine decreases with decreasing temperature. By controlling the cooling process, the solution is brought to a supersaturated state, promoting the orderly precipitation of the solute. To achieve controllable crystallization, the supersaturation needs to be controlled within the metastable region—that is, the metastable region between the solubility curve and the supersolubility curve. Ideally, the solution should slowly pass through the metastable region, achieving controlled growth on the existing crystal surface.

[0004] However, the metastable region width of L-threonine solution is not a fixed value, but is significantly affected by various operating parameters such as cooling rate, stirring intensity, and the presence of impurities. When the cooling rate is too fast, the supersaturation of the solution accumulates rapidly in a short period of time, quickly crossing the metastable region and exceeding the supersolubility curve, thus triggering two closely related serious problems: First, the system experiences violent and uncontrollable explosive nucleation, instantly generating a large number of fine crystals, forming a so-called "fine crystal storm," leading to difficulties in subsequent filtration and product loss; Second, excessively rapid cooling creates localized supercooled zones, forming "freezing fog" metastable particles near the seed crystal addition site. These particles are completely lost during subsequent mother liquor discharge or washing, producing a "phantom seed" effect, causing the seed crystal induction function to fail and the seed material to be wasted. The combination of these two problems often leads to uncontrolled crystallization, resulting in fine product particle size, low purity, and decreased yield, becoming a prominent technical bottleneck restricting the high-quality production of L-threonine.

[0005] Therefore, this invention provides a method for the extraction and purification of L-threonine after fermentation based on temperature gradient control. Summary of the Invention

[0006] In order to overcome the shortcomings of the prior art, at least one technical problem raised in the background art is solved.

[0007] The technical solution adopted by this invention to solve its technical problem is: 1. A method for extracting and purifying L-threonine after fermentation based on temperature gradient control, characterized by comprising the following steps:

[0008] Step S10: Collect time series data of solution turbidity and particle count under different cooling rates, stirring intensities, and impurity concentrations; establish a metastable region model with varying metastable region boundaries as a function of each parameter; output the sensitivity ranking of each parameter to critical supersaturation; and determine whether the cooling rate is the parameter with the highest influence.

[0009] Step S20: If so, fix the stirring intensity and impurity concentration, conduct a gradient experiment on the cooling rate, simultaneously collect particle chord length distribution data, identify the abrupt change point where the total number of particles shows an exponential increase, and output the critical rate of the fine crystal storm.

[0010] Step S30: Subdivide the cooling gradient within the range below the critical rate of the fine crystal storm, collect data on the change in submicron particle count in the local area before and after the addition of the seed crystal, identify the critical point at which the microcrystals rapidly decay to the baseline level, and output the critical rate of the phantom seed crystal.

[0011] Step S40: Take the smaller critical rate between the fine crystal storm and the phantom seed as the safety upper limit, combine it with the metastable region model, back-calculate the maximum allowable cooling rate for each temperature range, and output the segmented rate control curve.

[0012] Step S50: Based on the segmented rate control curve, dynamically adjust the cooling rate according to the deviation between the real-time supersaturation and the critical value.

[0013] As a further aspect of the present invention: the process of establishing a metastable region model in step S10, which shows the boundary of the metastable region as varying with each parameter, is as follows:

[0014] The critical supersaturation measured experimentally is used as the output, and the corresponding cooling rate, stirring speed, and impurity concentration are used as the input to form multiple rows of data. After standardization, a model is established using a multiple regression method. The model includes linear terms, square terms, and interaction terms between parameters for each parameter.

[0015] The model coefficients are fitted using the least squares method. The significance value of each coefficient is calculated, and terms with significance values ​​greater than 0.05 are removed. After each term is removed, the remaining coefficients are refitted until the significance values ​​of all remaining terms are less than 0.05, thus obtaining the initial metastable region model.

[0016] The reliability of the model is verified by cross-validation. The data is randomly divided into multiple parts, and one part is taken in turn as the validation set, while the rest are used as the training set. The mean absolute deviation between the predicted value and the measured value is calculated. If the mean absolute deviation is less than a preset threshold, the model is verified and the metastable region model is finally obtained.

[0017] As a further aspect of the present invention: the process of sorting the sensitivity of each parameter to the critical supersaturation in step S10 is as follows:

[0018] The standardized regression coefficients for each parameter are obtained from the metastable region model. The absolute values ​​of the standardized regression coefficients are arranged in descending order, and the sorting result is the sensitivity ranking of each parameter to the critical supersaturation. In the sensitivity ranking, if the cooling rate ranks first, then the cooling rate is determined to be the parameter with the highest influence.

[0019] As a further aspect of the present invention: the process of identifying the abrupt change point where the total number of particles exhibits exponential growth and outputting the critical rate of the fine-grained storm in step S20 is as follows:

[0020] Plot the total number of particles at each parameter level with time on the x-axis and the common logarithmic value of the total number of particles on the y-axis. Calculate the first-order difference between adjacent sampling points of the particle logarithmic value to obtain the difference sequence. Take the maximum value of the difference within the first 60 seconds after the start of the experiment as the baseline value. Traverse the difference sequence. If the difference values ​​of three consecutive sampling points reach or exceed ten times the baseline value, mark it as a mutation point. Otherwise, mark it as no mutation point. Arrange the marked results in ascending order of parameter values. Find the parameter interval from no mutation point to mutation point. Take the lower limit of the interval as the critical rate of fine crystal storm.

[0021] As a further aspect of the present invention: the process of identifying the critical point where the microcrystals rapidly decay to the baseline level and outputting the critical rate of the phantom seed crystal in step S30 is as follows:

[0022] For each cooling rate level, a curve was plotted with time on the x-axis and submicron particle count on the y-axis. The average submicron particle count within the first 60 seconds after the start of the experiment was calculated as the baseline value. The first-order difference of particle count over time was calculated, and the difference value was used to mark whether the seed crystal was effective or ineffective.

[0023] The results are sorted and labeled according to the rate from smallest to largest. The rate range from seed effectiveness to seed failure is found, and the lower limit of the range is taken as the critical rate of phantom seed.

[0024] As a further aspect of the present invention: in step S30, the method for marking the effectiveness or ineffectiveness of the seed crystal is as follows:

[0025] If there are three consecutive sampling points with negative difference values ​​and decreasing absolute values, and the particle count subsequently falls back to within ±10% of the baseline value, then the seed crystal marking is valid.

[0026] If the difference between three consecutive sampling points is negative and the absolute value continues to increase, and the particle count drops to less than 50% of the baseline value, then the seed crystal marker fails.

[0027] As a further aspect of the present invention: the process of taking the smaller value between the fine-grained storm critical rate and the phantom seed critical rate as the safety upper limit in step S40 is as follows:

[0028] Compare the values ​​of the fine crystal storm critical rate output in step S20 and the phantom seed critical rate output in step S30, and select the smaller value as the safety upper limit; the safety upper limit is used to calculate the maximum allowable cooling rate for each temperature range in subsequent steps.

[0029] As a further aspect of the present invention: the process of outputting the segmented rate control curve in step S40 is as follows:

[0030] Determine the starting temperature and the ending temperature, and divide the temperature range between the starting temperature and the ending temperature into multiple temperature sub-intervals; for the midpoint temperature of each temperature sub-interval, calculate the saturation concentration at that midpoint temperature using the solubility curve; set a critical supersaturation safety value by substituting the safety upper limit, the fixed stirring speed, and the impurity concentration into the metastable region model.

[0031] For each temperature sub-interval, calculate the maximum allowable supersaturation increment, back-calculate the maximum allowable cooling range based on the definition of supersaturation, calculate the ratio of the maximum cooling range to the target residence time of the temperature sub-interval, and obtain the maximum allowable cooling rate of the sub-interval.

[0032] Plot a piecewise rate control curve with the midpoint temperature of the temperature sub-interval as the x-axis and the maximum allowable cooling rate as the y-axis.

[0033] As a further aspect of the present invention: the process of dynamically adjusting the cooling rate based on the deviation between the real-time supersaturation and the critical value in step S50 is as follows:

[0034] The segmented rate control curve is used as the baseline cooling rate curve; solution data is collected every 10 seconds by an online supersaturation monitoring device, converted into real-time concentration values, and the real-time supersaturation is calculated by combining the solubility curve.

[0035] The reference cooling rate is obtained from the segmented rate control curve based on the current temperature; the deviation between the real-time supersaturation and the critical supersaturation safety value is calculated; if the deviation is greater than 0.05, the cooling rate adjustment is the reference cooling rate multiplied by 0.1; if the deviation is less than or equal to 0.01, the cooling rate adjustment is the reference cooling rate multiplied by -0.2; after setting upper and lower limits for the adjustment, the actual cooling rate is calculated and output to the temperature control system.

[0036] As a further aspect of the present invention: the process of setting upper and lower limits for the cooling rate adjustment in step S50 is as follows:

[0037] Under acceleration conditions with a deviation greater than 0.05, the actual cooling rate shall not exceed 1.2 times the baseline cooling rate; under deceleration conditions with a deviation less than or equal to 0.01, the actual cooling rate shall not be less than 0.02 degrees Celsius per minute; if the calculated actual cooling rate exceeds the upper limit, the upper limit value shall be used; if it is lower than the lower limit, the lower limit value shall be used. The upper and lower limits are used to prevent excessive single adjustment from causing system oscillation or oversaturation out of control.

[0038] The beneficial effects of this invention are as follows: Step S10 establishes a quantitative model of the metastable zone boundary changes with cooling rate, stirring intensity, and impurity concentration, and outputs a sensitivity ranking. After confirming that the cooling rate is the most influential parameter, steps S20 and S30 respectively calibrate the critical rates of fine crystal storm and phantom seed crystals, and use the smaller of the two as the safety upper limit. This fundamentally solves the problem of two failure modes being triggered simultaneously by excessively fast cooling rates in traditional processes. Step S40 combines the safety upper limit with the metastable zone model to deduce the maximum allowable cooling rate for each temperature range, outputting a segmented rate control curve, thus realizing a precise temperature control strategy of "slow first, then fast." Step S50 further establishes a closed-loop feedback control mechanism based on real-time supersaturation monitoring, dynamically adjusting the cooling rate according to the deviation between the measured supersaturation and the critical safety value, ensuring that the entire crystallization process always operates within the safe range of the metastable zone. This invention, through precise control of the cooling rate, simultaneously solves the problems of fine crystal storm and phantom seed crystals, improving the particle size uniformity, purity, and yield of L-threonine crystal products. Attached Figure Description

[0039] The invention will now be further described with reference to the accompanying drawings.

[0040] Figure 1 This is a flowchart of the steps of the L-threonine extraction and purification method based on temperature gradient control according to an embodiment of the present invention;

[0041] Figure 2 This is a logic diagram of the L-threonine extraction and purification method based on temperature gradient control described in the embodiments of the present invention. Detailed Implementation

[0042] To make the technical means, creative features, objectives and effects of this invention easier to understand, the invention will be further described below in conjunction with specific embodiments.

[0043] Example: Please refer to Figure 1-2 As shown in the embodiment of the present invention, the method for extracting and purifying L-threonine after fermentation based on temperature gradient control includes the following steps:

[0044] Step S10: Collect time series data of solution turbidity and particle count under different cooling rates, stirring intensities, and impurity concentrations; establish a metastable region model with varying metastable region boundaries as a function of each parameter; output the sensitivity ranking of each parameter to critical supersaturation; and determine whether the cooling rate is the parameter with the highest influence.

[0045] In step S10, the process of collecting time-series data on solution turbidity and particle count under different cooling rates, stirring intensities, and impurity concentrations is as follows:

[0046] Prepare a saturated L-threonine solution at 50 degrees Celsius, and confirm the initial concentration using the drying method.

[0047] Three variable parameters were set: cooling rate at 5 levels (0.1, 0.2, 0.5, 1.0, 2.0 degrees Celsius per minute), stirring speed at 3 levels (100, 200, 300 rpm), and impurity concentration at 3 levels (0.5, 1.0, 2.0 g per liter), for a total of 45 sets of experiments; in each set of experiments, the solution was first heated to 55 degrees Celsius and held at that temperature for 30 minutes to ensure complete dissolution;

[0048] The temperature was lowered at a set rate while a laser turbidimeter and a focused beam reflectometer were running simultaneously. Both recorded data every 5 seconds. The turbidimeter recorded the turbidity value, and the focused beam reflectometer recorded the total number of particles. When the turbidity value increased three times consecutively and the increase exceeded three times the standard deviation of the baseline noise, it was determined that the supersolubility curve boundary point had been reached, and the temperature and supersaturation were recorded.

[0049] Simultaneously, particle counting is used to verify that the moment when the total number of particles starts from zero should coincide with the moment when the turbidity increases. If the deviation is large, the average value should be taken.

[0050] Each experiment was repeated three times, and the arithmetic mean of the three critical supersaturation measurements was taken as the final critical supersaturation.

[0051] It should be noted that the process for preparing a saturated L-threonine solution can be as follows: Add pure L-threonine to deionized water, heat to 50 degrees Celsius and stir to dissolve. After stirring at a constant temperature for 30 minutes, take a sample and filter it through a 0.45-micron filter membrane. Determine the concentration of the filtrate by the drying method: Take 5 ml of the filtrate in a weighing bottle, dry it at 105 degrees Celsius to constant weight, calculate the ratio of solute mass to solvent mass, and obtain the solubility value at that temperature; repeat three times and take the average value as the initial concentration.

[0052] In step S10, the process of establishing a metastable region model with varying parameters for the metastable region boundary is as follows:

[0053] The critical supersaturation measured in 45 sets of experiments is used as the output, and the corresponding cooling rate, stirring speed, and impurity concentration are used as the input to form 45 rows of data, each row containing one output value and three input values.

[0054] After standardizing the input data, a model was built using the multiple regression method. The model includes the linear term, square term, and interaction term between each parameter. The 45 rows of data were substituted into the model, and the least squares method was used to fit the model coefficients.

[0055] Calculate the significance value p for each coefficient, and check the p value for each coefficient one by one: for terms with a p value greater than 0.05, remove them from the model. After removing each term, refit the remaining coefficients and check the p value of each coefficient again until the p value of all remaining terms is less than 0.05. The terms and corresponding coefficients that are finally retained constitute the initial metastable region model.

[0056] The 45 sets of data were randomly divided into five parts, with 9 sets of data in each part. One part was used as the validation set in turn, and the other four parts (36 sets) were used as the training set to fit the model. After fitting the model with the training set each time, the model was used to predict the output value of the 9 sets of data in the validation set. The average absolute deviation between the predicted value and the measured value was calculated in each validation. If the average absolute deviation is less than 0.05, the model is validated and finally the metastable region model with the boundary of the metastable region changing with each parameter is obtained.

[0057] In step S10, the process of ranking the sensitivity of each parameter to the critical supersaturation is as follows:

[0058] The standardized regression coefficients for each parameter are obtained from the metastable region model. The absolute values ​​of the standardized regression coefficients are arranged in descending order, and the sorting result is the ranking of the sensitivity of each parameter to the critical supersaturation.

[0059] In the sensitivity ranking, if the cooling rate is ranked first, it means that the cooling rate is the parameter with the greatest influence; otherwise, it is not.

[0060] It should be noted that if the cooling rate is not the most influential parameter, it means that the width of the metastable region is affected by the impurity concentration and stirring speed. Parameter fluctuation compensation can be achieved by adopting a multi-parameter joint control strategy or establishing a closed-loop control system based on real-time supersaturation feedback.

[0061] Understandably, the significance of step S10 lies in: by collecting turbidity and particle count data under different cooling rates, stirring intensities, and impurity concentrations, establishing a metastable model of the metastable zone boundary as a function of various parameters, and outputting a ranking of the sensitivity of each parameter to the critical supersaturation; the core is to quantitatively identify the dominant factors affecting the width of the metastable zone and determine whether the cooling rate is the most influential parameter; if the cooling rate is the dominant factor, then subsequent optimization through temperature control is feasible; this provides a data foundation and decision-making basis for the optimization of the entire crystallization process, avoiding inefficiency or control failure caused by blind temperature control;

[0062] Step S20: If so, fix the stirring intensity and impurity concentration, conduct a gradient experiment on the cooling rate, simultaneously collect particle chord length distribution data, identify the abrupt change point where the total number of particles shows an exponential increase, and output the critical rate of the fine crystal storm.

[0063] In step S20, the process of fixing the stirring intensity and impurity concentration, conducting a gradient experiment on the cooling rate, and simultaneously collecting particle chord length distribution data is as follows:

[0064] The stirring speed and impurity concentration were fixed. The stirring speed was selected from the least affected level in the sensitivity ranking (e.g., 200 rpm), and the impurity concentration was selected from a typical value in actual production (e.g., 1.0 g per liter).

[0065] Multiple cooling rate gradients are selected: within the experimental range of step S10, at least 8 rate levels are set from low to high, such as 0.05, 0.08, 0.10, 0.12, 0.15, 0.20, 0.25, and 0.30 degrees Celsius per minute, covering the entire range from safe to potentially triggering a fine crystal storm;

[0066] It should be noted that before each experiment, a saturated L-threonine solution was prepared according to the method in step S10, and heated to 55 degrees Celsius and kept at a constant temperature for 30 minutes to ensure complete dissolution.

[0067] Cool down at the set cooling rate, and at the same time run the focused beam reflection measuring instrument to record the particle chord length distribution data (total number of particles) every 5 seconds.

[0068] In step S20, the process of identifying the abrupt change point where the total number of particles shows an exponential increase and outputting the critical rate of the fine-grained storm is as follows:

[0069] Plot the total number of particles curves at each parameter level with time as the x-axis and the common logarithm of the total number of particles as the y-axis.

[0070] Based on the total number of particles curve, the first difference between the sampling points of adjacent 5-second intervals of the particle pair values ​​is calculated to obtain the difference sequence. The maximum value of the difference within the first 60 seconds after the start of the experiment is taken as the baseline value.

[0071] Traverse the difference sequence. If there are three consecutive sampling points whose difference values ​​all reach or exceed 10 times the baseline value, mark them as having a mutation point; otherwise, mark them as having no mutation point.

[0072] The results are sorted and marked according to the parameter values ​​from smallest to largest. The parameter interval from no mutation point to mutation point is found. If the difference between the upper and lower limits of the interval is greater than the minimum parameter interval, a new level is inserted at the midpoint of the interval and the above process is repeated until the interval difference is no greater than the minimum parameter interval. At this time, the lower limit of the interval is taken as the critical value, that is, the critical rate of fine crystal storm.

[0073] It should be noted that the minimum parameter spacing refers to the minimum difference between two adjacent parameter levels set during the data acquisition process in step S20. For example, if the parameter levels are set to 0.10, 0.15, 0.20, 0.25, and 0.30, the spacing between adjacent levels is 0.05, and the minimum parameter spacing is 0.05.

[0074] Understandably, the significance of step S20 lies in the following: Under the premise that the cooling rate is the most influential parameter, with fixed stirring intensity and impurity concentration, the total number of particles is monitored in real time through a gradient cooling experiment combined with a focused beam reflectance meter. This identifies the point of abrupt change where the total number of particles increases exponentially, and outputs the critical rate of the fine crystal storm. The core is to quantify and calibrate the cooling rate threshold that triggers explosive nucleation, providing a safety boundary for subsequent process design. Once this critical rate is clearly defined, the excessive generation of fine crystals due to excessively rapid cooling in actual production can be avoided, thereby reducing problems such as filtration difficulties, product loss, and decreased yield.

[0075] Step S30: Subdivide the cooling gradient within the range below the critical rate of the fine crystal storm, collect data on the change in submicron particle count in the local area before and after the addition of the seed crystal, identify the critical point at which the microcrystals rapidly decay to the baseline level, and output the critical rate of the phantom seed crystal.

[0076] In step S30, the process of subdividing the cooling gradient within a range below the critical rate of the fine-grained storm and collecting data on the change in submicron particle counts in local areas before and after seed addition is as follows:

[0077] The cooling gradient is subdivided, starting from 0.05 degrees Celsius per minute and ending at the critical rate of the fine crystal storm. At least 5 cooling rate levels are set within the range below the critical rate of the fine crystal storm (e.g., if the critical rate of the fine crystal storm is 0.20, it is subdivided into 0.05, 0.08, 0.11, 0.14, 0.17, and 0.20 degrees Celsius per minute, for a total of 6 levels). The stirring speed and impurity concentration are kept constant, consistent with step S20.

[0078] It should be noted that before each experiment, a saturated L-threonine solution was prepared according to the method in step S10 and heated to 55 degrees Celsius for 30 minutes.

[0079] Before cooling begins, add seed crystals to the solution. The amount of seed crystals used is two percent of the theoretical crystal yield, and the seed crystal particle size is controlled between one-fifth and one-tenth of the target product particle size.

[0080] Cool down at a set cooling rate, place the probe of the focused beam reflection measuring instrument near the seed crystal addition location, and record the count data of submicron particles (chord length less than 10 micrometers) every 2 seconds.

[0081] In step S30, the process of identifying the critical point at which the microcrystal rapidly decays to the baseline level and outputting the critical rate of the phantom seed crystal is as follows:

[0082] For the experiment at each cooling rate level, a curve was plotted with time on the x-axis and submicron particle count on the y-axis.

[0083] The average value of submicron particle counts within the first 60 seconds after the start of the experiment is used as the baseline value B0. The first-order difference of particle count over time is calculated. If there are three consecutive sampling points with negative difference values ​​and decreasing absolute values ​​(i.e., the rate of decrease continues to slow down), and the particle count subsequently falls back to the preset range, then the seed crystal is marked as valid. The preset range is: [B0-0.1*B0, B0+0.1*B0];

[0084] If there are three consecutive sampling points with negative difference values ​​and continuously increasing absolute values ​​(i.e., the rate of decrease is getting faster and faster), and the particle count drops to less than 50% of the baseline value B0, then it is marked as seed failure (i.e., phantom seed phenomenon).

[0085] Collect the labeled results corresponding to all cooling rate levels, arrange them in ascending order of rate, and find the rate range from seed effectiveness to seed failure.

[0086] If the interval width is greater than the distance between two adjacent rate levels, a new rate level is inserted into the interval, and the above process is repeated until the interval width is no greater than the distance between two adjacent rate levels. At this point, the lower limit of the interval is taken as the critical value, which is the critical rate of the phantom seed.

[0087] It should be noted that the specific operation of inserting a new rate level within the interval and repeating the above process is as follows: take the arithmetic mean of the upper and lower limits of the current interval as the new rate level, add the level to the test sequence, redistribute all levels evenly according to the original spacing, and repeat data collection and analysis until the interval width is less than or equal to the spacing between adjacent levels.

[0088] Understandably, the significance of step S30 is to further subdivide the cooling gradient within the range below the critical rate of the fine crystal storm, identify the critical point at which the microcrystals rapidly decay to the baseline level by the change trend of submicron particle count before and after the addition of the seed crystal, and output the critical rate of the phantom seed crystal. The core is to quantify and calibrate the cooling rate threshold that leads to seed-induced failure. After clarifying this critical rate, the loss of metastable particles formed by local overcooling after the addition of the seed crystal can be avoided, ensuring that the seed crystal truly plays its induction growth function and preventing waste of seed crystal materials and loss of control of the crystallization process.

[0089] Step S40: Take the smaller critical rate between the fine crystal storm and the phantom seed as the safety upper limit, combine it with the metastable region model, back-calculate the maximum allowable cooling rate for each temperature range, and output the segmented rate control curve.

[0090] In step S40, the process of outputting the segmented rate control curve is as follows:

[0091] Based on the requirements of the extraction and purification process, determine the starting temperature T_start (i.e., the saturation temperature, such as 50 degrees Celsius) and the ending temperature T_end (such as 20 degrees Celsius) of the crystallization process.

[0092] Divide the temperature interval between the starting temperature T_start and the ending temperature T_end into N (N ranges from 10 to 20) temperature sub-intervals, and the width of each temperature sub-interval is (T_start-T_end) / N;

[0093] For the midpoint temperature Ti of the temperature sub-interval (i represents the number of temperature sub-intervals), the saturation concentration at the midpoint temperature Ti is calculated using the solubility curve.

[0094] The solubility curve can be obtained by: preparing an L-threonine solution (containing actual impurities) consistent with the production conditions; selecting multiple temperature points (e.g., 20°C to 50°C) within the target temperature range (e.g., 20°C to 50°C); stirring at each temperature until the solution reaches saturation; sampling and filtering; determining the threonine concentration in the filtrate using the drying method or high-performance liquid chromatography; and fitting the concentration values ​​at each temperature to obtain the solubility curve.

[0095] The current safe value of the critical supersaturation is set to S_safe, where the critical supersaturation is obtained by substituting the safe upper limit, the fixed stirring speed, and the impurity concentration into the metastable region model;

[0096] For the i-th temperature sub-interval, calculate the maximum allowable supersaturation increment: ΔS_max_i = S_safe - current supersaturation (the current supersaturation is taken as the actual value at the end of the previous interval, and the initial interval is 0);

[0097] Based on the definition of supersaturation (supersaturation = actual concentration / saturated concentration - 1), we can inversely deduce the maximum allowable temperature drop ΔT_max_i when decreasing from the current temperature to the next temperature in the i-th temperature range.

[0098] Calculate the ratio between the maximum cooling amplitude ΔT_max_i and the target residence time in the temperature sub-interval to obtain the maximum allowable cooling rate R_max_i in the temperature sub-interval. Traverse each temperature sub-interval to obtain the maximum cooling rate corresponding to the temperature sub-interval.

[0099] The target residence time in the temperature sub-interval is calculated proportionally to the width of the temperature sub-interval and the corresponding maximum cooling rate.

[0100] Piecewise rate control curves are plotted with the midpoint temperature of the temperature sub-interval as the x-axis and the maximum allowable cooling rate as the y-axis.

[0101] Understandably, the significance of step S40 lies in taking the smaller of the fine-crystal storm critical rate and the phantom seed critical rate as the safety upper limit, and using the metastable region model to deduce the maximum allowable cooling rate for each temperature range, outputting a segmented rate control curve. The core is to transform discrete critical thresholds into continuous, segmented, controllable process operation boundaries. Through this curve, operators can clearly define the maximum allowable cooling rate for each temperature range, achieving a precise temperature control strategy of "slow first, then fast," maximizing crystallization efficiency while ensuring product quality.

[0102] Step S50: Based on the segmented rate control curve, dynamically adjust the cooling rate according to the deviation between the real-time supersaturation and the critical value;

[0103] In step S50, the process of dynamically outputting the cooling rate adjustment is as follows:

[0104] The segmented rate control curve output in step S40 is used as the reference cooling rate curve. An online supersaturation monitoring device is deployed inside the crystallizer. The spectral data of the solution is collected every 10 seconds using an attenuated total reflection-Fourier transform infrared spectrometer or a focused beam reflectance meter, and converted into a real-time concentration value C_real(t) through a pre-established concentration calibration model.

[0105] The concentration calibration model is established by preparing multiple L-threonine standard solutions of known concentration, collecting the spectra corresponding to each concentration, fitting the correlation equation between concentration and spectrum using partial least squares method, and finally obtaining the concentration calibration model.

[0106] Based on the current temperature, query the solubility curve to determine the saturation concentration C_sat(T) at the current temperature, and calculate the real-time supersaturation S_real(t) = C_real(t) / C_sat(T) - 1;

[0107] Based on the current temperature, the baseline cooling rate R_base(t) at the current temperature is obtained from the segmented rate control curve;

[0108] It should be noted that if the current temperature happens to be on the boundary of two temperature sub-intervals, the average rate of the two intervals is taken.

[0109] Set a critical oversaturation safety value S_limit, which is obtained by substituting the safety upper limit into the metastable region model in step S10.

[0110] Calculate the deviation between the current supersaturation and the critical safe value of supersaturation to obtain the supersaturation deviation D(t) = S_limit - S_real(t);

[0111] The cooling operation judgment range is set according to the supersaturation deviation D(t):

[0112] If the oversaturation deviation D(t) > 0.5, it indicates that the current oversaturation is far below the critical value, and there is room for acceleration.

[0113] If the oversaturation deviation D(t) ≤ 0.01, it means that the current oversaturation is close to or exceeds the critical value, and deceleration is required.

[0114] If 0.01 < oversaturation deviation D(t) ≤ 0.5, it means that the current oversaturation is within a safe range, and the baseline rate should be maintained.

[0115] Calculate the cooling rate adjustment ΔR(t) based on the cooling operation range:

[0116] If the oversaturation deviation D(t) > 0.5, then the cooling rate adjustment ΔR(t) = base cooling rate R_base(t) * 0.1 (i.e., speed up by 10%).

[0117] It should be noted that the basis for the 10% speed-up is as follows: In the setting of step S50, the speed-up is only triggered when the oversaturation difference D(t) is greater than 0.05 (far below the critical value); if the oversaturation is still far below the critical value after the speed-up, the speed-up will continue to be increased slightly in the next round until it enters the safe range; the 10% single step size can achieve smooth approximation and avoid excessive adjustment in a single step, which may cause the critical point to be exceeded.

[0118] If 0.01 < oversaturation deviation D(t) ≤ 0.5, then ΔR(t) is 0;

[0119] If the oversaturation deviation D(t) ≤ 0.01, then the cooling rate adjustment ΔR(t) = -base cooling rate R_base(t) * 0.2 (i.e., a 20% reduction).

[0120] It should be noted that the 20% deceleration is based on the following: once the supersaturation approaches the critical value, it indicates a significant increase in the nucleation risk within the crystallizer. A 20% deceleration can rapidly reduce the accumulation rate of supersaturation, allowing the system to return to a safe range as quickly as possible. There is a time delay (usually 30 seconds to 2 minutes) between the adjustment of the cooling command and the actual temperature change within the crystallizer, the solute response, and the update of the supersaturation feedback signal. During this delay, the supersaturation may continue to rise. Therefore, the deceleration amplitude needs to be large enough to offset or compensate for the accumulation of supersaturation during this response delay period. A 20% deceleration amplitude provides sufficient safety margin. The deceleration trigger condition is D(t) ≤ 0.01, meaning that the measured supersaturation is very close to or has exceeded the critical value. This is an emergency response state, requiring a larger adjustment amplitude. If the supersaturation is still in the danger zone after a 20% deceleration, the next cycle will continue to decelerate until the supersaturation falls back to a safe range.

[0121] Set upper and lower limits for ΔR(t): after acceleration, the rate shall not exceed 1.2 times the base cooling rate R_base(t), and after deceleration, the rate shall not be lower than 0.02 degrees Celsius per minute;

[0122] Calculate the actual cooling rate R_actual(t) = base cooling rate R_base(t) + cooling rate adjustment ΔR(t);

[0123] The actual cooling rate R_actual(t) is output to the temperature control system of the crystallizer as the target cooling rate in the next time interval (10 seconds);

[0124] Understandably, the significance of step S50 lies in the following: based on the segmented rate control curve, with real-time supersaturation monitoring value as input, deviation is judged through the set control rules, and the cooling rate adjustment is dynamically output. The core is to transform the static segmented rate curve into dynamic closed-loop feedback control. By monitoring the deviation between supersaturation and the critical safety value in real time, it automatically determines whether the current state is to accelerate, maintain, or decelerate, and outputs the adjustment amount to the temperature control system. This can compensate for uncertainties such as raw material batch fluctuations and equipment response delays, ensuring that the entire crystallization process always operates within the metastable safe range, and achieving stable and controllable industrial production.

[0125] The foregoing has shown and described the basic principles, main features, and advantages of the present invention. Those skilled in the art should understand that the present invention is not limited to the above embodiments. The embodiments and descriptions in the specification are merely illustrative of the principles of the invention. Various changes and modifications can be made to the invention without departing from its spirit and scope, and all such changes and modifications fall within the scope of the present invention as claimed. The scope of protection of the present invention is defined by the appended claims and their equivalents.

Claims

1. A method for extracting and purifying L-threonine after fermentation based on temperature gradient control, characterized in that: Includes the following steps: step S10: Collect time-series data of solution turbidity and particle count under different cooling rates, stirring intensities, and impurity concentrations; establish a metastable region model with varying metastable region boundaries as a function of each parameter; output the sensitivity ranking of each parameter to critical supersaturation; and determine whether the cooling rate is the parameter with the highest influence. Step S20: If so, fix the stirring intensity and impurity concentration, conduct a gradient experiment on the cooling rate, simultaneously collect particle chord length distribution data, identify the abrupt change point where the total number of particles shows an exponential increase, and output the critical rate of the fine crystal storm. Step S30: Subdivide the cooling gradient within the range below the critical rate of the fine crystal storm, collect data on the change in submicron particle count in the local area before and after the addition of the seed crystal, identify the critical point at which the microcrystals rapidly decay to the baseline level, and output the critical rate of the phantom seed crystal. Step S40: Take the smaller critical rate between the fine crystal storm and the phantom seed as the safety upper limit, combine it with the metastable region model, back-calculate the maximum allowable cooling rate for each temperature range, and output the segmented rate control curve. Step S50: Based on the segmented rate control curve, dynamically adjust the cooling rate according to the deviation between the real-time supersaturation and the critical value.

2. The method for extracting and purifying L-threonine after fermentation based on temperature gradient control according to claim 1, characterized in that: The process of establishing a metastable region model in step S10, showing how the boundary of the metastable region varies with each parameter, is as follows: The critical supersaturation measured experimentally is used as the output, and the corresponding cooling rate, stirring speed, and impurity concentration are used as the input to form multiple rows of data. After standardization, a model is established using a multiple regression method. The model includes linear terms, square terms, and interaction terms between parameters for each parameter. The model coefficients are fitted using the least squares method. The significance value of each coefficient is calculated, and terms with significance values ​​greater than 0.05 are removed. After each term is removed, the remaining coefficients are refitted until the significance values ​​of all remaining terms are less than 0.05, thus obtaining the initial metastable region model. The reliability of the model is verified by cross-validation. The data is randomly divided into multiple parts, and one part is taken in turn as the validation set, while the rest are used as the training set. The mean absolute deviation between the predicted value and the measured value is calculated. If the mean absolute deviation is less than a preset threshold, the model is verified and the metastable region model is finally obtained.

3. The method for extracting and purifying L-threonine after fermentation based on temperature gradient control according to claim 3, characterized in that: The process of sorting the sensitivity of each parameter to the critical supersaturation in step S10 is as follows: The standardized regression coefficients for each parameter are obtained from the metastable region model. The absolute values ​​of the standardized regression coefficients are arranged in descending order, and the sorting result is the sensitivity ranking of each parameter to the critical supersaturation. In the sensitivity ranking, if the cooling rate ranks first, then the cooling rate is determined to be the parameter with the highest influence.

4. The method for extracting and purifying L-threonine after fermentation based on temperature gradient control according to claim 1, characterized in that: The process of identifying the abrupt change point where the total number of particles increases exponentially and outputting the critical rate of the fine-grained storm in step S20 is as follows: Plot the total number of particles at each parameter level with time on the x-axis and the common logarithmic value of the total number of particles on the y-axis. Calculate the first-order difference between adjacent sampling points of the particle logarithmic value to obtain the difference sequence. Take the maximum value of the difference within the first 60 seconds after the start of the experiment as the baseline value. Traverse the difference sequence. If the difference values ​​of three consecutive sampling points reach or exceed ten times the baseline value, mark it as a mutation point. Otherwise, mark it as no mutation point. Arrange the marked results in ascending order of parameter values. Find the parameter interval from no mutation point to mutation point. Take the lower limit of the interval as the critical rate of fine crystal storm.

5. The method for extracting and purifying L-threonine after fermentation based on temperature gradient control according to claim 1, characterized in that: The process of identifying the critical point where the microcrystal rapidly decays to the baseline level and outputting the critical rate of the phantom seed crystal in step S30 is as follows: For each cooling rate level, a curve was plotted with time on the x-axis and submicron particle count on the y-axis. The average submicron particle count within the first 60 seconds after the start of the experiment was calculated as the baseline value. The first-order difference of particle count over time was calculated, and the difference value was used to mark whether the seed crystal was effective or ineffective. The results are sorted and labeled according to the rate from smallest to largest. The rate range from seed effectiveness to seed failure is found, and the lower limit of the range is taken as the critical rate of phantom seed.

6. The method for extracting and purifying L-threonine after fermentation based on temperature gradient control according to claim 5, characterized in that: In step S30, the method for marking the effectiveness or ineffectiveness of the seed crystal is as follows: If there are three consecutive sampling points with negative difference values ​​and decreasing absolute values, and the particle count subsequently falls back to within ±10% of the baseline value, then the seed crystal marking is valid. If the difference between three consecutive sampling points is negative and the absolute value continues to increase, and the particle count drops to less than 50% of the baseline value, then the seed crystal marker fails.

7. The method for extracting and purifying L-threonine after fermentation based on temperature gradient control according to claim 1, characterized in that: The process of taking the smaller value between the fine-grained storm critical rate and the phantom seed critical rate as the safety upper limit in step S40 is as follows: Compare the values ​​of the fine crystal storm critical rate output in step S20 and the phantom seed critical rate output in step S30, and select the smaller value as the safety upper limit; the safety upper limit is used to calculate the maximum allowable cooling rate for each temperature range in subsequent steps.

8. The method for extracting and purifying L-threonine after fermentation based on temperature gradient control according to claim 1, characterized in that: The process of outputting the piecewise rate control curve in step S40 is as follows: Determine the starting temperature and the ending temperature, and divide the temperature range between the starting temperature and the ending temperature into multiple temperature sub-intervals; for the midpoint temperature of each temperature sub-interval, calculate the saturation concentration at that midpoint temperature using the solubility curve; set a critical supersaturation safety value by substituting the safety upper limit, the fixed stirring speed, and the impurity concentration into the metastable region model. For each temperature sub-interval, calculate the maximum allowable supersaturation increment, back-calculate the maximum allowable cooling range based on the definition of supersaturation, calculate the ratio of the maximum cooling range to the target residence time of the temperature sub-interval, and obtain the maximum allowable cooling rate of the sub-interval. Plot a piecewise rate control curve with the midpoint temperature of the temperature sub-interval as the x-axis and the maximum allowable cooling rate as the y-axis.

9. The method for extracting and purifying L-threonine after fermentation based on temperature gradient control according to claim 8, characterized in that: The process of dynamically adjusting the cooling rate based on the deviation between the real-time supersaturation and the critical value in step S50 is as follows: The segmented rate control curve is used as the baseline cooling rate curve; solution data is collected every 10 seconds by an online supersaturation monitoring device, converted into real-time concentration values, and the real-time supersaturation is calculated by combining the solubility curve. The reference cooling rate is obtained from the segmented rate control curve based on the current temperature; Calculate the deviation between the real-time supersaturation and the critical safe value for supersaturation; If the deviation is greater than 0.05, the cooling rate adjustment is the base cooling rate multiplied by 0.1; if the deviation is less than or equal to 0.01, the cooling rate adjustment is the base cooling rate multiplied by -0.

2. After setting the upper and lower limits for the adjustment, the actual cooling rate is calculated and output to the temperature control system.

10. The method for extracting and purifying L-threonine after fermentation based on temperature gradient control according to claim 1, characterized in that: The process of setting upper and lower limits for the cooling rate adjustment in step S50 is as follows: Under acceleration conditions with a deviation greater than 0.05, the actual cooling rate shall not exceed 1.2 times the baseline cooling rate; under deceleration conditions with a deviation less than or equal to 0.01, the actual cooling rate shall not be less than 0.02 degrees Celsius per minute; if the calculated actual cooling rate exceeds the upper limit, the upper limit value shall be used; if it is lower than the lower limit, the lower limit value shall be used. The upper and lower limits are used to prevent excessive single adjustment from causing system oscillation or oversaturation out of control.

Images