An extraction process for the efficient extraction of demethylchlortetracycline hydrochloride from fermentation broth

By using a holographic sensing closed-loop production line and a Raman spectroscopy monitoring system to detect impurity concentration and temperature in real time, the problems of impurity generation and insufficient cooling capacity during the extraction of demethylchlortetracycline hydrochloride were solved, thus achieving consistency in finished product quality and process safety.

CN123102000APending Publication Date: 2026-11-17JIANGXI YONGTONG TECHNOLOGY CO LTD
View PDF 0 Cites 0 Cited by

Patent Information

Application Number
CN202611566115.6
Authority / Receiving Office
CN · China
Patent Type
Applications(China)
Current Assignee / Owner
Filing Date
2026-10-10
Publication Date
2026-11-17

AI Technical Summary

Technical Problem

The existing demethylchlortetracycline hydrochloride extraction process lacks real-time impurity detection methods, resulting in the inability to control the generation and accumulation of impurities in a timely manner. Insufficient cooling capacity and impurity accumulation form a vicious cycle, and conflicting adjustments of operating variables lead to quality control failure.

Method used

A holographic perception closed-loop production line is built, integrating a Raman spectroscopy monitoring system to detect impurity concentration and temperature in real time. The process feature vector is obtained through an online monitoring system, and the decision logic is executed to match and adjust control commands, thereby realizing real-time monitoring of impurity status index and thermal safety margin, as well as concurrent risk analysis of dual indicators.

Benefits of technology

Real-time monitoring of impurity concentration and temperature control during the extraction process of demethylchlortetracycline hydrochloride were achieved, avoiding accelerated dehydration and isomer formation, solving the problems of insufficient cooling capacity and impurity accumulation, and ensuring the consistency of finished product quality and process safety.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN123102000A_ABST
    Figure CN123102000A_ABST
Patent Text Reader

Abstract

This invention relates to the field of drug extraction control technology, specifically disclosing an efficient extraction process for demethylchlortetracycline hydrochloride from fermentation broth. The process includes: establishing an online monitoring system integrating the extraction process flow with Raman spectroscopy monitoring to form a holographic perception closed-loop production line; acquiring the operating characteristic vector of the back-extraction section in real time through the online monitoring system; performing single-index initial screening to identify preliminary analysis conditions; if a condition of interest is identified, conducting dual-index concurrent risk analysis to determine the final analysis condition and performing condition status judgment; executing decision logic matching; and outputting a set of adjustment and control commands based on the matched decision logic. The decision logic includes quality constraint adaptive optimization, safety mode switching based on dual-index conflict, and quality-driven online material path physical isolation, achieving deterministic control of quality and safety under extreme conditions and ensuring the consistency of the finished demethylchlortetracycline hydrochloride quality and process safety.
Need to check novelty before this filing date? Find Prior Art

Description

Technical Field

[0001] This invention relates to the field of drug extraction control technology, specifically to an extraction process for efficiently extracting demethylchlortetracycline hydrochloride from fermentation broth. Background Technology

[0002] Demethylchlortetracycline hydrochloride (also known as demethylchlortetracycline hydrochloride or demethylchlortetracycline) is an important derivative of tetracycline antibiotics. It has strong antibacterial activity and better stability than tetracycline and oxytetracycline. It is also a key starting material for the synthesis of new antibiotics such as minocycline and tigecycline. At present, the mainstream extraction process in industry is solvent extraction, which involves acidification of fermentation broth, plate and frame filtration, extraction with butyl acetate / butanol system, back-extraction with hydrochloric acid, decolorization, crystallization, centrifugation, washing and drying to obtain the finished product. Although this process is widely used in industrial practice, the following technical bottlenecks have long existed.

[0003] First, there is a lack of online real-time detection methods for key impurities. Demethylchlortetracycline hydrochloride is prone to two typical degradation reactions during extraction: under high-temperature conditions, the molecule easily dehydrates to form dehydrated demethylchlortetracycline; in acidic solutions, it easily undergoes epimerization to form tetra-demethylchlortetracycline, which has no antibacterial activity. In traditional processes, these two impurities are mainly detected by offline sampling followed by high-performance liquid chromatography (HPLC), which suffers from long sampling intervals and delayed results. This fails to reflect the dynamic changes in impurity concentrations in the back-extraction section in real time, making it difficult to provide data support for timely intervention.

[0004] Secondly, the recycling of crystallization mother liquor and the accumulation of impurities present an inherent contradiction. To improve the overall product yield, industrial processes often return a portion of the crystallization mother liquor to the back-extraction section for recycling. However, the tetradextrin enriched in the mother liquor continuously accumulates in the extraction-back-extraction closed loop during recycling, making it difficult to control its concentration within safe limits using only conventional discharge rates. After a certain number of recycling cycles, the impurity accumulation effect becomes significant, directly threatening the quality of the finished product.

[0005] Third, the degradation of cooling capacity and the accelerated formation of impurities create a vicious cycle. Back-extraction is an exothermic process, requiring the jacket cooling medium to maintain the process temperature. When the cooling system approaches its heat dissipation limit, the temperature inside the unit rises. This accelerates the formation rate of dehydrated methylchlortetracycline on one hand, and the circulation of the mother liquor containing impurities further exacerbates the heat load, creating a vicious cycle of temperature rise, accelerated impurity formation, and further temperature increases. Traditional control strategies manage cooling system protection and impurity control as independent loops, failing to establish a coordinated early warning and response mechanism between the two.

[0006] Fourth, conflicting operating variables under extreme conditions can lead to control failure. When the system simultaneously faces insufficient cooling capacity and excessive accumulation of impurities in the mother liquor, the conventional control logic has completely opposite requirements for adjusting the core operating variable, the hydrochloric acid aqueous solution feed flow rate: suppressing dehydration impurities requires reducing the flow rate to shorten the high-temperature residence time, while discharging differential isomers requires increasing the flow rate to accelerate mother liquor replacement. This conflicting demand can easily cause controller oscillation or unilateral exceedances, and may even lead to irreversible deterioration of the quality of the entire batch of materials.

[0007] To address the above problems, this invention proposes an extraction process for the efficient extraction of demethylchlortetracycline hydrochloride from fermentation broth. Summary of the Invention

[0008] The purpose of this invention is to provide an efficient extraction process for demethylchlortetracycline hydrochloride from fermentation broth, in order to solve the problems caused by the existing demethylchlortetracycline hydrochloride back-extraction section under concurrent critical conditions, where the cooling capacity approaches its limit, leading to the accelerated high-temperature generation of dehydrated demethylchlortetracycline, and the continuous accumulation of four different types of demethylchlortetracycline due to the repeated use of the crystallization mother liquor, and the control conflict and quality control problems caused by the opposite adjustment direction of the hydrochloric acid feed flow rate of the two.

[0009] The objective of this invention can be achieved through the following technical solution: an extraction process for efficiently extracting demethylchlortetracycline hydrochloride from fermentation broth, comprising the following steps: An online monitoring system integrating extraction process flow and Raman spectroscopy monitoring was established to form a holographic perception closed-loop production line; For the holographic perception closed-loop production line, the operation characteristic vector of the back-extraction section is obtained in real time through the online monitoring system, including the impurity status index, thermal safety margin and temperature field distribution characteristics. Based on the operating characteristic vector of the back-extraction section, a single-index preliminary screening including cooling margin preliminary screening and isomer accumulation preliminary screening is performed to identify the preliminary analysis conditions. The preliminary analysis conditions include normal conditions and conditions of concern. If it is a condition of concern, a dual-index concurrent risk analysis is performed to determine the final analysis conditions and to perform condition status determination. The condition status includes the concurrent critical condition preparation state and the irreversible cooling failure state. Based on the final analysis of the working conditions and the matching of the working conditions, the system outputs a set of adjustment and control instructions according to the matched decision logic. The decision logic includes quality constraint adaptive optimization, safety mode switching based on dual index conflict, and quality-driven online material path physical isolation.

[0010] Furthermore, the method for forming a holographic perception closed-loop production line is as follows: A demethylchlortetracycline hydrochloride extraction production line was constructed, including an acidification tank, a heating and insulation device, a plate and frame filter press, a filtrate return and salting-out agent dosing device, a multi-stage countercurrent centrifugal extractor, a back-extraction device, a decolorization kettle, a filtration and carbon removal device, a crystallization kettle, a centrifuge, a washing device, and a vacuum dryer, which are connected in sequence through process pipelines. The extractant is a mixed solvent. The aqueous phase inlet of the back-extraction device is connected to the fresh hydrochloric acid aqueous solution supply pipeline and the crystallization mother liquor circulation pipeline through a three-way confluence pipeline. An optical fiber flow cell is connected in series on the pipeline where the discharge pipeline of the back-extraction unit and the crystallization mother liquor circulation pipeline converge. The optical fiber flow cell is connected to a Raman spectrometer with a built-in quantitative model of dehydrated demethylchlortetracycline and four-dimensional demethylchlortetracycline via optical fiber. Temperature sensing networks and flow sensing networks are deployed inside the back-extraction unit and on the discharge pipeline, the jacket cooling medium pipeline, the fresh hydrochloric acid aqueous solution supply pipeline, and the crystallization mother liquor circulation pipeline, respectively. The online monitoring system is connected to the industrial controller via fieldbus.

[0011] Furthermore, the impurity status index is obtained as follows: The mixed liquid flowing through the optical fiber flow cell is excited in real time and the original spectrum is obtained by performing spectral acquisition through a Raman spectrometer. In each sampling cycle, the Raman spectrometer performs preprocessing and concentration prediction on the original spectrum through the built-in dehydrated demethylchlortetracycline and tetradextrin combined quantitative model, and simultaneously outputs the predicted concentration values ​​of dehydrated demethylchlortetracycline and tetradextrin in the mixed liquid, and sorts them to obtain the impurity concentration level. The concentration prediction sequences of dehydrated demethylchlortetracycline and four-dimensional demethylchlortetracycline from the most recent preset number of sampling periods are extracted from the real-time database and processed separately to obtain the concentration change rate of dehydrated demethylchlortetracycline and the concentration change rate of four-dimensional demethylchlortetracycline. These are integrated and recorded as the impurity change trend. The impurity situation index is composed of the impurity concentration level and the impurity change trend.

[0012] Furthermore, the methods for obtaining the thermal safety margin and temperature field distribution characteristics are as follows: The acquisition of thermal safety margin includes: continuously acquiring the inlet temperature, outlet temperature and volumetric flow rate of the cooling medium through temperature sensors and flow meters on the jacketed cooling medium pipeline; processing the data in conjunction with the pre-calibrated specific heat capacity of the cooling medium to obtain the current actual heat exchange; and processing the current actual heat exchange with the pre-stored rated maximum heat exchange to obtain the cooling capacity saturation. Extract the cooling capacity saturation sequence of the most recent preset number of sampling periods from the real-time database, process the cooling capacity saturation sequence to obtain the cooling capacity saturation change trend, and the thermal safety margin is composed of the cooling capacity saturation and the cooling capacity saturation change trend. The acquisition of temperature field distribution characteristics includes: obtaining the material temperature at the top, middle and bottom of the back-extraction unit through temperature sensors installed at the top, middle and bottom positions along the material flow direction inside the back-extraction unit; obtaining the back-extraction liquid outlet temperature through temperature sensors on the back-extraction unit's discharge pipeline; and integrating these to obtain the temperature field distribution characteristics.

[0013] Furthermore, the method for identifying preliminary analysis conditions is as follows: The evaluation cycle is defined, and the execution of the initial screening of cooling margin includes: extracting the cooling capacity saturation at the current moment from the thermal safety margin of the back-extraction section's operating characteristic vector; if it exceeds the first-level cooling warning line, the initial screening of cooling margin is marked as triggered. The execution of the initial screening of isomer accumulation includes: extracting the predicted value of the four-dimensional demethylchlortetracycline concentration at the current moment from the impurity status index of the back-extraction section's operating characteristic vector; if it exceeds the first-level isomer warning line, the initial screening of isomer accumulation is marked as triggered. If neither the cooling margin screening nor the isomer accumulation screening is triggered, the preliminary analysis condition of the current evaluation cycle will be identified as a normal condition; otherwise, it will be identified as a condition of concern.

[0014] Furthermore, the method for determining the final analysis conditions is as follows: If the initial analysis indicates that the operating condition is of concern, the system will enter the dual-indicator concurrent risk analysis stage. The cooling capacity saturation at the current moment will be extracted from the thermal safety margin and compared with the preset second-level cooling conflict warning line. At the same time, the predicted value of the four-dimensional demethylchlortetracycline concentration at the current moment will be extracted from the impurity status index and compared with the preset second-level isomer conflict warning line. If the cooling capacity saturation does not exceed the second-level cooling conflict warning line, or the predicted value of the four-dimensional demethylchlortetracycline concentration does not exceed the second-level isomer conflict warning line, the final analysis condition of the current evaluation cycle will be identified as a condition of concern, and the sampling cycle of the Raman spectrometer will be shortened to the encrypted monitoring frequency; otherwise, it will be identified as a concurrent critical condition.

[0015] Furthermore, the method for determining the operating condition is as follows: The cooling capacity saturation sequence within the most recent preset evaluation period is extracted from the thermal safety margin. The cooling capacity saturation sequence is processed to obtain the rate of increase of cooling capacity saturation. The dehydrated demethylchlortetracycline concentration change rate sequence within the most recent preset evaluation period is extracted from the impurity status index. The dehydrated demethylchlortetracycline concentration change rate sequence is processed to obtain the acceleration of dehydrated demethylchlortetracycline concentration generation. If the cooling capacity saturation has reached the preset cooling saturation limit, and the generation acceleration of dehydrated methylchlortetracycline concentration is greater than the preset impurity acceleration limit criterion, the current evaluation cycle's operating condition will be judged as an irreversible cooling failure state; otherwise, it will be judged as a preparatory state for concurrent critical operating conditions.

[0016] Furthermore, the method for executing decision logic matching and outputting adjustment and control command groups is as follows: When the final analysis condition is the condition of interest, the matching quality constraint adaptive optimization is used as the current decision logic to generate a set of adjustment and control instructions. When the final analysis condition is a concurrent critical condition and the condition status is a concurrent critical condition preparation state, the safety mode switching based on dual-index conflict is matched as the current decision logic. The optimization solution process of quality constraint adaptive optimization is immediately suspended. The flow rate of the controlled replenishing pump is forcibly limited to the preset minimum safe flow rate. The opening command of the electric regulating valve on the jacket cooling medium pipeline is forcibly set to the fully open state. The opening of the electromagnetic discharge valve on the crystallization mother liquor circulation pipeline is increased to the preset high discharge opening at the maximum allowable rate. The circulating mother liquor is discharged at the preset maximum discharge flow rate. Based on the minimum safe flow rate, the controlled replenishing pump increases the replenishing flow rate synchronously according to the mother liquor flow rate discharged by the electromagnetic discharge valve to make equal replacement. The adjustment command group is then obtained. When the final analysis condition is a concurrent critical condition and the condition status is an irreversible cooling failure, the physical isolation of the online material path driven by the matching quality is used as the current decision logic. On the basis of maintaining the adjustment command group generated by the safety mode switching based on the dual-index conflict, a switching command is issued to the three-way switching valve on the discharge pipeline of the back-extraction unit. The flow direction of the three-way switching valve is switched from the normal downstream pipeline connected to the decolorization kettle to the isolation pipeline connected to the temporary storage tank. An interlock pause signal is generated and sent to the feed pump of the aqueous phase to be extracted and the feed pump of the extractant of the multi-stage countercurrent centrifugal extractor through the fieldbus. The supply of organic extract phase and aqueous phase to be extracted to the back-extraction unit is forcibly stopped, and the adjustment command group is obtained.

[0017] Furthermore, the method for generating the adjustment and control instruction set under the decision logic of quality constraint adaptive optimization is as follows: Using the ratio of the circulating flow rate of the crystallization mother liquor to the feed flow rate of fresh hydrochloric acid aqueous solution as the independent variable, an optimization problem is established with the two optimization objectives of maximizing product yield and minimizing energy consumption. The objective function is constructed as follows: the product yield is obtained by reverse normalization, and the energy consumption is obtained by forward normalization. The predicted concentration of the four-dimensional demethylchlortetracycline at the current moment is extracted from the impurity status index and processed to obtain the normalized value of the isomer concentration. The normalized value of the isomer concentration is used as the independent variable to construct a penalty term, which is set to zero when the normalized value of the isomer concentration is lower than the preset penalty activation threshold. The objective function is obtained by summing the weighted fusion value of the normalized energy consumption value and the yield deviation with the penalty term. The objective function is solved online in a rolling manner to obtain the optimal ratio of the circulating flow rate of the crystallization mother liquor to the feed flow rate of the fresh hydrochloric acid aqueous solution within the current evaluation period, which minimizes the objective function. The optimal ratio is then processed to obtain two operational variables: the discharge flow rate of the crystallization mother liquor and the replenishment flow rate of the fresh hydrochloric acid aqueous solution. The discharge flow rate of the crystallization mother liquor is converted into the opening command of the electromagnetic discharge valve on the crystallization mother liquor circulation pipeline, and the replenishment flow rate of the fresh hydrochloric acid aqueous solution is converted into the speed command of the controlled replenishment pump on the fresh hydrochloric acid aqueous solution supply pipeline. The resulting set of adjustment and control commands is then obtained.

[0018] Furthermore, the product yield is determined by the proportion of the recovered demethylchlortetracycline hydrochloride in the total aqueous phase flow rate entering the back-extraction unit to the circulating flow rate of the crystallization mother liquor. The total aqueous phase flow rate entering the back-extraction unit is obtained by summing the fresh hydrochloric acid aqueous solution feed flow rate and the crystallization mother liquor circulating flow rate. The fresh hydrochloric acid aqueous solution feed flow rate and the crystallization mother liquor circulating flow rate are collected in real time by electromagnetic flowmeters installed on the corresponding pipelines. Energy consumption consists of the heating medium consumption of the heating and insulation device and the cooling medium consumption of the jacket of the back-extraction unit. The heating medium consumption and the jacket cooling medium consumption are collected in real time by temperature sensors and flowmeters installed on the corresponding pipelines and the data is processed.

[0019] The beneficial effects of this invention are as follows: 1. This invention connects an optical fiber flow cell in series with the outlet pipeline of the back-extraction device and the circulation pipeline of the crystallization mother liquor. Combined with a Raman spectrometer that incorporates a combined quantitative model of dehydrated demethylchlortetracycline and four-dimensional demethylchlortetracycline, it achieves simultaneous online real-time prediction of the concentrations of two key impurities, overcoming the technical shortcomings of traditional offline sampling and detection with lag. By combining a temperature sensing network and a flow sensing network, a holographic operating feature vector is formed that covers the impurity status index, thermal safety margin, and temperature field distribution characteristics, providing a real-time data foundation for multi-index joint operating condition determination.

[0020] 2. This invention solves the conflict between operational variables in extreme conditions by setting two-level early warning lines to jointly determine concurrent critical operating conditions and rigidly anchoring the hydrochloric acid feed flow rate to the minimum safe flow rate after switching to the safety mode. Simultaneously, it executes the maximum discharge of the mother liquor discharge valve and the equal-volume replacement of the replenishment pump. Furthermore, it isolates the unqualified back-extraction liquid in real time and interlocks to stop the upstream feed through a three-way switching valve, avoiding irreversible deterioration of the quality of the entire batch of materials. This invention achieves a graded and progressive quality and safety control from conventional optimization to safety mode switching and then to material isolation. Attached Figure Description

[0021] The invention will now be further described with reference to the accompanying drawings.

[0022] Figure 1This is a flowchart of the extraction process for efficiently extracting demethylchlortetracycline hydrochloride from fermentation broth according to an embodiment of the present invention; Figure 2 This is a flowchart illustrating the specific steps involved in determining the final analytical conditions and performing condition status determination in an extraction process for efficiently extracting demethylchlortetracycline hydrochloride from fermentation broth, as described in an embodiment of the present invention. Figure 3 This is the iterative convergence curve of adaptive optimization of mass constraints under different impurity loads in the embodiments of the present invention. Detailed Implementation

[0023] 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.

[0024] Example 1

[0025] Please see Figure 1 As shown in the embodiment of the present invention, an extraction process for efficiently extracting demethylchlortetracycline hydrochloride from fermentation broth aims to solve the problems of control conflict and quality loss caused by the opposite direction of hydrochloric acid feed flow regulation in the existing demethylchlortetracycline hydrochloride back-extraction section under concurrent critical conditions. This is due to the near-limited cooling capacity leading to accelerated high-temperature formation of dehydrated demethylchlortetracycline, the continuous accumulation of four different types of demethylchlortetracycline caused by the repeated use of the crystallization mother liquor, and the resulting quality loss. The goal is to achieve consistency in the quality of the finished demethylchlortetracycline hydrochloride product and ensure process safety. Specifically, the process includes the following steps: S1: Establish an online monitoring system for the extraction process and integrate Raman spectroscopy monitoring to form a holographic perception closed-loop production line; Specifically, for the extraction process, a demethylchlortetracycline hydrochloride extraction production line is built, including an acidification tank, a heating and insulation device, a plate and frame filter press, a filtrate return and salting-out agent dosing device, a multi-stage countercurrent centrifugal extractor, a back-extraction device, a decolorization kettle, a filtration and decarbonization device, a crystallization kettle, a centrifuge, a washing device, and a vacuum dryer, which are connected in sequence through process pipelines. The acidification tank is used to receive fermentation broth containing demethylchlortetracycline hydrochloride. The pH value of the fermentation broth is adjusted to the preset acidification endpoint by adding an acid regulator to complete the acidification. The outlet of the acidification tank is connected to a heating and heat preservation device, which heats the acidified fermentation broth through a heating medium and maintains it within the preset heat preservation temperature range. The feed inlet of the plate and frame filter press is connected to the discharge outlet of the heating and insulation device to filter the heated fermentation broth, separating the acidic filtrate rich in demethylchlortetracycline hydrochloride from the mycelial filter cake. The filtrate return and salting-out agent dosing device is connected to the filtrate outlet of the plate and frame filter press. By adding alkaline regulator and salting-out agent to the acidic filtrate, the pH value of the filtrate is returned to the preset extraction pH window and the salting-out effect is introduced to form the aqueous phase to be extracted. It should be noted that the amount of salting-out agent added is pre-calibrated based on the relationship curve between the concentration of the target product in the filtrate and the preset salting-out agent concentration-extraction partition coefficient. The heavy phase inlet of the multi-stage countercurrent centrifugal extractor is connected to the outlet of the filtrate return and salting-out agent dosing device, and the light phase inlet is connected to the extractant supply pipeline. Through multi-stage countercurrent contact, the efficient mass transfer of demethylchlortetracycline hydrochloride from the aqueous phase to be extracted to the organic extraction phase is achieved. The outlet of the organic extraction phase is connected to the back-extraction device, and the raffinate aqueous phase is discharged to the waste liquid treatment unit. The extractant is a mixed solvent composed of butyl acetate, n-butanol, and methyl isobutyl ketone. The back-extraction device is used to back-extract the organic extract phase loaded with demethylchlortetracycline into an aqueous hydrochloric acid solution, so that the demethylchlortetracycline hydrochloride is transferred from the organic phase to the aqueous phase, resulting in a back-extraction solution rich in demethylchlortetracycline hydrochloride. The organic phase inlet of the back-extraction device is connected to the organic extract phase outlet of the multi-stage countercurrent centrifugal extractor, and the aqueous phase inlet is connected to the fresh aqueous hydrochloric acid solution supply line and the crystallization mother liquor circulation line through a three-way confluence pipe, and a pipeline mixer is installed at the confluence. It should be noted that the crystallization mother liquor circulation pipeline is used to return part or all of the crystallization mother liquor generated in subsequent crystallization and centrifugation processes to the back-extraction section for reuse. The back-extraction device delivers the back-extraction liquid to the decolorization kettle. The decolorization kettle contains activated carbon, and the back-extraction liquid is stirred to ensure full contact with the activated carbon. The outlet of the decolorization kettle is connected to a filtration and carbon removal device, which separates the activated carbon from the decolorization liquid through precision filtration. The decolorization liquid then enters the crystallization kettle. The crystallization kettle regulates the temperature of the decolorizing solution through a jacketed cooling medium and cools and crystallizes according to a preset cooling curve. The outlet of the crystallization kettle is connected to a centrifuge, and wet crystals of demethylchlortetracycline hydrochloride and crystallization mother liquor are obtained by centrifugation. The crystallization mother liquor is collected and connected to the crystallization mother liquor circulation pipeline, while the wet crystals enter the washing device. After being washed with pre-cooled purified water and organic solvent, they are sent to a vacuum dryer for vacuum drying to obtain the finished product of demethylchlortetracycline hydrochloride. It should be noted that the finished product of demethylchlortetracycline hydrochloride is a raw material product that meets the pharmacopoeia standards. Key quality indicators include the content of the main component, moisture, impurity profile and solvent residue. The extraction process flow represents a complete closed-loop process for obtaining demethylchlortetracycline hydrochloride product through the demethylchlortetracycline hydrochloride extraction production line. Specifically, the fermentation broth is acidified in an acidification tank, heated and kept warm by a heating and insulation device, and separated into solid and liquid components by a plate and frame filter press to produce an acidic filtrate. The acidic filtrate is then adjusted to pH and salted out by a filtrate return and salting-out agent addition device to become the aqueous phase to be extracted. The aqueous phase to be extracted is then extracted by a multi-stage countercurrent centrifugal extractor to produce an organic extract phase loaded with demethylchlortetracycline. The organic extract phase is then back-extracted by a back-extraction device to produce an aqueous solution of demethylchlortetracycline hydrochloride, i.e., the back-extraction liquid. The back-extraction liquid is then decolorized by activated carbon in a decolorizing kettle and filtered to remove carbon to produce a decolorized liquid. The decolorized liquid is then crystallized in a crystallizing kettle and centrifuged to produce wet crystals of demethylchlortetracycline hydrochloride and a mother liquor for crystallization. The wet crystals are washed by a washing device and vacuum dried by a vacuum dryer to obtain the demethylchlortetracycline hydrochloride product. The mother liquor for crystallization is then returned to the aqueous phase feed end of the back-extraction device via a mother liquor circulation pipeline for reuse. An online monitoring system integrating Raman spectroscopy monitoring was deployed on the hydrochloric acid demethylchlortetracycline extraction production line to form a holographic perception closed-loop production line. The online monitoring system includes a Raman spectrometer, a temperature sensing network, and a flow sensing network. Specifically, a fiber optic flow cell is connected in series between the discharge pipeline of the back-extraction unit and the crystallization mother liquor circulation pipeline. The fiber optic flow cell is connected to a Raman spectrometer with a built-in quantitative model of dehydrated demethylchlortetracycline and four-dimensional demethylchlortetracycline via optical fiber. It should be noted that the combined quantitative model of dehydrated demethylchlortetracycline and tetradextrin is not based solely on the spectrum of pure standard solutions as the training set. Instead, it is pre-established using a matrix-matched standard addition method combined with calibration of actual production samples. Its specific construction and online prediction methods include training sample preparation, spectral acquisition, spectral preprocessing and overlapping peak unmixing, combined quantitative model construction, model validation and maintenance, and online prediction and anomaly detection. Specifically, the training sample preparation includes: taking the mixed liquid from the back-extraction section, removing the target impurities to obtain a matrix blank; adding known concentrations of dehydrated demethylchlortetracycline standard and tetradextrin standard to the matrix blank to prepare a series of concentration samples, covering a concentration range of 0.1 μg / mL to 100 μg / mL; simultaneously changing the sample's pH value, salting-out agent concentration, residual organic solvent ratio, crystallization mother liquor ratio, and temperature to cover the actual extract matrix changes; and taking samples from actual production batches to determine the reference concentrations of dehydrated demethylchlortetracycline and tetradextrin using high performance liquid chromatography as external validation and model calibration samples. Spectral acquisition includes: using a Raman spectrometer with an excitation wavelength of 785nm and a resolution of 4cm-1, a laser power of 100-300mW, an integration time of 1-5s, and 8-32 accumulations; the temperature of the fiber optic flow cell is controlled within the set value ±0.5℃; each sample is acquired at least 3 times, the average spectrum is taken, and the acquisition temperature, pH, flow rate and other operating parameters are recorded. Spectral preprocessing and overlapping peak unmixing include: performing dark current subtraction, baseline correction, standard normal variable transformation, multivariate scattering correction, wavelet denoising, and first or second derivative processing on the original spectrum in sequence; dividing the 500–1800 cm⁻¹ fingerprint region into several sub-intervals, and using competitive adaptive reweighted sampling or continuous projection algorithms to screen characteristic bands, focusing on extracting the second derivative difference information of dehydrated norchlortetracycline and tetradiffusion relative to the principal component norchlortetracycline hydrochloride in the 1000–1100 cm⁻¹, 1200–1350 cm⁻¹, and 1450–1650 cm⁻¹ intervals; and using multivariate curve-resolved alternating least squares method to unmix overlapping peaks to obtain the spectral contribution matrices of the principal component, dehydrated norchlortetracycline, and tetradiffusion. The joint quantitative model construction includes: using the unmixed impurity spectrum contribution matrix, principal component concentration covariates, and operating parameters as inputs, establishing a PLS2 model or support vector regression model, and simultaneously outputting the predicted concentration values ​​of dehydrated demethylchlortetracycline and tetracycline-differential demethylchlortetracycline; introducing non-negativity constraints and total tetracycline concentration constraints during modeling to suppress the misjudgment of principal component concentration fluctuations as impurity signals; the model uses the Kennard-Stone method to divide the calibration set and validation set, and determines the number of principal components or kernel parameters through cross-validation. Model validation and maintenance include: model validation metrics include root mean square error of prediction, coefficient of determination, recovery rate, and limit of detection; preferably, the recovery rate of dehydrated demethylchlortetracycline and tetradextrin demethylchlortetracycline is 90%–110%, the root mean square error of prediction is no greater than 10%, and the limit of detection is no higher than 0.5 μg / mL; actual samples are extracted for verification by high performance liquid chromatography at the time of each production batch, and if the prediction deviation exceeds the set threshold continuously, the model is updated; the model update uses new batch matrix-matched samples and HPLC reference values ​​for incremental correction; Online prediction and anomaly detection include: During online operation, the Raman spectrometer acquires the raw spectrum of the mixed liquid in the fiber optic flow cell according to the sampling period. After the same preprocessing, it is input into the joint quantitative model and outputs the predicted concentration values ​​of dehydrated demethylchlortetracycline, the predicted concentration values ​​of four-dimensional demethylchlortetracycline, and the prediction confidence level. At the same time, the Mahalanobis distance, residual sum of squares, and Q statistic of the spectrum are calculated. If the anomaly threshold is exceeded or the prediction confidence level is lower than the set value, the previous valid prediction value is retained and encrypted sampling, offline HPLC arbitration, or model correction is triggered. It should be noted that the joint quantitative model can achieve simultaneous online quantitative prediction of two key impurities under the conditions that the Raman characteristic peaks of hydrochloric acid demethylchlortetracycline, dehydrated demethylchlortetracycline and tetra-differential demethylchlortetracycline are highly overlapping and the actual extract matrix is ​​complex. Temperature and flow sensing networks are deployed at multiple locations within the back-extraction unit; Specifically, temperature sensors are installed at the upper, middle and lower parts of the material flow direction inside the back-extraction unit, temperature sensors are installed on the discharge pipe of the back-extraction unit, temperature sensors and flow meters are installed on the inlet and outlet pipes of the jacket cooling medium, and electromagnetic flow meters are installed on the fresh hydrochloric acid aqueous solution supply pipe and the crystallization mother liquor circulation pipe. The online monitoring system is connected to the industrial controller via a fieldbus, and the industrial controller is equipped with a wireless communication module to enable communication with the manufacturing execution system and the cloud-based quality management platform, thus completing the construction of a holographic perception closed-loop production line and entering a state of continuous operation and online monitoring. It should be noted that the Raman spectrometer, fiber optic flow cell, fiber optic probe, temperature sensing network, flow sensing network, and industrial controller are all industrial-grade online equipment. After installation, the following debugging and verification were completed: Raman spectrometer wavelength calibration, dark current calibration, and fiber optic flow cell optical path calibration; point-by-point calibration of the temperature sensor and flow meter; training, verification, and blind sample testing of the joint quantitative model using matrix-matched standard samples and actual production samples; determination of the concentrations of dehydrated demethylchlortetracycline and tetradextrin in the actual back-extraction sample using high-performance liquid chromatography, comparison with Raman predicted values, and generation of model verification reports, installation and debugging records, operation logs, and maintenance records. These records serve as the basis for the online monitoring system's ability to achieve synchronous online quantitative analysis. S2: For the holographic perception closed-loop production line, the operating feature vector of the back-extraction section is obtained in real time through the online monitoring system; Specifically, after the holographic perception closed-loop production line enters continuous operation and online monitoring status, the industrial controller obtains the operating characteristic vector of the back-extraction section in real time through the online monitoring system, including the impurity status index, thermal safety margin and temperature field distribution characteristics. The impurity state index is obtained as follows: In the online monitoring system integrating Raman spectroscopy monitoring, the optical fiber flow cell is connected in series in the pipeline after the discharge pipeline of the back-extraction device and the crystallization mother liquor circulation pipeline are combined. The mixed liquid flowing through the optical fiber flow cell is excited in real time under non-contact conditions and the original spectrum is obtained by performing spectral acquisition through Raman spectrometer. Within each sampling period, the Raman spectrometer performs preprocessing on the raw spectrum, including baseline correction, standard normal variable transformation, and characteristic band interception, using the built-in dehydrated demethylchlortetracycline and four-differential demethylchlortetracycline joint quantitative model, to obtain the spectral matrix. The spectral matrix is ​​then input into the solidified dehydrated demethylchlortetracycline and four-differential demethylchlortetracycline joint quantitative model for concentration prediction output. Simultaneously, the predicted concentration values ​​of dehydrated demethylchlortetracycline and four-differential demethylchlortetracycline in the mixed liquid are output, and the impurity concentration level is obtained by sorting. It should be noted that the sampling period refers to the time interval during which the dehydrated demethylchlortetracycline and four-dimensional demethylchlortetracycline combined quantitative model built into the Raman spectrometer performs one spectral acquisition, preprocessing and concentration prediction output on the mixed liquid flowing through the fiber optic flow cell. The mixed liquid refers to the liquid after the back-extraction liquid and the crystallization mother liquor returned through the crystallization mother liquor circulation pipeline are combined and thoroughly mixed in the pipeline mixer. Its impurity concentration distribution directly reflects the superposition state of the current back-extraction effect and the cumulative effect of the mother liquor. The impurity status index includes impurity concentration level and impurity change trend. When the industrial controller receives a set of impurity concentration levels, it binds the corresponding timestamp and stores it in the corresponding real-time database. The industrial controller extracts the most recent preset number of sampling period's dehydrated demethylchlortetracycline concentration prediction value sequence and four-way demethylchlortetracycline concentration prediction value sequence from the corresponding real-time database. It then uses the least squares method to perform first-order linear fitting on the two sets of sequences. The slopes obtained from the fitting are used as the dehydrated demethylchlortetracycline concentration change rate and the four-way demethylchlortetracycline concentration change rate, respectively. These are combined and recorded as the impurity change trend. It should be noted that the positive or negative sign of the impurity change trend indicates the direction of concentration increase or decrease, the absolute value of the impurity change trend indicates the rate of concentration change, and the impurity situation index characterizes the concentration level and change direction of dehydrated demethylchlortetracycline and tetra-differential demethylchlortetracycline in the back-extraction section at the current moment and in recent historical periods, providing a data basis for the cumulative screening of isomers in the subsequent single-index primary screening. The thermal safety margin is obtained as follows: the inlet temperature Tin, outlet temperature Tout, and volumetric flow rate V of the cooling medium are continuously acquired by temperature sensors and flow meters installed on the jacketed cooling medium pipeline; the volumetric flow rate is converted into mass flow rate m = ρ(Tavg)·V based on the pre-calibrated density-temperature curve ρ(T) of the cooling medium, where Tavg = (Tin + Tout) / 2; the specific heat capacity cp(Tavg) at the average temperature is obtained based on the pre-calibrated specific heat capacity-temperature curve cp(T); and the actual heat exchange rate Q is calculated in real time using the formula Q = m·cp(Tavg)·(Tout - Tin). The rated maximum heat exchange is corrected based on the current inlet temperature Tin of the cooling medium: Q_max(Tin)=mmax·cp(Tin)·(Tout,max-Tin), where mmax is the design maximum mass flow rate of the jacket cooling medium and Tout,max is the design maximum allowable outlet temperature; the cooling capacity saturation S=Q / Qmax(Tin), when S is greater than 1, S=1 and the cooling capacity over-limit alarm is triggered; To ensure data authenticity, the industrial controller performs data validity verification on the cooling capacity saturation calculation: the timestamp alignment error between the temperature sensor and the flow meter does not exceed one sampling period; if the absolute value of the temperature difference between the inlet and outlet of the cooling medium is less than the preset minimum effective temperature difference, the sampling point is marked as invalid; if the volumetric flow rate is lower than the lower limit of the flow meter's range or higher than the upper limit of the range, the sampling point is marked as invalid; if Q is less than 0 or Q is greater than 1.2 times Qmax(Tin), it is determined to be a sensor abnormality or calculation abnormality, marked as invalid data, and a maintenance alarm is pushed; the real-time database stores the original V, Tin, Tout, ρ, cp, Q, Qmax(Tin), S, and data quality code; subsequent cooling capacity saturation change trends are only fitted using the least squares data with valid quality codes; It should be noted that, therefore, when the inlet temperature, outlet temperature, and volumetric flow rate of the cooling medium are different, the actual heat exchange Q and the corrected rated maximum heat exchange Qmax(Tin) change synchronously, and the cooling capacity saturation S varies with the operating conditions. There will not be a situation where the same fixed value is obtained under different temperature and flow rate combinations. If a large number of identical saturations appear in the historical data, they should be verified by the original flow rate, temperature difference, density, specific heat capacity, and data quality code, and invalid data should be deleted or corrected. The thermal safety margin includes cooling capacity saturation and cooling capacity saturation change trend. The industrial controller binds the received cooling capacity saturation to the corresponding timestamp and stores it in the corresponding real-time database. It extracts the cooling capacity saturation sequence of the most recent preset number of sampling periods from the corresponding real-time database, performs first-order linear fitting using the least squares method, and uses the slope obtained from the fitting as the cooling capacity saturation change trend. It should be noted that cooling capacity saturation is a dimensionless index that characterizes the current heat dissipation margin of the back-extraction unit. The closer its value is to the upper limit, the closer the cooling system is to its heat dissipation capacity limit, and the smaller the remaining available cooling margin. The value and trend of cooling capacity saturation together constitute the thermal safety margin, which provides a data basis for the initial screening of cooling margin in the subsequent single-index screening. The temperature field distribution characteristics are obtained by using temperature sensors installed at the upper, middle and lower positions along the material flow direction inside the back-extraction device to obtain the material temperature at the top, middle and lower parts of the back-extraction device, and the temperature of the back-extraction liquid outlet is obtained by using temperature sensors on the discharge pipe of the back-extraction device. A set of temperature field distribution characteristics along the material flow direction is then obtained. It should be noted that the temperature field distribution characteristics are used to determine whether there is an abnormal state of local overheating or excessive temperature gradient in the back-extraction unit. Together with the thermal safety margin, they constitute a complete characterization of the thermal state and cooling margin of the back-extraction section. When the temperature field distribution characteristics are overheated, even if the thermal safety margin shows that the overall cooling capacity saturation has not yet reached the warning line, it still indicates that there may be a risk of uneven heat exchange or local degradation intensification inside the back-extraction unit. It should be noted that the industrial controller performs unified data governance on data stored in the real-time database, including cooling capacity saturation, actual heat exchange, volumetric flow rate, inlet temperature, outlet temperature, predicted concentration of dehydrated methylchlortetracycline, and predicted concentration of four-dimensional methylchlortetracycline. Each data entry is bound to a timestamp, device number, data quality code, and calculation version number. The calculation version number corresponds to the density-temperature curve, specific heat capacity-temperature curve, correction formula for rated maximum heat exchange, and unit conversion factor. Before data is entered into the database, range verification, rate of change verification, time alignment verification, and material balance verification are performed. Abnormal data is not included in trend fitting or early warning determination, and abnormal records are automatically generated. Through the above data governance, abnormal situations where the cooling capacity saturation shows the same value under different temperature and flow rate combinations are avoided due to inconsistent units, sensor drift, fixed specific heat capacity values, or manual rounding. It should be noted that the impurity concentration or temperature reading at a single moment cannot reflect the dynamic deterioration trend of the back-extraction section. Traditional single-point threshold alarms are prone to false alarms due to instantaneous fluctuations and lack quantitative indicators that characterize the degree of deterioration, such as the trend of cooling capacity saturation change and the acceleration of impurity generation. It is difficult to identify the precursors of irreversible deterioration in an early manner. The role of this step is to obtain the impurity change trend by fitting the concentration prediction value sequence with the least squares method, obtain the cooling capacity saturation change trend by fitting the cooling capacity saturation sequence, and integrate the temperature field distribution characteristics to form an operating feature vector that covers the current level, the direction of change, and the rate of change. This enables the system to distinguish between instantaneous fluctuations and substantial deterioration, and provides a quantitative basis for graded early warning. S3: Based on the operating characteristic vector of the back-extraction section, perform single-index preliminary screening including cooling margin preliminary screening and isomer accumulation preliminary screening to identify preliminary analysis conditions. Preliminary analysis conditions include normal conditions and conditions of concern. If it is a condition of concern, perform dual-index concurrent risk analysis to determine the final analysis condition and perform condition status judgment. Condition status includes concurrent critical condition preparation state and irreversible cooling failure state. like Figure 2As shown, the specific method for determining the final analysis working condition and performing the working condition status judgment is as follows; Specifically, for the operating characteristic vector of the back-extraction section, the industrial controller defines a fixed-duration evaluation cycle. Within each evaluation cycle, the risk screening and calibration are performed sequentially in three stages: single-index initial screening, dual-index concurrent risk analysis, and operating condition determination. The duration of the evaluation cycle is consistent with the sampling cycle of the Raman spectrometer, ensuring that each update of the impurity status index triggers a complete risk identification and operating condition determination process. The single-index initial screening stage includes two parallel screening operations: cooling margin initial screening and isomer accumulation initial screening. The initial screening of cooling margin is performed as follows: the industrial controller extracts the current cooling capacity saturation from the thermal safety margin of the back-extraction section's operating feature vector and compares it with the preset first-level cooling warning line. If the cooling capacity saturation exceeds the first-level cooling warning line, the initial screening of cooling margin is marked as triggered. The execution method of cumulative screening of isomers is as follows: the industrial controller extracts the current time's predicted concentration of tetra-differential demethylchlortetracycline from the impurity status index of the back-extraction section's operating characteristic vector, and compares it with the preset first-level isomer warning line. If the predicted concentration of tetra-differential demethylchlortetracycline exceeds the first-level isomer warning line, the cumulative screening of isomers is marked as triggered. It should be noted that both the first-level cooling warning line and the first-level isomer warning line are pre-calibrated through offline experiments and statistical analysis of historical batch data, and are stored in the industrial controller as configurable parameters, supporting remote revision through the manufacturing execution system; When neither the cooling margin preliminary screening nor the isomer cumulative preliminary screening is triggered, the industrial controller will identify the preliminary analysis condition of the current evaluation cycle as a normal condition, maintain routine operation status monitoring, and wait for the next evaluation cycle to re-execute the single index preliminary screening. Otherwise, the industrial controller will identify the preliminary analysis condition of the current evaluation cycle as the condition of interest and write a trigger record of the condition of interest in the real-time database, which includes the trigger timestamp, the trigger indicator type and the indicator value at the trigger time. It should be noted that the index represents two core parameters: cooling capacity saturation and four-dimensional demethylchlortetracycline concentration prediction values, which are extracted by the industrial controller from the operating feature vector of the back-extraction section and used for threshold comparison at each level. If the focus is on operating conditions, the industrial controller enters the dual-indicator concurrent risk analysis stage; Specifically, the industrial controller extracts the current cooling capacity saturation from the thermal safety margin and compares it with the preset second-level cooling conflict warning line; in parallel, it extracts the current four-dimensional demethylchlortetracycline concentration prediction value from the impurity status index and compares it with the preset second-level isomer conflict warning line. It should be noted that the second-level cooling conflict warning line is higher than the first-level cooling warning line, and the second-level isomer conflict warning line is higher than the first-level isomer warning line. Both were pre-calibrated through offline experiments and statistical analysis of historical batch data. If the cooling capacity saturation does not exceed the second-level cooling conflict warning line, or the predicted value of the four-dimensional demethylchlortetracycline concentration does not exceed the second-level isomer conflict warning line, the industrial controller determines that no dual-indicator concurrent risk has been formed, and the final analysis condition of the current assessment cycle is determined as the condition of concern. Under monitored operating conditions, the industrial controller performs enhanced monitoring operations: the sampling period of the Raman spectrometer is shortened from the usual preset value to a higher monitoring frequency, accumulating data with higher temporal resolution for subsequent dual-indicator concurrent risk analysis; It should be noted that the working condition being monitored is the lowest risk level output. The purpose is to obtain more comprehensive situational awareness information through encrypted monitoring and trend tracking without triggering any control intervention. If the cooling capacity saturation exceeds the second-level cooling conflict warning line, and the predicted value of the four-dimensional demethylchlortetracycline concentration exceeds the second-level isomer conflict warning line, the industrial controller determines that the risk of concurrent dual indicators is established, determines the final analysis condition of the current assessment cycle as the concurrent critical condition, and enters the condition status determination stage. It should be noted that setting the simultaneous breach of the second-level conflict warning line of each of the two indicators as the trigger condition for concurrent critical conditions is to avoid misjudgment due to the instantaneous fluctuation of a single indicator. The excessively high cooling capacity saturation indicates that the heat exchange system is approaching its limit, and the excessively high predicted value of the four-dimensional demethylchlortetracycline concentration indicates that the cumulative effect of the crystallization mother liquor circulation is significant. The simultaneous occurrence of the two points to a substantial deterioration in the back-extraction conditions. During the condition determination phase, the industrial controller performs a further deterioration assessment on the evaluation cycle that is identified as a concurrent critical condition in order to distinguish the condition status. Specifically, the industrial controller extracts the cooling capacity saturation sequence within the most recent preset evaluation period from the thermal safety margin, calculates the cooling capacity saturation change trend as the rate of increase of cooling capacity saturation; in parallel, it extracts the dehydrated demethylchlortetracycline concentration change rate sequence within the most recent preset evaluation period from the impurity status index, and performs first-order linear fitting using the least squares method, using the slope obtained from the fitting as the generation acceleration of dehydrated demethylchlortetracycline concentration. It should be noted that the rate of increase of cooling capacity saturation characterizes how quickly the heat dissipation capacity of the back-extraction device is depleted, and the rate of increase of dehydrated demethylchlortetracycline concentration characterizes whether the chemical degradation reaction has entered the self-accelerating stage. Both serve as auxiliary indicators for judging whether the system has entered irreversible deterioration. The industrial controller compares the rate of increase of cooling capacity saturation with a preset cooling rate limit criterion, and at the same time compares the generation acceleration of dehydrated demethylchlortetracycline concentration with a preset impurity acceleration limit criterion. It should be noted that both the cooling rate limit criterion and the impurity acceleration limit criterion were calibrated by combining the heat transfer capacity boundary analysis of the back-extraction device with experimental data on the degradation kinetics of demethylchlortetracycline hydrochloride. If the cooling capacity saturation has reached the preset cooling saturation limit, and the generation acceleration of dehydrated demethylchlortetracycline concentration is greater than the impurity acceleration limit criterion, the industrial controller will determine the current evaluation cycle's operating condition as an irreversible cooling failure state, and write a highest-level operating condition trigger record into the real-time database, including the trigger timestamp, current cooling capacity saturation, cooling capacity saturation rise rate, current predicted dehydrated demethylchlortetracycline concentration, generation acceleration of dehydrated demethylchlortetracycline concentration, and current predicted four-dimensional demethylchlortetracycline concentration. It should be noted that the irreversible cooling failure state indicates that the heat dissipation capacity of the back-extraction device has been completely exhausted and cannot be restored by adjusting the flow rate of the cooling medium. At the same time, the concentration of dehydrated demethylchlortetracycline is rising rapidly, indicating that the chemical degradation reaction has entered the self-accelerating stage. If not intervened immediately, it will lead to irreversible deterioration of the quality of the entire batch of back-extraction solution. If the cooling capacity saturation does not reach the cooling saturation limit, or the generation acceleration of dehydrated demethylchlortetracycline concentration does not exceed the impurity acceleration limit criterion, the industrial controller will determine the current operating condition of the current evaluation cycle as a concurrent critical operating condition preparation state, maintain the encrypted monitoring frequency and continuously repeat the auxiliary indicator evaluation of the operating condition determination stage until the operating condition is upgraded to an irreversible cooling failure state or the risk is eliminated. It should be noted that the concurrent critical condition standby state indicates that the system has entered the critical zone of dual-indicator conflict, but the degree of deterioration has not yet reached the irreversible stage, and intervention and rescue can still be carried out through safe mode switching; S4: Based on the final analysis of the working condition and the working condition status, the decision logic is matched, and the adjustment control command group is output according to the matched decision logic. The decision logic includes quality constraint adaptive optimization, safety mode switching based on dual index conflict, and quality-driven online material path physical isolation. Specifically, after receiving the final analysis working condition and working condition status output, the industrial controller reads the decision logic that matches the current final analysis working condition and working condition status from the preset decision rule base, and synchronously sends the generated adjustment control command group to each actuator through the fieldbus for execution within the same evaluation cycle, driving the physical state of the back-extraction section to be adjusted in accordance with the risk level. The decision rule base is stored in the non-volatile memory of the industrial controller in the form of a lookup table. Each rule consists of the final analysis condition identifier, condition status identifier, decision logic name, actuator target status and execution priority, and supports online updates through the manufacturing execution system. Specifically, when the final analysis condition is the condition of interest, the industrial controller uses adaptive optimization to match quality constraints as the current decision-making logic; It should be noted that the core feature of quality-constrained adaptive optimization is that the predicted concentration of demethylchlortetracycline in the four-way difference of the impurity situation index is embedded as a quality constraint term into the multi-objective optimization process. This enables the control system to automatically adjust the discharge ratio of crystallization mother liquor in the direction of diluting impurities while maintaining the optimization of yield and energy consumption, thereby achieving quality adaptive adjustment under optimization guidance. Under the decision logic of quality-constrained adaptive optimization, the industrial controller generates the following set of adjustment and control instructions: Specifically, the industrial controller uses the ratio of the circulating flow rate of the crystallization mother liquor to the feed flow rate of the fresh hydrochloric acid aqueous solution as the independent variable to establish an optimization problem with the two optimization objectives of maximizing product yield and minimizing energy consumption. The product yield is determined by the proportion of the recovered amount of demethylchlortetracycline hydrochloride in the total aqueous phase flow rate entering the back-extraction unit to the circulating flow rate of the crystallization mother liquor. The total aqueous phase flow rate entering the back-extraction unit is obtained by summing the fresh hydrochloric acid aqueous solution feed flow rate and the crystallization mother liquor circulating flow rate. The fresh hydrochloric acid aqueous solution feed flow rate is collected in real time by an electromagnetic flow meter installed on the fresh hydrochloric acid aqueous solution supply pipeline, and the crystallization mother liquor circulating flow rate is collected in real time by an electromagnetic flow meter installed on the crystallization mother liquor circulating pipeline. The energy consumption consists of the heating medium consumption of the heating and heat preservation device and the cooling medium consumption of the jacket of the back extraction device. Specifically, the heating medium consumption is calculated in real time by installing temperature sensors and flow meters on the inlet and outlet pipes of the heating medium in the heating and insulation device to obtain the volumetric flow rate, inlet temperature, and outlet temperature of the heating medium. This data is then combined with the pre-calibrated specific heat capacity of the heating medium. Similarly, the jacket cooling medium consumption is calculated in real time by installing temperature sensors and flow meters on the inlet and outlet pipes of the jacket cooling medium to obtain the inlet temperature, outlet temperature, and volumetric flow rate of the cooling medium. This data is then combined with the pre-calibrated specific heat capacity of the cooling medium. It should be noted that the higher the proportion of the mother liquor circulation flow rate to the total aqueous phase flow rate, the more hydrochloric acid demethylchlortetracycline is returned to the back-extraction section, the higher the total product yield, the larger the fresh hydrochloric acid aqueous solution feed flow rate, the larger the total aqueous phase flow rate entering the back-extraction unit, and the higher the energy consumption for subsequent heating and cooling. An optimization objective function is constructed. Specifically, the product yield and energy consumption are normalized separately. Energy consumption is normalized by forward normalization, which is the ratio of the current energy consumption to a preset energy consumption reference upper limit. The closer the normalized energy consumption value is to zero, the lower the energy consumption. The energy consumption reference upper limit is calibrated based on the sum of the rated power of the heating medium of the heating and insulation device and the rated power of the jacket cooling medium of the back extraction device. Product yield is normalized by reverse normalization, which is the ratio of the difference between the preset yield reference upper limit and the current product yield to the yield reference upper limit. The closer the yield deviation is to zero, the closer the current yield is to the optimal level. The yield reference upper limit is determined based on the historical batch optimal yield statistics of the demethylchlortetracycline hydrochloride extraction line. The predicted concentration of tetra-differential demethylchlortetracycline at the current moment is extracted from the impurity status index. The ratio of the predicted concentration of tetra-differential demethylchlortetracycline to the preset upper limit of the isomer concentration is processed to obtain the normalized value of the isomer concentration. The penalty term is constructed with the normalized value of the isomer concentration as the independent variable. The penalty term is constructed by using the product of the normalized value of the isomer concentration and the preset penalty ratio coefficient as the base value of the penalty term. If the normalized value of the isomer concentration is lower than the preset penalty activation threshold, the penalty term is set to zero. Otherwise, the penalty ratio coefficient increases linearly with the increase of the normalized value of the isomer concentration. It should be noted that the upper limit of the isomer concentration reference is determined by back-calculating the safe limit of the corresponding tetra-differential demethylchlortetracycline concentration in the mixed liquid of the back-extraction section based on the upper limit of the allowable content of tetra-differential demethylchlortetracycline in the finished product of demethylchlortetracycline hydrochloride. The objective function is obtained by weighted fusion of the normalized energy consumption value and the yield deviation and summing with the penalty term. The weight coefficients are pre-calibrated through offline experiments and statistical analysis of historical batch data while satisfying the normalization constraints, and are stored in the industrial controller as configurable parameters. It supports remote revision through the manufacturing execution system. The goal of the optimization solution is to minimize the objective function. The industrial controller uses an online rolling optimization method to solve the objective function; Specifically, within each evaluation cycle, the industrial controller reads the impurity status index at the current moment, obtains the predicted value of the four-dimensional demethylchlortetracycline concentration, recalculates the penalty term and updates the optimization objective function, and uses the sequential quadratic programming method to solve the optimization objective function to obtain the optimal ratio of the crystallization mother liquor circulation flow rate to the fresh hydrochloric acid aqueous solution feed flow rate that minimizes the optimization objective function within the current evaluation cycle. This ratio is denoted as the optimal ratio. The industrial controller decomposes the optimal ratio obtained by solving into two operational variables: the crystallization mother liquor discharge flow rate and the fresh hydrochloric acid aqueous solution replenishment flow rate. The crystallization mother liquor discharge flow rate is equal to the difference between the current crystallization mother liquor circulation flow rate and the optimal crystallization mother liquor circulation flow rate, and the fresh hydrochloric acid aqueous solution replenishment flow rate is equal to the optimal fresh hydrochloric acid aqueous solution feed flow rate. The flow rate of the mother liquor discharge is converted into the opening command of the electromagnetic discharge valve installed on the mother liquor circulation pipeline, and the flow rate of the fresh hydrochloric acid aqueous solution replenishment is converted into the speed command of the controlled replenishment pump installed on the fresh hydrochloric acid aqueous solution supply pipeline. The resulting set of adjustment and control commands is then obtained. It should be noted that while recycling the mother liquor from crystallization is beneficial for increasing the overall product yield, it continuously carries tetrachlorophenoxyacetic acid back to the back-extraction aqueous phase, causing impurity accumulation. While replenishing with fresh hydrochloric acid can dilute the impurity concentration, excessive replenishment increases wastewater discharge and subsequent wastewater treatment energy consumption. Introducing the predicted concentration of tetrachlorophenoxyacetic acid as a penalty term into the optimization objective function ensures that when the impurity concentration is within a safe range, the penalty term remains inactive, and the optimization result tends to maintain a high proportion of mother liquor recycling to maximize yield; in this case, the system prioritizes economic efficiency. When the impurity concentration increases and exceeds the penalty activation threshold, the penalty term is activated and increases with further increases in impurity concentration. The optimization result automatically shifts towards increasing the proportion of mother liquor discharge and increasing the flow rate of fresh hydrochloric acid replenishment; in this case, the system prioritizes quality and safety. It should be noted that this approach of embedding quality constraints into the optimization of economic objectives achieves an online dynamic balance between yield, energy consumption, and impurity control, avoiding the rigid switching between one or the other in traditional fixed threshold control, and also avoiding batch quality fluctuations caused by the lag in manual intervention. like Figure 3 As shown, Figure 3 The iterative convergence curves for adaptive optimization of quality constraints under different impurity loads demonstrate the iterative convergence process of the industrial controller using sequential quadratic programming (SQP) to solve the objective function online under different impurity loads. The horizontal axis represents the number of optimization iterations, and the vertical axis represents the normalized objective function value. The three curves correspond to the optimization convergence trajectories of tetradextrin (ETCM) in the mixed solution at three different concentration levels: safe (6.5 μg / mL), transitional (22.0 μg / mL), and hazardous (55.0 μg / mL). It can be seen that... Regardless of the initial ETCM concentration and its associated safety risk level, the optimization algorithm can converge to the minimum value within 5 to 7 iterations, indicating that the multi-objective optimization strategy has good real-time response capability. As the ETCM concentration increases, the penalty term is activated, and the basic convergence value of the objective function increases significantly. This means that while pursuing low energy consumption and high yield, the control system must actively adjust the ratio of the discharge flow rate of the crystallization mother liquor to the replenishment flow rate of the fresh hydrochloric acid aqueous solution to dilute impurities. Ultimately, this reflects the core of quality-constrained adaptive optimization, which is to automatically seek a dynamic balance between economic indicators and quality safety under different impurity loads. When the final analysis condition is a concurrent critical condition and the condition status is a concurrent critical condition preparation state, the industrial controller matches the safety mode switching based on the dual-index conflict as the current decision logic. It should be noted that the core feature of the safety mode switching based on dual-index conflict is to establish that the cooling capacity saturation and the four-dimensional detechlorin concentration prediction value simultaneously break through their respective second-level conflict warning lines as the trigger criterion for switching from the optimization mode to the safety mode. The set of rigid safety actions executed after the switch constitutes a set of preset safety modes that do not require online solution, realizing a deterministic transition from computational optimization to rule-based enforcement. Under the decision logic based on the safety mode switching of dual-index conflict, the industrial controller immediately suspends the optimization process of quality constraint adaptive optimization and generates the following set of adjustment control instructions: For the controlled replenishment pump, the industrial controller forces the flow rate to operate at a preset minimum safe flow rate and no longer responds to the output of the optimization objective function solution, ensuring that the back-extraction unit still has a basic supply of fresh hydrochloric acid aqueous solution to maintain a minimum extraction driving force; For the electric regulating valve on the jacket cooling medium pipeline, the industrial controller forces its opening command to the fully open state, calls all the remaining cooling medium flow to dissipate heat from the back-extraction unit, and slows down the accelerated generation of tetra-differential demethylchlortetracycline caused by the temperature rise. For the electromagnetic discharge valve on the crystallization mother liquor circulation pipeline, the industrial controller increases its opening at the maximum allowable rate to the preset high discharge opening, and discharges the circulating mother liquor rich in tetradextrin at the preset maximum discharge flow rate, thereby rapidly reducing the cumulative concentration of tetradextrin in the back-extraction aqueous phase. Based on the minimum safe flow rate, the industrial controller synchronously increases the replenishment flow rate of the controlled replenishment pump according to the mother liquor flow rate discharged by the electromagnetic discharge valve to make an equal amount of replenishment. That is, the replenishment flow rate is composed of the minimum safe flow rate and the compensation increment of the discharged mother liquor flow rate to maintain the balance of liquid level and flow rate in the back-extraction unit. It should be noted that the minimum safe flow rate, maximum discharge flow rate, and high-level discharge opening are all pre-calibrated based on the design volume of the back-extraction unit, the lower limit of the mass transfer residence time, and the pressure limit of the pipeline, and are stored in the industrial controller as read-only parameters. It should be noted that the decision logic for safety mode switching based on dual-index conflict is used in special scenarios where the back-extraction section is judged to be in a state of readiness for concurrent critical operation. By rigidly anchoring the hydrochloric acid feed flow rate to the minimum safe flow rate and simultaneously executing the maximum discharge of the mother liquor discharge valve and the equal amount of replenishment by the replenishment pump, it solves the control conflict problem of two completely opposite directions of adjustment for the same operating variable: the need to reduce the flow rate to shorten the high-temperature residence time to suppress dehydration demethylchlortetracycline and the need to increase the flow rate to accelerate the replacement of mother liquor to discharge four differential demethylchlortetracycline. When the final analysis condition is a concurrent critical condition and the condition status is an irreversible cooling failure, the industrial controller uses the quality-driven online material path physical isolation as the current decision logic. It should be noted that the core feature of quality-driven online material path physical isolation is that when the system determines that thermal failure and chemical degradation have entered the irreversible self-accelerating stage, it does not rely on any adjustment means to salvage the situation. Instead, it uses a three-way switching valve to strip the back-extraction liquid from the normal downstream production line to the temporary storage tank in real time, and simultaneously interlocks to cut off the feed of the upstream extractor, thus completing the physical reconstruction of the process material path and supply energy, and extending the emergency shutdown concept from equipment safety protection to quality safety protection. Under the quality-driven decision logic of physical isolation of online material paths, the industrial controller, while maintaining the continued effectiveness of all regulation and control commands generated by the safety mode switching based on dual-index conflict, additionally generates the following set of regulation and control commands: The industrial controller sends a switching command to the three-way switching valve installed on the discharge pipeline of the back-extraction unit, switching the flow direction of the three-way switching valve from the normal downstream pipeline connected to the decolorization kettle to the isolation pipeline connected to the temporary storage tank, so that all the back-extraction liquid produced thereafter will no longer enter the decolorization kettle and subsequent processes, but will be completely diverted to the temporary storage tank for isolation and sealing. The industrial controller generates an interlock pause signal, which is sent to the feed pump of the aqueous phase to be extracted and the feed pump of the extractant in the multi-stage countercurrent centrifugal extractor via the fieldbus. This forces the supply of organic extract phase and aqueous phase to be extracted to the back-extraction unit to stop, thus cutting off the material input for the continued generation and accumulation of dehydrated demethylchlortetracycline and tetra-differential demethylchlortetracycline from the source. It should be noted that the isolated back-extraction solution in the temporary storage tank is sampled and tested by quality management personnel after the batch ends. It is then determined whether to be recycled or discarded based on the impurity content. The management rules are defined by the batch disposal strategy issued by the manufacturing execution system. All adjustment and control command groups are issued and executed synchronously within the same evaluation cycle in which the final analysis working condition and working condition status are calibrated. The industrial controller continuously monitors the actual opening feedback of the electromagnetic discharge valve, the actual flow feedback of the controlled replenishment pump, the actual opening feedback of the electric regulating valve, the actual valve position feedback of the three-way switching valve, and the start and stop status feedback of the water phase feed pump and the extractant feed pump to confirm that each command has been fully executed. If any actuator fails to operate as instructed, the industrial controller immediately pushes an execution fault alarm to the operator station and designated mobile terminal via the wireless communication module. It should be noted that from quality-constrained adaptive optimization to safety mode switching based on dual-index conflict, and then to quality-driven online material path physical isolation, the entire decision logic matching and adjustment control command group issuance process is automatically triggered and executed by the industrial controller based on the final analysis of the working conditions and working conditions, while retaining the operator's authority to remotely intervene in emergencies through the manufacturing execution system. The technical solution of this invention is as follows: An online monitoring system integrating the extraction process flow and Raman spectroscopy monitoring is established to form a holographic perception closed-loop production line. For the holographic perception closed-loop production line, the online monitoring system acquires the operating characteristic vector of the back-extraction section in real time. Based on the operating characteristic vector of the back-extraction section, a single-index preliminary screening, including cooling margin preliminary screening and isomer accumulation preliminary screening, is performed to identify the preliminary analysis conditions. The preliminary analysis conditions include normal conditions and conditions of concern. If it is a condition of concern, a dual-index concurrent risk analysis is performed to determine the final analysis condition and execute the condition status judgment. The condition status includes a concurrent critical condition preparation state and an irreversible cooling failure state. Based on the final analysis condition and the condition status, a decision logic matching is performed, and an adjustment control command group is output according to the matched decision logic. The decision logic includes quality constraint adaptive optimization, safety mode switching based on dual-index conflict, and quality-driven online material path physical isolation.

[0026] The foregoing has provided a detailed description of one embodiment of the present invention, but this description is merely a preferred embodiment and should not be construed as limiting the scope of the invention. All equivalent variations and modifications made within the scope of the present invention should still fall within the scope of the present invention.

Claims

1. An extraction process for efficiently extracting demethylchlortetracycline hydrochloride from fermentation broth, characterized in that: Includes the following steps: An online monitoring system integrating extraction process flow and Raman spectroscopy monitoring was established to form a holographic perception closed-loop production line; For the holographic perception closed-loop production line, the operation characteristic vector of the back-extraction section is obtained in real time through the online monitoring system, including the impurity status index, thermal safety margin and temperature field distribution characteristics. Based on the operating characteristic vector of the back-extraction section, a single-index preliminary screening including cooling margin preliminary screening and isomer accumulation preliminary screening is performed to identify the preliminary analysis conditions. The preliminary analysis conditions include normal conditions and conditions of concern. If it is a condition of concern, a dual-index concurrent risk analysis is performed to determine the final analysis conditions and to perform condition status determination. The condition status includes the concurrent critical condition preparation state and the irreversible cooling failure state. Based on the final analysis of the working conditions and the matching of the working conditions, the system outputs a set of adjustment and control instructions according to the matched decision logic. The decision logic includes quality constraint adaptive optimization, safety mode switching based on dual index conflict, and quality-driven online material path physical isolation.

2. The extraction process for efficiently extracting demethylchlortetracycline hydrochloride from fermentation broth according to claim 1, characterized in that: The method for forming a holographic perception closed-loop production line is as follows: A demethylchlortetracycline hydrochloride extraction production line was constructed, including an acidification tank, a heating and insulation device, a plate and frame filter press, a filtrate return and salting-out agent dosing device, a multi-stage countercurrent centrifugal extractor, a back-extraction device, a decolorization kettle, a filtration and carbon removal device, a crystallization kettle, a centrifuge, a washing device, and a vacuum dryer, which are connected in sequence through process pipelines. The extractant is a mixed solvent. The aqueous phase inlet of the back-extraction device is connected to the fresh hydrochloric acid aqueous solution supply pipeline and the crystallization mother liquor circulation pipeline through a three-way confluence pipeline. An optical fiber flow cell is connected in series on the pipeline where the discharge pipeline of the back-extraction unit and the crystallization mother liquor circulation pipeline converge. The optical fiber flow cell is connected to a Raman spectrometer with a built-in quantitative model of dehydrated demethylchlortetracycline and four-dimensional demethylchlortetracycline via optical fiber. Temperature sensing networks and flow sensing networks are deployed inside the back-extraction unit and on the discharge pipeline, the jacket cooling medium pipeline, the fresh hydrochloric acid aqueous solution supply pipeline, and the crystallization mother liquor circulation pipeline, respectively. The online monitoring system is connected to the industrial controller via fieldbus.

3. The extraction process for efficiently extracting demethylchlortetracycline hydrochloride from fermentation broth according to claim 2, characterized in that: The impurity state index is obtained as follows: The mixed liquid flowing through the optical fiber flow cell is excited in real time and the original spectrum is obtained by performing spectral acquisition through a Raman spectrometer. In each sampling cycle, the Raman spectrometer performs preprocessing and concentration prediction on the original spectrum through the built-in dehydrated demethylchlortetracycline and tetradextrin combined quantitative model, and simultaneously outputs the predicted concentration values ​​of dehydrated demethylchlortetracycline and tetradextrin in the mixed liquid, and sorts them to obtain the impurity concentration level. The concentration prediction sequences of dehydrated demethylchlortetracycline and four-dimensional demethylchlortetracycline from the most recent preset number of sampling periods are extracted from the real-time database and processed separately to obtain the concentration change rate of dehydrated demethylchlortetracycline and the concentration change rate of four-dimensional demethylchlortetracycline. These are integrated and recorded as the impurity change trend. The impurity situation index is composed of the impurity concentration level and the impurity change trend.

4. The extraction process for efficiently extracting demethylchlortetracycline hydrochloride from fermentation broth according to claim 2, characterized in that: The methods for obtaining thermal safety margin and temperature field distribution characteristics are as follows: The acquisition of thermal safety margin includes: continuously acquiring the inlet temperature, outlet temperature and volumetric flow rate of the cooling medium through temperature sensors and flow meters on the jacketed cooling medium pipeline; processing the data in conjunction with the pre-calibrated specific heat capacity of the cooling medium to obtain the current actual heat exchange; and processing the current actual heat exchange with the pre-stored rated maximum heat exchange to obtain the cooling capacity saturation. Extract the cooling capacity saturation sequence of the most recent preset number of sampling periods from the real-time database, process the cooling capacity saturation sequence to obtain the cooling capacity saturation change trend, and the thermal safety margin is composed of the cooling capacity saturation and the cooling capacity saturation change trend. The acquisition of temperature field distribution characteristics includes: obtaining the material temperature at the top, middle and bottom of the back-extraction unit through temperature sensors installed at the top, middle and bottom positions along the material flow direction inside the back-extraction unit; obtaining the back-extraction liquid outlet temperature through temperature sensors on the back-extraction unit's discharge pipeline; and integrating these to obtain the temperature field distribution characteristics.

5. The extraction process for efficiently extracting demethylchlortetracycline hydrochloride from fermentation broth according to claim 1, characterized in that: The method for identifying preliminary analysis conditions is as follows: The evaluation cycle is defined, and the execution of the initial screening of cooling margin includes: extracting the cooling capacity saturation at the current moment from the thermal safety margin of the back-extraction section's operating characteristic vector; if it exceeds the first-level cooling warning line, the initial screening of cooling margin is marked as triggered. The execution of the initial screening of isomer accumulation includes: extracting the predicted value of the four-dimensional demethylchlortetracycline concentration at the current moment from the impurity status index of the back-extraction section's operating characteristic vector; if it exceeds the first-level isomer warning line, the initial screening of isomer accumulation is marked as triggered. If neither the cooling margin screening nor the isomer accumulation screening is triggered, the preliminary analysis condition of the current evaluation cycle will be identified as a normal condition; otherwise, it will be identified as a condition of concern.

6. The extraction process for efficiently extracting demethylchlortetracycline hydrochloride from fermentation broth according to claim 5, characterized in that: The method for determining the final analysis conditions is as follows: If the initial analysis indicates that the operating condition is of concern, the system will enter the dual-indicator concurrent risk analysis stage. The cooling capacity saturation at the current moment will be extracted from the thermal safety margin and compared with the preset second-level cooling conflict warning line. At the same time, the predicted value of the four-dimensional demethylchlortetracycline concentration at the current moment will be extracted from the impurity status index and compared with the preset second-level isomer conflict warning line. If the cooling capacity saturation does not exceed the second-level cooling conflict warning line, or the predicted value of the four-dimensional demethylchlortetracycline concentration does not exceed the second-level isomer conflict warning line, the final analysis condition of the current evaluation cycle will be identified as a condition of concern, and the sampling cycle of the Raman spectrometer will be shortened to the encrypted monitoring frequency; otherwise, it will be identified as a concurrent critical condition.

7. The extraction process for efficiently extracting demethylchlortetracycline hydrochloride from fermentation broth according to claim 6, characterized in that: The method for determining the operating condition status is as follows: The cooling capacity saturation sequence within the most recent preset evaluation period is extracted from the thermal safety margin. The cooling capacity saturation sequence is processed to obtain the rate of increase of cooling capacity saturation. The dehydrated demethylchlortetracycline concentration change rate sequence within the most recent preset evaluation period is extracted from the impurity status index. The dehydrated demethylchlortetracycline concentration change rate sequence is processed to obtain the acceleration of dehydrated demethylchlortetracycline concentration generation. If the cooling capacity saturation has reached the preset cooling saturation limit, and the generation acceleration of dehydrated methylchlortetracycline concentration is greater than the preset impurity acceleration limit criterion, the current evaluation cycle's operating condition will be judged as an irreversible cooling failure state; otherwise, it will be judged as a preparatory state for concurrent critical operating conditions.

8. The extraction process for efficiently extracting demethylchlortetracycline hydrochloride from fermentation broth according to claim 1, characterized in that: The method for executing decision logic matching and outputting adjustment and control command groups is as follows: When the final analysis condition is the condition of interest, the matching quality constraint adaptive optimization is used as the current decision logic to generate a set of adjustment and control instructions; When the final analysis condition is a concurrent critical condition and the condition status is a concurrent critical condition preparation state, the safety mode switching based on dual-index conflict is matched as the current decision logic. The optimization solution process of quality constraint adaptive optimization is immediately suspended. The flow rate of the controlled replenishing pump is forcibly limited to the preset minimum safe flow rate. The opening command of the electric regulating valve on the jacket cooling medium pipeline is forcibly set to the fully open state. The opening of the electromagnetic discharge valve on the crystallization mother liquor circulation pipeline is increased to the preset high discharge opening at the maximum allowable rate. The circulating mother liquor is discharged at the preset maximum discharge flow rate. Based on the minimum safe flow rate, the controlled replenishing pump increases the replenishing flow rate synchronously according to the mother liquor flow rate discharged by the electromagnetic discharge valve to make equal replacement. The adjustment command group is then obtained. When the final analysis condition is a concurrent critical condition and the condition status is an irreversible cooling failure, the physical isolation of the online material path driven by the matching quality is used as the current decision logic. On the basis of maintaining the adjustment command group generated by the safety mode switching based on the dual-index conflict, a switching command is issued to the three-way switching valve on the discharge pipeline of the back-extraction unit. The flow direction of the three-way switching valve is switched from the normal downstream pipeline connected to the decolorization kettle to the isolation pipeline connected to the temporary storage tank. An interlock pause signal is generated and sent to the feed pump of the aqueous phase to be extracted and the feed pump of the extractant of the multi-stage countercurrent centrifugal extractor through the fieldbus. The supply of organic extract phase and aqueous phase to be extracted to the back-extraction unit is forcibly stopped, and the adjustment command group is obtained.

9. The extraction process for efficiently extracting demethylchlortetracycline hydrochloride from fermentation broth according to claim 8, characterized in that: The method for generating the adjustment and control instruction set under the decision logic of quality constraint adaptive optimization is as follows: Using the ratio of the circulating flow rate of the crystallization mother liquor to the feed flow rate of fresh hydrochloric acid aqueous solution as the independent variable, an optimization problem is established with the two optimization objectives of maximizing product yield and minimizing energy consumption. Construct an optimization objective function: use reverse normalization on the product yield to obtain the yield deviation, and use forward normalization on the energy consumption to obtain the energy consumption normalized value; The predicted concentration values ​​of the four-dimensional detechlortetracycline at the current moment are extracted from the impurity status index and processed to obtain the normalized value of the isomer concentration. The penalty term is constructed with the normalized value of the isomer concentration as the independent variable. The penalty term is set to zero when the normalized value of the isomer concentration is lower than the preset penalty activation threshold. The objective function is obtained by summing the weighted fusion value of the normalized energy consumption value and the yield deviation, plus the penalty term; The objective function is solved online in a rolling manner to obtain the optimal ratio of the circulating flow rate of the crystallization mother liquor to the feed flow rate of the fresh hydrochloric acid aqueous solution within the current evaluation period, which minimizes the objective function. The optimal ratio is then processed to obtain two operational variables: the discharge flow rate of the crystallization mother liquor and the replenishment flow rate of the fresh hydrochloric acid aqueous solution. The discharge flow rate of the crystallization mother liquor is converted into the opening command of the electromagnetic discharge valve on the crystallization mother liquor circulation pipeline, and the replenishment flow rate of the fresh hydrochloric acid aqueous solution is converted into the speed command of the controlled replenishment pump on the fresh hydrochloric acid aqueous solution supply pipeline. The resulting set of adjustment and control commands is then obtained.

10. The extraction process for efficiently extracting demethylchlortetracycline hydrochloride from fermentation broth according to claim 9, characterized in that: The product yield is determined by the proportion of the recovered amount of demethylchlortetracycline hydrochloride in the total aqueous phase flow rate entering the back-extraction unit to the circulating flow rate of the crystallization mother liquor. The total aqueous phase flow rate entering the back-extraction unit is obtained by summing the fresh hydrochloric acid aqueous solution feed flow rate and the crystallization mother liquor circulating flow rate. The fresh hydrochloric acid aqueous solution feed flow rate and the crystallization mother liquor circulating flow rate are collected in real time by electromagnetic flow meters installed on the corresponding pipelines. The energy consumption consists of the heating medium consumption of the heating and insulation device and the cooling medium consumption of the jacket of the back-extraction unit. The heating medium consumption and the cooling medium consumption of the jacket are collected in real time by temperature sensors and flow meters installed on the corresponding pipelines and the data is processed.