Low-temperature extraction and co-production process for high-value utilization of whole components of calligonum

Abstract

This invention relates to the fields of bio-extraction technology and deep processing technology of agricultural products, and discloses a low-temperature extraction and co-production process for the high-value utilization of all components of Japanese raisin tree. By constructing a three-stage countercurrent extraction system consisting of supercritical CO2 extraction, low-temperature ethanol countercurrent extraction, and low-temperature water enzymatic extraction, the non-polar, moderately polar, and strongly polar active components in Japanese raisin tree are systematically separated. On this basis, it further combines ultrafiltration membrane molecular retention and fractionation, macroporous adsorption resin directional purification, and extraction and fermentation conversion of dietary fiber from the extraction residue, and co-produces Japanese raisin tree seed oil, neutral / acidic polysaccharides, total flavonoids, total saponins, high-purity polyphenols, SOD enzyme powder, dihydromyricetin, natural flavor, soluble / insoluble dietary fiber, Japanese raisin tree nutritional syrup, and low-alcohol fermented beverage. The residue generated in the entire process is made into organic fertilizer, and the process water is completely reused after reverse osmosis treatment, ultimately achieving the circular economy goal of high-value co-production of all components of Japanese raisin tree and near-zero waste.

Description

Technical Field

[0001] This invention relates to the fields of bio-extraction technology and deep processing technology of agricultural products, specifically a low-temperature extraction and co-production process for the high-value utilization of all components of Japanese raisin tree. Background Technology

[0002] Japanese raisin tree (Hovenia dulcis Thunb.), also known as Chinese raisin tree, is a plant belonging to the genus Hovenia in the family Rhamnaceae. Its fleshy fruit stalks and seeds are rich in various bioactive components such as polysaccharides, flavonoids, saponins, polyphenols, dihydromyricetin, and superoxide dismutase (SOD), possessing multiple health benefits including hangover relief, liver protection, antioxidant effects, and immune regulation. In recent years, with the continuous expansion of the planting area and the year-on-year increase in yield, the deep processing and high-value utilization of Japanese raisin tree resources have become an important direction for industrial development.

[0003] Currently, the processing and utilization of Japanese raisin tree (Hovenia dulcis) faces several technical bottlenecks: First, the extraction process is crude, resulting in significant loss of active ingredients. Existing methods for extracting active ingredients from Japanese raisin tree mainly employ traditional hot reflux extraction, decoction, or ultrasound-assisted extraction. These processes typically involve high extraction temperatures (50-90℃), while SOD, dihydromyricetin, and natural flavorings in Japanese raisin tree are heat-sensitive components, easily leading to denaturation, degradation, or volatilization loss due to high-temperature treatment. Furthermore, traditional processes often use a single solvent for single-stage extraction, making it difficult to achieve systematic dissolution of active ingredients with different polarities, resulting in a large amount of effective components remaining unused in the residue. Second, the extraction process lacks precise control, leading to poor batch stability. Existing extraction parameters (such as stirring speed, solvent flow rate, and extraction time) are mostly set based on experience, lacking real-time monitoring and feedback adjustment methods for the flow field, concentration field, and mass transfer efficiency within the extraction tower. Because the raw materials for *Hovenia dulcis* vary significantly due to factors such as origin, year, and maturity, fixed-parameter operating models are difficult to adapt to raw material fluctuations, resulting in large differences in product yield and quality between different batches, severely restricting the stability and controllability of industrial production. Thirdly, resource utilization is low, and by-product waste is severe. Current *Hovenia dulcis* processing mainly focuses on single products such as *Hovenia dulcis* wine, juice, and vinegar. The residue after extracting active ingredients is usually discarded directly as waste or used only as low-value-added feed. The valuable resources in *Hovenia dulcis* residue, rich in dietary fiber, residual sugars, and organic matter, are not fully recovered and utilized, causing not only a huge waste of resources but also increasing the environmental burden and economic cost of solid waste treatment. Fourthly, there is a lack of systematic co-production design and economic optimization. Existing research mainly focuses on optimizing the extraction process of single active ingredients, lacking a systematic co-production process design from the perspective of full component utilization. Meanwhile, the optimization of extraction processes largely relies on traditional methods such as single-factor or orthogonal experiments, and has not yet introduced multi-objective intelligent optimization algorithms to collaboratively optimize multiple indicators such as product yield, purity, energy consumption, and cost, making it difficult to achieve the optimal balance between economic benefits and resource utilization efficiency.

[0004] Therefore, there is an urgent need to develop a deep processing technology for Japanese raisin tree that integrates low-temperature activity preservation, full-component tiered utilization, and intelligent and precise control, in order to break through the above-mentioned technical bottlenecks and achieve efficient, high-value, and clean comprehensive utilization of Japanese raisin tree resources. Summary of the Invention

[0005] (a) Technical problems to be solved

[0006] To address the shortcomings of existing technologies, this invention provides a low-temperature extraction and co-production process for the high-value utilization of all components of Japanese raisin tree. This process has the advantages of tiered utilization of all components, preservation of low-temperature activity, and intelligent and precise control. It solves the problems of easy deactivation of heat-sensitive components, low resource utilization, poor batch stability, and lack of systematic co-production design and economic optimization in existing Japanese raisin tree processing technologies.

[0007] (II) Technical Solution

[0008] To achieve the above objectives, the present invention provides the following technical solution: a low-temperature extraction and co-production process for high-value utilization of all components of *Hovenia dulcis*, comprising the following steps:

[0009] S1. Raw material harvesting and low-temperature property database construction: Harvest, sort, clean, dry at low temperature and grade and crush the Japanese raisin tree, and simultaneously establish a complete component property database;

[0010] S2. Low-temperature multi-stage countercurrent extraction: CO2, ethanol and purified water are used in sequence to carry out three-stage countercurrent extraction of jujube at low temperature. At the same time, the velocity field and concentration field in the extraction tower are predicted in real time by CFD model to guide the setting of solvent flow rate and stirring speed.

[0011] S3. Liquid phase fractionation: After centrifugation and microfiltration, the extract mixture is initially separated into active components of different molecular weights by passing it through an ultrafiltration membrane that retains the molecular weight gradient, to obtain fractions F1 to F5.

[0012] S4. Targeted purification: Gradient elution of F2 to F5 fractions was carried out using macroporous adsorption resins of different polarities to prepare high-purity Hovenia dulcis polysaccharides, total flavonoids, total saponins and total polyphenols respectively.

[0013] S5, Low-temperature refined heat-sensitive components: Superoxide dismutase, dihydromyricetin and natural flavorings are extracted from the permeate F5 of S3 and the volatile condensate of the extract using membrane separation, salting out, chromatography and molecular distillation techniques under low-temperature conditions throughout the process.

[0014] S6. Utilization of extraction residue: The residue after S2 extraction and the by-product streams of washing water and permeate generated in S3 to S5 are converted into dietary fiber, jujube syrup and organic fertilizer, respectively, to achieve zero waste of all components.

[0015] S7. Product Value Co-production Assessment: Based on market value, establish a co-product value coefficient model, reasonably allocate total costs, and calculate the return on investment and break-even point under different production capacities.

[0016] S8. Multi-objective optimization: Record all experimental, simulation and financial data from S1 to S7, construct a hybrid agent model based on the data, use the NSGA-II algorithm to simultaneously maximize product yield and purity, minimize energy consumption and cost, and send the optimal operating parameters to the DCS system to achieve closed-loop control.

[0017] Preferably, in step S1, the raw material harvesting and low-temperature property database construction includes:

[0018] S1.1 Harvesting and Maturity Grading: Harvest the Japanese raisin tree jujubes within 15-30 days after the first frost. They are classified into special grade, first grade, and second grade according to the color and plumpness of the fruit stalk. The latitude, longitude, altitude, average daily temperature, and rainfall are recorded for each batch.

[0019] S1.2 Ultrasonic low-temperature cleaning: Use 30-40kHz ultrasonic waves combined with bubbling cleaning. The cleaning solution is softened water with a conductivity of ≤5μS / cm. The cleaning temperature is controlled at 8-12℃ and the time is 5-8min. After cleaning, drain until there is no visible water on the surface.

[0020] S1.3 Gradient freeze drying: After draining, the jujubes are spread on trays and pre-frozen at -40 to -30℃ for 6 to 8 hours; the first drying temperature is -10 to -5℃ and the vacuum degree is 5 to 8 Pa, lasting for 20 to 25 hours; the second drying temperature is 25 to 30℃ and the vacuum degree is 10 to 15 Pa, lasting for 4 to 8 hours, until the final moisture content is ≤5%;

[0021] S1.4 Low-temperature embrittlement and ultrafine classification: After drying, the material is subjected to impact embrittlement and pulverization at -5 to 0℃. It is then passed through 40 mesh, 80 mesh, 120 mesh and 200 mesh standard sieves in sequence to collect four particle size grades of powder, which are numbered F40, F80, F120 and F200 respectively.

[0022] S1.5 Construction of a complete component property database: For each batch of raw materials, the basic physicochemical indicators, the content of characteristic active ingredients and process compatibility parameters are measured, and all data are entered into the complete component property database.

[0023] Preferably, the low-temperature multi-stage countercurrent extraction in S2:

[0024] S2.1 Supercritical CO2 Extraction of Hovenia dulcis Seed Oil: Take F40-F80 grade powder from S1, load it into a supercritical extraction vessel, introduce food-grade CO2, set the pressure to 25-30 MPa, temperature to 35-38℃, CO2 flow rate to 3-5 L / min, and extraction time to 100-150 min. After pressure reduction and separation in a separation vessel at 6-8 MPa and 35-40℃, collect the Hovenia dulcis seed oil precipitated in the separation vessel. The residue after extraction is retained as raw material for the next stage of extraction.

[0025] S2.2 Low-temperature ethanol countercurrent extraction: The residue after supercritical extraction is transferred to a heat-insulated jacketed countercurrent extraction tower. The extraction tower has a built-in paddle stirrer with a speed of 80-100 rpm. 50%-55% edible ethanol is used as the solvent. The material-liquid ratio is set to 1:15-1:25, the extraction temperature is 20-30℃, and the residence time is 60-120 min. After solid-liquid separation, extract A rich in flavonoids and saponins is collected. The residue after extraction is reserved as raw material for the next stage of extraction.

[0026] S2.3 Low-temperature water + compound enzyme-assisted extraction: Add deionized water to the residue after ethanol extraction at a material-to-liquid ratio of 1:20 to 1:30, stir and extract at 15 to 25°C for 50 to 60 minutes, and add 0.2% to 0.5% cellulase and 0.1% to 0.3% pectinase at the same time. Adjust the pH to 6.0 to 6.8, and then briefly raise the temperature to 40 to 50°C for enzymatic hydrolysis for 60 to 70 minutes. After enzymatic hydrolysis, filter and collect the extract B rich in polysaccharides and proteins. The remaining residue is sent to section S6 for comprehensive utilization.

[0027] S2.4, CFD Real-Time Simulation and Parameter Adaptation: Based on the property database of S1, a three-dimensional rotating blade-porous medium coupled model is established, and input... , The particle size distribution, solvent viscosity, and diffusion coefficient of the powder are used to simulate the turbulent kinetic energy and local mass transfer coefficient in the extraction tower through finite element numerical solution. Based on this, the stirring speed and feed rate in step S2.2 are adjusted online to ensure that the extraction rate fluctuation is ≤5%.

[0028] Preferably, the parameters in S2.4 are adaptive: the stirring speed and feed rate in step S2.2 are adjusted online using real-time CFD simulation data. When the simulation shows that the dead zone ratio in the extraction tower is >15%, the stirring speed is increased by 10% to 20% and the feed rate is simultaneously reduced by 5% to 10%; when the simulation shows that the local mass transfer coefficient K <1×10 -5 When the feed rate is m / s, increase the feed flow rate by 10% to 15%; when both are within the normal range, keep the current parameters unchanged.

[0029] Preferably, in S3, liquid phase fractionation occurs:

[0030] S3.1 Combined Clarification: Combine extracts A and B from S2, first centrifuge them in a tubular centrifuge at a speed of 10000-12000 rpm and a feed flow rate of 1-3 L / min to remove large particles, and then microfilter them through a 0.45 μm ceramic membrane to obtain a clarified liquid.

[0031] S3.2 Ultrafiltration membrane fractionation: Four spiral wound ultrafiltration membrane modules with molecular weight cutoffs of 100kDa, 50kDa, 10kDa, and 3kDa were used in series at 15–25℃ and a transmembrane pressure difference of 0.1–0.2MPa. The retentate and permeate were collected sequentially and numbered F1–F5 respectively. The contents of total sugar, protein, polyphenols, and flavonoids were then determined.

[0032] Preferably, in step S3.2, the following are collected sequentially:

[0033] (1) Retention solution I: a protein-polysaccharide complex with a value >100kDa, designated as F1;

[0034] (2) Retention solution II: a medium molecular weight polysaccharide of 50-100 kDa, and designated as F2;

[0035] (3) Retained solution III: a complex containing oligosaccharides and some polyphenols with a capacity of 10-50 kDa, and designated as F3;

[0036] (4) Retained solution IV: is rich in flavonoid glycosides and small molecule polyphenols with a concentration of 3-10 kDa, and is designated as F4;

[0037] (5) Permeate: It is a small molecule active substance with <3kDa containing free amino acids, monosaccharides, organic acids and SOD, and is numbered F5.

[0038] Preferably, the directed purification process in S4 is as follows:

[0039] S4.1 Polysaccharide Purification: F2 and F3 were combined and concentrated to 1 / 4 of their original volume under low-temperature vacuum at ≤40℃ and vacuum degree of -0.08 to -0.095MPa. The mixture was then loaded onto a DEAE-cellulose column and eluted using a linear gradient of 0 to 0.5M NaCl. The neutral polysaccharide peak and the acidic polysaccharide peak were collected. After low-temperature vacuum concentration at <40℃, dialysis desalting with a dialysis bag with a 3.5kDa cutoff, and vacuum freeze-drying at a pre-freezing temperature of -40 to -30℃, a sublimation drying vacuum degree of 10 to 30Pa, and a desorption drying temperature of 25 to 30℃, neutral polysaccharide of Hovenia dulcis with a purity of ≥86% and acidic polysaccharide with a purity of ≥82% were obtained.

[0040] S4.2 Separation of flavonoids and saponins: F4 was concentrated under low-temperature vacuum at a temperature ≤40℃ and a vacuum degree of -0.08 to -0.095 MPa, and then loaded onto a D101 macroporous adsorption resin column. It was eluted sequentially with pure water, 10% ethanol, 40% ethanol, 60% ethanol, and 85% ethanol. The 40% ethanol eluent rich in flavonoid glycosides and the 60% ethanol eluent rich in saponins were collected. The two eluents were respectively concentrated under low-temperature vacuum at a temperature ≤40℃ and a vacuum degree of -0.08 to -0.095 MPa, and then freeze-dried under vacuum at a pre-freezing temperature of -40 to -30℃, a sublimation drying vacuum degree of 10 to 30 Pa, and a desorption drying temperature of 25 to 30℃ to obtain total flavonoids of jujube (calculated as rutin) ≥70% and total saponins (calculated as oleanolic acid) ≥65%.

[0041] S4.3 Polyphenol Refining: The 85% ethanol eluent rich in polyphenols is rotary evaporated to dryness at <40℃, then dissolved in 50% ethanol at a ratio of 1:2 to 1:5, and allowed to stand at 1 to 4℃ for crystallization for 10 to 12 hours. The crude polyphenol crystals are obtained by vacuum filtration at a vacuum degree of -0.06 to -0.08 MPa and a filter membrane pore size of 0.45 μm. The crystals are then dissolved by heating with ethyl acetate and recrystallized 1 to 2 times after cooling. Finally, high-purity Hovenia dulcis polyphenols with a content of ≥95% based on gallic acid are obtained.

[0042] Preferably, the low-temperature refining of the heat-sensitive component in step S5 includes the following sub-steps:

[0043] S5.1 Superoxide dismutase extraction: Take F5 solution, concentrate it using a 30kDa ultrafiltration membrane, add ammonium sulfate to saturation to 58%–60%, salt out overnight at 1–4℃, centrifuge at 0–4℃ and 10,000–12,000 rpm for 10–15 min, collect the precipitate after standing, redissolve it in Tris-HCl buffer at pH 7.8, perform Sephadex G-100 gel column chromatography, collect the SOD activity peak, desalt it by ultrafiltration using an ultrafiltration membrane with a molecular weight cutoff of 3kDa, and freeze-dry it under vacuum at a pre-freezing temperature of -40 to -30℃, a sublimation drying vacuum of 10–30 Pa, and a desorption drying temperature of 25–30℃ to obtain superoxide dismutase powder with a specific activity ≥3000 U / mg;

[0044] S5.2, Dihydromyricetin Crystallization: Take the 40% ethanol eluent rich in flavonoid glycosides from S4 and further use polyamide column chromatography, eluting with a gradient of 30% to 60% ethanol. Collect the fraction containing DMY, concentrate it to a small volume under low temperature vacuum at a temperature ≤40℃ and a vacuum degree of -0.08 to -0.095MPa, and recrystallize it with 20% to 25% ethanol at 0 to 5℃ for 10 to 12 hours to obtain white crystals of dihydromyricetin with a purity ≥98%.

[0045] S5.3 Recovery of natural fragrance from Japanese raisin tree: The gas released by supercritical extraction in S2 and the volatile components recovered by condensation during low-temperature extraction are passed into an ice-water condenser and condensed at 0-2℃. The condensate is separated by an oil-water separator, and the upper light phase is the crude fragrance. It is then refined by short-path molecular distillation at 40-50℃ and a vacuum of 1-5Pa to obtain colorless to pale yellow Japanese raisin tree essential oil.

[0046] Preferably, the utilization step of the extraction residue in step S6 is as follows:

[0047] S6.1 Preparation of insoluble dietary fiber: Wash the residue after S2 extraction with water until neutral, treat with 0.3%-0.5% citric acid solution at 55-60℃ for 2-2.5h, centrifuge the residue at 4000-5000rpm for 10-15min, let it stand to precipitate, wash with 90%-95% ethanol and acetone in sequence, vacuum dry at ≤50℃, pulverize through 100 mesh to obtain water-insoluble dietary fiber with a purity ≥90%;

[0048] S6.2 Extraction of soluble dietary fiber: Combine the supernatant from the acid treatment in S6.1, add 4-5 times the volume of 95% ethanol and precipitate overnight. Centrifuge at 4-6℃ and 9000-10000rpm for 10-12min. After standing, take the precipitate, redissolve it in water, dialyze, and freeze dry to obtain water-soluble dietary fiber with a purity ≥75%.

[0049] S6.3, Japanese raisin tree syrup and fermented products: The microfiltration permeate from S3 and the washing liquid from each chromatography step in S5 are combined, desalted by a nanofiltration membrane with a 200 Da cutoff, and concentrated at a low temperature of <50℃ to a soluble solids content ≥65°Brix to obtain a Japanese raisin tree syrup with a total sugar content ≥55%. Take 1 / 3 of the syrup and dilute it to a soluble solids content of 10-15°Brix. Add 0.1%-0.3% activated yeast and ferment at 20-22℃ for 5-7 days to make a low-alcohol fermented beverage of Japanese raisin tree.

[0050] S6.4 Water Circulation and Residue Fertilization: All clean wastewater generated by the process is treated by multi-media filtration + activated carbon + RO reverse osmosis and reused for S1 cleaning and S2 solvent preparation; the final extraction residue is composted and decomposed into organic fertilizer.

[0051] Preferably, in step S7, the value coefficient is determined by calculating the value ratio of each product based on the average market price over the past three years and the actual product quality obtained during the process. :

[0052]

[0053] in, Let i be the reference unit price for the i-th product. Let be the output of the i-th product per ton of raw materials. Let j be the reference unit price of the j-th product. Let J be the output of the j-th product in ton of raw materials.

[0054] Compared with existing technologies, this invention provides a low-temperature extraction and co-production process for the high-value utilization of all components of *Hovenia dulcis*, which has the following beneficial effects:

[0055] 1. This invention constructs a three-stage countercurrent extraction system consisting of supercritical CO2 extraction, low-temperature ethanol countercurrent extraction, and low-temperature water enzymatic extraction. Combined with ultrafiltration membrane molecular retention and fractionation, macroporous adsorption resin directional purification, and extraction and fermentation conversion of dietary fiber from the residue, this invention achieves the beneficial effect of high-value co-production of all components of *Hovenia dulcis*. It also co-produces more than ten products, including *Hovenia dulcis* seed oil, neutral / acidic polysaccharides, total flavonoids, total saponins, high-purity polyphenols, SOD enzyme powder, dihydromyricetin, natural flavorings, soluble / insoluble dietary fiber, *Hovenia dulcis* nutritional syrup, and low-alcohol fermented beverages. Furthermore, the residue is used to make organic fertilizer, and all process water is reused, ultimately achieving a near-zero waste circular economy goal.

[0056] 2. This invention achieves the beneficial effect of effectively protecting the natural configuration and biological activity of the heat-sensitive active ingredients in Japanese jujube by using gradient freeze-drying in the raw material pretreatment stage, a combination of supercritical CO2, low-temperature ethanol and low-temperature water enzymatic method in the extraction stage, and low-temperature vacuum concentration, low-temperature freeze-drying and low-temperature recrystallization in the purification stage, and maintaining a low temperature of 1~4℃ throughout the SOD extraction process. Furthermore, the SOD specific activity is ≥3000U / mg and the dihydromyricetin purity is ≥98%, overcoming the problem of large-scale degradation and inactivation of heat-sensitive components caused by traditional high-temperature processes.

[0057] 3. This invention establishes a dedicated physical property database for Japanese raisin tree and uses a CFD three-dimensional mass transfer model to regulate extraction parameters in real time. Furthermore, it constructs a surrogate model using Gaussian process regression and combines it with the NSGA-II multi-objective genetic algorithm to collaboratively optimize product yield, purity, energy consumption, and cost. Finally, the optimal parameters are solidified into the DCS system and connected to an online near-infrared spectrometer combined with a partial least squares model for real-time feedback regulation. This achieves the beneficial effects of intelligent and precise regulation of extraction rate fluctuations and batch-to-batch product index fluctuations of ≤5%, as well as batch stability assurance. It solves the problem of poor process adaptability and consistency under the traditional experience-based operation mode. Attached Figure Description

[0058] Figure 1 This is a flow chart of the co-production process of the present invention. Detailed Implementation

[0059] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0060] Please see Figure 1 A low-temperature extraction and co-production process for high-value utilization of all components of *Hovenia dulcis* includes the following steps:

[0061] S1. Raw material harvesting and low-temperature property database construction: Harvest, sort, clean, dry at low temperature and grade and crush the Japanese raisin tree, and simultaneously establish a dedicated database covering the chemical composition and thermal properties of all components to provide basic data for subsequent simulation and optimization.

[0062] S2. Low-temperature multi-stage countercurrent extraction: CO2, ethanol and purified water are used in sequence to carry out three-stage countercurrent extraction of jujube at low temperature. At the same time, the velocity field and concentration field in the extraction tower are predicted in real time by CFD model to guide the setting of solvent flow rate and stirring speed.

[0063] S3. Liquid phase fractionation: After centrifugation and microfiltration, the extract mixture is initially separated into multiple fractions by using an ultrafiltration membrane that cuts off the molecular weight gradient.

[0064] S4. Targeted purification: Gradient elution of F2 to F5 fractions was carried out using macroporous adsorption resins of different polarities to prepare high-purity Hovenia dulcis polysaccharides, total flavonoids, total saponins and total polyphenols respectively.

[0065] S5, Low-temperature refined heat-sensitive components (SOD, dihydromyricetin, fragrance): Superoxide dismutase, dihydromyricetin and natural fragrance are extracted from the permeate F5 of S3 and the volatile condensate of extraction using membrane separation, salting out, chromatography and molecular distillation techniques under low-temperature conditions throughout the process.

[0066] S6. Utilization of extraction residue: The residue after S2 extraction and the by-product streams of washing water and permeate generated in S3 to S5 are converted into dietary fiber, jujube syrup and organic fertilizer, respectively, to achieve zero waste of all components.

[0067] S7. Product Value Co-production Assessment: Based on market value, establish a co-product value coefficient model, reasonably allocate total costs, and calculate the return on investment and break-even point under different production capacities.

[0068] S8. Multi-objective optimization: Integrate all experimental, simulation and financial data from S1 to S7 to construct a hybrid agent model. Use the NSGA-II algorithm to simultaneously maximize product yield and purity, minimize energy consumption and cost, and send the optimal operating parameters to the DCS system to achieve closed-loop control.

[0069] Specifically, the product value assessment process in S7:

[0070] S7.1 Determination of Value Coefficient: Based on the average market price (RMB / kg) over the past three years and the actual product quality obtained during the process (kg / ton of raw material), calculate the value ratio of each product. :

[0071]

[0072] in, Let i be the reference unit price for the i-th product. The output of the i-th product in ton of raw materials (kg / ton of raw materials). Let j be the reference unit price of the j-th product. Let J be the output of the j-th product in ton of raw material (kg / ton of raw material).

[0073] S7.2 Cost Allocation: The total manufacturing cost (raw materials, energy consumption, water consumption, labor, equipment depreciation, maintenance, management, etc.) is allocated according to the value ratio of each product. The cost is proportionally allocated to each joint product and then divided by the output of that product to obtain the unit product cost.

[0074] S7.3 Economic Sensitivity Analysis: Establish an economic model in Excel or Python to calculate the IRR (benchmark discount rate 12%), NPV (project period 10 years), and dynamic investment payback period under the production capacity (500kg / batch, 2000kg / batch, 10t / batch), and draw a spider diagram of sensitivity to single factors (raw material price, product selling price, energy price).

[0075] Specifically, the multi-objective optimization steps in S8 are as follows:

[0076] S8.1 Process Big Data Platform: The physical property data of S1, the CFD simulation results of S2, the measured output data of S3 to S6, and the economic data of S7 are uniformly stored in the time series database (InfluxDB), and outliers are cleaned (3σ principle).

[0077] S8.2, surrogate model training: Gaussian process regression (GPR) is used to establish predictive models for the yield, purity, total energy consumption, and total water consumption of each co-product. The input features are 25 dimensions (including feedstock gradation, extraction temperature, time, solvent ratio, membrane cutting pressure, elution flow rate, etc.), and the output consists of 7 targets. Training is performed on 80% of the data, and validation is performed on 20% of the data; the coefficient of determination R² must be ≥0.92.

[0078] S8.3, NSGA-II Multi-Objective Optimization: The objective function is set to maximize total flavonoid yield, polysaccharide yield, DMY purity, and SOD activity retention, while minimizing energy consumption and wastewater generation per unit product. The population size is 200, the number of iterations is 500, the crossover probability is 0.9, and the mutation probability is 0.1, yielding the Pareto front solution set. The optimal solution is selected by combining CRITIC weighting and TOPSIS sorting.

[0079] S8.4, CFD-AI joint simulation: The optimal extraction operating conditions (temperature, speed, flow rate) recommended by NSGA-II are fed back to the CFD model to simulate and verify the proportion of dead zones and concentration distribution in the extraction column again. If the requirements are met, it is determined as the final process package.

[0080] S8.5 Online Real-time Control System: The optimal parameters are solidified into the DCS system, and an online near-infrared (NIR) spectrometer (collection wavelength 900-1700nm) and temperature / pressure sensors are connected. The partial least squares (PLS) model is used to predict the total polyphenol and flavonoid content in the extract in real time, and the extraction time is automatically fine-tuned (±10%) according to the prediction deviation to ensure that the batch product index fluctuation is ≤5%.

[0081] The present invention was disassembled into four embodiments according to the above procedures, and each was paired with a corresponding comparative example 1 to 4 for horizontal comparison. The evaluation results are as follows:

[0082] Example 1 (Construction of Raw Material Pretreatment System and Low-Temperature Property Database)

[0083] The complete pretreatment process in S1 was adopted. Specifically, the premium raw materials were harvested 25 days after the first frost and washed at 10℃ using 35kHz ultrasonication combined with bubbling in softened water with a conductivity ≤5μS / cm for 6 minutes; gradient freeze-drying was then performed (pre-freezing at -35℃ / 7h, first drying at -8℃ / 22h, and second drying at 28℃ / 6h) until the moisture content was ≤5%; brittle pulverization was carried out at -3℃, and the powders of four particle sizes (F40, F80, F120, and F200) were strictly collected by sieve. Subsequently, the physical properties of each batch of raw materials were fully measured, and the density, specific heat, and particle size distribution were recorded. / / 25 parameters, including oil / polysaccharide / flavonoid content, are entered into a dedicated database to create a multidimensional index.

[0084] Comparative Example 1 (Traditional hot air drying and non-grading pulverization)

[0085] The washed Japanese raisins were dried using hot air drying (temperature set at 60℃, duration 12h), a common method in food processing plants, until the moisture content dropped to ≤5%. After drying, the jujubes were directly pulverized at room temperature (approximately 25℃) using a regular hammer mill. The jujubes were simply sieved through a single-layer 80-mesh screen, and the material passing through the screen was used as the raw material for subsequent extraction. No systematic physical property database was established, and the subsequent extraction process parameters were set based on the experience of experienced technicians.

[0086] Summary of Results 1: In Comparative Example 1, hot air drying at 60℃ resulted in a 15%–20% loss of heat-sensitive components (SOD enzyme essentially inactivated, dihydromyricetin partially degraded) in the jujube, and single-stage pulverization led to an extremely wide particle size distribution (10–300 mesh mixed together), causing extraction rate fluctuations exceeding ±10% in subsequent supercritical extraction due to uneven particle size. In contrast, Example 1 protected the heat-sensitive active components through gradient freeze-drying, and the staged pulverization combined with the establishment of a database provided accurate input parameters for the S2 step CFD simulation (especially...). The use of particle size data improved the simulation accuracy and verified the fundamental role of the low-temperature property database in the entire process.

[0087] Example 2 (Low-temperature three-stage countercurrent extraction and CFD adaptive control)

[0088] The complete S2.1–S2.4 steps were performed: First, supercritical CO2 (28 MPa / 36℃ / 4 L / min, 120 min) was used to extract the oil from the seeds of *Hovenia dulcis*. Then, the residue was transferred to a jacketed tower with a paddle stirrer and subjected to countercurrent extraction with 55% ethanol (1:20 feed-to-liquid ratio, 25℃, 90 rpm). Finally, the residue was enzymatically hydrolyzed with deionized water (1:25 feed-to-liquid ratio) and a mixture of 0.3% cellulase and 0.2% pectinase at 45℃ for 65 min. Throughout the process, online CFD simulation was maintained. When the model showed a dead zone ratio >15%, the stirring speed was automatically increased from 90 rpm to 105 rpm (+16.7%), and the feed rate was reduced by 8%. When the local mass transfer coefficient K <1×10⁻⁶, the simulation was performed. -5 When the flow rate is m / s, the feed velocity is automatically increased by 12% to enhance turbulence.

[0089] Comparative Example 2 (Traditional hot reflux extraction vs. fixed parameters):

[0090] A heated reflux extraction tank, commonly used in food extraction workshops, was employed. The mixed powders of F40 to F80 from Example 1 were added to the tank, along with 6 times the volume (mL / g) of 70% ethanol. The temperature was set at 80°C, and the stirring speed was kept constant at 60 rpm. Extraction was performed twice in a single-batch, intermittent manner (2 hours each time). The extracts were combined and filtered; no further water extraction or enzymatic hydrolysis was performed on the residue, which was discarded directly. This process did not involve CFD simulation; the stirring speed and feed rate were controlled solely by visual inspection by the operator.

[0091] Summary of Results 2: In Comparative Example 2, although the high temperature of 80℃ accelerated molecular motion, it led to the loss of a large amount of heat-sensitive fragrances through volatilization, and the highly polar polysaccharides could not be dissolved by 70% ethanol, resulting in extremely low yields of water-soluble active components (polysaccharides and proteins) (polysaccharide yield less than 15%). More importantly, the fixed stirring parameters could not cope with batch-to-batch differences in materials. When encountering raw materials with high fiber content, extraction "dead zones" were easily formed in the tank, causing flavonoid content fluctuations of up to 22% between different batches. Example 2 achieved full polarity coverage (non-polar oils → moderately polar flavonoid saponins → polar polysaccharides) through a three-stage enzymatic extraction method using low-temperature CO2 + ethanol + water. Combined with real-time CFD control to eliminate mass transfer dead zones, the total polyphenol yield was increased by more than 40%, and the batch-to-batch fluctuation of the extraction rate was strictly controlled within ≤5%.

[0092] Example 3 (Liquid-phase fractionation of membrane modules and targeted purification of thermosensitive components)

[0093] The S3-S5 steps were performed in full: the extract was centrifuged in a tube (11000 rpm) and microfiltered through a ceramic membrane. Then, under a transmembrane pressure difference of 0.15 MPa, it was passed in series through ultrafiltration membrane modules of 100 kDa, 50 kDa, 10 kDa, and 3 kDa to precisely obtain five fractions, F1 to F5. For the F5 permeate, ammonium sulfate was added to 60% saturation at 4°C for salting out. After purification by Sephadex G-100 column chromatography and freeze-drying, SOD enzyme powder with a specific activity ≥3000 U / mg was obtained. For the F4 fraction, polyamide column chromatography combined with low-temperature recrystallization was used to obtain white crystals of dihydromyricetin (DMY) with a purity ≥98%. At the same time, the supercritical gas was condensed at 0°C and short-path molecular distilled (45°C, 3 Pa) to obtain the characteristic aroma essential oil of Japanese raisin tree.

[0094] Comparative Example 3 (Conventional alcohol precipitation and high-temperature concentration)

[0095] The commercially available alcohol precipitation + spray drying process was adopted. The ethanol extract obtained in Example 2 was combined with the water extract, and industrial ethanol was directly added until the ethanol concentration of the system reached 80%. The mixture was allowed to stand overnight to precipitate the polysaccharides, and crude polysaccharides were obtained by plate and frame filtration. The filtrate was concentrated and the ethanol was recovered by vacuum rotary evaporation at 60°C. For SOD and dihydromyricetin, only routine detection was performed without special low-temperature chromatography purification (because high temperature easily deactivates them); the fragrance was simply collected by a single-stage coil condenser without molecular distillation purification.

[0096] Summary of Results 3: In Comparative Example 3, a single-stage precipitation with 80% ethanol resulted in the loss of a large amount of medium molecular weight active polysaccharides (10-50 kDa) in the crude polysaccharide due to co-precipitation. Simultaneously, a significant amount of flavonoid glycosides were also precipitated, leading to a decrease of approximately 35% in the subsequent total flavonoid recovery rate. While the evaporation temperature of 60°C seems mild, it is sufficient to cause irreversible denaturation of SOD enzymes (optimal temperature approximately 37°C), resulting in a specific activity residue of less than 200 U / mg. Example 3 employed a molecular weight cutoff gradient membrane to achieve precise cutting of active components, avoiding cross-contamination. Furthermore, the entire process was strictly controlled under low-temperature conditions, successfully protecting the natural stereoconfiguration of SOD and dihydromyricetin, and achieving the parallel production of high-value-added rare ingredients.

[0097] Example 4 (Zero waste of extraction residue, co-production assessment and AI closed-loop optimization)

[0098] Complete steps S6, S7, and S8: The residue from step S2 is treated with 0.4% citric acid (58℃, 2.2h), centrifuged, and precipitated to obtain insoluble dietary fiber with a purity ≥90%. The supernatant is precipitated with alcohol to obtain soluble dietary fiber. The washing water and membrane permeate are combined, desalted by nanofiltration, and concentrated at low temperature to produce ≥65°Brix jujube syrup (total sugar ≥55%), which is partially fermented into a low-alcohol beverage. RO reverse osmosis treatment is used to achieve water reuse. Based on this, the economic model of S7 was used to calculate the IRR (benchmark discount rate of 12%) for a production capacity of 10t / batch; the GPR surrogate model of S8 (R²≥0.92) combined with the NSGA-II algorithm (population 200, generation 500) was used to obtain the Pareto front solution set with the goal of maximizing total flavonoid yield, DMY purity, and SOD activity retention while minimizing energy consumption and wastewater, and the optimal extraction temperature / speed / flow rate parameters were sent to the DCS system; at the same time, online NIR (900~1700nm) combined with the PLS model was used to provide real-time feedback to fine-tune the extraction time, thereby achieving closed-loop control.

[0099] Comparative Example 4 (Residue Disposal, Empirical Production Decisions, and Open-Loop Control)

[0100] The extraction residue undergoes no dietary fiber extraction or fermentation treatment and is directly incinerated or landfilled as solid waste, incurring additional solid waste disposal costs. Production scheduling and process temperature settings are based entirely on the workshop supervisor's past experience, without NPV / IRR economic calculations. The DCS system is used only as a display panel, with a fixed extraction time; online NIR detection and AI proxy model optimization are not employed, and samples are only sent to the laboratory for offline testing before final product bottling (results lag 6-8 hours).

[0101] Summary of Results 4: In Comparative Example 4, the disposal of residue not only resulted in the loss of economic value from dietary fiber and fermentable sugars, but also increased the environmental protection cost of solid waste treatment by 200-300 yuan per ton, leading to a decrease in the overall profit margin. More importantly, the open-loop control mode could not cope with the fluctuations in physical properties caused by changes in the origin of raw materials. When the fiber content in the raw materials changed abruptly, the fixed extraction time resulted in a batch-to-batch deviation of flavonoid content as high as ±18%, seriously affecting the stability of brand quality. Example 4 truly realized data-driven (CFD+AI+online NIR) to replace manual experience, while achieving optimal economic and technical indicators under multi-objective synergy, verifying the advancement and robustness of this invention within the Industry 4.0 framework.

[0102] In summary, through a point-by-point comparison of the four embodiments and comparative examples, it is fully demonstrated that the present invention is not a single extraction process, but a systematic engineering project that integrates low-temperature activity preservation, membrane-based fine segmentation, tiered utilization of residues, and AI closed-loop optimization. Specifically, Examples 1-4 demonstrate that the low-temperature property database and CFD control solve the problems of traditional extraction; membrane grading enables multi-product production to replace single crude extracts; residue utilization transforms a burden into a benefit; and the linkage between NSGA-II and DCS provides a reusable smart factory solution for the jujube processing industry. The four embodiments are interconnected, and the absence of any one link will lead to a decrease in yield, an increase in energy consumption, or a waste of resources.

[0103] Although embodiments of the invention have been shown and described, it will be understood by those skilled in the art that various changes, modifications, substitutions and alterations can be made to these embodiments without departing from the principles and spirit of the invention, the scope of which is defined by the appended claims and their equivalents.

Claims

1. A low-temperature extraction and co-production process for the high-value utilization of all components of *Hovenia dulcis*, characterized in that, Includes the following steps: S1. Raw material harvesting and low-temperature property database construction: Harvest, sort, clean, dry at low temperature and grade and crush the Japanese raisin tree, and simultaneously establish a complete component property database; S2. Low-temperature multi-stage countercurrent extraction: CO2, ethanol and purified water are used in sequence to carry out three-stage countercurrent extraction of jujube at low temperature. At the same time, the velocity field and concentration field in the extraction tower are predicted in real time by CFD model to guide the setting of solvent flow rate and stirring speed. S3. Liquid phase fractionation: After centrifugation and microfiltration, the extract mixture is initially separated into active components of different molecular weights by passing it through an ultrafiltration membrane that retains the molecular weight gradient, to obtain fractions F1 to F5. S4. Targeted purification: Gradient elution of F2 to F5 fractions was carried out using macroporous adsorption resins of different polarities to prepare high-purity Hovenia dulcis polysaccharides, total flavonoids, total saponins and total polyphenols respectively. S5, Low-temperature refined heat-sensitive components: Superoxide dismutase, dihydromyricetin and natural flavorings are extracted from the permeate F5 of S3 and the volatile condensate of the extract using membrane separation, salting out, chromatography and molecular distillation techniques under low-temperature conditions throughout the process. S6. Utilization of extraction residue: The residue after S2 extraction and the by-product streams of washing water and permeate generated in S3 to S5 are converted into dietary fiber, jujube syrup and organic fertilizer, respectively, to achieve zero waste of all components. S7. Product Value Co-production Assessment: Based on market value, establish a co-product value coefficient model, reasonably allocate total costs, and calculate the return on investment and break-even point under different production capacities. S8. Multi-objective optimization: Record all experimental, simulation and financial data from S1 to S7, construct a hybrid agent model based on the data, use the NSGA-II algorithm to simultaneously maximize product yield and purity, minimize energy consumption and cost, and send the optimal operating parameters to the DCS system to achieve closed-loop control.

2. The low-temperature extraction and co-production process for high-value utilization of all components of *Hovenia dulcis* according to claim 1, characterized in that, The raw material harvesting and low-temperature property database construction in S1: S1.1 Harvesting and Maturity Grading: Harvest the Japanese raisin tree jujubes within 15-30 days after the first frost. They are classified into special grade, first grade, and second grade according to the color and plumpness of the fruit stalk. The latitude, longitude, altitude, average daily temperature, and rainfall are recorded for each batch. S1.2 Ultrasonic low-temperature cleaning: Use 30-40kHz ultrasonic waves combined with bubbling cleaning. The cleaning solution is softened water with a conductivity of ≤5μS / cm. The cleaning temperature is controlled at 8-12℃ and the time is 5-8min. After cleaning, drain until there is no visible water on the surface. S1.3 Gradient freeze drying: After draining, the jujubes are spread on trays and pre-frozen at -40 to -30℃ for 6 to 8 hours; the first drying temperature is -10 to -5℃ and the vacuum degree is 5 to 8 Pa, lasting for 20 to 25 hours; the second drying temperature is 25 to 30℃ and the vacuum degree is 10 to 15 Pa, lasting for 4 to 8 hours, until the final moisture content is ≤5%; S1.4 Low-temperature embrittlement and ultrafine classification: After drying, the material is subjected to impact embrittlement and pulverization at -5 to 0℃. It is then passed through 40 mesh, 80 mesh, 120 mesh and 200 mesh standard sieves in sequence to collect four particle size grades of powder, which are numbered F40, F80, F120 and F200 respectively. S1.5 Construction of a complete component property database: For each batch of raw materials, the basic physicochemical indicators, the content of characteristic active ingredients and process compatibility parameters are measured, and all data are entered into the complete component property database.

3. The low-temperature extraction and co-production process for high-value utilization of all components of *Hovenia dulcis* according to claim 1, characterized in that, The low-temperature multi-stage countercurrent extraction in S2: S2.1 Supercritical CO2 Extraction of Hovenia dulcis Seed Oil: Take F40-F80 grade powder from S1, load it into a supercritical extraction vessel, introduce food-grade CO2, set the pressure to 25-30 MPa, temperature to 35-38℃, CO2 flow rate to 3-5 L / min, and extraction time to 100-150 min. After pressure reduction and separation in a separation vessel at 6-8 MPa and 35-40℃, collect the Hovenia dulcis seed oil precipitated in the separation vessel. The residue after extraction is retained as raw material for the next stage of extraction. S2.2 Low-temperature ethanol countercurrent extraction: The residue after supercritical extraction is transferred to a heat-insulated jacketed countercurrent extraction tower. The extraction tower has a built-in paddle stirrer with a speed of 80-100 rpm. 50%-55% edible ethanol is used as the solvent. The material-liquid ratio is set to 1:15-1:25, the extraction temperature is 20-30℃, and the residence time is 60-120 min. After solid-liquid separation, extract A rich in flavonoids and saponins is collected. The residue after extraction is reserved as raw material for the next stage of extraction. S2.3 Low-temperature water + compound enzyme-assisted extraction: Add deionized water to the residue after ethanol extraction at a material-to-liquid ratio of 1:20 to 1:30, stir and extract at 15 to 25°C for 50 to 60 minutes, and add 0.2% to 0.5% cellulase and 0.1% to 0.3% pectinase at the same time. Adjust the pH to 6.0 to 6.8, and then briefly raise the temperature to 40 to 50°C for enzymatic hydrolysis for 60 to 70 minutes. After enzymatic hydrolysis, filter and collect the extract B rich in polysaccharides and proteins. The remaining residue is sent to section S6 for comprehensive utilization. S2.4, CFD Real-Time Simulation and Parameter Adaptation: Based on the property database of S1, a three-dimensional rotating blade-porous medium coupled model is established, and input... , The particle size distribution, solvent viscosity, and diffusion coefficient of the powder are used to simulate the turbulent kinetic energy and local mass transfer coefficient in the extraction tower through finite element numerical solution. Based on this, the stirring speed and feed rate in step S2.2 are adjusted online to ensure that the extraction rate fluctuation is ≤5%.

4. The low-temperature extraction and co-production process for high-value utilization of all components of *Hovenia dulcis* according to claim 3, characterized in that, In step S2.4, the parameters are adaptively adjusted online using real-time CFD simulation data. When the simulation shows a dead zone ratio > 15% in the extraction tower, the stirring speed is increased by 10%–20%, and the feed rate is simultaneously reduced by 5%–10%. When the simulation shows a local mass transfer coefficient K < 1 × 10⁻⁶, the stirring speed is adjusted accordingly. -5 When the feed rate is m / s, increase the feed flow rate by 10% to 15%; when both are within the normal range, keep the current parameters unchanged.

5. The low-temperature extraction and co-production process for high-value utilization of all components of *Hovenia dulcis* according to claim 1, characterized in that, Liquid phase fractionation in S3: S3.1 Combined Clarification: Combine extracts A and B from S2, first centrifuge them in a tubular centrifuge at a speed of 10000-12000 rpm and a feed flow rate of 1-3 L / min to remove large particles, and then microfilter them through a 0.45 μm ceramic membrane to obtain a clarified liquid. S3.2 Ultrafiltration membrane fractionation: Four spiral wound ultrafiltration membrane modules with molecular weight cutoffs of 100kDa, 50kDa, 10kDa, and 3kDa were used in series at 15–25℃ and a transmembrane pressure difference of 0.1–0.2MPa. The retentate and permeate were collected sequentially and numbered F1–F5 respectively. The contents of total sugar, protein, polyphenols, and flavonoids were then determined.

6. The low-temperature extraction and co-production process for high-value utilization of all components of *Hovenia dulcis* according to claim 5, characterized in that, In step S3.2, the following are collected sequentially: (1) Retention solution I: a protein-polysaccharide complex with a value >100kDa, designated as F1; (2) Retention solution II: a medium molecular weight polysaccharide of 50-100 kDa, and designated as F2; (3) Retained solution III: a complex containing oligosaccharides and some polyphenols with a capacity of 10-50 kDa, and designated as F3; (4) Retained solution IV: is rich in flavonoid glycosides and small molecule polyphenols with a concentration of 3-10 kDa, and is designated as F4; (5) Permeate: It is a small molecule active substance with <3kDa containing free amino acids, monosaccharides, organic acids and SOD, and is numbered F5.

7. The low-temperature extraction and co-production process for high-value utilization of all components of *Hovenia dulcis* according to claim 6, characterized in that, The targeted purification process in S4: S4.1 Polysaccharide Purification: F2 and F3 were combined and concentrated to 1 / 4 of their original volume under low-temperature vacuum at ≤40℃ and vacuum degree of -0.08 to -0.095MPa. The mixture was then loaded onto a DEAE-cellulose column and eluted using a linear gradient of 0 to 0.5M NaCl. The neutral polysaccharide peak and the acidic polysaccharide peak were collected. After low-temperature vacuum concentration at <40℃, dialysis desalting with a dialysis bag with a 3.5kDa cutoff, and vacuum freeze-drying at a pre-freezing temperature of -40 to -30℃, a sublimation drying vacuum degree of 10 to 30Pa, and a desorption drying temperature of 25 to 30℃, neutral polysaccharide of Hovenia dulcis with a purity of ≥86% and acidic polysaccharide with a purity of ≥82% were obtained. S4.2 Separation of flavonoids and saponins: F4 was concentrated under low-temperature vacuum at a temperature ≤40℃ and a vacuum degree of -0.08 to -0.095 MPa, and then loaded onto a D101 macroporous adsorption resin column. It was eluted sequentially with pure water, 10% ethanol, 40% ethanol, 60% ethanol, and 85% ethanol. The 40% ethanol eluent rich in flavonoid glycosides and the 60% ethanol eluent rich in saponins were collected. The two eluents were respectively concentrated under low-temperature vacuum at a temperature ≤40℃ and a vacuum degree of -0.08 to -0.095 MPa, and then freeze-dried under vacuum at a pre-freezing temperature of -40 to -30℃, a sublimation drying vacuum degree of 10 to 30 Pa, and a desorption drying temperature of 25 to 30℃ to obtain total flavonoids of jujube (calculated as rutin) ≥70% and total saponins (calculated as oleanolic acid) ≥65%. S4.3 Polyphenol Refining: The 85% ethanol eluent rich in polyphenols is rotary evaporated to dryness at <40℃, then dissolved in 50% ethanol at a ratio of 1:2 to 1:5, and allowed to stand at 1 to 4℃ for crystallization for 10 to 12 hours. The crude polyphenol crystals are obtained by vacuum filtration at a vacuum degree of -0.06 to -0.08 MPa and a filter membrane pore size of 0.45 μm. The crystals are then dissolved by heating with ethyl acetate and recrystallized 1 to 2 times after cooling. Finally, high-purity Hovenia dulcis polyphenols with a content of ≥95% based on gallic acid are obtained.

8. The low-temperature extraction and co-production process for high-value utilization of all components of *Hovenia dulcis* according to claim 1, characterized in that, The low-temperature refining of the heat-sensitive component in S5 includes the following sub-steps: S5.1 Superoxide dismutase extraction: Take F5 solution, concentrate it using a 30kDa ultrafiltration membrane, add ammonium sulfate to saturation to 58%–60%, salt out overnight at 1–4℃, centrifuge at 0–4℃ and 10,000–12,000 rpm for 10–15 min, collect the precipitate after standing, redissolve it in Tris-HCl buffer at pH 7.8, perform Sephadex G-100 gel column chromatography, collect the SOD activity peak, desalt it by ultrafiltration using an ultrafiltration membrane with a molecular weight cutoff of 3kDa, and freeze-dry it under vacuum at a pre-freezing temperature of -40 to -30℃, a sublimation drying vacuum of 10–30 Pa, and a desorption drying temperature of 25–30℃ to obtain superoxide dismutase powder with a specific activity ≥3000 U / mg; S5.2, Dihydromyricetin Crystallization: Take the 40% ethanol eluent rich in flavonoid glycosides from S4 and further use polyamide column chromatography, eluting with a gradient of 30% to 60% ethanol. Collect the fraction containing DMY, concentrate it to a small volume under low temperature vacuum at a temperature ≤40℃ and a vacuum degree of -0.08 to -0.095MPa, and recrystallize it with 20% to 25% ethanol at 0 to 5℃ for 10 to 12 hours to obtain white crystals of dihydromyricetin with a purity ≥98%. S5.3 Recovery of natural fragrance from Japanese raisin tree: The gas released by supercritical extraction in S2 and the volatile components recovered by condensation during low-temperature extraction are passed into an ice-water condenser and condensed at 0-2℃. The condensate is separated by an oil-water separator, and the upper light phase is the crude fragrance. It is then refined by short-path molecular distillation at 40-50℃ and a vacuum of 1-5Pa to obtain colorless to pale yellow Japanese raisin tree essential oil.

9. The low-temperature extraction and co-production process for high-value utilization of all components of *Hovenia dulcis* according to claim 1, characterized in that, The step of utilizing the extraction residue in S6: S6.1 Preparation of insoluble dietary fiber: Wash the residue after S2 extraction with water until neutral, treat with 0.3%-0.5% citric acid solution at 55-60℃ for 2-2.5h, centrifuge the residue at 4000-5000rpm for 10-15min, let it stand to precipitate, wash with 90%-95% ethanol and acetone in sequence, vacuum dry at ≤50℃, pulverize through 100 mesh to obtain water-insoluble dietary fiber with a purity ≥90%; S6.2 Extraction of soluble dietary fiber: Combine the supernatant from the acid treatment in S6.1, add 4-5 times the volume of 95% ethanol and precipitate overnight. Centrifuge at 4-6℃ and 9000-10000rpm for 10-12min. After standing, take the precipitate, redissolve it in water, dialyze, and freeze dry to obtain water-soluble dietary fiber with a purity ≥75%. S6.3, Japanese raisin tree syrup and fermented products: The microfiltration permeate from S3 and the washing liquid from each chromatography step in S5 are combined, desalted by a nanofiltration membrane with a 200 Da cutoff, and concentrated at a low temperature of <50℃ to a soluble solids content ≥65°Brix to obtain a Japanese raisin tree syrup with a total sugar content ≥55%. Take 1 / 3 of the syrup and dilute it to a soluble solids content of 10-15°Brix. Add 0.1%-0.3% activated yeast and ferment at 20-22℃ for 5-7 days to make a low-alcohol fermented beverage of Japanese raisin tree. S6.4 Water Circulation and Residue Fertilization: All clean wastewater generated by the process is treated by multi-media filtration + activated carbon + RO reverse osmosis and reused for S1 cleaning and S2 solvent preparation; the final extraction residue is composted and decomposed into organic fertilizer.

10. The low-temperature extraction and co-production process for high-value utilization of all components of *Hovenia dulcis* according to claim 1, characterized in that, The value coefficient in S7 is determined by calculating the value ratio of each product based on the average market price over the past three years and the actual product quality obtained during the process. : in, Let i be the reference unit price for the i-th product. Let be the output of the i-th product per ton of raw materials. Let j be the reference unit price of the j-th product. Let J be the output of the j-th product in ton of raw materials.

Images