Modularized food-nourishing system based on intelligent sensing and preparation method of modularized food-nourishing system

Abstract

The invention relates to the technical field of large health food, in particular to a modular food-nourishing system based on intelligent sensing and a preparation method thereof.The system comprises an intelligent sensing module, a physique identification and formula decision module, a food-nourishing product preparation module and a time sequence management module; a day, ground and human homology medicine and food formula system and a four-section variable temperature gradient precise extraction process are provided, precise analysis is carried out by utilizing a dynamic physique identification model based on a federal learning framework, and personalized adaptation can be carried out on a reference formula through a generative fine tuning technology; and performing real-time simulation and parameter optimization on the extraction process by combining a digital twinning and process optimization model, and planning a taking scheme through a dynamic time sequence management module with cognitive feedback capability. According to the invention, a technical closed loop from intelligent perception, accurate decision and adaptive production to closed loop optimization is realized, and the accuracy, effectiveness and user compliance of food and nutrition intervention are greatly improved.

Description

Technical Field

[0001] This application relates to the field of health food technology, and in particular to a modular dietary system based on intelligent sensing and its preparation method. Background Technology

[0002] Currently, food-based health products are facing a series of systemic technical bottlenecks: First, the products have limited functions and do not match the complex sub-health conditions of the human body; second, traditional health preservation theories lack a quantitative path to connect with modern production technology, resulting in crude formulation and processing; a more prominent contradiction lies in the preparation stage—the industry generally adopts a "one-pot stew" process where all raw materials are cooked at the same temperature for a long time. This linear thinking completely ignores the essential differences in the physical texture and thermal stability of different raw materials, leading to at least three defects: 1) Heat-sensitive components (such as volatile oils and some glycosides) in flower and leaf raw materials are destroyed or dispersed at high temperatures; 2) Active ingredients in dense root, stem, and seed raw materials are not fully extracted due to insufficient temperature or time; 3) Large fluctuations in the process result in unstable product quality between batches. In addition, the "bitter taste of good medicine" seriously affects user experience and long-term adherence, and the product application is out of sync with the human body's circadian rhythm, further weakening the intervention effect.

[0003] More importantly, the existing solutions are only superficially "intelligent," mostly limited to data collection and simple rule matching. They lack continuous learning capabilities based on data, precise modeling and control of the production process, and the closed-loop ability to self-optimize based on feedback. This results in static and rigid intervention solutions that cannot truly achieve dynamic and precise adaptation to individual needs.

[0004] While intelligent technologies such as federated learning and digital twins have made progress in their respective fields, the field of health and nutrition still lacks a systematic solution that can deeply couple these cutting-edge intelligent technologies with a fixed, modular formula system designed based on a specific theory (such as the Three Powers Theory) and its dedicated physical production process (such as four-stage variable temperature gradient extraction), forming a complete technological closed loop with continuous optimization capabilities. The core contradiction that existing technologies have failed to address lies in how a static formula process system can respond to dynamic individual physiological states, and how complex personalized decisions can accurately drive and optimize the physical production process.

[0005] Therefore, there is an urgent need for a solution that can revolutionize all aspects of the problem, from theory to practice and from design to application, and can deeply integrate cutting-edge intelligent technologies, in order to systematically solve the above-mentioned problems. Summary of the Invention

[0006] This application provides a modular dietary therapy system based on intelligent sensing to solve problems such as the limited functionality of traditional dietary therapy products, crude production processes, low user compliance, insufficient system intelligence, and lack of adaptive optimization capabilities.

[0007] The first aspect of this application provides a modular dietary therapy system based on intelligent sensing, comprising: an intelligent sensing module, a constitution identification and formulation decision-making module, a dietary therapy product preparation module, and a timing management module. The intelligent sensing module collects user physiological state data and environmental data through a multi-source sensor fusion network and incorporates a data reliability assessment unit to clean and quality-label the raw data. The constitution identification and formulation decision-making module analyzes the user's constitution using a dynamic constitution identification model built on a federated learning framework, matches a benchmark formulation from a predefined system of medicinal and edible raw material formulations, and outputs a personalized formulation adjustment plan through a generative formulation fine-tuning engine to obtain a target formulation. The dietary therapy product preparation module uses a four-stage variable temperature gradient precise extraction process to prepare the raw materials in the target formulation to obtain the prepared product, and integrates a process digital twin model and process parameter optimization algorithm to simulate, predict, and dynamically optimize the production process. The timing management module plans the consumption sequence for the prepared product and dynamically adjusts the timing rules based on user feedback data and physiological responses.

[0008] Optionally, the predefined food-medicine homology raw material formulation system consists of complementary and independently prepared celestial, human, and terrestrial formulations.

[0009] Optionally, the step of analyzing the user's constitution by using the physiological state data through a dynamic constitution identification model built on a federated learning framework, matching a benchmark formula from a predefined system of medicinal and edible raw material formulas, and outputting a personalized formula adjustment scheme through a generative formula fine-tuning engine includes: constructing a general constitution identification benchmark model, training the general constitution identification benchmark model on a central server based on a federated learning framework, and performing personalized fine-tuning of the model on the user's terminal device using local data under privacy protection to form a dynamically updated user-specific constitution identification model; inputting the physiological state data into the user-specific constitution identification model, and inputting the feature vector output by the user-specific constitution identification model into a conditional generative adversarial network to generate a raw material ratio fine-tuning scheme for the current user's state.

[0010] Optionally, the integrated process digital twin model and process parameter optimization algorithm for simulating, predicting, and dynamically optimizing the production process includes: establishing a digital twin model of key equipment in a four-stage variable temperature gradient precision extraction process; inputting real-time production data and preset optimization targets into the process optimization algorithm; performing simulation calculations through the digital twin model; and dynamically recommending and adjusting process parameters at each stage.

[0011] Optionally, the raw material composition and weight ratio of the Heavenly, Human, and Earthly formulas include: Heavenly Module: 8-12 parts rose, 4-6 parts chrysanthemum, 2-4 parts mint, 4-6 parts lotus leaf, 8-12 parts Poria cocos, 3-5 parts dried tangerine peel, 5-7 parts almond, 2-4 parts bamboo leaf, 1-3 parts bitter orange blossom, 1-2 parts tea; Human Module: 8-12 parts jujube seed, 6-10 parts longan pulp, 8-12 parts lily bulb, 8-12 parts Poria cocos, 8 parts lotus seed. -12 parts, walnut kernels 6-8 parts, red ginseng / ginseng / American ginseng slices 1-2 parts, dried plums 2-4 parts, osmanthus 1-3 parts, tea 1-2 parts, where the weight ratio of walnut kernels to red ginseng / ginseng / American ginseng slices is (3.5-4.5):1; Earthly modules: black sesame 10-14 parts, polygonatum 8-12 parts, wolfberry 8-12 parts, mulberry 8-12 parts, five-finger peach 13-17 parts, yam 10-14 parts, black dates 5-7 parts, black tea 4-6 parts.

[0012] The second aspect of this application provides a modular food preparation method based on intelligent sensing, which employs a four-stage variable temperature gradient precise extraction process. The raw materials of the Heaven, Human, and Earth components are classified according to their physical texture and added sequentially. The method includes the following steps: adding and extracting root and seed raw materials in the first high-temperature stage; adding and extracting fruit and peel raw materials in the second medium-temperature stage; adding and extracting flower and leaf raw materials in the third low-temperature stage; and integrating and blending the flavor in the final integration stage.

[0013] Optionally, the specific parameters of the four-stage variable temperature gradient precision extraction process are as follows: the extraction temperature in the first high-temperature stage is 98±2℃, and the extraction time is 35-45 minutes; the extraction temperature in the second medium-temperature stage is 88±2℃, and the extraction time is 15-25 minutes; the extraction temperature in the third low-temperature stage is 78±2℃, and the extraction time is 5-10 minutes; and the temperature in the final integration stage is 70±2℃, and the integration time is 2-5 minutes.

[0014] Optionally, when preparing the human part formula, Poria cocos, lotus seeds, and red ginseng / ginseng / American ginseng slices are added in the first high-temperature stage; longan pulp, lily bulb, walnut kernels, and dried plum are added in the second medium-temperature stage; and jujube kernels and osmanthus are added in the third low-temperature stage.

[0015] Optionally, during the final integration stage or subsequent blending stage, one or more of the following can be added as natural flavoring agents or form modifiers: kudzu root powder, cold-extracted moringa leaf powder, acerola cherry powder, konjac powder, honey, mogrosides, and natural maple syrup, to meet different flavor and product form requirements.

[0016] The beneficial effects of using the present invention are as follows: This application proposes and establishes a three-part basic formula of "Heaven, Earth, and Man," which follows the "Three Powers" theory. The Heavenly formula (mainly flowers and leaves) functions to clear and disperse pathogens; the Man formula (mainly nuts) functions to calm the mind and promote harmony; and the Earthly formula (mainly roots and stems) functions to nourish. Addressing the significant differences in the texture of the raw materials within the formula system, a four-stage variable temperature gradient precise extraction process is employed. Furthermore, by introducing digital twins and process optimization algorithms, static process parameters are upgraded to an intelligent process that dynamically optimizes based on raw material batches and environmental conditions, ensuring that each production run approaches optimal efficiency. A dynamic constitution identification model that continuously evolves with user data is constructed using a federated learning framework, breaking away from static models. This approach overcomes the limitations of traditional food-based products by employing generative fine-tuning technology to achieve personalized and precise adjustments to the baseline formula; it optimizes the user's eating experience and improves user compliance through the scientific application of natural flavoring agents; it facilitates industrial expansion through a modular architecture; and it upgrades the fixed process into an intelligent agent with learning and optimization capabilities through a time-series management module with cognitive feedback capabilities, better addressing individual differences and changes, and truly realizing the intervention concept of "dynamic balance." User stickiness and intervention effectiveness are cumulatively improved, and the overall intervention effectiveness is expected to show a cumulative upward trend over time, forming a complete intelligent closed loop of "perception-decision-execution-feedback-optimization." This solves the problems of traditional food-based health products, such as limited functionality, crude production processes, low user compliance, insufficient system intelligence, and lack of adaptive optimization capabilities.

[0017] Additional aspects and advantages of this application will be set forth in part in the description which follows, and in part will be obvious from the description, or may be learned by practice of this application. Attached Figure Description

[0018] The above and / or additional aspects and advantages of this application will become apparent and readily understood from the following description of the embodiments taken in conjunction with the accompanying drawings, wherein: Figure 1 This is a schematic diagram of the structure of a modular dietary therapy system based on intelligent sensing provided in an embodiment of this application; Figure 2 This is a flowchart illustrating the collaborative work of federated learning and generative fine-tuning according to embodiments of this application; Figure 3 This is a schematic diagram illustrating the principle of digital twin-driven process optimization according to an embodiment of this application; Figure 4 This is an extraction process curve diagram of the four-stage variable temperature gradient precision extraction process provided in the embodiments of this application; Figure 5 This is a schematic diagram illustrating the relationship between the four-stage variable temperature gradient precision extraction process and the formulation adaptation provided in the embodiments of this application. Detailed Implementation

[0019] The embodiments of this application are described in detail below. Examples of these embodiments are shown in the accompanying drawings, wherein the same or similar reference numerals denote the same or similar elements or elements having the same or similar functions throughout. The embodiments described below with reference to the accompanying drawings are exemplary and intended to explain this application, and should not be construed as limiting this application.

[0020] The following description, with reference to the accompanying drawings, describes a modular dietary therapy system based on intelligent sensing, according to embodiments of this application. Addressing the problems of limited functionality, rudimentary processes, and disorganized applications mentioned in the background section, this application provides a modular dietary therapy system based on intelligent sensing. In this system, Specifically, Figure 1 This is a schematic diagram of the structure of a modular dietary therapy system based on intelligent sensing provided in an embodiment of this application.

[0021] like Figure 1 As shown, the modular dietary therapy system based on intelligent sensing includes: an intelligent sensing module 100, a constitution identification and formula decision-making module 200, a dietary therapy product preparation module 300, and a time-series management module 400.

[0022] The intelligent sensing module 100 collects users' physiological state data and environmental data through a multi-source sensor fusion network, and has a built-in data credibility assessment unit to clean and quality-label the raw data; the constitution identification and prescription decision module 200 analyzes users' constitutions using a dynamic constitution identification model built on a federated learning framework, matches a benchmark formula from a predefined system of medicinal and edible raw material formulas, and outputs a personalized formula adjustment plan through a generative formula fine-tuning engine to obtain the target formula; the food product preparation module 300 uses a four-stage variable temperature gradient precision extraction process to prepare the raw materials in the target formula to obtain the prepared product, and integrates a process digital twin model and process parameter optimization algorithm to realize the simulation, prediction and dynamic optimization of the production process; the timing management module 400 plans the consumption sequence for the prepared product and dynamically adjusts the timing rules according to user feedback data and physiological response.

[0023] It should be noted that the multi-dimensional data collected by the multi-source sensor network in this application, encompassing "heaven" (circadian rhythm), "earth" (environment), and "human" (physiology), provides unique, cross-dimensional training features for the federated learning model. This allows the model's output of physical constitution assessment results to transcend the limitations of single physiological data analysis, providing the data foundation for accurate "three-dimensional" adaptation. The personalized formula output by the generative fine-tuning engine does not merely adjust the proportions of raw materials; its adjustment logic (such as increasing or decreasing the proportion of a certain raw material) is directly related to the physical texture and thermosensitive properties of that raw material. Therefore, this scheme serves as a key input parameter, triggering the digital twin model to re-optimize the subsequent four stages of temperature gradient process parameters, ensuring the optimal reproduction of the personalized formula on the production line. The process parameters and predicted quality indicators optimized in real time by the digital twin model during production, along with the physiological feedback data after user consumption, serve as reinforcing signals to optimize the federated learning model and generation strategy, thereby enabling the entire system to continuously evolve around the user's experience.

[0024] Understandably, the intelligent sensing module integrates multi-source data and assesses its reliability; the constitution identification and prescription decision-making module employs an evolutionary federated learning model and a generative fine-tuning engine; the dietary product preparation module can optionally be equipped with digital twins for process optimization; and the time-series management module has the ability to dynamically adjust based on feedback. These modules work together to form an intelligent closed loop.

[0025] In the embodiments of this application, the predefined formula system of medicinal and edible raw materials consists of complementary and independently prepared formulas of the celestial part, the human part, and the terrestrial part.

[0026] It should be noted that the "Heaven, Earth, and Man" formula design follows the "Three Powers" theory. The Heaven formula is positioned as "clearing and dispersing," aiming to support the rise of Yang energy and the activation of metabolism in the morning (Lesser Yang phase), and to resolve the stagnation of heat in the "Upper Jiao." The ingredient combination is mainly composed of light and upward-moving flowers and leaves (such as mint and chrysanthemum), supplemented with qi-regulating and dampness-resolving ingredients (such as dried tangerine peel and poria). The Man formula is positioned as "connecting the heart and kidneys, calming the mind and improving intelligence," aiming to regulate the physiological transition from evening to night (Lesser Yin and Greater Yin phases) when the heart fire descends and the kidney water ascends, promoting mental tranquility. The ingredient combination is based on nourishing the heart and calming the mind (sour jujube seed and lily bulb) and nourishing the kidneys and clearing the heart (walnut kernel and red ginseng slices). The Earth formula is positioned as "nourishing and protecting the foundation," aiming to provide deep nourishment and energy replenishment after high-energy-consuming activities during the day (Solar Yang phase), and to strengthen the foundation of the "Lower Jiao." The ingredients are mainly for nourishing the liver and kidneys (black sesame seeds, polygonatum) and strengthening the spleen and replenishing qi (five-finger peach, yam).

[0027] It is understandable that the embodiments of this application, through the "three-element" modular design, transform the macro-level health concept into industrialized product units with clear objectives that can be independently produced, verified, sold, and flexibly combined, ensuring the systematic nature of efficacy while possessing production flexibility, thus solving the problems of vague functions and arbitrary compatibility of traditional compound products.

[0028] In the embodiments of this application, the raw material composition and weight ratio of the Heavenly Formula, Human Formula, and Earthly Formula include: Heavenly Module: 8-12 parts rose petals, 4-6 parts chrysanthemum petals, 2-4 parts mint petals, 4-6 parts lotus leaves, 8-12 parts poria cocos, 3-5 parts dried tangerine peel, 5-7 parts almond petals, 2-4 parts bamboo leaves, 1-3 parts bitter orange blossom, 1-2 parts tea. Human Part Module: 8-12 parts of jujube seed, 6-10 parts of longan pulp, 8-12 parts of lily bulb, 8-12 parts of poria cocos, 8-12 parts of lotus seed, 6-8 parts of walnut kernel, 1-2 parts of red ginseng / ginseng / American ginseng slices, 2-4 parts of dried plum, 1-3 parts of osmanthus, and 1-2 parts of tea. The weight ratio of walnut kernel to red ginseng / ginseng / American ginseng slices is (3.5-4.5):1. Earthly Module: 10-14 parts black sesame, 8-12 parts polygonatum, 8-12 parts wolfberry, 8-12 parts mulberry, 13-17 parts five-finger peach, 10-14 parts yam, 5-7 parts black dates, 4-6 parts black tea.

[0029] It is understood that the embodiments of this application, by clearly defining the raw material composition, weight proportions and key ratios of each module, transform traditional compatibility experience into quantifiable and standardized technical parameters, providing a clear basis for industrial production and solving the problems of vague raw material ratios and large batch differences in traditional food and nutrition products.

[0030] In this embodiment, physiological state data is analyzed using a dynamic constitution identification model built on a federated learning framework. A baseline formula is matched from a predefined system of medicinal and edible ingredients, and a personalized formula adjustment plan is output through a generative formula fine-tuning engine. This includes: A general physical fitness identification benchmark model is constructed. The general physical fitness identification benchmark model is trained on a central server based on a federated learning framework. The model is then personalized and fine-tuned on user terminal devices using local data under privacy protection, forming a dynamically updated user-specific physical fitness identification model. Physiological state data is input into a user-specific constitution identification model, and the feature vector output by the user-specific constitution identification model is input into a conditional generative adversarial network to generate a fine-tuning scheme for the raw material ratio based on the current user state.

[0031] It should be noted that the user-local data processed by the federated learning framework in this embodiment is a structured feature vector representing the imbalance of the "two constitutions" after preprocessing by the aforementioned "heterogeneous sensor fusion network," rather than the original physiological signals. The central server does not aggregate general health data, but rather model parameters specifically used to optimize the "mapping relationship from multidimensional data to specific formula systems." This gives the federated learning process a distinct scenario specificity, and its optimization objective directly serves the core decision-making logic of this system.

[0032] The model training process is guided by the "Two Elements" theory in the Eastern life system perspective, classifying constitution types into two basic categories: "predominantly Yang deficiency" and "predominantly Yin deficiency," as well as several intermediate subtypes (such as Yang deficiency with liver stagnation, Yin deficiency with spleen deficiency, etc.). The training sample library contains over 1200 sets of constitution label data confirmed by professional TCM physicians. Each set of samples corresponds to a complete multi-dimensional physiological parameter dataset, ensuring that the model can learn the objective mapping relationship between constitution type and physiological characteristics. Cross-validation is used to optimize hyperparameters during model training, ultimately achieving an accuracy rate of no less than 85% in constitution type determination, with an accuracy rate of no less than 90% for determining the core basic constitution categories.

[0033] The accuracy of the dynamic model based on federated learning can continuously improve with the user's usage time. After 3 months of federated learning iterations, the confidence level of the user's physical condition identification model can be improved by about 10-15 percentage points from the initial level, which makes the "one person, one prescription" approach change from static matching to dynamic tracking.

[0034] For example, to improve the accuracy and personalization of body constitution identification, the system is deployed as follows: In the cloud, a deep neural network based on the "Two Elements" theory is trained using anonymized labeled data to serve as a baseline physical constitution identification model. This baseline model is then distributed to user terminals (such as mobile apps) using a federated learning framework. Figure 2 As shown, the terminal uses locally stored private historical physiological data to fine-tune the model, uploading only encrypted model parameter updates to the cloud for aggregation, forming a privacy-preserving distributed learning system. Each user has a locally-owned federated learning model that continuously adapts to their own changes. When formula fine-tuning is needed, the high-dimensional feature vector output by the user-owned model, along with the target formula identifier (such as "human formula"), is input into a pre-trained conditional generative adversarial network. This network can generate highly personalized suggestions for fine-tuning the ingredient ratios (e.g., slightly increasing the amount of jujube seed by 0.5 parts and slightly decreasing the amount of red ginseng slices by 0.1 parts on the baseline formula) to achieve more refined formula optimization.

[0035] Understandably, this application achieves more comprehensive "heaven, earth, and human" data collection (physiological and environmental) through a multi-source sensor fusion network. It utilizes a federated learning framework to construct a dynamic constitution identification model that continuously evolves with user data, overcoming the limitations of static models. Through generative fine-tuning technology, it enables personalized and precise adjustments to the baseline formula. Finally, through a time-series management module with cognitive feedback capabilities, it forms a complete intelligent closed loop of "perception-decision-execution-feedback-optimization".

[0036] In this embodiment of the application, the integration of a process digital twin model and a process parameter optimization algorithm to achieve simulation, prediction, and dynamic parameter optimization of the production process includes: Establish digital twin models of key equipment in the four-stage variable temperature gradient precision extraction process; Real-time production data and preset optimization targets are input into the process optimization algorithm, which is then simulated and calculated using a digital twin model to dynamically recommend and adjust process parameters at each stage.

[0037] It should be noted that the key to the digital twin model constructed in this invention lies in its physical mapping rules, which strictly follow the specific mass and heat transfer kinetics of root, fruit, and leaf raw materials at different temperature stages in the 'four-segment variable temperature gradient precise extraction process'. A schematic diagram of the process optimization principle driven by the digital twin is shown below. Figure 3 As shown, this model is not a general process simulator, but a dedicated simulation and optimization tool that is deeply bound to the specific formulation system and segmented process of this invention. It can make sensitive and accurate process parameter responses to changes in raw material ratios caused by formulation fine-tuning.

[0038] Digital twin-driven process optimization can, on the basis of achieving the advantages of basic processes (extraction rate improvement of more than 32.5%, RSD < 3.5%), adaptive parameter re-optimization can be performed for each personalized formulation fine-tuning. According to simulation calculations, the batch stability index (RSD) can be further reduced by about 50% on the basis of the original high level.

[0039] For example, to improve the accuracy and stability of the process, the following steps are implemented: A high-fidelity digital twin model is established for the extraction production line. Before each production run, real-time monitoring data of the raw materials for this "ground-based formula" (such as moisture content and particle size) and the production target (such as "maximizing polysaccharide extraction rate") are input into the process optimization algorithm. This algorithm can be a Bayesian optimization algorithm. The optimization algorithm drives the digital twin model to perform high-speed simulation calculations, seeking the best from a massive number of parameter combinations and recommending the optimal set of process parameters for this production run (e.g., first stage: 99℃, 42 minutes; second stage: 87℃, 18 minutes…). During production, the digital twin runs synchronously with the physical equipment, enabling virtual monitoring, deviation prediction, and proactive control.

[0040] Next, referring to the accompanying drawings, a modular food preparation method based on intelligent sensing, according to an embodiment of this application, is described.

[0041] This method employs a four-stage variable temperature gradient precise extraction process, and the raw materials for the Heaven, Human, and Earth components are classified according to their physical properties and added sequentially, including the following steps: The root and seed ingredients in the formula are added and extracted in the first high-temperature stage; The fruit and peel ingredients in the formula are added and extracted in the second medium-temperature stage; The flower and leaf ingredients in the formula are added and extracted in the third low-temperature stage; Flavor integration and blending are carried out in the final integration stage.

[0042] In this embodiment, the specific parameters of the four-stage variable temperature gradient precision extraction process are as follows: the extraction temperature in the first high-temperature stage is 98±2℃, and the extraction time is 35-45 minutes; the extraction temperature in the second medium-temperature stage is 88±2℃, and the extraction time is 15-25 minutes; the extraction temperature in the third low-temperature stage is 78±2℃, and the extraction time is 5-10 minutes; the final integration stage temperature is 70±2℃, and the integration time is 2-5 minutes. The extraction process curve is shown in the figure below. Figure 4 As shown.

[0043] In the embodiments of this application, when preparing the human part formula, Poria cocos, lotus seeds and red ginseng / ginseng / American ginseng slices are added in the first high temperature stage; longan pulp, lily bulb, walnut kernel and dried plum are added in the second medium temperature stage; and jujube seed and osmanthus are added in the third low temperature stage.

[0044] It should be noted that this invention integrates formula design and process design. The raw material composition of the Heavenly, Human, and Earthly formulas dictates that the process of this invention is necessary to achieve optimal results. Any attempt to uniformly add the above raw materials and cook them at a uniform temperature will result in insufficient extraction of some components while destroying others.

[0045] like Figure 5 As shown, root and seed raw materials such as Poria cocos and lotus seeds have a dense texture and require long-term extraction at the first high-temperature stage (98℃) to fully dissolve polysaccharides and other components. Fruit and peel raw materials such as longan pulp and walnut kernels contain sugars, oils, and some heat-sensitive components, and are suitable for extraction at the second medium-temperature stage (88℃) to avoid high-temperature charring. Flower and leaf raw materials such as jujube seed and osmanthus are rich in volatile oils and glycosides and are extremely delicate, requiring short-term extraction at the third low-temperature stage (78℃) to preserve their core flavor and activity.

[0046] It is understood that, according to the physical characteristics and heat sensitivity of active ingredients of different raw materials such as roots / seeds (high temperature resistant), fruits / peels (medium temperature required), and flowers / leaves (high temperature sensitive), the feeding and extraction are carried out in four precisely temperature-controlled stages, thereby maximizing the dissolution rate of various functional ingredients and protecting their activity in a single extraction system.

[0047] In the embodiments of this application, during the final integration stage or subsequent blending stage, one or more of the following are added as natural flavoring agents or form modifiers: kudzu root powder, cold-extracted moringa leaf powder, acerola cherry powder, konjac powder, honey, mogroside, and natural maple syrup, to meet different flavor and product form requirements.

[0048] It should be noted that the selection and amount of natural flavor enhancers should be designed specifically for the flavor characteristics of each module: the Heaven module focuses on a refreshing taste and can add a small amount of honey (1%-2%); the Human module needs to soften the bitterness and can add monk fruit glycosides (0.01%-0.03%); the Earth module needs to enhance the mellowness and can add natural maple syrup (2%-3%).

[0049] The taste of traditional Chinese medicine (TCM) is a significant factor affecting patient medication adherence and is also an important consideration in the design of TCM formulations. Bitter-tasting TCMs constitute a large proportion of existing TCM products. Therefore, improving the bitterness of TCMs and their preparations to make them more palatable to patients is of great significance for the application and development of TCM.

[0050] It is understandable that the embodiments of this application, through the scientific selection and proportioning of natural flavoring agents, significantly optimize the eating experience without affecting the efficacy of the product, solve the problem that traditional health products cannot balance efficacy and taste, and improve user medication compliance.

[0051] The following two examples and one comparative example illustrate the four-segment variable temperature gradient precision extraction process: Example

[0052] Human body module four-stage variable temperature gradient extraction (basic ratio) Ingredients: 8 parts jujube seed, 6 parts longan pulp, 8 parts lily bulb, 8 parts poria cocos, 8 parts lotus seed, 6 parts walnut kernel, 1.5 parts red ginseng slices, 2 parts dried plum, and 1 part osmanthus.

[0053] Extraction steps Add the root and seed raw materials (Poria cocos, lotus seeds and red ginseng slices) and 18 times the weight of deionized water to the extraction tank, heat to 98℃, and extract at a constant temperature for 40 minutes with a stirring rate of 50 r / min. Lower the temperature inside the extraction tank to 88°C, add fruit and peel ingredients (longan pulp, lily bulb, walnut kernels and dried plum), and extract at a constant temperature for 20 minutes; Lower the temperature inside the extraction tank to 78°C, add the flower and leaf raw materials (jujube seed and osmanthus), and extract at a constant temperature for 8 minutes; Cool naturally to 70°C, gently stir and combine for 3 minutes, filter through a 200-mesh sieve, and collect the extract; Add 0.02% mogroside to the extract, stir well, and then concentrate and encapsulate under vacuum. Example

[0054] Human body module four-stage variable temperature gradient extraction (high ratio) Ingredients: 12 parts jujube seed, 10 parts longan pulp, 12 parts lily bulb, 12 parts poria cocos, 12 parts lotus seed, 8 parts walnut kernel, 2 parts red ginseng slices, 4 parts dried plum, and 3 parts osmanthus.

[0055] Extraction steps Add the root and seed raw materials (Poria cocos, lotus seeds and red ginseng slices) and 18 times the weight of deionized water to the extraction tank, heat to 98℃, and extract at a constant temperature for 45 minutes with a stirring rate of 50 r / min. Lower the temperature inside the extraction tank to 88°C, add fruit and peel ingredients (longan pulp, lily bulb, walnut kernels and dried plum), and extract at a constant temperature for 25 minutes; Lower the temperature inside the extraction tank to 78°C, add the flower and leaf raw materials (jujube seed and osmanthus), and extract at a constant temperature for 10 minutes; Cool naturally to 70°C, gently stir and combine for 5 minutes, filter through a 200-mesh sieve, and collect the extract; Add 0.02% natural maple syrup to the extract, stir well, and then vacuum concentrate and encapsulate.

[0056] Comparative Example 1 Traditional constant temperature extraction process for human body module (basic formula) Ingredients: 8 parts jujube seed, 6 parts longan pulp, 8 parts lily bulb, 8 parts poria cocos, 8 parts lotus seed, 6 parts walnut kernel, 1.5 parts red ginseng slices, 2 parts dried plum, and 1 part osmanthus.

[0057] Extraction steps Add the mixed raw materials and 18 times the weight of deionized water to the extraction tank, heat to 100℃, and extract at a constant temperature for 60 minutes with a stirring rate of 50 r / min. Filter directly through a 200-mesh sieve and collect the extract.

[0058] The extracts from Examples 1-2 and Comparative Example 1 were subjected to performance testing and analysis. The performance test results are shown in Table 1.

[0059] Test sample Total flavonoid extraction rate (mg / g) Polysaccharide extraction rate (mg / g) Batch RSD (total flavonoids) (%) Long-term use compliance (%) Example 1 7.15 16.83 0.95% 86.7 Example 2 7.36 17.26 1.07% 90.0 Comparative Example 1 5.21 12.75 8.47% 63.3 Table 1. Results of Extraction Process Tests As shown in Table 1, compared with Comparative Example 1, Example 1 showed a 37.2% increase in total flavonoid extraction rate, a 32% increase in polysaccharide extraction rate, and an approximately 8-fold improvement in batch stability. This indicates that the four-stage variable temperature gradient extraction method of this application can extract the active ingredients of plants more efficiently, resulting in extracts with higher nutritional value than those obtained using traditional extraction methods. Furthermore, the addition of natural flavoring agents to improve the flavor of traditional Chinese medicine significantly enhances user compliance.

[0060] The data fully demonstrates that the four-stage variable temperature gradient precision extraction process of this application, combined with a specific "heaven, earth, and human" formulation system, produces a synergistic effect. While significantly improving the extraction rate of active ingredients, it achieves extremely high product quality stability, with technical effects far exceeding traditional processes, exhibiting outstanding substantive characteristics and significant progress. Furthermore, simulation calculations based on the constructed digital twin optimization model show that the production batch stability (based on the RSD of total flavonoid content) of the above process can achieve approximately 50% further optimization potential when dealing with different raw material batches or personalized fine-tuning of the formulation. Simultaneously, the federated learning framework, through simulation, enables the dedicated model deployed on the user terminal to improve the confidence level of user constitution assessment by approximately 10-15 percentage points after three months of continuous learning. This fully demonstrates the continuous self-optimization technical characteristics of the system of this invention.

[0061] The following is a detailed description of a modular dietary therapy system based on intelligent sensing, using a specific embodiment: User Zhang, 35 years old, was initially diagnosed with a "Yang deficiency" constitution using a federated learning personalized model. The system primarily matched a "Ground-Based Formula." During a damp and cold season, the sensor fusion network detected changes in environmental temperature and humidity, and the generative fine-tuning engine suggested temporarily increasing the proportion of five-finger peach in the formula. Based on this specially formulated recipe, the preparation module used digital twin technology to optimize the model and generate exclusive extraction parameters for production. After Zhang took the product, his sleep data improved significantly. This positive feedback was captured by the cognitive feedback unit of the time-series management module, reinforcing the decision path of "initiating fine-tuning under specific environmental parameters." Throughout the process, the Eastern theoretical concepts of "Three Powers" and "Four Symbols" were implemented in a data-driven and dynamic manner through intelligent technology.

[0062] In the description of this specification, the references to terms such as "one embodiment," "some embodiments," "example," "specific example," or "some examples," etc., indicate that a specific feature, structure, material, or characteristic described in connection with that embodiment or example is included in at least one embodiment or example of this application. In this specification, the illustrative expressions of the above terms do not necessarily refer to the same embodiment or example. Furthermore, the specific features, structures, materials, or characteristics described may be combined in any suitable manner in one or more embodiments or examples. Moreover, without contradiction, those skilled in the art can combine and integrate the different embodiments or examples described in this specification, as well as the features of different embodiments or examples.

[0063] Furthermore, the terms "first" and "second" are used for descriptive purposes only and should not be construed as indicating or implying relative importance or implicitly specifying the number of technical features indicated. Thus, a feature defined as "first" or "second" may explicitly or implicitly include at least one of that feature. In the description of this application, "N" means at least two, such as two, three, etc., unless otherwise explicitly specified.

[0064] Any process or method described in the flowchart or otherwise herein can be understood as representing a module, segment, or portion of code comprising one or more N executable instructions for implementing custom logic functions or processes, and the scope of the preferred embodiments of this application includes additional implementations in which functions may be performed not in the order shown or discussed, including substantially simultaneously or in reverse order depending on the functions involved, as should be understood by those skilled in the art to which embodiments of this application pertain.

[0065] It should be understood that the various parts of this application can be implemented using hardware, software, firmware, or a combination thereof. In the above embodiments, the N steps or methods can be implemented using software or firmware stored in memory and executed by a suitable instruction execution system. For example, if implemented in hardware as in another embodiment, it can be implemented using any one or a combination of the following techniques known in the art: discrete logic circuits having logic gates for implementing logical functions on data signals, application-specific integrated circuits (ASICs) having suitable combinational logic gates, programmable gate arrays (PGAs), field-programmable gate arrays (FPGAs), etc.

[0066] Those skilled in the art will understand that all or part of the steps of the methods in the above embodiments can be implemented by a program instructing related hardware. The program can be stored in a computer-readable storage medium, and when executed, the program includes one or a combination of the steps of the method embodiments.

Claims

1. A modular dietary therapy system based on intelligent sensing, characterized in that, include: The module includes an intelligent sensing module, a constitution identification and formulation decision-making module, a dietary product preparation module, and a time-series management module. The intelligent sensing module is used to collect users' physiological state data and environmental data through a multi-source sensor fusion network, and has a built-in data credibility assessment unit to clean and quality mark the raw data. The constitution identification and formula decision module is used to analyze the user's constitution by using the physiological state data through a dynamic constitution identification model built based on a federated learning framework, match the benchmark formula from the predefined medicinal and food homology raw material formula system, and output a personalized formula adjustment plan through a generative formula fine-tuning engine to obtain the target formula. The dietary product preparation module is used to prepare the raw materials in the target formula using a four-stage variable temperature gradient precision extraction process to obtain the prepared product, and integrates a process digital twin model and process parameter optimization algorithm to realize the simulation, prediction and dynamic optimization of the production process. The timing management module is used to plan the dosage sequence of the prepared product and dynamically adjust the timing rules based on user feedback data and physiological responses.

2. The modular dietary therapy system based on intelligent sensing according to claim 1, characterized in that, The predefined formula system for medicinal and edible raw materials consists of three complementary and independently prepared formulas: the Heavenly Formula, the Human Formula, and the Earthly Formula.

3. The modular dietary therapy system based on intelligent sensing according to claim 1, characterized in that, The process involves analyzing the user's constitution using a dynamic constitution identification model built on a federated learning framework based on the physiological state data, matching a baseline formula from a predefined system of medicinal and edible ingredients, and outputting a personalized formula adjustment plan through a generative formula fine-tuning engine. This includes: A general physical constitution identification benchmark model is constructed. The general physical constitution identification benchmark model is trained on a central server based on a federated learning framework. The model is then personalized and fine-tuned on a user terminal device using local data under privacy protection, forming a dynamically updated user-specific physical constitution identification model. The physiological state data is input into the user-specific constitution identification model, and the feature vector output by the user-specific constitution identification model is input into the conditional generative adversarial network to generate a raw material ratio fine-tuning scheme for the current user state.

4. The modular dietary therapy system based on intelligent sensing according to claim 1, characterized in that, The integrated process digital twin model and process parameter optimization algorithm enable the simulation, prediction, and dynamic parameter optimization of the production process, including: Establish digital twin models of key equipment in the four-stage variable temperature gradient precision extraction process; Real-time production data and preset optimization targets are input into the process optimization algorithm, which is then simulated and calculated using the digital twin model to dynamically recommend and adjust process parameters at each stage.

5. The modular dietary therapy system based on intelligent sensing according to claim 2, characterized in that, The raw material composition and weight ratio of the Heavenly Formula, Human Formula, and Earthly Formula include: Heavenly Module: 8-12 parts rose petals, 4-6 parts chrysanthemum petals, 2-4 parts mint petals, 4-6 parts lotus leaves, 8-12 parts poria cocos, 3-5 parts dried tangerine peel, 5-7 parts almond petals, 2-4 parts bamboo leaves, 1-3 parts bitter orange blossom, 1-2 parts tea. Human Part Module: 8-12 parts of jujube seed, 6-10 parts of longan pulp, 8-12 parts of lily bulb, 8-12 parts of poria cocos, 8-12 parts of lotus seed, 6-8 parts of walnut kernel, 1-2 parts of red ginseng / ginseng / American ginseng slices, 2-4 parts of dried plum, 1-3 parts of osmanthus, and 1-2 parts of tea. The weight ratio of walnut kernel to red ginseng / ginseng / American ginseng slices is (3.5-4.5):

1. Earthly Module: 10-14 parts black sesame, 8-12 parts polygonatum, 8-12 parts wolfberry, 8-12 parts mulberry, 13-17 parts five-finger peach, 10-14 parts yam, 5-7 parts black dates, 4-6 parts black tea.

6. A modular dietary supplement preparation method based on intelligent sensing, characterized in that, The process employs a four-stage variable temperature gradient precision extraction technology, and the raw materials for the Heaven, Human, and Earth components are categorized according to their physical properties and added sequentially, including the following steps: The root and seed ingredients in the formula are added and extracted in the first high-temperature stage; The fruit and peel ingredients in the formula are added and extracted in the second medium-temperature stage; The flower and leaf ingredients in the formula are added and extracted in the third low-temperature stage; Flavor integration and blending are carried out in the final integration stage.

7. The modular dietary supplement preparation method based on intelligent sensing according to claim 6, characterized in that, The specific parameters of the four-stage variable temperature gradient precision extraction process are as follows: the extraction temperature in the first high-temperature stage is 98±2℃, and the extraction time is 35-45 minutes; the extraction temperature in the second medium-temperature stage is 88±2℃, and the extraction time is 15-25 minutes; the extraction temperature in the third low-temperature stage is 78±2℃, and the extraction time is 5-10 minutes; the temperature in the final integration stage is 70±2℃, and the integration time is 2-5 minutes.

8. The modular dietary supplement preparation method based on intelligent sensing according to claim 6, characterized in that, When preparing the human part formula, Poria cocos, lotus seeds, and red ginseng / ginseng / American ginseng slices are added in the first high temperature stage; longan pulp, lily bulb, walnut kernels, and dried plum are added in the second medium temperature stage; and jujube kernels and osmanthus are added in the third low temperature stage.

9. The modular dietary supplement preparation method based on intelligent sensing according to claim 6, characterized in that, In the final integration stage or subsequent blending stage, one or more of the following are added as natural flavoring agents or form modifiers: kudzu root powder, cold-extracted moringa leaf powder, acerola cherry powder, konjac powder, honey, monk fruit glycosides, and natural maple syrup, to meet different flavor and product form requirements.

Images