Unmanned medicine decocting system capable of achieving artificial intelligence and accurate quantity control of traditional Chinese medicine

Abstract

The invention relates to the cross technical field of artificial intelligence and traditional Chinese medicine, and discloses a traditional Chinese medicine artificial intelligence accurate quantity control unmanned medicine decocting system, the system constructs nine modules, a prescription data analysis module realizes intelligent analysis of prescription information, a process scheme intelligent formulation module realizes automatic generation of a personalized decoction scheme, and a traditional Chinese medicine decoction module realizes automatic control of the traditional Chinese medicine decoction. The multi-modal sensing data acquisition module realizes real-time data acquisition in the whole process, the dynamic regulation and control module realizes online process deviation correction, the equipment management module realizes accurate execution of fire power, pressure, water quantity, feeding and split charging, and the intelligent decoction end point judgment module realizes objective quantitative judgment of a decoction end point. The self-cleaning and sterilization maintenance module realizes no cross contamination in the equipment, the man-machine interaction module realizes local and cloud convenient operation, the safety early warning module realizes multi-fault active defense, and all the modules cooperate to realize intelligent control. The method has the advantages of intelligent prescription analysis, dynamic process adaptation, multi-dimensional sensing fusion, accurate judgment and special medicinal material differentiation treatment.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of artificial intelligence and traditional Chinese medicine, in particular to an unmanned decoction system for traditional Chinese medicine with artificial intelligence precise control. BACKGROUND

[0002] With the acceleration of the modernization process of traditional Chinese medicine, the traditional Chinese medicine decoction process is facing the demand for transformation and upgrading from experiential operation to standardized and intelligent production. However, the current mainstream automatic decoction equipment still has significant technical bottlenecks: in terms of individualized decoction capacity, the existing equipment generally lacks autonomous decision-making mechanism based on prescription information, and cannot dynamically adjust the core process parameters such as water addition ratio, decoction time, and fire control according to the type, dosage and compatibility of traditional Chinese medicines, making it difficult to achieve individualized decoction of thousands of prescriptions; for special medicinal material processing, such as long and slow decoction of mineral medicinal materials (turtle shell and turtle shell) and late addition of aromatic medicinal materials (mint and cardamom), the mechanical process of existing equipment cannot accurately reproduce the traditional processing specifications, resulting in a gap between the effective component dissolution rate and manual decoction; in the decoction endpoint determination link, the traditional experience-dependent boiling observation and timing has not been effectively combined with modern chromatographic analysis technology, and due to factors such as batch-to-batch quality difference of medicinal materials and extraction process working condition fluctuation, the actual extraction time often deviates from the optimal window period, causing unstable efficacy; in addition, the limitations of equipment performance also restrict the development of the industry - small decoction machines have single function, poor stability and need manual assistance, while large industrial equipment has improved production capacity but lacks flexibility, and generally has problems such as low cost performance and lack of remote control. These technical shortcomings seriously hinder the standardization and intelligent development of traditional Chinese medicine decoction process, and it is urgent to build a precise control unmanned decoction system with artificial intelligence decision-making ability, which breaks through the mechanical limitations of traditional equipment by integrating prescription analysis algorithm, multi-parameter dynamic control model and intelligent sensing technology, and realizes the technical leap from experiential decoction to data-driven precise decoction. SUMMARY

[0003] (I) Technical problems solved In view of the deficiencies of the prior art, the present application provides an unmanned decoction system for traditional Chinese medicine with artificial intelligence precise control, which has the advantages of intelligent prescription analysis and dynamic process adaptation, multi-dimensional sensing fusion precise judgment, and differential processing of special medicinal materials, and solves the problems of lack of individualized decoction capacity, strong subjectivity in decoction endpoint determination, and insufficient flexibility and stability of the equipment of the existing automatic decoction equipment.

[0004] (II) Technical solutions To achieve the above object, the present application provides the following technical solution: An unmanned decoction system for traditional Chinese medicine with precise control of artificial intelligence, comprising a prescription data analysis module, an intelligent process scheme formulation module, a multi-modal sensing data acquisition module, a dynamic control module, an equipment management module, an intelligent judgment module for decoction end point, a self-cleaning and sterilization maintenance module, a man-machine interaction module and a safety warning module. The prescription data analysis module is responsible for receiving and understanding prescription information. The intelligent process scheme formulation module generates individualized decoction scheme based on prescription information. The multi-modal sensing data acquisition module acquires full-dimensional data of the decoction process in real time. The dynamic control module monitors the full-dimensional data of the decoction process online and makes dynamic adjustments to the decoction scheme in real time. The equipment management module is responsible for managing the heating power, pressure adjustment, medicinal material feeding, water injection, medicinal liquid filtration and sub-packaging operation of the decoction equipment. The intelligent judgment module for decoction end point fuses and calculates the effective component concentration change curve, temperature and pressure curve and machine vision features measured by near-infrared spectroscopy to determine the end point. The self-cleaning and sterilization maintenance module performs all-round cleaning and sterilization of the inner container, pipeline and medicinal liquid channel of the equipment through in-situ cleaning, ultraviolet sterilization and high-temperature steam disinfection. The man-machine interaction module provides an operation interface, displays the current prescription, progress and real-time data curve, and supports remote monitoring of the decoction state through App / mini program, receiving completion notification and viewing historical records. The safety warning module monitors abnormal conditions such as dry burning, boiling over, excessively high pressure and sensor failure in real time, and when the module detects the above faults, automatically starts the safety program, and at the same time, performs hierarchical warning prompts through the sound-light alarm and operation interface pop-up window.

[0005] Preferably, the prescription data analysis module is built-in with a traditional Chinese medicine knowledge base and a machine learning model, the traditional Chinese medicine knowledge base contains the attributes of thousands of traditional Chinese medicines, conventional decoction rules, medicinal material origin information, processing technology parameters, quality difference data of different batches of medicinal materials and compatibility contraindications, and is connected with a cloud database through a network for real-time updating of the latest technological parameters and efficacy verification data of traditional Chinese medicine research results, clinical decoction optimization cases.

[0006] Preferably, the intelligent process scheme formulation module is built-in with an AI process parameter generation model and a special decoction method adaptation algorithm, the AI process parameter generation model is based on the traditional Chinese medicine knowledge base and the machine learning model, and generates an individualized decoction scheme automatically according to the input prescription and special marking of special decoction methods.

[0007] Preferably, the multi-modal perception data acquisition module comprises a physicochemical index sensing unit and a machine vision unit; the physicochemical index sensing unit acquires physicochemical data in real time through a temperature sensor, a pressure sensor, a liquid level sensor, a pH sensor, a conductivity sensor and an online near-infrared spectrum monitoring device; and the machine vision unit monitors the liquid surface boiling state, the amount of foam, the morphological change of medicinal materials, the color change of medicinal liquid and the expansion degree of medicinal residue in real time through a high-definition camera.

[0008] Preferably, the dynamic regulation module is built-in with a model predictive control (MPC) algorithm, simultaneously receives sensing data, compares with a preset process curve, and adjusts the boiling fire and pressure in real time by using the model predictive control (MPC) algorithm.

[0009] Preferably, the dynamic regulation module dynamically regulates the fire by calculating the real-time heating power, and the calculation formula is:

[0010] In the formula, represents the real-time heating power at time t, represents the set temperature, represents the current real-time temperature, represents the proportional coefficient, represents the integral coefficient, represents the differential coefficient.

[0011] Preferably, the equipment management module comprises a multi-section parameter control unit and a dosage control unit; the multi-section parameter control unit adjusts the simmering, boiling and holding through electromagnetic heating, and switches the multiple modes of normal pressure, pressurization and depressurization through a pressure control valve; the dosage control unit is built-in with a rear-down medicinal material bin, a high-precision water pump and a flowmeter, the rear-down medicinal material bin sets the dosing time according to the preset rear-down medicinal material dosing time node in the individualized decoction scheme, realizes the precise control of the water addition amount through the high-precision water pump and the flowmeter, and calculates the estimated evaporation amount in the decoction process and the total water addition amount .

[0012] Preferably, the dosage control unit first calculates the estimated evaporation amount in the decoction process , and the calculation formula is:

[0013] In the formula, represents the estimated evaporation amount in the decoction process, represents the evaporation coefficient, represents the current real-time temperature, represents the ambient temperature, represents the cumulative decoction time, represents the medicinal liquid surface area. The total water addition amount is calculated, and the calculation formula is:

[0014] In the formula, W represents the total water addition amount, k represents the water absorption coefficient of medicinal materials, W represents the total mass of decoction pieces, and µ represents the target decoction liquid yield, W represents the estimated water evaporation amount during the decoction process.

[0015] Preferably, the intelligent judgment module of the decoction endpoint fuses an algorithm to calculate a comprehensive judgment index of the decoction endpoint to judge the optimal decoction endpoint, and the calculation formula is:

[0016] In the formula, represents the change rate of the effective component concentration with time, and represents the diffusion rate constant, represents the saturation concentration of the effective component under the current temperature and pressure, represents the real-time concentration of the effective component at t.

[0017] Preferably, the self-cleaning and sterilization maintenance module comprises a traditional Chinese medicine filtering unit, a sterilization processing unit, a sub-packaging unit and a self-cleaning unit. The traditional Chinese medicine filtering unit is internally provided with a precision filter screen, and after the traditional Chinese medicine decoction is completed, the traditional Chinese medicine liquid is automatically removed from the dregs through the precision filter screen; The sterilization processing unit irradiates the filtered traditional Chinese medicine liquid with ultraviolet light with a wavelength of 254 nm for sterilization processing, and the processing time is 30-45 seconds; The sub-packaging unit sub-packages the traditional Chinese medicine liquid into a sterile packaging bag according to the set dose through a quantitative pump and a peristaltic pump after sterilization processing, and automatically seals the packaging bag; The self-cleaning unit automatically starts the self-cleaning program for cleaning after the traditional Chinese medicine is sub-packaged.

[0018] Compared with the prior art, the present application provides an unmanned decoction system for traditional Chinese medicine with artificial intelligence precise control, which has the following beneficial effects: 1. The present application evaluates the deviation trend of the current temperature and the target temperature by calculating the real-time heating power at t , and then dynamically controls the firepower, so that the temperature curve closely follows the preset process path, and finally achieves the effect of shortening the decoction period and improving the effective component dissolution rate.

[0019] 2. The present application controls the temperature and pressure in the multi-stage decoction process through the multi-stage parameter control unit, sets the dosing time through the dosing time node set by the dosing control unit, to realize the precise control of the water addition amount, and calculates the total water addition amount​​​ (mL) to determine whether the medicinal material dissolution and subsequent concentration requirements are met, and through the coordinated work of the multi-section parameter control unit and the dosage control unit, precise execution and material balance control of the decoction process are achieved, thereby reducing human operation errors and improving the stability of medicinal liquid quality.

[0020] 3、 The present application calculates the change rate of the effective ingredient concentration with time The calculation is used in the dynamic threshold determination logic of the dynamic control module to objectively quantify the decoction endpoint, avoiding the subjectivity of traditional experience-based judgment, and finally achieving the effects of improving the consistency of medicinal effects and reducing excessive decoction. BRIEF DESCRIPTION OF DRAWINGS

[0021] Figure 1 The flowchart of the decoction system of the present application. DETAILED DESCRIPTION

[0022] The technical solutions in the embodiments of the present application will be described clearly and completely below with reference to the drawings in the embodiments of the present application. Obviously, the described embodiments are only a part of the embodiments of the present application, rather than all the embodiments. Based on the embodiments in the present application, all other embodiments obtained by those skilled in the art without creative labor fall within the scope of protection of the present application.

[0023] Please refer to Figure 1 An unmanned decoction system of traditional Chinese medicine with artificial intelligence precise quantity control, comprising a prescription data analysis module, an intelligent process scheme formulation module, a multi-modal sensing data acquisition module, a dynamic control module, an equipment management module, an intelligent decoction endpoint judgment module, a self-cleaning and sterilization maintenance module, a man-machine interaction module and a safety warning module; The prescription data analysis module is responsible for receiving and understanding the prescription information; The intelligent process scheme formulation module generates individualized decoction schemes based on the prescription information; The multi-modal sensing data acquisition module acquires full-dimensional data of the decoction process in real time; The dynamic control module monitors the full-dimensional data of the decoction process online and makes real-time dynamic adjustments to the decoction scheme; The equipment management module is responsible for managing the heating power, pressure adjustment, medicinal material feeding, water injection, medicinal liquid filtration and sub-packaging operations of the decoction equipment; The intelligent decoction endpoint judgment module fuses and calculates the effective ingredient concentration change curve measured by near-infrared spectroscopy, the temperature and pressure curve and the machine vision features (liquid level state, medicinal material morphology) to determine the endpoint. The self-cleaning and sterilization maintenance module performs omnibearing cleaning and sterilization on the inner container, pipeline and liquid channel of the equipment through cleaning in place (CIP), ultraviolet sterilization and high-temperature steam disinfection, so as to ensure long-term stable operation of the equipment and avoid cross contamination of liquid medicine. The man-machine interaction module provides an operation interface, displays the current prescription, progress and real-time data curve, and supports remote monitoring of the decoction state, receiving of completion notification and viewing of historical records through an App / mini program. The safety warning module monitors abnormal conditions such as dry burning, pot overflow, excessive pressure and sensor failure in real time, and when the module detects the above faults, automatically starts a safety program (such as stopping heating, pressure relief), and at the same time, performs hierarchical warning prompts through an audible and visual alarm and a pop-up window on the operation interface.

[0024] Advantages: through the establishment of the prescription data analysis module, the process scheme intelligent formulation module, the multi-modal sensing data acquisition module, the dynamic regulation and control module, the equipment management module, the decoction endpoint intelligent judgment module, the self-cleaning and sterilization maintenance module, the man-machine interaction module and the safety warning module, the prescription data analysis module realizes intelligent analysis of prescription information, the process scheme intelligent formulation module realizes automatic generation of individualized decoction scheme, the multi-modal sensing data acquisition module realizes real-time data acquisition in the whole process, the dynamic regulation and control module realizes online correction of process, the equipment management module realizes accurate execution of firepower, pressure, water quantity, feeding and sub-packaging, the decoction endpoint intelligent judgment module realizes objective and quantitative determination of the decoction endpoint, the self-cleaning and sterilization maintenance module realizes no cross contamination in the equipment, the man-machine interaction module realizes convenient operation on the local and cloud ends, the safety warning module realizes active defense against multiple faults, and finally the system solves the problems of lack of individualized decoction capability, strong subjectivity of decoction endpoint determination, and insufficient flexibility and stability of the existing automatic decoction equipment.

[0025] The prescription data analysis module is internally provided with a traditional Chinese medicine knowledge base and a machine learning model, the traditional Chinese medicine knowledge base contains attributes (such as medicinal properties, active ingredients, solubility and volatility) of thousands of traditional Chinese medicines, conventional decoction rules (first decoction, later addition, bag decoction and melting, etc.), information of medicinal material producing areas, processing technology parameters, quality difference data of different batches of decoction pieces and compatibility contraindications, and is connected with a cloud database through a network to update traditional Chinese medicine research results, latest process parameters of clinical decoction optimization cases and efficacy verification data in real time.

[0026] Through the combination of the above traditional Chinese medicine knowledge base and machine learning model, deep analysis of prescription information and intelligent recommendation of process parameters are realized, and the scientificity and repeatability of the decoction process are improved.

[0027] The intelligent process planning module incorporates an AI-powered process parameter generation model and a special decoction method adaptation algorithm. Based on a traditional Chinese medicine knowledge base and machine learning model, the AI ​​model automatically generates a personalized decoction plan according to the input prescription (herb types, dosages, and efficacy), and specifically labels herbs requiring special decoction methods. This plan includes total water volume, soaking time, heat / temperature and pressure at each stage, estimated decoction time, timing of adding later-added herbs, time for adding earlier-added herbs, filtration method for herbs to be wrapped, heating temperature and timing of mixing in the softened herbs, concentration rate of the decoction, standard dosage for dispensing, and cleaning and sterilization procedures. Under the given conditions, the aforementioned machine learning model employs a hybrid architecture of Gradient Boosting Decision Tree (GBDT) and Convolutional Neural Network (CNN). The GBDT is used to process structured data such as medicinal herb properties, while the CNN is used to analyze near-infrared spectral feature maps. For data encoding, the properties and meridian tropism of medicinal herbs are transformed into One-Hot vectors, dosages are standardized and normalized, and a knowledge graph adjacency matrix is ​​constructed for compatibility issues. During the model training phase, supervised learning is performed using over 100,000 valid cases from a hospital's historical decoction database, and hyperparameters are optimized through cross-validation. The final model achieves R on the test set. 2 The prediction accuracy is 0.92.

[0028] By using the aforementioned AI process parameter generation model and special frying method adaptation algorithm, the automated arrangement of complex frying processes can be achieved, realizing a precise control effect that takes into account both traditional experience and modern science.

[0029] The multimodal perception data acquisition module includes a physicochemical index sensing unit and a machine vision unit; The physicochemical index sensing unit collects physicochemical data in real time through temperature sensors, pressure sensors, liquid level sensors, pH sensors, conductivity sensors, and near-infrared (NIR) spectroscopy online monitoring equipment. This includes real-time decoction temperature (°C), pressure inside the pot (MPa), liquid level (cm), pH value, conductivity (mS / cm), active ingredient concentration (mg / mL), water absorption rate of medicinal materials (%), evaporation water volume (mL), heating power (W), and density of the concentrated liquid (g / cm³). 3 The parameters (%) and residual moisture of the dregs are used to dynamically adjust process parameters and assist in determining the decoction endpoint and evaluating the quality of the liquid. The machine vision unit uses a high-definition camera to monitor the boiling state of the liquid surface, the amount of foam, the changes in the shape of the medicinal materials, the color changes of the liquid (RGB value), and the degree of expansion of the dregs in real time, in order to help determine the decoction process.

[0030] By integrating the above-mentioned physicochemical indicators with machine vision data, a comprehensive digital representation of the decocting process can be achieved, providing high-confidence data support for dynamic regulation.

[0031] The dynamic regulation module is built-in model predictive control (MPC) algorithm, receives the sensing data, compares with the preset process curve, and uses the model predictive control (MPC) algorithm to adjust the decoction fire and pressure in real time, so as to ensure that the decoction process is always on the optimal path.

[0032] The dynamic regulation module dynamically regulates the fire by calculating the real-time heating power, and the calculation formula is:

[0033] In the formula, represents the real-time heating power (W) at time t, represents the set temperature (℃), represents the current real-time temperature (℃), represents the proportional coefficient (value range 5-15, used for fast response to temperature deviation), and the coefficient selection rules are as follows: According to the current temperature deviation absolute value, it is adaptively selected when |>10℃, take 12-15 to quickly heat up; when 5℃≤ |≤10℃, take 8-11; when |<5℃, take 5-7 to stabilize the temperature control; According to the static deviation duration adjustment, when the deviation lasts more than 30 seconds, take 0.7-1.0 to strengthen the integral compensation; when the deviation lasts for 10-30 seconds, take 0.4-0.6; when the deviation lasts for <10 seconds, take 0.1-0.3; According to the temperature change rate adjustment, when |d( ) / dt|>2℃ / s, take 0.07-0.1 to suppress overshoot; when 0.5℃ / s≤|d ( ) / dt|≤2℃ / s, take 0.04-0.06; when |d ( ) / dt|<0.5℃ / s, take 0.01-0.03, represents the integral coefficient (value range 0.1-1, used for eliminating static temperature deviation), represents the differential coefficient (value range 0.01-0.1, used for suppressing temperature overshoot), and the above coefficients are dynamically determined according to the decoction stage and the type of medicinal materials by table lookup method: the system presets a five-stage parameter table of soaking-warming-boiling-keeping-concentrating, and the values of each stage , , are adjusted in a ladder form based on the medicinal material category (root and stem, flower and grass, mineral, and glue); specifically, in the warming stage take 12-15 to quickly heat up; in the boiling stage reduce to 8-10 to prevent violent boiling; in the glue type medicinal material keeping stage Take 0.8-1.0 to eliminate static bias, and take 0.1-0.3 for herbal medicine to avoid over-cooking; in the concentration stage of root and stem type medicinal materials Take 0.05-0.1 to inhibit temperature overshoot, and take 0.01-0.03 for other types; the intermediate value is used for initialization when the system is first run, and subsequent fine tuning is performed through incremental PID parameter self-tuning algorithm according to the feedback of the decoction effect.

[0034] The advantage is: by calculating the real-time heating power at t time To evaluate the deviation trend of the current temperature and the target temperature, and then dynamically regulate the firepower, so that the temperature curve closely follows the preset process path, and finally achieves the effect of shortening the decoction period and improving the dissolution rate of effective ingredients.

[0035] The equipment management module includes a multi-section parameter control unit and a dosage control unit; The multi-section parameter control unit adjusts the simmering, high heat and holding through electromagnetic heating, and switches between normal pressure, pressurization (high heat quick decoction) and depressurization (simmering slow cooking) through the pressure control valve; The dosage control unit is built-in with a rear-down medicinal material bin, a high-precision water pump and a flowmeter. The rear-down medicinal material bin sets the dosing time according to the preset rear-down medicinal material dosing time node in the individualized decoction scheme, and realizes precise control of the water addition amount through the high-precision water pump and flowmeter, and calculates the estimated evaporation amount and the total water addition amount (mL) to determine whether the medicinal material dissolution and subsequent concentration requirements are met.

[0036] Through the cooperative work of the above multi-section parameter control unit and dosage control unit, the precise execution and material balance control of the decoction process are realized, and the effects of reducing human operation errors and improving the stability of medicinal liquid quality are achieved.

[0037] The dosage control unit first calculates the estimated evaporation amount in the decoction process to determine whether the automatic water replenishment program needs to be started, and assists in determining the total water addition amount (mL), and the calculation formula is:

[0038] In the formula, represents the estimated evaporation amount (mL) in the decoction process, represents the evaporation coefficient (value 0.002-0.005 mL / (℃・s・m 2 ), calibrated by heating power and pressure in the pot), represents the current real-time temperature (℃), represents the ambient temperature (℃), represents the cumulative decoction time (s), Liquid surface area (m 2 , determined by the size of the decoction pot), estimated evaporated water volume The calculation is dynamic: during the decoction process, the system samples the current temperature at fixed time intervals (e.g., every second) and recalculates , thereby updating the total water addition volume V in real time, and the evaporation coefficient is calibrated through experiments, tested under different heating powers and pressures, and a relationship table is established between power and pressure for dynamic adjustment of the value; The total water addition volume (mL) is then calculated, with the formula:

[0039] In the formula, represents the total water addition volume (mL), k represents the water absorption coefficient of medicinal materials (mL / g, obtained by weighted average of medicinal material texture, 1.5-2.0 for root and stem, 2.0-2.5 for flowers and grasses, weighted average calculation method: ① determine the weight according to the mass proportion of each medicinal material in the prescription, single medicinal material weight = / m (the mass of the i-th medicinal material, m is the total mass of the decoction pieces); ② determine the basic water absorption coefficient according to the medicinal material texture classification: 1.5-2.0 mL / g for root and stem, 2.0-2.5 mL / g for flowers and grasses, 1.8-2.2 mL / g for fruits, 1.0-1.2 mL / g for minerals, 2.3-2.8 mL / g for animals; ③ weighted average formula k=Σ( × ) (the basic water absorption coefficient corresponding to the i-th medicinal material); example: prescription contains root and stem medicinal material A (mass 100g, =1.8), flowers and grasses medicinal material B (mass 50g, =2.2), then =100 / (100+50)=0.67, =50 / (100+50)=0.33, k=0.67×1.8+0.33×2.2=1.93 mL / g), represents the total mass of decoction pieces (g), µ represents the target decoction yield (value 0.7-0.8, i.e., 70%-80%), represents the estimated evaporated water volume (mL) during the decoction process.

[0040] Advantages: through total water addition volume ​The calculation is used in the dynamic water replenishment logic of the dynamic control module to dynamically balance the water volume during the decoction process, thereby ensuring that the medicinal materials are fully soaked and the final concentration of the decoction meets the standard, so that the system can improve the extraction rate of effective ingredients and reduce the waste of decoction.

[0041] The intelligent decoction endpoint determination module uses real-time measurements of the effective component concentration change curve from infrared spectroscopy, temperature and pressure curves, and machine vision features to calculate a comprehensive decoction endpoint determination index through a fusion algorithm, thereby determining the optimal decoction endpoint. The calculation formula is as follows:

[0042] In the formula, The value represents the rate of change of the concentration of the active ingredient over time (mg / (mL·s)), and α represents the diffusion rate constant (s). -1 (Positively correlated with the particle size of the medicinal materials and the decoction temperature) This indicates the saturation concentration (mg / mL) of the active ingredient under the current temperature and pressure. This represents the real-time concentration (mg / mL) of the active ingredient at time t; when When the value is less than 0.001, the boiling point is considered reached.

[0043] The advantage is that it allows us to measure the rate of change of the concentration of the active ingredient over time. The calculation is used in the dynamic threshold determination logic of the dynamic control module to make an objective quantitative judgment on the decoction endpoint, avoiding the subjectivity of traditional experience judgment, and finally achieving the effect of improving the consistency of efficacy and reducing over-decoction.

[0044] The self-cleaning and sterilization maintenance module includes a traditional Chinese medicine filtration unit, a sterilization unit, a dispensing unit, and a self-cleaning unit. The herbal medicine filtration unit has a built-in precision filter screen. After the herbal medicine is decocted, the liquid is automatically passed through the precision filter screen to remove the dregs. The sterilization unit sterilizes the filtered Chinese medicine liquid by irradiating it with ultraviolet light with a wavelength of 254nm for 30 to 45 seconds. After sterilization, the dispensing unit dispenses the traditional Chinese medicine liquid into sterile packaging bags according to the set dosage using a metering pump and a peristaltic pump, and then automatically seals the bags. After the Chinese medicine is packaged, the self-cleaning unit automatically starts the self-cleaning program. Through the spray arm and high-temperature steam, it automatically and thoroughly cleans the inner tank and pipes after each decoction to remove residual medicine and dregs, avoid cross-contamination of subsequent batches of medicine, and maintain the cleanliness of the equipment.

[0045] Through the above-mentioned full-process automatic processing of filtering, sterilization, sub-packaging and self-cleaning, high-purity sub-packaging of medicinal liquid and long-term stable operation of the equipment are realized, and the effect of meeting the GMP standard and ensuring the safety of medication is achieved.

[0046] The process of the decoction system of the present application is applied to actual scenes as follows: Example 1 A certain Grade III-A hospital pharmacy needs to process 200 personalized prescriptions daily, involving resolving agents, tonifying agents and other various dosage forms. The system automatically identifies the processing requirements of each medicine through the prescription data analysis module, generates differentiated decoction schemes through the process scheme intelligent generation module, and adjusts the firepower according to the real-time temperature and pressure through the dynamic control module. Finally, 200 medicinal liquids are processed in a single day, with a qualified rate of 99.6%, which is 4 times more efficient than traditional manual decoction.

[0047] Example 2 A certain chain of Chinese medicine clinics provides customized paste prescription services for chronic disease patients. The system monitors the melting state of donkey-hide gelatin and other gelatinous medicinal materials through the multi-modal sensing data acquisition module, accurately determines the paste collection time through the near-infrared spectrum and visual features combined with the decoction endpoint intelligent judgment module, and ensures that each batch of paste has no cross-contamination through the self-cleaning and sterilization maintenance module. Clinical verification shows that the fluctuation range of the effective ingredient content of the paste is controlled within ±3%.

[0048] Example 3 An internet hospital provides remote decoction services for patients in remote areas. After the patient uploads the electronic prescription to the system, the human-computer interaction module generates a real-time decoction progress push, and the safety warning module detects the packaging integrity before transportation. The system realizes full-process unmanned operation from receiving to distribution, with an average delivery time of 6 hours and a patient satisfaction rate of 98%.

[0049] Summary: Example 1 verifies the efficiency and stability of the system in large-scale prescription processing; Example 2 verifies the precise control ability of the system in complex dosage form (such as paste) preparation; and Example 3 verifies the feasibility of full-process unmanned operation of the system in the remote medical scene, thus confirming that the decoction system of the present application can be applied to various scenes in medical institutions, pharmaceutical enterprises and internet medical platforms.

[0050] The model combination method is a conventional means: the hybrid architecture of Gradient Boosting Decision Tree (GBDT) and Convolutional Neural Network (CNN) is a known, conventional technical solution for processing multi-modal data in the field of artificial intelligence. Among them, GBDT is responsible for processing structured table data (such as medicinal material types, dosages), and CNN is responsible for processing unstructured atlas data (such as near-infrared spectra). This division and cooperation mode has been widely used in the prior art. The "combination" of them in specific implementation refers to processing different types of data during model training and prediction, and finally performing feature fusion at a certain layer (e.g., a fully connected layer) of the model, which does not need to be described repeatedly. Near-infrared spectrum data is not input at the initial development of the process scheme. The development of the initial scheme mainly depends on the structured data (medicinal material attributes, dosages, etc.) output by the prescription analysis module. The near-infrared spectrum data is collected in real time by the multi-modal perception data acquisition module after the start of the decoction process. These real-time data are provided to the dynamic control module and the intelligent judgment module for the end of decoction to monitor and dynamically adjust the started decoction process and determine the end point. This shows that the system processes different data at different stages.

[0051] The knowledge graph is part of the model input: constructing the incompatibility of compatibility as a knowledge graph adjacency matrix is a standard data preprocessing method, and the purpose is to convert the compatibility relationship of traditional Chinese medicine into a computer processable numerical feature. This adjacency matrix as a feature will be input into the AI model (especially GBDT) together with the One-Hot vector of medicinal materials, the normalized dosage, etc. for training and prediction. This method is very common in the fields of recommendation systems and medical data analysis.

[0052] The formula is a standard PID control variant: the formula is a standard discretization implementation form of the PID controller (proportional-integral-derivative) relied on by the model predictive control (MPC) algorithm in industrial control. The expression with integral and derivative cited in the review is the standard form in the continuous time domain. Among them, the integral term , here the integral is the historical total of the temperature deviation ( ) accumulated from the current control period to the current time t, and in the actual digital system implementation, this is achieved by accumulating the deviation value of each sampling period. In addition, regarding the derivative term , the derivative here represents the instantaneous change rate of the temperature deviation ( ) at the current time t. In the actual system, it is approximated by calculating the difference between the "current time deviation" and the "last sampling time deviation", and then dividing by the sampling time interval. The above operation is a general engineering practice in the field of automation control, and those skilled in the art can directly implement it based on their basic knowledge.

[0053] The seven core modules of the system and their connection relationships and data flow directions have been fully disclosed in the specification. From prescription analysis, scheme generation, data acquisition, dynamic control to endpoint determination, a clear closed-loop control system is formed, and there are actual cases. For example, near-infrared spectroscopy technology is used for substance composition identification and endpoint determination, which has been successfully applied in other fields (such as plastic classification) and in this field (Chinese medicine decoction). For another example, the application of large language models (LLM) and artificial intelligence technology in Chinese medicine knowledge processing and prescription recommendation is increasingly mature, which proves the technical rationality and realizability of the "prescription data analysis module" and the "process scheme intelligent formulation module" of the present application.

[0054] In the specification, clear and reasonable value ranges are given for key parameters (such as λ1, λ2, λ3, evaporation coefficient b, medicinal material water absorption coefficient k, etc.), and it is explained how these parameters are dynamically adjusted according to the decoction stage and medicinal material type.

[0055] Although embodiments of the present application have been shown and described, it will be understood by those having ordinary skill in the art that various changes, modifications, substitutions and alterations can be made therein without departing from the principles and spirit of the present application, the scope of which is defined by the appended claims and their equivalents.

Claims

1. A pre-controlled, artificial intelligence-based unmanned decoction system for precise dosage control of traditional Chinese medicine, characterized in that: It includes a prescription data analysis module, a process plan intelligent formulation module, a multimodal perception data acquisition module, a dynamic control module, an equipment management module, a decoction endpoint intelligent judgment module, a self-cleaning and sterilization maintenance module, a human-computer interaction module, and a safety early warning module; The prescription data parsing module is responsible for receiving and understanding prescription information; The intelligent process planning module generates personalized decoction plans based on prescription information; The multimodal sensing data acquisition module collects real-time data from all dimensions of the simmering process; The dynamic control module monitors all dimensions of data during the simmering process online and makes dynamic adjustments to the simmering plan in real time. The equipment management module is responsible for managing the heating power, pressure regulation, medicinal material feeding, water injection, liquid filtration, and dispensing operations of the decoction equipment. The intelligent determination module for the end point of decoction determines the end point by fusing the effective ingredient concentration change curve measured by near-infrared spectroscopy, the temperature and pressure curve, and machine vision features. The self-cleaning and sterilization maintenance module performs comprehensive cleaning and sterilization of the equipment's inner tank, pipes, and liquid channels through in-situ cleaning, ultraviolet sterilization, and high-temperature steam disinfection. The human-computer interaction module provides an operation interface that displays the current prescription, progress, and real-time data curves, and supports remote monitoring of the decoction status, receiving completion notifications, and viewing historical records via an App / mini-program; The safety early warning module monitors abnormal situations such as dry burning, overflow, excessive pressure, and sensor malfunction in real time. When the module detects the above faults, it automatically starts the safety program and provides tiered early warning prompts through audible and visual alarms and pop-up windows on the operation interface.

2. The unmanned decoction system for precise dosage control of traditional Chinese medicine using artificial intelligence according to claim 1, characterized in that: The prescription data parsing module has a built-in TCM knowledge base and machine learning model. The TCM knowledge base contains the properties of thousands of Chinese medicines, conventional decoction methods, information on the origin of medicinal materials, processing parameters, quality differences of different batches of decoction pieces, and contraindications. It is connected to a cloud database via the network to update the latest process parameters and efficacy verification data of TCM research results, clinical decoction optimization cases, etc. in real time.

3. The unmanned decoction system for precise dosage control of traditional Chinese medicine using artificial intelligence according to claim 1, characterized in that: The intelligent process planning module has a built-in AI process parameter generation model and a special decoction method adaptation algorithm. The AI ​​process parameter generation model is based on a traditional Chinese medicine knowledge base and a machine learning model. Based on the input prescription, it automatically generates a personalized decoction plan by specially annotating Chinese herbs with special decoction methods.

4. The unmanned decoction system for precise dosage control of traditional Chinese medicine using artificial intelligence according to claim 1, characterized in that: The multimodal sensing data acquisition module includes a physicochemical index sensing unit and a machine vision unit. The physicochemical index sensing unit collects physicochemical data in real time through temperature sensors, pressure sensors, liquid level sensors, pH sensors, conductivity sensors, and near-infrared spectroscopy online monitoring equipment. The machine vision unit monitors the boiling state of the liquid surface, foam volume, changes in the morphology of medicinal materials, changes in the color of the medicinal liquid, and the degree of expansion of the medicinal residue in real time through a high-definition camera.

5. The unmanned decoction system for precise dosage control of traditional Chinese medicine using artificial intelligence according to claim 1, characterized in that: The dynamic control module has a built-in Model Predictive Control (MPC) algorithm. It also receives sensor data, compares it with the preset process curve, and uses the MPC algorithm to adjust the cooking heat and pressure in real time.

6. The unmanned decoction system for precise dosage control of traditional Chinese medicine using artificial intelligence according to claim 1, characterized in that: The dynamic control module dynamically adjusts the heating power by calculating the real-time heating power, and the calculation formula is as follows: ； In the formula, This represents the real-time heating power at time t. Indicates the set temperature. This indicates the current real-time temperature. Represents the proportionality coefficient. Represents the integral coefficient. This represents the differential coefficient.

7. The unmanned decoction system for precise dosage control of traditional Chinese medicine using artificial intelligence according to claim 1, characterized in that: The equipment management module includes a multi-segment parameter control unit and a dosage control unit. The multi-segment parameter control unit uses electromagnetic heating to adjust the heat (low, high, and heat preservation), and a pressure control valve to switch between normal pressure, pressurized, and depressurized modes. The dosage control unit has a built-in post-addition herb bin, a high-precision water pump, and a flow meter. The post-addition herb bin sets the addition time according to the preset post-addition herb time nodes in the personalized decoction plan. The high-precision water pump and flow meter achieve precise control of the water addition and calculate the estimated water evaporation during the decoction process. With total water addition .

8. The unmanned decoction system for precise dosage control of traditional Chinese medicine using artificial intelligence according to claim 7, characterized in that: The dosage control unit first calculates the estimated amount of water to evaporate during the decocting process. The calculation formula is as follows: ； In the formula, This indicates the estimated amount of water that will evaporate during the frying or boiling process. Indicates the evaporation coefficient. This indicates the current real-time temperature. Indicates ambient temperature. Indicates the total cooking time. Indicates the surface area of ​​the liquid medicine; Then calculate the total water volume. The calculation formula is as follows: ； In the formula, This represents the total amount of water added, and k represents the water absorption coefficient of the medicinal material. The value represents the total mass of the medicinal slices, and µ represents the target yield of the decoction. This indicates the estimated amount of water that will evaporate during the frying or boiling process.

9. The unmanned decoction system for precise dosage control of traditional Chinese medicine using artificial intelligence according to claim 1, characterized in that: The intelligent determination module for the boiling endpoint uses a fusion algorithm to calculate a comprehensive determination index for the boiling endpoint to determine the optimal boiling endpoint. The calculation formula is as follows: ； In the formula, The value represents the rate of change of the concentration of the active ingredient over time, and α represents the diffusion rate constant. This indicates the saturation concentration of the active ingredient under the current temperature and pressure. This represents the real-time concentration of the active ingredient at time t.

10. The unmanned decoction system for precise dosage control of traditional Chinese medicine using artificial intelligence according to claim 1, characterized in that: The self-cleaning and sterilization maintenance module includes a traditional Chinese medicine filtration unit, a sterilization processing unit, a dispensing unit, and a self-cleaning unit. The herbal medicine filtration unit has a built-in precision filter screen, which automatically removes the dregs from the herbal medicine liquid after the decoction is completed. The sterilization unit sterilizes the filtered Chinese medicine liquid by irradiating it with ultraviolet light with a wavelength of 254nm for 30 to 45 seconds. After sterilization, the dispensing unit dispenses the traditional Chinese medicine liquid into sterile packaging bags according to the set dosage using a metering pump and a peristaltic pump, and then automatically seals the bags. After completing the dispensing of traditional Chinese medicine, the self-cleaning unit automatically starts the self-cleaning program.

Images