Milk production control method and system combined with digital twinning

Abstract

The application discloses a rice milk production control method and system combined with digital twinning, relates to the field of digital twinning, and comprises the following steps: based on digital twinning technology, a rice milk double simulation model is constructed, including a rice milk production simulation model and a rice milk application simulation model which are logically sequentially connected; process parameters of a production line are collected in real time to obtain a rice milk production dataset, the rice milk production dataset is input into the rice milk production simulation model, and a rice milk performance data matrix is obtained through virtual simulation; the rice milk performance data matrix is input into the rice milk application simulation model for quality defect early warning, reverse regulation is performed according to the early warning result, an optimization instruction is generated, and the optimization instruction is fed back to the production line to perform real-time rice milk production control. The application solves the technical problem that the existing rice milk production control only focuses on production link monitoring, leading to poor rice milk application adaptability, achieves the technical effect of accurately controlling the rice milk production process from the application effect, and further improves the rice milk application adaptability.

Description

TECHNICAL FIELD

[0001] The present application relates to the field of digital twinning, in particular to a rice milk production control method and system combined with digital twinning. BACKGROUND

[0002] Under the background of the continuous expansion and upgrading of the coffee beverage market, rice milk as a characteristic additive ingredient in coffee beverages, its production quality and application effect compatible with coffee directly determine the taste, flavor and quality of coffee beverages, and thus affect the competitiveness of the product in the market, so it is crucial to ensure the high-quality application of rice milk in coffee. At present, to solve the problem of rice milk production quality monitoring, the traditional production monitoring method is mainly used, that is, only the process parameters in the production process of rice milk are monitored, and whether the production is normal is judged according to the preset parameter range, so as to ensure the basic production quality of rice milk. However, this traditional method only focuses on the production link of rice milk and does not fully consider the actual effect of rice milk in coffee application, resulting in that the produced rice milk may have compatibility problems such as taste discordance and poor flavor when mixed with coffee, which cannot meet the needs of consumers for high-quality coffee beverages.

[0003] At present, in the related technology, the rice milk production control only focuses on the production link monitoring, which leads to the technical problem of poor application compatibility of rice milk. SUMMARY

[0004] The present application provides a rice milk production control method and system combined with digital twinning, which adopts a rice milk dual simulation model (including production and application simulation models) constructed based on digital twinning, real-time collection of production line process parameters input into the production simulation model to obtain a performance data matrix, input into the application simulation model to warn quality defects, reverse generation of optimization instructions and feedback to the production line to realize real-time control, etc. Technical means solve the technical problem that the existing rice milk production control only focuses on the production link monitoring, which leads to poor application compatibility of rice milk, and achieve the technical effect of accurately controlling the rice milk production process from the application effect, and thus improving the application compatibility of rice milk.

[0005] The present application provides a rice milk production control method combined with digital twinning, comprising: based on digital twinning technology, constructing a rice milk dual simulation model containing bidirectional mapping of physical entities and virtual models, the rice milk dual simulation model including logically sequentially connected rice milk production simulation model and rice milk application simulation model; real-time collection of process parameters of the production line to obtain a rice milk production data set, inputting the rice milk production data set into the rice milk production simulation model to obtain a rice milk performance data matrix through virtual simulation; inputting the rice milk performance data matrix into the rice milk application simulation model for quality defect warning, reverse regulation according to the warning result, generating optimization instructions, and feeding back the optimization instructions to the production line to execute real-time rice milk production control.

[0006] In a possible implementation, the construction of the rice milk production simulation model performs the following processing: constructing an enzyme hydrolysis reaction kinetics prediction sub-model, synchronously outputting the degree of saccharification and the reducing sugar content through a temperature-time-enzyme concentration three-dimensional relationship matrix; constructing a homogenization effect prediction sub-model, synchronously outputting the fat particle size distribution and the emulsion stability index through the interactive calculation of homogenization pressure and flow; based on the output of the enzyme hydrolysis reaction kinetics prediction sub-model and the homogenization effect prediction sub-model, establishing a dynamic viscosity calculation module; establishing a fusion output layer, integrating the outputs of the enzyme hydrolysis reaction kinetics prediction sub-model, the homogenization effect prediction sub-model and the dynamic viscosity calculation module into a rice milk performance data matrix.

[0007] In a possible implementation, the dynamic viscosity calculation module is further established to perform the following processing: presetting a saccharification degree limit value, when the saccharification degree output by the enzyme hydrolysis reaction kinetics prediction sub-model is greater than the saccharification degree limit value, activating high-viscosity compensation, and compensating the dynamic viscosity calculation module.

[0008] In a possible implementation, the construction of the rice milk application simulation model performs the following processing: constructing a stability adaptation prediction sub-model according to the influence of the fat particle size distribution and the emulsion stability index on the emulsion stability of rice milk coffee mixing; constructing a sensory adaptation prediction sub-model according to the influence of the degree of saccharification and the reducing sugar content on the sweetness balance of rice milk coffee mixing; constructing a function adaptation prediction sub-model according to the influence of viscosity and the fat particle size distribution on the performance of rice milk in coffee drawing; connecting the stability adaptation prediction sub-model, the sensory adaptation prediction sub-model and the function adaptation prediction sub-model in parallel to construct a rice milk application simulation model.

[0009] In a possible implementation, the rice milk performance data matrix is input into the rice milk application simulation model for quality defect early warning, and according to the early warning result, an optimization instruction is generated to perform the following processing: when the stability adaptation prediction sub-model detects that the emulsion stability index is less than a preset first stability index limit value, a first quality defect early warning is generated; when the sensory adaptation prediction sub-model detects that the sweetness deviation is greater than a preset sweetness deviation limit value, a second quality defect early warning is generated; when the function adaptation prediction sub-model detects that the drawing foam decay rate is greater than a preset decay rate standard value, a third quality defect early warning is generated; according to the trigger state of the first quality defect early warning, the second quality defect early warning and the third quality defect early warning, the optimization instruction is generated through reverse regulation.

[0010] In a possible implementation, according to the triggering state of the first quality defect early warning, the second quality defect early warning and the third quality defect early warning, reverse regulation is performed, an optimization instruction is generated, and the following processing is performed: when and only when one prediction sub-model triggers an early warning, according to the first quality defect early warning or the second quality defect early warning or the third quality defect early warning, a corresponding first optimization instruction or a second optimization instruction or a third optimization instruction is generated; when greater than or equal to two prediction sub-models trigger an early warning at the same time, multi-parameter collaborative optimization is performed, and a collaborative optimization instruction is generated.

[0011] In a possible implementation, multi-parameter collaborative optimization is performed, a collaborative optimization instruction is generated, and the following processing is performed: according to the difference of the coffee mode, a dynamic weight coefficient is assigned to each prediction sub-model, wherein the sum of the dynamic weight coefficients of the respective prediction sub-models is 1; an instruction intensity factor of each prediction sub-model is calculated according to the dynamic weight coefficient; and multi-parameter collaborative optimization is performed through a weighted fusion algorithm to synthesize the instruction intensity factors, and a collaborative optimization instruction is generated.

[0012] In a possible implementation, according to the difference of the coffee mode, a dynamic weight coefficient is assigned to each prediction sub-model, and the following processing is performed: if it is a deep roasting coffee mode, the dynamic weight coefficient is assigned in a first order, wherein the first order is: the sensory adaptation prediction sub-model weight > the function adaptation prediction sub-model weight > the stability adaptation prediction sub-model weight; if it is a light roasting coffee mode, the dynamic weight coefficient is assigned in a second order, wherein the second order is: the function adaptation prediction sub-model weight > the stability adaptation prediction sub-model weight > the sensory adaptation prediction sub-model weight; when the emulsification stability index in the first quality defect early warning is less than a preset second stability index limit value, a stability crisis mode is triggered, and the first order and the second order are forcibly adjusted to a third order, wherein the second stability index limit value is less than the first stability index limit value, and the third order is: the stability adaptation prediction sub-model weight > the function adaptation prediction sub-model weight > the sensory adaptation prediction sub-model weight.

[0013] In a possible implementation, the following processing is further performed: accessing a barista skill level database, and dynamically adjusting the size of the attenuation rate standard value according to the operator level, wherein the level is positively correlated with the size of the attenuation rate standard value.

[0014] The application also provides a rice milk production control system combined with digital twinning, comprising: a rice milk dual simulation model construction module, configured to construct a rice milk dual simulation model comprising a bidirectional mapping between a physical entity and a virtual model based on a digital twinning technology, the rice milk dual simulation model comprising a rice milk production simulation model and a rice milk application simulation model logically and sequentially connected; a production simulation module, configured to collect process parameters of a production line in real time to obtain a rice milk production dataset, input the rice milk production dataset into the rice milk production simulation model, and obtain a rice milk performance data matrix through virtual simulation; and a reverse regulation and optimization module, configured to input the rice milk performance data matrix into the rice milk application simulation model to perform quality defect early warning, perform reverse regulation according to a warning result, generate an optimization instruction, and feed back the optimization instruction to the production line to perform real-time rice milk production control.

[0015] The rice milk production control method and system combined with digital twinning provided in the application first construct a rice milk dual simulation model comprising a bidirectional mapping between a physical entity and a virtual model based on a digital twinning technology, the rice milk dual simulation model comprising a rice milk production simulation model and a rice milk application simulation model logically and sequentially connected, then collect process parameters of a production line in real time to obtain a rice milk production dataset, input the rice milk production dataset into the rice milk production simulation model, and obtain a rice milk performance data matrix through virtual simulation, and finally input the rice milk performance data matrix into the rice milk application simulation model to perform quality defect early warning, perform reverse regulation according to a warning result, generate an optimization instruction, and feed back the optimization instruction to the production line to perform real-time rice milk production control. The technical effect of accurately regulating the rice milk production process from the application effect and improving the rice milk application adaptability is achieved. BRIEF DESCRIPTION OF DRAWINGS

[0016] In order to more clearly illustrate the technical solutions of the embodiments of the application, the drawings of the embodiments of the application will be briefly introduced below. In the present application, a flowchart is used to illustrate the operations performed by the system according to the embodiments of the application. It should be understood that the foregoing or the following operations are not necessarily performed in sequence. On the contrary, various steps can be processed in reverse order or simultaneously as needed. Meanwhile, other operations can be added to these processes, or a step or several steps of operation can be removed from these processes.

[0017] Figure 1 The flowchart of the rice milk production control method combined with digital twinning provided in the embodiments of the application is shown.

[0018] Figure 2 The structural diagram of the rice milk production control system combined with digital twinning provided in the embodiments of the application is shown.

[0019] Marker explanation: rice milk dual simulation model construction module 10, production simulation module 20, reverse regulation and optimization module 30. DETAILED DESCRIPTION

[0020] The above description is only a summary of the technical solutions of the present application. In order to make the technical means of the present application more clear, the present application can be implemented according to the content of the description, and in order to make the above and other purposes, characteristics and advantages of the present application more obvious and easy to understand, the following specific embodiments of the present application are described.

[0021] In order to make the purposes, technical solutions and advantages of the present application more clear, the following will combine the drawings to make further detailed description of the present application. The described embodiments should not be regarded as limitation of the present application. All other embodiments obtained by those skilled in the art without creative labor are within the scope of protection of the present application.

[0022] In the following description, "some embodiments" are related to a subset of all possible embodiments, but it can be understood that "some embodiments" can be the same subset or different subset of all possible embodiments, and can be combined with each other without conflict. The term "first\second" is only to distinguish similar objects, and does not represent the specific order of the object. The terms "include" and "have" and any variations, are intended to cover non-exclusive inclusion, for example, a process, method, system, product or server including a series of steps or units does not have to be limited to those steps or units clearly listed, but can include other steps or modules not clearly listed or inherent to these processes, methods, products or devices. Unless otherwise defined, all technical and scientific terms used herein have the same meaning as understood by those skilled in the art to which the present application belongs. The terms used herein are only for the purpose of describing the embodiments of the present application.

[0023] The embodiments of the present application provide a rice milk production control method combined with digital twinning, as shown in the following Figure 1 The method comprises the following steps:

[0024] In step S100, based on digital twinning technology, a rice milk dual simulation model containing bidirectional mapping of physical entity and virtual model is constructed, and the rice milk dual simulation model comprises a rice milk production simulation model and a rice milk application simulation model which are logically sequentially connected.

[0025] Specifically, digital twinning is a technology that creates a virtual model of a physical entity and enables two-way mapping and data interaction between the physical entity and the virtual model. It can reflect the state of the physical entity in real time and guide the operation of the physical entity through simulation and optimization of the virtual model. The rice milk dual simulation model is a virtual model that includes a rice milk production simulation model and a rice milk application simulation model. It is used to simulate the production process and application effect of rice milk. The rice milk production simulation model is a virtual model used to simulate the production process of rice milk, including raw material processing, mixing, homogenization, sterilization, and other processes. The rice milk application simulation model is a virtual model used to simulate the application effect of rice milk, including performance indicators such as taste.

[0026] Three-dimensional modeling of each device and link of the rice milk production line, including raw material processing, mixing, homogenization, sterilization, and filling. In the digital twinning modeling software (such as Siemens NX, ANSYS Twin Builder, etc.), according to the three-dimensional model of the physical entity, the corresponding virtual model is constructed, which includes the rice milk production simulation model and the rice milk application simulation model. In the digital twinning platform, the two-way mapping relationship between the physical entity and the virtual model is set to ensure that the real-time data of the physical entity can be synchronized to the virtual model, and the optimization instructions of the virtual model can be fed back to the physical entity.

[0027] For example, use Siemens NX to model the homogenizer of the rice milk production line, including the shell, internal structure, motor, and other components of the homogenizer. In ANSYS Twin Builder, according to the physical model of the homogenizer, a virtual model is constructed, and virtual sensors for homogenization pressure, flow, and other parameters are set. Through the OPC UA protocol, the real-time pressure data of the homogenizer is transmitted from the physical entity to the virtual model, and the optimization instructions in the virtual model are set to be fed back to the physical entity through the same protocol.

[0028] In one possible implementation, the construction of the rice milk production simulation model further includes step S110 of constructing an enzyme hydrolysis reaction kinetics predictor model that synchronously outputs the degree of saccharification and the content of reducing sugar through a temperature-time-enzyme concentration three-dimensional relationship matrix. Specifically, an experimental scheme is designed, and the degree of saccharification and the content of reducing sugar under different temperatures (such as 40°C, 50°C, and 60°C), times (such as 10 minutes, 20 minutes, and 30 minutes), and enzyme concentrations (such as 0.1%, 0.2%, and 0.3%) are collected. The degree of saccharification refers to the degree of conversion of starch into sugar in the enzyme hydrolysis reaction. The content of reducing sugar refers to the content of reducing sugar generated in the enzyme hydrolysis reaction. Experimental data are collected by using a spectrometer or the like, and the degree of saccharification and the content of reducing sugar are recorded. According to the experimental data, a temperature-time-enzyme concentration three-dimensional relationship matrix is constructed by using MATLAB or Python, and a prediction model is established by regression analysis or the like. The prediction accuracy of the model is verified by experimental data, and it is ensured that the model can accurately predict the degree of saccharification and the content of reducing sugar. For example, the Michaelis-Menten kinetics model is selected as a basic model, and the parameters of the basic model are fitted according to the experimental data to obtain the enzyme hydrolysis reaction kinetics predictor model.

[0029] In step S120, a homogenization effect predictor model is constructed, which synchronously outputs the fat particle size distribution and the emulsion stability index through the interactive calculation of the homogenization pressure and the flow rate. Specifically, an experimental scheme is designed, and the fat particle size distribution and the emulsion stability index under different homogenization pressures (such as 10 MPa, 20 MPa, and 30 MPa) and flow rates (such as 10 L / min, 20 L / min, and 30 L / min) are collected. The fat particle size distribution refers to the particle size distribution range of the fat particles in the homogenization process. The emulsion stability index is an index for measuring the stability of the emulsion, and the higher the value, the better the stability. Experimental data are collected by using a particle size analyzer or the like, and the fat particle size distribution and the emulsion stability index are recorded. According to the experimental data, a homogenization pressure and flow rate interactive calculation model is constructed by using MATLAB or Python, and a prediction model is established by regression analysis or the like. The prediction accuracy of the model is verified by experimental data, and it is ensured that the model can accurately predict the fat particle size distribution and the emulsion stability index. For example, a model based on the Reynolds number and the Weber number is selected as a basic model, and the parameters of the basic model are fitted according to the experimental data to obtain the homogenization effect predictor model.

[0030] Step S130, based on the output of the enzyme hydrolysis reaction kinetics prediction sub-model and the homogenization effect prediction sub-model, a dynamic viscosity calculation module is established. Specifically, an experimental scheme is designed, and dynamic viscosity data of rice milk under different process parameters (such as temperature, enzyme concentration, homogenization pressure, etc.) conditions are collected. Viscosity is used to characterize the viscosity of rice milk in the flow process. Use a viscometer or other equipment to collect experimental data and record the dynamic viscosity of rice milk. According to the experimental data, use MATLAB or Python to build a dynamic viscosity calculation model, and establish a prediction model by regression analysis and other methods. The prediction accuracy of the model is verified by experimental data to ensure that the model can accurately predict the dynamic viscosity of rice milk. For example, select the power law model or the Herschel-Bulkley model as the basic model, and fit the parameters of the basic model according to the experimental data to obtain the dynamic viscosity calculation module.

[0031] Step S140, establish a fusion output layer, integrate the outputs of the enzyme hydrolysis reaction kinetics prediction sub-model, the homogenization effect prediction sub-model and the dynamic viscosity calculation module into a rice milk performance data matrix. Specifically, the output data of the enzyme hydrolysis reaction kinetics prediction sub-model, the homogenization effect prediction sub-model and the dynamic viscosity calculation module are integrated to generate a rice milk performance data matrix. Use MATLAB or Python to process the integrated data to ensure the consistency and integrity of the data. The prediction accuracy of the integrated model is verified by experimental data to ensure that the model can accurately predict the performance of rice milk. This implementation can quickly adjust the production formula and process to adapt to market changes by building enzyme hydrolysis reaction kinetics prediction sub-model, homogenization effect prediction sub-model and dynamic viscosity calculation module. For example, if the market demand changes and the sweetness of rice milk needs to be adjusted, the parameters of the enzyme hydrolysis reaction kinetics prediction sub-model and the homogenization effect prediction sub-model can be adjusted to quickly generate optimization instructions and adjust the production process to meet market demand.

[0032] In one possible implementation, the dynamic viscosity calculation module is established, and step S130 further includes step S131, presetting a saccharification degree limit value, when the saccharification degree output by the enzyme hydrolysis reaction kinetics prediction sub-model is greater than the saccharification degree limit value, activating high viscosity compensation, and compensating the dynamic viscosity calculation module.

[0033] Specifically, an experimental scheme is designed to collect dynamic viscosity data of rice milk under different degrees of saccharification (e.g., 30%, 40%, 50%, 60%, 70%). A viscometer or other equipment is used to collect experimental data and record the dynamic viscosity of rice milk. Based on the experimental data, a relationship model between dynamic viscosity and degree of saccharification is constructed using MATLAB or Python, and a prediction model is established through regression analysis or other methods. A preset degree of saccharification limit (e.g., 60%) is set, and when the degree of saccharification output by the enzyme reaction kinetics prediction sub-model is greater than the limit, a high viscosity compensation mechanism is activated to adjust the parameters of the dynamic viscosity calculation module to compensate for the impact of high viscosity on calculation accuracy. The effectiveness of the compensation mechanism is verified through experimental data to ensure that the model can accurately predict the dynamic viscosity of rice milk. This implementation method can effectively adjust the parameters of the dynamic viscosity calculation module when the degree of saccharification exceeds the preset limit through the high viscosity compensation mechanism, compensate for the impact of high viscosity on calculation accuracy, improve the accuracy of dynamic viscosity calculation, and further improve the controllability of the production process and the consistency of product quality.

[0034] In one possible implementation, the construction of the rice milk simulation model further includes step S150 of constructing a stable adaptation prediction sub-model based on the influence of the fat particle size distribution and the emulsion stability index on the emulsion stability of rice milk coffee mixing. Specifically, an experimental scheme is designed to collect emulsion stability data of rice milk coffee mixing under different fat particle size distributions (e.g., 100-200 nm, 80-150 nm, 70-130 nm) and emulsion stability indexes (e.g., 70, 80, 90). A particle size analyzer and stability testing equipment are used to collect experimental data and record the emulsion stability. Based on the experimental data, a stable adaptation prediction sub-model is constructed using MATLAB or Python, and a prediction model is established through regression analysis or other methods. The prediction accuracy of the model is verified through experimental data to ensure that the model can accurately predict the emulsion stability of rice milk coffee mixing. For example, the Ostwald ripening theory is selected as the basic theory, and the parameters of the basic theory are fitted based on experimental data to obtain a stable adaptation prediction sub-model.

[0035] Step S160, according to the influence of the degree of saccharification and the content of reducing sugar on the sweetness balance of rice milk coffee mixture, a sensory adaptation predictor model is constructed. Specifically, an experimental scheme is designed, and the sweetness balance data of rice milk coffee mixture under different degrees of saccharification (such as 30%, 40%, 50%) and reducing sugar content (such as 5g / 100g, 7g / 100g, 9g / 100g) are collected. Use spectrometer and other equipment to collect experimental data and record the sweetness balance. According to the experimental data, use MATLAB or Python to construct a sensory adaptation predictor model, and establish a prediction model by regression analysis and other methods. Verify the prediction accuracy of the model through experimental data to ensure that the model can accurately predict the sweetness balance of rice milk coffee mixture. For example, select the Hedonic scale as the basic method, and fit the parameters of the basic method according to the experimental data to obtain the sensory adaptation predictor model.

[0036] Step S170, according to the influence of viscosity and the fat particle size distribution on the performance of rice milk in coffee, a function adaptation predictor model is constructed. Specifically, an experimental scheme is designed, and the data of rice milk in coffee under different viscosities (such as 10 mPa·s, 15 mPa·s, 20 mPa·s) and fat particle size distributions (such as 100-200 nm, 80-150 nm, 70-130 nm) are collected. Use viscometer and latte test equipment to collect experimental data and record the performance of latte. According to the experimental data, use MATLAB or Python to construct a function adaptation predictor model, and establish a prediction model by regression analysis and other methods. Verify the prediction accuracy of the model through experimental data to ensure that the model can accurately predict the performance of rice milk in coffee. For example, select the power law model as the basic theory, and fit the parameters of the basic theory according to the experimental data to obtain the function adaptation predictor model.

[0037] Step S180, the stable adaptation predictor model, the sensory adaptation predictor model and the function adaptation predictor model are connected in parallel to construct a rice milk application simulation model. Specifically, the stable adaptation predictor model, the sensory adaptation predictor model and the function adaptation predictor model are logically connected in parallel to obtain a rice milk application simulation model, which is used to predict the comprehensive application performance of rice milk in coffee. This implementation can accurately predict the emulsion stability of rice milk coffee mixture through the stable adaptation predictor model, ensuring that the rice milk does not separate in coffee and maintains a uniform mixed state. Through the sensory adaptation predictor model, the sweetness balance of rice milk coffee mixture can be accurately predicted, ensuring moderate sweetness and meeting the taste needs of consumers. Through the function adaptation predictor model, the performance of rice milk in coffee can be accurately predicted, ensuring good latte performance and improving the visual appeal of the product.

[0038] Step S200, real-time acquisition of process parameters of the production line to obtain rice milk production data set, input the rice milk production data set into the rice milk production simulation model, and obtain rice milk performance data matrix through virtual simulation.

[0039] Specifically, a variety of sensors are deployed at key links of the rice milk production line to real-time acquisition of process parameters. The sensor deployment and data acquisition example is shown in Table 1.

[0040] Table 1: Sensor deployment and data acquisition example

[0041]

[0042] The sensor data is transmitted to the central server in real time using industrial automation systems (such as PLC, DCS) and data acquisition software (such as LabVIEW, NI DAQ). The collected data is stored in a database (such as SQL Server, MongoDB), and data processing tools (such as MATLAB, Python) are used to clean the collected data, remove abnormal data and noise data, use filtering algorithms (such as low-pass filter) to smooth the data, and standardize the data to a unified dimension.

[0043] The preprocessed rice milk production data set is input into the rice milk production simulation model, and virtual simulation software (such as ANSYS Fluent, COMSOL Multiphysics) is used to simulate the rice milk production process, generating a rice milk performance data matrix, including but not limited to dynamic viscosity, fat particle size distribution, emulsion stability index, saccharification degree and reducing sugar content, etc.

[0044] Step S300, input the rice milk performance data matrix into the rice milk application simulation model to generate quality defect early warning, according to the early warning result, carry out reverse regulation, generate optimization instruction, feedback the optimization instruction to the production line to realize real-time rice milk production control.

[0045] Specifically, the rice milk performance data matrix obtained through virtual simulation will be input into the rice milk application simulation model, including dynamic viscosity, fat particle size distribution, emulsion stability index, saccharification degree, and reducing sugar content, etc. Using early warning algorithms based on statistical analysis or machine learning, the rice milk performance data matrix is analyzed to predict whether there are potential problems in the application performance of rice milk in coffee. For example: if the emulsion stability index is below the preset threshold, it indicates that the rice milk will have a layered phenomenon in coffee; if the saccharification degree or reducing sugar content is not within the preset range, it will affect the sweetness balance of rice milk coffee; if the dynamic viscosity or fat particle size distribution is not within the preset range, it will affect the pull flower effect of rice milk in coffee. According to the corresponding quality defect early warning, the corresponding early warning signal is generated, indicating the potential quality problem and its severity. According to the early warning result, optimization instructions are generated, which can include adjusting the temperature, time or enzyme concentration of enzymatic reaction, adjusting the pressure or flow of homogenization, or adjusting other related process parameters.

[0046] Before feeding the optimization instructions back to the production line, the optimization instructions need to be verified through the rice milk dual simulation model. The rice milk dual simulation model includes a rice milk production simulation model and a rice milk application simulation model. Through the joint verification of the two models, the effectiveness and safety of the optimization instructions in actual production are ensured. The verification process includes: inputting the generated optimization instructions into the rice milk dual simulation model, simulating the production process through the rice milk production simulation model, simulating the application performance of rice milk in coffee through the rice milk application simulation model, verifying whether the optimization instructions can effectively solve the early warning problem, and evaluating the simulation results. If the optimization instructions can effectively solve the early warning problem and do not introduce new problems, the verification is passed; otherwise, the optimization instructions need to be regenerated and verified again.

[0047] The optimization instructions that pass the verification are fed back to the production line through an industrial automation control system (such as PLC, DCS). The automation equipment on the production line adjusts the process parameters in real time according to the feedback optimization instructions, ensuring the optimization and quality control of the production process. Continue to collect the process parameters of the production line in real time to form a closed loop control, ensuring the stability of the production process and the consistency of the product quality.

[0048] In a possible implementation, the rice milk performance data matrix is input into the rice milk application simulation model for quality defect early warning, and according to the early warning result, reverse regulation is performed to generate an optimization instruction. Step S300 further includes step S310: when the emulsion stability index detected by the stability adaptive prediction sub-model is less than a preset first stability index limit value, a first quality defect early warning is generated. Specifically, the emulsion stability index in the rice milk performance data matrix is input into the stability adaptive prediction sub-model. If the emulsion stability index is less than a preset first stability index limit value (for example, 80), a first quality defect early warning is generated, and a warning signal is recorded, indicating that there may be a problem with the emulsion stability.

[0049] Step S320: when the sensory adaptive prediction sub-model detects that the sweetness deviation is greater than a preset sweetness deviation limit value, a second quality defect early warning is generated. Specifically, the saccharification degree and the reducing sugar content in the rice milk performance data matrix are input into the sensory adaptive prediction sub-model. The model calculates the actual sweetness according to a preset sweetness balance formula (for example, sweetness = saccharification degree x reducing sugar content), and calculates the deviation between the actual sweetness and the target sweetness (for example, sweetness deviation = actual sweetness - target sweetness). If the sweetness deviation is greater than a preset sweetness deviation limit value (for example, ±2 g / 100 g), a second quality defect early warning is generated, and a warning signal is recorded, indicating that there may be a problem with the sweetness balance.

[0050] Step S330: when the function adaptive prediction sub-model detects that the pull flower foam decay rate is greater than a preset decay rate standard value, a third quality defect early warning is generated. Specifically, the dynamic viscosity and the fat particle size distribution in the rice milk performance data matrix are input into the function adaptive prediction sub-model. The model calculates the decay rate of the pull flower foam according to a preset pull flower performance formula, for example, decay rate = f(dynamic viscosity, fat particle size distribution). If the pull flower foam decay rate is greater than a preset decay rate standard value (for example, 0.5 mm / s), a third quality defect early warning is generated, and a warning signal is recorded, indicating that there may be a problem with the pull flower performance.

[0051] Step S340, according to the trigger state of the first quality defect early warning, the second quality defect early warning and the third quality defect early warning, reverse regulation is carried out, and optimization instruction is generated. Specifically, according to the trigger state of the first, second and third quality defect early warning, the application performance of rice milk in coffee is comprehensively evaluated, and the corresponding optimization instruction is generated, for example: if the emulsion stability index is lower than the threshold value, the instruction of "increasing the homogenization pressure by 5%" is generated; if the sweetness deviation exceeds the limit value, the instruction of "increasing the enzymolysis reaction temperature by 2°C" or "increasing the enzyme concentration by 0.1%" is generated; if the pull flower foam decay rate exceeds the standard value, the instruction of "increasing the homogenization time by 10 seconds" or "adjusting the homogenization pressure to 25 MPa" is generated. The generated optimization instruction is input into the rice milk double simulation model for verification to ensure the effectiveness and safety of the optimization instruction in actual production. The optimization instruction that passes the verification is fed back to the production line through PLC or DCS to adjust the process parameters in real time. This implementation mode predicts the application performance of rice milk in coffee through the model, discovers potential quality problems in advance, and optimizes the production process through reverse regulation and real-time feedback, so that the application performance of rice milk in coffee reaches the best.

[0052] In a possible implementation, according to the trigger state of the first quality defect early warning, the second quality defect early warning and the third quality defect early warning, reverse regulation is carried out, and optimization instruction is generated, and step S340 further includes step S341. When and only when one prediction sub-model triggers an early warning, according to the first quality defect early warning or the second quality defect early warning or the third quality defect early warning, reverse regulation is carried out, and the corresponding first optimization instruction or second optimization instruction or third optimization instruction is generated. Specifically, it is checked whether only one prediction sub-model triggers an early warning. If it is the first quality defect early warning (the emulsion stability index is lower than the threshold value), the first optimization instruction such as "increasing the homogenization pressure by 5%" is generated. If it is the second quality defect early warning (the sweetness deviation exceeds the limit value), the second optimization instruction such as "increasing the enzymolysis reaction temperature by 2°C" is generated. If it is the third quality defect early warning (the pull flower foam decay rate exceeds the standard value), the third optimization instruction such as "increasing the homogenization time by 10 seconds" is generated. Wherein, the optimization instruction is the corresponding adjustment instruction automatically generated by analyzing the key indicators (such as emulsion stability index, sweetness deviation, pull flower foam decay rate, etc.) in the rice milk performance data matrix according to the preset threshold value and rules to correct the parameters deviating from the target range.

[0053] Step S342, when more than or equal to two predictor models trigger the early warning at the same time, multi-parameter collaborative optimization is performed to generate a collaborative optimization instruction. Specifically, it is checked whether more than one predictor model triggers the early warning. For example, if the first and second quality defect early warnings trigger at the same time, a collaborative optimization instruction such as “increase homogenization pressure by 5% and increase enzymolysis reaction temperature by 2°C” is generated; if the first and third quality defect early warnings trigger at the same time, a collaborative optimization instruction such as “increase homogenization pressure by 5% and increase homogenization time by 10 seconds” is generated; if the second and third quality defect early warnings trigger at the same time, a collaborative optimization instruction such as “increase enzymolysis reaction temperature by 2°C and increase homogenization time by 10 seconds” is generated; and if the three early warnings trigger at the same time, a collaborative optimization instruction such as “increase homogenization pressure by 5%, increase enzymolysis reaction temperature by 2°C, and increase homogenization time by 10 seconds” is generated. In this implementation, when only one early warning is triggered, a targeted optimization instruction is generated to ensure that the problem can be quickly solved. When multiple early warnings are triggered at the same time, multiple problems are solved at the same time through the collaborative optimization instruction, avoiding secondary problems that may be caused by a single optimization instruction, and achieving the technical effect of precise optimization.

[0054] In a possible implementation, multi-parameter collaborative optimization is performed to generate a collaborative optimization instruction, and step S342 further includes step S3421 of assigning a dynamic weight coefficient to each predictor model according to differences in coffee modes, where the sum of the dynamic weight coefficients of the predictor models is 1. Specifically, multiple coffee modes are defined according to different coffee recipes and application scenarios. According to historical data and expert experience, different weight coefficients are assigned to each predictor model under different coffee modes. At the same time, the weight coefficients can be dynamically adjusted according to real-time data and production requirements.

[0055] Step S3422, calculating an instruction intensity factor of each predictor model according to the dynamic weight coefficients. Specifically, a preset formula (such as instruction intensity = early warning severity x dynamic weight coefficient) is used to calculate the instruction intensity factor of each predictor model. The calculated instruction intensity factors are standardized to ensure that they are comparable under the same dimension.

[0056] Step S3423, performing multi-parameter collaborative optimization by using a weighted fusion algorithm to integrate the instruction intensity factors to generate a collaborative optimization instruction. Specifically, a weighted fusion algorithm (such as a weighted average method) is used to integrate the instruction intensity factors of each predictor model. According to the integrated instruction intensity factors, a specific optimization instruction is generated. For example, if the integrated instruction intensity factors indicate that the homogenization pressure and the enzymolysis reaction temperature need to be adjusted at the same time, a corresponding collaborative optimization instruction is generated. This implementation dynamically assigns weight coefficients according to differences in coffee modes, ensuring that the optimization instruction is more suitable for actual production requirements.

[0057] In a possible implementation, the dynamic weight coefficients are assigned to each prediction sub-model according to the difference of coffee modes, and step S3421 further includes step S34211. If the deep-roasted coffee mode is selected, the dynamic weight coefficients are assigned according to a first order, where the first order is: the sensory adaptation prediction sub-model weight > the function adaptation prediction sub-model weight > the stability adaptation prediction sub-model weight. Specifically, the current production mode is identified as the deep-roasted coffee mode through the production management system or operator input. According to the characteristics of the deep-roasted coffee mode, the dynamic weight coefficients are assigned. Since the deep-roasted coffee has a strong taste and a strong bitter taste, the sweetness balance is required to be higher. If the sweetness balance is poor, the taste experience of consumers will be affected. Therefore, the dynamic weight coefficients are assigned according to the preset first order, to ensure that the weight of the sensory adaptation prediction sub-model is the highest, the weight of the function adaptation prediction sub-model is the second, and the weight of the stability adaptation prediction sub-model is the last. For example, the weight of the sensory adaptation prediction sub-model is 0.5, the weight of the function adaptation prediction sub-model is 0.3, and the weight of the stability adaptation prediction sub-model is 0.2.

[0058] In step S34212, if the light-roasted coffee mode is selected, the dynamic weight coefficients are assigned according to a second order, where the second order is: the function adaptation prediction sub-model weight > the stability adaptation prediction sub-model weight > the sensory adaptation prediction sub-model weight. Specifically, the current production mode is identified as the light-roasted coffee mode through the production management system or operator input. According to the characteristics of the light-roasted coffee mode, the dynamic weight coefficients are assigned. Since the light-roasted coffee has a light taste and a good latte art effect, the latte art performance is required to be higher. If the latte art performance is poor, the visual appeal of the product will be affected. Therefore, the dynamic weight coefficients are assigned according to the preset second order, to ensure that the weight of the function adaptation prediction sub-model is the highest, the weight of the stability adaptation prediction sub-model is the second, and the weight of the sensory adaptation prediction sub-model is the last. For example, the weight of the function adaptation prediction sub-model is 0.5, the weight of the stability adaptation prediction sub-model is 0.3, and the weight of the sensory adaptation prediction sub-model is 0.2.

[0059] Step S34213, when the emulsification stability index in the first quality defect early warning is less than a preset second stability index limit value, triggering a stability crisis mode, and forcibly adjusting the first order and the second order to a third order, wherein the second stability index limit value is less than the first stability index limit value, and the third order is: stability adaptation predictor model weight > function adaptation predictor model weight > sensory adaptation predictor model weight. Specifically, the emulsification stability index is detected by the stability adaptation predictor model. If the emulsification stability index is less than the preset second stability index limit value (such as 60), the stability crisis mode is triggered. When the emulsification stability index is lower than the preset second stability index limit value, it indicates that there is a serious problem with the emulsification stability of the rice milk in the coffee, which will cause delamination and affect product quality. In order to prioritize the solution of the emulsification stability problem, the weight order is forcibly adjusted to the third order, ensuring that the weight of the stability adaptation predictor model is the highest, followed by the function adaptation predictor model, and finally the sensory adaptation predictor model. For example, the stability adaptation predictor model weight is 0.6, the function adaptation predictor model weight is 0.3, and the sensory adaptation predictor model weight is 0.1. This implementation dynamically adjusts the weight coefficient according to the characteristics of different coffee modes, ensuring that the optimization instruction is more in line with actual production needs. In the stability crisis mode, the emulsification stability is prioritized, ensuring the stability of the production process and product quality.

[0060] In one possible implementation, step S330 further comprises step S331 of accessing a barista skill level database and dynamically adjusting the size of the attenuation rate standard value according to the operator level, wherein the level is positively correlated with the size of the attenuation rate standard value.

[0061] Specifically, the barista skill level database is accessed through the production management system, which records the skill levels of different baristas. The skill level of the current operator is read from the database. According to the skill level, the attenuation rate standard value of the latte foam is dynamically adjusted using a preset adjustment algorithm. For example, for every increase in skill level, the attenuation rate standard value increases by 0.1 mm / s. This implementation dynamically adjusts the attenuation rate standard value according to the skill level of the barista, ensuring that the optimization instruction is more in line with the actual operating ability of different baristas, achieving the technical effect of personalized optimization.

[0062] The embodiment of the application adopts a rice milk dual simulation model (including a production simulation model and an application simulation model) based on digital twinning to collect real-time production line process parameters and input the production simulation model to obtain a performance data matrix, input the application simulation model to warn of quality defects, generate optimization instructions in reverse, and feed back to the production line to achieve real-time control and other technical means, solving the technical problem of poor rice milk application adaptability caused by focusing on production link monitoring in existing rice milk production control, achieving the technical effect of accurately regulating the rice milk production process from the application effect, and further improving the rice milk application adaptability.

[0063] In the foregoing, reference is made to Figure 1 The rice milk production control method combined with digital twinning according to the embodiment of the application is described in detail. Next, the rice milk production control system combined with digital twinning according to the embodiment of the application will be described with reference to Figure 2 The rice milk production control system combined with digital twinning according to the embodiment of the application is described in detail. Next, the rice milk production control system combined with digital twinning according to the embodiment of the application will be described with reference to

[0064] The rice milk production control system combined with digital twinning according to the embodiment of the application is used to solve the technical problem of poor rice milk application adaptability caused by focusing on production link monitoring in existing rice milk production control, achieve the technical effect of accurately regulating the rice milk production process from the application effect, and further improve the rice milk application adaptability. The rice milk production control system combined with digital twinning includes a rice milk dual simulation model construction module 10, a production simulation module 20, and a reverse regulation optimization module 30.

[0065] The rice milk dual simulation model construction module 10 is used to construct a rice milk dual simulation model including a physical entity and a virtual model bidirectional mapping based on digital twinning technology, and the rice milk dual simulation model includes a rice milk production simulation model and a rice milk application simulation model logically connected in sequence. The production simulation module 20 is used to collect real-time production line process parameters to obtain a rice milk production data set, input the rice milk production data set into the rice milk production simulation model, and obtain a rice milk performance data matrix through virtual simulation. The reverse regulation optimization module 30 is used to input the rice milk performance data matrix into the rice milk application simulation model for quality defect warning, perform reverse regulation according to the warning result, generate optimization instructions, and feed back the optimization instructions to the production line to perform real-time rice milk production control.

[0066] In the following, the specific configuration of the rice milk dual simulation model construction module 10 will be described in detail. As described above, for the construction of the rice milk production simulation model, the rice milk dual simulation model construction module 10 can further include: an enzymatic hydrolysis reaction kinetics predictor model construction unit for constructing an enzymatic hydrolysis reaction kinetics predictor model, which synchronously outputs the degree of saccharification and the reducing sugar content through a temperature-time-enzyme concentration three-dimensional relationship matrix; a homogenization effect predictor model construction unit for constructing a homogenization effect predictor model, which synchronously outputs the fat particle size distribution and the emulsion stability index through the interactive calculation of homogenization pressure and flow rate; a dynamic viscosity calculation module establishment unit for establishing a dynamic viscosity calculation module based on the output of the enzymatic hydrolysis reaction kinetics predictor model and the homogenization effect predictor model; and a fusion output layer establishment unit for establishing a fusion output layer, which integrates the output of the enzymatic hydrolysis reaction kinetics predictor model, the homogenization effect predictor model and the dynamic viscosity calculation module into a rice milk performance data matrix.

[0067] Wherein, the dynamic viscosity calculation module is established, the rice milk dual simulation model construction module 10 can further include: a compensation unit for presetting a saccharification degree limit value, when the saccharification degree output by the enzymatic hydrolysis reaction kinetics predictor model is greater than the saccharification degree limit value, activating high viscosity compensation, and compensating the dynamic viscosity calculation module.

[0068] Wherein, for the construction of the rice milk application simulation model, the rice milk dual simulation model construction module 10 can further include: a stability adaptation predictor model construction unit for constructing a stability adaptation predictor model according to the influence of the fat particle size distribution and the emulsion stability index on the emulsion stability of rice milk coffee mixing; a sensory adaptation predictor model construction unit for constructing a sensory adaptation predictor model according to the influence of the degree of saccharification and the reducing sugar content on the sweetness balance of rice milk coffee mixing; a function adaptation predictor model construction unit for constructing a function adaptation predictor model according to the influence of viscosity and the fat particle size distribution on the performance of rice milk in coffee drawing; and a sub-model connection unit for connecting the stability adaptation predictor model, the sensory adaptation predictor model and the function adaptation predictor model side by side to construct a rice milk application simulation model.

[0069] The specific configuration of the reverse regulation optimization module 30 will be described in detail below. As described above, the rice milk performance data matrix is input into the rice milk application simulation model for quality defect early warning, and according to the early warning result, the reverse regulation optimization module 30 can further include: a first quality defect early warning generation unit for generating a first quality defect early warning when the emulsion stability index detected by the stability adaptive prediction sub-model is less than the preset first stability index limit value; a second quality defect early warning generation unit for generating a second quality defect early warning when the sweetness deviation detected by the sensory adaptive prediction sub-model is greater than the preset sweetness deviation limit value; a third quality defect early warning generation unit for generating a third quality defect early warning when the pull flower foam decay rate detected by the functional adaptive prediction sub-model is greater than the preset decay rate standard value; and a reverse regulation unit for performing reverse regulation according to the trigger state of the first quality defect early warning, the second quality defect early warning and the third quality defect early warning to generate an optimization instruction.

[0070] According to the trigger state of the first quality defect early warning, the second quality defect early warning and the third quality defect early warning, the reverse regulation unit can further include: a single optimization instruction generation sub-unit for generating a corresponding first optimization instruction or a second optimization instruction or a third optimization instruction according to the first quality defect early warning or the second quality defect early warning or the third quality defect early warning when and only when one prediction sub-model triggers an early warning; and a collaborative optimization instruction generation sub-unit for performing multi-parameter collaborative optimization to generate a collaborative optimization instruction when greater than or equal to two prediction sub-models trigger an early warning at the same time.

[0071] According to the trigger state of the first quality defect early warning, the second quality defect early warning and the third quality defect early warning, the reverse regulation unit can further include: a single optimization instruction generation sub-unit for generating a corresponding first optimization instruction or a second optimization instruction or a third optimization instruction according to the first quality defect early warning or the second quality defect early warning or the third quality defect early warning when and only when one prediction sub-model triggers an early warning; and a collaborative optimization instruction generation sub-unit for performing multi-parameter collaborative optimization to generate a collaborative optimization instruction when greater than or equal to two prediction sub-models trigger an early warning at the same time.

[0072] Wherein, according to the difference of coffee mode, a dynamic weight coefficient is assigned to each prediction sub-model, and the dynamic weight coefficient assignment component can further include: a first order assignment sub-component for assigning the dynamic weight coefficient in a first order if the coffee mode is deep roasting, wherein the first order is: sensory adaptation prediction sub-model weight > function adaptation prediction sub-model weight > stability adaptation prediction sub-model weight; a second order assignment sub-component for assigning the dynamic weight coefficient in a second order if the coffee mode is light roasting, wherein the second order is: function adaptation prediction sub-model weight > stability adaptation prediction sub-model weight > sensory adaptation prediction sub-model weight; and a third order assignment sub-component for triggering a stability crisis mode when the emulsion stability index in the first quality defect early warning is less than a preset second stability index limit value, and forcibly adjusting the first order and the second order to a third order, wherein the second stability index limit value is less than the first stability index limit value, and the third order is: stability adaptation prediction sub-model weight > function adaptation prediction sub-model weight > sensory adaptation prediction sub-model weight.

[0073] Wherein, the third quality defect early warning generation unit can further include: a decay rate standard value dynamic adjustment sub-unit for accessing a barista skill level database, and dynamically adjusting the size of the decay rate standard value according to the operator level, wherein the level is positively correlated with the size of the decay rate standard value.

[0074] The rice milk production control system combined with digital twinning provided by the embodiments of the present application can execute the rice milk production control method combined with digital twinning provided by any of the embodiments of the present application, and has the corresponding function modules and beneficial effects of the execution method.

[0075] Although the present application makes various references to certain modules in the system according to the embodiments of the present application, however, any number of different modules can be used and run on the user terminal and / or server, and each unit and module included is only divided according to the functional logic, but is not limited to the above division, as long as the corresponding functions can be realized; in addition, the specific names of each functional unit are only for easy mutual differentiation, and do not limit the protection scope of the present application.

[0076] The foregoing DETAILED DESCRIPTION, including the above section titled "Detailed Description," is not to be taken as limiting the scope of the application. Various modifications, combinations, and equivalents can be apparent to those skilled in the art and can be made once the nature of the application is understood. Any modification, combination, or equivalent, which falls within the principles and the scope of the present application, is intended to be included in the present application. In some instances, the actions or steps can be performed in different order from those described herein, and still achieve desirable results. Additionally, the process depicted in the figures can not necessarily require the particular order shown or sequential order to achieve the desired results. In certain implementations, multitasking and parallel processing can be advantageous.

Claims

1. A rice milk production control method incorporating digital twinning, characterized by, The method comprises: Based on digital twinning technology, a rice milk dual simulation model containing a bidirectional mapping of a physical entity and a virtual model is constructed, the rice milk dual simulation model comprising a rice milk production simulation model and a rice milk application simulation model logically connected in sequence; Real-time acquisition of process parameters of the production line to obtain a rice milk production dataset, input of the rice milk production dataset into the rice milk production simulation model, and obtaining of a rice milk performance data matrix through virtual simulation; Input of the rice milk performance data matrix into the rice milk application simulation model for quality defect early warning, reverse regulation according to the early warning result, generation of an optimization instruction, and feedback of the optimization instruction to the production line for real-time rice milk production control; The construction of the rice milk production simulation model comprises: Construction of an enzymatic reaction kinetics prediction sub-model, synchronous output of the degree of saccharification and the content of reducing sugar through a temperature-time-enzyme concentration three-dimensional relationship matrix; Construction of a homogenization effect prediction sub-model, synchronous output of the fat particle size distribution and the emulsion stability index through interactive calculation of the homogenization pressure and the flow rate; Based on the output of the enzymatic reaction kinetics prediction sub-model and the homogenization effect prediction sub-model, a dynamic viscosity calculation module is established; An output fusion layer is established, and the outputs of the enzymatic reaction kinetics prediction sub-model, the homogenization effect prediction sub-model and the dynamic viscosity calculation module are integrated into a rice milk performance data matrix; The dynamic viscosity calculation module further comprises: A preset saccharification degree limit value, when the saccharification degree output by the enzymatic reaction kinetics prediction sub-model is greater than the saccharification degree limit value, high viscosity compensation is activated to compensate the dynamic viscosity calculation module; The construction of the rice milk application simulation model comprises: Construction of a stability adaptation prediction sub-model according to the influence of the fat particle size distribution and the emulsion stability index on the emulsion stability of rice milk coffee mixing; Construction of a sensory adaptation prediction sub-model according to the influence of the degree of saccharification and the content of reducing sugar on the sweetness balance of rice milk coffee mixing; Construction of a functional adaptation prediction sub-model according to the influence of viscosity and the fat particle size distribution on the performance of rice milk in coffee drawing; The stability adaptation prediction sub-model, the sensory adaptation prediction sub-model and the functional adaptation prediction sub-model are connected in parallel to construct a rice milk application simulation model.

2. The rice milk production control method incorporating digital twinning of claim 1, wherein, Input of the rice milk performance data matrix into the rice milk application simulation model for quality defect early warning, reverse regulation according to the early warning result, generation of an optimization instruction, comprising: When the stability adaptation prediction sub-model detects that the emulsion stability index is less than a preset first stability index limit value, a first quality defect early warning is generated; When the sensory adaptation prediction sub-model detects that the sweetness deviation is greater than a preset sweetness deviation limit value, a second quality defect early warning is generated; When the functional adaptation prediction sub-model detects that the drawing foam decay rate is greater than a preset decay rate standard value, a third quality defect early warning is generated; Reverse regulation according to the trigger state of the first quality defect early warning, the second quality defect early warning and the third quality defect early warning, and generation of an optimization instruction.

3. The rice milk production control method incorporating digital twinning of claim 2, characterized by, According to the triggering state of the first quality defect early warning, the second quality defect early warning and the third quality defect early warning, reverse regulation is performed to generate optimization instructions, including: According to the triggering state of the first quality defect early warning, the second quality defect early warning and the third quality defect early warning, reverse regulation is performed to generate optimization instructions, including: According to the triggering state of the first quality defect early warning, the second quality defect early warning and the third quality defect early warning, reverse regulation is performed to generate optimization instructions, including:

4. The rice milk production control method incorporating digital twinning of claim 3, characterized by, According to the triggering state of the first quality defect early warning, the second quality defect early warning and the third quality defect early warning, reverse regulation is performed to generate optimization instructions, including: When more than two prediction sub-models trigger early warning at the same time, multi-parameter collaborative optimization is performed to generate collaborative optimization instructions. When more than two prediction sub-models trigger early warning at the same time, multi-parameter collaborative optimization is performed to generate collaborative optimization instructions. When more than two prediction sub-models trigger early warning at the same time, multi-parameter collaborative optimization is performed to generate collaborative optimization instructions.

5. The rice milk production control method incorporating digital twinning of claim 4, wherein, When more than two prediction sub-models trigger early warning at the same time, multi-parameter collaborative optimization is performed to generate collaborative optimization instructions. According to the difference of coffee mode, dynamic weight coefficients are assigned to each prediction sub-model, wherein the sum of the dynamic weight coefficients of each prediction sub-model is 1. According to the difference of coffee mode, dynamic weight coefficients are assigned to each prediction sub-model, wherein the sum of the dynamic weight coefficients of each prediction sub-model is 1. According to the difference of coffee mode, dynamic weight coefficients are assigned to each prediction sub-model, wherein the sum of the dynamic weight coefficients of each prediction sub-model is 1. 6.The rice milk production control method with digital twinning of claim 2, wherein, If it is a deep roasting coffee mode, the dynamic weight coefficients are assigned in a first order, wherein the first order is: sensory adaptation prediction sub-model weight > function adaptation prediction sub-model weight > stability adaptation prediction sub-model weight. If it is a deep roasting coffee mode, the dynamic weight coefficients are assigned in a first order, wherein the first order is: sensory adaptation prediction sub-model weight > function adaptation prediction sub-model weight > stability adaptation prediction sub-model weight.

7. A rice milk production control system incorporating digital twinning, characterized by, When the emulsion stability index in the first quality defect early warning is less than a preset second stability index limit value, a stability crisis mode is triggered, and the first order and the second order are forcibly adjusted to a third order, wherein the second stability index limit value is less than the first stability index limit value, and the third order is: stability adaptation prediction sub-model weight > function adaptation prediction sub-model weight > sensory adaptation prediction sub-model weight. Further comprising: Accessing a barista skill level database, dynamically adjusting the size of the attenuation rate standard value according to the operator level, wherein the level is positively correlated with the size of the attenuation rate standard value. The system is used to implement the rice milk production control method combined with digital twinning according to any one of claims 1-6, and the system comprises: A rice milk dual simulation model construction module is configured to construct a rice milk dual simulation model comprising a bidirectional mapping of a physical entity and a virtual model based on digital twinning technology, wherein the rice milk dual simulation model comprises a rice milk production simulation model and a rice milk application simulation model connected in logical sequence. A production simulation module is configured to collect process parameters of a production line in real time to obtain a rice milk production data set, input the rice milk production data set into the rice milk production simulation model, and obtain a rice milk performance data matrix through virtual simulation. A reverse regulation optimization module is configured to input the rice milk performance data matrix into the rice milk application simulation model to perform quality defect early warning, perform reverse regulation according to the early warning result, generate optimization instructions, and feed back the optimization instructions to the production line to perform real-time rice milk production control.

Images