Fruit and vegetable drying production line real-time monitoring and optimizing method based on digital twinning

Abstract

The invention discloses a fruit and vegetable drying production line real-time monitoring and optimizing method based on digital twinning, and the method comprises the steps: S1, building a production line digital twinning model based on a fruit and vegetable drying physical production line; s2, collecting data in a fruit and vegetable drying physical production line, reversely updating the data to the production line digital twinborn model in real time, and monitoring the data in the production line digital twinborn model; s3, establishing a production line digital twin model knowledge base; and S4, optimizing fruit and vegetable drying production indexes based on the knowledge base. According to the invention, real-time monitoring and optimization of the fruit and vegetable drying production line can be realized.

Description

Technical Field

[0001] The present invention relates to the field of monitoring and optimization of dried fruit and vegetable production lines, and in particular to real-time monitoring and optimization of dried fruit and vegetable production lines based on digital twins. Background Art

[0002] Existing fruit and vegetable drying production lines use manual inspections or simple PLC control, which has problems such as data collection delays, high energy consumption and lagging process parameter adjustments, resulting in unstable product quality (such as uneven moisture and heat energy waste). Currently, digital twin technology is mostly used for predictive maintenance of industrial equipment or complex process simulation, and has not yet achieved real-time monitoring and optimization of the entire process based on the dynamic characteristics of fruit and vegetable drying production lines. Summary of the Invention

[0003] The purpose of the present invention is to provide a real-time monitoring and optimization method for a dried fruit and vegetable production line based on digital twins, aiming to solve the monitoring and optimization of the dried fruit and vegetable production line.

[0004] The present invention provides a real-time monitoring and optimization method for a dried fruit and vegetable production line based on digital twins, comprising:

[0005] Real-time monitoring and optimization method of dried fruit and vegetable production line based on digital twin, including:

[0006] S1. Establish a digital twin model of the production line based on the physical production line of dried fruits and vegetables;

[0007] S2. Collect data from the physical production line for dried fruits and vegetables and update it to the production line digital twin model in real time, monitoring the data in the production line digital twin model;

[0008] S3. Establish a knowledge base for the production line digital twin model;

[0009] S4. Optimize the production indicators of dried fruits and vegetables based on the knowledge base.

[0010] The embodiments of the present invention are intended to solve the monitoring and optimization of a fruit and vegetable drying production line.

[0011] The above description is only an overview of the technical solution of the present invention. In order to more clearly understand the technical means of the present invention, it is implemented in accordance with the contents of the specification, and in order to make the above and other purposes, features and advantages of the present invention more obvious and easy to understand, the specific implementation methods of the present invention are specifically listed below. BRIEF DESCRIPTION OF THE DRAWINGS

[0012] In order to more clearly illustrate the specific embodiments of the present invention or the technical solutions in the prior art, the following briefly introduces the drawings required for use in the specific embodiments or the description of the prior art. Obviously, the drawings described below are some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying any creative work.

[0013] Figure 1 This is a flow chart of a method for real-time monitoring and optimization of a dried fruit and vegetable production line based on digital twins according to an embodiment of the present invention; DETAILED DESCRIPTION

[0014] The following will clearly and completely describe the technical solutions of the present invention in conjunction with the embodiments. Obviously, the embodiments described are only some embodiments of the present invention, not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without making creative efforts are within the scope of protection of the present invention.

[0015] Method Example

[0016] According to an embodiment of the present invention, a real-time monitoring and optimization method for a dried fruit and vegetable production line based on digital twins is provided. Figure 1 This is a flow chart of a method for real-time monitoring and optimization of a dried fruit and vegetable production line based on digital twins according to an embodiment of the present invention. Figure 1 As shown, specifically including:

[0017] S1. Establish a digital twin model of the production line based on the physical production line of dried fruits and vegetables;

[0018] The S1 specifically includes:

[0019] S11. Perform physical mapping on the physical production line of dried fruits and vegetables to obtain a mapping relationship;

[0020] S12, establishing a geometric model based on the mapping relationship and rendering to obtain a rendering model;

[0021] The S12 specifically includes: establishing a geometric model based on the equipment layer, perception layer, and control layer of the production line and rendering to obtain a rendering model.

[0022] S13, configuring motion parameters for devices in the rendering model and establishing linkage relationships between devices;

[0023] S14. Add a simulation engine to obtain a digital twin model of the production line.

[0024] S2. Collect data from the physical production line for dried fruits and vegetables and update it to the production line digital twin model in real time, monitoring the data in the production line digital twin model;

[0025] Collecting data from the physical production line of dried fruits and vegetables and updating it to the digital twin model of the production line in real time specifically includes: collecting data from the physical production line of dried fruits and vegetables based on the OPCUA protocol, and updating it to the digital twin model of the production line in real time through the MQTT protocol.

[0026] S3. Establish a knowledge base for the production line digital twin model;

[0027] S3 specifically includes:

[0028] A knowledge graph for the digital twin model of the first production line was established based on data from the production line digital twin model, historical equipment parameters, and fruit and vegetable drying process data.

[0029] Establish an expert experience knowledge graph, establish a drying mechanism knowledge graph, and establish a drying machine learning model graph.

[0030] S4. Optimize the production indicators of dried fruits and vegetables based on the knowledge base.

[0031] S4 specifically includes:

[0032] Based on the knowledge graph of the digital twin model of the first production line, the expert experience knowledge graph, the drying mechanism knowledge graph and the drying machine learning model graph, the production indicators of fruit and vegetable drying are optimized. The production indicators include: production line energy consumption, drying time and finished product qualification rate. The optimization constraints include: equipment temperature and humidity fluctuations.

[0033] In an embodiment of the present invention, when the degree of match between the optimized production indicators and the corresponding data in the production line digital twin model knowledge base is less than a certain threshold, the production line digital twin model knowledge base is updated. The production line digital twin model is visualized.

[0034] In the embodiment of the present invention, in S11, the physical mapping of the fruit and vegetable drying physical production line to obtain the mapping relationship is specifically implemented as follows:

[0035] Physical production line mapping:

[0036] Geometric modeling: Based on the actual composition of the production line, it is divided into the equipment layer, perception layer, and control layer.

[0037] The equipment layer includes drying boxes, conveyors, dampers and other equipment. The geometric models of each device are accurately constructed through 3D modeling software, and then rendered and portrayed so that they present an appearance and structure consistent with the actual equipment in the virtual environment.

[0038] The perception layer includes temperature sensors, humidity sensors, moisture content sensors, etc. The installation locations of the sensors are accurately marked during modeling to facilitate subsequent data collection and mapping.

[0039] The control layer includes control devices such as PLC controllers and frequency converters, and establishes a connection between them and the device layer, laying the foundation for the construction of linkage logic between devices.

[0040] In the embodiment of the present invention, a geometric model is established based on the mapping relationship and rendered to obtain a rendering model. The specific implementation relationship of configuring motion parameters for devices in the rendering model and establishing a linkage relationship between devices is as follows:

[0041] Use Unity's Animator component to configure equipment motion parameters, such as conveyor speed (adjustable from 0.5-2m / s) and damper opening (0-100%), and establish linkage logic between devices (for example, start the conveyor when the drying box temperature reaches the set value);

[0042] In an embodiment of the present invention, adding a simulation engine to obtain a digital twin model of the production line specifically includes: building a high-fidelity three-dimensional model based on the physical production line, integrating a simulation engine of multiple physical fields, and mapping physical equipment and material properties in real time.

[0043] For translation in the physical field, the Translation function is used to define the linear displacement logic of equipment such as conveyors and transmission belts, and horizontal or vertical position migration is achieved through coordinate system transformation.

[0044] For rotation in the physical field, the Rotate function is used to set the angular motion parameters of rotating components such as fan blades and valves, and circular motion control is achieved based on Euler angles or quaternion calculations.

[0045] In the physics engine, if two devices converge at a position and motion relationship node, the rigid body dynamics component integrates the compound motion of translation and rotation to establish spatial constraints between the devices (such as the pose matching between the material transfer between the conveyor and the drying box). Ultimately, a complete digital model of the production line is generated, forming a virtual twin with both geometric accuracy and motion logic.

[0046] In an embodiment of the present invention, a hierarchical modeling strategy is adopted to construct a digital twin system mapping between physical equipment and material properties through three levels: three-dimensional solid modeling, multi-physics field coupling solution, and real-time data interaction.

[0047] The core mathematical model in the digital twin system can be expressed as:

[0048] Temperature field model;

[0049]

[0050] Where: C P is the specific heat capacity, t is the time, k is the thermal conductivity; ρ is the density, air velocity v; pressure p; D is the diffusion coefficient, is the Laplace operator, T is the temperature field, M is the moisture field, Q loss is the convective heat dissipation term, S(M) is the shrinkage rate function, σ thermal is the thermal stress tensor, ε(T) is a function related to temperature T, E is the elastic modulus, β(T) is the thermal expansion coefficient, which is related to temperature T, △T is the temperature change, is related to the moisture content gradient Related functions.

[0051] Thermodynamic model: Based on the unsteady heat transfer equation, discretized into finite element grid, real-time solution of temperature field distribution: Among them, Q loss Calculate convection heat dissipation by CFD simulation of the air flow field in the drying box;

[0052]

[0053] Where: is the heat conduction term; Q heater is the heating power (unit: W / m 3 ).

[0054] Material drying model: Combine Fick's diffusion law with shrinkage effect correction to predict moisture content changes:

[0055]

[0056] Where: D eff (T,M) is the effective diffusion coefficient (unit: m 2 / s); is the Laplace operator of water content (unit: % / m 2 ); S(M) is the shrinkage rate function.

[0057] Multi-physics coupling modeling process:

[0058] (Thermodynamic modeling) uses the discretization method of the unsteady heat transfer equation:

[0059]

[0060] Where: T n+1 、T n is the temperature field at the n+1th and nth time steps (unit: K); Δt is the time step (unit: s); h is the convective heat transfer coefficient (unit: W / (m 2 K)) is obtained through CFD simulation in ANSYS Fluent, using the k-ε turbulence model to solve the Navier-Stokes equations:

[0061] Where: v i ,v jis the velocity component (unit: m / s); p is the pressure field (unit: Pa); μ: fluid dynamic viscosity (unit: Pa·s); g i is the acceleration due to gravity (unit: m / s 2 ).

[0062] (Material Drying Kinetics) An improved diffusion equation is established based on Fick's second law:

[0063]

[0064] Where: v shrink Represents the shrinkage rate, and the shrinkage rate function adopts the exponential correction form:

[0065] Where: M0 is the initial moisture content; M c is the critical moisture content; A, B, C are material characteristic parameters; n is the shrinkage index.

[0066] (Thermal-mechanical coupling analysis) Establish a two-way coupling equation in COMSOL:

[0067]

[0068] Where: T˙ is the first-order derivative of temperature with respect to time (unit: K / s); q latent (M) represents the latent heat term (unit: W / m 3 ); M˙ is the first-order derivative of water content with respect to time (unit: % / s); α T is the thermal expansion coupling coefficient (unit: 1 / K), which is used to perform stress-strain analysis using SimScale:

[0069]

[0070] Where: σ is the stress tensor (unit: Pa); E is the elastic modulus (unit: Pa); ε is the strain tensor (dimensionless); β is the thermal expansion coefficient (unit: 1 / K); γ is the moisture expansion coefficient (unit: 1 / %); ΔM is the change in moisture content (unit: %); η is the viscoelastic damping coefficient (unit: Pa·s); ε˙ is the strain rate (unit: 1 / s).

[0071] The above model accurately simulates the drying process, provides a basis for optimizing process parameters, and ensures the accuracy of the digital twin.

[0072] In an embodiment of the present invention, data from the physical production line for dried fruits and vegetables is collected and reversely updated to the digital twin model of the production line in real time, specifically including: collecting data from the physical production line for dried fruits and vegetables based on the OPCUA protocol, and reversely updating to the digital twin model of the production line in real time through the MQTT protocol.

[0073] Reverse updating uses dynamic parameter calibration to reversely correct simulation model parameters (such as the diffusion coefficient D_eff and the convective heat transfer coefficient h) using real-time sensor data (such as moisture content, temperature, and humidity). Specifically, the real-time data collected by the sensors is compared with the predicted data of the virtual model, the error is calculated, and then the model parameters are adjusted through an optimization algorithm to align the output of the virtual model with the actual state of the physical entity, ensuring the accuracy and reliability of the digital twin.

[0074] In this embodiment of the present invention, a knowledge graph of the digital twin model of the first production line is established based on the data in the digital twin model of the production line, historical equipment parameters, and fruit and vegetable drying process data, specifically including:

[0075] The time series database (InfluxDB) stores historical data and supports time window sliding queries. The relational database (PostgreSQL) stores knowledge graphs (including equipment parameters, fruit and vegetable drying process formulas, historical cases, etc.). The historical cases include: previous batch drying parameter settings and drying effect data.

[0076] In the embodiment of the present invention, the expert experience knowledge graph, the drying mechanism knowledge graph, and the drying machine learning model graph are established in the following manner:

[0077] Establish a three-layer digital twin knowledge graph: A [expert experience library], B [mechanism model library], and C [machine learning model library];

[0078] a. Expert experience database: This database collects the experience rules of drying process experts, such as "the initial temperature of apple drying is set to 60°C and the humidity is 70%", and uses ontology to structure the representation. This database converts the expert experience into computer-recognizable and processable knowledge, facilitating system calls and reasoning.

[0079] b. Mechanism model library: Contains thermodynamic models, drying models, and equipment dynamics models. Thermodynamic models describe the heat transfer and temperature distribution during the drying process; drying models predict changes in material moisture content; and equipment dynamics models simulate the operating characteristics and dynamic response of equipment, providing a theoretical basis for system simulation and optimization.

[0080] c Machine Learning Model Library: LSTM models are trained based on historical data to predict moisture content. The LSTM model's ability to process time series data is leveraged to capture the long-term dependencies of moisture content changes. Random Forest models are trained to predict energy consumption. The accuracy of energy consumption predictions is improved through ensemble learning of multiple decision trees.

[0081] In this embodiment of the present invention, the fruit and vegetable drying production indicators are optimized based on the first production line digital twin model knowledge graph, expert experience knowledge graph, drying mechanism knowledge graph, and drying machine learning model graph. The production indicators include: temperature deviation, humidity deviation, production line energy consumption, drying time, and finished product qualification rate. The optimization constraints include: equipment temperature and humidity fluctuations, specifically including:

[0082] The optimization process for temperature deviation, humidity deviation and production line energy consumption is as follows:

[0083] Three-degree-of-freedom model predictive controller (MPC): The control variables are heating power, damper opening, and conveyor speed. The prediction time is 10 steps (10 seconds per step). The optimization objective is to minimize temperature deviation (setpoint ±2°C), humidity deviation (setpoint ±5%), and production line energy consumption. Constraints: heating power ≤ 150kW, damper opening ≥ 30%, and conveyor speed ≤ 2m / s.

[0084] The energy consumption optimization process for energy consumption (kW·h / kg), drying time (h), and finished product qualification rate is as follows:

[0085] An improved NSGA-II algorithm was used to optimize the objective functions of energy consumption (kW·h / kg), drying time (h), and finished product yield. Constraints included equipment temperature ≤ 180°C and humidity fluctuation ≤ ±5%. The algorithm used selection, crossover, and mutation operations to search for the optimal combination of process parameters within the solution space, achieving a balance between multiple optimization objectives.

[0086] The improved NSGA-II algorithm adopts one of the following methods:

[0087] 1. Improve the crossover operator

[0088] SBAX operator: This operator combines the improved arithmetic crossover operator with the simulated binary crossover (SBX) operator to form the SBAX operator. This operator expands the search range, helping the SBX operator escape local optima, improving search efficiency and solution quality. For example, in multi-objective optimization problems such as water conservancy and reservoir operation, the SBAX operator can enable the NSGA-II algorithm to converge faster and achieve better optimization results.

[0089] The seru-swap crossover operator is used to solve specific multi-objective optimization problems, such as the allocation of multi-skilled workers to a seru production system. This operator can effectively improve the performance of the NSGA-II algorithm for such problems.

[0090] Four-point binary crossover and Laplace crossover operators: used for efficient bandwidth allocation in spectrum sharing networks, enabling the NSGA-II algorithm to effectively solve resource allocation problems.

[0091] 2. Improve initialization and selection strategies

[0092] Tent Map Initialization: Using a tent map to initialize the population makes the initial solution more uniform, facilitates exploration of different areas, and enhances initial search capabilities. Furthermore, an adaptive elite selection method based on norm and average distance elimination strategies is proposed. This method selects solutions with good convergence and diversity in the early stages of evolution. Later, a selection method based on average distance elimination is used to evenly distribute the population on the Pareto frontier, promoting algorithmic diversity.

[0093] Hybrid local search strategy: A hybrid local search strategy is added to the offspring generation process, which randomly updates the solution between the optimal individual and its neighbors, enhancing the solution search capability. The effectiveness of this improved algorithm is demonstrated by combining it with the DSSAT model.

[0094] 3. Improve fitness allocation and crowding distance calculation

[0095] Improvements to fitness allocation and crowding distance calculation methods allow for more accurate differentiation between individuals, enhancing the algorithm's ability to maintain diversity and convergence. For example, a new fitness allocation mechanism dynamically adjusts the fitness of individuals based on their distribution and density in the target space, enabling the algorithm to more accurately select high-quality individuals during evolution and improving optimization performance.

[0096] 4. Hybrid Strategy

[0097] Combine NSGA-II with other algorithms to form a hybrid algorithm. For example:

[0098] Combined with Particle Swarm Optimization (PSO): By sharing information and conducting collaborative searches, the algorithm's global convergence speed and local search capabilities are improved, avoiding premature convergence. This allows for more efficient finding of Pareto optimal solutions when solving complex multi-objective optimization problems.

[0099] Combined with simulated annealing (SA): SA's random search capability and Metropolis criterion are used to help NSGA-II escape from local optimal solutions, enhancing the algorithm's global search capability and robustness.

[0100] 5. Adaptive parameter adjustment

[0101] Adaptive adjustment of crossover and mutation probabilities: Dynamically adjust crossover and mutation probabilities based on population diversity and evolutionary state. For example, when population diversity is high, the mutation probability is reduced to speed up convergence; when diversity is low, the mutation probability is increased to prevent premature convergence.

[0102] Adaptive adjustment of evolutionary parameters: Dynamically adjust evolutionary parameters based on multiple factors, including population distribution, evolutionary generations, and individual fitness. For example, in the early stages of evolution, global search capabilities are prioritized, with parameter settings biased towards larger-scale searches. As evolution progresses, localized searches are gradually adopted, and parameters are adjusted accordingly to improve search accuracy.

[0103] In an embodiment of the present invention, when the matching degree between the optimized production index and the corresponding data in the production line digital twin model knowledge base is less than a certain threshold, the method for updating the production line digital twin model knowledge base is as follows:

[0104] When the optimized production metric matches a historical case with a similarity score of less than 0.7, it is marked as a "novel case," triggering machine learning model training. By feeding new case data into the machine learning model library and updating model parameters, the system continuously learns new knowledge and experience, enhancing its adaptability to changing drying scenarios.

[0105] Optimized production indicators refer to the optimal combination of process parameters during the fruit and vegetable drying process.

[0106] In an embodiment of the present invention, the specific implementation method of the visual production line digital twin model is as follows:

[0107] (1) Operation status visualization;

[0108] The Unity XR SDK enables a first-person AR interface, overlaying virtual instrument panels (such as thermometers, tachometers, and hygrometers) onto physical equipment. Operators wearing AR devices can view equipment operating parameters in real time during on-site inspections, creating an immersive monitoring experience.

[0109] Establish a real-time dashboard for equipment operating parameters to intuitively display the real-time changes in equipment parameters such as temperature, humidity, speed, and power in the form of charts and curves, so that operators can understand the operating status of the production line in real time and establish an automatic fault alarm device based on the operating parameters.

[0110] (2) Visualization of the production process;

[0111] The 3D animation displays the material flow path in the production line, overlaying real-time moisture content, temperature and other data, and supports timeline playback function.

[0112] The beneficial effects of the present invention are as follows:

[0113] 1. Multi-source data fusion and state perception

[0114] 1. Integrate dispersed data: This system integrates real-time sensor data, equipment operation logs, environmental parameters, and other multi-dimensional data to build a unified database. This system uses OPCUA and MQTT protocols to achieve efficient data collection and synchronization, resolving the data isolation issues of traditional monitoring systems.

[0115] 2. Improve perception accuracy: Use digital twins to map the physical production line status in real time. Combined with multi-physics field coupling modeling and dynamic parameter calibration, this enables accurate state perception of the entire drying production line process, providing accurate data support for process optimization and control.

[0116] 2. Dynamic Parameter Adaptive Optimization

[0117] 1. Real-time closed-loop correction: Based on real-time data and a historical experience database, the system automatically adjusts drying process parameters (such as temperature, humidity, and wind speed) through a three-degree-of-freedom model predictive controller and a multi-objective optimization engine. When the optimized production indicators do not match the historical cases, the knowledge base self-evolution mechanism is triggered to update the model parameters.

[0118] 2. Significantly improved accuracy: Compared with manual adjustment, parameter control accuracy is improved by more than 30%, adapting to material batch differences and environmental changes.

[0119] 3. Virtual-Real Bidirectional Interaction and Closed-Loop Control

[0120] 1. Bidirectional physical-virtual mapping: Digital twins mirror the physical production line status in real time, with virtual simulation results providing indirect guidance for actual control. This creates a closed-loop optimization mechanism of "data acquisition - simulation prediction - control adjustment," forming a bidirectional interactive production control model.

[0121] 2. Full-process dynamic optimization: Utilizing multi-physics field coupling simulation and model predictive control, the drying process is dynamically optimized throughout the entire process to achieve a synergistic improvement in production efficiency and quality.

[0122] 4. Energy consumption optimization and fault warning

[0123] 1. Energy consumption prediction and optimization: Using machine learning models to predict energy consumption distribution and combining it with a multi-objective optimization engine, we optimize energy distribution and reduce overall energy consumption by more than 15%, meeting low-carbon factory standards.

[0124] 2. Equipment failure warning: Based on historical data and real-time monitoring, the equipment operating status is analyzed and predicted to detect potential equipment failures in advance, reducing downtime by 40% and improving equipment utilization.

[0125] 5. Hybrid Knowledge Base and Self-Evolution Capabilities

[0126] 1. Multimodal knowledge fusion: Integrate expert experience, physical mechanism models, and machine learning models to build a three-layer knowledge graph.

[0127] 2. Adaptive update mechanism: When the matching degree between optimized production indicators and historical cases is lower than 0.7, model training is automatically triggered to enhance system adaptability.

[0128] 6. 3D Visualization and AR Interaction

[0129] 1. Immersive monitoring experience: The Unity XR SDK enables an AR interface, overlaying virtual instruments onto physical devices and supporting first-person inspections.

[0130] 2. Transparency of the production process: 3D dynamic display of production line operation status, material flow and process parameters improves operational convenience.

[0131] 7. Actual application effect

[0132] 1. Efficiency improvement: Production cycle shortened by 20% and equipment utilization increased by 15%.

[0133] 2. Quality assurance: Product uniformity is improved and the browning rate is reduced to below 5%.

[0134] 3. Reduced operation and maintenance costs: Preventive maintenance reduces downtime by 40%, and manual inspection efficiency increases by 5 times.

[0135] 4. Green production: The energy consumption optimization plan reduces carbon emissions by 15%, meeting the standards of a low-carbon factory.

[0136] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention, rather than to limit it. Although the present invention has been described in detail with reference to the above embodiments, those skilled in the art should understand that they can still modify the technical solutions described in the above embodiments, or replace some or all of the technical features therein with equivalents. These modifications or replacements of the technical solutions of the embodiments of the present invention do not cause the essence of the corresponding technical solutions to deviate from the scope of this solution.

Claims

1. A real-time monitoring and optimization method for a dried fruit and vegetable production line based on digital twins, characterized in that: include, S1. Establish a digital twin model of the production line based on the physical production line of dried fruits and vegetables; S2. Collect data from the physical production line for dried fruits and vegetables and update it to the production line digital twin model in real time, monitoring the data in the production line digital twin model; S3. Establish a knowledge base for the production line digital twin model; S4. Optimize the production indicators of dried fruits and vegetables based on the knowledge base.

2. The method according to claim 1, characterized in that The S1 specifically includes: S11. Perform physical mapping on the physical production line of dried fruits and vegetables to obtain a mapping relationship; S12, establishing a geometric model based on the mapping relationship and rendering to obtain a rendering model; S13, configuring motion parameters for devices in the rendering model and establishing linkage relationships between devices; S14. Add a simulation engine to obtain a digital twin model of the production line.

3. The method according to claim 2, characterized in that The S12 specifically includes: establishing a geometric model based on the equipment layer, perception layer, and control layer of the production line and rendering to obtain a rendering model.

4. The method according to claim 3, characterized in that The data collected from the physical production line for dried fruits and vegetables and updated in real time to the digital twin model of the production line specifically include: collecting data from the physical production line for dried fruits and vegetables based on the OPC UA protocol, and updating in real time to the digital twin model of the production line through the MQTT protocol.

5. The method according to claim 4, characterized in that The S3 establishes a production line digital twin model knowledge base, specifically including: a first production line digital twin model knowledge graph, an expert experience knowledge graph, a drying mechanism knowledge graph, and a drying machine learning model graph; The first production line digital twin model knowledge graph is established based on the data in the production line digital twin model, historical equipment parameters and fruit and vegetable drying process data.

6. The method according to claim 5, characterized in that The S4 specifically includes: Based on the knowledge graph of the digital twin model of the first production line, the expert experience knowledge graph, the drying mechanism knowledge graph and the drying machine learning model graph, the production indicators of fruit and vegetable drying are optimized. The production indicators include: production line energy consumption, drying time and finished product qualification rate. The optimization constraints include: equipment temperature and humidity fluctuations.

7. The method according to claim 5, characterized in that When the matching degree between the optimized production indicators and the corresponding data in the production line digital twin model knowledge base is less than a certain threshold, the production line digital twin model knowledge base is updated.

8. The method according to claim 1, characterized in that Also includes: The equipment parameters and production environment parameters in the digital twin model of the visual production line are visualized in a visual interface.

9. The method according to claim 8, characterized in that Provide fault warning based on the parameters in the visual interface.

Images