A menthane amide cooling agent reaction condition prediction system and method
By optimizing the reaction conditions of menthol amide cooling agent using data-driven methods and multi-scale prediction models, the problem of time-consuming and labor-intensive traditional methods has been solved, enabling rapid identification of optimal reaction conditions and efficient production.
Patent Information
- Application Number
- CN202510790732.3
- Authority / Receiving Office
- CN · China
- Patent Type
- Patents(China)
- Current Assignee / Owner
- Filing Date
- 2025-06-13
- Publication Date
- 2026-05-15
- Estimated Expiration
- 2045-06-13
AI Technical Summary
Traditional methods for synthesizing menthol amide cooling agents require extensive experiments and repeated adjustments, which are time-consuming and labor-intensive. They also fail to quickly identify multi-factor interactions, resulting in high experimental costs and wasted resources.
By employing methods such as data collection and preprocessing, establishing correlation models, real-time monitoring and optimization, multi-scale prediction, and closed-loop control, combined with molecular dynamics simulation and macroscopic reactor models, reaction conditions are optimized. Through real-time data feedback and automatic adjustment, the optimal reaction conditions can be quickly identified.
This reduces the number of blind experiments, improves experimental efficiency, lowers material and equipment costs, and ensures optimal reaction conditions and product quality.
Smart Images
Figure CN120656576B_ABST
Abstract
Description
Technical Field
[0001] This invention relates to the technical field of predicting the reaction conditions of menthol amide cooling agents, and specifically to a system and method for predicting the reaction conditions of menthol amide cooling agents. Background Technology
[0002] In the synthesis of menthol amide cooling agents, reaction conditions (such as temperature, reaction time, concentration, and catalyst type) significantly affect the yield and purity of the product. Traditional experimental methods typically require numerous trials and repeated adjustments to find the optimal reaction conditions. This is not only time-consuming and labor-intensive, but the optimization process is often haphazard, making it difficult to guarantee efficient experimental design and optimal reaction results. The experimental design is inefficient: traditional manual experimental design and reaction condition adjustment methods usually require significant time and resources. It cannot fully assess the complex interactions between multiple factors: different reaction factors may have complex interactions, and traditional methods cannot quickly identify the impact of these interactions on the reaction results. In the research and production of chemical reactions, especially before large-scale production, extensive experimental verification can lead to high experimental costs. Traditional methods often rely on numerous experiments and cannot quickly determine the optimal reaction conditions, resulting in unnecessary waste of resources such as experimental equipment, reagents, and catalysts. Summary of the Invention
[0003] A method for predicting the reaction conditions of menthol amide cooling agent includes the following steps;
[0004] S1. Data Collection and Preparation: Identify the main raw materials of menthol amide cooling agent, including menthol, amidation reagent, solvent, and catalyst. Design different reaction conditions, including temperature, reaction time, solvent type, catalyst type, and reactant concentration. Record the product characteristics under different reaction conditions, including yield, purity, reaction rate, and cooling effect. Use molecular dynamics simulation to simulate the reaction mechanism, determine the reaction pathway and energy barrier, determine the interaction between reactants, solvent, and catalyst, and analyze their impact on product selectivity.
[0005] S2. Data Preprocessing and Feature Extraction: Process experimental data, remove outliers and missing values to ensure data accuracy and integrity, extract key variables from experimental data: temperature, time, concentration, solvent, and catalyst as input features, and extract reaction kinetic features, including reaction rate and activation energy, using the results of molecular simulations;
[0006] S3. Establish a correlation model between reaction conditions and product properties: Use supervised learning methods to model the data, analyze the relationship between reaction conditions and product properties, optimize the selection of reaction conditions, and gradually optimize product yield or purity through interaction with the environment (experimental results). Combine molecular-level reaction mechanism models and macroscopic reactor models to establish a multi-scale prediction model.
[0007] S4. Model Training and Validation: Divide the data into training and validation sets, train the model using the training set, evaluate the model's accuracy using cross-validation, adjust the model parameters, and evaluate the model's predictive accuracy using metrics such as mean squared error (MSE) and R². Based on the validated model, input new reaction conditions to make predictions and obtain the expected product characteristics (such as yield, purity, etc.).
[0008] S5. Optimization and Practical Validation of Reaction Conditions: The Design of Experiments (DoE) method is used to systematically optimize reaction conditions, generate different experimental schemes and evaluate the performance of the products. Online monitoring technologies (such as spectral analysis, infrared analysis, etc.) are introduced to track the reaction process in real time, obtain real-time data to optimize reaction conditions, and adjust the model according to the experimental results. The reaction conditions are optimized through real-time monitoring data and experimental feedback.
[0009] S6. Multi-factor synergistic optimization: Consider multiple factors (such as solvent polarity, pH value, reactor type, etc.) for synergistic optimization to ensure the optimal combination of reaction conditions, while optimizing the thermodynamic and kinetic conditions in the reaction process to ensure the energy efficiency and reaction rate of the reaction;
[0010] S7. Real-time data-driven closed-loop control: Real-time data is collected using an automated experimental platform and dynamically adjusted in conjunction with a reaction condition model. Based on real-time feedback from real-time reaction data and prediction models, reaction conditions are automatically adjusted to ensure that the reaction is always in the optimal state.
[0011] Furthermore, a method for predicting the reaction conditions of menthol amide cooling agents is provided.
[0012] In step S3, the product yield and purity are optimized step by step through interaction with the environment. A multi-scale prediction model is established by combining the molecular-level reaction mechanism model and the macroscopic reactor model. The specific steps are as follows:
[0013] S31. Environmental Interaction Data Collection and Definition: Reaction Conditions: including temperature, time, reactant concentration, catalyst type and concentration, and solvent type;
[0014] Reactor characteristics: including reactor volume, stirring rate, and heat exchange efficiency;
[0015] Real-time monitoring data: Collect real-time data on temperature, concentration, pH value, and product concentration during the reaction process using sensors or online analysis tools (such as infrared spectroscopy, mass spectrometry, and gas chromatography);
[0016] Based on the above data, a molecular-level reaction mechanism model was established: through quantum chemical calculation methods, an intermolecular interaction model was established to describe the interaction and transformation of reactants, catalysts, and solvents at the molecular level;
[0017] By combining fluid dynamics, thermodynamic behavior, and mass transfer effects in the reactor, a macroscopic model of the reactor is established to simulate the overall reaction process.
[0018] S32. Model Building and Environmental Interaction Design: Quantum chemical methods are used to simulate the interaction between reactants and catalysts, predict key information such as reaction pathways and activation energies, molecular dynamics (MD) simulations are used to analyze molecular motion and reaction processes, predict reaction rates and pathways, and heat and mass transfer equations are used to simulate heat exchange and mass transfer within the reactor, optimize reaction rates and material conversion efficiency, and combine molecular mechanism models with macroscopic reactor models to form a multi-scale model from micro to macro, describing the microscopic mechanisms of the reaction process at the molecular level and predicting the overall reaction situation in the reactor at the macroscopic level.
[0019] S33. Environmental Interaction Optimization and Prediction: Real-time monitoring data, including temperature, concentration, and pH, are used as input parameters for the model and dynamically adjusted. By monitoring the reaction process through real-time data, reaction conditions are automatically adjusted. Feedback from real-time environmental data helps to quickly identify changes in the reaction process, thereby optimizing product yield and purity. After the experiment and reaction process are completed, the feedback system—the integration of the control system and the model prediction system—adjusts the conditions for the next reaction, making the prediction closer to the actual results. Based on environmental feedback, the model parameters are automatically adjusted, and the model is updated regularly to continuously improve the accuracy of predictions.
[0020] S34. Experimental Verification and Model Calibration: Based on the prediction results of the multi-scale model, conduct small-scale and pilot-scale experimental verification to ensure the reliability of the model prediction. Verify the accuracy of the model by comparing the monitoring data of the reaction process (such as yield, purity, etc.) with the model prediction results. Use experimental data to calibrate the model, especially when new reaction mechanisms or behavioral characteristics appear during the reaction process, which require adjustment of the model through feedback.
[0021] S35. Continuous optimization and intelligent adjustment: The multi-scale model is combined with the reactor control system to form a closed-loop control system. The reaction conditions are adjusted according to real-time data to gradually optimize the reaction process. The model prediction and actual data are continuously fed back, and the parameters in the control system, such as temperature, pressure and reaction time, are automatically adjusted. During the experiment, reinforcement learning algorithm is introduced to continuously optimize the reaction process through continuous interaction with the environment, thereby improving the yield and purity of the product.
[0022] S36. Achieve industrialization and large-scale production: After successful verification in the laboratory and pilot-scale, apply the multi-scale prediction model to the large-scale production process. Through feedback and optimization of the multi-scale model, ensure that the scaled-up production process can achieve similar yield and purity as under laboratory conditions. As data accumulates during the production process, continuously optimize the multi-scale model and upgrade it to meet new process requirements or new reaction conditions.
[0023] Furthermore, a method for predicting the reaction conditions of menthol amide cooling agents is provided.
[0024] In step S5, the Design of Experiments (DoE) method is used to systematically optimize reaction conditions, generate different experimental schemes, and evaluate the performance of the products. The specific steps are as follows.
[0025] S51. Define experimental objectives and response variables: Clarify the optimization objectives of the experiment, including improving the yield, purity and other key properties of the product: reaction rate, cooling effect. The response variables are indicators used to measure the experimental results, including the reaction yield, product purity, reaction time and temperature. These variables need to be quantified in order to analyze their relationship with the reaction conditions.
[0026] S52. Determine Factors and Levels: Factors are those that influence the reaction outcome. In the optimization of menthol amide cooling agent, factors include: reaction temperature, reaction time, reactant concentration, catalyst type and concentration, solvent type, and stirring rate. Each factor should be set with several different levels: value ranges including: reaction temperature: low (20°C), medium (40°C), high (60°C); reaction time: short (1 hour), medium (3 hours), long (5 hours); reactant concentration: low (0.1 mol / L), medium (0.5 mol / L), high (1 mol / L). The number of factor levels depends on the experimental requirements and computational capabilities.
[0027] S53. Perform experiments and collect data: Conduct experiments according to the designed experimental plan: full factorial design. In each experiment, ensure that the reaction conditions, temperature, time, and concentration meet the predetermined factors and levels, and accurately record the reaction results. After each experiment, measure and record the response variables (such as yield, purity, etc.). During the experiment, ensure that the detailed conditions and response data of each experiment are recorded, and indicate the experiment number and other relevant information.
[0028] S54. Data Analysis and Modeling: Use analysis of variance to test whether each factor and its interaction significantly affects the response variable; use regression model to analyze the relationship between the response variable and each factor; and use regression analysis to obtain the mathematical relationship between the response conditions and the response variable.
[0029] S55. Optimization and Prediction: Experimental data is used to verify the accuracy and reliability of the regression model. The predictive ability of the model is evaluated using goodness-of-fit (R²) and mean squared error. Response surface methodology is used to predict product yield and purity under different reaction conditions. The optimal combination of reaction conditions is found using gradient descent to achieve the highest product quality. Reaction conditions are adjusted based on the optimization results, and experiments are conducted to verify the reliability of the optimization results. This ensures that the predicted reaction conditions can achieve optimal results in actual operation.
[0030] S56. Continuous Improvement and Iteration: Based on the deviation between experimental results and model predictions, continuously adjust experimental design and model parameters. During the optimization process, increase or decrease the levels of certain factors to further improve the accuracy and optimization effect of the model. On the basis of laboratory scale, gradually expand the experimental scale to ensure the feasibility of optimized reaction conditions in large-scale production.
[0031] A mentholamide cooling agent reaction condition prediction system is provided, which is used to implement any method for predicting the reaction conditions of a mentholamide cooling agent; the mentholamide cooling agent reaction condition prediction system includes: a data acquisition module, an experimental design module, a mathematical modeling and analysis module, an optimization and prediction module, an experimental feedback module, and a multi-scale integration module;
[0032] Data acquisition module: Real-time acquisition of data on various reaction conditions and response variables during the experiment, and real-time monitoring of product yield, purity, and reaction rate response variables;
[0033] Experimental Design Module: Design different experimental schemes, including factor selection, level setting, and experimental combination, and generate experimental plans based on different experimental design methods: full factorial design, partial factorial design, and response surface methodology;
[0034] Mathematical Modeling and Analysis Module: Based on experimental data, use regression analysis to build mathematical models, analyze the influence of reaction conditions (factors) on product properties (response variables), and calculate the interaction between factors and their contribution to the response;
[0035] Optimization and Prediction Module: Utilizes regression models for optimization, predicts the optimal combination of reaction conditions, and adjusts reaction conditions based on optimization algorithms to achieve the best product yield and purity.
[0036] Experimental Feedback Module: This module validates the optimized predictions experimentally, checking the discrepancies between the predictions and actual responses. Through experimental feedback, the model is adjusted and optimized, gradually improving prediction accuracy.
[0037] Multi-scale integration module: Integrates molecular-level reaction mechanism models with macroscopic reactor models to form a multi-scale model, which combines microscopic reaction processes (such as molecular dynamics simulations) and macroscopic reactor processes (such as fluid dynamics models) for prediction.
[0038] The beneficial effects of this invention are as follows: By employing techniques such as Design of Experiments (DoE) and Response Surface Methodology (RSM), the number of blind trials can be effectively reduced. Optimized experimental design allows for the rapid identification of key reaction factors and optimal reaction conditions, avoiding extensive trial-and-error processes. Optimized experimental design enables the acquisition of accurate reaction conditions in a shorter time, avoiding lengthy debugging processes and improving experimental response speed. Intelligent optimization of experimental design effectively reduces waste of raw materials, reagents, and catalysts. Reliable results are obtained with fewer experiments, saving on expensive experimental materials and equipment costs. Reduced manual adjustments and ineffective inputs during the experimental process make the experimental process more automated and efficient, reducing the workload of personnel. Attached Figure Description
[0039] Figure 1 A flowchart of a method for predicting reaction conditions of a menthol amide cooling agent; Detailed Implementation
[0040] A method for predicting the reaction conditions of menthol amide cooling agent includes the following steps;
[0041] S1. Data Collection and Preparation: Identify the main raw materials of menthol amide cooling agent, including menthol, amidation reagent, solvent, and catalyst. Design different reaction conditions, including temperature, reaction time, solvent type, catalyst type, and reactant concentration. Record the product characteristics under different reaction conditions, including yield, purity, reaction rate, and cooling effect. Use molecular dynamics simulation to simulate the reaction mechanism, determine the reaction pathway and energy barrier, determine the interaction between reactants, solvent, and catalyst, and analyze their impact on product selectivity.
[0042] S2. Data Preprocessing and Feature Extraction: Process experimental data, remove outliers and missing values to ensure data accuracy and integrity, extract key variables from experimental data: temperature, time, concentration, solvent, and catalyst as input features, and extract reaction kinetic features, including reaction rate and activation energy, using the results of molecular simulations;
[0043] S3. Establish a correlation model between reaction conditions and product properties: Use supervised learning methods to model the data, analyze the relationship between reaction conditions and product properties, introduce reinforcement learning algorithms to optimize the selection of reaction conditions, and gradually optimize product yield or purity through interaction with the environment (experimental results). Combine molecular-level reaction mechanism models and macroscopic reactor models to establish a multi-scale prediction model.
[0044] S4. Model Training and Validation: Divide the data into training and validation sets, train the model using the training set, evaluate the model's accuracy using cross-validation, adjust the model parameters, and evaluate the model's predictive accuracy using metrics such as mean squared error (MSE) and R². Based on the validated model, input new reaction conditions to make predictions and obtain the expected product characteristics (such as yield, purity, etc.).
[0045] S5. Optimization and Practical Validation of Reaction Conditions: The Design of Experiments (DoE) method is used to systematically optimize reaction conditions, generate different experimental schemes and evaluate the performance of the products. Online monitoring technologies (such as spectral analysis, infrared analysis, etc.) are introduced to track the reaction process in real time, obtain real-time data to optimize reaction conditions, and adjust the model according to the experimental results. The reaction conditions are optimized through real-time monitoring data and experimental feedback.
[0046] S6. Multi-factor synergistic optimization: Consider multiple factors (such as solvent polarity, pH value, reactor type, etc.) for synergistic optimization to ensure the optimal combination of reaction conditions, while optimizing the thermodynamic and kinetic conditions in the reaction process to ensure the energy efficiency and reaction rate of the reaction;
[0047] S7. Real-time data-driven closed-loop control: Real-time data is collected using an automated experimental platform and dynamically adjusted in conjunction with a reaction condition model. Based on real-time feedback from real-time reaction data and prediction models, reaction conditions are automatically adjusted to ensure that the reaction is always in the optimal state.
[0048] Furthermore, a method for predicting the reaction conditions of menthol amide cooling agents is provided.
[0049] In step S3, the product yield and purity are optimized step by step through interaction with the environment. A multi-scale prediction model is established by combining the molecular-level reaction mechanism model and the macroscopic reactor model. The specific steps are as follows:
[0050] S31. Environmental Interaction Data Collection and Definition: Reaction Conditions: including temperature, time, reactant concentration, catalyst type and concentration, and solvent type;
[0051] Reactor characteristics: including reactor volume, stirring rate, and heat exchange efficiency;
[0052] Real-time monitoring data: Collect real-time data on temperature, concentration, pH value, and product concentration during the reaction process using sensors or online analysis tools (such as infrared spectroscopy, mass spectrometry, and gas chromatography);
[0053] Based on the above data, a molecular-level reaction mechanism model was established: through quantum chemical calculation methods, an intermolecular interaction model was established to describe the interaction and transformation of reactants, catalysts, and solvents at the molecular level;
[0054] By combining fluid dynamics, thermodynamic behavior, and mass transfer effects in the reactor, a macroscopic model of the reactor is established to simulate the overall reaction process.
[0055] S32. Model Building and Environmental Interaction Design: Quantum chemical methods are used to simulate the interaction between reactants and catalysts, predict key information such as reaction pathways and activation energies, molecular dynamics (MD) simulations are used to analyze molecular motion and reaction processes, predict reaction rates and pathways, and heat and mass transfer equations are used to simulate heat exchange and mass transfer within the reactor, optimize reaction rates and material conversion efficiency, and combine molecular mechanism models with macroscopic reactor models to form a multi-scale model from micro to macro, describing the microscopic mechanisms of the reaction process at the molecular level and predicting the overall reaction situation in the reactor at the macroscopic level.
[0056] S33. Environmental Interaction Optimization and Prediction: Real-time monitoring data, including temperature, concentration, and pH, are used as input parameters for the model and dynamically adjusted. By monitoring the reaction process through real-time data, reaction conditions are automatically adjusted. Feedback from real-time environmental data helps to quickly identify changes in the reaction process, thereby optimizing product yield and purity. After the experiment and reaction process are completed, the feedback system—the integration of the control system and the model prediction system—adjusts the conditions for the next reaction, making the prediction closer to the actual results. Based on environmental feedback, the model parameters are automatically adjusted, and the model is updated regularly to continuously improve the accuracy of predictions.
[0057] S34. Experimental Verification and Model Calibration: Based on the prediction results of the multi-scale model, conduct small-scale and pilot-scale experimental verification to ensure the reliability of the model prediction. Verify the accuracy of the model by comparing the monitoring data of the reaction process (such as yield, purity, etc.) with the model prediction results. Use experimental data to calibrate the model, especially when new reaction mechanisms or behavioral characteristics appear during the reaction process, which require adjustment of the model through feedback.
[0058] S35. Continuous optimization and intelligent adjustment: The multi-scale model is combined with the reactor control system to form a closed-loop control system. The reaction conditions are adjusted according to real-time data to gradually optimize the reaction process. The model prediction and actual data are continuously fed back, and the parameters in the control system, such as temperature, pressure and reaction time, are automatically adjusted. During the experiment, reinforcement learning algorithm is introduced to continuously optimize the reaction process through continuous interaction with the environment, thereby improving the yield and purity of the product.
[0059] S36. Achieve industrialization and large-scale production: After successful verification in the laboratory and pilot-scale, apply the multi-scale prediction model to the large-scale production process. Through feedback and optimization of the multi-scale model, ensure that the scaled-up production process can achieve similar yield and purity as under laboratory conditions. As data accumulates during the production process, continuously optimize the multi-scale model and upgrade it to meet new process requirements or new reaction conditions.
[0060] Furthermore, a method for predicting the reaction conditions of menthol amide cooling agents is provided.
[0061] In step S5, the Design of Experiments (DoE) method is used to systematically optimize reaction conditions, generate different experimental schemes, and evaluate the performance of the products.
[0062] S51. Define experimental objectives and response variables: Clarify the optimization objectives of the experiment, including improving the yield, purity and other key properties of the product: reaction rate, cooling effect. The response variables are indicators used to measure the experimental results, including the reaction yield, product purity, reaction time and temperature. These variables need to be quantified in order to analyze their relationship with the reaction conditions.
[0063] S52. Determine Factors and Levels: Factors are those that influence the reaction outcome. In the optimization of menthol amide cooling agent, factors include: reaction temperature, reaction time, reactant concentration, catalyst type and concentration, solvent type, and stirring rate. Each factor should be set with several different levels: value ranges including: reaction temperature: low (20°C), medium (40°C), high (60°C); reaction time: short (1 hour), medium (3 hours), long (5 hours); reactant concentration: low (0.1 mol / L), medium (0.5 mol / L), high (1 mol / L). The number of factor levels depends on the experimental requirements and computational capabilities.
[0064] S53. Perform experiments and collect data: Conduct experiments according to the designed experimental plan: full factorial design. In each experiment, ensure that the reaction conditions, temperature, time, and concentration meet the predetermined factors and levels, and accurately record the reaction results. After each experiment, measure and record the response variables (such as yield, purity, etc.). During the experiment, ensure that the detailed conditions and response data of each experiment are recorded, and indicate the experiment number and other relevant information.
[0065] S54. Data Analysis and Modeling: Use analysis of variance to test whether each factor and its interaction significantly affects the response variable; use regression model to analyze the relationship between the response variable and each factor; and use regression analysis to obtain the mathematical relationship between the response conditions and the response variable.
[0066] S55. Optimization and Prediction: Experimental data is used to verify the accuracy and reliability of the regression model. The predictive ability of the model is evaluated using goodness-of-fit (R²) and mean squared error. Response surface methodology is used to predict product yield and purity under different reaction conditions. The optimal combination of reaction conditions is found using gradient descent to achieve the highest product quality. Reaction conditions are adjusted based on the optimization results, and experiments are conducted to verify the reliability of the optimization results. This ensures that the predicted reaction conditions can achieve optimal results in actual operation.
[0067] S56. Continuous Improvement and Iteration: Based on the deviation between experimental results and model predictions, continuously adjust experimental design and model parameters. During the optimization process, increase or decrease the levels of certain factors to further improve the accuracy and optimization effect of the model. On the basis of laboratory scale, gradually expand the experimental scale to ensure the feasibility of optimized reaction conditions in large-scale production.
[0068] A mentholamide cooling agent reaction condition prediction system is provided, which is used to implement any method for predicting the reaction conditions of a mentholamide cooling agent; the mentholamide cooling agent reaction condition prediction system includes: a data acquisition module, an experimental design module, a mathematical modeling and analysis module, an optimization and prediction module, an experimental feedback module, and a multi-scale integration module;
[0069] Data acquisition module: Real-time acquisition of data on various reaction conditions and response variables during the experiment, and real-time monitoring of product yield, purity, and reaction rate response variables;
[0070] Experimental Design Module: Design different experimental schemes, including factor selection, level setting, and experimental combination, and generate experimental plans based on different experimental design methods: full factorial design, partial factorial design, and response surface methodology;
[0071] Mathematical Modeling and Analysis Module: Based on experimental data, use regression analysis to build mathematical models, analyze the influence of reaction conditions (factors) on product properties (response variables), and calculate the interaction between factors and their contribution to the response;
[0072] Optimization and Prediction Module: Utilizes regression models for optimization, predicts the optimal combination of reaction conditions, and adjusts reaction conditions based on optimization algorithms to achieve the best product yield and purity.
[0073] Experimental Feedback Module: This module validates the optimized predictions experimentally, checking the discrepancies between the predictions and actual responses. Through experimental feedback, the model is adjusted and optimized, gradually improving prediction accuracy.
[0074] Multi-scale integration module: Integrates molecular-level reaction mechanism models with macroscopic reactor models to form a multi-scale model, which combines microscopic reaction processes (such as molecular dynamics simulations) and macroscopic reactor processes (such as fluid dynamics models) for prediction.
Claims
1. A method for predicting the reaction conditions of menthol amide cooling agent, characterized in that, Includes the following steps; S1. Data Collection and Preparation: Determine the main raw materials of menthol amide cooling agent, including menthol, amidation reagent, solvent, and catalyst. Design different reaction conditions, including temperature, reaction time, solvent type, catalyst type, and reactant concentration. Record the yield, purity, reaction rate, and cooling effect under different reaction conditions. Use molecular dynamics simulation to simulate the reaction mechanism, determine the reaction path and energy barrier, determine the interaction between reactants, solvent, and catalyst, and analyze their impact on product selectivity. S2. Data Preprocessing and Feature Extraction: Process experimental data, remove outliers and missing values to ensure data accuracy and integrity, extract key variables from experimental data: temperature, time, concentration, solvent, and catalyst as input features, and extract reaction kinetic features, including reaction rate and activation energy, using the results of molecular simulations; S3. Establish a correlation model between reaction conditions and product properties: Use supervised learning methods to model the data, analyze the relationship between reaction conditions and product properties, optimize the selection of reaction conditions, and gradually optimize product yield and purity through interaction with the environment. Combine molecular-level reaction mechanism models and macroscopic reactor models to establish a multi-scale prediction model. S4. Model Training and Validation: Divide the data into training and validation sets, train the model using the training set, evaluate the model's accuracy using cross-validation, adjust the model parameters, evaluate the model's prediction accuracy using mean squared error and R², and based on the validated model, input new reaction conditions to make predictions and obtain the expected product characteristics: yield and purity. S5. Optimization and Practical Verification of Reaction Conditions: The Design of Experiments (DoE) method is used to systematically optimize reaction conditions, generate different experimental schemes and evaluate the performance of the products. Online monitoring technology, spectral analysis, is introduced to track the reaction process in real time, obtain real-time data to optimize reaction conditions, and adjust the model based on experimental results. Through real-time monitoring data and experimental feedback, reaction conditions are optimized. S6. Multi-factor synergistic optimization: Consider solvent polarity, pH value, and reactor type for synergistic optimization to ensure the optimal combination of reaction conditions, while optimizing the thermodynamic and kinetic conditions in the reaction process to ensure the energy efficiency and reaction rate of the reaction; S7. Real-time data-driven closed-loop control: Real-time data is collected using an automated experimental platform and dynamically adjusted in conjunction with a reaction condition model. Based on real-time feedback from real-time reaction data and prediction models, reaction conditions are automatically adjusted to ensure that the reaction is always in the optimal state.
2. The method for predicting reaction conditions of menthol amide cooling agent as described in claim 1, characterized in that, In step S3, the product yield and purity are optimized step by step through interaction with the environment. A multi-scale prediction model is established by combining the molecular-level reaction mechanism model and the macroscopic reactor model. The specific steps are as follows: S31. Environmental Interaction Data Collection and Definition: Reaction Conditions: including temperature, time, reactant concentration, catalyst type and concentration, and solvent type; Reactor characteristics: including reactor volume, stirring rate, and heat exchange efficiency; Real-time monitoring data: Real-time data on temperature, concentration, pH value, and product concentration are collected through sensors during the reaction process; A molecular-level reaction mechanism model was established based on real-time monitoring data: through quantum chemical calculation methods, an intermolecular interaction model was established to describe the interaction and transformation of reactants, catalysts, and solvents at the molecular level; By combining fluid dynamics, thermodynamic behavior, and mass transfer effects in the reactor, a macroscopic model of the reactor is established to simulate the overall reaction process. S32. Model Building and Environmental Interaction Design: Quantum chemical methods are used to simulate the interaction between reactants and catalysts, predict key information such as reaction pathways and activation energies, molecular dynamics (MD) simulations are used to analyze molecular motion and reaction processes, predict reaction rates and pathways, and heat and mass transfer equations are used to simulate heat exchange and mass transfer within the reactor, optimize reaction rates and material conversion efficiency, and combine molecular mechanism models with macroscopic reactor models to form a multi-scale model from micro to macro, describing the microscopic mechanisms of the reaction process at the molecular level and predicting the overall reaction situation in the reactor at the macroscopic level. S33. Environmental Interaction Optimization and Prediction: Real-time monitored temperature, concentration, and pH are used as input parameters for the model and dynamically adjusted. The reaction process is monitored through real-time data, and reaction conditions are automatically adjusted. Feedback from real-time environmental data can help quickly identify changes in the reaction process, thereby optimizing product yield and purity. After the experiment and reaction process are completed, the feedback system integrates the control system and the model prediction system to adjust the next reaction conditions, making the prediction closer to the actual results. Based on environmental feedback, the model parameters are automatically adjusted and the model is updated regularly to continuously improve the accuracy of prediction. S34. Experimental Verification and Model Calibration: Based on the prediction results of the multi-scale model, conduct small-scale and pilot-scale experimental verification to ensure the reliability of the model prediction. Verify the accuracy of the model by comparing the monitoring data of the reaction process, such as yield and purity, with the model prediction results. Use experimental data to calibrate the model. If new reaction mechanisms or behavioral characteristics appear during the reaction process, the model needs to be adjusted through feedback. S35. Continuous optimization and intelligent adjustment: The multi-scale model is combined with the reactor control system to form a closed-loop control system. The reaction conditions are adjusted according to real-time data to gradually optimize the reaction process. The model prediction and actual data are continuously fed back, and the parameters in the control system, such as temperature, pressure and reaction time, are automatically adjusted. During the experiment, reinforcement learning algorithm is introduced to continuously optimize the reaction process through continuous interaction with the environment, thereby improving the yield and purity of the product. S36. Achieve industrialization and large-scale production: After successful verification in the laboratory and pilot-scale, apply the multi-scale prediction model to the large-scale production process. Through feedback and optimization of the multi-scale model, ensure that the scaled-up production process can achieve similar yield and purity as under laboratory conditions. As data accumulates during the production process, continuously optimize the multi-scale model and upgrade it to meet new process requirements or new reaction conditions.
3. The method for predicting reaction conditions of menthol amide cooling agent as described in claim 1, characterized in that, In step S5, the Design of Experiments (DoE) method is used to systematically optimize reaction conditions, generate different experimental schemes, and evaluate the performance of the products. The specific steps are as follows: S51. Define experimental objectives and response variables: Clarify the optimization objectives of the experiment, including improving the yield, purity and other key properties of the product: reaction rate, cooling effect. The response variables are indicators used to measure the experimental results, including the reaction yield, product purity, reaction time and temperature. These variables need to be quantified in order to analyze their relationship with the reaction conditions. S52. Determine the factors and levels: Factors are the factors that affect the reaction results. In the optimization of menthol amide cooling agent, factors include: reaction temperature, reaction time, reactant concentration, catalyst type and concentration, solvent type, and stirring rate. Each factor should be set with several different levels. The number of factor levels depends on the experimental requirements and computing power. S53. Perform experiments and collect data: Conduct experiments according to the designed experimental plan: full factorial design. In each experiment, ensure that the reaction conditions: temperature, time, and concentration meet the predetermined factors and levels, and accurately record the reaction results. After each experiment, measure and record the response variables: yield and purity. During the experiment, ensure that the detailed conditions and response data of each experiment are recorded, and indicate the experiment number and other relevant information. S54. Data Analysis and Modeling: Use analysis of variance to test whether each factor and its interaction significantly affects the response variable; use regression model to analyze the relationship between the response variable and each factor; and use regression analysis to obtain the mathematical relationship between the response conditions and the response variable. S55. Optimization and Prediction: Use experimental data to verify the accuracy and reliability of the regression model, evaluate the predictive ability of the model by goodness of fit: R² value and mean square error, use response surface methodology to predict product yield and purity under different reaction conditions, find the optimal combination of reaction conditions by gradient descent method to achieve the highest product quality, adjust the reaction conditions according to the optimization results, and conduct experiments to verify the reliability of the optimization results. S56. Continuous Improvement and Iteration: Based on the deviation between experimental results and model predictions, continuously adjust experimental design and model parameters. During the optimization process, increase or decrease the levels of certain factors to further improve the accuracy and optimization effect of the model. Gradually expand the experimental scale based on the laboratory scale.
4. A system for predicting reaction conditions of menthol amide cooling agent, characterized in that, The mentholamide cooling agent reaction condition prediction system is used to implement the mentholamide cooling agent reaction condition prediction method as described in any one of claims 1-3; the mentholamide cooling agent reaction condition prediction system includes: a data acquisition module, an experimental design module, a mathematical modeling and analysis module, an optimization and prediction module, an experimental feedback module, and a multi-scale integration module; Data acquisition module: Real-time acquisition of data on various reaction conditions and response variables during the experiment, and real-time monitoring of product yield, purity, and reaction rate response variables; Experimental Design Module: Design different experimental schemes, including factor selection, level setting, and experimental combination, and generate experimental plans based on different experimental design methods: full factorial design, partial factorial design, and response surface methodology; Mathematical Modeling and Analysis Module: Based on experimental data, mathematical models are established using regression analysis to analyze the influence of reaction conditions on product properties, calculate the interaction between factors and their contribution to the response; Optimization and Prediction Module: Utilizes regression models for optimization, predicts the optimal combination of reaction conditions, and adjusts reaction conditions based on optimization algorithms to achieve the best product yield and purity. Experimental Feedback Module: Based on the optimized prediction results, experimental verification is performed to check the difference between the prediction and the actual response results. Through experimental feedback, the model is adjusted and optimized to gradually improve the prediction accuracy. Multi-scale integration module: Integrates molecular-level reaction mechanism models with macroscopic reactor models to form a multi-scale model, which combines microscopic reaction processes and macroscopic reactor processes for prediction.