Cyclone separator optimization system and method based on machine learning
By constructing a performance database and prediction model based on the impulse theorem and random forest algorithm for cyclone separator optimization, the problems of accuracy and efficiency in cyclone separator optimization are solved, and efficient and accurate industrial adaptive optimization is achieved.
Patent Information
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- NANJING AOBO IND INTELLIGENCE TECH RES INST CO LTD
- Filing Date
- 2025-12-29
- Publication Date
- 2026-05-01
AI Technical Summary
Existing cyclone separator optimization methods suffer from insufficient accuracy and low efficiency, and are difficult to perform large-scale parameter combination calculations, thus failing to meet the high-efficiency and precision requirements of industrial production.
Numerical simulations were performed using a coarse-grained model based on the impulse theorem. The results were verified by combining experimental data and literature. A performance database for cyclone separators was constructed, and a performance prediction model was established using the random forest algorithm. Industrial scenario constraints were set, and parameter combinations were optimized to form an optimal solution.
It achieves efficient and precise optimization of cyclone separators, adapts to fluctuations in industrial scenarios, provides a systematic and practical optimization solution, and improves separation efficiency and reduces operating pressure drop.
Smart Images
Figure CN121960133A_ABST
Abstract
Description
Technical Field
[0001] This invention relates to the field of separation device technology, specifically to a cyclone separator optimization system and method based on machine learning. Background Technology
[0002] Cyclone separators are core gas-solid separation equipment in chemical, power, and metallurgical industries. Their separation efficiency, operating pressure drop, and stability directly affect the energy consumption and product quality of industrial production, thus becoming a key target for industrial optimization.
[0003] Currently, optimization methods for cyclone separators mainly include empirical formula methods, experimental optimization methods, and traditional numerical simulation methods. Empirical formula methods rely on engineering experience for derivation, resulting in limited applicability and insufficient accuracy. Experimental optimization methods require the construction of physical devices, leading to long cycles, high costs, and limitations in parameter adjustment. While traditional numerical simulation methods can reflect flow field characteristics, they are difficult to perform large-scale parameter combination calculations, resulting in low efficiency. Therefore, there is an urgent need for an efficient, accurate, and industrially applicable cyclone separator optimization method to overcome the bottlenecks of existing technologies. Summary of the Invention
[0004] The purpose of this invention is to provide a cyclone separator optimization system and method based on machine learning, which solves the problems existing in the background technology.
[0005] To solve the above-mentioned technical problems, the present invention adopts the following technical solution: a cyclone separator optimization method based on machine learning, specifically including the following steps: S1, based on the impulse theorem, develops coarse-grained... The model will be coarse-grained. The model was applied to the numerical simulation of a cyclone separator and compared with experimental data and literature results for verification. S2 combines the structural parameters of the cyclone separator with the operating conditions to form a parameter combination, which is then coarsened. The model performs large-scale parameter calculations to obtain performance indicators for each parameter combination, and organizes the performance indicators of each parameter combination into a cyclone separator performance database. S3, the data in the cyclone separator performance database is processed using data processing methods, and a cyclone separator performance prediction model is constructed based on the processed data; S4. Based on the performance requirements of cyclone separators in industrial scenarios, set constraints for cyclone separators, and obtain the optimal parameter combination through the cyclone separator performance prediction model. Then, form an optimization scheme for cyclone separators based on the optimal parameter combination. S5, applying the cyclone separator optimization scheme to coarse-grained The model was validated by comparing the differences in performance indicators of cyclone separators, and the performance prediction model of cyclone separators was optimized based on the comparison results.
[0006] Preferably, the coarse-grained development based on the impulse theorem... The model will be coarse-grained. The model was applied to cyclone separator simulation and compared with experimental data and literature results for verification, including the following steps: Constructing a coarse-grained model based on the impulse theorem The model framework, combined with the particle motion characteristics within the cyclone separator, sets parameters for coarse-grained particles and establishes a mapping relationship between the mechanical properties of coarse-grained particles and the actual particle group, thus completing the coarsening process. The basic construction of the model; Based on coarse-grained construction The model was designed with multiple sets of numerical simulation experiments for different coarse-grained particle sizes, while keeping other parameters consistent. The relationship between the calculation results and the coarse-grained particle size was verified by comparing the results of each set of simulation experiments. coarse-grained The model is applied to the numerical simulation of cyclone separators. Based on the simulation results, the separation efficiency and operating pressure drop are extracted. The obtained separation efficiency and operating pressure drop are integrated to form performance data, which are then compared with publicly available experimental data or literature results of cyclone separators. Error analysis is used to verify the simulation results.
[0007] Preferably, the combination of cyclone separator structural parameters and operating conditions to form a parameter combination is achieved through coarse-grained processing. The model performs large-scale parameter calculations to obtain performance indicators for each parameter combination, and then organizes these performance indicators into a cyclone separator performance database. This includes the following steps: S21, combining the industrial application scenarios of cyclone separators and literature research, sets the structural parameters and operating conditions of cyclone separators, and combines the structural parameters and operating conditions to form multiple sets of parameter combinations; S22, based on validated coarse-grained The model performs simulation calculations based on parameter combinations. During the simulation calculation of each parameter combination, only one set of parameters is adjusted at a time, and the simulation results of each parameter combination are recorded. S23. Based on the simulation results of each parameter combination, the adjusted structural parameters, operating conditions and performance indicators are collected, and the collected data are integrated to form a cyclone separator performance database.
[0008] Preferably, the process of processing the data in the cyclone separator performance database using a data processing method and constructing a cyclone separator performance prediction model based on the processed data includes the following steps: S31, based on the physical constraints and data rationality standards of cyclone separator operation, the data in the cyclone separator performance database is screened to remove data that does not conform to physical laws and data standards; S32. Based on the screened cyclone separator performance data, the Z-score standardization method is used to process the data to obtain the standardized cyclone separator performance data. The Z-score standardization formula is shown below: ; in For standardized cyclone separator performance data, The performance data of the filtered cyclone separators, respectively Mean and standard deviation of all samples in the cyclone separator performance database; S33. Based on the standardized cyclone separator performance data, a cyclone separator performance prediction model is established using the random forest algorithm, and the prediction accuracy of the cyclone separator performance prediction model is verified.
[0009] Preferably, the step of establishing a cyclone separator performance prediction model using a random forest algorithm based on the standardized cyclone separator performance data, and verifying the prediction accuracy of the cyclone separator performance prediction model, includes the following steps: The standardized cyclone separator performance data is divided into a training set and a test set. The training set is used to build the cyclone separator performance prediction model, and the test set is used to verify the performance of the cyclone separator performance prediction model. The number of decision trees and the maximum tree depth are set based on the training set data. The structural parameters and operational condition parameters in the training set are used as input features, and the performance index is used as the output target. The training set data is then input into the initialized random forest model for iterative training. During training, information gain is used as the splitting index of decision tree nodes, and decision trees are constructed one by one until all decision trees meet the set maximum tree depth, forming an integrated cyclone separator performance prediction model. The test set data is input into the cyclone separator performance prediction model to obtain the performance prediction results. The difference between the prediction results and the actual test set data is compared by calculating the coefficient of determination. A threshold for the prediction accuracy is set. When the prediction accuracy is higher than the set threshold, the final cyclone separator performance prediction model is obtained.
[0010] The formula for calculating the coefficient of determination is as follows: ; in As the coefficient of determination, For the test set The actual performance index value corresponding to each sample For the test set The predicted performance index values of each sample output by the cyclone separator performance prediction model. This represents the mean of the true performance metrics values for all samples in the test set. This represents the number of samples in the test set.
[0011] Preferably, the process of setting cyclone separator constraints based on industrial scenario performance requirements, obtaining the optimal parameter combination through a cyclone separator performance prediction model, and forming an optimized cyclone separator scheme based on the optimal parameter combination includes the following steps: Based on actual production data from industrial scenarios, the performance requirements for cyclone separator separation efficiency and operating pressure drop are extracted and used as performance constraints. The range of values for cyclone separator structural parameters and the adaptation interval of operating conditions are set and integrated to form a set of constraints. Based on the constraint set, the core objectives are to maximize separation efficiency and minimize operating pressure drop. The weights of separation efficiency and operating pressure drop are set, and the structural parameters and operating conditions of the cyclone separator are used as optimization variables. The fitness function is combined to evaluate the fitness and the parameter combinations are sorted from high to low fitness. The fitness function is shown below: ; in This represents the fitness value for the current parameter combination. To constrain the penalty coefficient, The weights for separation efficiency and operating pressure drop are respectively. The separation efficiency of the current parameter combination. To achieve the target separation efficiency required for industrial scenarios. To meet the target operating pressure reduction requirements of industrial scenarios, This refers to the operating voltage drop corresponding to this parameter combination; The parameter combination with the previous optimal fitness is adjusted based on the fluctuation range of industrial parameters to generate a new parameter combination, fitness evaluation is performed, and invalid combinations that exceed the constraint condition set are filtered out. A fitness fluctuation threshold is set. When the fitness fluctuation of the parameter combination with the optimal fitness is less than the set threshold, this parameter combination is taken as the optimal parameter combination. The optimal parameter combination is substituted into the cyclone separator performance prediction model for repeated evaluation, and after combining the manufacturing and operation feasibility verification in industrial scenarios, the cyclone separator optimization scheme is obtained. Preferably, the cyclone separator optimization scheme is applied to coarse-grained... The model was validated by comparing the differences in cyclone separator performance indicators, and the cyclone separator performance prediction model was optimized based on the comparison results, including the following steps: The resulting cyclone separator optimization scheme was incorporated into the coarse-grained processing. Model, conduct numerical simulation calculations, and obtain the performance data of the optimized cyclone separator based on the calculation results; Using the simulation results of the parameter combination before optimization as a reference, the simulation results before and after optimization are compared and analyzed based on performance indicators, and the performance improvement of the optimization scheme and the existing gaps are recorded according to the comparison results. If the optimized performance indicators meet the requirements of the industrial scenario, the optimization scheme is confirmed to be effective. If the expected results are not achieved, the reasons for the failure are analyzed, the corresponding parameter combinations are fed back to step S2, the cyclone separator performance database is expanded, and the cyclone separator performance prediction model is optimized until an optimization scheme that meets the requirements is obtained.
[0012] This embodiment also discloses a system for optimizing cyclone separators based on machine learning, including: coarsening... The module consists of a model module, a parameter database module, a data processing and model building module, and a solution iteration module. The coarsening The model module is used to construct coarse-grained models based on the impulse theorem. The model was developed and applied to numerical simulation of a cyclone separator, outputting performance data. The parameter database module is used to set the structural parameters and operating conditions of the cyclone separator and combine them into multiple sets of parameter combinations. The structural parameters, operating conditions and performance indicators corresponding to each set of parameter combinations are integrated to form a cyclone separator performance database. The data processing and model building module is used to process performance data using standardized methods and to build a cyclone separator performance prediction model using the random forest algorithm. The scheme iteration module is used to obtain the optimal parameter combination through the cyclone separator performance prediction model, and form an optimized cyclone separator scheme based on the optimal parameter combination. It then determines whether the optimized scheme meets the requirements. If it does not meet the requirements, it optimizes the prediction model and scheme based on the corresponding parameter combination.
[0013] The beneficial effects of this invention are as follows: (1) This invention develops coarse-grained [processes] based on the impulse theorem. The model will be coarse-grained. The model was applied to the numerical simulation of a cyclone separator, and its rationality and accuracy were verified by comparing it with experimental data and literature results. Furthermore, the structural parameters and operating conditions of the cyclone separator were combined to form a parameter set, which was then coarsened. The model performs large-scale parameter calculations to obtain performance indicators for each parameter combination, and organizes these indicators into a cyclone separator performance database. Data processing methods are then used to process the data in this database, and a cyclone separator performance prediction model is constructed based on the processed data. Simultaneously, based on the performance requirements of industrial scenarios, constraints are set for the cyclone separator, and the optimal parameter combination is derived using the performance prediction model. An optimization scheme for the cyclone separator is then developed based on this optimal parameter combination, and finally, this optimization scheme is applied to coarse-grained [processing / optimization]. The model was validated, the differences in the performance indicators of cyclone separators were compared, and the performance prediction model of cyclone separators was optimized based on the comparison results, providing a systematic and practical technical solution for the efficient optimization of cyclone separators.
[0014] (2) The coarsening of the present invention is supported by the impulse theorem With the model at its core, combined with a performance database and random forest algorithm, the constructed predictive model can accurately capture the correlation between parameters and performance. By setting constraints based on industrial needs and verifying across operating conditions, the optimization scheme can be adapted to actual operating condition fluctuations, providing efficient and practical support for cyclone separator optimization. Attached Figure Description
[0015] Figure 1 This is a schematic diagram of the machine learning-based cyclone separator optimization system and method of the present invention. Detailed Implementation
[0016] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.
[0017] Example 1 Please see Figure 1 This embodiment discloses a cyclone separator optimization method based on machine learning, which specifically includes the following steps: S1, based on the impulse theorem, develops coarse-grained... The model will be coarse-grained. The model was applied to the numerical simulation of a cyclone separator and compared with experimental data and literature results for verification. The development of coarse-grained [process] based on the impulse theorem The model will be coarse-grained. The model was applied to cyclone separator simulation and compared with experimental data and literature results for verification, including the following steps: Constructing a coarse-grained model based on the impulse theorem The model framework, combined with the particle motion characteristics within the cyclone separator, sets parameters for coarse-grained particles and establishes a mapping relationship between the mechanical properties of coarse-grained particles and the actual particle group, thus completing the coarsening process. The basic construction of the model; Based on coarse-grained construction The model was designed with multiple sets of numerical simulation experiments for different coarse-grained particle sizes, while keeping other parameters consistent. The relationship between the calculation results and the coarse-grained particle size was verified by comparing the results of each set of simulation experiments. coarse-grained The model is applied to the numerical simulation of cyclone separators. Based on the simulation results, the separation efficiency and operating pressure drop are extracted. The obtained separation efficiency and operating pressure drop are integrated to form performance data, which are then compared with publicly available experimental data or literature results of cyclone separators. Error analysis is used to verify the simulation results.
[0018] S2 combines the structural parameters of the cyclone separator with the operating conditions to form a parameter combination, which is then coarsened. The model performs large-scale parameter calculations to obtain performance indicators for each parameter combination, and organizes the performance indicators of each parameter combination into a cyclone separator performance database. The process involves combining the structural parameters of the cyclone separator with operating conditions to form a parameter combination, which is then coarsened. The model performs large-scale parameter calculations to obtain performance indicators for each parameter combination, and then organizes these performance indicators into a cyclone separator performance database. This includes the following steps: S21, combining the industrial application scenarios of cyclone separators and literature research, sets the structural parameters and operating conditions of cyclone separators, and combines the structural parameters and operating conditions to form multiple sets of parameter combinations; S22, based on validated coarse-grained The model performs simulation calculations based on parameter combinations. During the simulation calculation of each parameter combination, only one set of parameters is adjusted at a time, and the simulation results of each parameter combination are recorded. S23. Based on the simulation results of each parameter combination, the adjusted structural parameters, operating conditions and performance indicators are collected, and the collected data are integrated to form a cyclone separator performance database.
[0019] S3, the data in the cyclone separator performance database is processed using data processing methods, and a cyclone separator performance prediction model is constructed based on the processed data; The process of processing data in the cyclone separator performance database using data processing methods and constructing a cyclone separator performance prediction model based on the processed data includes the following steps: S31, based on the physical constraints and data rationality standards of cyclone separator operation, the data in the cyclone separator performance database is screened to remove data that does not conform to physical laws and data standards; S32. Based on the screened cyclone separator performance data, the Z-score standardization method is used to process the data to obtain the standardized cyclone separator performance data. The Z-score standardization formula is shown below: ; in For standardized cyclone separator performance data, The performance data of the filtered cyclone separators, respectively Mean and standard deviation of all samples in the cyclone separator performance database; S33. Based on the standardized cyclone separator performance data, a cyclone separator performance prediction model is established using the random forest algorithm, and the prediction accuracy of the cyclone separator performance prediction model is verified.
[0020] The process of establishing a cyclone separator performance prediction model using a random forest algorithm based on standardized cyclone separator performance data, and verifying the prediction accuracy of the cyclone separator performance prediction model, includes the following steps: The standardized cyclone separator performance data is divided into a training set and a test set. The training set is used to build the cyclone separator performance prediction model, and the test set is used to verify the performance of the cyclone separator performance prediction model. The number of decision trees and the maximum tree depth are set based on the training set data. The structural parameters and operational condition parameters in the training set are used as input features, and the performance index is used as the output target. The training set data is then input into the initialized random forest model for iterative training. During training, information gain is used as the splitting index of decision tree nodes, and decision trees are constructed one by one until all decision trees meet the set maximum tree depth, forming an integrated cyclone separator performance prediction model. The test set data is input into the cyclone separator performance prediction model to obtain the performance prediction results. The difference between the prediction results and the actual test set data is compared by calculating the coefficient of determination. A threshold for the prediction accuracy is set. When the prediction accuracy is higher than the set threshold, the final cyclone separator performance prediction model is obtained.
[0021] The formula for calculating the coefficient of determination is as follows: ; in As the coefficient of determination, For the test set The actual performance index value corresponding to each sample For the test set The predicted performance index values of each sample output by the cyclone separator performance prediction model. This represents the mean of the true performance metrics values for all samples in the test set. This represents the number of samples in the test set.
[0022] S4. Based on the performance requirements of cyclone separators in industrial scenarios, set constraints for cyclone separators, and obtain the optimal parameter combination through the cyclone separator performance prediction model. Then, form an optimization scheme for cyclone separators based on the optimal parameter combination. The process of setting constraints on cyclone separator performance based on industrial scenarios, determining optimal parameter combinations using a cyclone separator performance prediction model, and formulating an optimization scheme for the cyclone separator based on these optimal parameter combinations includes the following steps: Based on actual production data from industrial scenarios, the performance requirements for cyclone separator separation efficiency and operating pressure drop are extracted and used as performance constraints. The range of values for cyclone separator structural parameters and the adaptation interval of operating conditions are set and integrated to form a set of constraints. Based on the constraint set, the core objectives are to maximize separation efficiency and minimize operating pressure drop. The weights of separation efficiency and operating pressure drop are set, and the structural parameters and operating conditions of the cyclone separator are used as optimization variables. The fitness function is combined to evaluate the fitness and the parameter combinations are sorted from high to low fitness. The fitness function is shown below: ; in This represents the fitness value for the current parameter combination. To constrain the penalty coefficient, The weights for separation efficiency and operating pressure drop are respectively. The separation efficiency of the current parameter combination. To achieve the target separation efficiency required for industrial scenarios. To meet the target operating pressure reduction requirements of industrial scenarios, This refers to the operating voltage drop corresponding to this parameter combination; The parameter combination with the previous optimal fitness is adjusted based on the fluctuation range of industrial parameters to generate a new parameter combination, fitness evaluation is performed, and invalid combinations that exceed the constraint condition set are filtered out. A fitness fluctuation threshold is set. When the fitness fluctuation of the parameter combination with the optimal fitness is less than the set threshold, this parameter combination is taken as the optimal parameter combination. The optimal parameter combination is substituted into the cyclone separator performance prediction model for repeated evaluation, and after combining the manufacturing and operation feasibility verification in industrial scenarios, the cyclone separator optimization scheme is obtained. S5, applying the cyclone separator optimization scheme to coarse-grained The model was validated by comparing the differences in performance indicators of cyclone separators, and the performance prediction model of cyclone separators was optimized based on the comparison results.
[0023] The cyclone separator optimization scheme is applied to coarse-grained processing. The model was validated by comparing the differences in cyclone separator performance indicators, and the cyclone separator performance prediction model was optimized based on the comparison results, including the following steps: The resulting cyclone separator optimization scheme was incorporated into the coarse-grained processing. Model, conduct numerical simulation calculations, and obtain the performance data of the optimized cyclone separator based on the calculation results; Using the simulation results of the parameter combination before optimization as a reference, the simulation results before and after optimization are compared and analyzed based on performance indicators, and the performance improvement of the optimization scheme and the existing gaps are recorded according to the comparison results. If the optimized performance indicators meet the requirements of the industrial scenario, the optimization scheme is confirmed to be effective. If the expected results are not achieved, the reasons for the failure are analyzed, the corresponding parameter combinations are fed back to step S2, the cyclone separator performance database is expanded, and the cyclone separator performance prediction model is optimized until an optimization scheme that meets the requirements is obtained.
[0024] Example 2 This embodiment also discloses a system for optimizing cyclone separators based on machine learning, including: coarsening... The module consists of a model module, a parameter database module, a data processing and model building module, and a solution iteration module. The coarsening The model module is used to construct coarse-grained models based on the impulse theorem. The model was developed and applied to numerical simulation of a cyclone separator, outputting performance data. The parameter database module is used to set the structural parameters and operating conditions of the cyclone separator and combine them into multiple sets of parameter combinations. The structural parameters, operating conditions and performance indicators corresponding to each set of parameter combinations are integrated to form a cyclone separator performance database. The data processing and model building module is used to process performance data using standardized methods and to build a cyclone separator performance prediction model using the random forest algorithm. The scheme iteration module is used to obtain the optimal parameter combination through the cyclone separator performance prediction model, and form an optimized cyclone separator scheme based on the optimal parameter combination. It then determines whether the optimized scheme meets the requirements. If it does not meet the requirements, it optimizes the prediction model and scheme based on the corresponding parameter combination.
[0025] It should be noted that, The above content is merely an example and illustration of the concept of the present invention. Those skilled in the art can make various modifications or additions to the specific embodiments described, or use similar methods to replace them, as long as they do not deviate from the concept of the invention or exceed the scope defined by the present invention, and all such modifications and additions should fall within the protection scope of the present invention.
Claims
1. A cyclone separator optimization method based on machine learning, characterized in that, Includes the following steps: S1, based on the impulse theorem, develops coarse-grained... The model will be coarse-grained. The model was applied to the numerical simulation of a cyclone separator and compared with experimental data and literature results for verification. S2 combines the structural parameters of the cyclone separator with the operating conditions to form a parameter combination, which is then coarsened. The model performs large-scale parameter calculations to obtain performance indicators for each parameter combination, and organizes the performance indicators of each parameter combination into a cyclone separator performance database. S3, the data in the cyclone separator performance database is processed using data processing methods, and a cyclone separator performance prediction model is constructed based on the processed data; S4. Based on the performance requirements of cyclone separators in industrial scenarios, set constraints for cyclone separators, and obtain the optimal parameter combination through the cyclone separator performance prediction model. Then, form an optimization scheme for cyclone separators based on the optimal parameter combination. S5, applying the cyclone separator optimization scheme to coarse-grained The model was validated by comparing the differences in performance indicators of cyclone separators, and the performance prediction model of cyclone separators was optimized based on the comparison results.
2. The cyclone separator optimization method based on machine learning according to claim 1, characterized in that, The development of coarse-grained [process] based on the impulse theorem The model will be coarse-grained. The model was applied to cyclone separator simulation and compared with experimental data and literature results for verification, including the following steps: Constructing a coarse-grained model based on the impulse theorem The model framework, combined with the particle motion characteristics within the cyclone separator, sets parameters for coarse-grained particles and establishes a mapping relationship between the mechanical properties of coarse-grained particles and the actual particle group, thus completing the coarsening process. The basic construction of the model; Based on coarse-grained construction The model was designed with multiple sets of numerical simulation experiments for different coarse-grained particle sizes, while keeping other parameters consistent. The relationship between the calculation results and the coarse-grained particle size was verified by comparing the results of each set of simulation experiments. coarse-grained The model is applied to the numerical simulation of cyclone separators. Based on the simulation results, the separation efficiency and operating pressure drop are extracted. The obtained separation efficiency and operating pressure drop are integrated to form performance data, which are then compared with publicly available experimental data or literature results of cyclone separators. Error analysis is used to verify the simulation results.
3. The cyclone separator optimization method based on machine learning according to claim 1, characterized in that, The process involves combining the structural parameters of the cyclone separator with operating conditions to form a parameter combination, which is then coarsened. The model performs large-scale parameter calculations to obtain performance indicators for each parameter combination, and then organizes these performance indicators into a cyclone separator performance database. This includes the following steps: S21, combining the industrial application scenarios of cyclone separators and literature research, sets the structural parameters and operating conditions of cyclone separators, and combines the structural parameters and operating conditions to form multiple sets of parameter combinations; S22, based on validated coarse-grained The model performs simulation calculations based on parameter combinations. During the simulation calculation of each parameter combination, only one set of parameters is adjusted at a time, and the simulation results of each parameter combination are recorded. S23. Based on the simulation results of each parameter combination, the adjusted structural parameters, operating conditions and corresponding performance indicators are collected, and the collected data are integrated to form a cyclone separator performance database.
4. The cyclone separator optimization method based on machine learning according to claim 1, characterized in that, The process of processing data in the cyclone separator performance database using data processing methods and constructing a cyclone separator performance prediction model based on the processed data includes the following steps: S31, based on the physical constraints and data rationality standards of cyclone separator operation, the data in the cyclone separator performance database is screened to remove data that does not conform to physical laws and data standards; S32. Based on the screened cyclone separator performance data, the Z-score standardization method is used to process the data to obtain the standardized cyclone separator performance data. The Z-score standardization formula is shown below: ； in For standardized cyclone separator performance data, The performance data of the filtered cyclone separators, respectively Mean and standard deviation of all samples in the cyclone separator performance database; S33. Based on the standardized cyclone separator performance data, a cyclone separator performance prediction model is established using the random forest algorithm, and the prediction accuracy of the cyclone separator performance prediction model is verified.
5. The cyclone separator optimization method based on machine learning according to claim 4, characterized in that, The process of establishing a cyclone separator performance prediction model using a random forest algorithm based on standardized cyclone separator performance data, and verifying the prediction accuracy of the cyclone separator performance prediction model, includes the following steps: The standardized cyclone separator performance data is divided into a training set and a test set. The training set is used to build the cyclone separator performance prediction model, and the test set is used to verify the performance of the cyclone separator performance prediction model. The number of decision trees and the maximum tree depth are set based on the training set data. The structural parameters and operational condition parameters in the training set are used as input features, and the performance index is used as the output target. The training set data is then input into the initialized random forest model for iterative training. During training, information gain is used as the splitting index of decision tree nodes, and decision trees are constructed one by one until all decision trees meet the set maximum tree depth, forming an integrated cyclone separator performance prediction model. The test set data is input into the cyclone separator performance prediction model to obtain the performance prediction results. The difference between the prediction results and the actual test set data is compared by calculating the coefficient of determination. A threshold for the prediction accuracy is set. When the prediction accuracy is higher than the set threshold, the final cyclone separator performance prediction model is obtained. The formula for calculating the coefficient of determination is as follows: ； in As the coefficient of determination, For the test set The actual performance index value corresponding to each sample For the test set The predicted performance index values of each sample output by the cyclone separator performance prediction model. This represents the mean of the true performance metrics values for all samples in the test set. This represents the number of samples in the test set.
6. The cyclone separator optimization method based on machine learning according to claim 1, characterized in that, The process of setting constraints on cyclone separator performance based on industrial scenarios, determining optimal parameter combinations using a cyclone separator performance prediction model, and formulating an optimization scheme for the cyclone separator based on these optimal parameter combinations includes the following steps: Based on actual production data from industrial scenarios, the performance requirements for cyclone separator separation efficiency and operating pressure drop are extracted and used as performance constraints. The range of values for cyclone separator structural parameters and the adaptation interval of operating conditions are set and integrated to form a set of constraints. Based on the constraint set, the core objectives are to maximize separation efficiency and minimize operating pressure drop. The weights of separation efficiency and operating pressure drop are set, and the structural parameters and operating conditions of the cyclone separator are used as optimization variables. The fitness function is combined to evaluate the fitness and the parameter combinations are sorted from high to low fitness. The fitness function is shown below: ； in This represents the fitness value for the current parameter combination. To constrain the penalty coefficient, The weights for separation efficiency and operating pressure drop are respectively. The separation efficiency of the current parameter combination. To achieve the target separation efficiency required for industrial scenarios. To meet the target operating pressure reduction requirements of industrial scenarios, This refers to the operating voltage drop corresponding to this parameter combination; Based on the fluctuation range of industrial parameters, the parameter combination structure parameters and operating conditions of the previous optimal fitness are adjusted to generate a new parameter combination, fitness is evaluated, and invalid combinations that exceed the constraint condition set are filtered out. A fitness fluctuation threshold is set. When the fitness fluctuation of the parameter combination of the optimal fitness is less than the set threshold, this parameter combination is taken as the optimal parameter combination. The optimal parameter combination is substituted into the cyclone separator performance prediction model for repeated evaluation. After combining the feasibility verification of manufacturing and operation in industrial scenarios, the optimized cyclone separator scheme is obtained.
7. The cyclone separator optimization method based on machine learning according to claim 1, characterized in that, The cyclone separator optimization scheme is applied to coarse-grained processing. The model was validated by comparing the differences in cyclone separator performance indicators, and the cyclone separator performance prediction model was optimized based on the comparison results, including the following steps: The resulting cyclone separator optimization scheme was incorporated into the coarse-grained processing. Model, conduct numerical simulation calculations, and obtain the performance data of the optimized cyclone separator based on the calculation results; Using the simulation results of the parameter combination before optimization as a reference, the simulation results before and after optimization are compared and analyzed based on performance indicators, and the performance improvement of the optimization scheme and the existing gaps are recorded according to the comparison results. If the optimized performance indicators meet the requirements of the industrial scenario, the optimization scheme is confirmed to be effective. If the expected results are not achieved, the reasons for the failure are analyzed, the corresponding parameter combinations are fed back to step S2, the cyclone separator performance database is expanded, and the cyclone separator performance prediction model is optimized until an optimization scheme that meets the requirements is obtained.
8. A system for implementing the machine learning-based cyclone separator optimization method according to claims 1-7, characterized in that, include: coarse-grained The module consists of a model module, a parameter database module, a data processing and model building module, and a solution iteration module. The coarsening The model module is used to construct coarse-grained models based on the impulse theorem. The model was developed and applied to numerical simulation of a cyclone separator, outputting performance data. The parameter database module is used to set the structural parameters and operating conditions of the cyclone separator and combine them into multiple sets of parameter combinations. The structural parameters, operating conditions and performance indicators corresponding to each set of parameter combinations are integrated to form a cyclone separator performance database. The data processing and model building module is used to process performance data using standardized methods and to build a cyclone separator performance prediction model using the random forest algorithm. The scheme iteration module is used to obtain the optimal parameter combination through the cyclone separator performance prediction model, and form an optimized cyclone separator scheme based on the optimal parameter combination. It then determines whether the optimized scheme meets the requirements. If it does not meet the requirements, it optimizes the prediction model and scheme based on the corresponding parameter combination.