Intelligent online optimization control method for pulverizing system of double-inlet and double-outlet coal mill

Abstract

The invention discloses an intelligent online optimization control method for a pulverizing system of a double-inlet and double-outlet coal mill, and relates to the technical field of automatic control. The method is realized based on an optimization control system, and the optimization control system comprises a data acquisition module which comprises a sensor and a data acquisition card for data acquisition and is used for acquiring operation data of a coal mill in real time; and the data processing module comprises a data preprocessing algorithm and a data storage unit and is used for cleaning, filtering and normalizing the collected data and storing the processed data for subsequent use. By collecting and processing the operation data of the coal mill in real time, the change of the system can be quickly responded, the operation parameters are adjusted in time, and the optimal working state of the coal mill is ensured; a machine learning algorithm is adopted for model training and optimization control strategy making, automatic learning and adaptation to different working conditions can be achieved, and the self-adaptive capacity and the intelligent level of the system are improved.

Description

Technical Field

[0001] The invention relates to the technical field of automatic control, and in particular to an intelligent online optimization control method for a double-inlet and double-outlet coal mill pulverizing system. Background Art

[0002] With the continuous growth of energy demand and the improvement of environmental protection requirements, coal mill is one of the important equipment in thermal power plants, and its operating efficiency and energy consumption level have a great impact on the entire power generation process. The traditional coal mill pulverizing system mainly relies on manual experience and simple control strategies for operation and management, which has certain limitations and shortcomings. Therefore, a more intelligent and efficient online optimization control method is needed to achieve precise control and management of coal mills.

[0003] After searching, the application scheme of Chinese patent application number CN202111236190.3 discloses a coal mill control optimization method based on model offline planning, including the following steps: S1. Collect and process the accumulated historical data related to the operation of the coal mill from the power plant; S2. Learn the dynamic model of the coal mill based on the processed data set; S3. Use the model predictive control framework to build a coal mill control optimization model; S4. Solve the constructed optimization model to obtain the control strategy of the coal mill. The control optimization method in the above patent has the following shortcomings: it relies on historical data for learning and prediction, and cannot respond well to the immediate changes in the operating status of the coal mill, and needs to be improved. Summary of the invention

[0004] The purpose of the present invention is to solve the shortcomings of the prior art and to propose an intelligent online optimization control method for a double-inlet and double-outlet coal pulverizing system.

[0005] In order to achieve the above object, the present invention adopts the following technical solutions:

[0006] An intelligent online optimization control method for a double-inlet and double-outlet coal mill pulverizing system is implemented based on an optimization control system, wherein the optimization control system comprises:

[0007] Data acquisition module: including sensors and data acquisition cards for data acquisition, which collect the operation data of the coal mill in real time;

[0008] Data processing module: including data preprocessing algorithm and data storage unit, used to clean, filter and normalize the collected data, and store the processed data for subsequent use;

[0009] Modeling and optimization module: including machine learning algorithms, optimization control algorithms and multi-objective optimization algorithms, used to establish key parameter models and perform optimization calculations;

[0010] Control execution module: including control algorithm and actuator, used to automatically adjust the operating parameters of the coal mill according to the optimization results;

[0011] The optimization control method comprises the following steps:

[0012] S1: Real-time data collection, through the preset sensors to collect the system operation data in real time, including raw coal bunker material level, primary air volume, inlet and outlet temperature, coal mill power, coal powder fineness and other parameters; the collected data is transmitted to the data processing module via wired or wireless means;

[0013] S2: Data preprocessing: cleaning, filtering and normalizing the collected data in the data processing module; the preprocessed data will be stored in the data storage unit for subsequent use;

[0014] S3: Model training: select historical data from the data storage unit as training sets and test sets; select a machine learning algorithm as a modeling tool, use the training set data to train the selected machine learning algorithm to obtain a preliminary model; use cross-validation technology to avoid overfitting problems during training; adjust the model's hyperparameters;

[0015] S4: Model testing and verification. Use the test set data to test the trained model to evaluate its prediction accuracy and generalization ability. If the test result does not meet the requirements, return to S3 to reselect the model or adjust the hyperparameters until it meets the requirements.

[0016] S5: Optimization control strategy formulation: after the model is verified, the optimization control strategy is formulated according to its prediction results; including adjusting the primary air volume and changing the coal feed rate;

[0017] S6: Control execution and feedback adjustment. The formulated optimization control strategy is sent to the control execution module, which automatically adjusts the operating parameters of the coal mill to achieve the best operating state. At the same time, the operating state of the system is monitored in real time, and adjusted and optimized in real time. New operating data is regularly collected to update the model parameters in the model library.

[0018] Preferably: in S1, real-time data collection includes:

[0019] S11: Sensor selection and installation: Select appropriate sensor types according to system requirements, including temperature, pressure, and flow sensors, and install them at the detection location;

[0020] S12: Signal conditioning: amplify, filter and linearize the analog signal output by the sensor and convert it into a standard signal suitable for processing by the data acquisition card;

[0021] S13: analog-to-digital conversion: converting the analog signal after signal conditioning into a digital signal through an analog-to-digital converter;

[0022] S14: Data formatting: Formatting the collected digital signals, including adding timestamps and unit conversion.

[0023] Preferably: in S2, data preprocessing includes:

[0024] S21: Data cleaning: remove outliers and noise data, fill in missing values, and ensure data quality;

[0025] S22: Filtering: Smoothing the data to reduce the impact of fluctuations and noise;

[0026] S23: normalization processing;

[0027] S24: Feature extraction: Extract useful information from the original data to form a feature vector.

[0028] Preferably: in S3, the model training includes:

[0029] S31: data segmentation;

[0030] S32: Feature selection: Select the features with the best information content from the feature vector;

[0031] S33: Algorithm selection: Choose a neural network or support vector machine based on the nature of the problem and the characteristics of the data;

[0032] S34: Parameter setting: Set the algorithm's hyperparameters to control the model training process;

[0033] S35: Initialize weights: initialize the weights of the model;

[0034] S36: Forward propagation: The input data is calculated through the model to obtain the prediction result;

[0035] S37: Loss calculation: Calculate the value of the loss function based on the predicted results and the true label to evaluate the performance of the model;

[0036] S38: Back propagation: Calculate the gradient according to the value of the loss function and update the weight of the model;

[0037] S39: Iterative optimization: Repeat the forward propagation and back propagation process until the loss function converges or reaches a predetermined number of iterations.

[0038] Preferably: In S3, the specific method of model training is as follows:

[0039] Data segmentation: divide the data set into training set, validation set and test set; the training set is used to train the model; the validation set is used to adjust the model parameters and select the best model; the test set is used to finally evaluate the performance of the model;

[0040] Feature selection, which selects the most useful features from the original data to reduce the dimension of the data. Specifically, one or more of the filtering method, wrapping method and embedding method are used;

[0041] Parameter settings, including learning rate, regularization parameter, kernel function type, number of hidden layers and number of neurons.

[0042] Preferably, in S3, when a neural network algorithm is used, the details are as follows:

[0043] Forward propagation, during the forward propagation process, the input data is processed by layers of neurons and finally the output is obtained; the calculation formula for each layer is as follows:

[0044]

[0045]

[0046] in, It is The weighted input of the layer, and They are The weights and biases of the layers, It is The activation value of the layer, is the activation function.

[0047] Preferably, in said S3, when a neural network algorithm is used, it also includes:

[0048] Loss calculation,The loss function is used to measure the difference between the model prediction value and the true value, using mean square error and cross entropy loss, as follows:

[0049] Mean Square Error:

[0050]

[0051] Cross Entropy Loss:

[0052]

[0053] Back propagation, calculate the gradient of the loss function with respect to each parameter, and update the parameters to minimize the loss; the gradient calculation formula is as follows:

[0054]

[0055]

[0056] in, It is The error term of the layer, Represents element-wise product.

[0057] Preferably: in S4, the model testing and verification includes:

[0058] S41: Performance evaluation: Use the test set data to test the model and evaluate its prediction accuracy, recall rate, and F1 score indicators;

[0059] S42: Cross-validation: Use cross-validation techniques to evaluate the generalization ability of the model;

[0060] S43: Hyperparameter adjustment: Adjust the model's hyperparameters, including learning rate and regularization parameters, based on the test results;

[0061] S44: Model saving: Save the trained model to disk or database.

[0062] Preferably, in S5, the optimization control strategy formulation includes:

[0063] S51: Goal determination: clearly define the optimization goals, including maximizing production efficiency, minimizing energy consumption, and meeting environmental emission standards;

[0064] S52: Constraint analysis: Analyze the constraints of each objective function, including the maximum output limit of the coal mill and the range requirements of coal powder fineness;

[0065] S53: Strategy generation: Generate a preliminary optimization control strategy based on the prediction results of the model, including specific measures such as how to adjust the primary air volume and change the coal supply amount;

[0066] S54: Strategy evaluation: Use simulation or actual operation data to evaluate the generated strategy and check its feasibility and effectiveness;

[0067] S55: Strategy iteration: Adjust and improve the strategy based on the evaluation results until satisfactory results are obtained.

[0068] Preferably: in S6, the control execution and feedback adjustment include:

[0069] S61: Control instruction generation: converting the formulated optimization control strategy into specific control instructions, including the opening of the electric control valve and the frequency of the inverter;

[0070] S62: Sending instructions: sending control instructions to the control execution module, which automatically adjusts the operating parameters of the coal mill to achieve the best operating state;

[0071] S63: Status monitoring: real-time monitoring of the system's operating status, including the current, voltage, temperature parameters of the coal mill and the quality indicators of the pulverized coal;

[0072] S64: Feedback collection: collects system feedback information, including coal mill operation data and coal powder quality test results;

[0073] S65: Strategy adjustment: Adjust and optimize the optimization control strategy based on feedback information to ensure the stability and reliability of the control system while improving production efficiency and reducing energy consumption levels.

[0074] The beneficial effects of the present invention are:

[0075] 1. The present invention can quickly respond to system changes and adjust operating parameters in time to ensure the optimal working state of the coal mill by real-time collection and processing of coal mill operation data; it adopts machine learning algorithm for model training and optimization control strategy formulation, can automatically learn and adapt to different working conditions, and improve the system's adaptability and intelligence level.

[0076] 2. The present invention can achieve the goals of maximizing production efficiency and minimizing energy consumption through modeling and optimizing calculation of key parameters, thereby improving the working efficiency and economic benefits of the coal mill; regularly collect new operating data to update the model parameters in the model library, maintain the stability and reliability of the system, and facilitate subsequent maintenance and management. BRIEF DESCRIPTION OF THE DRAWINGS

[0077] Figure 1 The present invention provides a flow chart of an intelligent online optimization control method for a double-inlet and double-outlet coal mill pulverizing system. DETAILED DESCRIPTION

[0078] The technical solution of the present invention is further described in detail below in conjunction with specific implementation methods.

[0079] Embodiment 1:

[0080] An intelligent online optimization control method for a double-inlet and double-outlet coal mill pulverizing system is implemented based on an optimization control system, wherein the optimization control system comprises:

[0081] Data acquisition module: including various sensors (such as temperature sensor, pressure sensor, flow sensor, power sensor, etc.) and data acquisition card, used to collect the operation data of coal mill in real time;

[0082] Data processing module: including data preprocessing algorithm and data storage unit, used to clean, filter and normalize the collected data, and store the processed data for subsequent use;

[0083] Modeling and optimization module: including machine learning algorithms, optimization control algorithms and multi-objective optimization algorithms, etc., used to establish key parameter models and perform optimization calculations;

[0084] Control execution module: includes control algorithms and actuators (such as electric control valves, frequency converters, etc.), which are used to automatically adjust the operating parameters of the coal mill according to the optimization results.

[0085] The optimization control method comprises the following steps:

[0086] S1: Real-time data collection, through the preset sensors to collect the system operation data in real time, including raw coal bunker material level, primary air volume, inlet and outlet temperature, coal mill power, coal powder fineness and other parameters; the collected data is transmitted to the data processing module via wired or wireless means;

[0087] S2: Data preprocessing: cleaning, filtering and normalizing the collected data in the data processing module; the preprocessed data will be stored in the data storage unit for subsequent use;

[0088] S3: Model training, select historical data from the data storage unit as training sets and test sets; select machine learning algorithms (such as neural networks, support vector machines, etc.) as modeling tools, use the training set data to train the selected machine learning algorithms to obtain a preliminary model; use cross-validation and other techniques to avoid overfitting during the training process; adjust the model's hyperparameters (such as learning rate, regularization parameters, etc.) to obtain better performance;

[0089] S4: Model testing and verification. Use the test set data to test the trained model to evaluate its prediction accuracy and generalization ability. If the test result does not meet the requirements, return to S3 to reselect the model or adjust the hyperparameters until it meets the requirements.

[0090] S5: Optimization control strategy formulation: After the model is verified, the optimization control strategy is formulated according to its prediction results; including specific measures such as adjusting the primary air volume and changing the coal feed rate to achieve the best coal grinding output and energy consumption level;

[0091] S6: Control execution and feedback adjustment. The formulated optimization control strategy is sent to the control execution module, which automatically adjusts the operating parameters of the coal mill to achieve the best operating state. At the same time, the operating state of the system is monitored in real time, and adjusted and optimized in real time. New operating data is regularly collected to update the model parameters in the model library.

[0092] Wherein, in said S1, real-time data collection includes:

[0093] S11: Sensor selection and installation: Select appropriate sensor types according to system requirements, including temperature, pressure, flow sensors, etc., and correctly install them on the coal mill and related equipment;

[0094] S12: Signal conditioning: amplify, filter and linearize the analog signal output by the sensor and convert it into a standard signal suitable for processing by the data acquisition card;

[0095] S13: Analog-to-digital conversion: converting the conditioned analog signal into a digital signal through an analog-to-digital converter (ADC);

[0096] S14: Data formatting: Format the collected digital signals, including adding timestamps, unit conversion, etc., to meet subsequent processing requirements.

[0097] Wherein, in said S2, data preprocessing includes:

[0098] S21: Data cleaning: remove outliers and noise data, fill in missing values, and ensure data quality;

[0099] S22: Filtering: Smoothing the data to reduce the impact of fluctuations and noise;

[0100] S23: Normalization: Map the data to a uniform scale for easy subsequent processing;

[0101] S24: Feature extraction: Extract useful information from raw data to form feature vectors for subsequent modeling and optimization.

[0102] Wherein, in S3, model training includes:

[0103] S31: data segmentation;

[0104] S32: Feature selection: Select the features with the best information content from the feature vector;

[0105] S33: Algorithm selection: Select appropriate machine learning algorithms (including neural networks, support vector machines, etc.) based on the nature of the problem and the characteristics of the data;

[0106] S34: Parameter setting: Set the algorithm's hyperparameters to control the model training process;

[0107] S35: Initialize weights: initialize the weights of the model;

[0108] S36: Forward propagation: The input data is calculated through the model to obtain the prediction result;

[0109] S37: Loss calculation: Calculate the value of the loss function based on the predicted results and the true label to evaluate the performance of the model;

[0110] S38: Back propagation: Calculate the gradient according to the value of the loss function and update the weight of the model;

[0111] S39: Iterative optimization: Repeat the forward propagation and back propagation process until the loss function converges or reaches a predetermined number of iterations.

[0112] Among them, in S3, the specific method of model training is as follows:

[0113] Data segmentation: divide the data set into training set, validation set and test set; the training set is used to train the model; the validation set is used to adjust the model parameters and select the best model; the test set is used to finally evaluate the performance of the model;

[0114] Feature selection, which selects the most useful features from the original data to reduce the dimension of the data. Specifically, one or more of the filtering method, wrapping method and embedding method are used;

[0115] Parameter settings, including learning rate, regularization parameter, kernel function type (for SVM), number of hidden layers and neurons (for neural network), etc.

[0116] Wherein, in said S3, when a neural network algorithm is adopted, the details are as follows:

[0117] Forward propagation, during the forward propagation process, the input data is processed by layers of neurons and finally the output is obtained; the calculation formula for each layer is as follows:

[0118]

[0119]

[0120] in, It is The weighted input of the layer, and They are The weights and biases of the layers, It is The activation value of the layer, is the activation function;

[0121] Loss calculation,The loss function is used to measure the difference between the model prediction value and the true value, using mean square error (MSE) and cross entropy loss, as follows:

[0122] Mean Square Error:

[0123]

[0124] Cross Entropy Loss:

[0125]

[0126] Back propagation, calculate the gradient of the loss function with respect to each parameter, and update the parameters to minimize the loss; the gradient calculation formula is as follows:

[0127]

[0128]

[0129] in, It is The error term of the layer, Represents element-wise product.

[0130] Wherein, in said S4, the model testing and verification includes:

[0131] S41: Performance evaluation: Use the test set data to test the model and evaluate its prediction accuracy, recall rate, F1 score and other indicators;

[0132] S42: Cross-validation: Use cross-validation techniques to evaluate the generalization ability of the model and avoid overfitting problems;

[0133] S43: Hyperparameter adjustment: Adjust the model's hyperparameters, including learning rate, regularization parameters, etc., based on the test results to achieve better performance;

[0134] S44: Model saving: Save the trained model to disk or database for subsequent use and deployment.

[0135] Among them, in said S5, the optimization control strategy formulation includes:

[0136] S51: Goal determination: clearly define the optimization goals, including maximizing production efficiency, minimizing energy consumption, and meeting environmental emission standards;

[0137] S52: Constraint analysis: Analyze the constraints of each objective function, including the maximum output limit of the coal mill, the range requirements of coal powder fineness, etc.;

[0138] S53: Strategy generation: Generate a preliminary optimization control strategy based on the prediction results of the model, including specific measures such as how to adjust the primary air volume and change the coal supply amount;

[0139] S54: Strategy evaluation: Use simulation or actual operation data to evaluate the generated strategy and check its feasibility and effectiveness;

[0140] S55: Strategy iteration: Adjust and improve the strategy based on the evaluation results until satisfactory results are obtained.

[0141] Among them, in said S6, the control execution and feedback adjustment include:

[0142] S61: Control instruction generation: converting the formulated optimization control strategy into specific control instructions, including the opening of the electric control valve, the frequency of the inverter, etc.;

[0143] S62: Sending instructions: sending control instructions to the control execution module, which automatically adjusts the operating parameters of the coal mill to achieve the best operating state;

[0144] S63: Status monitoring: real-time monitoring of the system's operating status, including key parameters such as current, voltage, temperature of the coal mill and quality indicators of pulverized coal;

[0145] S64: Feedback collection: Collect system feedback information, including coal mill operation data, coal powder quality test results, etc., for subsequent analysis and decision support;

[0146] S65: Strategy adjustment: Adjust and optimize the optimization control strategy based on feedback information to ensure the stability and reliability of the control system while improving production efficiency and reducing energy consumption levels.

[0147] The above description is only a preferred specific implementation manner of the present invention, but the protection scope of the present invention is not limited thereto. Any technician familiar with the technical field can make equivalent replacements or changes according to the technical scheme and inventive concept of the present invention within the technical scope disclosed by the present invention, which should be covered by the protection scope of the present invention.

Claims

1. An intelligent online optimization control method for a double-inlet and double-outlet coal mill pulverizing system, characterized in that: It is implemented based on an optimized control system, which includes: Data acquisition module: including sensors and data acquisition cards for data acquisition, which collect the operation data of the coal mill in real time; Data processing module: including data preprocessing algorithm and data storage unit, used to clean, filter and normalize the collected data, and store the processed data for subsequent use; Modeling and optimization module: including machine learning algorithms, optimization control algorithms and multi-objective optimization algorithms, used to establish key parameter models and perform optimization calculations; Control execution module: including control algorithm and actuator, used to automatically adjust the operating parameters of the coal mill according to the optimization results; The optimization control method comprises the following steps: S1: Real-time data collection, through the preset sensors, real-time collection of system operation data, including raw coal bunker material level, primary air volume, inlet and outlet temperature, coal mill power, coal powder fineness and other parameters; the collected data is transmitted to the data processing module via wired or wireless means; S2: Data preprocessing: cleaning, filtering and normalizing the collected data in the data processing module; the preprocessed data will be stored in the data storage unit for subsequent use; S3: Model training: select historical data from the data storage unit as training sets and test sets; select a machine learning algorithm as a modeling tool, use the training set data to train the selected machine learning algorithm to obtain a preliminary model; use cross-validation technology to avoid overfitting problems during training; adjust the model's hyperparameters; S4: Model testing and verification. Use the test set data to test the trained model to evaluate its prediction accuracy and generalization ability. If the test result does not meet the requirements, return to S3 to reselect the model or adjust the hyperparameters until it meets the requirements. S5: Optimization control strategy formulation: after the model is verified, the optimization control strategy is formulated according to its prediction results; including adjusting the primary air volume and changing the coal feed rate; S6: Control execution and feedback adjustment. The formulated optimization control strategy is sent to the control execution module, which automatically adjusts the operating parameters of the coal mill to achieve the best operating state. At the same time, the operating state of the system is monitored in real time, and adjusted and optimized in real time. New operating data is regularly collected to update the model parameters in the model library.

2. The intelligent online optimization control method for a double-inlet and double-outlet coal pulverizing system according to claim 1 is characterized in that: In S1, real-time data collection includes: S11: Sensor selection and installation: Select appropriate sensor types according to system requirements, including temperature, pressure, and flow sensors, and install them at the detection location; S12: Signal conditioning: amplify, filter and linearize the analog signal output by the sensor and convert it into a standard signal suitable for processing by the data acquisition card; S13: analog-to-digital conversion: converting the analog signal after signal conditioning into a digital signal through an analog-to-digital converter; S14: Data formatting: Formatting the collected digital signals, including adding timestamps and unit conversion.

3. The intelligent online optimization control method for a double-inlet and double-outlet coal mill pulverizing system according to claim 1 is characterized in that: In S2, data preprocessing includes: S21: Data cleaning: remove outliers and noise data, fill in missing values, and ensure data quality; S22: Filtering: Smoothing the data to reduce the impact of fluctuations and noise; S23: normalization processing; S24: Feature extraction: Extract useful information from the original data to form a feature vector.

4. The intelligent online optimization control method for a double-inlet and double-outlet coal mill pulverizing system according to claim 1 is characterized in that: In S3, model training includes: S31: data segmentation; S32: Feature selection: Select the features with the best information content from the feature vector; S33: Algorithm selection: Choose a neural network or support vector machine based on the nature of the problem and the characteristics of the data; S34: Parameter setting: Set the algorithm's hyperparameters to control the model training process; S35: Initialize weights: initialize the weights of the model; S36: Forward propagation: The input data is calculated through the model to obtain the prediction result; S37: Loss calculation: Calculate the value of the loss function based on the predicted results and the true label to evaluate the performance of the model; S38: Back propagation: Calculate the gradient according to the value of the loss function and update the weight of the model; S39: Iterative optimization: Repeat the forward propagation and back propagation process until the loss function converges or reaches a predetermined number of iterations.

5. The intelligent online optimization control method for a double-inlet and double-outlet coal mill pulverizing system according to claim 4 is characterized in that: In S3, the specific method of model training is as follows: Data segmentation: divide the data set into training set, validation set and test set; the training set is used to train the model; the validation set is used to adjust the model parameters and select the best model; the test set is used to finally evaluate the performance of the model; Feature selection, which selects the most useful features from the original data to reduce the dimension of the data. Specifically, one or more of the filtering method, wrapping method and embedding method are used; Parameter settings, including learning rate, regularization parameter, kernel function type, number of hidden layers and number of neurons.

6. The intelligent online optimization control method for a double-inlet and double-outlet coal mill pulverizing system according to claim 5 is characterized in that: In S3, when a neural network algorithm is used, the details are as follows: Forward propagation, during the forward propagation process, the input data is processed by layers of neurons and finally the output is obtained; the calculation formula for each layer is as follows: in, It is The weighted input of the layer, and They are The weights and biases of the layers, It is The activation value of the layer, is the activation function.

7. The intelligent online optimization control method for a double-inlet and double-outlet coal mill pulverizing system according to claim 6 is characterized in that: In the S3, when a neural network algorithm is used, it also includes: Loss calculation,The loss function is used to measure the difference between the model prediction value and the true value, using mean square error and cross entropy loss, as follows: Mean Square Error: Cross Entropy Loss: Back propagation, calculate the gradient of the loss function with respect to each parameter, and update the parameters to minimize the loss; the gradient calculation formula is as follows: in, It is The error term of the layer, Represents element-wise product.

8. The intelligent online optimization control method for a double-inlet and double-outlet coal mill pulverizing system according to claim 7 is characterized in that: In S4, the model testing and verification includes: S41: Performance evaluation: Use the test set data to test the model and evaluate its prediction accuracy, recall rate, and F1 score indicators; S42: Cross-validation: Use cross-validation techniques to evaluate the generalization ability of the model; S43: Hyperparameter adjustment: Adjust the model's hyperparameters, including learning rate and regularization parameters, based on the test results; S44: Model saving: Save the trained model to disk or database.

9. The intelligent online optimization control method for a double-inlet and double-outlet coal mill pulverizing system according to claim 8 is characterized in that: In S5, the optimization control strategy formulation includes: S51: Goal determination: clearly define the optimization goals, including maximizing production efficiency, minimizing energy consumption, and meeting environmental emission standards; S52: Constraint analysis: Analyze the constraints of each objective function, including the maximum output limit of the coal mill and the range requirements of coal powder fineness; S53: Strategy generation: Generate a preliminary optimization control strategy based on the prediction results of the model, including specific measures such as how to adjust the primary air volume and change the coal supply amount; S54: Strategy evaluation: Use simulation or actual operation data to evaluate the generated strategy and check its feasibility and effectiveness; S55: Strategy iteration: Adjust and improve the strategy based on the evaluation results until satisfactory results are obtained.

10. The intelligent online optimization control method for a double-inlet and double-outlet coal mill pulverizing system according to claim 9 is characterized in that: In S6, the control execution and feedback adjustment include: S61: Control instruction generation: converting the formulated optimization control strategy into specific control instructions, including the opening of the electric control valve and the frequency of the inverter; S62: Sending instructions: sending control instructions to the control execution module, which automatically adjusts the operating parameters of the coal mill to achieve the best operating state; S63: Status monitoring: real-time monitoring of the system's operating status, including the current, voltage, temperature parameters of the coal mill and the quality indicators of the pulverized coal; S64: Feedback collection: collects system feedback information, including coal mill operation data and coal powder quality test results; S65: Strategy adjustment: Adjust and optimize the optimization control strategy based on feedback information to ensure the stability and reliability of the control system while improving production efficiency and reducing energy consumption levels.

Images