Training method and device of crushing production line automatic control model, equipment and medium

Abstract

The application belongs to the technical field of automatic control, and discloses a training method and device, equipment and medium for an automatic control model of a crushing production line. The historical operation parameters of a grinding machine of the crushing production line are obtained, and the corresponding feature data and label data of each sample moment are obtained by processing. The training set and the verification set are divided. The training set is input into the constructed regression network model, and the predicted optimal parameters of the target equipment output by the model are obtained. The loss value between the predicted optimal parameters and the corresponding label data is calculated, the model weight parameters are iteratively optimized until the performance of the model on the verification set meets the standard, the training is terminated, and the automatic control model is obtained. The automatic control model can dynamically optimize and adjust the operation parameters of the crushing production line equipment based on the grinding machine parameters, improve the ability of automatic optimization and adaptation to changes in production practice, and thus improve the production efficiency and production quality, and ensure the efficiency and stability of the entire production process.

Description

Technical Field

[0001] This invention belongs to the field of automatic control technology, specifically relating to a training method, device, equipment, and storage medium for an automatic control model of a crushing production line. Background Technology

[0002] In modern crushing production lines, programmable logic controllers (PLCs) are widely used to control various operational processes from material input to final product output. These control systems can process input signals from various sensors and adjust the operation of mechanical equipment according to preset logic, such as adjusting speed, temperature, and pressure.

[0003] However, PLCs require manual setting of machine parameters before production begins. This reliance on manual input is not only time-consuming but also prone to affecting production efficiency and product quality due to operator inexperience or improper parameter settings. Furthermore, PLC programs are typically fixed and lack flexibility, resulting in an ineffective response to changes in production conditions or material characteristics, leading to inefficiency and resource waste.

[0004] While PLCs can process real-time data, they typically lack the ability to learn from and optimize processes based on that data. Therefore, PLCs cannot automatically improve or adjust their control strategies based on production practices, limiting their capabilities in production process optimization.

[0005] In modern crushing production lines, although PLCs provide basic automated control, their lack of flexibility in responding to changes and the need for manual parameter presets lead to inefficiency and inconsistent production quality. Furthermore, the fixed programs of PLCs and insufficient utilization of real-time data limit their ability to automatically optimize and adapt to changes in production practices. Summary of the Invention

[0006] The purpose of this invention is to provide a training method, device, equipment, and storage medium for an automatic control model of a crushing production line, which enables the trained automatic control model to optimize and adjust the operating parameters of the crushing production line equipment, improve its ability to automatically optimize and adapt to changes in production practice, and thus improve production efficiency and quality.

[0007] The first aspect of this invention discloses a training method for an automatic control model of a crushing production line, comprising:

[0008] The historical operating parameters of the grinding mill in the crushing production line are obtained, and feature data at multiple sample times are sampled from the historical operating parameters.

[0009] Based on the feature data of each sample time, the label data corresponding to each sample time is obtained. The label data is the true optimal parameters of the target equipment in the crushing production line. The target equipment is any one or more equipment in the crushing production line other than the grinding mill.

[0010] The feature data of each sample time point are preprocessed to obtain sample data corresponding to each sample time point; the sample data is then divided into a training set and a validation set.

[0011] Input each sample data in the training set into the constructed regression network model to obtain the optimal prediction parameters of the target device for each sample time point output by the regression network model.

[0012] Calculate the loss value between the predicted optimal parameters of the target device and the corresponding label data for each sample time point;

[0013] Based on the loss value, the weight parameters of the regression network model are iteratively optimized until the performance index of the regression network model on the validation set reaches the performance threshold, and the training is terminated to obtain the automatic control model; wherein, the automatic control model is used to predict the target operating parameters of the target equipment of the crushing production line based on the real-time operating parameters of the grinding mill of the crushing production line, and control the target equipment to operate at the target operating parameters.

[0014] In some embodiments, the feature data at each sample time point are preprocessed to obtain sample data corresponding to each sample time point, including:

[0015] Calculate the median of the feature data at each sample time point, and fill in the missing values ​​in the feature data at each sample time point based on the median;

[0016] A method based on interquartile range is used to remove outliers from the padded feature data.

[0017] The feature data after removing outliers is normalized to obtain the sample data corresponding to each sample time.

[0018] In some embodiments, before inputting the sample data from the training set into the constructed regression network model to obtain the optimal prediction parameters of the target device corresponding to each sample time point output by the regression network model, the method further includes:

[0019] Based on PyTorch, load the ResNet50 pre-trained model and set the number of nodes in the output layer according to the number of parameters of the target equipment in the crushing production line that need to be predicted;

[0020] Remove the activation function of the last layer in the ResNet50 pre-trained model and readjust it to suit the needs of the regression task to obtain the regression network model.

[0021] In some embodiments, the weight parameters of the regression network model are iteratively optimized based on the loss value, including:

[0022] The Adam optimizer is used to automatically adjust the learning rate and calculate the gradient of the loss value with respect to each weight parameter of the regression network model.

[0023] The Adam optimizer is used to update the weight parameters of the regression network model based on the learning rate and gradient.

[0024] In some embodiments, calculating the loss value between the predicted optimal parameters of the target device at each sample time point and the corresponding label data includes:

[0025] Using mean squared error as the loss function, the loss value between the predicted optimal parameters of the target device and the corresponding label data at each sample time point is calculated. The loss value is expressed as:

[0026]

[0027] Where L represents the loss value, MSE represents the mean squared error, and y i It is the label data corresponding to the i-th sample time. The optimal parameters for the regression network model to predict the i-th sample data are given by n, where n represents the total number of sample data in the training set.

[0028] A second aspect of this invention discloses a training device for an automatic control model of a crushing production line, comprising:

[0029] The sampling unit is used to acquire historical operating parameters of the grinding mill in the crushing production line, and to sample feature data from multiple sample times from the historical operating parameters.

[0030] The tag acquisition unit is used to acquire the tag data corresponding to each sample time based on the feature data of each sample time. The tag data is the true optimal parameters of the target equipment in the crushing production line. The target equipment is any one or more equipment in the crushing production line other than the grinding mill.

[0031] The sample processing unit is used to preprocess the feature data of each sample time to obtain the sample data corresponding to each sample time.

[0032] A partitioning unit is used to divide the sample data into a training set and a validation set;

[0033] The prediction unit is used to input the sample data from the training set into the constructed regression network model to obtain the optimal prediction parameters of the target device for each sample time point output by the regression network model.

[0034] The calculation unit is used to calculate the loss value between the predicted optimal parameters of the target device and the corresponding label data at each sample time.

[0035] An iterative unit is used to iteratively optimize the weight parameters of the regression network model based on the loss value until the performance index of the regression network model on the validation set reaches a performance threshold, and then terminate training to obtain an automatic control model; wherein, the automatic control model is used to predict the target operating parameters of the target equipment of the crushing production line based on the real-time operating parameters of the grinding mill of the crushing production line, and control the target equipment to operate at the target operating parameters.

[0036] In some embodiments, the sample processing unit includes:

[0037] The filling sub-unit is used to calculate the median of the feature data at each sample time and fill in the missing values ​​in the feature data at each sample time according to the median;

[0038] Sub-cell removal is used to remove outliers from padded feature data using an interquartile range-based method.

[0039] The normalization sub-unit is used to normalize the feature data after removing outliers to obtain the sample data corresponding to each sample time.

[0040] In some embodiments, it also includes:

[0041] The loading unit is used to load the ResNet50 pre-trained model based on PyTorch before the prediction unit inputs the sample data from the training set into the constructed regression network model and obtains the optimal prediction parameters of the target device corresponding to each sample time point output by the regression network model.

[0042] The building blocks are used to set the number of nodes in the output layer according to the number of parameters of the target equipment in the crushing production line as needed; at the same time, the activation function of the last layer in the ResNet50 pre-trained model is removed and readjusted to adapt to the needs of the regression task to obtain the regression network model.

[0043] A third aspect of the present invention discloses an electronic device, including a memory storing executable program code and a processor coupled to the memory; the processor calls the executable program code stored in the memory to execute the training method of the automatic control model of the crushing production line disclosed in the first aspect.

[0044] The fourth aspect of the present invention discloses a computer-readable storage medium storing a computer program, wherein the computer program causes a computer to execute the training method for the automatic control model of the crushing production line disclosed in the first aspect.

[0045] The beneficial effects of this invention are as follows: By sampling the historical operating parameters of the grinding mill in the crushing production line, feature data at multiple sample times are obtained, and label data corresponding to each sample time is acquired, dividing the data into training and validation sets. The training set is input into a constructed regression network model to obtain the predicted optimal parameters of the target equipment at each sample time output by the regression network model. The loss value between the predicted optimal parameters and the corresponding label data is calculated, and the weight parameters of the regression network model are iteratively optimized until the performance index of the regression network model on the validation set reaches the performance threshold. Training is then terminated to obtain an automatic control model. This automatic control model is used to predict the target operating parameters of the target equipment in the crushing production line based on the real-time operating parameters of the grinding mill and to control the operation of the target equipment. Therefore, the trained automatic control model can dynamically optimize and adjust the operating parameters of the crushing production line equipment based on the grinding mill parameters, improving the ability to automatically optimize and adapt to changes in production practice, thereby improving production efficiency and quality, and ensuring the high efficiency and stability of the entire production process. Attached Figure Description

[0046] The accompanying drawings illustrate specific examples of the technical solutions described in this invention and, together with the detailed embodiments, form part of the specification, serving to explain the technical solutions, principles, and effects of this invention.

[0047] Unless otherwise specified or defined, the same reference numerals in different figures represent the same or similar technical features, and different reference numerals may be used to represent the same or similar technical features.

[0048] Figure 1 This is a flowchart of a training method for an automatic control model of a crushing production line disclosed in an embodiment of the present invention;

[0049] Figure 2 This is a schematic diagram of the structure of a training device for an automatic control model of a crushing production line disclosed in an embodiment of the present invention;

[0050] Figure 3 This is a schematic diagram of the structure of an electronic device disclosed in an embodiment of the present invention.

[0051] Explanation of reference numerals in the attached figures:

[0052] 201. Sampling unit; 202. Tag acquisition unit; 203. Sample processing unit; 204. Segmentation unit; 205. Prediction unit; 206. Calculation unit; 207. Iteration unit; 301. Memory; 302. Processor. Detailed Implementation

[0053] Unless otherwise specified or defined, all technical and scientific terms used herein have the same meaning as commonly understood by one of ordinary skill in the art. When combined with the technical solutions of the invention in a real-world scenario, all technical and scientific terms used herein may also have meanings corresponding to the purpose of achieving the technical solutions of the invention. The terms "first," "second," etc., used herein are merely for distinguishing names and do not represent a specific number or order. The term "and / or" as used herein includes any and all combinations of one or more of the associated listed items.

[0054] It should be noted that when a component is considered "fixed" to another component, it can be directly fixed to the other component or there can be an intervening component; when a component is considered "connected" to another component, it can be directly connected to the other component or there can be an intervening component; when a component is considered "mounted" on another component, it can be directly mounted on the other component or there can be an intervening component; when a component is considered "placed" on another component, it can be directly placed on the other component or there can be an intervening component.

[0055] Unless otherwise specified or defined, the terms "described" or "the" as used herein refer to the technical features or technical content mentioned or described prior to the relevant section, which may be the same as or similar to the technical features or technical content mentioned herein. Furthermore, the terms "comprising" and "having," and any variations thereof, as used herein, are intended to cover non-exclusive inclusion. For example, a process, method, system, product, or apparatus that includes a series of steps or units is not limited to the steps or units listed, but may optionally include steps or units not listed, or may optionally include other steps or units inherent to such processes, methods, products, or apparatus.

[0056] This invention discloses a training method for an automatic control model of a crushing production line, which can be implemented through computer programming. The execution subject of this method can be an electronic device such as a computer, laptop, or tablet, or a training device for the automatic control model of the crushing production line embedded in an electronic device; this invention does not limit this. To facilitate understanding of this invention, specific embodiments will be described in more detail below with reference to the accompanying drawings.

[0057] like Figure 1 As shown, the training method for the automatic control model of the crushing production line includes the following steps 110-160:

[0058] 110. Obtain the historical operating parameters of the grinding mill in the crushing production line, and sample feature data from multiple sample times from the historical operating parameters.

[0059] Sensors are installed on the grinding mill in the crushing production line to monitor and record the mill's operating parameters in real time, including speed, pressure, and temperature. The operating parameters for a specified historical period (such as the past month or the past week) are obtained from the mill's operating parameters as historical operating parameters. Feature data from multiple sample moments are then sampled from the historical operating parameters for subsequent analysis.

[0060] 120. Based on the feature data of each sample time, obtain the label data corresponding to each sample time. The label data is the true optimal parameters of the target equipment in the crushing production line. The target equipment is any one or more equipment in the crushing production line other than the grinding mill.

[0061] Crushing production lines are widely used in various industrial fields, including but not limited to ore processing, chemical raw material crushing, building material preparation, and ceramic material production. Furthermore, similar production lines are also widely used in the pharmaceutical field, especially in the crushing and delivery of drugs.

[0062] In the crushing production line, the material undergoes a first crushing by a crusher, followed by a second crushing by a grinding mill. It then passes through a cyclone separator and a baghouse dust collector for spray drying. The dried material enters a delivery tank and is then transferred to a buffer silo for storage. Next, controlled by a rotary valve, it enters an air classifier for coarse and fine separation. The separated coarse material is returned to the crusher for further processing, while the fine material is conveyed by a first vacuum conveyor to a first ribbon mixer. After mixing with other auxiliary materials or pure water, it enters a powder silo, then is conveyed by a first screw conveyor to a powder tank, and finally by a second vacuum conveyor to a high-speed mixer. Simultaneously, various auxiliary materials (such as liquids A, B, C, and D) are mixed together under the temperature control of a mold temperature controller and fed into the high-speed mixer to mix with the fine material. The mixture is then cooled by a chiller and enters a second ribbon mixer for further mixing before being sent to the powder tank. Finally, it is conveyed by a second screw conveyor to the delivery tank, and finally, a third vacuum conveyor sends the mixture out, completing the entire production process.

[0063] The operating parameters of a grinding mill directly affect the particle size, uniformity, and flowability of materials. If the grinding mill's parameters change, the parameters of subsequent equipment must be adjusted accordingly to adapt to the changes in materials. Failure to make adaptive adjustments may lead to reduced material processing efficiency, unstable product quality, equipment blockage, or even damage, ultimately affecting the overall operating efficiency and product qualification rate of the production line.

[0064] Therefore, the target equipment specifically refers to other equipment located after the grinding mill in the crushing production line, such as cyclone separators, bag filters, air classifiers, first vacuum conveyors, first ribbon mixers, first screw conveyors, second vacuum conveyors, mold temperature controllers, high-speed mixers, second ribbon mixers, second screw conveyors, and third vacuum conveyors. The true optimal parameters of the target equipment are empirical values ​​derived from theoretical and experimental research. They represent the operating parameters of other equipment in the production line when production efficiency and quality reach their optimal performance under different grinding mill parameter conditions.

[0065] For example, feature data x = {v g ,v g_avg ,l g ,s g_in ,s g_out ,h g_in ,h g_out ,t g}, where vg is the grinding speed, v g_avg The average speed of the grinder in the first hour, l g For the grinding machine load, s g_in For the feed particle size of the grinding mill, s g_out For the output particle size of the grinding mill, h g_in For the feed moisture of the grinding mill, h g_out For the output moisture of the grinding mill, t g This refers to the temperature of the grinding machine.

[0066] Tag data y = {v cy ,v bdc ,v si ,v vac1 ,v r1 ,v sc1 ,v vac2 ,t mtc ,v h ,v r2 ,v sc2 ,v vac3}, where v cy For cyclone separator speed, v bdc For bag filter dust collector speed, v si For airflow sieve velocity, v vac1 For the speed of the first vacuum conveyor, v r1 For the speed of the first ribbon mixer, v sc1 For the speed of the first screw conveyor, v vac2 For the speed of the second vacuum conveyor, t mtc For mold temperature controller temperature, v h For high-speed mixer speed, v r2 For the speed of the second ribbon mixer, v sc2For the speed of the second screw conveyor, v vac3 This refers to the speed of the third vacuum conveyor.

[0067] 130. Preprocess the feature data of each sample time to obtain the sample data corresponding to each sample time, and divide the sample data into training set and validation set.

[0068] The feature data of each sample time point are preprocessed, including filling missing values, removing outliers, and normalization, and then divided into training set and validation set in an 8:2 ratio.

[0069] The preprocessing implementation may specifically include the following steps S11 to S13 (not shown):

[0070] S11. Calculate the median of the feature data at each sample time point, and fill in the missing values ​​in the feature data at each sample time point according to the median.

[0071] S12. Based on the interquartile range method, outliers are removed from the filled feature data.

[0072] First, the interquartile range (IQR) in the feature data is calculated as shown in the following formula (1):

[0073] IQR = Q3 - Q1 (1)

[0074] Where Q1 represents the feature value that ranks at the 25th percentile after sorting the feature data of the same feature from smallest to largest, and Q3 represents the feature value that ranks at the 75th percentile.

[0075] Then, feature values ​​in the feature data that are lower than Q1-1.5*IQR or higher than Q3+1.5*IQR are identified as outliers and filtered out.

[0076] S13. Normalize the feature data after removing outliers to obtain sample data corresponding to each sample time.

[0077] Scale each feature value in the feature data to between 0 and 1, as shown in the following formula (2):

[0078] X′ i =(X i -min(X)) / (max(X)-min(X)) (2)

[0079] Among them, X i Let X′ be the i-th feature value in the feature data. i Let X be the normalized value of the i-th eigenvalue, min(X) be the minimum eigenvalue in the eigendata, and max(X) be the maximum eigenvalue in the eigendata.

[0080] Optionally, after dividing the training set, data augmentation can be performed on multiple samples within the training set that correspond one-to-one with multiple sample times. By augmenting the sample data in the training set, the generalization ability and robustness of the model can be improved, aiming to simulate various changes that may be encountered in real-world operating environments. Data augmentation processes include adding noise, scaling, and random occlusion.

[0081] Firstly, adding noise refers to adding noise to the sample data in the training set to simulate the errors and uncertainties in the actual measurement process. The mathematical expression is shown in the following formula (3):

[0082] x noisy =x+τ,τ~N(0,1)(3)

[0083] Where x is the sample data before adding noise, N(0,1) is a standard normal distribution with a mean of 0 and a variance of 1, and τ is the noise drawn from the standard normal distribution. noisy These are sample data with noise added.

[0084] Secondly, scaling refers to simulating changes in sensor sensitivity or different operating conditions by scaling sample data. The mathematical expression is shown in the following formula (4):

[0085] x scaled =αx (4)

[0086] Where α is the scaling factor, and x is the sample data before scaling. scaled This is scaled sample data.

[0087] Third, random occlusion refers to randomly occluding a portion of the sample data to simulate sensor failure or data loss. The mathematical expression is shown in the following formula (5):

[0088]

[0089] Where i is the index of the sample data, M is the randomly selected index of the occluded training set, and x[i] is the value of the original sample data x at index i. masked[i] It is the value of the sample data at index i after random occlusion.

[0090] 140. Input each sample data in the training set into the constructed regression network model to obtain the optimal prediction parameters of the target device corresponding to each sample time point output by the regression network model.

[0091] Before performing step 140, the regression network model can be adjusted based on the existing ResNet50 pre-trained model. For example, by loading the ResNet50 pre-trained model using PyTorch, the number of nodes in the output layer is set according to the number of parameters of the target equipment on the crushing production line to be predicted. Simultaneously, the activation function of the last layer in the ResNet50 pre-trained model is removed and readjusted to adapt to the requirements of the regression task, resulting in a (ResNet50 Regression, or ResNet50-R) regression network model. Adapting to the requirements of the regression task means enabling the output layer to directly predict continuous values.

[0092] Specifically, the output layer of the last layer of the ResNet50 pre-trained model is represented as z = σ(F fc (x out ), where F fc Indicates a fully connected layer, x out σ represents the feature output of the last convolutional layer, and σ represents the activation function. To ensure that the final output of the network is a continuous value, the activation function is removed and readjusted to adapt to the requirements of the regression task, thus obtaining the ResNet50-R regression network model. That is, the ResNet50-R regression network model is obtained by modifying and adjusting the ResNet50 pre-trained model. The automatic control model obtained after training it can be used to predict the optimal operating parameters of the target equipment in the crushing production line. The output layer of the ResNet50-R regression network model can be represented as shown in the following formula (6):

[0093]

[0094] Where W and b represent the weights and biases of the fully connected layer. This represents the final output of the regression network model, i.e., the predicted optimal parameters.

[0095] 150. Calculate the loss value between the predicted optimal parameters of the target device and the corresponding label data for each sample time.

[0096] During training, mean absolute error, Huber loss, or mean-square error (MSE) can be used as loss functions to evaluate the loss value between the predicted optimal parameters of the target device and the corresponding label data output by the model, so as to guide model training. For example, if MSE is used as the loss function to calculate the loss value between the predicted optimal parameters of the target device and the corresponding label data at each sample time, the loss value L can be expressed as shown in the following formula (7):

[0097]

[0098] Where L represents the loss value, MSE represents the mean squared error, and y i It is the label data corresponding to the i-th sample time. The optimal parameters for the regression network model to predict the i-th sample data are the same as those for the i-th sample time step, where n represents the total number of sample data in the training set.

[0099] 160. Based on the loss value, iteratively optimize the weight parameters of the regression network model until the performance index of the regression network model on the validation set reaches the performance threshold, then terminate training to obtain the automatic control model.

[0100] The model can be optimized using optimizers such as gradient descent, stochastic gradient descent, or Adam. For example, the Adam optimizer can be used to automatically adjust the learning rate and iteratively optimize the weight parameters of the regression network model based on the learning rate and loss value until the performance index of the regression network model on the validation set reaches the performance threshold, at which point training is terminated and an automatically controlled model is obtained.

[0101] During training, the Adam optimizer is first used to automatically adjust the learning rate and calculate the gradient of the loss value with respect to each weight parameter of the regression network model. Then, the Adam optimizer is used to update the weight parameters of the regression network model based on the learning rate and these gradients. The update rule can be shown in the following formula (8):

[0102]

[0103] Where ∈ is a small number to prevent the denominator from being 0, usually taken as 10. -8 θ t θ represents the weight parameters of the model after the t-th iteration. t+1 Represents the model's weight parameters after the (t+1)th iteration, where η represents the learning rate. It is the first moment estimate after bias correction (i.e., the exponential moving average of the gradient), expressed as shown in the following formula (9):

[0104]

[0105] Where, m t =β1m t-1 +(1-β1)g t g t It is the gradient after the t-th iteration, and β1 is the exponential decay rate estimated by the first moment;

[0106] The second moment estimate of the gradient (i.e., the exponential moving average of the square of the gradient) is expressed as shown in the following formula (10):

[0107]

[0108] in, β2 is the exponential decay rate estimated by the second moment.

[0109] Next, a performance threshold thresh is set, and the backpropagation iterative optimization process continues until the performance index of the regression network model on the validation set reaches the performance threshold or no longer shows significant improvement. At this point, it is determined that the loss function has been minimized and the performance of the regression network model has been maximized, the training process is terminated, and the regression network model with the last updated weight parameters is identified as the automatic control model obtained after training is complete.

[0110] Among them, accuracy (acc) can be used as the standard for judging performance, that is, the performance index of the regression network model. The formula for calculating acc can be expressed as shown in the following formula (11):

[0111]

[0112] Here, λ represents the allowable error range between the true values ​​(i.e., the true optimal parameters) and the predicted values ​​(i.e., the predicted optimal parameters). After each training iteration, the accuracy acc is calculated using sample data from the validation set and compared with a preset performance threshold thresh. If acc ≥ thresh, the model is considered to have achieved ideal performance on the validation set, and the training process terminates.

[0113] The trained automatic control model is used to predict the target operating parameters of the target equipment in the crushing production line based on the real-time operating parameters of the grinding mill, and to control the target equipment to operate at the target operating parameters. Specifically, the real-time operating parameters of the grinding mill in the crushing production line to be predicted (e.g., grinding mill speed, average speed of the grinding mill in the previous hour, grinding mill load, grinding mill feed particle size, grinding mill discharge particle size, grinding mill feed moisture, grinding mill discharge moisture, and grinding mill temperature) can be input into the automatic control model. Based on the model output, the target operating parameters of the target equipment in the crushing production line (e.g., cyclone separator speed, bag filter speed, air classifier speed, first vacuum conveyor speed, first ribbon mixer speed, first screw conveyor speed, second vacuum conveyor speed, mold temperature controller temperature, high-speed mixer speed, second ribbon mixer speed, second screw conveyor speed, and third vacuum conveyor speed) can be obtained, and the target equipment in the crushing production line can be controlled to operate according to the target operating parameters.

[0114] Therefore, the trained automatic control model can dynamically optimize and adjust the operating parameters of the crushing production line equipment based on the real-time operating parameters of the grinding mill, improve the ability to automatically optimize and adapt to changes in production practice, thereby improving production efficiency and quality, and ensuring the high efficiency and stability of the entire production process.

[0115] In summary, by integrating advanced sensor technology and a deep learning model (ResNet50-R), this invention enables real-time monitoring and automatic optimization of crushing line equipment parameters, thereby improving production efficiency and product quality. Utilizing the MSE and Adam optimizers, network weights can be precisely adjusted and optimized, ensuring the model's stability and high performance under various production conditions. Furthermore, through intelligent data processing and performance monitoring, this invention significantly reduces resource waste, supports sustainable production goals, and promotes the development of industrial automation and intelligent manufacturing.

[0116] like Figure 2 As shown in the figure, an embodiment of the present invention discloses a training device for an automatic control model of a crushing production line, including a sampling unit 201, a tag acquisition unit 202, a sample processing unit 203, a partitioning unit 204, a prediction unit 205, a calculation unit 206, and an iteration unit 207, wherein,

[0117] The sampling unit 201 is used to acquire the historical operating parameters of the grinding mill in the crushing production line and to sample feature data from multiple sample times from the historical operating parameters.

[0118] The tag acquisition unit 202 is used to acquire the tag data corresponding to each sample time based on the feature data of each sample time. The tag data is the true optimal parameters of the target equipment in the crushing production line. The target equipment is any one or more equipment in the crushing production line other than the grinding mill.

[0119] The sample processing unit 203 is used to preprocess the feature data of each sample time to obtain sample data corresponding to each sample time.

[0120] The partitioning unit 204 is used to divide the sample data into a training set and a validation set;

[0121] The prediction unit 205 is used to input the sample data of each sample in the training set into the constructed regression network model to obtain the optimal prediction parameters of the target device corresponding to each sample time point output by the regression network model.

[0122] The calculation unit 206 is used to calculate the loss value between the predicted optimal parameters of the target device and the corresponding label data at each sample time.

[0123] Iteration unit 207 is used to iteratively optimize the weight parameters of the regression network model based on the loss value until the performance index of the regression network model on the validation set reaches the performance threshold, and then terminate the training to obtain the automatic control model. The automatic control model is used to predict the target operating parameters of the target equipment of the crushing production line based on the real-time operating parameters of the grinding mill of the crushing production line, and control the target equipment to run at the target operating parameters.

[0124] As an optional implementation, the sample processing unit 203 includes the following sub-units (not shown):

[0125] The filling sub-unit is used to calculate the median of the feature data at each sample time and fill in the missing values ​​in the feature data at each sample time according to the median;

[0126] Sub-cell removal is used to remove outliers from padded feature data using an interquartile range-based method.

[0127] The normalization sub-unit is used to normalize the feature data after removing outliers to obtain the sample data corresponding to each sample time.

[0128] Optionally, the training device for the above-mentioned automatic control model of the crushing production line also includes the following units not shown:

[0129] The loading unit is used to load the ResNet50 pre-trained model based on PyTorch before the prediction unit 205 inputs the sample data from the training set into the constructed regression network model and obtains the optimal prediction parameters of the target device corresponding to each sample time point output by the regression network model.

[0130] The building blocks are used to set the number of nodes in the output layer according to the number of parameters of the target equipment in the crushing production line as needed; at the same time, the activation function of the last layer in the ResNet50 pre-trained model is removed and readjusted to adapt to the needs of the regression task to obtain the regression network model.

[0131] Optionally, the aforementioned iterative unit 207 is specifically used to automatically adjust the learning rate using the Adam optimizer and calculate the gradient of the loss value relative to each weight parameter of the regression network model. The Adam optimizer is used to update the weight parameters of the regression network model based on the learning rate and gradient until the performance index of the regression network model on the validation set reaches the performance threshold, thereby terminating the training and obtaining an automatically controlled model.

[0132] like Figure 3 As shown, an embodiment of the present invention discloses an electronic device, including a memory 301 storing executable program code and a processor 302 coupled to the memory 301;

[0133] The processor 302 calls the executable program code stored in the memory 301 to execute the training method of the automatic control model of the crushing production line described in the above embodiments.

[0134] This invention also discloses a computer-readable storage medium storing a computer program that causes a computer to execute the training method for the automatic control model of the crushing production line described in the above embodiments.

[0135] The purpose of the above embodiments is to reproduce and derive the technical solution of the present invention by way of example, and to fully describe the technical solution, purpose and effect of the present invention. The purpose is to enable the public to have a more thorough and comprehensive understanding of the disclosure of the present invention, and not to limit the scope of protection of the present invention.

[0136] The above embodiments are not an exhaustive list based on the present invention, and there may be many other embodiments not listed. Any substitutions and improvements made without departing from the concept of the present invention are within the protection scope of the present invention.

Claims

1. A training method for an automatic control model of a crushing production line, characterized in that, include: The historical operating parameters of the grinding mill in the crushing production line are obtained, and feature data at multiple sample times are sampled from the historical operating parameters. Based on the feature data of each sample time, the label data corresponding to each sample time is obtained. The label data is the true optimal parameters of the target equipment in the crushing production line. The target equipment is any one or more equipment in the crushing production line other than the grinding mill. The feature data of each sample time point are preprocessed to obtain sample data corresponding to each sample time point; the sample data is then divided into a training set and a validation set. Input each sample data in the training set into the constructed regression network model to obtain the optimal prediction parameters of the target device for each sample time point output by the regression network model. Calculate the loss value between the predicted optimal parameters of the target device and the corresponding label data for each sample time point; Based on the loss value, the weight parameters of the regression network model are iteratively optimized until the performance index of the regression network model on the validation set reaches the performance threshold, and the training is terminated to obtain the automatic control model; wherein, the automatic control model is used to predict the target operating parameters of the target equipment of the crushing production line based on the real-time operating parameters of the grinding mill of the crushing production line, and control the target equipment to operate at the target operating parameters; Before inputting the sample data from the training set into the constructed regression network model and obtaining the optimal predicted parameters of the target device for each sample time point output by the regression network model, the following steps are also included: Based on PyTorch, load the ResNet50 pre-trained model and set the number of nodes in the output layer according to the number of parameters of the target equipment in the crushing production line that need to be predicted; Remove the activation function of the last layer in the ResNet50 pre-trained model and readjust it to suit the needs of the regression task to obtain the regression network model.

2. The training method for the automatic control model of the crushing production line as described in claim 1, characterized in that, The feature data at each sample time point are preprocessed to obtain the sample data corresponding to each sample time point, including: Calculate the median of the feature data at each sample time point, and fill in the missing values ​​in the feature data at each sample time point based on the median; A method based on interquartile range is used to remove outliers from the padded feature data. The feature data after removing outliers is normalized to obtain the sample data corresponding to each sample time.

3. The training method for the automatic control model of the crushing production line as described in claim 1, characterized in that, Based on the loss value, the weight parameters of the regression network model are iteratively optimized, including: The Adam optimizer is used to automatically adjust the learning rate and calculate the gradient of the loss value with respect to each weight parameter of the regression network model. The Adam optimizer is used to update the weight parameters of the regression network model based on the learning rate and gradient.

4. The training method for the automatic control model of the crushing production line as described in any one of claims 1 to 3, characterized in that, Calculate the loss value between the predicted optimal parameters of the target device and the corresponding label data for each sample time point, including: Using mean squared error as the loss function, the loss value between the predicted optimal parameters of the target device and the corresponding label data at each sample time point is calculated. The loss value is expressed as: Where L represents the loss value, and MSE represents the mean squared error. It is the label data corresponding to the i-th sample time. The optimal parameters for the regression network model to predict the i-th sample data are given by n, where n represents the total number of sample data in the training set.

5. A training device for an automatic control model of a crushing production line, characterized in that, include: The sampling unit is used to acquire historical operating parameters of the grinding mill in the crushing production line, and to sample feature data from multiple sample times from the historical operating parameters. The tag acquisition unit is used to acquire the tag data corresponding to each sample time based on the feature data of each sample time. The tag data is the true optimal parameters of the target equipment in the crushing production line. The target equipment is any one or more equipment in the crushing production line other than the grinding mill. The sample processing unit is used to preprocess the feature data of each sample time to obtain the sample data corresponding to each sample time. A partitioning unit is used to divide the sample data into a training set and a validation set; The prediction unit is used to input the sample data from the training set into the constructed regression network model to obtain the optimal prediction parameters of the target device for each sample time point output by the regression network model. The calculation unit is used to calculate the loss value between the predicted optimal parameters of the target device and the corresponding label data at each sample time. An iterative unit is used to iteratively optimize the weight parameters of the regression network model based on the loss value until the performance index of the regression network model on the validation set reaches a performance threshold, and then terminate training to obtain an automatic control model; wherein, the automatic control model is used to predict the target operating parameters of the target equipment of the crushing production line based on the real-time operating parameters of the grinding mill of the crushing production line, and control the target equipment to operate at the target operating parameters; Also includes: The loading unit is used to load the ResNet50 pre-trained model based on PyTorch before the prediction unit inputs the sample data from the training set into the constructed regression network model and obtains the optimal prediction parameters of the target device corresponding to each sample time point output by the regression network model. The building blocks are used to set the number of nodes in the output layer according to the number of parameters of the target equipment in the crushing production line as needed; at the same time, the activation function of the last layer in the ResNet50 pre-trained model is removed and readjusted to adapt to the needs of the regression task to obtain the regression network model.

6. The training device for the automatic control model of the crushing production line as described in claim 5, characterized in that, The sample processing unit includes: The filling sub-unit is used to calculate the median of the feature data at each sample time and fill in the missing values ​​in the feature data at each sample time according to the median; Sub-cell removal is used to remove outliers from padded feature data using an interquartile range-based method. The normalization sub-unit is used to normalize the feature data after removing outliers to obtain the sample data corresponding to each sample time.

7. An electronic device, characterized in that, It includes a memory storing executable program code and a processor coupled to the memory; the processor calls the executable program code stored in the memory to execute the training method of the automatic control model of the crushing production line according to any one of claims 1 to 4.

8. A computer-readable storage medium, characterized in that, The computer-readable storage medium stores a computer program, wherein the computer program causes a computer to execute the training method for the automatic control model of the crushing production line according to any one of claims 1 to 4.

Images