A blank control method and system for producing warped sheets
Through automatic control of feeding system and neural network technology, the problems of uniformity and continuity of ratios in the production of copper tube outer curling sheets are solved, and an efficient and continuous production process is achieved, ensuring product quality and production efficiency.
Patent Information
- Application Number
- CN202410769347.6
- Authority / Receiving Office
- CN · China
- Patent Type
- Patents(China)
- Current Assignee / Owner
- Filing Date
- 2024-06-14
- Publication Date
- 2024-12-03
- Estimated Expiration
- 2044-06-14
AI Technical Summary
In the production process of copper tube outer flakes, ensuring the uniformity and continuity of the raw materials and mix ratios of each batch of products is a major challenge, affecting production efficiency and product quality.
Adopt automatic feeding system, by monitoring the working status of the hopper and processing equipment, build a hopper control subnet and processing control subnet, establish an automatic feeding model, adjust the feeding parameters and processing parameters in real time, and use neural networks and length memory. The algorithm predicts and adjusts key time nodes and parameters to ensure uniformity of material ratios and smooth production.
The strict mixing of raw materials and additives for each batch of outer flakes is achieved, which reduces product defects and variations, improves production fluency and continuity, reduces production interruptions, ensures accurate control of materials and processing parameters, and adapts to raw materials. Changes in supply and production demand.
Smart Images

Figure CN118789875B_ABST
Abstract
Description
Technical Field
[0001] The present invention relates to the technical field of warped sheet control, and in particular to a blank control method and system for producing warped sheets. Background Art
[0002] In modern industry, copper tubes are widely used in various applications, such as refrigeration and air-conditioning systems, water supply and natural gas transmission. The outer flap components of copper tubes play a key role in these applications. Their role is not only to connect and close, but also to ensure sealing and pressure resistance. For different application environments, the material ratio and processing accuracy of the outer flap need to be optimized according to their functions and environmental requirements. For example, in high-pressure environments, such as natural gas pipelines, the outer flap requires high strength and good corrosion resistance, and reinforced copper alloys or coating technology are often used to improve durability; while in refrigeration systems, the outer flap not only needs to prevent leakage, but also maintain good heat exchange performance, so specific thermal conductive fillers may be added to optimize thermal management. By precisely controlling the ratio and processing of these materials, the performance of the copper tube outer flap can be significantly improved to meet the stringent requirements of different application scenarios.
[0003] In the current production process of warped sheets, facing the diverse demands of different batches of products, ensuring the uniformity and continuity of the raw materials and mixed material ratios of each batch of warped sheets is a major challenge. Since each type of warped sheet may require a specific material ratio to meet specific mechanical and physical property requirements, adjusting and controlling these ratios has a direct impact on production efficiency and product quality.
[0004] The information disclosed in this background technology section is only intended to deepen the understanding of the overall background technology of the present disclosure, and should not be regarded as acknowledging or suggesting in any form that the information constitutes the prior art known to those skilled in the art. Summary of the invention
[0005] The present invention provides a blank control method and system for producing warped sheets, which can effectively solve the problems in the background technology.
[0006] In order to achieve the above object, the technical solution adopted by the present invention is:
[0007] A blank control method and system for producing warped sheets, the method comprising:
[0008] Obtaining material information of raw materials and additives, and determining processing ratio information of the raw materials and additives according to the processing target of the warped sheet blank;
[0009] Monitor the working status of the hopper and collect working parameter information of the hopper;
[0010] Monitor the working state of the warped sheet blank processing equipment and collect control parameter information of the warped sheet blank processing equipment;
[0011] According to the processing ratio information and the working parameter information, a hopper control subnetwork and a processing control subnetwork are respectively constructed corresponding to the hopper and the warped sheet blank processing equipment, wherein the hopper control subnetwork controls the material discharging parameters of the hopper, and the processing control subnetwork controls the processing parameters of the warped sheet blank processing equipment;
[0012] An automatic control feeding model is established based on the hopper control subnetwork and the processing control subnetwork. The automatic control feeding model obtains the working parameter information and the control parameter information, and controls the coordination parameters between the working parameter information and the control parameter information and the control time nodes.
[0013] Furthermore, a hopper control subnetwork is constructed, including:
[0014] Obtaining the processing ratio information of the warped sheet blank, monitoring the hopper, and integrating the monitoring data into a blank parameter subset;
[0015] The elements in the blank parameter subset are divided into blank internal driving parameter information and blank external driving parameter information, wherein the blank internal driving parameter information is the matching relationship between the relevant parameters in the hopper, and the blank external driving parameter information is the relevant parameters for the automatic control feeding model to adjust;
[0016] Establish a blank sample set grid, perform data management on the historical blank parameter subset according to the time series, set an identification label for the processing category of the warped sheet blank corresponding to the blank parameter subset, and predict the internal drive parameter information through the historical blank parameter subset;
[0017] A billet monitoring grid is established to receive data in the billet parameter subset and select a loss function, wherein the loss function adjusts the working parameters of the hopper according to the billet sample set grid.
[0018] Furthermore, a processing control subnetwork is constructed, including:
[0019] Acquire the processing target of the warped sheet blank, monitor the warped sheet blank processing equipment, and integrate the monitoring data into a processing parameter subset;
[0020] The elements in the processing parameter subset are divided into processing internal drive parameter information and processing external drive parameter information, wherein the processing internal drive parameter information is the coordination relationship between relevant parameters in the warped sheet blank processing equipment, and the processing external drive parameter information is the relevant parameters for the automatic control feeding model to allocate;
[0021] Establishing a processing sample set grid, performing data management on the historical processing parameter subset according to the time series, setting an identification label for the processing category of the warped sheet blank corresponding to the processing parameter subset, and predicting the internal drive parameter information through the historical processing parameter subset;
[0022] A processing monitoring grid is established to receive data in the processing parameter subset and select a loss function, wherein the loss function adjusts the working parameters of the warped sheet blank processing equipment according to the processing sample set grid.
[0023] Further, according to the establishment of the billet monitoring grid, receiving the data in the billet parameter subset, and selecting a loss function, the loss function adjusts the working parameters of the hopper according to the billet sample set grid, including:
[0024] Collecting the working parameters of the blank in real time, and preprocessing the collected data to obtain blank data parameters;
[0025] Constructing the number of nodes of the input layer according to the blank data parameters, each node represents one of the working parameters, and setting weights and biases for each of the nodes;
[0026] Setting a feature extraction layer and generating an activation function, and performing feature extraction on the working parameters of each node in the input layer through the activation function, wherein the weight is used to connect each of the feature extraction layers, and the bias is used to adjust the output of the activation function;
[0027] The first layer of the feature extraction layer receives data from the input layer, and the output of the feature extraction layer of the previous layer is used as the input of the feature extraction layer of the next layer until the output layer is reached.
[0028] Furthermore, the mean square error is selected as the loss function, and the mean square error is expressed as follows:
[0029]
[0030] Where n represents the number of samples; y i Indicates the actual monitoring value; represents the predicted value of the blank sample set grid or the processed sample set grid; MSE represents the mean square error.
[0031] Furthermore, the automatic control feeding model obtains the working parameter information and the control parameter information, and controls the matching parameters between the working parameter information and the control parameter information and the control time node, including:
[0032] Obtaining the material feeding speed in the working parameter information and the warped sheet forming speed in the control parameter information;
[0033] A time processing window is set by processing the warped sheet blanks of the same batch, a time anchor point is set within the time processing window by the material feeding speed and the warped sheet forming speed, and the working parameter information and the control parameter information are updated, saved, and forgotten at the time anchor point based on a long-short memory algorithm;
[0034] Acquire the quality information of the processing of the warped sheet blank, and provide feedback on the matching parameters between the working parameter information and the control parameter information according to the quality information of the processing of the warped sheet blank, which is set as a feedback time node within the time processing window;
[0035] Among them, the time anchor point and the feedback time node are the time points at which the automatic control feeding model performs adaptive control on the working parameter information and the control parameter information.
[0036] Furthermore, a training set and a verification set are constructed based on the data information stored in the hopper control subnetwork and the processing control subnetwork, and the automatic control feeding model is trained and verified.
[0037] Furthermore, the parameters of the automatic control feeding model are trained by a back propagation algorithm, including:
[0038] The training set is input into the neural network, and the output of each layer is obtained by the forward propagation process of the neural network, during which the input of each layer undergoes a linear transformation of weights and biases, and then undergoes a nonlinear transformation of an activation function;
[0039] Calculate the loss function value of the automatic control feeding model on the current training sample according to the adaptive adjustment amplitude of the automatic control feeding model at the time anchor point and the feedback time node;
[0040] Use the chain rule to propagate the gradient forward layer by layer, backpropagate the gradient of the loss function with respect to the output layer to the hidden layer and the input layer, and update the parameters of each layer at the same time;
[0041] According to the calculated gradient, the gradient descent method is used to update the parameters of the model so that the loss function gradually decreases;
[0042] Repeat the above steps and perform multiple iterations on the entire training set until the automatic control feeding model converges or reaches a predetermined number of iterations.
[0043] A blank control method and system for producing warped sheets, the system comprising:
[0044] A material information acquisition module, which acquires material information of raw materials and additives, and determines processing ratio information of the raw materials and additives according to the processing target of the warped sheet blank;
[0045] A material discharge status monitoring module monitors the working status of the hopper and collects the working parameter information of the hopper;
[0046] A blank processing monitoring module monitors the working state of the warped sheet blank processing equipment and collects control parameter information of the warped sheet blank processing equipment;
[0047] A subnetwork control module, which constructs a hopper control subnetwork and a processing control subnetwork corresponding to the hopper and the warped sheet blank processing equipment respectively according to the processing ratio information and the working parameter information, wherein the hopper control subnetwork controls the material discharge parameters of the hopper, and the processing control subnetwork controls the processing parameters of the warped sheet blank processing equipment;
[0048] The intelligent control module establishes an automatic control feeding model based on the hopper control subnetwork and the processing control subnetwork. The automatic control feeding model obtains the working parameter information and the control parameter information, and controls the coordination parameters between the working parameter information and the control parameter information and the control time node.
[0049] Furthermore, the intelligent control module includes:
[0050] A sub-network parameter acquisition unit, which acquires a material feeding speed in the working parameter information and a warped sheet forming speed in the control parameter information;
[0051] A time series parameter updating unit is used to set a time processing window for processing the warped sheet blanks of the same batch, set a time anchor point within the time processing window by the material feeding speed and the warped sheet forming speed, and update, save, and forget the working parameter information and the control parameter information at the time anchor point based on a long-short memory algorithm;
[0052] A blank processing quality feedback unit obtains the quality information of the processing of the warped sheet blank, and provides feedback on the matching parameters between the working parameter information and the control parameter information according to the quality information of the processing of the warped sheet blank, and is set as a feedback time node within the time processing window, wherein the time anchor point and the feedback time node are the time points at which the automatic control feeding model performs adaptive control on the working parameter information and the control parameter information.
[0053] The technical solution of the present invention can achieve the following technical effects:
[0054] The real-time monitoring and precise adjustment of the automatic feeding system enhance the uniformity of the ratio, ensuring that the raw materials and additives of each batch of warped flakes are mixed according to strict ratio information. The finely controlled ratio ensures the uniform mixing of the materials and the consistency from one batch to the next, thereby reducing product defects and variations caused by improper material ratios. The automatic feeding model combined with advanced algorithms can predict and adjust key time nodes and parameters in the production process, enhance the smoothness and continuity of the production process, predict and reduce production interruptions caused by sudden downtime or parameter adjustments, and can make advance adjustments when trends that may affect production continuity are detected. The system's built-in monitoring grid and feedback unit can collect and analyze processing data in real time. The real-time feedback mechanism ensures precise control of materials and processing parameters during the production process, thereby maintaining the continuity and uniformity of production. Through continuous monitoring and automatic adjustment, the system can quickly adapt to changes in raw material supply or changes in production demand without long manual reconfiguration or adjustment, maintaining the efficiency and continuous operation of the production process.
[0055] The above description is only an overview of the technical solution of the present application. In order to more clearly understand the technical means of the present application, it can be implemented in accordance with the contents of the specification. In order to make the above and other purposes, features and advantages of the present application more obvious and easy to understand, the specific implementation methods of the present application are listed below. BRIEF DESCRIPTION OF THE DRAWINGS
[0056] In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the drawings required for use in the embodiments or the description of the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments recorded in the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying creative work.
[0057] Figure 1 A schematic diagram of a process for controlling a blank for producing warped sheets;
[0058] Figure 2 A schematic diagram of the process of automatically controlling the adjustment parameters of the feeding model;
[0059] Figure 3 Schematic diagram of the training process for the back-propagation algorithm;
[0060] Figure 4 It is a schematic diagram of the structure of the blank control system for producing warped sheets;
[0061] Figure 5 It is the structural diagram of the hopper control subnetwork;
[0062] Figure 6 Schematic diagram of the structure of the processing control subnetwork. DETAILED DESCRIPTION
[0063] The technical solutions in the embodiments of the present invention will be described clearly and completely below in conjunction with the drawings in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, rather than all the embodiments.
[0064] Unless otherwise defined, all technical and scientific terms used herein have the same meaning as those commonly understood by those skilled in the art of the present invention. The terms used in the specification of the present invention are only for the purpose of describing specific embodiments and are not intended to limit the present invention. The term "and / or" used herein includes any and all combinations of one or more related listed items.
[0065] Embodiment 1
[0066] like Figure 1 As shown, the present application provides a blank control method and system for producing warped sheets, the method comprising:
[0067] S100: Obtaining material information of raw materials and additives, and determining processing ratio information of raw materials and additives according to the processing target of the warped sheet blank;
[0068] S200: monitoring the working state of the hopper and collecting working parameter information of the hopper;
[0069] S300: monitoring the working state of the warped sheet blank processing equipment and collecting control parameter information of the warped sheet blank processing equipment;
[0070] S400: constructing a hopper control subnetwork and a processing control subnetwork for the hopper and the warped sheet blank processing equipment respectively according to the processing ratio information and the working parameter information, wherein the hopper control subnetwork controls the material discharging parameters of the hopper, and the processing control subnetwork controls the processing parameters of the warped sheet blank processing equipment;
[0071] S500: An automatic control feeding model is established based on the hopper control subnetwork and the processing control subnetwork. The automatic control feeding model obtains working parameter information and control parameter information, and controls the coordination parameters between the working parameter information and the control parameter information and the control time node.
[0072] Specifically, before processing, first obtain the material information of all raw materials and additives, including the type, characteristics and necessary technical data of the materials. According to the predetermined processing goals of the warped sheet blanks, such as the required mechanical strength, corrosion resistance and other physical properties, calculate and determine the optimal processing ratio of the raw materials and additives. Monitoring the working status of the hopper can be achieved by installing sensors at the corresponding positions of the hopper, such as monitoring the capacity, weight and tilt angle of the hopper, and collecting the working parameter information of the hopper in real time, such as the unloading speed, the tilt angle of the hopper and the degree of material compression. Similarly, for the warped sheet blank processing equipment, it can also be achieved by installing sensors at the corresponding equipment positions, such as monitoring The speed, temperature, and pressure of the processing equipment are monitored, and the control parameter information of the warped sheet blank processing equipment is collected, such as molding speed, temperature setting value, and pressure setting value. The monitored data is stored and managed by establishing a hopper control subnetwork and a processing control subnetwork, and an automatic control model is established based on the two subnetworks. The three form an Internet of Things-based model from raw material selection to additive ratio to feeding and molding speed, and then combined with relevant algorithms for data processing and learning to complete the analysis and deep learning of the monitoring data, and connect this communication to the control end of the equipment, and select the corresponding time node to adjust the coordination between the relevant data parameters in the two subnetworks.
[0073] The technical solution of the present invention effectively solves the problem of uniformity and continuity of the ratio of raw materials and mixed materials in the production process of copper tube outer warping pieces. The real-time monitoring and precise adjustment of the automatic control feeding system enhance the uniformity of the ratio, ensuring that the raw materials and additives of each batch of outer warping pieces are mixed according to strict ratio information. The finely controlled ratio ensures the uniform mixing of the materials and the consistency from one batch to the next, thereby reducing product defects and variations caused by improper material ratios. The real-time feedback mechanism ensures the precise control of materials and processing parameters during the production process.
[0074] Further, if Figure 5 As shown, construct the hopper control subnetwork, including:
[0075] Obtain the processing ratio information of the warped sheet blank, monitor the hopper, and integrate the monitoring data into the blank parameter subset;
[0076] The elements in the blank parameter subset are divided into blank internal driving parameter information and blank external driving parameter information. The blank internal driving parameter information is the coordination relationship between the relevant parameters in the hopper, and the blank external driving parameter information is the relevant parameters for the automatic control feeding model to be deployed;
[0077] Establish a blank sample set grid, manage the historical blank parameter subsets according to the time series, set identification labels for the outer warp blank processing categories corresponding to the blank parameter subsets, and predict the internal drive parameter information through the historical blank parameter subsets;
[0078] A billet monitoring grid is established to receive data from a subset of billet parameters and select a loss function that adjusts the working parameters of the hopper according to the billet sample set grid.
[0079] As a preferred embodiment of the above embodiment, the monitoring data and the processing ratio information of the outer warped piece are combined to form a blank parameter subset. The blank parameter subset is divided into blank internal driving parameter information and blank external driving parameter information according to the data analysis software. The blank internal driving parameter information includes the material ratio uniformity, temperature and humidity inside the hopper, and the blank external driving parameter information includes external adjustment parameters such as the hopper unloading speed adjusted according to production needs. A blank sample set grid is established. Time series management data is used in the blank sample set grid, including processing ratio information data, recording the ratio of different raw materials and additives, and storing the historical blank parameter subsets corresponding to each raw material and additive ratio. According to the processing requirements, the processing category labels are pre-defined. For example, the copper tubes used in high-pressure pipelines and the copper tubes used in refrigeration systems are divided into two different categories. The grid can be deeply learned and the normalization of working parameters of the same raw material and additive ratio is generated according to the learning results. The data, whose normalized data is the accurate value for predicting the internal driving parameter information, refers to the relative mapping relationship between the control parameters of the feeding equipment, such as the relationship between the hopper capacity and the current weight, the relationship between the inclination angle and the feeding speed, etc. It does not determine a specific value, but the relative relationship between two data. In actual prediction, a unit can be defined. For example, how many additives are needed for 1 kg of raw material A. When the total amount of feeding is determined, how much inclination angle and vibration parameters of the hopper need to be controlled to ensure the feeding speed. The external driving parameter information considers the matching parameters between it and other equipment, and then determines the specific related values, establishes a billet monitoring network, receives and analyzes the data of the billet parameter subset in real time, selects appropriate loss functions for the parameters predicted by the sub-network, such as the mean square error, etc., which are used to measure the gap between the model prediction results and the true values, and adjusts the working parameters of the hopper according to the billet detection grid.
[0080] Further, if Figure 6 As shown, the processing control subnetwork is constructed, including:
[0081] Obtain the processing target of the warped sheet blank, monitor the processing equipment of the warped sheet blank, and integrate the monitoring data into a processing parameter subset;
[0082] The elements in the processing parameter subset are divided into processing internal drive parameter information and processing external drive parameter information. The processing internal drive parameter information is the coordination relationship between relevant parameters in the processing equipment for the warped sheet blanks, and the processing external drive parameter information is the relevant parameters for the automatic control feeding model to be deployed;
[0083] Establish a processing sample set grid, manage the data of historical processing parameter subsets according to the time series, set identification labels for the processing categories of warped sheet blanks corresponding to the processing parameter subsets, and predict the internal drive parameter information through the historical processing parameter subsets;
[0084] A processing monitoring grid is established to receive data in a processing parameter subset and select a loss function. The loss function adjusts the working parameters of the warped sheet blank processing equipment according to the processing sample set grid.
[0085] In this embodiment, the specific implementation method of constructing the processing control subnetwork is consistent with the specific implementation method of constructing the hopper control subnetwork, and will not be repeated here.
[0086] Further, according to establishing a billet monitoring grid, receiving data in a billet parameter subset, and selecting a loss function, the loss function adjusts the working parameters of the hopper according to the billet sample set grid, including:
[0087] Collect the working parameters of the blank in real time, and pre-process the collected data to obtain the blank data parameters;
[0088] The number of nodes of the input layer is constructed according to the blank data parameters, each node represents one of the working parameters, and the weight and bias are set for each node;
[0089] Set up the feature extraction layer and generate an activation function. The activation function is used to extract features from the working parameters of each node in the input layer. The weight is used to connect each feature extraction layer, and the bias is used to adjust the output of the activation function.
[0090] The first layer of the feature extraction layer accepts data from the input layer, and the output of the feature extraction layer of the previous layer is used as the input of the feature extraction layer of the next layer until it reaches the output layer.
[0091] In another embodiment, a hopper control subnetwork is constructed through a neural network. After the blank data parameters are collected, the collected data are preprocessed, including data cleaning, removal of outliers, normalization, etc. According to the dimensions of the blank data parameters, such as weight, tilt angle, temperature, etc., a corresponding number of input layer nodes are constructed. Each node represents a working parameter. Weights and biases are set for each node. The weight is used to adjust the importance of each working parameter. The bias is used to adjust the offset of each node. A feature extraction layer is set in the neural network. A multi-layer approach can be used to extract higher-level features. The number of nodes in each layer can be determined according to actual conditions and needs, and an activation function is generated for the feature extraction layer. Common activation functions include ReLU, Sigmoid, Tanh, etc., which are used to introduce nonlinearity, and the activation function is obtained by weighting and summing each node in the input layer. Add the bias to generate the output of the node. Specifically, the first layer of the feature extraction layer receives data from the input layer. The output value of each node is calculated by the data, weight and bias of the input layer node. The output of the feature extraction layer of each layer is used as the input of the next layer to form a connection between layers. In this way, the connection between layers and the weights and biases between nodes will form a forward propagation process, gradually extracting high-level features. The output of the feature extraction layer finally reaches the output layer after multiple layers of feature extraction. The number of nodes in the output layer is determined according to the task requirements, such as predicting the optimal unloading speed of the hopper. The node generates the final prediction result after the activation function, indicating the optimal unloading speed of the hopper. It should be noted that the weights and biases in the network can be adjusted as the data changes, which can also be applied to another embodiment of establishing a processing monitoring grid, which will not be repeated here.
[0092] Furthermore, the mean square error is selected as the loss function, and the mean square error is expressed as follows:
[0093]
[0094] Where n represents the number of samples; y i Indicates the actual monitoring value; It represents the predicted value of the blank sample set grid or the processed sample set grid; MSE represents the mean square error.
[0095] As a preferred embodiment of the above embodiment, the mean square error as a loss function can effectively measure the deviation between the predicted value and the actual value. By optimizing this loss function, the network can more accurately predict the feeding parameters of the hopper. The MSE optimization process can make the model gradually converge to the optimal solution during the training process, thereby improving the accuracy and generalization ability of the model. Optimizing MSE can reduce the gap between the predicted value and the actual value, thereby reducing the error of the predicted feeding speed or other working parameters.
[0096] Furthermore, if Figure 2 As shown, the automatic control feeding model obtains the working parameter information and the control parameter information, and controls the matching parameters between the working parameter information and the control parameter information and the control time node, including:
[0097] S510: Obtaining the material feeding speed in the working parameter information and the warped sheet forming speed in the control parameter information;
[0098] S520: Setting a time processing window for processing the same batch of warped sheet blanks, setting a time anchor point within the time processing window by the material feeding speed and the warped sheet forming speed, and updating, saving, and forgetting the working parameter information and the control parameter information at the time anchor point based on the long-short memory algorithm;
[0099] S530: obtaining quality information of the processing of the warped sheet blank, and providing feedback on the matching parameters between the working parameter information and the control parameter information according to the quality information of the processing of the warped sheet blank, which is set as a feedback time node within the time processing window;
[0100] Among them, the time anchor point and the feedback time node are the time points at which the automatic control feeding model performs adaptive control on the working parameter information and the control parameter information.
[0101] In this embodiment, a time processing window for processing the warped sheet blanks of the same batch is set, that is, the warped sheet blanks are processed within a certain time range. In the time processing window, multiple time anchor points are set. The setting of the time anchor points can be based on actual conditions and take into account various factors that affect the blanking parameter variables. For example, the cycle blanking time is determined according to the specifications and molding speed of the warped sheet blank production. This time node can be defined as a time anchor point. Since the production cycle of the same batch takes a long time, the long short-term memory algorithm can be used at each time anchor point to update, save and forget the working parameter information and control parameter information. The long short-term memory algorithm can remember long-term dependencies and help process changes and trends in parameter information in time series. For each time anchor point, the long short-term memory algorithm will update its internal memory state according to the current working parameter information and control parameter information to adapt to the needs of the current processing stage. The time anchor point is used to control the long short-term memory algorithm to update and adjust the working parameters and control parameters at different time points to ensure the continuity of the model in the time series. The feedback time node is a time point set according to the actual processing situation, which is used to feedback the processing quality information in real time, and then adjust the matching parameters between the working parameter information and the control parameter information.
[0102] Furthermore, a training set and a verification set are constructed according to the data information stored in the hopper control subnetwork and the processing control subnetwork to train and verify the automatic control feeding model.
[0103] Furthermore, if Figure 3 As shown, the parameters of the automatic control feeding model are trained through the back propagation algorithm, including:
[0104] S531: input the training set into the neural network, and go through the forward propagation process of the neural network. During the forward propagation process, the input of each layer undergoes a linear transformation of weights and biases, and then undergoes a nonlinear transformation of the activation function to obtain the output of each layer;
[0105] S532: Calculate the loss function value of the automatic control feeding model on the current training sample according to the adaptive adjustment amplitude of the automatic control feeding model at the time anchor point and the feedback time node;
[0106] S533: Use the chain rule to propagate the gradient forward layer by layer, back-propagate the gradient of the loss function to the output layer to the hidden layer and input layer, and update the parameters of each layer at the same time;
[0107] S534: According to the calculated gradient, the gradient descent method is used to update the parameters of the model so that the loss function gradually decreases;
[0108] S535: Repeat the above steps and perform multiple iterations on the entire training set until the automatic control feeding model converges or reaches a predetermined number of iterations.
[0109] As a preferred embodiment of the above, the nodes of each layer, i.e., neurons, are connected to all nodes of the previous layer. Each connection has a weight, and each node has a bias. In the forward propagation process, the input of each node is the output of the node of the previous layer multiplied by the corresponding weight, and then the bias term is added. The input of each layer is linearly transformed by the weight and bias, which can be expressed as follows:
[0110]
[0111] Among them, z represents the weighted input of the node in this layer, w i represents the weight of connecting the node, x i represents the output of the previous layer, and b represents the bias of the node;
[0112] At each time anchor point and feedback time node, the automatic control feeding model will calculate the loss function value of the model on the current training sample based on the current working parameter information and control parameter information. The loss function value measures the difference between the model's predicted results and the actual values. It is the goal of model optimization. The chain rule is used to propagate the gradient of the loss function to the output layer layer by layer, and the gradient is back-propagated to the hidden layer and the input layer. This step is to continuously adjust the model parameters through the gradient descent method so that the loss function gradually decreases and the model's predicted results are closer to the actual values. Repeat the above steps and perform multiple iterative training on the entire training set. Each iterative training will adjust the model parameters according to the loss function value of the current sample, thereby improving the model's prediction performance. During the training process, it is necessary to monitor the changes in the loss function value until the loss function value converges or reaches the predetermined number of iterations.
[0113] Embodiment 2
[0114] Based on the same inventive concept as the blank control method for producing warped sheets in the aforementioned embodiment, referring to Figure 4 As shown, the present invention also provides a blank control system for producing outer warped sheets, comprising:
[0115] A material information acquisition module acquires material information of raw materials and additives, and determines processing ratio information of raw materials and additives according to the processing target of the warped sheet blank;
[0116] The material unloading status monitoring module monitors the working status of the hopper and collects the working parameter information of the hopper;
[0117] The blank processing monitoring module monitors the working status of the outer warp blank processing equipment and collects the control parameter information of the outer warp blank processing equipment;
[0118] The subnetwork control module constructs a hopper control subnetwork and a processing control subnetwork for the hopper and the outer warp blank material processing equipment respectively according to the processing ratio information and the working parameter information. The hopper control subnetwork controls the material feeding parameters of the hopper, and the processing control subnetwork controls the processing parameters of the outer warp blank material processing equipment.
[0119] The intelligent control module establishes an automatic control feeding model based on the hopper control subnetwork and the processing control subnetwork. The automatic control feeding model obtains working parameter information and control parameter information, and controls the coordination parameters between the working parameter information and the control parameter information as well as the control time nodes.
[0120] The above-mentioned adjustment system in the present invention can effectively realize the blank control method for producing warped sheets, and the technical effects that can be achieved are as described in the above-mentioned embodiments, which will not be repeated here.
[0121] Specifically, the intelligent control module includes:
[0122] A sub-network parameter acquisition unit is used to acquire a material feeding speed in the working parameter information and a warped sheet forming speed in the control parameter information;
[0123] The time series parameter updating unit sets the time processing window based on the processing of the same batch of warped sheet blanks, sets the time anchor point within the time processing window based on the material feeding speed and the warped sheet forming speed, and updates, saves, and forgets the working parameter information and control parameter information at the time anchor point based on the long-short memory algorithm;
[0124] The blank processing quality feedback unit obtains the quality information of the processing of the warped sheet blank, and feeds back the matching parameters between the working parameter information and the control parameter information according to the quality information of the processing of the warped sheet blank. The feedback time node is set within the time processing window, wherein the time anchor point and the feedback time node are the time points at which the automatic control feeding model performs adaptive control on the working parameter information and the control parameter information.
[0125] Similarly, the above-mentioned optimization schemes for the system can also respectively achieve the optimization effects corresponding to the method in Example 1, which will not be repeated here.
[0126] Although the present application has been described in conjunction with specific features and embodiments thereof, it is obvious that various modifications and combinations may be made thereto without departing from the spirit and scope of the present application. Accordingly, this specification and the accompanying drawings are merely exemplary illustrations of the present application as defined therein, and are deemed to have covered any and all modifications, variations, combinations or equivalents within the scope of the present application. Obviously, a person skilled in the art may make various modifications and variations to the present application without departing from the scope of the present application. Thus, if these modifications and variations of the present application fall within the scope of the present application and its equivalents, the present application is intended to include these modifications and variations.
Claims
1. A blank control method for producing warped sheets, characterized in that: The method comprises: Obtaining material information of raw materials and additives, and determining processing ratio information of the raw materials and additives according to the processing target of the warped sheet blank; Monitor the working status of the hopper and collect working parameter information of the hopper; Monitor the working state of the warped sheet blank processing equipment and collect control parameter information of the warped sheet blank processing equipment; According to the processing ratio information and the working parameter information, a hopper control subnetwork and a processing control subnetwork are respectively constructed corresponding to the hopper and the warped sheet blank processing equipment, wherein the hopper control subnetwork controls the material discharging parameters of the hopper, and the processing control subnetwork controls the processing parameters of the warped sheet blank processing equipment; An automatic control feeding model is established based on the hopper control subnetwork and the processing control subnetwork, wherein the automatic control feeding model obtains the working parameter information and the control parameter information, and controls the matching parameters between the working parameter information and the control parameter information and the control time node; Build the hopper control subnetwork, including: Obtaining the processing ratio information of the warped sheet blank, monitoring the hopper, and integrating the monitoring data into a blank parameter subset; The elements in the blank parameter subset are divided into blank internal driving parameter information and blank external driving parameter information, wherein the blank internal driving parameter information is the matching relationship between the relevant parameters in the hopper, and the blank external driving parameter information is the relevant parameters for the automatic control feeding model to adjust; Establish a blank sample set grid, perform data management on the historical blank parameter subset according to the time series, set an identification label for the processing category of the warped sheet blank corresponding to the blank parameter subset, and predict the internal drive parameter information through the historical blank parameter subset; Establishing a billet monitoring grid, receiving data in the billet parameter subset, and selecting a loss function, the loss function adjusting the working parameters of the hopper according to the billet sample set grid; Construct the processing control subnetwork, including: Acquire the processing target of the warped sheet blank, monitor the warped sheet blank processing equipment, and integrate the monitoring data into a processing parameter subset; The elements in the processing parameter subset are divided into processing internal drive parameter information and processing external drive parameter information, wherein the processing internal drive parameter information is the coordination relationship between relevant parameters in the warped sheet blank processing equipment, and the processing external drive parameter information is the relevant parameters for the automatic control feeding model to allocate; Establishing a processing sample set grid, performing data management on the historical processing parameter subset according to the time series, setting an identification label for the processing category of the warped sheet blank corresponding to the processing parameter subset, and predicting the internal drive parameter information through the historical processing parameter subset; Establishing a processing monitoring grid, receiving data in the processing parameter subset, and selecting a loss function, wherein the loss function adjusts the working parameters of the warped sheet blank processing equipment according to the processing sample set grid; The automatic control feeding model obtains the working parameter information and the control parameter information, and controls the matching parameters between the working parameter information and the control parameter information and the control time node, including: Obtaining the material feeding speed in the working parameter information and the warped sheet forming speed in the control parameter information; A time processing window is set for processing the warped sheet blanks of the same batch, a time anchor point is set within the time processing window by the material feeding speed and the warped sheet forming speed, and the working parameter information and the control parameter information are updated, saved, and forgotten at the time anchor point based on the long-short memory algorithm; Acquire the quality information of the processing of the warped sheet blank, and provide feedback on the matching parameters between the working parameter information and the control parameter information according to the quality information of the processing of the warped sheet blank, which is set as a feedback time node within the time processing window; Among them, the time anchor point and the feedback time node are the time points at which the automatic control feeding model performs adaptive control on the working parameter information and the control parameter information.
2. The blank control method for producing warped sheets according to claim 1, characterized in that: According to the above, a billet monitoring grid is established, data in a billet parameter subset is received, and a loss function is selected. The loss function adjusts the working parameters of the hopper according to the billet sample set grid, including: Collecting the working parameters of the blank in real time, and preprocessing the collected data to obtain blank data parameters; Constructing the number of nodes of the input layer according to the blank data parameters, each node represents one of the working parameters, and setting weights and biases for each of the nodes; Setting a feature extraction layer and generating an activation function, and performing feature extraction on the working parameters of each node in the input layer through the activation function, wherein the weight is used to connect each of the feature extraction layers, and the bias is used to adjust the output of the activation function; The first layer of the feature extraction layer receives data from the input layer, and the output of the feature extraction layer of the previous layer is used as the input of the feature extraction layer of the next layer until the output layer is reached.
3. The blank control method for producing warped sheets according to claim 1, characterized in that: The mean square error is selected as the loss function, and the mean square error is expressed as follows: ; Where n represents the number of samples; Indicates the actual monitoring value; represents the predicted value of the blank sample set grid or the processed sample set grid; MSE represents the mean square error.
4. The blank control method for producing warped sheets according to claim 1, characterized in that: A training set and a verification set are constructed based on the data information stored in the hopper control subnetwork and the processing control subnetwork, and the automatic control feeding model is trained and verified.
5. The blank control method for producing warped sheets according to claim 4, characterized in that: The parameters of the automatic control feeding model are trained by a back propagation algorithm, including: The training set is input into the neural network, and the output of each layer is obtained by the forward propagation process of the neural network, during which the input of each layer undergoes a linear transformation of weights and biases, and then undergoes a nonlinear transformation of an activation function; Calculate the loss function value of the automatic control feeding model on the current training sample according to the adaptive adjustment amplitude of the automatic control feeding model at the time anchor point and the feedback time node; Use the chain rule to propagate the gradient forward layer by layer, backpropagate the gradient of the loss function with respect to the output layer to the hidden layer and the input layer, and update the parameters of each layer at the same time; According to the calculated gradient, the gradient descent method is used to update the parameters of the model so that the loss function gradually decreases; Repeat the above steps and perform multiple iterations on the entire training set until the automatic control feeding model converges or reaches a predetermined number of iterations.
6. A blank control system for producing warped sheets, characterized in that: The blank control method for producing warped sheets according to claim 1 is adopted, wherein the system comprises: A material information acquisition module, which acquires material information of raw materials and additives, and determines processing ratio information of the raw materials and additives according to the processing target of the warped sheet blank; A material discharge status monitoring module monitors the working status of the hopper and collects the working parameter information of the hopper; A blank processing monitoring module monitors the working state of the warped sheet blank processing equipment and collects control parameter information of the warped sheet blank processing equipment; A subnetwork control module, which constructs a hopper control subnetwork and a processing control subnetwork corresponding to the hopper and the warped sheet blank processing equipment respectively according to the processing ratio information and the working parameter information, wherein the hopper control subnetwork controls the material discharge parameters of the hopper, and the processing control subnetwork controls the processing parameters of the warped sheet blank processing equipment; An intelligent control module establishes an automatic control feeding model based on the hopper control subnetwork and the processing control subnetwork, wherein the automatic control feeding model obtains the working parameter information and the control parameter information, and controls the matching parameters between the working parameter information and the control parameter information and the control time node; The intelligent control module comprises: A sub-network parameter acquisition unit, which acquires a material feeding speed in the working parameter information and a warped sheet forming speed in the control parameter information; A time series parameter updating unit is used to set a time processing window for processing the warped sheet blanks of the same batch, set a time anchor point within the time processing window by the material feeding speed and the warped sheet forming speed, and update, save, and forget the working parameter information and the control parameter information at the time anchor point based on a long-short memory algorithm; A blank processing quality feedback unit obtains the quality information of the processing of the warped sheet blank, and provides feedback on the matching parameters between the working parameter information and the control parameter information according to the quality information of the processing of the warped sheet blank, and is set as a feedback time node within the time processing window, wherein the time anchor point and the feedback time node are the time points at which the automatic control feeding model performs adaptive control on the working parameter information and the control parameter information.
