Aluminum profile extrusion process energy consumption control method based on deep learning

Abstract

The invention discloses an aluminum profile extrusion process energy consumption control method based on deep learning, and relates to the technical field of aluminum profile processing, and the method comprises the steps: collecting technological parameters in an aluminum profile extrusion process in real time, inputting the technological parameters into a trained machine learning model, and outputting a prediction result. The method comprises the following steps: determining the opening time of an aluminum profile heating furnace, the opening time of a mold heating furnace, the preheating temperature required by an aluminum profile and the preheating temperature required by a mold, and constructing the opening time, the preheating temperature required by the aluminum profile and the preheating temperature required by the mold in a digital twin model; and furnace doors of the aluminum profile heating furnace and the mold heating furnace are controlled to be opened through the digital twin model. According to the method, the training data are safely stored for a long time through a fountain mechanism, a good data protection effect is achieved, it can be guaranteed that the machine learning model is trained subsequently through the safe and accurate training data, and the trained machine learning model is accurate.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of aluminum profile processing, in particular to an aluminum profile extrusion process energy consumption control method based on deep learning. BACKGROUND

[0002] The current global energy pattern is undergoing profound changes, and the "double carbon" goal has become a consensus among countries. Energy saving and emission reduction has become an inevitable requirement for sustainable development in all walks of life.

[0003] The method and system for controlling the energy consumption of the aluminum profile extrusion process disclosed in the publication No. CN119511830A include obtaining process parameters in the aluminum profile extrusion process from the MES system, inputting the real-time data of the process parameters into the machine learning model, and outputting optimized process parameters. The preheating temperature of the mold and aluminum bar and the opening time of the mold heating furnace door and the aluminum bar heating furnace door are automatically and accurately calculated in the software module according to the process parameters. Through the combination of industrial computers, MES systems and machine learning technology, and the accurate calculation of equipment control parameters, efficient energy consumption control and production optimization are realized.

[0004] The aluminum profile industry is an important part of the manufacturing industry, and its extrusion process is a high-energy consumption link. Aluminum billet heating, extrusion machine operation and other processes require a large amount of electricity and heat energy. At present, the deep learning model can optimize the energy consumption of the aluminum profile extrusion process, but the use of the deep learning model requires a large amount of historical data for training. Since a large amount of historical data plays an important role in the middle, there is a situation that external networks affect the historical data, thereby disturbing the training results of the deep learning model. It is easy to cause inaccurate training of the deep learning model, resulting in errors in the energy consumption control of the aluminum profile extrusion process. In addition, since aluminum profiles need to participate in multiple devices during production, it is difficult to control the parameters of multiple devices (heating furnaces) and the collaborative management of the furnace door opening. SUMMARY

[0005] The purpose of the present application is to provide an aluminum profile extrusion process energy consumption control method based on deep learning to solve the problems in the background art.

[0006] In order to achieve the above-mentioned purpose, the present application provides the following technical scheme: an aluminum profile extrusion process energy consumption control method based on deep learning, comprising the following steps: Collecting historical data in the aluminum profile extrusion process as training data, storing the training data in a control platform, and setting a fountain mechanism in the control platform to store and manage the training data, wherein the fountain mechanism includes a fountain package, a plurality of fountain points and a plurality of fountain elements. Training the machine learning model based on the training data to obtain a trained machine learning model and storing the trained machine learning model in the control platform; Real-time collection of process parameters in the aluminum profile extrusion process and input into the trained machine learning model, and output to obtain a prediction result, wherein the prediction result includes the predicted optimal heating temperature of the aluminum profile and the mold and the extrusion time, and based on the control platform, the temperature set value of the aluminum profile heating furnace and the mold heating furnace is dynamically adjusted according to the prediction result; Based on the time length and energy loss of the aluminum profile and the mold from the heating furnace to the installation on the extruder, the opening time of the aluminum profile heating furnace and the mold heating furnace and the required preheating temperature of the aluminum profile and the required preheating temperature of the mold are determined and constructed in the digital twin model, and when the required preheating temperature of the aluminum profile and the required preheating temperature of the mold are reached, the furnace door of the aluminum profile heating furnace and the mold heating furnace is opened respectively through the digital twin model, and the aluminum profile and the mold are installed on the extruder for extrusion forming.

[0007] In a preferred embodiment, the step of setting a fountain mechanism in the control platform to store and manage the training data comprises: Setting a fountain package in the control platform, setting an access port corresponding to the fountain package and recording an authorized port, and associating a fountain port corresponding to the access port, and setting the fountain package in the access interval; Setting a plurality of positioning points and a plurality of fountain points in the fountain package, wherein the positioning points and the fountain points are staggered between them; Configuring a data package corresponding to the positioning point and setting a fountain element corresponding to the fountain point, wherein the fountain element stores a plurality of pheromones, and adjacent data packages and fountain elements are connected to each other; Setting a plurality of traction lines in the fountain package, wherein the traction line is composed of a plurality of free packages, the free package has a plurality of connection ports, one end of the plurality of traction lines is connected with the center package in the fountain package, and the other end of the plurality of traction lines is connected with the fountain port; The fountain package, a plurality of fountain points and a plurality of fountain elements are used as a fountain mechanism; The training data is divided according to the data volume and stored in different data packages, and the order of the corresponding data packages is recorded according to the division order of the training data.

[0008] In a preferred embodiment, the step of setting a plurality of positioning points and a plurality of fountain points in the fountain package comprises: Determining a plurality of storage points in the fountain package, taking the center of the storage range of the fountain package as the center point, and configuring a center package corresponding to the center point; Staggered marking corresponding to the storage point, respectively obtaining a plurality of staggered positioning points and fountain points.

[0009] In a preferred embodiment, the step of arranging a plurality of traction lines in the fountain package comprises: A plurality of free packages are arranged in the fountain package, and a plurality of connection ports of the free packages are respectively connected with different fountain elements; By sequentially connecting between a plurality of free packages, the free package at one end is connected with the center package, and the free package at the other end is connected with the fountain port associated with the access port of the fountain package, to obtain a traction line. By arranging a plurality of preset numbers of free packages, a plurality of traction lines are obtained. A density rule is set for the plurality of traction lines, wherein the density rule is that the free packages on the plurality of traction lines cannot be connected with the same fountain element.

[0010] In a preferred embodiment, the step of training the machine learning model based on the training data to obtain a trained machine learning model and storing the trained machine learning model in the control platform comprises: The training data in the control platform is obtained through the authorized port, and when there is an unauthorized port accessing the control platform, the unauthorized port is resisted by the fountain mechanism; The training data is used to train the machine learning model to obtain a trained machine learning model and store the trained machine learning model in the control platform.

[0011] In a preferred embodiment, the step of resisting the unauthorized port by the fountain mechanism when the unauthorized port accesses the control platform comprises: When the unauthorized port accesses the access port of the fountain package, the unauthorized network is transferred to the associated fountain port through the access port, and a traction line is enabled; The free packages in the traction line are connected with different fountain elements through a plurality of connection ports, one of the fountain elements is selected as the binding point of the free package, the pheromones in the fountain elements are continuously acquired through the free package, and the pheromones are released through the fountain port, so as to dock the unauthorized port. The released pheromones are connected with each other and connected with the fountain port; When the pheromones in the fountain element connected with the free package in the traction line are insufficient, the free package is immediately disconnected from the fountain element with insufficient pheromones, and other fountain elements are connected to continue to supply pheromones to the fountain port for release. The fountain element with insufficient pheromones is supplemented until the unauthorized port exits the access.

[0012] In a preferred embodiment, the step of dynamically adjusting the temperature setting value of the aluminum profile heating furnace and the mold heating furnace based on the prediction result of the digital twin model comprises: Real-time acquisition of process parameters in the aluminum profile extrusion process and input into the trained machine learning model to obtain the predicted optimal heating temperature and extrusion time of the aluminum profile and the mold as the prediction result. Based on the aluminum profile and the mold, a digital twin model is built, the digital twin model is stored in the control platform, and the digital twin model is connected to the corresponding aluminum profile heating furnace and mold heating furnace according to the corresponding aluminum profile heating furnace and mold heating furnace; The prediction results of the digital twin model are converted into control instructions for the aluminum profile heating furnace and the mold heating furnace, and the control platform controls the temperature setting of the aluminum profile heating furnace and the mold heating furnace according to the control instructions.

[0013] In a preferred embodiment, when the aluminum profile reaches the required preheating temperature and the mold reaches the required preheating temperature, the aluminum profile and the mold are installed on the extruder, and the extrusion forming step is performed by controlling the aluminum profile heating furnace and the mold heating furnace through the digital twin model, including: The digital twin model in the control platform controls the aluminum profile heating furnace and the mold heating furnace respectively according to the opening time of the aluminum profile heating furnace and the mold heating furnace and the required preheating temperature of the aluminum profile and the required preheating temperature of the mold; When the aluminum profile reaches the required preheating temperature and the mold reaches the required preheating temperature, the doors of the aluminum profile heating furnace and the mold heating furnace are controlled to open respectively by the digital twin model, and the timing starts; The aluminum profile and the mold are installed on the extruder until the extrusion forming is completed.

[0014] In the above technical solution, the present application provides the technical effects and advantages: The present application stores the training data for a long time through the fountain mechanism, has good data protection effect, can guarantee that the subsequent machine learning model is trained by safe and accurate training data, so that the trained machine learning model is accurate, and the predicted aluminum profile required preheating temperature and mold required preheating temperature are accurate, and the heating furnace can be accurately and timely controlled; 2. The present application provides the prediction results of the AI model to the digital twin model for load, and then processes the prediction results and real-time aluminum profile extrusion process parameters in the digital twin model, which can prepare for the control of the heating furnace, can control the parameters of the heating furnace and the opening of the door at the correct time and temperature, and can realize the cooperation and accurate control effect between multiple devices in industrial production. BRIEF DESCRIPTION OF DRAWINGS

[0015] In order to more clearly illustrate the technical solutions in the embodiments or prior art, the drawings needed in the embodiments will be briefly introduced below. Obviously, the drawings in the following description are only some embodiments described in the present application, and other drawings can also be obtained by those skilled in the art according to these drawings.

[0016] Figure 1 A flow chart of the method of the present application.

[0017] Figure 2 A system block diagram of the fountain mechanism of the present application. DETAILED DESCRIPTION

[0018] For the purpose of making the objects, technical solutions and advantages of the embodiments of the present application clearer, the technical solutions in the embodiments of the present application will be described clearly and completely below with reference to the drawings in the embodiments of the present application. Obviously, the described embodiments are some but not all of the embodiments of the present application. Based on the embodiments in the present application, all other embodiments obtained by those of ordinary skill in the art without creative work fall within the protection scope of the present application.

[0019] Embodiment 1, please refer to Figure 1 and Figure 2 The aluminum profile extrusion process energy consumption control method based on deep learning described in the embodiment includes the following steps: S1, collect historical data in the aluminum profile extrusion process as training data, store the training data in the control platform, and set a fountain mechanism in the control platform to store and manage the training data, wherein the fountain mechanism includes a fountain package, a plurality of fountain points and a plurality of fountain elements; S2, train the machine learning model based on the training data, obtain the trained machine learning model and store it in the control platform; S3, real-time collect the process parameters in the aluminum profile extrusion process and input them into the trained machine learning model, and output the predicted results, wherein the predicted results include the predicted optimal heating temperature of the aluminum profile and the mold and the extrusion time, and the control platform adjusts the temperature set value of the aluminum profile heating furnace and the mold heating furnace based on the predicted results; S4, based on the length of time and energy loss of the aluminum profile and the mold from the heating furnace to the installation on the extruder, determine the opening time of the aluminum profile heating furnace and the mold heating furnace and the required preheating temperature of the aluminum profile and the required preheating temperature of the mold and build in the digital twin model, when the required preheating temperature of the aluminum profile and the required preheating temperature of the mold are reached, the furnace door of the aluminum profile heating furnace and the mold heating furnace is opened respectively through the digital twin model, the aluminum profile and the mold are installed on the extruder, and the extrusion forming is carried out. The actual required preheating temperature of the aluminum profile and the required preheating temperature of the mold and the opening time can be set according to the length of time and energy loss of the aluminum profile and the mold from the heating furnace to the installation on the extruder, so that the aluminum profile and the mold can exactly meet the extrusion temperature after being installed on the extruder, and there is no need to excessively heat the aluminum profile and the mold, which leads to the aluminum profile and the mold staying outside for too long to dissipate heat, wasting energy and increasing invalid energy consumption, and delaying the extrusion time.

[0020] As described in steps S1-S4 above, the training data is stored for a long time by the fountain mechanism, which has good data protection effect, can ensure that the subsequent machine learning model is trained by safe and accurate training data, so that the trained machine learning model is accurate, and the predicted required preheating temperature of the aluminum profile and the required preheating temperature of the mold are accurate, and the heating furnace can be accurately and timely controlled.

[0021] In one embodiment, the step S1 of setting the fountain mechanism to store and manage the training data in the control platform comprises: S11, setting a fountain bag in the control platform, setting an access port corresponding to the fountain bag and recording an authorized port, associating a fountain port corresponding to the access port, and setting the fountain bag in the access interval; S12, setting a plurality of positioning points and a plurality of fountain points in the fountain bag, wherein the positioning points and the fountain points are staggered; S13, configuring a data packet corresponding to the positioning point and setting a fountain element corresponding to the fountain point, wherein the fountain element stores a plurality of pheromones, and the adjacent data packet and the fountain element are connected to each other; S14, setting a plurality of traction lines in the fountain bag, wherein the traction line is composed of a plurality of free packets, the free packet has a plurality of connection ports, one end of the plurality of traction lines is connected with the center packet in the fountain bag, and the other end of the plurality of traction lines is connected with the fountain port; S15, taking the fountain bag, the plurality of fountain points and the plurality of fountain elements as the fountain mechanism; S16, dividing the training data according to the data volume and storing them in different data packets respectively, and recording the order of the corresponding data packets according to the division order of the training data.

[0022] In one embodiment, the step S12 of setting multiple positioning points and multiple fountain points in the fountain package comprises: S121, determining multiple storage points in the fountain package, taking the center point in the storage range of the fountain package as the center point, and configuring a center package corresponding to the center point; S122, staggered marking corresponding to the storage points, respectively obtaining multiple staggered marked positioning points and fountain points.

[0023] In one embodiment, the step S14 of setting multiple traction lines in the fountain package comprises: S141, setting multiple free packages in the fountain package, multiple connection ports of the free packages being respectively connected with different fountain elements; S142, sequentially connecting through a preset number of free packages, a free package at one end being connected with the center package, and a free package at the other end being connected with a fountain port associated with the access port of the fountain package, to obtain a traction line, multiple traction lines being obtained by setting through multiple preset numbers of free packages; S143, setting a density rule corresponding to the multiple traction lines, wherein the density rule is that the free packages on the multiple traction lines cannot be connected with the same fountain element; As described in steps S11-S16 above, the control platform is a system platform for data storage, analysis, and for instruction issuance. First, before controlling the aluminum profile extrusion process for the purpose of reducing energy consumption and producing with quality and quantity, a large amount of historical data of the aluminum profile extrusion process is needed as training data. The historical process parameters and production data are used as training data for subsequent training of the machine learning model. In the control platform, a fountain package is set, and an access port corresponding to the fountain package is set. The access port is associated with a fountain port. The access port is an authorized port for accessing training data. The training data can be extracted and used as subsequent training of the machine learning model. The training data can be stored for a long time. When the trained machine learning model fails, it can be retrained. First, a fountain package is set in the control platform. The fountain package is in the access interval. The access interval is the storage space in the control platform. The fountain package is the storage space in the access interval. In order to safely store the training data, a plurality of storage points are determined in the fountain package. The storage points are storage addresses. A central package is then configured on the storage point at the center of the fountain package. The central package is a virtual machine. The storage points in the corresponding area are mixed and staggered. A plurality of marked positioning points and fountain points are obtained. A data package is then configured corresponding to the positioning point. A fountain element is set corresponding to the fountain point. The data package, the fountain element, and the pheromone are virtual machines. The fountain element stores a plurality of pheromones. Adjacent data packages and fountain elements are connected to each other (they are connected to each other to provide a network architecture for moving free packages. The free packages can be transferred between the fountain elements as needed. The data package serves as a bridge between the free packages. The free packages cannot obtain the training data in the data package. The data package is not open to the free packages. The data package is open to the authorized port access). A close connection network is formed in the fountain package. A plurality of traction lines are then set in the fountain package. The traction lines are constructed by a plurality of free packages set in the fountain package. Specifically, for example, there are 10 free packages in the fountain package. Every 5 free packages are combined and sequentially connected. One end of the 5 free packages connected in a chain is connected to the central package. The other end of the 5 free packages connected in a chain is connected to the fountain port associated with the access port. The traction line is obtained. Since there are two traction lines, in order to comply with the density rule, the free packages on the two traction lines will be separated. The free packages on the two traction lines will not be connected to one fountain element. In this way, the pheromones in one fountain element can be used. The free packages can be constantly transferred to obtain more pheromones to resist unauthorized port access. This is an example of two. The actual number of traction lines is multiple. When an unauthorized port accesses the access port of the fountain package, the access port will allocate the access of the unauthorized port to the associated fountain port to resist unauthorized networks and ensure the security of data storage.

[0024] In one embodiment, the step S2 of training the machine learning model based on the training data, obtaining the trained machine learning model and storing the trained machine learning model in the control platform comprises: S21, obtaining the training data in the control platform through an authorized port, and resisting the unauthorized port through a fountain mechanism when the unauthorized port accesses the control platform; S22, training the machine learning model with the training data, obtaining the trained machine learning model and storing the trained machine learning model in the control platform.

[0025] In one embodiment, the step S21 of resisting the unauthorized port through the fountain mechanism when the unauthorized port accesses the control platform comprises: S211, when the unauthorized port accesses the access port of the fountain packet, transferring the unauthorized network to the associated fountain port through the access port, and enabling a towline; S212, connecting the free packet in the towline to different fountain elements through multiple connection ports respectively, selecting one of the fountain elements as the binding point of the free packet, continuously acquiring the pheromone in the fountain element through the free packet, and releasing the pheromone through the towline through the fountain port (which will be released between the fountain packet and the access interval, so that the unauthorized port accesses), docking the unauthorized port, and connecting the released pheromones to each other and to the fountain port; S213, when the pheromone in the fountain element connected by the free packet in the towline is insufficient, immediately disconnecting the free packet from the fountain element with insufficient pheromone, connecting other fountain elements to continue acquiring pheromone to supply to the fountain port for release, and supplementing the fountain element with insufficient pheromone until the unauthorized port exits the access.

[0026] As described in steps S21 and S22, the machine learning model includes but is not limited to supervised learning algorithm, unsupervised learning algorithm, reinforcement learning algorithm. The machine learning model is trained using historical data in the control platform to obtain the influence of different process parameters on energy consumption and production efficiency; using supervised learning algorithms such as random forest or linear regression, the data is trained in the industrial computer; learn the relationship between different process parameters and energy consumption, predict the influence of different process parameters on energy consumption, and find the optimal combination of production parameters. Among them, the random forest is suitable for finding the optimal combination from multiple process parameters to minimize energy consumption. The linear regression model is used to predict the relationship between a specific extrusion process parameter and energy consumption, and the machine learning model is trained by training data, which is obtained from the authorized port in the control platform. The training data can be increased in the future and can be saved for a long time. When the authorized port accesses the access port in the control platform, it can directly enter the access, and multiple data packets can provide training data to the authorized port access and acquisition in sequence.When there is an unauthorized port access control platform, the access port will transfer the unauthorized port to the associated fountain port. After enabling the fountain port, a towline will be enabled. The free packets in the towline are connected to different fountain elements through multiple connection ports. One connection port is connected to one fountain element. Due to the multiple connected fountain elements, a fountain element can be randomly selected as the network connection point of the free packet. The pheromone in the fountain element is continuously acquired through the free packet. Multiple pheromones exist in each fountain element and can be supplemented later. The specific supplement method is: the released pheromone is stored for a preset time without connecting the unauthorized network between the fountain packet and the access area, and then the pheromone is output to the fountain packet and supplemented into the fountain element. The pheromone is released through the fountain port (will be released between the fountain packet and the access area, so that the unauthorized port can access). The unauthorized port is connected to the towline, which is the connection between multiple free packets. The information can be transmitted to the fountain port for release through the connection between them. The released pheromones are connected to each other and connected to the fountain port. For example, when the unauthorized port is transferred to the fountain port, the unauthorized port is connected to the pheromone through the pheromone. The unauthorized port can access the pheromone through the fountain port. Because there are new pheromones constantly, even if the unauthorized port breaks through one pheromone, the next pheromone will be connected to the unauthorized port. The pheromone is the access object of the unauthorized port. As long as the unauthorized port does not exit the access, the pheromone will be continuously released. When multiple unauthorized ports access at the same time, they are all transferred to the fountain port. Different towlines are enabled for unauthorized ports, and all of them are accessed through the fountain port. The training data can be safely stored for a long time through the fountain mechanism, which has good data protection effect. It can ensure that the machine learning model is trained by safe and accurate training data, so that the trained machine learning model is accurate. The predicted preheating temperature of the aluminum profile and the preheating temperature of the mold are accurate, and the heating furnace can be accurately and timely controlled.

[0027] In one embodiment, the step S3 of dynamically adjusting the temperature set value of the aluminum profile heating furnace and the mold heating furnace according to the prediction result based on the digital twin model comprises: S31, real-time collection of process parameters in the aluminum profile extrusion process and input into the trained machine learning model to obtain the predicted optimal heating temperature of the aluminum profile and the mold and the extrusion time as the prediction result; S32, constructing a digital twin model based on the aluminum profile and the mold, storing the digital twin model in the control platform, and connecting the digital twin model and the corresponding aluminum profile heating furnace and mold heating furnace according to the corresponding connection; S33, convert the prediction results of the digital twin model into control instructions for the aluminum profile heating furnace and the mold heating furnace respectively, and control the temperature setting of the aluminum profile heating furnace and the mold heating furnace according to the control instructions through the control platform.

[0028] In one embodiment, when the required preheating temperature of the aluminum profile and the required preheating temperature of the mold are reached, the step S4 of installing the aluminum profile and the mold on the extruder and performing extrusion molding by opening the furnace door of the aluminum profile heating furnace and the mold heating furnace controlled by the digital twin model, comprises: S41, the digital twin model in the control platform controls the aluminum profile heating furnace and the mold heating furnace respectively according to the opening time of the aluminum profile heating furnace and the mold heating furnace and the required preheating temperature of the aluminum profile and the required preheating temperature of the mold (the digital twin model is constructed according to the shape of the aluminum profile and the mold, which can be realized by CAD drawing tool, to obtain a three-dimensional model, and corresponding environmental parameter model is set for the three-dimensional model (the environmental parameter model is a parameter marker point representing the environment outside the three-dimensional model, and the parameter of the environment is temperature), and in the subsequent process, the real-time collected process parameters of the aluminum profile and the mold during production can be constructed in the environmental parameter marker of the corresponding three-dimensional model to obtain the digital twin model, which can simulate the state of the aluminum profile and the mold during production according to the prediction result marked in the environmental parameter model by the AI model, and can simulate the predicted trend of the aluminum profile and the mold, so that the heating furnace can be directly controlled according to the digital twin model in the subsequent process, realizing the cooperation between multiple devices in the aluminum profile production process, making the control process more smooth and stable, avoiding the dissipation of heat caused by excessive external residence of the aluminum profile after heating, and better assisting the control of energy consumption); S42, when the required preheating temperature of the aluminum profile and the required preheating temperature of the mold are reached, the furnace door of the aluminum profile heating furnace and the mold heating furnace is controlled to open respectively by the data twin model, and the timing starts; S43, install the aluminum profile and the mold on the extruder, and adjust the extrusion speed according to the real-time data of the aluminum profile (adjust the extrusion speed according to the real-time temperature and pressure information to ensure product quality, and the aluminum profile here is aluminum bar) during extrusion until the extrusion is completed.

[0029] The process parameters in the aluminum profile extrusion process are obtained from the control platform, including the mold temperature, the aluminum bar extrusion temperature, the aluminum bar length, the extrusion speed, the extrusion rod idle stroke, the extrusion rod idle forward speed, and the extrusion rod idle backward speed, as described in steps S31-S33 and S41-43 above. The real-time data of the process parameters are input into the machine learning model, and the optimized process parameters are output. The preheated aluminum bar and the preheated mold are tested for temperature drop speed at room temperature. The time for the mold to be taken out of the mold heating furnace and installed on the extruder is tested. The aluminum bar cutting time and the aluminum bar installation time on the extruder are tested. The one-time aluminum bar extrusion cycle working time is calculated, which is the sum of the single aluminum bar extrusion time, the extrusion rod idle backward time, and the extrusion rod idle forward time. The time for the aluminum bar to stay outside the equipment is calculated, which is the sum of the aluminum bar pushing-out time from the heating furnace, the aluminum bar cutting time, and the aluminum bar installation time on the extruder. The aluminum bar pushing-out time from the heating furnace, the distance from the aluminum bar heating furnace door to the cutting knife, and the aluminum bar pushing-out speed from the heating furnace are calculated. The time corresponding to the extrusion cycle when the aluminum bar heating furnace door is opened and the time corresponding to the extrusion cycle when the mold heating furnace door is opened are calculated. According to the preset production requirements and the extrusion process parameters, the required preheating temperature of the aluminum bar and the required preheating temperature of the mold are calculated. The aluminum bar heating furnace and the mold heating furnace are controlled according to the obtained parameters, so that the aluminum bar and the mold are preheated to the required preheating temperature of the aluminum bar and the required preheating temperature of the mold, and the aluminum bar heating furnace door and the mold heating furnace door are opened according to the calculated time points.

[0030] The above merely describes specific embodiments of the present application, but the protection scope of the present application is not limited thereto. Any person skilled in the art can easily think of changes or replacements within the technical scope disclosed in the present application, which should be covered within the protection scope of the present application. Therefore, the protection scope of the present application should be subject to the protection scope of the claims.

Claims

1. A method for energy consumption control in aluminum profile extrusion process based on deep learning, characterized in that, Includes the following steps: Historical data from the aluminum profile extrusion process is collected as training data and stored in the control platform. A fountain mechanism is set up in the control platform to store and manage the training data. The fountain mechanism includes a fountain package, multiple fountain points, and multiple fountain elements. The machine learning model is trained based on the training data, and the trained machine learning model is stored in the control platform. The process parameters during the aluminum profile extrusion process are collected in real time and input into a trained machine learning model. The output is a prediction result, which includes the predicted optimal heating temperature and extrusion time for the aluminum profile and the die. The process parameters and prediction results during the aluminum profile extrusion process are constructed in a digital twin model. Based on the digital twin model, the temperature setpoints of the aluminum profile heating furnace and the die heating furnace are dynamically adjusted according to the prediction results. Based on the time and energy loss during the process from when the aluminum profile and the die come out of the heating furnace to when they are installed on the extruder, the opening time of the aluminum profile heating furnace and the die heating furnace, as well as the required preheating temperature of the aluminum profile and the die, are determined and constructed in a digital twin model. When the required preheating temperature of the aluminum profile and the die are reached, the furnace doors of the aluminum profile heating furnace and the die heating furnace are opened by controlling them respectively through the digital twin model, and the aluminum profile and the die are installed on the extruder for extrusion molding.

2. The energy consumption control method for aluminum profile extrusion process based on deep learning according to claim 1, characterized in that, The step of setting up a fountain mechanism in the control platform to store and manage training data includes: Set up a fountain package in the control platform, set up an access port for the corresponding fountain package and record the authorized port, associate the corresponding access port with the fountain port, and set the fountain package within the access range; Multiple positioning points and multiple fountain points are set in the fountain package, with the positioning points and fountain points being set alternately; Data packets are configured for corresponding positioning points, and fountain elements are set for corresponding fountain points. Each fountain element stores multiple pheromones, and adjacent data packets and fountain elements are interconnected. Multiple traction lines are set in the fountain package. Each traction line consists of multiple free packages with multiple connection ports. One end of each traction line is connected to the central package in the fountain package, and the other end of each traction line is connected to the fountain port. The fountain mechanism consists of fountain packages, multiple fountain points, and multiple fountain elements. The training data is divided into portions according to its size and stored in different data packets. The order of the corresponding data packets is recorded according to the order in which the training data is divided.

3. The energy consumption control method for aluminum profile extrusion process based on deep learning according to claim 2, characterized in that, The step of setting multiple positioning points and multiple fountain points in the fountain package includes: Multiple storage points are determined in the fountain package, and the center point of the fountain package storage range is taken as the center point. A center package is configured for the corresponding center point. The corresponding storage points are marked in an alternating pattern to obtain multiple locations and fountain points with alternating markings.

4. The energy consumption control method for aluminum profile extrusion process based on deep learning according to claim 3, characterized in that, The step of setting multiple traction lines in the fountain package includes: Multiple free-floating packets are set in the fountain package, and the multiple connection ports of the free-floating packets are connected to different fountain elements respectively; By sequentially connecting a preset number of free packets, one free packet at one end is connected to the central packet, and the free packet at the other end is connected to the fountain port associated with the fountain packet access port, a traction line is obtained. Multiple traction lines are obtained by setting multiple preset numbers of free packets. Density rules are set for multiple traction lines, where the density rule is that free packets on multiple traction lines cannot be connected to the same fountain element.

5. The energy consumption control method for aluminum profile extrusion process based on deep learning according to claim 1, characterized in that, The step of training the machine learning model based on training data, obtaining the trained machine learning model, and storing it in the control platform includes: Training data is obtained from the control platform through authorized ports. When unauthorized ports access the control platform, a fountain mechanism is used to resist unauthorized ports. The training data is used to train the machine learning model, and the trained machine learning model is then stored in the control platform.

6. The energy consumption control method for aluminum profile extrusion process based on deep learning according to claim 5, characterized in that, The steps for resisting unauthorized ports through a fountain mechanism when an unauthorized port access control platform exists include: When an unauthorized port accesses the fountain packet's access port, the unauthorized network is redirected to the associated fountain port via the access port, and a tractor is enabled. The free packet in the traction line is connected to different fountain elements through multiple connection ports. One of the fountain elements is selected as the binding point of the free packet. The pheromone in the fountain element is continuously acquired through the free packet, and the pheromone is released through the fountain port via the traction line. The unauthorized port is docked, and the released pheromone is interconnected with each other and all connected to the fountain port. When the pheromone in the fountain element connected to the free packet in the traction line is insufficient, the free packet is immediately disconnected from the fountain element with insufficient pheromone, and other fountain elements are connected to continue to obtain pheromone and supply it to the fountain port for release. The fountain element with insufficient pheromone is replenished until the unauthorized port exits access.

7. The energy consumption control method for aluminum profile extrusion process based on deep learning according to claim 1, characterized in that, The step of dynamically adjusting the temperature setpoints of the aluminum profile heating furnace and the mold heating furnace based on the prediction results using a digital twin model includes: The process parameters of aluminum profile extrusion are collected in real time and input into a trained machine learning model to obtain the predicted optimal heating temperature and extrusion time of aluminum profile and die, which are used as the prediction results. A digital twin model is constructed based on aluminum profiles and molds, and the digital twin model is stored in the control platform. The corresponding aluminum profile heating furnace and mold heating furnace are connected according to the digital twin model. The prediction results of the digital twin model are converted into control commands for the aluminum profile heating furnace and the mold heating furnace, respectively. The control platform controls the temperature settings of the aluminum profile heating furnace and the mold heating furnace according to the control commands.

8. The energy consumption control method for aluminum profile extrusion process based on deep learning according to claim 1, characterized in that, The step of opening the furnace doors of the aluminum profile heating furnace and the mold heating furnace respectively through a digital twin model when the preheating temperatures required for the aluminum profile and the mold are reached, and installing the aluminum profile and the mold onto the extrusion press for extrusion molding includes: The digital twin model in the control platform controls the aluminum profile heating furnace and the mold heating furnace separately by controlling the opening time of the aluminum profile heating furnace and the preheating temperature required for the aluminum profile and the mold respectively. When the required preheating temperatures for the aluminum profile and the mold are reached, the furnace doors of the aluminum profile heating furnace and the mold heating furnace are opened via data twin models, and timing begins. The aluminum profile and mold are mounted onto the extrusion press until they are extruded into shape.

Images