Extrusion and injection integrated process regulation and product defect avoidance system based on digital twinning

Abstract

The application discloses a kind of based on digital twinning's squeeze injection integrated process regulation and product defect avoidance system, comprising: squeeze injection integrated equipment, digital twinning, neural network model;Squeeze injection integrated equipment integrates the function of extrusion and injection molding, digital twinning includes: the three-dimensional model of squeeze injection integrated equipment, the force and torque of the electrical system and hydraulic system Abstract, the database that data collected by sensor is stored, processor, man-machine interface;Processor is used to filter out the process parameters that the squeeze injection integrated equipment needs to regulate when product is predicted to have defect;Neural network model includes input layer and output layer, input layer inputs sensor data received by digital twinning, and output layer outputs the judgment result whether product has defect.The application improves the calculation rate when equipment is running, and operation result is closer to the process parameter generated when squeeze injection integrated equipment is running;It can regulate process parameters in advance when the product of squeeze injection integrated equipment has defect.

Description

Technical Field

[0001] This invention relates to the field of extrusion injection molding, and more particularly to a digital twin-based integrated extrusion and injection process control and product defect avoidance system. Background Technology

[0002] With the development of injection molding technology, the demand for large-volume injection molding using small and medium-sized equipment has gradually emerged, leading to the development of integrated extrusion and injection molding equipment. This equipment integrates extrusion and injection molding, extrudes the molten plastic, and stores it in a storage tank, thus enabling large-volume injection using small and medium-sized extrusion equipment. Some domestic companies have improved their extrusion and injection mechanisms to meet the needs of different scenarios during production. However, the process control methods for integrated extrusion and injection molding equipment and the defect detection of the final product still rely on relatively traditional approaches. While traditional process control methods for integrated extrusion and injection molding equipment can produce plastic products with complex structures, they often result in numerous product defects, such as material shortages, shrinkage, surface bright marks, air bubbles, water inclusions, burrs, deformation, unclear surfaces, whitening, and porosity. These defects are mainly caused by fluctuations in process parameters during equipment operation, reducing raw material utilization and production efficiency. Summary of the Invention

[0003] To address the shortcomings of existing technologies, this invention proposes a digital twin-based integrated extrusion and injection process control and product defect avoidance system. This system enables bidirectional real-time communication between the integrated extrusion and injection equipment and its digital twin. This bidirectional information flow is used to monitor the process parameters of the physical integrated extrusion and injection equipment in real time. A trained neural network model is used to predict product defects, and the digital twin controls the physical integrated extrusion and injection equipment to reasonably avoid defects.

[0004] The specific technical solution is as follows:

[0005] A digital twin-based system for integrated extrusion and injection process control and product defect avoidance includes: integrated extrusion and injection equipment, digital twin, and neural network model;

[0006] The integrated extrusion and injection equipment integrates extrusion and injection molding functions, including: an extrusion assembly, a storage tank, an injection molding assembly, an electrical system, a hydraulic system, and a mold template; the extrusion assembly is connected to the storage tank, the storage tank is connected to the injection molding assembly, the electrical system provides a power source for the extrusion assembly, the hydraulic system provides power for the injection molding assembly, and the injection molding assembly is connected to the mold template; multiple sensors are arranged inside the extrusion channel of the extrusion assembly and the injection channel of the injection molding assembly, as well as at the power output points of the electrical system and the hydraulic system, and all sensors are connected to a digital twin;

[0007] The digital twin includes: a three-dimensional model of the integrated extrusion and injection equipment, forces and torques abstracted from the electrical and hydraulic systems, a database storing data collected by sensors, a processor, and a human-machine interface; the processor is used to filter out the process parameters that the integrated extrusion and injection equipment needs to be adjusted when product defects are predicted.

[0008] The neural network model includes an input layer and an output layer. The input layer receives sensor data received by the digital twin, and the output layer outputs the judgment result of whether the product is defective. After training, the accuracy of the neural network model in judging whether the manufactured product is defective reaches a threshold.

[0009] Furthermore, the integrated extrusion and injection equipment also includes: a feed funnel, a connecting body, and a switching valve plate; the extrusion assembly includes: an extrusion channel, an extrusion screw, a feed inlet, and a discharge outlet;

[0010] The feeding funnel is connected to the feeding port, and the raw material is fed into the extrusion assembly through the feeding funnel and subjected to melting and plasticizing treatment; the end of the extrusion channel is the discharge port, and the extrusion screw is coaxially arranged in the extrusion channel. The motor output port of the electrical system is connected to the extrusion screw, and the extrusion screw transports the molten raw material to the discharge port; the discharge port is connected to the storage tank through a connector; the storage tank is connected to the injection molding assembly through a conversion valve plate; the hydraulic system provides power to the injection molding assembly, injecting the molten raw material stored in the storage tank into the mold clamped in the mold template through the injection molding channel.

[0011] Furthermore, multiple temperature sensors, pressure sensors, viscosity sensors, flow rate sensors, and flow sensors are arranged in the extrusion channel of the extrusion assembly, the storage tank, the injection channel of the injection molding assembly, and the interior of the mold clamped in the mold template. Force and torque sensors are arranged at the power output of the electrical system and the hydraulic system.

[0012] A method for controlling the integrated extrusion and injection process and avoiding product defects based on digital twins, implemented using the aforementioned system for controlling the integrated extrusion and injection process and avoiding product defects, specifically includes the following steps:

[0013] S1: Based on the mechanical structure of the integrated extrusion and injection equipment, establish a 3D model of the integrated extrusion and injection equipment;

[0014] S2: Input the 3D model of the integrated extrusion and injection equipment into Unreal Engine, and calibrate and verify the 3D model of the integrated extrusion and injection equipment to obtain a digital twin that matches the operating data of the integrated extrusion and injection equipment;

[0015] S3: Collect process parameter data of the integrated extrusion equipment during operation by sensors placed at each key node of the flow channel and at the effect points of the electrical and hydraulic systems of the integrated extrusion equipment, and input these data into the digital twin to obtain the product dataset;

[0016] S4: Using the product dataset as input, build and train a neural network model;

[0017] S5: When the extrusion and injection equipment is running, the real-time data collected by the sensors is input into the trained neural network model to predict product defects in real time; if there are no defects, the real-time data is re-inputted and the next judgment is started; if there are defects, the extrusion and injection equipment is controlled by a digital twin to avoid defects.

[0018] Furthermore, in S1, when establishing the 3D model of the integrated extrusion and injection equipment, the modeling of the electrical and hydraulic systems of the integrated extrusion and injection equipment is omitted, and the force and torque provided by them during operation are used to represent the system itself in the model; at the same time, a separate 3D model of the flow channel structure in the integrated extrusion and injection equipment is established for real-time simulation and animation demonstration of the human-machine interface.

[0019] Furthermore, in step S2, the specific method for calibrating and verifying the 3D model of the integrated extrusion equipment is as follows:

[0020] Under the same preset operating conditions and speeds, obtain the operating data generated by each mechanical component of the physical extrusion equipment and its 3D model. Sequentially obtain the difference between the operating data of each mechanical component in the physical extrusion equipment and the operating data in the 3D model of the extrusion equipment, and determine whether the difference is within the standard deviation range. If not, re-obtain the frictional force generated by each mechanical component at a preset speed during the operation of the physical extrusion equipment under the preset operating conditions, and re-set the frictional force generated by each mechanical component at the preset speed into the corresponding mechanical component in the 3D model of the extrusion equipment, until the difference in the operating data of each mechanical component is within the corresponding standard deviation range.

[0021] Furthermore, the specific operation of S4 is as follows:

[0022] The product dataset is divided into a training set and a test set. The training set contains significantly more data than the test set, and both sets include data on defective and non-defective final products. The training set is then input into the neural network model for training. The test set is then input into the trained neural network model to verify whether the accuracy of the trained neural network model in predicting product defects reaches the threshold. If not, step S3 is repeated, increasing the amount of data in the training set and re-inputting the training set into the neural network model for training until the accuracy of the neural network model's prediction reaches the threshold. If it does, the model is considered the successfully trained neural network model.

[0023] Furthermore, in S5, when the trained neural network model predicts that the product may have defects, the method of avoiding defects by controlling the integrated extrusion and injection equipment through a digital twin is as follows:

[0024] The processor in the digital twin first uses a random module to randomly change one or more process parameter data, and then inputs the changed process parameter data into the trained neural network model for prediction. If the prediction result is that no defects will be produced, the modified parameter is returned; if the prediction result is still that defects will be produced, the data is randomly changed again and input into the trained neural network model again until the prediction result is that no defects will be produced, and the parameter is returned.

[0025] The digital twin communicates with the control system of the physical extrusion and injection equipment through a specific communication protocol, adjusting the process parameters during the operation of the extrusion and injection equipment to achieve early defect avoidance.

[0026] Furthermore, the specific communication protocol used is the OPC protocol.

[0027] The beneficial effects of this invention are:

[0028] (1) When establishing the three-dimensional model of the integrated extrusion and injection equipment, this invention only establishes the three-dimensional model of each mechanical structure in the equipment, omitting the modeling of the electrical system and hydraulic system of the equipment, and replacing them with the force and torque generated by them, which simplifies the three-dimensional model of the integrated extrusion and injection equipment, improves the calculation speed during equipment operation, and calibrates and verifies the established model so that its calculation results are closer to the process parameters generated during the operation of the actual integrated extrusion and injection equipment.

[0029] (2) The present invention avoids defects based on a neural network model. When the products produced by the integrated extrusion and injection equipment may be defective, the process parameters are adjusted in advance, which reduces the generation of defective products when the integrated extrusion and injection equipment is working, and improves the utilization rate of raw materials and the production efficiency of products.

[0030] (3) The present invention realizes real-time monitoring of the process flow during the operation of the integrated extrusion and injection equipment through various sensors, which can better understand the material conversion process during the operation of the integrated extrusion and injection equipment, and help to improve and enhance the various process flows and operations during the operation of the integrated extrusion and injection equipment. Attached Figure Description

[0031] Figure 1 This is a schematic diagram of the structure of the extrusion-injection integrated process control and product defect avoidance system based on digital twins of the present invention.

[0032] Figure 2 This is a schematic diagram of the integrated extrusion and injection equipment used in the embodiments of the present invention.

[0033] Figure 3 This is a flowchart of the method for integrated extrusion and injection process control and product defect avoidance based on digital twins, as proposed in this invention.

[0034] Figure 4 This is a flowchart of the calibration and testing process for the digital twin of the integrated extrusion and injection equipment of this invention.

[0035] Figure 5 This is a flowchart of the defect prediction and avoidance operation performed by the machine learning neural network model of this invention.

[0036] In the diagram, 1 is the electrical system, 2 is the feed hopper, 3 is the extrusion assembly, 4 is the connector, 5 is the hydraulic system, 6 is the storage tank, 7 is the conversion valve plate, 8 is the injection molding assembly, 9 is the mold template, and 10 is the base. Detailed Implementation

[0037] The present invention will be described in detail below with reference to the accompanying drawings and preferred embodiments. The objectives and effects of the present invention will become clearer as a result. The present invention will be further described in detail below with reference to the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are merely illustrative of the invention and are not intended to limit the invention.

[0038] Before providing a further detailed description of the embodiments of the present invention, the nouns and terms involved in the embodiments of the present invention will be explained, and the nouns and terms involved in the embodiments of the present invention shall be subject to the following interpretations.

[0039] Digital twin technology: Digital twin technology is based on a digital mirror of a physical object, thereby further describing the changes of the physical object in the real environment, and simulating the behavior and impact of the physical object in the real environment, thus realizing functions such as status monitoring, fault diagnosis, trend prediction, and comprehensive optimization. The main difference between digital twins and simulation lies in whether the information transfer between the physical entity and the digital twin is reciprocal. Simulation typically does not benefit from obtaining real-time data; that is, in simulation technology, the information transfer between the physical entity and the digital twin is not reciprocal. Digital twins are designed around bidirectional information flow; that is, in digital twin technology, the information transfer between the physical entity and the digital twin is reciprocal. This information flow first appears when the object's sensors provide relevant data to the system processor; then, it reappears when the system processor shares its analysis results with the original source object.

[0040] An integrated extrusion and injection molding machine combines extrusion and injection molding equipment. Its working principle differs from injection molding mainly in that it extrudes the molten and plasticized plastic and stores it in a storage tank. Its advantage is that it can achieve large-volume injection using small and medium-sized extrusion and injection equipment.

[0041] like Figure 1 As shown, a digital twin-based extrusion and injection integrated process control and product defect avoidance system includes: an integrated extrusion and injection device, a digital twin, and a neural network model.

[0042] like Figure 2 As shown, the main components of the integrated extrusion and injection molding equipment include: electrical system 1, feeding hopper 2, extrusion assembly 3, connecting body 4, hydraulic system 5, storage tank 6, switching valve plate 7, injection molding assembly 8, mold template 9, and base 10. Electrical system 1, feeding hopper 2, extrusion assembly 3, connecting body 4, hydraulic system 5, storage tank 6, switching valve plate 7, injection molding assembly 8, and mold template 9 are all fixedly connected to the base 10. The extrusion assembly 3 includes: extrusion channel, extrusion screw, feed port, and discharge port.

[0043] The feeding hopper 2 is connected to the feeding port of the extrusion assembly 3. Raw material is fed into the extrusion assembly 3 through the feeding hopper 2 and undergoes melting and plasticizing. The extrusion screw is coaxially arranged within the extrusion channel. The motor output port of the electrical system 1 is connected to the extrusion screw inside the extrusion assembly 3, providing a power source for the extrusion assembly 3. The end of the extrusion channel is the discharge port. The extrusion screw rotates under the action of the electrical system 1, thereby extruding the molten and plasticized plastic from the extrusion channel to the discharge port. The discharge port of the extrusion assembly 3 is connected to the storage tank 6 via the connector 4, storing the molten and plasticized plastic in the storage tank 6. The storage tank 6 is connected to the injection molding assembly 8 via the conversion valve plate 7. The hydraulic system 5 provides power to the injection molding assembly 8, injecting the molten raw material stored in the storage tank 6 into the mold clamped in the mold template 9 in the injection molding printing area, thus forming the final product.

[0044] Temperature sensors, pressure sensors, viscosity sensors, flow rate sensors, and flow sensors are installed at key nodes in the feed funnel 2, extrusion channel of extrusion assembly 3, storage tank 6, injection channel of injection assembly 8, and inside the mold of the integrated extrusion and injection equipment. Force and torque sensors are installed at the power output of electrical system 1 and hydraulic system 5. All sensors are connected to a digital twin.

[0045] The digital twin includes: a 3D model of the integrated extrusion equipment; forces and torques abstracted from the electrical system 1 and hydraulic system 5 of the integrated extrusion equipment; a database storing data collected by sensors; a processor; and a human-machine interface. The processor is mainly used to filter out the process parameters that need to be adjusted when product defects are predicted.

[0046] The neural network model includes an input layer and an output layer. The input layer receives sensor data received by the digital twin, and the output layer outputs the judgment result of whether the product is defective. The neural network model can be trained to obtain a neural network model with high accuracy in judging whether the manufactured product is defective. In this embodiment, the neural network model is selected as a BP neural network.

[0047] like Figure 3 As shown, based on the above-mentioned digital twin-based integrated extrusion process control and product defect avoidance system, a digital twin-based integrated extrusion process control and product defect avoidance method is proposed, which specifically includes the following steps:

[0048] S1: Based on the components contained in the mechanical structure of the integrated extrusion equipment and the structural relationships between the components and between the mechanical structures, establish a 3D model of the integrated extrusion equipment.

[0049] In building the 3D model of the integrated extrusion and injection equipment, the modeling of the electrical system 1 and hydraulic system 5 of the integrated extrusion and injection equipment is omitted. Instead, the forces and torques provided by these systems during operation are used to represent the systems themselves in the model, ensuring the simplicity and completeness of the model. At the same time, a separate 3D model of the flow channel structure in the integrated extrusion and injection equipment (i.e., the extrusion flow channel of the extrusion component 3 and the injection flow channel of the injection component 8) is built for real-time simulation and animation demonstration of the human-machine interface.

[0050] Experiments were conducted using a physical extrusion and injection equipment to obtain the magnitude of the frictional force between the various mechanical components when the equipment was running under rated conditions and at rated speed. The obtained frictional force data was then input into the 3D model of the extrusion and injection equipment to complete the construction of the 3D model.

[0051] S2: Render the obtained 3D model texture of the integrated extrusion and injection equipment into Unreal Engine, and calibrate and verify the model to obtain a digital twin that matches the operating data of the integrated extrusion and injection equipment; in this embodiment, Unity 3D is used as the Unreal Engine. The specific method for calibrating and verifying the model is as follows:

[0052] like Figure 4 As shown, the system acquires the operating data of each mechanical component during the operation of the physical integrated extrusion equipment and its 3D model under the same preset working conditions and speed. The system then sequentially acquires the difference between the operating data of each mechanical component in the physical integrated extrusion equipment and the operating data in the 3D model of the integrated extrusion equipment, and determines whether this difference is within the standard deviation range. If not, the system reacquires the frictional force generated by each mechanical component at a preset speed during the operation of the physical integrated extrusion equipment under the preset working conditions, and re-inputs the frictional force generated by each mechanical component at the preset speed into the corresponding mechanical component in the 3D model of the integrated extrusion equipment, until the difference in the operating data of each mechanical component is within the corresponding standard deviation range.

[0053] S3: By using sensors placed at key nodes of each flow channel of the integrated extrusion equipment and force and torque sensors placed at the effect points of electrical system 1 and hydraulic system 5, process parameter data of the integrated extrusion equipment during operation are collected. This data is then transmitted through a data server to a digital twin created in Unreal Engine to obtain a product dataset, thus realizing the information flow transmission from the physical entity to the virtual digital twin.

[0054] S4: Build and train a neural network model for machine learning. The specific steps are as follows:

[0055] like Figure 5As shown, the data in the product dataset (i.e., the sensor data received by the digital twin) can be divided into two categories: data showing defects in the final product and data showing no defects in the final product. The data in the product dataset is further divided into training and testing sets according to a certain ratio. The training set contains significantly more data than the testing set, and both sets contain data showing both defects and no defects in the final product. The training set is input into the neural network model for training, and then the testing set is input into the trained neural network model to obtain the prediction results of the trained neural network model for product defects. These predictions are then compared with the actual production results to determine if the prediction accuracy is a relatively high value (i.e., whether it reaches a threshold). If not, step S3 is repeated, increasing the amount of data in the training set and re-inputting the training set into the neural network model for training, until the prediction accuracy of the trained neural network model for the product production results reaches a relatively high value. This trained model is then used for subsequent applications.

[0056] S5: During the operation of the integrated extrusion and injection equipment, the real-time data collected by the sensors is input into the trained neural network model to predict product defects in real time. If no defects are detected, the real-time data is re-inputted, and the cycle of judgment begins again. When defects may occur, the integrated extrusion and injection equipment is controlled by a digital twin to avoid defects. That is, an early warning is issued through the human-machine interface, the processor selects the process parameters that should be adjusted, and communicates with the control system of the integrated extrusion and injection equipment through a specific communication protocol to control the physical entity of the integrated extrusion and injection equipment and adjust the process parameters during the operation of the integrated extrusion and injection equipment, thereby achieving early defect avoidance. In this embodiment, the specific communication protocol selected is the OPC protocol.

[0057] When a trained neural network model predicts potential product defects, the processor uses the following method to select the process parameters that should be adjusted:

[0058] The processor first uses a random module to randomly change one or more process parameter data, and then inputs the changed process parameter data into the trained neural network model for prediction. If the prediction result is that no defects will be produced, the modified parameter is returned; if the prediction result is still that defects will be produced, the data is randomly changed again and input into the trained neural network model again until the prediction result is that no defects will be produced, and then the parameter is returned.

[0059] It will be understood by those skilled in the art that the above descriptions are merely preferred examples of the invention and are not intended to limit the invention. Although the invention has been described in detail with reference to the foregoing examples, those skilled in the art can still modify the technical solutions described in the foregoing examples or make equivalent substitutions for some of the technical features. All modifications and equivalent substitutions made within the spirit and principles of the invention should be included within the scope of protection of the invention.

Claims

1. A digital twin-based integrated extrusion process control and product defect avoidance system, characterized in that, include: Integrated extrusion equipment, digital twin, neural network model; The integrated extrusion and injection equipment integrates extrusion and injection molding functions, including: an extrusion assembly, a storage tank, an injection molding assembly, an electrical system, a hydraulic system, and a mold template; the extrusion assembly is connected to the storage tank, the storage tank is connected to the injection molding assembly, the electrical system provides a power source for the extrusion assembly, the hydraulic system provides power for the injection molding assembly, and the injection molding assembly is connected to the mold template; multiple sensors are arranged inside the extrusion channel of the extrusion assembly and the injection channel of the injection molding assembly, as well as at the power output points of the electrical system and the hydraulic system, and all sensors are connected to a digital twin; The digital twin includes: a three-dimensional model of the integrated extrusion and injection equipment, forces and torques abstracted from the electrical and hydraulic systems, a database storing data collected by sensors, a processor, and a human-machine interface; the processor is used to filter out the process parameters that the integrated extrusion and injection equipment needs to be adjusted when product defects are predicted. The neural network model includes an input layer and an output layer. The input layer receives sensor data received by the digital twin, and the output layer outputs the judgment result of whether the product is defective. After training, the accuracy of the neural network model in judging whether the manufactured product is defective reaches a threshold.

2. The extrusion-injection integrated process control and product defect avoidance system based on digital twins according to claim 1, characterized in that, The integrated extrusion and injection equipment also includes: a feed funnel, a connecting body, and a switching valve plate; the extrusion assembly includes: an extrusion channel, an extrusion screw, a feed inlet, and a discharge outlet; The feeding funnel is connected to the feeding port, and the raw material is fed into the extrusion assembly through the feeding funnel and subjected to melting and plasticizing treatment; the end of the extrusion channel is the discharge port, and the extrusion screw is coaxially arranged in the extrusion channel. The motor output port of the electrical system is connected to the extrusion screw, and the extrusion screw transports the molten raw material to the discharge port; the discharge port is connected to the storage tank through a connector; the storage tank is connected to the injection molding assembly through a conversion valve plate; the hydraulic system provides power to the injection molding assembly, injecting the molten raw material stored in the storage tank into the mold clamped in the mold template through the injection molding channel.

3. The extrusion-injection integrated process control and product defect avoidance system based on digital twins according to claim 1, characterized in that, The extrusion channel of the extrusion assembly, the storage tank, the injection channel of the injection molding assembly, and the interior of the mold clamped in the mold template are all equipped with multiple temperature sensors, pressure sensors, viscosity sensors, flow rate sensors, and flow sensors. Force and torque sensors are arranged at the power output of the electrical and hydraulic systems.

4. A method for integrated extrusion and injection process control and product defect avoidance based on digital twins, implemented based on the integrated extrusion and injection process control and product defect avoidance system based on digital twins as described in any one of claims 1-3, characterized in that, Specifically, the following steps are included: S1: Based on the mechanical structure of the integrated extrusion and injection equipment, establish a 3D model of the integrated extrusion and injection equipment; S2: Input the 3D model of the integrated extrusion and injection equipment into Unreal Engine, and calibrate and verify the 3D model of the integrated extrusion and injection equipment to obtain a digital twin that matches the operating data of the integrated extrusion and injection equipment; S3: Collect process parameter data of the integrated extrusion equipment during operation by sensors placed at each key node of the flow channel and at the effect points of the electrical and hydraulic systems of the integrated extrusion equipment, and input these data into the digital twin to obtain the product dataset; S4: Using the product dataset as input, build and train a neural network model; S5: When the extrusion and injection equipment is running, the real-time data collected by the sensors is input into the trained neural network model to predict product defects in real time; if there are no defects, the real-time data is re-inputted and the next judgment is started; if there are defects, the extrusion and injection equipment is controlled by a digital twin to avoid defects.

5. The method for integrated extrusion and injection process control and product defect avoidance based on digital twins according to claim 4, characterized in that, In S1, when establishing the 3D model of the integrated extrusion and injection equipment, the modeling of the electrical and hydraulic systems of the integrated extrusion and injection equipment is omitted, and the force and torque provided by them during operation are used to represent the system itself in the model; at the same time, a separate 3D model of the flow channel structure in the integrated extrusion and injection equipment is established for real-time simulation and animation demonstration of the human-machine interface.

6. The method for integrated extrusion and injection process control and product defect avoidance based on digital twins according to claim 4, characterized in that, In step S2, the specific method for calibrating and verifying the 3D model of the integrated extrusion equipment is as follows: Under the same preset operating conditions and speeds, obtain the operating data generated by each mechanical component of the physical extrusion equipment and its 3D model. Sequentially obtain the difference between the operating data of each mechanical component in the physical extrusion equipment and the operating data in the 3D model of the extrusion equipment, and determine whether the difference is within the standard deviation range. If not, re-obtain the frictional force generated by each mechanical component at a preset speed during the operation of the physical extrusion equipment under the preset operating conditions, and re-set the frictional force generated by each mechanical component at the preset speed into the corresponding mechanical component in the 3D model of the extrusion equipment, until the difference in the operating data of each mechanical component is within the corresponding standard deviation range.

7. The method for integrated extrusion and injection process control and product defect avoidance based on digital twins according to claim 4, characterized in that, The specific operation of S4 is as follows: The product dataset is divided into a training set and a test set. The training set contains significantly more data than the test set, and both sets include data on defective and non-defective final products. The training set is then input into the neural network model for training. The test set is then input into the trained neural network model to verify whether the accuracy of the trained neural network model in predicting product defects reaches the threshold. If not, step S3 is repeated, increasing the amount of data in the training set and re-inputting the training set into the neural network model for training until the accuracy of the neural network model's prediction reaches the threshold. If it does, the model is considered the successfully trained neural network model.

8. The method for integrated extrusion and injection process control and product defect avoidance based on digital twins according to claim 4, characterized in that, In step S5, when the trained neural network model predicts a product defect, the method for defect avoidance by controlling the integrated extrusion and injection equipment through a digital twin is as follows: The processor in the digital twin first uses a random module to randomly change one or more process parameter data, and then inputs the changed process parameter data into the trained neural network model for prediction. If the prediction result is that no defects will be produced, the process parameter data is returned; if the prediction result is still that defects will be produced, the data is randomly changed again and input into the trained neural network model again until the prediction result is that no defects will be produced, and the process parameter data is returned. The digital twin communicates with the control system of the physical extrusion and injection equipment through a specific communication protocol, adjusting the process parameters during the operation of the extrusion and injection equipment to achieve early defect avoidance.

9. The method for integrated extrusion and injection process control and product defect avoidance based on digital twins according to claim 8, characterized in that, The specific communication protocol used is the OPC protocol.

Images