Environment-friendly plastic product extrusion molding intelligent parameter control system and control method

Abstract

The invention relates to the technical field of plastic processing, and discloses an environment-friendly plastic product extrusion molding intelligent parameter control system and method, and the system comprises a data collection and fusion module, a causal association construction module, a digital twinning and virtual-real interaction verification module, an intelligent control module, a self-adaptive reconstruction module, and a man-machine interaction and visualization module. The method comprises the steps of collecting data in real time, and performing deep fusion to generate features; dynamically constructing a knowledge graph by fusing features, and synchronously updating causal topology; constructing a digital twinborn model, and performing virtual operation prediction; generating an instruction to regulate and control the microstructure; recognizing and classifying disturbances, and reconstructing a causal topology to guide strategy adjustment; the system state is displayed in real time, and operation intervention is supported. On-line microstructure data are acquired and deeply fused in real time through the data acquisition and fusion module, and a deep correlation model is constructed in combination with the microstructure guide reverse deduction unit, so that the problems of large batch difference and unstable physical properties of environment-friendly plastic materials are effectively solved.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of plastic processing, in particular to an intelligent parameter control system and method for extrusion molding of environment-friendly plastic products. BACKGROUND

[0002] Global environmental problems are becoming increasingly serious, which has promoted the rapid development and wide application of environment-friendly plastics (such as biodegradable plastics, recycled plastics, and bio-based plastics). These materials, with their unique environmentally friendly characteristics, are gradually replacing traditional plastics and showing great market potential in packaging, agricultural films, automobiles, electronics, and other fields. However, compared to traditional general-purpose plastics, the material properties of environment-friendly plastics are more complex and significantly different between batches.

[0003] Existing extrusion molding of environment-friendly plastics usually follows and improves the processing technology of general-purpose plastics, which is based on the rotation of the screw, heating, and the shaping effect of the mold to convert solid plastic particles into continuous products. In actual production, engineers mainly rely on preset process parameters (such as temperature, pressure, screw speed, and pulling speed) and combine with operating experience for manual adjustment and offline quality detection.

[0004] However, when the existing technology is applied to environment-friendly plastics with varying characteristics, the decisive influence of the microstructure on the final mechanical, optical, and barrier properties of environment-friendly plastics is ignored, making it difficult to control product performance from a micro level. When the material batch changes, the equipment wears out, or external environmental disturbances occur, the traditional system is difficult to identify and adapt to material batch changes independently and quickly, making system adjustments often lag behind and relying on repeated trial and error of human experience. Therefore, the present application provides an intelligent parameter control system and method for extrusion molding of environment-friendly plastic products to solve the problems existing in the prior art. SUMMARY

[0005] To overcome the shortcomings of the prior art, the present application provides an intelligent parameter control system and method for extrusion molding of environment-friendly plastic products, which solves the problems of complex and variable material properties, multi-parameter strong coupling, difficulty in online regulation of microstructure, and susceptibility to unexpected disturbances in the existing extrusion molding process of environment-friendly plastics.

[0006] To achieve the above purpose, the present application is implemented by the following technical solutions: The first aspect of the present application provides an intelligent parameter control system for extrusion molding of environment-friendly plastic products, which has the following structure: The system includes a data acquisition and fusion module, a causal correlation construction module, a digital twin and virtual-real interaction verification module, an intelligent control module, a self-adaptive reconstruction module, and a human-computer interaction and visualization module.

[0007] The data acquisition and fusion module is configured to acquire, in real time, macro process parameter data, product macro quality data, melt physical property data, and molecular microstructure data of the melt or semi-solid state through a configured macro process parameter sensor, a product macro quality sensor, an online physical property sensor, and an online microstructure characterization unit. The macro process parameter sensor can acquire the temperature of each heating zone of the extruder , melt pressure , die pressure , screw rotation speed , feed amount, motor electric , motor current , torque , and product output . The product macro quality sensor can acquire the product width , thickness , surface defect index , and surface finish . The online physical property sensor can acquire the melt apparent viscosity , elastic modulus , component characteristics , degradation index , recycled material purity , and specific additive content . The online microstructure characterization unit can employ an online small-angle X-ray scattering module, an online wide-angle X-ray diffraction module, or an online dielectric spectrum module to acquire the molecular orientation degree , crystallinity , crystal form characteristics , phase separation degree , molecular chain motion relaxation time , and polarity index . The data acquisition and fusion module further includes a data fusion unit configured to perform time stamp synchronization, data cleaning, missing value processing, noise filtering, and normalization processing on the real-time acquired macro process parameter data, product macro quality data, melt physical property data, and molecular microstructure data. The normalization processing can employ the following formula: ; wherein is the original data; is the minimum value of the data; is the maximum value of the data.

[0008] The data fusion unit then performs deep fusion on the processed data through tensor decomposition or a multi-modal deep neural network to generate fusion features .

[0009] The causal correlation modeling module is connected with the data acquisition and fusion module, and is configured to dynamically construct and update an extrusion molding knowledge graph and a dynamic causal correlation topology between multi-modal data, process parameters and product performance in real time according to the fusion features. The causal correlation modeling module comprises a knowledge graph construction unit and a dynamic causal correlation learning unit. The knowledge graph construction unit is configured to integrate field expert experience, historical production batch data and a material database to construct an initial extrusion molding knowledge graph , wherein is an entity set, is a relationship set. The knowledge graph construction unit dynamically updates the knowledge graph through a knowledge graph embedding technology (for example, a TransE model) and a graph neural network (for example, GCN or GraphSAGE). The message passing mechanism of the graph neural network can be represented as: ; , wherein is a node feature vector at the i-th layer; is a neighbor set of the node ; is a normalization constant; is a learnable weight matrix; is an activation function.

[0010] The dynamic causal correlation learning unit is configured to autonomously learn and dynamically construct a nonlinear causal relationship network between multi-modal data, process parameters and product performance according to the fusion features through a causal discovery algorithm (for example, a PC algorithm or an FCI algorithm) , wherein is a variable set, is a directed edge set representing a causal relationship. The nonlinear causal relationship network is the dynamic causal correlation topology. The dynamic causal correlation topology is updated in real time according to new data.

[0011] The digital twin and virtual-real interaction verification module is connected with the causal correlation modeling module, and is configured to construct a high-fidelity digital twin model of the environmentally friendly plastic extrusion molding process. The digital twin and virtual-real interaction verification module comprises a multi-physical field coupling modeling unit, a material constitutive model integration unit and a virtual-real interaction prediction unit. The multi-physical field coupling modeling unit is configured to construct a digital twin model comprising an extruder geometry, thermodynamics, fluid mechanics and mass transfer process. The physical model can comprise a continuity equation, a momentum equation and an energy equation. For example, the momentum equation can be represented as: ; , wherein is a density; ​is the velocity vector; is the pressure; is the stress tensor; is the gravity. The material constitutive model integration unit is configured to integrate a non-Newtonian rheological constitutive model (e.g., a power-law model , where is the apparent viscosity; is the consistency coefficient; is the shear rate; is the power-law index) of the environmentally friendly plastic, a thermodynamic parameter model, a crystallization kinetics model (e.g., an Avrami equation , where is the crystallinity; is the crystallization rate constant; is the Avrami index), and a degradation kinetics model into the digital twin model. The digital twin model can be corrected in real time according to the online physical property data. The virtual-real interaction prediction unit is configured to map the fused features to the digital twin model in real time, synchronously drive the digital twin model to perform virtual running, and predict potential effects of a control strategy generated by the intelligent control module on melt flow, temperature distribution, microstructure evolution, and product macroscopic performance.

[0012] The intelligent control module, connected with the digital twin and virtual-real interaction verification module, is configured to generate a process parameter adjustment instruction based on a prediction result of the digital twin model, the knowledge graph, and the dynamic causal correlation topology, to realize macroscopic performance regulation guided by a microstructure of a product. The intelligent control module includes a microstructure-guided reverse deduction unit, a reinforcement learning predictive control unit, and a meta-learning generalization unit. The microstructure-guided reverse deduction unit is configured to construct a deep correlation model of microstructure features and macroscopic performance by using a deep generative model (e.g., a variational autoencoder or a generative adversarial network). The microstructure-guided reverse deduction unit, in combination with the knowledge graph and the dynamic causal correlation topology, calculates, by using a reverse deduction optimization algorithm (e.g., a reverse optimization based on gradient descent or a Monte Carlo tree search), a required instantaneous physical field distribution (e.g., a specific shear rate , a temperature gradient , a pressure curve , or a cooling rate ) required for a target microstructure (e.g., a target molecular orientation degree, a crystallinity, or a crystal form feature) or a macroscopic performance of a product. The calculation can be represented as an optimization problem: ; wherein is the target microstructure; is the target macroscopic performance;Predict( ) is a physical field under which the microstructure and macro performance are predicted.

[0013] The microstructure guides the reverse deduction unit to further map the instantaneous physical field distribution to generate specific process parameter adjustment instructions (e.g., screw speed adjustment amount , heating zone temperature adjustment amount , die gap adjustment amount , pulling speed adjustment amount ).

[0014] The reinforcement learning predictive control unit is used to construct a dynamic environment model of the extrusion molding process, which can predict state transitions and rewards under different process parameter actions : ; wherein is the current state, is the current action. The reinforcement learning predictive control unit learns the process parameter adjustment strategy by interacting with the environment model or the digital twin model through a reinforcement learning agent (e.g., based on deep Q network or actor-critic architecture).

[0015] The reinforcement learning agent selects the optimal action by maximizing the long-term cumulative reward, and the reward function considers the product macro quality , production efficiency , energy consumption , process parameter deviation wherein is the current state, is the current action. The reinforcement learning predictive control unit learns the process parameter adjustment strategy by interacting with the environment model or the digital twin model through a reinforcement learning agent (e.g., based on deep Q network or actor-critic architecture). The reinforcement learning agent selects the optimal action by maximizing the long-term cumulative reward, and the reward function considers the product macro quality , production efficiency , energy consumption , process parameter deviation , and the deviation of the microstructure from the target value .

[0016] The meta-learning generalization unit is used to quickly adapt the reinforcement learning agent to new environments and new tasks by a meta-learning method (for example, MAML or Reptile), and quickly adjust the optimal control strategy after the dynamic causal correlation topology is reconfigured or a new batch of materials is encountered.

[0017] The adaptive reconfiguration module is connected with the data acquisition and fusion module and the intelligent control module, and is used to identify and classify unexpected disturbances, and drive the causal correlation construction module to adaptively reconfigure the dynamic causal correlation topology, thereby guiding the intelligent control module to adjust the control strategy. The adaptive reconfiguration module includes an abnormal pattern recognition and classification unit, a dynamic correlation topology reconfiguration unit, and an endogenous safety and trust unit. The abnormal pattern recognition and classification unit is used to monitor the data stream in the fusion features in real time by using deep metric learning or unsupervised anomaly detection algorithm (for example, Isolation Forest or One-Class SVM), identify and classify abnormal data streams that deviate significantly from known patterns, and the abnormal data streams indicate material property mutation, sensor failure, equipment wear or external environment change. When the deviation between the actual production effect and the prediction of the digital twin model exceeds the preset threshold, the endogenous safety and trust unit can switch the control strategy and trigger the rapid learning process.

[0018] The human-computer interaction and visualization module is connected with each of the above modules, and is used to display the running state, decision process, prediction result and abnormal warning information of the control system in real time, and support the operator to set the target and intervene the parameters. The human-computer interaction and visualization module includes a data visualization interface, a knowledge and decision visualization interface, a digital twin simulation result display interface, a warning and diagnosis interface, and a parameter setting and intervention interface.

[0019] The second aspect of the present application provides an intelligent parameter control method for extrusion molding of environmentally friendly plastic products, which is applied to the intelligent parameter control system for extrusion molding of environmentally friendly plastic products in the first aspect of the present application. The method comprises the following steps: S1, real-time acquisition of macro process parameters, product quality data, melt physical property data and molecular microstructure data of the melt or semi-solid state in the extrusion process by a plurality of source sensors, and deep fusion of the real-time acquired data to generate fusion features; S2, dynamically constructing and updating an extrusion molding knowledge graph according to the generated fusion features, and synchronously updating a dynamic causal correlation topology between multi-modal data, process parameters and product performance; S3, constructing a digital twin model of the environmentally friendly plastic extrusion molding process based on the knowledge graph, the dynamic causal correlation topology and the fusion features, and making the digital twin model perform virtual operation and effect prediction through real-time mapping of the fusion features; S4, generating process parameter adjustment instructions based on the prediction results of the digital twin model, the knowledge graph, and the dynamic causal correlation topology to regulate the microstructure of the product, for example, regulating the target molecular orientation degree, crystallinity, or crystal form characteristics determined through prediction analysis; S5, identifying and classifying unexpected disturbances, and reconstructing the dynamic causal correlation topology to guide the adjustment control strategy; S6, real-time display of the control system running state, decision-making process, prediction results, and abnormal warning information, and supporting operators to set targets and intervene in parameters.

[0020] The present application provides an intelligent parameter control system and method for extrusion molding of environmentally friendly plastic products. The following beneficial effects are achieved: 1. The present application acquires and deeply fuses online microstructure data in real time through the data acquisition and fusion module, and combines the microstructure guide reverse deduction unit in the intelligent control module to build a deep correlation model between micro features and macro performance, effectively addressing the problems of large batch differences and unstable physical properties of environmentally friendly plastic materials, ensuring that the mechanical properties, optical properties, or barrier properties of the final product always remain within the target range.

[0021] 2. The present application introduces a causal correlation construction module that can autonomously learn and dynamically update the extrusion molding knowledge graph and the dynamic causal correlation topology between multi-modal data, process parameters, and product performance. The self-adaptive reconstruction module can identify and classify abnormal data streams and drive the dynamic causal correlation topology to quickly reconstruct, responding in real time to changes in causal relationships caused by factors such as material batch changes, equipment wear, or environmental disturbances. This eliminates the need for pre-set fixed models or extensive manual experience adjustments, improving the system's automation level and reducing the difficulty of operation and the requirement for operator expertise.

[0022] 3. The high-fidelity digital twin model constructed by the present application can be driven by real-time data for virtual operation and effect prediction, providing a safe strategy verification environment for the intelligent control module, avoiding the risk of trial and error in actual production, enhancing the reliability and safety of system decision-making, and effectively avoiding production accidents or waste caused by improper parameter adjustment. BRIEF DESCRIPTION OF DRAWINGS

[0023] Figure 1 The present application provides an intelligent parameter control system and method for extrusion molding of environmentally friendly plastic products. The following beneficial effects are achieved: Figure 2 The present application provides an intelligent parameter control system and method for extrusion molding of environmentally friendly plastic products. The following beneficial effects are achieved: Figure 3 The present application provides an intelligent parameter control system and method for extrusion molding of environmentally friendly plastic products. The following beneficial effects are achieved: Figure 4A structure schematic diagram of the digital twin and virtual-real interaction verification module of the present application; Figure 5 A structure schematic diagram of the intelligent control module of the present application; Figure 6 A structure schematic diagram of the adaptive reconstruction module of the present application; Figure 7 A structure schematic diagram of the man-machine interaction and visualization module of the present application.

[0024] Among them, 100, data acquisition and fusion module; 200, causal correlation construction module; 300, digital twin and virtual-real interaction verification module; 400, intelligent control module; 500, adaptive reconstruction module; 600, man-machine interaction and visualization module. DETAILED DESCRIPTION

[0025] The technical solutions in the embodiments of the present application will be described clearly and completely below with reference to the drawings in the specification of the present application. Obviously, the described embodiments are only some of the embodiments of the present application, not all. Based on the embodiments in the present application, all other embodiments obtained by those skilled in the art without creative labor fall within the scope of protection of the present application.

[0026] Referring to the drawings Figure 1 , Figure 1 It is a schematic diagram of an intelligent parameter control system for extrusion molding of environmentally friendly plastic products according to an embodiment of the present application. The present application provides an intelligent parameter control system for extrusion molding of environmentally friendly plastic products, which comprises: a data acquisition and fusion module 100, a causal correlation construction module 200, a digital twin and virtual-real interaction verification module 300, an intelligent control module 400, an adaptive reconstruction module 500, and a man-machine interaction and visualization module 600.

[0027] The data acquisition and fusion module 100 is used to acquire multi-source data in the extrusion process in real time, specifically including macro process parameters, product macro quality data, melt physical property data, and molecular microstructure data of the melt or semi-solid state. This module synchronizes the timestamps of the collected heterogeneous data, cleans the data, processes missing values, filters out noise, and normalizes the data, and further generates fusion features through deep fusion technology.

[0028] The causal correlation construction module 200 is connected with the data acquisition and fusion module 100 and is used to receive fusion features. This module dynamically constructs and updates the extrusion molding knowledge graph based on the fusion features in real time, and synchronously updates the dynamic causal correlation topology between multi-modal data, process parameters, and product performance.

[0029] The digital twin and virtual-real interaction verification module 300 is connected with the causal correlation construction module 200, and is used for constructing a high-fidelity digital twin model of the plastic extrusion process. Based on the knowledge graph, the dynamic causal correlation topology and the fused features, the module maps the fused features in real time to drive the digital twin model to perform virtual operation and effect prediction.

[0030] The intelligent control module 400 is connected with the digital twin and virtual-real interaction verification module 300. Based on the prediction result of the digital twin model, the knowledge graph and the dynamic causal correlation topology, the module generates a process parameter adjustment instruction to regulate the microstructure of the product.

[0031] The adaptive reconstruction module 500 is connected with the data acquisition and fusion module 100 and the intelligent control module 400. The module is used for identifying and classifying unexpected disturbances appearing in the data stream. When an unexpected disturbance is identified, the module drives the causal correlation construction module 200 to adaptively reconstruct the dynamic causal correlation topology. The reconstructed dynamic causal correlation topology in turn guides the intelligent control module 400 to adjust the control strategy.

[0032] The man-machine interaction and visualization module 600 is connected with the above-mentioned modules. The module is used for real-time display of the running state, decision-making process, prediction result and abnormal warning information of the control system, and supports the operator to set targets and intervene in parameters.

[0033] The working process of the system is as follows: the data acquisition and fusion module 100 continuously acquires multi-dimensional real-time data from the extrusion production line, and processes and fuses the data into unified fused features. The fused features are transmitted to the causal correlation construction module 200 for continuously updating and perfecting the extrusion molding knowledge graph and the dynamic causal correlation topology. Meanwhile, the digital twin and virtual-real interaction verification module 300 utilizes the knowledge, topology and fused features to construct and drive a high-fidelity digital twin model in real time, to virtually simulate and predict the effect of the actual extrusion process. The intelligent control module 400 receives the prediction result of the digital twin model and the knowledge graph and the dynamic causal correlation topology, and generates an optimized process parameter adjustment instruction based on the information, and then controls the extruder to realize precise regulation of the microstructure of the product. The adaptive reconstruction module 500 monitors the data stream, and once an unexpected disturbance is identified, triggers the causal correlation construction module 200 to reconstruct the causal correlation topology, and guides the intelligent control module 400 to adjust its control strategy to cope with changes. The man-machine interaction and visualization module 600 provides comprehensive information display and operation interface to ensure the transparency and controllability of the system.

[0034] Referring to the accompanying drawings Figure 2 , Figure 2 is a structural schematic diagram of the data acquisition and fusion module 100 according to an embodiment of the application.

[0035] The data acquisition and fusion module 100 is used to acquire multi-source and multi-dimensional data in the extrusion process. The module includes macro process parameter sensors, product macro quality sensors, online physical property sensors, online microstructure characterization units, and a data fusion unit.

[0036] The macro process parameter sensors are deployed at corresponding positions of the extruder to acquire data such as the temperature of each heating zone of the extruder , the melt pressure , the die pressure , the screw rotation speed , the feed amount , the motor current , the torque , and the product output , in real time. These sensors output data at a preset frequency (e.g., 10 times per second).

[0037] The product macro quality sensors are deployed in the detection area after the extruded product is offline to acquire data such as the product width , the thickness , the surface defect index (e.g., identified and quantified by a machine vision system), and the surface finish , in real time.

[0038] The online physical property sensors are integrated in the extruder die or the melt conveying pipeline to acquire data such as the apparent viscosity of the melt , the elastic modulus , the component characteristics , the degradation index , the purity of recycled materials , and the content of specific additives , in real time. The online physical property sensors can use devices such as online rheometers, near-infrared spectrometers, or differential scanning calorimeters (DSC).

[0039] The online microstructure characterization units are deployed near the extrusion die outlet or in the cooling section to acquire data on the molecular microstructure of the melt or semi-solid state, including the molecular orientation degree , the crystallinity , the crystal form characteristics , the degree of phase separation , the molecular chain motion relaxation time , and the polarity index . The online microstructure characterization units can use any one or a combination of online small-angle X-ray scattering modules, online wide-angle X-ray diffraction modules, or online dielectric spectrum modules.

[0040] The data fusion unit is connected with the above-mentioned various sensors and characterization units. Its function is to process the raw data obtained from the macro-process parameter sensor, product macro-quality sensor, online physical property sensor and online microstructure characterization unit. First, the data fusion unit performs timestamp synchronization to ensure that the data from different sources are aligned on the time axis. Second, data cleaning and missing value processing are performed, for example, interpolation method or machine learning-based method is used to fill in the missing data. Third, noise filtering is performed, for example, Kalman filtering or wavelet denoising is used. Finally, the processed data is normalized.

[0041] After the preprocessing is completed, the data fusion unit performs deep fusion on the processed multi-dimensional and multi-modal data through tensor decomposition or multi-modal deep neural network to generate unified fusion features . The fusion features, as high-dimensional vectors, can comprehensively represent the macro-state, microstructure and physical property information of the current extrusion process.

[0042] Referring to the accompanying drawings Figure 3 , Figure 3 is a structural schematic diagram of a causal correlation construction module 200 according to an embodiment of the present application.

[0043] The causal correlation construction module 200 is connected with the multi-scale and multi-modal data acquisition and fusion module 100, and is used to receive the fusion features . The function of this module is to dynamically construct and real-time update the extrusion molding knowledge graph and the dynamic causal correlation topology between multi-modal data, process parameters and product performance. The causal correlation construction module 200 includes a knowledge graph construction unit and a dynamic causal correlation learning unit.

[0044] The knowledge graph construction unit is used to integrate the field expert experience, historical production batch data and material database to construct an initial extrusion molding knowledge graph . Among them, represents a set of entities, for example: environmental plastic grade, extruder model, process parameter (such as screw speed, temperature section), microstructure feature (such as crystallinity, molecular orientation degree) and product performance (such as tensile strength, impact toughness); represents a set of relationships, for example: influence, composed of, parameter of.

[0045] The knowledge graph construction unit dynamically updates the knowledge graph through knowledge graph embedding technology and graph neural network. For example, the TransE model can be used to learn the low-dimensional vector representation of entities and relationships, and the target function aims to make the sum of the head entity vector and the relationship vector close to the tail entity vector: ; In the formula, is a triple in the knowledge graph, are the embedding vectors of the head entity, relation, and tail entity respectively, is a negative sample triplet, is the interval parameter, Indicates a positive value.

[0046] In addition, graph neural networks (e.g., graph convolutional networks GCN or GraphSAGE) can be used to aggregate and update node features in the knowledge graph to capture more complex structural information. The message passing mechanism of graph neural networks can be expressed as: ; Where, is a node In the The feature vector of the layer, is a node The set of neighbor nodes of It is an aggregation function (e.g., sum, average, or max pooling) that aggregates information about neighboring nodes. By continuously receiving new fused features, the knowledge graph construction unit can identify new entities and relationships, or update the attributes of existing entities and relationships, thereby keeping the knowledge graph up-to-date and accurate.

[0047] Dynamic causal association learning unit is used to learn , autonomously learn and dynamically construct a nonlinear causal relationship network between multimodal data, process parameters and product performance Where, It is a set of variables, including various sensor data, process parameters, intermediate state variables and product performance indicators; is a set of directed edges, representing the causal relationship between variables.

[0048] This unit can employ causal discovery algorithms, such as the PC algorithm, the FCI algorithm, or deep learning-based causal discovery methods. These algorithms infer causal graphs by analyzing conditional independence in the data or leveraging structural equation models. A nonlinear causal network is a dynamic causal topology. This dynamic causal topology is updated in real time based on newly input fused feature data. For example, when new data patterns emerge, the dynamic causal learning unit can detect new causal paths or changes in the strength of existing causal paths and incrementally update the causal topology to ensure that it accurately reflects the true causal mechanisms in the current extrusion process.

[0049] Refer to the attached Figure 4 , Figure 4 It is a structural diagram of a digital twin and virtual-reality interaction verification module according to an embodiment of the present invention.

[0050] The digital twin and virtual-real interaction verification module 300 is connected with the causal correlation construction module 200, and is used for constructing a high-fidelity digital twin model of the plastic extrusion process. Based on the knowledge graph , dynamic causal correlation topology and fusion features , the fusion features are mapped in real time to drive the digital twin model to perform virtual operation and effect prediction. The digital twin and virtual-real interaction verification module 300 includes a multi-physical field coupling modeling unit, a material constitutive model integration unit, and a virtual-real interaction prediction unit.

[0051] The multi-physical field coupling modeling unit is used to construct a digital twin model including the geometry of the extruder, thermodynamics, fluid mechanics, and mass transfer process. The digital twin model is a multi-scale, multi-physical field coupling simulation model that can accurately simulate the complex behavior of the melt inside the extruder and the product forming process. The digital twin model couples the computational fluid dynamics (CFD) method to simulate the flow, heat transfer, mixing and shear behavior of the melt inside the screw, barrel and die. For example, the mass conservation equation (continuity equation) of the melt flow can be expressed as: ; In the formula, is the fluid density, is the fluid velocity vector, is the time.

[0052] At the same time, the digital twin model combines the finite element analysis (FEA) method to simulate the deformation of the die and the forming process of the product, including cooling shrinkage, stress distribution and the formation of the final shape.

[0053] The material constitutive model integration unit is used to integrate the non-Newtonian rheological constitutive model, thermodynamic parameter model, crystallization kinetics model and degradation kinetics model of the environmentally friendly plastic into the digital twin model. For example, the commonly used power-law model of non-Newtonian fluid can be expressed as: ; In the formula, is the shear stress, is the consistency coefficient, is the shear rate, is the power-law index.

[0054] These model parameters can be dynamically corrected according to the real-time data obtained by the online physical property sensor, ensuring that the digital twin model can accurately reflect the actual physical and chemical properties of the current batch of environmentally friendly plastic. The digital twin model can be corrected in real time according to the online physical property data, thereby improving the simulation accuracy and prediction accuracy.

[0055] The virtual-real interaction prediction unit is used to map the fusion features The real-time mapping is performed on the digital twin model, and the digital twin model is driven to run virtually. This real-time mapping mechanism makes the state of the digital twin model highly consistent with the actual extrusion process. The virtual-real interaction prediction unit then uses the digital twin model to perform virtual verification and effect prediction on the strategy to be generated by the intelligent control module 400, including the strategy's effect on the melt flow state, temperature distribution, microstructure evolution (for example, molecular orientation degree), etc. and crystallinity Through prediction, the system can evaluate the expected effects of different control strategies and select the optimal control solution.

[0056] Refer to the attached Figure 5 , Figure 5 FIG. 1 is a schematic structural diagram of an intelligent control module according to an embodiment of the present invention.

[0057] The intelligent control module 400 is connected to the digital twin and virtual-reality interaction verification module 300 to calculate the prediction results and knowledge graph based on the digital twin model. and dynamic causal topology , generating process parameter adjustment instructions to achieve control of the product microstructure. The intelligent control module 400 includes a microstructure-guided reverse deduction unit, a reinforcement learning prediction control unit, and a meta-learning generalization unit.

[0058] The microstructure-guided inverse deduction unit is used to build a deep correlation model between micro features and macro properties using deep generative models (e.g., variational autoencoders (VAEs) or generative adversarial networks (GANs). This deep correlation model can learn microstructures (such as molecular orientation) from fused features. , crystallinity , crystal characteristics ) and macroscopic properties (such as tensile strength, impact toughness, surface finish). Unit combined with knowledge graph and dynamic causal topology , calculate the instantaneous physical field distribution required to achieve the target microstructure or macroscopic performance through inverse optimization algorithms (e.g., gradient descent-based inverse optimization or Monte Carlo tree search MCTS) This calculation can be formulated as an optimization problem, which aims to find a set of physical parameters such that the microstructure and macroscopic properties simulated in the digital twin model are as close as possible to the target values: ; Where, In the physical field The microstructure and macro properties predicted by the digital twin model under the action of targeting a microstructure (e.g., a specific degree of molecular orientation or crystallinity), targeting a macroscopic property. The physical field distribution can include, but is not limited to, a specific shear rate , a temperature gradient , a pressure profile , a cooling rate . The microstructure-guided inverse deduction unit then reversely maps the instantaneous physical field distribution to generate concrete process parameter adjustment instructions, e.g., a screw rotation speed adjustment amount , a temperature adjustment amount of each heating zone , a die gap adjustment amount , or a pulling speed adjustment amount .

[0059] The reinforcement learning predictive control unit is used to build a dynamic environment model of the extrusion molding process . The environment model is capable of predicting the next state and the obtained reward in the case of a given current state and process parameter adjustment action , and the state transition process can be represented as: ; The reinforcement learning predictive control unit interacts with the environment model or digital twin model through a reinforcement learning agent (e.g., based on a deep Q network DQN or an actor-critic Actor-Critic architecture) to learn the optimal process parameter adjustment strategy. The reinforcement learning agent selects actions by maximizing the long-term cumulative reward.

[0060] The meta-learning generalization unit is used to enable the reinforcement learning agent to quickly adapt to new extrusion environments and new production tasks through a meta-learning method (e.g., a model-agnostic meta-learning MAML or Reptile algorithm). This unit enables the agent to quickly adjust its learned strategy with a small amount of new data or new task experience, rather than starting from scratch. Therefore, when the dynamic causal correlation topology is reconfigured or a new batch of materials is encountered, the meta-learning generalization unit can guide the reinforcement learning agent to quickly converge to a new optimal control strategy.

[0061] Referring to the accompanying Figure 6 , Figure 6 is a schematic diagram of an adaptive reconfiguration module structure according to an embodiment of the present application.

[0062] The adaptive reconstruction module 500 is connected with the multi-scale multi-modal data acquisition and fusion module 100 and the intelligent control module 400, and is used to identify and classify unexpected disturbances, and drive the causal correlation construction module 200 to reconstruct the dynamic causal correlation topology. The reconstructed dynamic causal correlation topology in turn guides the intelligent control module 400 to adjust the control strategy. The adaptive reconstruction module 500 includes an abnormal pattern identification and classification unit, a dynamic correlation topology reconstruction unit, and an endogenous safety and trust unit.

[0063] The abnormal pattern identification and classification unit is used to monitor the data stream in the fusion feature in real time by using deep metric learning or unsupervised anomaly detection algorithms (such as Isolation Forest or One-Class SVM). This unit identifies abnormal data streams that deviate significantly from known patterns by learning the distribution of normal data patterns. Abnormal data streams indicate possible unexpected disturbances, such as material property mutations (e.g., inconsistent batches of recycled materials), sensor failures, equipment wear and tear, or external environmental changes (e.g., fluctuations in workshop temperature). Anomaly detection usually involves calculating the "anomaly score" between data points and normal data distribution When the score exceeds a pre-set threshold, it is determined to be abnormal.

[0064] The dynamic correlation topology reconstruction unit is used to drive the dynamic causal correlation learning unit in the causal correlation construction module 200 to perform rapid re-execution or reinforcement learning process after identifying unexpected disturbances. This reconstruction process aims to update or correct the current dynamic causal correlation topology to accurately reflect the changes in internal causal relationships caused by disturbances. For example, when the material composition changes, it may cause the causal relationship strength between certain process parameters and product microstructure to change, or new causal paths to appear. The dynamic correlation topology reconstruction unit ensures that the causal topology can quickly adapt to these changes.

[0065] The endogenous safety and trust unit is used to introduce safety constraint reinforcement learning to ensure that process parameter adjustment instructions are always executed within the pre-set safe operation range . This unit achieves this goal by adding a penalty term to the reward function of reinforcement learning or by using constraint optimization methods. For example, the action selection can be constrained in the following way: ; where is the loss function of reinforcement learning, a safe operation region of the process parameters. In addition, when the actual production effect deviates from the prediction of the digital twin model 300 beyond a preset threshold (for example, the actual value of the product thickness deviates from the predicted value by more than a certain percentage), the endogenous safety and trust unit can trigger an emergency response mechanism, switch to a preset safety control strategy, and at the same time trigger the intelligent control module 400 to learn quickly to adapt to the new working condition as soon as possible and restore the optimal control.

[0066] Referring to the drawings Figure 7 , Figure 7 is a schematic diagram of a human-computer interaction and visualization module structure according to an embodiment of the present application.

[0067] The human-computer interaction and visualization module 600 is connected with each of the above modules, and is used to display the running state, decision-making process, prediction result and abnormal early warning information of the control system in real time, and support the operator to set the target and intervene the parameters. The human-computer interaction and visualization module 600 includes a data visualization interface, a knowledge and decision visualization interface, a digital twin simulation result display interface, a warning and diagnosis interface, and a parameter setting and intervention interface.

[0068] The data visualization interface is used to display various data collected by the multi-scale multi-modal data acquisition and fusion module 100 in a graphical manner in real time, including macro process parameters (such as temperature curve, pressure fluctuation), product macro quality data (such as width, thickness trend chart), melt physical property data (such as viscosity change), and molecular microstructure data (such as crystallinity change over time). The data can be presented in the form of line chart, column chart, scatter chart or real-time curve chart, etc. to facilitate the operator to monitor various indicators of the production process.

[0069] The knowledge and decision visualization interface is used to display the extrusion molding knowledge graph and dynamic causal relationship topology constructed by the causal relationship construction module 200. The knowledge graph can be displayed in the form of node-edge graph, and the entities and relationships are clear and visible. The dynamic causal relationship topology is presented in the form of causal graph, showing the causal relationship strength and direction between variables. In addition, the interface can also display the reasoning path and basis of the decision generated by the intelligent control module 400, for example, explaining that a certain process parameter adjustment is based on the reverse deduction result of the target microstructure.

[0070] The digital twin simulation result display interface is used to display the simulation result of the digital twin and virtual-real interaction verification module 300. The interface can display the dynamic three-dimensional visualization image of the temperature field, pressure field and velocity field distribution of the melt simulated by the digital twin model inside the screw, barrel and die, as well as the stress, strain and final shape prediction in the product forming process. For example, the shear rate distribution inside the die can be visualized, and the shear rate at a certain position can be represented as: ; wherein, is the deformation rate tensor; is the shear rate.

[0071] The early warning and diagnosis interface is used to receive and display the abnormality warning information sent by the adaptive reconstruction module 500. This interface can clearly indicate the type of abnormality (such as sensor failure, batch mutation, equipment wear) and provide a diagnosis report indicating the possible causes and the scope of impact. At the same time, when the endogenous safety and trust unit triggers the safety mode or the fast learning process, this interface will prompt the operator in real time.

[0072] The parameter setting and intervention interface is used to support the operator in setting the target and intervening the parameters of the system. The operator can input the desired product macroscopic performance target or microscopic structure target through this interface, for example, set the target tensile strength or target crystallinity. At the same time, the operator can also manually intervene or correct the process parameter adjustment instructions generated by the intelligent control module 400 in specific cases, and observe the system response after intervention and the prediction results of the digital twin model.

[0073] Although embodiments of the present application have been shown and described, it is to be understood that various modifications, substitutions, replacements and changes can be made to these embodiments without departing from the principles and spirit of the present application, and the scope of the present application is defined by the appended claims and their equivalents.

Claims

1. Intelligent parameter control system for extrusion molding of environmentally friendly plastic products, characterized by: include: The data acquisition and fusion module is used to obtain macro process parameters, product quality data, melt physical property data, and molecular microstructure data of the melt or semi-solid state in real time during the extrusion process through multi-source sensors, and deeply fuse the real-time data to generate fusion features; A causal association building module is used to dynamically build and update the extrusion molding knowledge graph in real time based on the generated fusion features, and to synchronously update the dynamic causal association topology between multimodal data, process parameters, and product performance; The digital twin and virtual-reality interaction verification module is used to build a digital twin model of the environmentally friendly plastic extrusion molding process based on knowledge graphs, dynamic causal association topology, and fusion features. By mapping fusion features in real time, the digital twin model can be used for virtual operation and effect prediction. An intelligent control module is used to generate process parameter adjustment instructions based on the prediction results, knowledge graph, and dynamic causal association topology of the digital twin model to regulate the microstructure of the product; The adaptive reconstruction module is used to identify and classify unexpected disturbances and reconstruct the dynamic causal topology through the causal construction module. The reconstructed dynamic causal topology guides the intelligent control module to adjust the control strategy; The human-computer interaction and visualization module is used to display the control system's operating status, decision-making process, prediction results, and abnormal warning information in real time, and supports operators in setting goals and intervening in parameters.

2. The intelligent parameter control system for extrusion molding of environmentally friendly plastic products according to claim 1 is characterized in that: The data acquisition and fusion module includes: Macro process parameter sensors are used to obtain the temperature of each heating zone of the extruder, melt pressure, die head pressure, screw speed, feed rate, motor current, torque and product output; Product macro-quality sensor, used to obtain product width, thickness, surface defect index and surface finish; Online physical property sensors are used to obtain melt apparent viscosity, elastic modulus, component characteristics, degradation index, recycled material purity, and specific additive content; Online microstructure characterization unit, used to obtain molecular orientation, crystallinity, crystal form characteristics, phase separation degree, molecular chain motion relaxation time and polarity index; The data fusion unit is used to synchronize timestamps, perform data cleaning, handle missing values, filter out noise, and normalize the acquired data, and deeply fuse the processed data through tensor decomposition or multimodal deep neural network to generate fusion features.

3. The intelligent parameter control system for extrusion molding of environmentally friendly plastic products according to claim 1 is characterized in that: The causal association building module includes: The knowledge graph construction unit is used to integrate historical production data and material databases to build an extrusion molding knowledge graph, and dynamically update the knowledge graph through knowledge graph embedding technology and graph neural network; A dynamic causal association learning unit is used to autonomously learn and dynamically construct a nonlinear causal relationship network between multimodal data, process parameters and product performance based on fusion features. The nonlinear causal relationship network is a dynamic causal association topology, and the dynamic causal association topology is updated in real time according to new data.

4. The intelligent parameter control system for extrusion molding of environmentally friendly plastic products according to claim 1 is characterized in that: The digital twin and virtual-reality interaction verification module includes: A multi-physics coupling modeling unit is used to build a digital twin model that includes the extruder geometry, thermodynamics, fluid mechanics, and mass transfer process. The digital twin model is coupled with computational fluid dynamics to simulate melt flow and heat transfer, and is combined with finite element analysis to simulate die deformation and product formation. A material constitutive model integration unit is used to integrate the non-Newtonian rheological constitutive model, thermodynamic parameter model, crystallization kinetic model, and degradation kinetic model of environmentally friendly plastics into the digital twin model, which can be modified in real time based on online physical property data; The virtual-reality interactive prediction unit is used to map the fusion features to the digital twin model in real time, synchronously drive the digital twin model to perform virtual operation, and predict the impact of the control strategy generated by the intelligent control module on melt flow, temperature distribution, microstructure evolution and product macroscopic performance.

5. The intelligent parameter control system for extrusion molding of environmentally friendly plastic products according to claim 1 is characterized in that: The intelligent control module includes: The microstructure-guided reverse deduction unit is used to build a deep correlation model between microscopic features and macroscopic performance using a deep generative model. Combining the knowledge graph and dynamic causal correlation topology, it calculates the instantaneous physical field distribution required for the target microstructure or macroscopic performance through a reverse deduction algorithm, and generates process parameter adjustment instructions through reverse mapping. A reinforcement learning prediction control unit is used to build a dynamic environment model of the extrusion process. Through the interaction between the reinforcement learning agent and the environment model or digital twin model, it learns the process parameter adjustment strategy; A meta-learning generalization unit is used to enable reinforcement learning agents to quickly adapt to new environments and tasks through meta-learning methods, and to quickly adjust the optimal control strategy after dynamic causal topology reconstruction or when encountering new material batches.

6. The intelligent parameter control system for extrusion molding of environmentally friendly plastic products according to claim 5, characterized in that: The intelligent control module guides the reverse deduction unit through the microstructure to regulate the target molecular orientation, crystallinity or crystal form characteristics determined by predictive analysis.

7. The intelligent parameter control system for extrusion molding of environmentally friendly plastic products according to claim 1 is characterized in that: The adaptive reconstruction module includes: An abnormal pattern recognition and classification unit, which is used to identify and classify abnormal data streams in the fused features using deep metric learning or unsupervised anomaly detection algorithms, wherein the abnormal data streams indicate sudden changes in material properties, sensor failure, equipment wear, or changes in the external environment; The dynamic association topology reconstruction unit is used to drive the dynamic causal association learning unit to quickly rerun or strengthen learning after identifying unexpected disturbances, and reconstruct the dynamic causal association topology; The intrinsic safety and trust unit is used to introduce safety-constrained reinforcement learning to ensure that process parameters are within the safe operating range. When the deviation between the actual effect and the digital twin prediction exceeds the threshold, the control strategy is switched and rapid learning is triggered.

8. The intelligent parameter control system for extrusion molding of environmentally friendly plastic products according to claim 1 is characterized in that: The human-computer interaction and visualization module includes: Data visualization interface for real-time display of sensor data, fusion features, and microstructure characterization results; Knowledge and decision visualization interface, used to visualize the current structure of the extrusion molding knowledge graph, dynamic causal association topology, control strategy and decision basis; Digital twin simulation result display interface, used to present the simulation results and prediction trends of the digital twin model in real time; Early warning and diagnosis interface, used to promptly display abnormality identification and tracing results, and provide fault causes and recommended solutions; Parameter setting and intervention interface is used to support operators in setting product quality targets, microstructure targets, and energy consumption targets.

9. The intelligent parameter control system for extrusion molding of environmentally friendly plastic products according to claim 2, characterized in that: The online microstructure characterization unit adopts an online small-angle X-ray scattering module, an online wide-angle X-ray diffraction module or an online dielectric spectroscopy module.

10. An intelligent parameter control method for extrusion molding of environmentally friendly plastic products, applied to the intelligent parameter control system for extrusion molding of environmentally friendly plastic products according to any one of claims 1 to 9, characterized in that: The following steps are involved: S1. Use multi-source sensors to acquire macro process parameters, product quality data, melt physical property data, and molecular microstructure data of the melt or semi-solid state in real time during the extrusion process, and deeply fuse the real-time acquired data to generate fusion features; S2. Based on the generated fusion features, dynamically build and update the extrusion molding knowledge graph in real time, and synchronously update the dynamic causal relationship topology between multimodal data, process parameters and product performance; S3. Constructing a digital twin model of the environmentally friendly plastic extrusion molding process based on the knowledge graph, dynamic causal association topology, and fusion features, and enabling the digital twin model to perform virtual operation and effect prediction through real-time mapping of fusion features; S4. Generate process parameter adjustment instructions based on the prediction results, knowledge graph, and dynamic causal association topology of the digital twin model to regulate the microstructure of the product; S5. Identify and classify unexpected disturbances, and reconstruct the dynamic causal topology to guide the adjustment of the control strategy; S6. Display the control system's operating status, decision-making process, prediction results, and abnormal warning information in real time, and support operators in setting targets and intervening in parameters.

Images