An intelligent temperature control system for the whole process of cable production

Abstract

The present invention relates to the technical field of temperature control, and particularly to an intelligent temperature control system for the whole process of cable production. It includes a cable parameter acquisition module, a production process modeling module, a temperature control decision-making module, an execution control module, and a monitoring and early warning module. The system collects cable production parameters and uses graph neural networks and symbolic regression algorithms to model the production temperature distribution and the temperature-quality relationship. Based on the model, reinforcement learning is used to generate the optimal temperature control strategy, and precise temperature control is achieved through adaptive fuzzy control. At the same time, the system monitors temperature parameters in real time, combines sparse coding and variational inference to detect abnormalities, and issues early warning signals when over-limit or equipment failures occur, ensuring production stability and product quality. The present invention comprehensively improves the temperature control accuracy and efficiency of the whole process of cable production through intelligent modeling, decision-making, control, and monitoring means.

Description

Technical Field

[0001] The present invention relates to the technical field of temperature control, and particularly to an intelligent temperature control system for the whole process of cable production. Background Art

[0002] With the continuous development of the power system and the wide application of cable transmission technology, cables play an increasingly important role in modern power transmission. The long-term stable operation of cables is directly related to the safety and reliability of the power system. However, during actual operation, due to changes in the load carried by the cable, fluctuations in the ambient temperature, and differences in heat dissipation conditions, the temperature of the cable may change significantly. When the cable temperature is too high, it will cause accelerated aging of the insulating material, shortened cable life, and may even lead to cable failures, seriously threatening the stable operation of the power system. There are currently the following problems: It is difficult for existing technologies to accurately model the temperature transfer path and dynamic changes during the whole process of cable production, and it is impossible to achieve accurate prediction and control; existing methods are difficult to deeply explore the implicit physical laws between temperature and product quality, resulting in insufficient optimization of temperature control strategies; the accuracy of existing technologies in temperature control execution is insufficient and cannot meet the high-quality requirements of cable production. Summary of the Invention

[0003] To solve the above problems, the present invention provides an intelligent temperature control system for the whole process of cable production, which solves the problems of how to accurately model the temperature transfer path and dynamic changes during the whole process of cable production, deeply explore the implicit laws between temperature and quality, and achieve high-precision temperature control execution to ensure the quality and operation reliability of cable production, thereby improving the temperature control accuracy and efficiency of the whole process of cable production.

[0004] To achieve the above object, the technical solution adopted by the present invention is:

[0005] An intelligent temperature control system for the whole process of cable production, comprising a cable parameter acquisition module, a production process modeling module, a temperature control decision-making module, an execution control module, and a monitoring and warning module that are communicatively connected in sequence;

[0006] The cable parameter acquisition module is used to collect parameter information during the cable production process and construct a parameter database;

[0007] The production process modeling module is used to model the temperature transfer path and spatial distribution during the production process based on the parameter information by using a graph neural network, generate a dynamic temperature distribution prediction model for the entire production process, and at the same time use a symbolic regression algorithm to extract the implicit physical laws in the temperature-quality relationship;

[0008] The temperature control decision-making module is used to generate an optimal temperature control strategy by using a reinforcement learning algorithm based on the dynamic temperature distribution prediction model and the extracted implicit physical laws;

[0009] The execution control module is used to receive the temperature control strategy generated by the temperature control decision module, and convert the regulation strategy into an accurate control signal through an adaptive fuzzy control algorithm to drive the production equipment to complete temperature adjustment;

[0010] The monitoring and warning module is used to monitor the temperature parameters in the production process in real time, detect potential anomalies using sparse coding and variational inference algorithms, determine whether there are abnormal situations, and issue a warning signal when the temperature exceeds the set range or equipment failure occurs.

[0011] Furthermore, the parameter information includes ambient temperature, raw material heat capacity, cable surface heat transfer coefficient, equipment temperature rise curve and dynamic production parameters.

[0012] Furthermore, the operation process of the production process modeling module includes the following steps:

[0013] Normalize the parameter information, remove outliers through a noise reduction algorithm, and use a graph neural network to construct a temperature transfer topology graph in the production equipment and cable production process to capture the heat transfer paths and their mutual relationships between nodes in the production process;

[0014] Based on the temperature transfer topology graph, combined with dynamic production parameters, train the graph neural network model to generate a dynamic temperature distribution prediction model for different production stages, and predict the temperature change situation in the production process in real time;

[0015] Use a symbolic regression algorithm to analyze the temperature data in the prediction model, mine the implicit physical relationship between temperature changes and product quality, and generate a mathematical expression of the temperature-quality relationship;

[0016] Verify the generated temperature distribution prediction model and the extracted implicit physical laws through historical production data and experimental data, and optimize the model according to the verification results.

[0017] Even further, the formula of the dynamic temperature distribution prediction model is as follows:

[0018]

[0019] Among them, represents the instantaneous temperature distribution of the system at time t; represents the heat change amount; m represents the mass of the cable or related node; represents the specific heat capacity of the cable material; , , and respectively represent heat influence, heat conduction diffusion, external heat source and environmental benchmark influence; represents the Laplace operator of the temperature distribution; Represents the input power of the external heat source, including the dynamic heat input of the production equipment.

[0020] Furthermore, the mathematical expression of the temperature-mass relationship is as follows:

[0021]

[0022] Among them, Represents the product quality index during the cable production process; T represents the temperature of the production equipment and the cable surface; H Represents the heat in the system at present; Represents the temperature change per unit time; t represents the production time; Represents the reference temperature; Represents the cable production process efficiency parameter; C represents the heat transfer coefficient of the cable surface; Represents the temperature transfer time constant; , , , and Represents the constant coefficients extracted by the symbolic regression algorithm; a and b represent the exponential parameters.

[0023] Further, the operation process of the temperature control decision-making module includes the following steps:

[0024] Input the dynamic temperature distribution data collected in real time during the production process and the extracted implicit physical laws into the temperature control decision-making module, and perform normalization and standardization processing based on historical data;

[0025] Based on the dynamic temperature distribution prediction model, use the reinforcement learning algorithm to define the objective function of the production process, including the constraints of maximizing product quality, minimizing temperature control energy consumption, and ensuring the stability of equipment operation;

[0026] Generate a preliminary temperature control strategy through the state space and action space, and use the policy gradient algorithm for policy optimization during the training process;

[0027] Conduct multi-objective optimization analysis on the generated temperature control strategy, and output the optimal temperature control strategy.

[0028] Furthermore, the formula of the optimal temperature control strategy is as follows:

[0029]

[0030] Among them, Represents the optimal temperature control strategy; Represents the temperature control energy consumption; Represents the equipment operation stability function; Represents the temperature T of the production equipment and the target temperature deviation; , and represent weight coefficients; represents the target temperature; T represents the temperature of the production equipment and the cable surface.

[0031] Furthermore, the execution control module adjusts the control rules and membership functions in real time based on the adaptive fuzzy control algorithm, and continuously optimizes the control parameters using the production feedback data.

[0032] Furthermore, the monitoring and warning module performs real-time analysis on the production data based on the distributed computing architecture, and realizes the hierarchical push of remote warning information through the industrial Internet of Things platform.

[0033] The beneficial effects of the present invention are as follows:

[0034] The present invention models the temperature transfer path and spatial distribution in the whole process of cable production by introducing a graph neural network, generates a dynamic temperature distribution prediction model, realizes accurate prediction of temperature changes and full-process optimized control, and improves production efficiency and product quality. The symbolic regression algorithm is used to extract the implicit physical laws in the relationship between temperature and quality, providing in-depth physical mechanism support for the optimization of temperature control strategies to ensure that the temperature control adjustment is more scientific and accurate. The optimal temperature control strategy is generated based on the reinforcement learning algorithm, enabling the system to adaptively adjust according to the dynamic prediction model and the extracted laws, avoiding errors caused by human experience, and improving the scientific and intelligent level of decision-making. The temperature control strategy is converted into accurate control signals through the adaptive fuzzy control algorithm to ensure that the production equipment can accurately execute the regulation instructions and maintain the temperature stability in the cable production process, improving product consistency. The monitoring and warning module realizes real-time monitoring and anomaly detection of the production process through sparse coding and variational inference algorithms, and can give early warnings in time when the temperature exceeds the limit or the equipment fails, preventing the further expansion of potential problems and ensuring the safety and reliability of production. BRIEF DESCRIPTION OF THE DRAWINGS

[0035] Figure 1 is a schematic diagram of the modules of an intelligent temperature control system for the whole process of cable production according to the present invention.

[0036] Figure 2 is a schematic flow chart of the operation process of the production process modeling module provided by an embodiment of the present invention.

[0037] Figure 3 is a schematic flow chart of the operation process of the temperature control decision module provided by an embodiment of the present invention. DETAILED DESCRIPTION OF THE EMBODIMENTS

[0038] Please refer to Figures 1 - 3 as shown, the present invention relates to an intelligent temperature control system for the whole process of cable production.

[0039] Embodiment

[0040] An intelligent temperature control system for the whole process of cable production, including a cable parameter acquisition module, a production process modeling module, a temperature control decision-making module, an execution control module, and a monitoring and warning module that are communicatively connected in sequence;

[0041] The cable parameter acquisition module is used to collect parameter information in the cable production process and construct a parameter database; the parameter information includes ambient temperature, raw material heat capacity, cable surface heat transfer coefficient, equipment temperature rise curve, and dynamic production parameters.

[0042] It should be noted that high-precision temperature sensors and thermocouples are arranged at key positions of the cable production equipment (such as extruders, cooling water tanks, wire drawing equipment, etc.) to collect dynamic production parameters such as ambient temperature, equipment temperature rise data, cable surface temperature, and cable internal temperature.

[0043] Equip high-precision temperature sensors, humidity sensors, heat flux sensors, and photoelectric sensors to collect information such as ambient temperature, raw material heat capacity, and cable surface heat transfer coefficient during the production process in real time.

[0044] Temperature sensors: Distributed at key nodes of the production line (such as the extruder outlet, the cooling tank inlet, and the cabling process) to achieve multi-point real-time monitoring.

[0045] Heat flux sensors: Monitor the dynamic characteristics of the heat exchange between the cable surface and the environment.

[0046] Dynamic parameter recording: Collect the temperature rise curve of the equipment (such as the extruder barrel, heater) and the production line speed and tension.

[0047] Preliminarily fuse and process the collected parameters through an edge computing device to achieve time synchronization and spatial alignment of multi-data sources, and ensure the consistency and accuracy of parameter data.

[0048] The construction of the parameter database is as follows:

[0049] Static parameters: Include physical properties such as the heat capacity of raw materials and the heat transfer coefficient of the cable surface, which are obtained through laboratory measurements and literature data correction.

[0050] Dynamic parameters: Include production line speed, raw material input, production environment temperature fluctuations, etc., which are obtained through a real-time data acquisition system (such as PLC or SCADA).

[0051] Data synchronization and storage: Use an edge computing device to synchronize multi-sensor data, and store the collected data into the parameter database through an industrial Internet of Things (IIoT) platform, supporting high-frequency (such as 100 Hz) data updates.

[0052] The production process modeling module is used to model the temperature transfer path and spatial distribution in the production process based on the parameter information by using a graph neural network, generate a dynamic temperature distribution prediction model for the entire production process, and extract the implicit physical laws in the temperature-quality relationship by using a symbolic regression algorithm;

[0053] Among them, the operation process of the production process modeling module includes the following steps:

[0054] Normalize the parameter information, remove outliers through a noise reduction algorithm, and use a graph neural network to construct a temperature transfer topology graph in the production equipment and cable production process, and capture the heat transfer path and its mutual relationship between each node in the production process;

[0055] Specifically, define a structured topology graph of the production equipment and the environment, including each production node (such as a heat source, a cooling device, a conveying device) and its connection relationship. Each node represents a key position in the production process, such as a cable preheating area, a forming area, a cooling area, etc.; the edge represents the temperature transfer path.

[0056] Node features: Collect the characteristics of each node such as temperature, heat flux density, and material thermal conductivity, and construct a multi-dimensional feature vector.

[0057] Edge features: Calculate the thermal resistance, heat flow direction, and time delay characteristics between nodes according to the heat conduction law.

[0058] Use a graph convolutional network (GCN) or a graph attention network (GAT) to input the temperature transfer topology graph into the network model. The network captures the local and global temperature transfer relationships in the production process through layer-by-layer transfer and aggregation operations, and generates a high-dimensional embedding representation.

[0059] Based on the temperature transfer topology graph, combined with dynamic production parameters, train the graph neural network model to generate a dynamic temperature distribution prediction model for different production stages, and predict the temperature change situation in the production process in real time;

[0060] Specifically, combine real-time production parameters (such as feeding speed, coolant flow rate, production time, etc.) as dynamic inputs. Use a time series encoding method (such as LSTM or time convolutional network TCN) to process the dynamic input information.

[0061] Define a loss function, such as the temperature prediction error based on the mean square error (MSE). Adopt a self-supervised learning strategy to automatically label data during the model training process to improve the training efficiency. Iteratively adjust the model weights through a gradient descent algorithm (such as the Adam optimizer) to ensure the efficient prediction performance of the model in each production stage.

[0062] Deploy the trained model to the production system, receive real-time data input, and generate temperature distribution prediction results for each production stage. The output includes a heat map of the node temperature distribution and the temperature change trend of key nodes.

[0063] Use the symbolic regression algorithm to analyze the temperature data in the prediction model, mine the implicit physical relationship between temperature changes and product quality, and generate a mathematical expression of the temperature-quality relationship.

[0064] Verify the generated temperature distribution prediction model and the extracted implicit physical laws through historical production data and experimental data, and optimize the model according to the verification results.

[0065] Specifically, use experimental data and production historical data to verify the generated mathematical expression, and evaluate its accuracy and universality. Feed the expression results back to the production system for quality prediction and process optimization. Compare the predicted temperature distribution of the model with the actual temperature data collected in production, and calculate the prediction error. Use metrics such as mean squared error (MSE) and mean absolute error (MAE) to evaluate the model performance. Through simulation experiments or trial production, collect the temperature and quality data of key nodes and conduct comparative analysis with the prediction results. According to the verification results, adjust the hyperparameters of the graph neural network (such as the number of layers, learning rate, activation function). Introduce more constraints or features in the symbolic regression algorithm to improve the practical application effect of the expression.

[0066] Furthermore, the formula of the dynamic temperature distribution prediction model is as follows:

[0067]

[0068] Among them, represents the instantaneous temperature distribution of the system at time t; represents the amount of heat change; m represents the mass of the cable or related node; represents the specific heat capacity of the cable material; , , and respectively represent the heat influence, heat conduction diffusion, external heat source and environmental benchmark influence; represents the Laplace operator of the temperature distribution; represents the input power of the external heat source, including the dynamic heat input of the production equipment.

[0069] The mathematical expression of the temperature-quality relationship is as follows:

[0070]

[0071] Among them, Indicates the product quality indicators during the cable production process; T represents the temperature of the production equipment and the cable surface; H Indicates the current heat in the system; Indicates the temperature change per unit time; t represents the production time; Indicates the reference temperature; Indicates the cable production process efficiency parameter; C represents the heat transfer coefficient of the cable surface; Indicates the temperature transfer time constant; 、 、 、 and Indicates the constant coefficients extracted by the symbolic regression algorithm; a and b represent the exponential parameters.

[0072] The temperature control decision-making module is used to generate an optimal temperature control strategy by using a reinforcement learning algorithm based on the dynamic temperature distribution prediction model and the extracted implicit physical laws;

[0073] Among them, the operation process of the temperature control decision-making module includes the following steps:

[0074] Input the dynamic temperature distribution data collected in real time during the production process and the extracted implicit physical laws into the temperature control decision-making module, and perform normalization and standardization processing based on historical data;

[0075] Based on the dynamic temperature distribution prediction model, use the reinforcement learning algorithm to define the objective function of the production process, including the constraints of maximizing product quality, minimizing temperature control energy consumption, and equipment operation stability;

[0076] Specifically, based on the dynamic temperature distribution prediction model, clarify the optimization objectives of the production process:

[0077] Maximizing product quality: Using the prediction model and implicit physical laws to ensure that the production process reaches the optimal quality state, such as conductivity, insulation performance, etc.

[0078] Minimizing temperature control energy consumption: Reducing the energy consumption of heating, cooling, and other temperature control measures, and improving the energy utilization efficiency of the production process.

[0079] Equipment operation stability: Reducing the temperature fluctuations during equipment operation, extending the equipment life, and reducing the failure rate.

[0080] According to the specific production requirements, allocate the weights of each optimization objective. For example, in the high-quality production stage, focus on product quality; in the energy-saving stage, pay more attention to minimizing energy consumption. Through the multi-objective optimization method, combine these objectives into a comprehensive decision-making criterion.

[0081] Generate a preliminary temperature control strategy through the state space and action space, and use the policy gradient algorithm for policy optimization during the training process;

[0082] Specifically, the state space is constructed by defining the state space of the production process, including: current dynamic temperature distribution data (such as temperatures at different nodes); current production parameters (such as production speed, coolant flow rate); and production equipment status (such as power level of heating equipment).

[0083] The action space is designed by defining the actions that the temperature control module can perform, such as: adjusting the power of the heating equipment (increasing or decreasing); changing the coolant flow rate or cooling time of the cooling equipment; and changing the production line speed to meet the temperature control requirements.

[0084] Initialize the reinforcement learning model (such as a deep Q-network or a policy gradient-based algorithm), and randomly generate a preliminary temperature control strategy based on the current state.

[0085] Utilize the dynamic temperature data and state space in the production process to train the reinforcement learning algorithm and gradually optimize the temperature control strategy:

[0086] Based on the results of the objective function, such as improvement in product quality, reduction in energy consumption, and improvement in equipment stability, give corresponding rewards or punishments. Use a policy gradient algorithm (such as PPO or REINFORCE) to optimize the parameters of the policy network. Achieve a dynamic balance between exploring new temperature control strategies and exploiting existing strategies through an ε-greedy policy or random perturbation. Continuously iterate the training, and through a large amount of simulation and actual production process data, continuously optimize the execution effect of the temperature control strategy. Dynamically adjust the learning rate and training parameters to avoid problems such as overfitting or slow policy updates.

[0087] Conduct a multi-objective optimization analysis on the generated temperature control strategy and output the optimal temperature control strategy.

[0088] Furthermore, the formula for the optimal temperature control strategy is as follows:

[0089]

[0090] Where, represents the optimal temperature control strategy; represents the temperature control energy consumption; represents the equipment operation stability function; represents the deviation between the temperature T of the production equipment and the target temperature ; , and represent the weight coefficients; represents the target temperature; T represents the temperature of the production equipment and the cable surface.

[0091] The execution control module is used to receive the temperature control strategy generated by the temperature control decision module, and convert the regulation strategy into an accurate control signal through an adaptive fuzzy control algorithm to drive the production equipment to complete the temperature adjustment; based on the adaptive fuzzy control algorithm, the execution control module adjusts the control rules and membership functions in real time, and continuously optimizes the control parameters by using the production feedback data.

[0092] It should be noted that the temperature control strategy is converted into an instruction signal executable by the equipment through a controller (such as a PLC or an industrial PC):

[0093] For heating equipment: convert the temperature control strategy into a PID signal output, such as controlling the input current of the heating resistor.

[0094] For the cooling system: adjust the opening of the cooling water valve, the flow rate of the coolant, or the refrigeration power.

[0095] For the traction equipment: dynamically adjust the production line speed to avoid overheating or insufficient cooling.

[0096] Signal interface:

[0097] Analog signal: Use a DAC module to convert the digital signal into an analog signal (such as a 4-20mA current).

[0098] Digital signal: Use industrial communication protocols such as Modbus or Profinet to send instructions to each device.

[0099] The design of the fuzzy control rules is as follows:

[0100] Fuzzy control establishes a fuzzy rule base by associating input variables (such as temperature deviation and deviation change rate) with output variables (such as heating power and cooling flow rate).

[0101] Example rules:

[0102] Rule 1: If the temperature deviation is large and the deviation change rate is fast, the heating power should increase rapidly.

[0103] Rule 2: If the temperature is close to the target value and the deviation change rate is small, the heating power should decrease gradually.

[0104] Membership functions of input variables: Temperature deviation: Divided into "large deviation", "medium deviation", and "small deviation". Deviation change rate: Divided into "rapid change", "medium-speed change", and "slow change". Membership functions of output variables (control signals): Heating power or cooling power: Divided into "high power", "medium power", and "low power".

[0105] Adjust the membership function based on real-time feedback data. For example, when the environmental temperature fluctuates violently, expand the range of the "medium deviation" membership function to enhance the system's response ability to medium deviations. Continuously learn the temperature control decision effect and dynamically adjust the fuzzy rule base through reinforcement learning algorithms. For example, when the cooling efficiency is lower than expected, increase the priority of the cooling flow response.

[0106] The heating equipment control method realizes precise temperature control of the extruder barrel and wire drawing equipment by adjusting the heating power (by adjusting the current or voltage). When the temperature deviation is large, quickly increase the power; when approaching the target value, gradually reduce the power to avoid overshoot.

[0107] The cooling system control method adjusts the cooling water flow rate, coolant temperature, or the operating power of the refrigerator. When the cooling water temperature rises, automatically increase the coolant flow rate or lower the cooling water temperature. When insufficient cooling causes the surface temperature of the cable to rise, give priority to adjusting the operating state of the cooling system.

[0108] The production line speed coordination control method dynamically adjusts the production line speed (such as the motor speed of the traction equipment) to maintain the temperature stability of the cable at all stages of production. When the temperature at the extruder outlet is too high, temporarily reduce the production line speed and extend the cooling time. After the temperature returns to normal, gradually increase the production line speed to ensure production efficiency.

[0109] The monitoring and warning module is used to monitor various temperature parameters in real time during the production process, detect potential anomalies using sparse coding and variational inference algorithms, judge whether there are abnormal situations, and issue warning signals when the temperature exceeds the set range or equipment failure occurs; the monitoring and warning module performs real-time analysis of production data based on a distributed computing architecture and realizes hierarchical push of remote warning information through an industrial Internet of Things platform.

[0110] Specifically, sparse coding is used to construct the feature space of normal temperature data. The specific process is as follows:

[0111] Use historical temperature data during normal production processes to train a model based on sparse coding and extract low-dimensional features of temperature parameters. During real-time monitoring, when the features of temperature data exceed the feature distribution range during training, they are marked as potential anomalies.

[0112] Example application: If the surface temperature of the cable suddenly rises and exceeds the sparse feature distribution, it may indicate the failure of the cooling system.

[0113] Use variational inference to estimate the probability distribution of real-time data and identify data that does not conform to the normal production state: Model the real-time temperature parameters as a probability distribution (such as a Gaussian distribution). When the temperature data deviates from the normal distribution (such as outside the mean + 3σ range), trigger anomaly detection.

[0114] Based on comprehensive multi-parameter analysis, the anomalies are classified into the following categories:

[0115] Minor anomaly: The temperature deviation is small but exceeds the set range, which may have a slight impact on product quality.

[0116] Major anomaly: The temperature deviation is large, which may cause abnormal operation of the equipment or unqualified products.

[0117] Emergency failure: The equipment fails (such as the cooling system stops working or the heater power is out of control), and immediate handling is required.

[0118] Based on the collected real-time data and combined with the anomaly detection results, multi-level warning rules are set:

[0119] Threshold warning: If a certain temperature parameter exceeds the set threshold (such as the target temperature ±5°C), a warning is triggered.

[0120] Trend warning: If the temperature change rate exceeds the preset value (such as the heating rate exceeds 2°C / s), a trend warning is issued in advance.

[0121] Composite warning: Combine multiple parameters (such as ambient temperature, equipment temperature rise curve, and cable surface temperature) to judge whether there is a comprehensive risk.

[0122] The types of warning signals are as follows:

[0123] Acoustic and optical alarm: At the production line site, the operator is reminded to pay attention through alarm lights and buzzers.

[0124] Remote notification: Push the warning information to the terminal devices (such as mobile phones, tablets) of relevant personnel through the Industrial Internet of Things (IIoT) platform.

[0125] Production line control: In case of emergency, automatically trigger the production line shutdown protection function, for example, stop heating when the extruder temperature is too high.

[0126] To sum up, through the integration of high-precision sensors, edge computing, graph neural networks, symbolic regression algorithms, and reinforcement learning algorithms, the system can achieve comprehensive data collection, analysis, and control from all dimensions of ambient temperature, equipment status to production parameters, construct an intelligent temperature control closed-loop management, and improve the automation level of the production process. By introducing a graph neural network to construct a topological graph of the temperature transfer path and combining dynamic production parameters, the system can generate a dynamic temperature distribution prediction model, greatly improving the accuracy of temperature prediction and avoiding the temperature control instability problem caused by prediction errors in traditional temperature control methods.

[0127] The present invention utilizes symbolic regression to extract implicit physical laws in the temperature-mass relationship, combines reinforcement learning algorithms to generate optimal temperature control strategies, can significantly reduce temperature control energy consumption while ensuring product quality, and optimize production energy efficiency. Based on sparse coding and variational inference algorithms, precise detection of potential anomalies in the production process is achieved. Combining a hierarchical early warning mechanism with the remote notification function of the industrial Internet of Things platform significantly improves the safety of equipment operation and the timeliness of fault response.

[0128] The present invention adopts an adaptive fuzzy control algorithm to convert the temperature control strategy into precise device control signals. By adjusting the membership function and fuzzy rules in real time, the system can adapt to complex requirements under different production conditions, ensuring the efficiency and precision of temperature control execution. Dynamically adjusting the production line speed, heating power, and cooling flow rate can quickly respond to temperature change requirements, reduce downtime and product rejection rates caused by overheating or insufficient cooling, while extending the service life of the equipment and reducing maintenance costs.

[0129] The above embodiments are only descriptions of the preferred embodiments of the present invention, and do not limit the scope of the present invention. Without departing from the design spirit of the present invention, various deformations and improvements made by those of ordinary skill in the art to the technical solutions of the present invention shall fall within the protection scope determined by the claims of the present invention.

Claims

1. An intelligent temperature control system for the whole process of cable production, characterized in that, It includes a cable parameter acquisition module, a production process modeling module, a temperature control decision module, an execution control module and a monitoring and early warning module which are sequentially connected in communication; The cable parameter acquisition module is used to collect parameter information during the cable production process and build a parameter database; The production process modeling module is used to model the temperature transfer path and spatial distribution in the production process based on the parameter information using a graph neural network, generate a dynamic temperature distribution prediction model for the entire production process, and use a symbolic regression algorithm to extract implicit physical laws in the temperature-quality relationship; The temperature control decision module is used to generate an optimal temperature control strategy using a reinforcement learning algorithm based on the dynamic temperature distribution prediction model and the extracted implicit physical laws; The execution control module is used to receive the temperature control strategy generated by the temperature control decision module, and convert the control strategy into a precise control signal through an adaptive fuzzy control algorithm to drive the production equipment to complete temperature regulation; The monitoring and early warning module is used to monitor various temperature parameters in the production process in real time, use sparse coding and variational inference algorithms to detect potential anomalies, determine whether there are abnormal conditions, and issue an early warning signal when the temperature exceeds the set range or the equipment fails; The operation process of the production process modeling module includes the following steps: The parameter information is normalized, and outliers are removed through a noise reduction algorithm. A graph neural network is used to construct a temperature transfer topology map of the production equipment and cable production process, capturing the heat transfer paths and their relationships between nodes in the production process; Based on the temperature transfer topology map and combined with dynamic production parameters, the graph neural network model is trained to generate a dynamic temperature distribution prediction model for different production stages, and the temperature changes in the production process are predicted in real time. The symbolic regression algorithm is used to analyze the temperature data in the prediction model, explore the implicit physical relationship between temperature change and product quality, and generate a mathematical expression of the temperature-quality relationship; The generated temperature distribution prediction model and the extracted implicit physical laws are verified through historical production data and experimental data, and the model is optimized based on the verification results.

2. The intelligent temperature control system for the whole process of cable production according to claim 1, characterized in that, The parameter information includes ambient temperature, raw material heat capacity, cable surface heat transfer coefficient, equipment temperature rise curve and dynamic production parameters.

3. The intelligent temperature control system for the entire process of cable production according to claim 1, wherein The formula of the dynamic temperature distribution prediction model is as follows: ； Among them, represents the instantaneous temperature distribution of the system at time t; represents the amount of heat change; m represents the mass of the cable or related node; represents the specific heat capacity of the cable material; , , and respectively represent the heat influence, heat conduction diffusion, external heat source and environmental reference influence; represents the Laplace operator of the temperature distribution; represents the external heat source input power, including the dynamic heat input of the production equipment.

4. An intelligent temperature control system for the whole process of cable production according to claim 1, characterized in that, The temperature-mass relationship mathematical expression is as follows: ； Among them, represents the product quality index during the cable production process; T represents the temperature of the production equipment and the cable surface; H represents the current heat in the system; represents the temperature change per unit time; t represents the production time; represents the reference temperature; represents the cable production process efficiency parameter; C represents the heat transfer coefficient of the cable surface; represents the temperature transfer time constant; , , , and represent the constant coefficients extracted by the symbolic regression algorithm; a and b represent the exponential parameters.

5. An intelligent temperature control system for the whole process of cable production according to claim 1, characterized in that, The operation process of the temperature control decision module includes the following steps: The dynamic temperature distribution data collected in real time during the production process and the extracted implicit physical laws are input into the temperature control decision module, and normalized and standardized based on historical data; Based on the dynamic temperature distribution prediction model, the objective function of the production process is defined using a reinforcement learning algorithm, including constraints such as maximizing product quality, minimizing temperature control energy consumption, and equipment operation stability. Generate a preliminary temperature control strategy through state space and action space, and use the policy gradient algorithm to optimize the strategy during training; Perform multi-objective optimization analysis on the generated temperature control strategy and output the optimal temperature control strategy.

6. An intelligent temperature control system for the whole process of cable production according to claim 1, characterized in that, The formula of the optimal temperature control strategy is as follows: ； Among them, represents the optimal temperature control strategy; represents the temperature control energy consumption; represents the equipment operation stability function; represents the deviation between the temperature T of the production equipment and the target temperature ; , and represent the weight coefficients; represents the target temperature; T represents the temperature of the production equipment and the cable surface.

7. An intelligent temperature control system for the whole process of cable production according to claim 1, characterized in that, The execution control module adjusts the control rules and membership functions in real time based on the adaptive fuzzy control algorithm, and continuously optimizes the control parameters by using the production feedback data.

8. An intelligent temperature control system for the whole process of cable production according to claim 1, characterized in that, The monitoring and warning module conducts real-time analysis of production data based on the distributed computing architecture, and realizes hierarchical push of remote warning information through the industrial Internet of Things platform.

Images