A continuous winding automatic control method for CICC type high temperature superconducting coil of a stellarator

Abstract

The application discloses a kind of CICC type high-temperature superconducting coil continuous winding automatic control method for star simulator, it is related to intelligent manufacturing field, it solves the technical problem of low forming precision and poor environmental adaptability in the continuous winding forming process of irregular shape coil;The application identifies target data, cable material and coil finished product shape;Establish parameter identification model, obtain the control parameter when coil firing forming based on parameter identification model;Establish the digital twin model of firing furnace, and the standard control parameter is obtained by environmental calibration to control parameter based on digital twin model;Standard control parameter is based on the operation of firing machine control;The application can improve the performance of finished coil and the reliability of equipment operation, so as to avoid the low efficiency, unstable precision and individual difference caused by traditional manual experience setting parameter.

Description

Technical Field

[0001] This invention relates to the field of intelligent manufacturing, specifically to an automatic control method for continuous winding of CICC-type high-temperature superconducting coils for stellarators. Background Technology

[0002] In modern industrial production, irregularly shaped coils, such as stator coils for special motors, special transformer coils, and special electromagnetic component coils for aerospace, are increasingly widely used because they can meet the spatial adaptability, electromagnetic performance optimization, and structural compactness requirements of specific equipment. However, their geometric contours are complex, often including non-circular curves, variable curvature arc segments, and three-dimensional spatial transitions. Traditional winding methods that rely on manual or semi-automatic equipment suffer from poor consistency, low production efficiency, and strong dependence on operating skills.

[0003] Currently, most automatic control methods for continuous winding of irregularly shaped coils suffer from low forming accuracy and poor environmental adaptability, which leads to unstable coil performance, poor equipment reliability, and affects product quality and production efficiency.

[0004] Therefore, the present invention provides an automatic control method for continuous winding of CICC-type high-temperature superconducting coils for stellarators, to solve the above-mentioned technical problems. Summary of the Invention

[0005] The present invention aims to solve at least one of the technical problems existing in the prior art; to this end, the present invention proposes an automatic control method for continuous winding of CICC-type high-temperature superconducting coils for stellarators, which is used to solve the technical problems of low forming accuracy and poor environmental adaptability in the continuous winding forming process of irregularly shaped coils.

[0006] To achieve the above objectives, a first aspect of the present invention provides an automatic control method for continuous winding of CICC-type high-temperature superconducting coils for stellarators, comprising: Identify target data, cable material, and finished coil shape; among which, target data includes furnace temperature and oxygen content; A parameter identification model is established, and the control parameters during coil firing are obtained based on the parameter identification model; a digital twin model of the firing furnace is established, and the control parameters are calibrated environmentally based on the digital twin model to obtain standard control parameters; among them, the control parameters include the winding inflection point position, the inflection point pressure, the cable feed speed, and the feed length; The firing machine is operated based on standard control parameters.

[0007] Preferably, the method for controlling the operation of the firing machine based on standard control parameters includes: When the control mode is local control, all motors can be jogged by the on-site operation buttons. When the control mode is the idle state in remote control, extract the winding inflection point position, inflection point pressure, cable feed speed and feed length from the standard control parameters. When the control mode is in manual mode under remote control, the spindle and slave axis are manually adjusted and positioned; the spindle is a virtual axis with a fixed position set manually, and the slave axis is the feed motor in the bending forming machine. When the control mode is in the automatic state of remote control, the winding inflection point position, inflection point pressure, cable feed speed and feed length are identified and input through the parameter setting interface of the host computer control platform. The conductor feeding machine is controlled to run based on the cable feed speed, and the bending forming machine is controlled to apply pressure at the corresponding position of the cable based on the winding inflection point position and inflection point pressure. When the control mode is in an error state in remote control, perform emergency handling operations.

[0008] Preferably, the emergency treatment operation includes: The firing machine is powered off, decoupled, and deenabled, and specific error messages are displayed. After the error is reset, it returns to an idle state and continues running from the breakpoint.

[0009] Preferably, the identification target data, cable material, and coil finished product shape include: The temperature inside the firing furnace is identified by a temperature sensor installed inside the furnace, and the oxygen content inside the furnace is identified by an oxygen sensor installed inside the furnace. The material of the cable being processed is extracted from the cable material database, and the preset shape of the finished coil is extracted from the coil shape database.

[0010] Preferably, the establishment of the parameter identification model includes: Extract the cable material, coil shape, and corresponding winding inflection point position, inflection point pressure, cable feed speed, and feed length from historical reference data. The historical reference data includes the cable material, coil shape, and winding inflection point position, inflection point pressure, cable feed speed, and feed length of coils that have been successfully fired and meet the qualification requirements. The qualification requirements are manually set. The cable material, coil shape, and corresponding winding inflection point position, inflection point pressure, cable feed speed, and feed length are integrated into several sets of training and testing data. The training data is used to train the artificial intelligence model, and the testing data is used to test the trained artificial intelligence model. The artificial intelligence model is then adjusted based on the testing results. The final result is a parameter recognition model with the cable material and coil shape as inputs and the winding inflection point position, inflection point pressure, cable feed speed, and feed length as outputs. The artificial intelligence model includes a BP neural network model and an RBF neural network model.

[0011] Preferably, the step of obtaining the control parameters during coil firing based on the parameter identification model includes: By inputting the cable material and the finished coil shape into the parameter recognition model, the winding inflection point position, inflection point pressure, cable feed speed, and feed length of the coil to be fired are obtained.

[0012] Preferably, establishing a digital twin model of the firing furnace includes: Extract the 3D model of the firing furnace, target data, and cable product information from the database; the cable product information includes the cable's appearance and physical characteristics. A cable model is constructed based on product information of the cable, an environment model is constructed based on target data, and a simulation model is constructed based on the 3D model of the firing furnace; the cable model, environment model and simulation model are combined to generate a digital twin model of the firing furnace.

[0013] Preferably, the step of obtaining standard control parameters by environmental calibration of control parameters based on a digital twin model includes: The simulated coil is fed into the firing furnace using a digital twin model of the furnace and control parameters. When the simulated coil shape differs from the preset coil shape, the control parameters are re-acquired until the simulated coil shape is identical to the preset coil shape. At this point, the corresponding control parameters are used as the standard control parameters.

[0014] Compared with the prior art, the beneficial effects of the present invention are: 1. This invention solves the technical problems of low forming accuracy and poor environmental adaptability in the continuous winding forming process of irregularly shaped coils by identifying target data, cable material, and finished coil shape; establishing a parameter identification model; obtaining control parameters during coil firing based on the parameter identification model; establishing a digital twin model of the firing furnace; performing environmental calibration on the control parameters based on the digital twin model to obtain standard control parameters; and controlling the firing machine to operate based on the standard control parameters. This invention can improve the performance of finished coils and the reliability of equipment operation.

[0015] 2. This invention, by introducing a systematic extraction, organization, and modeling training of historical reference data, realizes the intelligent prediction and generation of key process parameters in the continuous winding and forming process of irregularly shaped coils. It can make full use of the implicit experience and rules contained in the successful and compliant product examples in the past, and transform these experience data into quantifiable, reusable, and generalizable control parameter generation capabilities through artificial intelligence models. This avoids the inefficiency, unstable accuracy, and individual differences caused by the traditional manual reliance on experience to set parameters. Attached Figure Description

[0016] To more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0017] Figure 1 This is a schematic diagram of the operation steps of the present invention; Figure 2 This is a schematic diagram illustrating the operational steps for establishing the parameter recognition model in this invention; Figure 3 This is the linkage interface of the host computer control platform of the present invention; Figure 4 This is a software control flowchart of the host computer control platform of the present invention. Detailed Implementation

[0018] The technical solution of the present invention will be clearly and completely described below with reference to the embodiments. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0019] Please see Figure 1 The first aspect of this invention provides an automatic control method for continuous winding of CICC-type high-temperature superconducting coils for stellarators, comprising: Identify target data, cable material, and finished coil shape; among which, target data includes furnace temperature and oxygen content; A parameter identification model is established, and the control parameters during coil firing are obtained based on the parameter identification model; a digital twin model of the firing furnace is established, and the control parameters are calibrated environmentally based on the digital twin model to obtain standard control parameters; among them, the control parameters include the winding inflection point position, the inflection point pressure, the cable feed speed, and the feed length; The firing machine is operated based on standard control parameters.

[0020] It is worth noting that this invention achieves fully automated and intelligent control of the continuous winding process for irregularly shaped coils. First, by integrating target data, cable material characteristics, and coil shape parameters, a multi-dimensional input database is constructed to ensure precise adaptation of material properties and geometric features during winding, avoiding problems such as winding deformation and wire breakage caused by material differences or irregular shapes. Second, the parameter recognition model innovatively combines a parameter recognition model with a digital twin model of the firing furnace. Based on historical process data and real-time collected material and shape parameters, the parameter recognition model dynamically generates core control parameters such as winding inflection point position, pressure intensity, cable feed speed, and length. Simultaneously, the digital twin model is used to virtually simulate and calibrate environmental factors such as the temperature field and oxygen concentration distribution within the firing furnace, effectively eliminating the adaptation deviation of traditional empirical parameters in complex environments and enabling standard control parameters to respond in real-time to fluctuations in the furnace environment. This invention significantly improves the robustness and accuracy of parameters. Furthermore, by driving the actuator based on calibrated standard control parameters, it achieves closed-loop automation from data perception and parameter optimization to execution control, greatly reducing manual intervention. Especially for continuous winding of irregular coils, precise positioning of inflection points and dynamic adjustment of pressure ensure the coil's forming accuracy under complex geometric paths. Simultaneously, coordinated control of feed speed and length avoids quality defects caused by cable accumulation or insufficient tension. In addition, the introduction of a digital twin model not only supports the pre-optimization of control parameters but also reduces physical testing costs through virtual debugging. Combined with multi-source data fusion capabilities, it forms a continuous accumulation and iteration mechanism for process knowledge, providing rapid adaptation solutions for coil products of different specifications and materials, significantly improving production efficiency and product consistency. This invention, through a "data-driven - model optimization - precise execution" technical path, breaks through the technical bottlenecks of traditional irregular coil winding, which relies on manual experience, has low forming accuracy, and poor environmental adaptability. It achieves multi-factor coordinated control of material properties, geometry, and firing environment, providing a systematic solution for the automated mass production of high-precision, complex coil products.

[0021] The method used in this application to control the operation of a firing machine based on standard control parameters includes: When the control mode is local control, all motors can be jogged by the on-site operation buttons. When the control mode is the idle state in remote control, extract the winding inflection point position, inflection point pressure, cable feed speed and feed length from the standard control parameters. When the control mode is in manual mode under remote control, the spindle and slave axis are manually adjusted and positioned; the spindle is a virtual axis with a fixed position set manually, and the slave axis is the feed motor in the bending forming machine. When the control mode is in the automatic state of remote control, the winding inflection point position, inflection point pressure, cable feed speed and feed length are identified and input through the parameter setting interface of the host computer control platform. The conductor feeding machine is controlled to run based on the cable feed speed, and the bending forming machine is controlled to apply pressure at the corresponding position of the cable based on the winding inflection point position and inflection point pressure. When the control mode is in an error state in remote control, perform emergency handling operations.

[0022] Emergency handling procedures performed in this application include: The firing machine is powered off, decoupled, and deenabled, and specific error messages are displayed. After the error is reset, it returns to an idle state and continues running from the breakpoint.

[0023] It should be noted that the host computer control platform has a built-in communication system based on the ETH protocol and provides a large number of components for these systems and technologies, thus ensuring optimal communication between the touchscreen and the system and achieving a high degree of integration with the system's automated network. This platform provides a user-friendly interface for winding superconducting coils, divided into seven sub-interfaces: a welcome screen, a status parameter setting screen, an automatic operation screen, a manual operation screen, and a linkage screen; the linkage screen is as follows... Figure 3 As shown, each interface performs different functions according to requirements, simplifying operator operations to the greatest extent and improving operator efficiency. It also provides a complete human-machine interface and monitoring display, facilitating operator operation, real-time monitoring, and further processing. After the coil winding control system was built, simulated operation showed that: the switching between system states was normal, and the functions in different states worked normally; during automatic operation, it could automatically complete tasks according to the predetermined timing relationship, and all motors could run synchronously according to the predetermined position cam curve; during operation, various protection functions worked normally, and once a system error occurred, it could immediately complete the alarm display of the error state and implement corresponding protection measures; it realized the functions of power failure protection and breakpoint continuation. Once the system was powered off, it could promptly save the parameter data of each motor and the superconducting cable feed length data during operation. Upon power-up, the system could continue operation from the breakpoint without affecting coil winding.

[0024] It should be noted that the software control flowchart of the host computer control platform is as follows: Figure 4As shown, the main functions include: winding superconducting cables into specific shapes according to specific motor linkage relationships; dividing the continuous winding automatic control system into local control and remote control. In the local control mode, all motors are jogged by operating buttons on-site, while in the remote control mode, the system is divided into four states: idle, manual, automatic, and error. Different states perform specific functions and achieve seamless switching and interlocking between specific states; realizing one-button automatic operation of the entire continuous winding automatic control system for cable laying, straightening, bending, and winding; and implementing a series of protection functions such as speed protection, limit position protection, and torque protection to ensure the continuous and stable operation of large-scale production machinery.

[0025] It should be noted that the specific parameters of the automatic control system for continuous winding include: 1. In the continuous winding forming automatic control system, each motion motor is a synchronous servo motor supporting the ETH bus. The PLC uses an embedded controller with a multi-core processor. Hardware interfaces: It provides two independent Ethernet ports, four USB 2.0 ports, and one DV interface. The controller has multi-axis coupling and linkage functions, and can control up to 255 servo axes; the TW real-time core can achieve a task cycle of up to 50 microseconds to ensure the real-time performance and synchronization requirements of the servo axes. 2. The continuous winding forming automatic control system is equipped with a UPS and can recover data after an abnormal power outage; 3. The continuous winding forming automatic control system needs to control a total of 4 servo motors, including: one Huichuan motor for the rotary platform to control the rotation of the rotary platform; one Mitsubishi motor for the bending forming machine to be responsible for conductor feeding; and two Mitsubishi motors for bending forming. Equipped with a multi-turn absolute encoder, the maximum speed can reach 6000rpm; The black box function can save the servo data before and after an alarm to non-volatile memory; Torque fluctuation <0.5%, overall protection rating IP67, high shock resistance design, ensuring safety and durability under various working conditions; One speed reducer is used to control the deceleration of the rotary platform; One absolute encoder is used to measure the conductor feed length. Two linear displacement sensors are used to monitor the movement of the bending head. 4. The continuous winding automatic control system can track the coil winding trajectory in real time according to the conductor feed speed; 5. The continuous winding forming automatic control system can perform linkage control of each piece of equipment in the winding production line, as well as individual control of each piece of equipment in the winding production line; in the later stage of coil winding, when the conductor is completely separated from the conductor feeding system / straightening system, the automatic control system can perform shielding treatment on the corresponding system. 6. The continuous winding forming automatic control system uses one industrial computer to ensure stable performance and a user-friendly human-machine interface; 7. The control cabinet of the continuous winding automatic control system ensures good internal heat dissipation, has a reasonable internal layout, and the wire harnesses have corresponding numbers.

[0026] This application identifies target data, cable material, and finished coil shape, including: The temperature inside the firing furnace is identified by a temperature sensor installed inside the furnace, and the oxygen content inside the furnace is identified by an oxygen sensor installed inside the furnace. The material of the cable being processed is extracted from the cable material database, and the preset shape of the finished coil is extracted from the coil shape database.

[0027] Please see Figure 2 In this application, a parameter identification model is established, including: Extract the cable material, coil shape, and corresponding winding inflection point position, inflection point pressure, cable feed speed, and feed length from historical reference data. The historical reference data includes the cable material, coil shape, and winding inflection point position, inflection point pressure, cable feed speed, and feed length of coils that have been successfully fired and meet the qualification requirements. The qualification requirements are manually set. The cable material, coil shape, and corresponding winding inflection point position, inflection point pressure, cable feed speed, and feed length are integrated into several sets of training and testing data. The training data is used to train the artificial intelligence model, and the testing data is used to test the trained artificial intelligence model. The artificial intelligence model is then adjusted based on the testing results. The final result is a parameter recognition model with the cable material and coil shape as inputs and the winding inflection point position, inflection point pressure, cable feed speed, and feed length as outputs. The artificial intelligence model includes a BP neural network model and an RBF neural network model.

[0028] It is worth noting that this invention, by introducing a systematic extraction, organization, and modeling training of historical reference data, achieves intelligent prediction and generation of key process parameters during the continuous winding of irregularly shaped coils. It fully utilizes the implicit experience and patterns contained in previously successfully fired and compliant product examples, transforming this experiential data into quantifiable, reusable, and generalizable control parameter generation capabilities through an artificial intelligence model. This avoids the inefficiency, instability in accuracy, and individual differences inherent in traditional manual parameter setting based on experience. This method constructs training and validation sets by grouping data such as cable material, coil shape, corresponding winding inflection point position, pressure, feed speed, and length. This allows BP neural network and RBF neural network models to learn the optimal process matching relationship for different materials under different forming shapes during training, and to approximate complex nonlinear mapping relationships. This model, learned through multi-dimensional input-output relationships, can automatically derive the optimal winding control parameters when producing new products by only inputting the cable material and the target finished product shape, significantly shortening process debugging time. Meanwhile, the model's verification and iterative optimization process ensures the stability and accuracy of the prediction results, enabling the parameter identification model to not only adapt to conventional production scenarios but also possess a certain degree of self-adaptability to cope with real-world disturbances such as batch differences in raw materials and changes in equipment status. By establishing the parameter identification model, the automation and intelligence levels of the production line can be significantly improved, achieving consistency and traceability in batch production. Furthermore, in terms of control precision, it ensures accurate inflection point positioning, uniform force application, and smooth feed during the winding process, effectively reducing the scrap rate and improving the structural stability and electrical performance of the coil. Simultaneously, because the parameter identification model can continuously incorporate the latest historical data for retraining, it possesses dynamic evolution and continuous optimization characteristics. This means that as production data accumulates, the ability to identify and output control parameters will become increasingly accurate, forming a data-driven process closed loop, ultimately achieving efficient, stable, and low-cost automatic control for the firing of complex and irregular coils.

[0029] Specifically, the steps for testing the trained AI model using validation data and adjusting the AI ​​model based on the validation results are as follows: The cable material and coil shape from the inspection data are input into the trained AI model to obtain the corresponding winding inflection point position, inflection point pressure, cable feed speed, and feed length. These parameters are then compared with the corresponding parameters in the inspection data. If conditions one, two, and three are all satisfied, no parameter adjustment is needed, and the next set of inspection data is tested. If one or more of conditions one, two, and three are not satisfied, the corresponding parameters are adjusted until conditions one, two, and three are all satisfied before proceeding to the next set of inspection data. When the number of inspections meeting all three conditions accounts for 90% or more of the total inspection data, a parameter identification model is obtained, with the input being cable material and coil finished product shape, and the output being winding inflection point position, inflection point pressure intensity, cable feed speed, and feed length. Among them, condition one is that the positional distance between the corresponding winding inflection point position and inflection point pressure intensity and the corresponding winding inflection point position and inflection point pressure intensity in the inspection data is less than a distance threshold; condition two is that the difference between the corresponding cable feed speed and the corresponding cable feed speed in the inspection data is less than a speed difference threshold; condition three is that the difference between the corresponding feed length and the corresponding feed length in the inspection data is less than a length difference threshold. The distance threshold, speed difference threshold, and length difference threshold are all obtained manually.

[0030] It should be noted that the artificial intelligence model includes both BP neural network and RBF neural network models. The model type is selected based on the elastic modulus of the cable material: when the cable material is a high-elasticity material (elastic modulus E > 200 GPa, such as steel wire), the RBF neural network is used to improve the accuracy of nonlinear fitting; when the cable material is a low-elasticity material (E ≤ 200 GPa, such as copper or aluminum), the BP neural network is used. If there are mixed-material samples in the historical data, the final control parameters are output through a weighted fusion algorithm. The fusion formula is as follows: ; in, , These are the output parameters of the BP and RBF models, respectively. This represents the mean squared error of the model on the test data.

[0031] It should be noted that the BP neural network adopts a 3-layer structure: the number of neurons in the input layer is (cable material feature dimension + shape feature dimension, such as material features including parameters such as density and elastic modulus, and shape features including parameters such as number of inflection points and maximum curvature, with 13 neurons in the input layer); there are 2 hidden layers with 64 and 32 neurons respectively, and the activation function is ReLU; the number of neurons in the output layer is (3 × number of inflection points + 2, 3 coordinate dimensions × number of inflection points + feed speed and feed length), and the activation function is a linear function; during training, the Adam optimizer is used, with an initial learning rate of 0.001, 500 iterations, and a batch size of 32; the radial basis function of the RBF neural network is a Gaussian function, and the center vector is selected from the training data by the K-means clustering algorithm, with the number of cluster centers being 1 / 10 of the sample size.

[0032] In this application, control parameters for coil firing are obtained based on a parameter identification model, including: By inputting the cable material and the finished coil shape into the parameter recognition model, the winding inflection point position, inflection point pressure, cable feed speed, and feed length of the coil to be fired are obtained.

[0033] This application establishes a digital twin model of the firing furnace, including: Extract the 3D model of the firing furnace, target data, and cable product information from the database; the cable product information includes the cable's appearance and physical characteristics. A cable model is constructed based on product information of the cable, an environment model is constructed based on target data, and a simulation model is constructed based on the 3D model of the firing furnace; the cable model, environment model and simulation model are combined to generate a digital twin model of the firing furnace.

[0034] It is worth noting that the process of establishing a digital twin model of the firing furnace highly replicates the firing furnace equipment, cable products, and actual operating environment in the physical world within a virtual space. This achieves dynamic, integrated modeling of equipment structure, product characteristics, and environmental parameters. This process enables precise simulation, prediction, and optimization of the firing process without affecting actual production, thereby significantly reducing trial-and-error costs and production risks. By extracting the 3D model of the firing furnace, target data, and cable product information from the database, the model comprehensively covers the mechanical structure, temperature field, gas composition, and geometric and physical characteristics of the cables during the firing process, ensuring that the virtual model remains synchronously linked with the real equipment. This integrated digital twin construction allows for virtual experiments to simulate the molding effects of different cable materials and finished product shapes under various environmental conditions such as temperature and oxygen content before firing. This enables early prediction of potential structural deformation, material stress distribution, and winding accuracy issues, ensuring the accuracy of one-time molding in production. Digital twin models can perform real-time virtual verification and optimization of control parameters during the production process. By combining environmental and cable models, they can not only simulate the thermodynamic and fluid dynamic behavior inside the furnace, but also calculate detailed control requirements such as inflection point locations and force distribution through precise geometric interactions. This provides accurate environmental calibration for the initial control parameters output by the parameter identification model, enabling standardized control parameter generation. Compared to traditional calibration methods based on single-point measurements or experience-based judgments, digital twins can dynamically calculate on a global scale, ensuring that every adjustment is globally optimal and repeatable. Furthermore, this digital twin model is not only suitable for static analysis, but can also be dynamically mapped based on sensor data during production, maintaining a high degree of consistency between the virtual model and the actual operating state. This not only improves the real-time performance and accuracy of control, but also enhances robustness to equipment state fluctuations, environmental changes, and batch material variations. In the long term, the establishment of digital twins provides a foundation for closed-loop optimization of the entire production line, continuously accumulating comparative data between virtual simulation and actual production. This provides more accurate sample support for subsequent artificial intelligence training, thereby driving the continuous evolution of control strategies. This method allows companies to achieve full-process visualization, predictability, and optimization of the irregular coil firing process, as well as improve production efficiency, reduce scrap rates, extend equipment life, and optimize energy consumption.

[0035] It should be noted that the appearance characteristics in the cable product information include the shape and size of the cable, while the physical characteristics include the material, heat resistance and toughness of the cable.

[0036] It should be noted that: 1. The three-dimensional model of the firing furnace: a 1:1 scale geometric model is constructed based on SolidWorks, including the digital representation of the furnace chamber (dimensional error ≤0.5mm), heating elements (power density distribution error ≤2%), cable conveying mechanism (positioning accuracy ±0.02mm) and temperature sensor (sampling frequency 1kHz, measurement error ≤±1℃). The model file format is STL or STEP. 2. Environmental Model: Constructed based on computational fluid dynamics and heat transfer equations, including: Temperature field: The temperature distribution inside the furnace was simulated using the energy conservation equation (spatial resolution 1 mm³, time step 0.1 s). Oxygen concentration field: The oxygen partial pressure distribution was simulated based on the Navier-Stokes equation coupled with the diffusion equation (error ≤ ±0.5%). Thermal radiation: Consider the radiative heat transfer between the inner wall of the furnace and the surface of the cables (emissivity is set according to the furnace wall material, such as 0.6 for stainless steel). 3. Cable model: Based on multibody dynamics theory, including: Material properties: Elastic modulus (E), Poisson's ratio (ν), coefficient of thermal expansion (αT), and yield strength (σs) all change dynamically with temperature (e.g., copper has E=110GPa at 200℃, which is 15% lower than at room temperature). Geometric characteristics: cable diameter (error ≤ ±0.01mm), cross-sectional shape (circular / rectangular), surface roughness (Ra≤1.6μm); Mechanical behavior: An elastoplastic constitutive model (such as the Johnson-Cook model) is used to simulate the bending deformation (curvature radius error ≤ 0.1 mm) and creep effect at high temperature (strain rate ≤ 1e-5 / s) during the winding process.

[0037] In this application, standard control parameters are obtained by environmental calibration of control parameters based on a digital twin model, including: The simulated coil is fed into the firing furnace using a digital twin model of the furnace and control parameters. When the simulated coil shape differs from the preset coil shape, the control parameters are re-acquired until the simulated coil shape is identical to the preset coil shape. At this point, the corresponding control parameters are used as the standard control parameters.

[0038] It should be noted that when the simulated coil shape differs from the preset coil shape, the control parameters are re-acquired until the simulated coil shape is identical to the preset shape. The quantification of shape difference is as follows: The following two indicators are used to evaluate the deviation between the simulation result and the preset shape: Global error: The root mean square error (RMSE) of the analog coil and the three-dimensional point cloud of the preset shape is ≤0.05mm (calculated after registration by ICP algorithm, with a sampling point count of ≥1000 points / coil). Local error: The radius of curvature error at the critical inflection point is ≤0.03mm (curvature calculation formula κ=|dθ / ds|, where θ is the tangent angle and s is the arc length).

[0039] Some of the data in the above formula are calculated by removing dimensions and taking their numerical values. The formula is the closest to the real situation obtained by software simulation of a large amount of collected data. The preset parameters and preset thresholds in the formula are set by those skilled in the art according to the actual situation or obtained through simulation of a large amount of data.

[0040] Working principle of the invention: Identify target data, cable material, and finished coil shape; establish a parameter identification model, and obtain control parameters during coil firing based on the parameter identification model; establish a digital twin model of the firing furnace, and perform environmental calibration of the control parameters based on the digital twin model to obtain standard control parameters; control the firing machine to operate based on the standard control parameters.

[0041] The above embodiments are only used to illustrate the technical methods of the present invention and are not intended to limit it. Although the present invention has been described in detail with reference to preferred embodiments, those skilled in the art should understand that modifications or equivalent substitutions can be made to the technical methods of the present invention without departing from the spirit and scope of the technical methods of the present invention.

Claims

1. An automatic control method for continuous winding of CICC-type high-temperature superconducting coils for stellarators, characterized in that, include: Identify target data, cable material, and finished coil shape; among which, target data includes furnace temperature and oxygen content; A parameter identification model is established, and the control parameters during coil firing are obtained based on the parameter identification model; a digital twin model of the firing furnace is established, and the control parameters are calibrated environmentally based on the digital twin model to obtain standard control parameters; among them, the control parameters include the winding inflection point position, the inflection point pressure, the cable feed speed, and the feed length; The firing machine is operated based on standard control parameters. The establishment of the parameter identification model includes: Extract the cable material, coil shape, and corresponding winding inflection point position, inflection point pressure, cable feed speed, and feed length from historical reference data; wherein, the historical reference data includes the cable material, coil shape, and winding inflection point position, inflection point pressure, cable feed speed, and feed length of coils that have been successfully fired and meet the qualification requirements. The cable material, coil shape, and corresponding winding inflection point position, inflection point pressure, cable feed speed, and feed length are integrated into several sets of training and testing data. The training data is used to train the artificial intelligence model, and the testing data is used to test the trained artificial intelligence model. The artificial intelligence model is then adjusted based on the testing results. The final result is a parameter recognition model with the cable material and coil shape as inputs and the winding inflection point position, inflection point pressure, cable feed speed, and feed length as outputs. The artificial intelligence model includes a BP neural network model and an RBF neural network model. The standard control parameters obtained by environmental calibration of control parameters based on the digital twin model include: The simulated coil is fed into the firing furnace using a digital twin model of the furnace and control parameters. When the simulated coil shape differs from the preset coil shape, the control parameters are re-acquired until the simulated coil shape is identical to the preset coil shape. At this point, the corresponding control parameters are used as the standard control parameters.

2. The automatic control method for continuous winding of CICC-type high-temperature superconducting coils for stellarators according to claim 1, characterized in that, The operation of the firing machine based on standard control parameters includes: When the control mode is local control, all motors can be jogged by the on-site operation buttons. When the control mode is the idle state in remote control, extract the winding inflection point position, inflection point pressure, cable feed speed and feed length from the standard control parameters. When the control mode is in manual mode under remote control, the spindle and the slave axis are manually adjusted and positioned; the spindle is a virtual axis with a fixed position set manually, and the slave axis is the feed motor in the bending forming machine. When the control mode is in the automatic state of remote control, the winding inflection point position, inflection point pressure, cable feed speed and feed length are identified and input through the parameter setting interface of the host computer control platform. The conductor feeding machine is controlled to run based on the cable feed speed, and the bending forming machine is controlled to apply pressure at the corresponding position of the cable based on the winding inflection point position and inflection point pressure. When the control mode is in an error state in remote control, perform emergency handling operations.

3. The automatic control method for continuous winding of CICC-type high-temperature superconducting coils for stellarators according to claim 2, characterized in that, The emergency handling procedures include: The firing machine is powered off, decoupled, and deenabled, and specific error messages are displayed. After the error is reset, it returns to an idle state and continues running from the breakpoint.

4. The automatic control method for continuous winding of CICC-type high-temperature superconducting coils for stellarators according to claim 1, characterized in that, The identified target data, cable material, and coil finished product shape include: The temperature inside the firing furnace is identified by a temperature sensor installed inside the furnace, and the oxygen content inside the furnace is identified by an oxygen sensor installed inside the furnace. The material of the cable being processed is extracted from the cable material database, and the preset shape of the finished coil is extracted from the coil shape database.

5. The automatic control method for continuous winding of CICC-type high-temperature superconducting coils for stellarators according to claim 1, characterized in that, The control parameters obtained during coil firing based on the parameter identification model include: By inputting the cable material and the finished coil shape into the parameter recognition model, the winding inflection point position, inflection point pressure, cable feed speed, and feed length of the coil to be fired are obtained.

6. The automatic control method for continuous winding of CICC-type high-temperature superconducting coils for stellarators according to claim 1, characterized in that, The establishment of the digital twin model of the firing furnace includes: Extract the 3D model of the firing furnace, target data, and cable product information from the database; the cable product information includes the cable's appearance and physical characteristics. A cable model is constructed based on product information of the cable, an environment model is constructed based on target data, and a simulation model is constructed based on the 3D model of the firing furnace; the cable model, environment model and simulation model are combined to generate a digital twin model of the firing furnace.

Images