Control method and system for high-pressure glass fiber pipeline winding equipment

Abstract

The invention is suitable for the technical field of composite material manufacturing, and provides a high-pressure glass fiber pipeline winding equipment control method and system.The method comprises the steps that an upper computer automatically calculates and generates a digital winding process file according to input structural parameters and performance requirements of a target pipeline; a multi-axis motion controller analyzes the digital winding process file and generates a synchronous motion instruction set; when the tension closed-loop control unit executes the digital winding process file, independent and adjustable tension closed-loop control among different winding layers and constant tension closed-loop control in the layers are realized; the resin immersion control unit maintains the resin immersion amount of the fiber in a process setting range. Compared with separation of process planning and equipment execution in traditional winding equipment, the method has the advantages that a digital winding process file is automatically generated through an upper computer, a bottom layer control unit is directly driven, and a seamless digital chain from design to manufacturing is constructed.

Description

Technical Field

[0001] This invention relates to the field of composite material manufacturing technology, specifically a control method and system for high-pressure glass fiber pipeline winding equipment. Background Technology

[0002] In current industrial practice, the control system of high-pressure glass fiber pipeline winding equipment generally consists of a host computer, a programmable logic controller or motion control card, a servo drive system, and supporting actuators. The operation process typically relies on operators manually calculating or using basic software to preliminarily plan the winding line parameters based on design drawings, including winding angle, layer distribution, etc., and setting process variables such as fiber tension and resin impregnation. During equipment operation, the control system mainly achieves basic linkage between the mandrel rotation axis and the guide head translation axis to complete the spatial trajectory laying of the fiber. However, with the continuous improvement of requirements for pipeline pressure rating, operational reliability, and production economy in the energy and chemical industries, existing control methods have revealed multiple technical bottlenecks in achieving stable, efficient, and intelligent production of high-performance pipelines.

[0003] For example, existing systems rely heavily on operators' personal experience to convert process parameters. The process documents generated by the host computer and the control instructions of the underlying equipment lack close integration, failing to form a complete integrated digital process chain. This disconnect makes the parameter adjustment process cumbersome and inefficient, making it difficult to quickly respond to the flexible production needs of products with different specifications and performance requirements. At the same time, frequent human intervention introduces operational errors, affecting the consistency between product batches. When switching to small-batch customized products, operators need to repeatedly adjust the winding angle and tension parameters, which prolongs the production preparation time and is prone to process execution deviations due to differences in experience.

[0004] Therefore, how to provide an intelligent device control process driven by the device itself to address the high-frequency iteration of products is the technical problem that this invention aims to solve. Summary of the Invention

[0005] The purpose of this invention is to provide a control method and system for high-pressure glass fiber pipeline winding equipment to solve the problems mentioned in the background art.

[0006] To achieve the above objectives, the present invention provides the following technical solution: This invention provides a control method for a high-pressure glass fiber pipeline winding device. The method is applied to a control system including a host computer, a multi-axis motion controller, a servo drive system, a tension closed-loop control unit, and a resin impregnation control unit. The method includes the following steps: S1: The host computer automatically calculates and generates a digital winding process file containing parameters such as layer sequence, winding angle, yarn trajectory, tension reference value and resin content based on the input structural parameters and performance requirements of the target pipeline pipe through the built-in winding line algorithm library. S2: The multi-axis motion controller parses the digital winding process file, generates a synchronous motion instruction set based on the linkage mathematical model between the mandrel spindle and the guide head motion axis, and drives the servo drive system to accurately lay the impregnated fiber onto the mandrel surface according to a predetermined spatial trajectory. S3: When executing the digital winding process file, the tension closed-loop control unit calls the corresponding tension reference value according to the current winding layer sequence and collects the fiber tension signal in real time. By adjusting the output torque or yarn feeding speed of the yarn feeding mechanism, it realizes independent and adjustable tension closed-loop control between different winding layers and constant tension within the layer. S4: The resin impregnation control unit adjusts the set temperature of the impregnation tank constant temperature device and the gap pressure of the extrusion roller according to the resin content parameters in the digital winding process file, so that the resin impregnation amount of the fiber is kept within the process setting range.

[0007] As a further aspect of the present invention: in step S2, the linkage mathematical model is defined as follows: the ratio of the displacement increment of the guide wire head along the mandrel axis to the circumference at the current winding position of the mandrel is equal to the tangent of the current winding angle; the multi-axis motion controller uses the mandrel spindle rotation angle as a reference and calculates the target position of the guide wire head axially in real time through electronic cam table mapping.

[0008] As a further aspect of the present invention: the setting of the tension reference value in step S3 follows an interlayer decreasing strategy: during the winding process from the inner layer to the outer layer, the tension reference value of the outer layer winding is set to 0.92 to 0.98 times the tension reference value of the adjacent inner layer.

[0009] As a further aspect of the present invention: the method further includes step S5: online vision-assisted calibration; deploying an industrial camera near the guide wire head, triggering image acquisition of the core mold baseline or the edge of the wound layer at the starting or key section of each layer being laid; identifying the deviation between the actual laying position and the theoretical trajectory through an image processing algorithm, and feeding this deviation as a position compensation amount into the multi-axis motion controller in real time to dynamically correct the synchronous motion command set.

[0010] As a further aspect of the present invention: in the online vision-assisted calibration step, the position compensation amount is decomposed into fine-tuning amounts of the axial position and / or tangential sway angle of the guide wire head through a coordinate transformation module.

[0011] As a further aspect of the present invention, the method further includes step S6: self-evolutionary optimization of process parameters; establishing a data interface for communication with the production management system, and collecting and storing quality data including equipment control parameter sequences, environmental parameters, and corresponding pipeline finished product water pressure burst strength and ultrasonic test results over a long period of time; training a machine learning model based on historical data to establish a predictive relationship from control parameters to product quality; when producing new specifications of products or switching raw materials, the host computer calls the machine learning model, takes the maximization of target performance as the optimization objective, iteratively optimizes the core parameters in the digital winding process file, and outputs a recommended set of process parameters.

[0012] As a further aspect of the present invention: the machine learning model is a deep neural network or a gradient boosting decision tree model; the core parameters include at least the tension reference value of each layer, the winding speed, and the resin temperature; the optimization objective is to pursue Pareto optimality of material usage or production cycle while satisfying the minimum burst strength constraint.

[0013] The present invention also provides a control system for a high-pressure glass fiber pipeline winding equipment. The system is used to implement the aforementioned control method for the high-pressure glass fiber pipeline winding equipment. The system includes: The process planning module is used to execute step S1; A multi-axis motion control module is used to execute step S2; Tension coordination control module, used to execute step S3; The adhesive impregnation steady-state control module is used to execute step S4; And optionally, a vision-assisted calibration module for performing step S5; And optionally, a self-evolutionary optimization module for performing step S6.

[0014] Compared with the prior art, the beneficial effects of the present invention are: the present invention automatically generates digital winding process files through the host computer and directly drives the underlying control unit, thus constructing a seamless digital chain from design to manufacturing. This greatly reduces manual intervention, eliminates errors caused by reliance on experience and manual conversion, and enables flexible production of products of different specifications to respond quickly, which is extremely suitable for the current high-frequency product iteration scenario. Attached Figure Description

[0015] To more clearly illustrate the technical solutions in the embodiments of the present invention, 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.

[0016] Figure 1 A flowchart illustrating the control method for high-pressure glass fiber pipeline winding equipment.

[0017] Figure 2 This is a structural diagram of the control system for a high-pressure glass fiber pipeline winding equipment. Detailed Implementation

[0018] To make the technical problems to be solved, the technical solutions, and the beneficial effects of the present invention clearer, the present invention will be further described in detail below with reference to the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are merely illustrative of the present invention and are not intended to limit the present invention.

[0019] Figure 1 This is a flowchart illustrating a control method for a high-pressure glass fiber tubing winding device. In this embodiment of the invention, a control method for a high-pressure glass fiber tubing winding device is provided. The method is applied to a control system including a host computer, a multi-axis motion controller, a servo drive system, a tension closed-loop control unit, and a resin impregnation control unit. The method includes the following steps: S1: The host computer automatically calculates and generates a digital winding process file containing parameters such as layer sequence, winding angle, yarn trajectory, tension reference value and resin content based on the input structural parameters and performance requirements of the target pipeline pipe through the built-in winding line algorithm library. S2: The multi-axis motion controller parses the digital winding process file, generates a synchronous motion instruction set based on the linkage mathematical model between the mandrel spindle and the guide head motion axis, and drives the servo drive system to accurately lay the impregnated fiber onto the mandrel surface according to a predetermined spatial trajectory. S3: When executing the digital winding process file, the tension closed-loop control unit calls the corresponding tension reference value according to the current winding layer sequence and collects the fiber tension signal in real time. By adjusting the output torque or yarn feeding speed of the yarn feeding mechanism, it realizes independent and adjustable tension closed-loop control between different winding layers and constant tension within the layer. S4: The resin impregnation control unit adjusts the set temperature of the impregnation tank constant temperature device and the gap pressure of the extrusion roller according to the resin content parameters in the digital winding process file to keep the resin impregnation amount of the fiber within the process set range. In one embodiment of the technical solution of this invention, the host computer, based on the input structural parameters and performance requirements of the target pipeline, automatically calculates and generates a digital winding process file containing parameters such as layer sequence, winding angle, yarn trajectory, tension reference value, and resin content through a built-in winding pattern algorithm library; this achieves seamless integration between process design and equipment execution. For example, the winding pattern algorithm library can calculate the basic winding path of the fiber on the mandrel based on a preset geometric model and empirical formula. For a cylindrical mandrel, the algorithm can generate a series of helical trajectories. The tension reference value can be set to a constant value or set according to a simple rule of increasing or decreasing the number of layers. The resin content parameter can be set to a fixed percentage. These parameters are then organized into a structured electronic file, such as a table or a script file, for subsequent read by the control unit.

[0020] Then, the multi-axis motion controller parses the digital winding process file and, based on the linkage mathematical model between the mandrel spindle and the guide head's motion axis, generates a set of synchronous motion instructions to drive the servo drive system, ensuring that the impregnated fibers are precisely laid onto the mandrel surface along a predetermined spatial trajectory. The core of this step is ensuring the accuracy of fiber placement. For example, the linkage mathematical model can be a simple proportional relationship; for instance, for every fixed angle the mandrel rotates, the guide head moves a fixed distance axially. Based on this model, the multi-axis motion controller converts the yarn trajectory defined in the process file into real-time position instructions for the mandrel spindle and guide head's motion axis. These instructions are then sent to the servo drive system, which executes them precisely, guiding the impregnated fibers to be laid along the calculated path.

[0021] Furthermore, when executing the digital winding process document, the tension closed-loop control unit calls the corresponding tension reference value according to the current winding sequence and collects fiber tension signals in real time. By adjusting the output torque or yarn feeding speed of the unwinding mechanism, it achieves independent and adjustable tension closed-loop control between different winding layers while maintaining constant tension within each layer. This step aims to optimize the fiber tension distribution during the winding process, thereby affecting the mechanical properties of the final product. For example, the tension reference value can be selected from a preset tension value list based on the layer sequence information in the process document. The tension closed-loop control unit continuously monitors the actual fiber tension through tension sensors installed on the fiber path. When the actual tension deviates from the tension reference value corresponding to the current layer sequence, the control unit sends an adjustment signal to the unwinding mechanism. For example, it adjusts the output torque by changing the current or frequency of the drive motor, or changes the yarn feeding speed by controlling the rotation speed of the yarn feeding roller, so that the fiber tension approaches and is maintained at the reference value.

[0022] Specifically, the resin impregnation control unit, based on the resin content parameters in the digital winding process file, adjusts the set temperature of the impregnation tank temperature control device and the gap pressure of the extrusion rollers to maintain the resin impregnation amount of the fibers within the process setting range. This step ensures the quality of the resin impregnation of the fibers and has a direct impact on the performance of the composite material. For example, the resin impregnation control unit can set the target temperature of the impregnation tank temperature control device according to the resin content parameters specified in the process file. The temperature control device maintains the resin temperature near the set value through a heating or cooling system. Simultaneously, the gap pressure of the extrusion rollers can be adjusted mechanically or pneumatically to control the amount of resin carried away by the fibers as they pass through the extrusion rollers. Through the coordinated adjustment of these two parameters, a uniform and compliant resin impregnation amount can be ensured before winding.

[0023] In a preferred embodiment of the present invention, in step S2, the linkage mathematical model is defined as follows: the ratio of the displacement increment of the guide wire head along the mandrel axis to the circumference at the current winding position of the mandrel is equal to the tangent of the current winding angle; the multi-axis motion controller uses the mandrel spindle rotation angle as a reference and calculates the target position of the guide wire head axially in real time through electronic cam table mapping, so as to achieve nonlinear precision synchronization between the two axes.

[0024] The above describes a synchronous control process. The linked mathematical model precisely describes the geometric relationship between the axial movement of the guide head and the rotation of the mandrel during the winding process. Specifically, it equates the ratio of the incremental displacement of the guide head along the mandrel axis to the circumference of the mandrel at the current winding point to the tangent of the current winding angle. This definition ensures the geometric accuracy of the fiber placement trajectory and is the foundation for achieving precision winding. For example, for a cylindrical mandrel with a constant diameter, its circumference is fixed, and the change in the winding angle directly determines the rate of axial movement of the guide head; while for a conical or irregularly shaped mandrel, the circumference changes with the axial position, and the model can adaptively calculate the corresponding axial displacement. In the winding equipment, the rotation of the mandrel spindle is the core motion of the entire winding process. Setting its rotation angle as the reference means that the motion of all other driven axes (such as the guide head axis) will be synchronized with the real-time rotation angle of the mandrel spindle as a reference. This control strategy based on the spindle rotation angle can effectively cope with speed fluctuations or acceleration / deceleration that may occur during the winding process, ensuring that the relative relationship of the motion of each axis remains accurate at all times. The axial target position of the guide head is calculated in real time using an electronic cam meter. An electronic cam meter is an advanced motion control technology that maps a specific position of the main shaft (mandrel main shaft) to a specific target position of the slave shaft (guide head shaft) through a preset or real-time calculated functional relationship. This mapping relationship can be linear or non-linear, and is particularly suitable for the complex non-linear synchronization requirements between the guide head and mandrel in this application. The electronic cam meter can be pre-calculated based on the aforementioned linkage mathematical model and stored in the controller for real-time querying; alternatively, in some advanced systems, the cam curve can be dynamically adjusted or generated based on real-time process parameters. Achieving non-linear precision synchronization between the two axes refers to establishing a highly accurate linkage between the axial movement of the guide head and the rotation of the mandrel main shaft, capable of adapting to non-linear geometric relationships. Traditional linear synchronization often fails to meet the requirements of complex winding trajectories. However, by combining the aforementioned linkage mathematical model and the electronic cam meter, a precise non-linear correspondence can be maintained between the axial position and winding angle of the guide head during mandrel rotation, ensuring that the fiber is accurately laid according to a predetermined spatial trajectory and avoiding fiber stacking, gaps, or winding angle deviations caused by synchronization errors.

[0025] This invention refines the precision trajectory collaborative control step S2 in the control method for high-pressure glass fiber pipeline winding equipment, aiming to solve the problem of insufficient nonlinear synchronization accuracy between the guide head and the mandrel. Specifically, after the host computer generates a digital winding process file based on the structural parameters and performance requirements of the target pipeline, the multi-axis motion controller, when parsing this file, no longer relies solely on simple linkage relationships but introduces a precise linkage mathematical model. This model explicitly defines that the ratio of the displacement increment of the guide head along the mandrel axis to the circumference of the mandrel at its current winding position must be equal to the tangent of the current winding angle. This mathematical definition ensures the accuracy of the fiber placement trajectory from a geometric perspective. Based on this, the multi-axis motion controller uses the real-time rotation angle of the mandrel spindle as a reference and utilizes the advanced motion control mechanism of an electronic cam gauge to precisely map the rotation angle of the mandrel spindle to the target position along the guide head axis. The electronic cam gauge can handle complex nonlinear mapping relationships, thereby achieving highly precise and nonlinear synchronous motion between the guide head and the mandrel spindle. In this way, the multi-axis motion controller can generate a more precise set of synchronous motion instructions to drive the servo drive system, enabling the impregnated fibers to be precisely laid onto the mandrel surface according to a predetermined spatial trajectory. This precise trajectory coordination control effectively compensates for the insufficient fiber laying accuracy caused by the lack of specific mathematical models and precise synchronization mechanisms in traditional methods, laying a solid foundation for the accurate superposition of subsequent winding layers and the improvement of the final product performance.

[0026] As a preferred embodiment of the present invention, the setting of the tension reference value in step S3 follows an interlayer decreasing strategy: during the winding process from the inner layer to the outer layer, the tension reference value of the outer layer winding is set to 0.92 to 0.98 times the tension reference value of the adjacent inner layer, so as to optimize the residual stress distribution during the winding process.

[0027] In one example of the technical solution of this invention, "the setting of the tension reference value follows an interlayer decreasing strategy" means that during the multi-layer winding process, as the winding proceeds from the mandrel surface to the outer layers, the target tension value of the fiber winding in each layer is reduced relative to the target tension value of its inner adjacent layers. This strategy aims to influence the residual stress state of the composite material after curing by controlling the tension differences between different layers. For example, a linear decreasing method can be used, where the tension reference value decreases by a fixed proportion or a fixed value with each additional layer; or a non-linear decreasing method can be used, for example, determining the tension of each layer using an exponential decreasing or piecewise decreasing functional relationship based on material properties or layer distribution.

[0028] "The winding process from the inner layer to the outer layer" refers to the process in fiber winding molding where fiber tapes or yarn bundles are piled up layer by layer from the surface of the mandrel outwards until the preset wall thickness is reached. This is a typical winding sequence in the manufacture of high-pressure glass fiber pipelines, ensuring the continuity and integrity of the pipe wall.

[0029] The statement that "the outer layer tension reference value is set to 0.92 to 0.98 times the tension reference value of the adjacent inner layer" defines the specific range of interlayer tension reduction. This means that the outer layer tension reference value will be between 92% and 98% of the tension reference value of its immediately adjacent inner layer. For example, if the inner layer tension reference value is 100N, then its outer layer tension reference value will be between 92N and 98N. This ratio can be a fixed value, such as always being 0.95 times; or it can be a parameter that is dynamically adjusted within the range of 0.92 to 0.98 based on factors such as the diameter, wall thickness, fiber type, or resin system of the pipeline.

[0030] "Optimizing the residual stress distribution during the winding process" refers to the precise control of winding tension so that the residual stress inside the pipeline pipe (caused by material shrinkage, differences in thermal expansion coefficients, etc.) can be more evenly and reasonably distributed after curing, avoiding local stress concentration or excessive tensile / compressive stress, thereby improving the overall mechanical properties of the pipeline pipe, such as compressive strength, fatigue resistance and service life.

[0031] The technical solution of this invention introduces a layer-by-layer decreasing tension setting strategy into the above control method, which is closely integrated with the dynamic tension collaborative control function in step S3. This allows the tension closed-loop control unit to not only achieve independently adjustable tension control between different winding layers and constant tension control within each layer when executing the digital winding process file, but more importantly, these independently adjustable tension reference values ​​are set according to a preset optimization strategy. Specifically, during the winding process from the inner layer to the outer layer, by setting the tension reference value of the outer layer winding to 0.92 to 0.98 times the tension reference value of the adjacent inner layer, the residual stress generated inside the pipe wall due to factors such as resin curing shrinkage and fiber prestress can be effectively alleviated. This decreasing strategy can avoid excessive radial compressive stress or circumferential tensile stress on the inner layer after the outer layer fibers have cured, thereby making the stress distribution of the entire pipe wall more balanced and uniform. This refined tension management allows the fibers of each layer to more effectively cooperate in bearing the force when the pipeline is subjected to internal pressure, significantly improving the overall load-bearing capacity and reliability of the pipeline.

[0032] As a preferred embodiment of the present invention, the method further includes step S5: online vision-assisted calibration; deploying an industrial camera near the guide wire head, triggering image acquisition of the core mold baseline or the edge of the wound layer at the starting or key section of each layer being laid; identifying the deviation between the actual laying position and the theoretical trajectory through an image processing algorithm, and feeding this deviation as a position compensation amount into the multi-axis motion controller in real time to dynamically correct the synchronous motion command set.

[0033] In one example of the technical solution of this invention, online vision-assisted calibration refers to a technical means of using optical imaging equipment to monitor the winding area in real time during the winding production process, and obtaining actual winding state information through image analysis technology, thereby correcting the movement trajectory of the equipment. Its core lies in providing non-contact, high-precision position feedback to compensate for the shortcomings of traditional open-loop or encoder-based closed-loop control in dealing with external interference. For example, a vision system based on structured light or laser scanning can be used to accurately reconstruct the three-dimensional contour of the mandrel or the wound layer by projecting specific patterns or light rays onto the surface of the mandrel and acquiring its deformed image, thereby calculating the actual position of the fiber placement. Alternatively, a binocular stereo vision-based solution can be used, where two industrial cameras simultaneously acquire images from different angles, and the spatial coordinates of the target point are calculated using the principle of triangulation, achieving accurate measurement of the fiber placement position.

[0034] Deploying an industrial camera near the guide head aims to acquire high-resolution images of the winding point or its vicinity, enabling precise capture of the actual fiber placement. Proximity to the guide head reduces measurement errors, improves real-time performance, and ensures a direct correlation between image content and the winding process. The industrial camera can be fixedly mounted on the guide head body, moving synchronously with it to maintain a relatively fixed viewing angle and distance from the winding point. Alternatively, the industrial camera can be deployed via a separate robotic arm or sliding rail system, allowing it to automatically adjust its position and orientation based on changes in the winding progress or mandrel size to optimize image acquisition.

[0035] At the beginning or critical section of each layer laying, image acquisition of the mandrel baseline or the edge of the already wound layer is triggered to optimize the efficiency and resource utilization of visual inspection. The beginning section is a crucial stage for winding accuracy, directly affecting the laying of subsequent layers; critical sections may refer to areas with drastic changes in winding angle, complex geometry, or susceptibility to interference. Image acquisition of the mandrel baseline or the edge of the already wound layer provides a clear reference target for comparison with the theoretical trajectory. The triggering mechanism can be based on a winding layer counter or a specific angular position of the mandrel spindle. For example, image acquisition is automatically triggered when the number of winding layers reaches a preset value or the mandrel rotates to a specific angle. Simultaneously, a preset winding line algorithm can be used to automatically mark critical sections requiring visual calibration when generating process documents, and the host computer issues image acquisition commands when the equipment reaches these marked points.

[0036] Identifying the deviation between the actual placement position and the theoretical trajectory using image processing algorithms is the core intelligent component of visual calibration. Its function is to transform the acquired raw image data into deviation information that can be used for control. By analyzing the image, the actual position, width, and edge features of the fiber bundle can be accurately extracted and compared with the preset theoretical trajectory, thereby quantifying the deviation between the actual and theoretical values. For example, edge detection algorithms (such as Canny and Sobel operators) combined with Hough transform can be used to identify the edges of the fiber bundle or the mandrel baseline. Then, the actual placement trajectory is fitted using the least squares method and geometrically compared with the theoretical trajectory in the digital winding process file to calculate the positional deviation. Alternatively, deep learning-based image segmentation or object detection models can be used to perform pixel-level or region-level identification of the fiber bundle in the image, directly outputting the centerline or edge coordinates of the fiber bundle, which are then compared with the theoretical trajectory to obtain high-precision deviation data.

[0037] Feeding this deviation as a position compensation value into the multi-axis motion controller in real time is a key step in vision feedback closed-loop control. It transforms the deviation information detected by the vision system into compensation commands that the control system can understand and execute. Real-time feeding ensures the control system's response speed, allowing deviations to be corrected promptly and preventing error accumulation. Deviation data can be transmitted to the multi-axis motion controller via high-speed industrial Ethernet (such as EtherCAT, Profinet) or a dedicated serial communication interface (such as RS485). Upon receiving the compensation value, the controller superimposes it onto the original motion command. Alternatively, the processed deviation signal can be directly input to the external compensation input port of the multi-axis motion controller via analog or digital I / O interfaces, where the controller's internal interpolation algorithm automatically corrects the error.

[0038] Dynamic correction of the synchronous motion command set is the final execution stage of visual calibration. The multi-axis motion controller adjusts the original synchronous motion command set in real time based on the received position compensation. This dynamic correction ensures that the guide head can accurately place the fiber at the target position, maintaining high precision even in the presence of external interference or system errors. The multi-axis motion controller can employ a combination of feedforward and feedback control. The position compensation is used as a feedforward signal to directly correct the target position or speed command of the guide head's axial movement; simultaneously, the controller's internal servo closed-loop system continues high-precision position tracking. Furthermore, the controller can incorporate an adaptive compensation module that dynamically adjusts the compensation gain or compensation strategy based on historical deviation data and compensation effects to adapt to nonlinear error sources that may occur during the winding process.

[0039] This invention addresses the problem of actual fiber placement deviating from the theoretical trajectory during winding by introducing online visual-assisted calibration and establishing a real-time visual feedback mechanism, thus ensuring fiber placement accuracy. In the basic precision trajectory collaborative control (S2), the multi-axis motion controller, based on the digital winding process file and a mathematical model of the linkage between the mandrel spindle and the guide head's motion axis, generates a synchronous motion instruction set to drive the servo drive system, ensuring the impregnated fibers are precisely laid onto the mandrel surface along a predetermined spatial trajectory. However, due to unavoidable factors such as mandrel installation errors, mold thermal deformation, and fiber conveying slippage, the actual placement position may deviate from the theoretical trajectory. To address this, this application further deploys an industrial camera near the guide head, triggering image acquisition of the mandrel baseline or the edge of the wound layer at the beginning or key section of each layer's laying. The industrial camera captures images of the winding area in real time and transmits these images to the image processing module. The image processing algorithm analyzes the acquired images to accurately identify the actual fiber bundle placement position. Subsequently, this actual position is compared with the preset theoretical trajectory in the digital winding process file to calculate the precise deviation between the two. This deviation is quantified as a position compensation amount and fed into the multi-axis motion controller in real time with extremely low latency. Upon receiving this compensation amount, the multi-axis motion controller superimposes it onto the existing synchronous motion command set, dynamically adjusting the guide head's trajectory. For example, if the vision system detects that the fiber placement deviates from the theoretical trajectory in a certain direction, the controller immediately issues a correction command, causing the guide head to fine-tune in the opposite direction, thereby guiding the fiber back to the correct path. The entire process forms a high-speed, high-precision closed-loop control loop, ensuring that the actual fiber placement trajectory closely tracks the theoretical trajectory, effectively counteracting the influence of various interference factors. In this way, online vision-assisted calibration and the existing precision trajectory collaborative control (S2) form a close collaborative relationship. The existing precision trajectory collaborative control provides basic, high-precision motion planning and execution capabilities, while online vision-assisted calibration, on this basis, provides real-time, externally environmentally adaptive feedback correction capabilities. This combination enables the winding equipment not only to move according to a preset mathematical model but also to perceive and correct deviations in the actual production environment through its "eyes," thereby raising fiber placement accuracy to a new level.

[0040] In a preferred embodiment of the present invention, in the online vision-assisted calibration step, the position compensation amount is decomposed into a fine adjustment amount for the axial position and / or tangential sway angle of the guide head through a coordinate transformation module, thereby achieving comprehensive compensation for core mold installation error, mold thermal deformation and fiber conveying slippage.

[0041] In one embodiment of the technical solution of this invention, in the above-mentioned online vision-assisted calibration step, an industrial camera acquires images of the mandrel baseline or the edge of the wound layer near the guide head. After the image processing algorithm identifies the deviation between the actual placement position and the theoretical trajectory, the quantity obtained for correcting the synchronous motion command set is the position compensation amount. This amount is usually expressed as a deviation value in a two-dimensional image coordinate system or a displacement vector in a spatial coordinate system, reflecting the difference between the actual fiber placement position and the ideal position. The coordinate transformation module is a functional unit whose function is to transform data from one coordinate system to another. In this application, it is responsible for transforming the position compensation amount identified by the vision system from the image coordinate system or sensor coordinate system to the motion control coordinate system of the winding device, and further decomposes it into specific adjustment amounts for the guide head motion axis. As one implementation, the coordinate transformation module can be a software algorithm module integrated inside a multi-axis motion controller, which processes visual feedback data in real time through a preset geometric model and transformation matrix. As another implementation, the coordinate transformation module can be a standalone hardware processing unit, such as a dedicated module based on an FPGA (Field-Programmable Gate Array) or DSP (Digital Signal Processor), capable of high-speed parallel processing of complex coordinate transformation calculations to meet real-time control requirements. Decomposing the fine-tuning of the axial position and / or tangential sway angle of the guide head refers to breaking down a single position compensation amount into two or more independent, controllable motion parameters. Specifically, the axial position fine-tuning of the guide head refers to the minute displacement adjustment of the guide head along the mandrel spindle direction (i.e., the Z-axis direction of the winding device). This adjustment is mainly used to compensate for deviations along the length of the mandrel, such as axial dimensional changes caused by mandrel misalignment, thermal expansion or contraction of the mandrel. The tangential sway angle fine-tuning refers to the minute angle adjustment of the yarn exit point relative to the tangential direction of the mandrel surface during winding. This adjustment is mainly used to compensate for lateral offset or slippage that may occur during fiber placement, such as fiber path deviation caused by fiber tension fluctuations, frictional changes, or minor vibrations of the guide head itself. The use of "and / or" implies that in practical applications, one can choose to adjust only the axial position, only the tangential yaw angle, or both, depending on the specific error type and compensation requirements.

[0042] This invention addresses the problem of insufficient compensation accuracy in traditional methods by refining the position compensation amount obtained during the online vision-assisted calibration process. When the online vision-assisted calibration system identifies the deviation between the actual fiber placement position and the theoretical trajectory and generates a position compensation amount, this amount is first sent to the coordinate transformation module. This module does not simply feed the compensation amount directly back to the multi-axis motion controller; instead, it performs in-depth analysis and transformation of the position compensation amount based on a preset equipment geometric model and kinematic relationships. Through the coordinate transformation module, the single, comprehensive position compensation amount is intelligently decomposed into two or more independent fine-tuning commands targeting the guide head's motion axis: fine-tuning of the guide head's axial position and / or fine-tuning of the tangential sway angle. This decomposition is based on a deep understanding of the influence mechanism of different error sources (such as mandrel installation errors, mold thermal deformation, and fiber conveying slippage) on the fiber placement trajectory. For example, axial deviations in mandrel installation or thermal expansion and contraction of the mold mainly manifest as an overall offset of the fiber placement trajectory along the mandrel axis. In this case, axial position fine-tuning plays a dominant role in compensation. Slight slippage of the fiber at the guide head or lateral swaying caused by uneven tension is primarily corrected by precise tangential sway angle fine-tuning. Subsequently, these decomposed and transformed fine-tuning values ​​are fed into the multi-axis motion controller in real time. When executing its synchronous motion command set generated based on a linkage mathematical model, the multi-axis motion controller superimposes these fine-tuning values, thereby dynamically correcting the actual motion trajectory of the guide head. This refined compensation mechanism allows for targeted and comprehensive compensation for various interference factors such as mandrel installation errors, mold thermal deformation, and fiber conveying slippage, rather than a general overall correction. Compared to simply feeding the position compensation value directly into the controller as a whole deviation, the technical solution of this invention can more accurately locate and eliminate errors, significantly improving the accuracy of the impregnated fiber placement onto the mandrel surface along a predetermined spatial trajectory, thus ensuring the winding quality and performance consistency of the high-pressure glass fiber pipeline.

[0043] As a preferred embodiment of the present invention, the method further includes step S6: self-evolution optimization of process parameters; establishing a data interface for communication with the production management system, and collecting and storing quality data including equipment control parameter sequences, environmental parameters, and corresponding pipeline finished product water pressure burst strength and ultrasonic test results over a long period of time; training a machine learning model based on historical data to establish a predictive relationship from control parameters to product quality; when producing new specifications of products or switching raw materials, the host computer calls the machine learning model, takes the maximization of target performance as the optimization objective, iteratively optimizes the core parameters in the digital winding process file, and outputs a recommended set of process parameters.

[0044] In one embodiment of the technical solution of this invention, establishing a data interface for communication with the production management system aims to achieve real-time and accurate transmission and integration of production process data. This data interface can employ widely used communication protocols in the industrial field, such as OPC UA or MQTT protocols, or interact with Enterprise Resource Planning (ERP) systems or Manufacturing Execution Systems (MES) through customized API interfaces, ensuring seamless access to the data analysis platform for equipment data from the production site.

[0045] The long-term collection and storage of quality data, including equipment control parameter sequences, environmental parameters, and corresponding pipeline finished product hydraulic burst strength and ultrasonic test results, is aimed at building a comprehensive historical production database. The equipment control parameter sequences can include winding speed, tension benchmark values ​​for each layer, impregnation tank temperature, and extrusion roller gap; environmental parameters can cover workshop temperature and humidity; while the quality data is automatically acquired through integrated hydraulic burst testing equipment such as hydraulic flaw detectors, or manually entered, and is associated with production batches, timestamps, and other information, stored in a relational database or time-series database, providing a rich data foundation for subsequent machine learning model training.

[0046] The core of intelligent optimization lies in training machine learning models based on historical data to establish predictive relationships between control parameters and product quality. This process utilizes a large amount of collected historical production data and supervised learning algorithms, such as deep neural networks, gradient boosting decision tree models, or support vector machines, to learn and construct a complex nonlinear mapping relationship between control parameters in the winding process and the final product quality (such as hydraulic burst strength and defect rate). This model can quantify the impact of different combinations of control parameters on product quality, thereby predicting product performance under given parameters.

[0047] When producing new specifications or switching raw materials, the host computer invokes this machine learning model to iteratively optimize the core parameters in the digital winding process file, with the goal of maximizing target performance. This means that when faced with new production demands or changes in conditions, the host computer does not need to rely on human experience for trial and error. Instead, it uses built-in optimization algorithms (such as genetic algorithms, particle swarm optimization algorithms, or Bayesian optimization algorithms) combined with a trained machine learning model to automatically explore and evaluate different combinations of core parameters under preset product performance goals (such as maximizing burst strength, minimizing material consumption, or shortening the production cycle). Through repeated prediction and adjustment, the system can efficiently find the optimal set of parameters that meets or exceeds the target performance.

[0048] Finally, the system outputs a recommended set of process parameters. This set of parameters is the optimal combination of production parameters optimized by a machine learning model. It can be directly distributed to the underlying actuators such as multi-axis motion controllers, tension closed-loop control units, and resin impregnation control units in the form of updated digital winding process files to guide actual production, thereby achieving intelligent adjustment of process parameters and closed-loop optimization of the production process.

[0049] This invention establishes a data interface to achieve comprehensive collection and storage of production data. Based on this data, a machine learning model is trained, building a predictive capability from control parameters to product quality. When production conditions change, the host computer can intelligently invoke this model, guided by preset performance targets, to iteratively optimize the core parameters in the digital winding process file and output the optimal recommended set of process parameters. This transforms the entire winding production process from traditional experience-driven to data-driven intelligent optimization.

[0050] In a preferred embodiment of the present invention, the machine learning model is a deep neural network or a gradient boosting decision tree model; the core parameters include at least the tension reference value of each layer, the winding speed, and the resin temperature; the optimization objective is to pursue Pareto optimality of material usage or production cycle while satisfying the minimum burst strength constraint.

[0051] In one embodiment of the technical solution of this invention, the machine learning model can be a deep neural network or a gradient boosting decision tree model. A deep neural network is a neural network containing multiple hidden layers, capable of extracting and predicting features from high-dimensional input data by learning complex data patterns and nonlinear relationships. Its implementation can include, but is not limited to, architectures such as feedforward neural networks, convolutional neural networks, or recurrent neural networks, trained using backpropagation algorithms and optimizers to minimize prediction errors. A gradient boosting decision tree model is an ensemble learning method that iteratively trains a series of weak predictors (usually decision trees) and accumulates their results to gradually reduce prediction errors. Its implementation can include, but is not limited to, frameworks such as XGBoost, LightGBM, or CatBoost, which are optimized for efficiency and performance, capable of handling large-scale datasets and effectively preventing overfitting. Both models possess strong nonlinear fitting capabilities and the ability to model complex multivariate relationships, accurately capturing the complex mapping relationship between winding process parameters and product quality.

[0052] The core parameters refer to process variables that decisively influence the final performance and production efficiency of high-pressure fiberglass pipelines. Among them, the tension reference value for each layer is the set value controlling the tension on the fibers when laid on the mandrel; it directly affects the fiber density, interlayer bonding strength, and residual stress distribution after pipe curing. Precise control of the tension reference value for each layer is crucial for optimizing the mechanical properties of the pipe. Winding speed refers to the relative speed of the mandrel rotation and the guide head movement; it determines production efficiency and the uniformity of fiber laying. A reasonable winding speed ensures stable fiber laying, avoiding fiber accumulation or excessive stretching. Resin temperature is a key parameter affecting resin viscosity, wetting properties, and curing reaction rate. Appropriate resin temperature ensures sufficient fiber wetting, forming a uniform composite material, and controls the curing process, thereby affecting the mechanical properties and appearance quality of the final product.

[0053] The optimization objectives refer to the performance indicators to be achieved during the optimization of process parameters. The minimum burst strength constraint is a mandatory quality requirement that the product must meet to ensure that the pipeline will not fail due to excessive pressure during use. This is a fundamental guarantee of product safety and reliability. Material usage refers to the total amount of fiber and resin required to produce a unit length of pipeline; optimizing material usage aims to reduce production costs and improve resource utilization. Production cycle time refers to the time required to complete the winding of a unit length of pipeline; optimizing the production cycle aims to improve production efficiency and shorten delivery time. Pareto optimality is a multi-objective optimization concept that seeks a balance among multiple conflicting objectives, such that no objective can be improved without sacrificing any of them. Here, it means seeking the optimal trade-off between material usage and production cycle time while ensuring that the product's burst strength meets the requirements; for example, minimizing material usage while also minimizing the production cycle time, or vice versa.

[0054] This invention constructs a data-driven, self-evolving optimization closed loop by introducing an advanced machine learning model, clarifying key process parameters, and setting a multi-objective optimization strategy. In step S6 of the self-evolving optimization of process parameters, a data interface is established with the production management system to collect and store long-term data on equipment control parameter sequences, environmental parameters, and the quality data of the finished pipeline pipe's hydraulic burst strength and ultrasonic testing results. This historical data forms the basis for training the machine learning model. To more accurately capture the complex nonlinear relationship between process parameters and product quality, this application specifically specifies that the machine learning model can be a deep neural network or a gradient boosting decision tree model. These models can learn from massive amounts of historical data and establish predictive relationships from control parameters to product quality. For example, they can predict performance indicators such as burst strength, material usage, and production cycle of the pipeline pipe under a given combination of core parameters such as tension benchmark values ​​for each layer, winding speed, and resin temperature. When it is necessary to produce new specifications of products or switch raw materials, the host computer no longer relies solely on experience or trial and error but instead calls upon the pre-trained machine learning model. At this point, the optimization objective is set to achieve Pareto optimality in material usage or production cycle while meeting the preset minimum burst strength constraint. This means the system intelligently explores different combinations of core parameters and uses models to predict the impact of these combinations on burst strength, material usage, and production cycle. Through a multi-objective optimization algorithm, the system can identify a set of Pareto optimal process parameters, that is, the best balance between ensuring product quality (burst strength) and simultaneously considering production cost (material usage) and production efficiency (production cycle). Finally, the host computer outputs the recommended set of process parameters, which will be used to generate a digital winding process file to guide subsequent steps such as precision trajectory collaborative control, dynamic tension collaborative control, and steady-state control of the impregnation process. This mechanism transforms the entire winding equipment control method from traditional experience-driven or static optimization to data-driven dynamic adaptive optimization. By using deep neural networks or gradient boosting decision tree models to accurately model complex processes, and by comprehensively considering key parameters such as tension reference values ​​for each layer, winding speed, and resin temperature, combined with Pareto optimal multi-objective optimization, the system can more intelligently generate and adjust process parameters, thereby significantly improving production efficiency and economic benefits while ensuring product quality.

[0055] As a preferred embodiment of the technical solution of the present invention, an intelligent control system for a high-pressure glass fiber pipeline winding device is also provided, for implementing the control method described in any of the above embodiments, the system comprising: The process planning module is used to execute step S1; A multi-axis motion control module is used to execute step S2; Tension coordination control module, used to execute step S3; The adhesive impregnation steady-state control module is used to execute step S4; And optionally, a vision-assisted calibration module for performing step S5; And optionally, a self-evolutionary optimization module for performing step S6.

[0056] In one example of the technical solution of this invention, the process planning module is a functional unit used to convert the structural parameters and performance requirements of high-pressure glass fiber pipelines into digital winding process documents. Its concept lies in providing a unified interface and computing core to achieve automated conversion from product design to production instructions. Possible implementation methods include: using an industrial control computer or dedicated workstation, running professional CAD / CAM software integrated with a winding line type algorithm library, receiving user input through a graphical user interface and generating process documents; or using an embedded system or high-performance programmable logic controller, pre-configuring multiple winding line type algorithms, receiving product parameters from external systems (such as MES systems) through a data interface, and automatically generating digital process documents.

[0057] The multi-axis motion control module is responsible for parsing the digital winding process file and generating a synchronous motion instruction set based on the linkage mathematical model between the mandrel spindle and the guide head motion axis. This drive system ensures that the impregnated fibers are precisely laid onto the mandrel surface along a predetermined spatial trajectory. Its function is to achieve high-precision, high-synchronization control of the equipment's moving parts. Possible implementation methods include: using a high-performance multi-axis motion controller, such as a dedicated controller based on a digital signal processor or field-programmable gate array, to communicate at high speed with the servo drive system via real-time Ethernet; or it can be implemented using an industrial PC paired with a professional motion control card, using software algorithms for complex motion trajectory planning, interpolation, and electronic cam functions.

[0058] The tension coordination control module is used to call the corresponding tension reference value according to the current winding layer sequence when executing the digital winding process file, and to collect fiber tension signals in real time. By adjusting the output torque or yarn feeding speed of the yarn unloading mechanism, it achieves independent and adjustable tension closed-loop control between different winding layers and constant tension within each layer. Its core lies in the refined and dynamic management of fiber tension. Possible implementation methods include: consisting of an independent tension controller, a high-precision tension sensor (such as a weighing sensor or piezoelectric sensor), and a high-response actuator (such as a magnetic powder brake or a servo motor-driven yarn unloading roller), and achieving closed-loop adjustment through PID or other advanced adaptive control algorithms; or the tension control function can be integrated into the main motion controller or programmable logic controller, and data interaction and control with the tension sensor and actuator can be achieved through an analog input / output module.

[0059] The resin impregnation steady-state control module maintains the resin impregnation amount of the fiber within the process setting range by adjusting the set temperature of the impregnation tank temperature control device and the gap pressure of the extrusion rollers, based on the resin content parameters in the digital winding process file. Its function is to ensure the stability and consistency of the resin impregnation process. Possible implementation methods include: consisting of a temperature sensor, heating / cooling unit, pressure sensor, extrusion roller gap adjustment mechanism (such as a stepper motor or servo motor drive), and an independent temperature / pressure controller, maintaining the set parameters through feedback control; or a control unit based on a programmable logic controller or microcontroller, controlling the heater, cooling valve, and extrusion roller driver through analog output and receiving sensor feedback signals.

[0060] The vision-assisted calibration module is used during the winding process to acquire images of the mandrel baseline or the edge of the wound layer using an industrial camera deployed near the guide head. Image processing algorithms identify the deviation between the actual placement position and the theoretical trajectory, and this deviation is fed into the multi-axis motion control module in real time as a position compensation amount to dynamically correct the synchronous motion command set. Its function is to provide online, real-time trajectory correction capabilities. Possible implementation methods include: incorporating a high-resolution industrial camera, image acquisition card, embedded vision processing system, or high-performance industrial PC, running image processing algorithms based on edge detection, template matching, or feature point recognition; or using a smart camera integrated with a deep learning inference engine, which directly identifies placement deviations and outputs compensation amounts through a trained neural network model.

[0061] The self-evolutionary optimization module establishes a data interface for communication with the production management system. It continuously collects and stores quality data including equipment control parameter sequences, environmental parameters, and corresponding pipeline finished product water pressure burst strength and ultrasonic test results. Based on historical data, it trains a machine learning model to establish a predictive relationship between control parameters and product quality. When producing new specifications or switching raw materials, this machine learning model is invoked to iteratively optimize the core parameters in the digital winding process file. Its function is to achieve self-learning and continuous optimization of process parameters. Possible implementation methods include: consisting of a data acquisition gateway, an industrial database, and a data analysis server (running machine learning frameworks such as TensorFlow and PyTorch in programming languages ​​like Python and R), interacting with the host computer and MES system; or using a cloud-based industrial IoT platform, preprocessing data through edge computing devices, and performing large-scale model training and optimization in the cloud.

[0062] The intelligent control system for the high-pressure glass fiber tubing winding equipment described above, through modular design, decomposes the complex winding process control into multiple collaborative professional functional modules, thereby achieving comprehensive and refined management of the winding process. The entire system operation begins with the process planning module. Based on product design requirements, this module utilizes a built-in winding line algorithm library to automatically generate a digital winding process document containing parameters such as layer sequence, winding angle, yarn trajectory, tension reference value, and resin content. This digital document serves as the foundation and basis for subsequent system control, digitizing and standardizing the traditional process planning process that relies on manual experience, effectively solving the problem of disconnect between process planning and equipment execution. Subsequently, this digital process document is jointly parsed and executed by the multi-axis motion control module, tension collaborative control module, and impregnation steady-state control module. The multi-axis motion control module, based on the linkage mathematical model between the mandrel spindle and the guide head motion axis, transforms the yarn trajectory in the digital process document into a precise set of synchronous motion instructions, driving the servo drive system to ensure that the impregnated fiber can be accurately laid on the mandrel surface according to the predetermined spatial trajectory. This guarantees precise reproduction of the winding trajectory, overcoming the problem of insufficient trajectory accuracy in traditional control. Meanwhile, the tension coordination control module, based on the tension benchmark value set in the digital process file and acquiring fiber tension signals in real time, adjusts the yarn feeding mechanism through closed-loop control to achieve independently adjustable tension between different winding layers and constant tension within each layer. This refined tension control strategy is crucial for optimizing the residual stress distribution during the winding process. The resin impregnation steady-state control module, based on the resin content parameters in the digital process file, adjusts the temperature of the impregnation tank and the gap pressure of the extrusion rollers to ensure that the resin impregnation amount of the fiber remains within the process setting range, effectively addressing the problem of the impregnation process being susceptible to environmental fluctuations. To further improve winding accuracy and adaptability, the system can also be configured with a vision-assisted calibration module. This module monitors the fiber placement position in real time using an industrial camera and calculates the deviation between the actual position and the theoretical trajectory as a compensation amount, feeding it back to the multi-axis motion control module in real time to dynamically correct the synchronous motion command set. This online calibration mechanism can effectively compensate for interference factors such as mandrel installation errors, mold thermal deformation, and fiber conveying slippage, significantly improving the accuracy of the winding trajectory and the first layer bonding accuracy. In addition, to achieve continuous process optimization and intelligent upgrading, the system can also be configured with a self-evolutionary optimization module. This module establishes a data interface to collect long-term data on equipment control parameters, environmental parameters, and product quality during the production process. It then utilizes machine learning models to establish predictive relationships between these control parameters and product quality. When production demands change, this module can invoke the model to iteratively optimize core process parameters and output a recommended set of process parameters, thereby achieving self-learning and continuous optimization of process parameters. Through the close collaboration and information exchange among these modules, this system forms a highly integrated, intelligent, and collaborative control loop.Digital process documents serve as the core link, seamlessly connecting process planning with equipment execution; each professional control module works independently yet collaboratively, ensuring the accuracy and stability of the winding process; while optional visual calibration and self-evolution optimization modules further enhance the system's adaptability and intelligence, enabling the production of high-pressure glass fiber pipeline pipes to stably and efficiently reach higher pressure rating standards.

[0063] The above description is merely a preferred embodiment of the present invention and is not intended to limit the present invention. Any modifications, equivalent substitutions, and improvements made within the spirit and principles of the present invention should be included within the protection scope of the present invention.

Claims

1. A control method for a high-pressure glass fiber pipeline winding device, characterized in that, The method is applied to a control system including a host computer, a multi-axis motion controller, a servo drive system, a tension closed-loop control unit, and a resin impregnation control unit. The method includes the following steps: S1: The host computer automatically calculates and generates a digital winding process file containing parameters such as layer sequence, winding angle, yarn trajectory, tension reference value and resin content based on the input structural parameters and performance requirements of the target pipeline pipe through the built-in winding line algorithm library. S2: The multi-axis motion controller parses the digital winding process file, generates a synchronous motion instruction set based on the linkage mathematical model between the mandrel spindle and the guide head motion axis, and drives the servo drive system to accurately lay the impregnated fiber onto the mandrel surface according to a predetermined spatial trajectory. S3: When executing the digital winding process file, the tension closed-loop control unit calls the corresponding tension reference value according to the current winding layer sequence and collects the fiber tension signal in real time. By adjusting the output torque or yarn feeding speed of the yarn feeding mechanism, it realizes independent and adjustable tension closed-loop control between different winding layers and constant tension within the layer. S4: The resin impregnation control unit adjusts the set temperature of the impregnation tank constant temperature device and the gap pressure of the extrusion roller according to the resin content parameters in the digital winding process file, so that the resin impregnation amount of the fiber is kept within the process setting range.

2. The control method for high-pressure glass fiber pipeline winding equipment according to claim 1, characterized in that, In step S2, the linkage mathematical model is defined as follows: the ratio of the displacement increment of the guide wire head along the mandrel axis to the circumference at the current winding position of the mandrel is equal to the tangent of the current winding angle; the multi-axis motion controller uses the mandrel spindle rotation angle as a reference and calculates the target position of the guide wire head in real time through electronic cam table mapping.

3. The control method for high-pressure glass fiber pipeline winding equipment according to claim 1, characterized in that, The setting of the tension reference value in step S3 follows an interlayer decreasing strategy: during the winding process from the inner layer to the outer layer, the tension reference value of the outer layer winding is set to 0.92 to 0.98 times the tension reference value of the adjacent inner layer.

4. The control method for high-pressure glass fiber pipeline winding equipment according to claim 1, characterized in that, The method further includes step S5: online vision-assisted calibration; deploying an industrial camera near the guide wire head, triggering image acquisition of the core mold baseline or the edge of the wound layer at the beginning or key section of each layer being laid; identifying the deviation between the actual laying position and the theoretical trajectory through an image processing algorithm, and feeding this deviation as a position compensation amount into the multi-axis motion controller in real time to dynamically correct the synchronous motion command set.

5. The control method for high-pressure glass fiber pipeline winding equipment according to claim 4, characterized in that, In the online vision-assisted calibration step, the position compensation amount is decomposed into fine-tuning amounts for the axial position and / or tangential sway angle of the guide wire head through the coordinate transformation module.

6. The control method for high-pressure glass fiber pipeline winding equipment according to claim 1, characterized in that, The method further includes step S6: self-evolution optimization of process parameters; establishing a data interface for communication with the production management system, and collecting and storing quality data including equipment control parameter sequences, environmental parameters, and corresponding pipeline finished product water pressure burst strength and ultrasonic test results over a long period of time; training a machine learning model based on historical data to establish a predictive relationship from control parameters to product quality; when producing new specifications of products or switching raw materials, the host computer calls the machine learning model, takes the maximization of target performance as the optimization objective, iteratively optimizes the core parameters in the digital winding process file, and outputs a recommended set of process parameters.

7. The control method for high-pressure glass fiber pipeline winding equipment according to claim 1, characterized in that, The machine learning model is a deep neural network or a gradient boosting decision tree model; the core parameters include at least the tension baseline value of each layer, the winding speed, and the resin temperature; the optimization objective is to pursue Pareto optimality of material usage or production cycle while satisfying the minimum burst strength constraint.

8. A control system for a high-pressure glass fiber pipeline winding equipment, the system being used to implement the control method for the high-pressure glass fiber pipeline winding equipment as described in any one of claims 1 to 7, characterized in that, The system includes: The process planning module is used to execute step S1; A multi-axis motion control module is used to execute step S2; Tension coordination control module, used to execute step S3; The adhesive impregnation steady-state control module is used to execute step S4; And optionally, a vision-assisted calibration module for performing step S5; And optionally, a self-evolutionary optimization module for performing step S6.

Images