Pipe forming system based on real-time defect feedback

Abstract

The present application belongs to the field of pipe manufacturing and intelligent control technology, and solves the problems of longitudinal cracks, scratches, pitting, uneven size and excessive residual stress on the surface caused by thermal coupling imbalance in the existing pipe extrusion molding. The present application discloses a pipe forming system based on real-time defect feedback, which comprises an execution device, a central control device, a multi-modal sensing device and a three-section temperature control device. The three-zone temperature control structure of the initial cooling section, the constant temperature section and the final cooling section is adopted, and the multi-modal sensing device is arranged at the outlet of each section to collect surface image, wall thickness and temperature data in real time, construct a defect feature map and match a process knowledge base to output thermal coupling control instructions. Through the constant temperature section partition heating and the final cooling section gradient cooling, stress relaxation and shaping control are realized, and at the same time, the process parameters are dynamically corrected by combining closed-loop feedback compensation. The present application can effectively inhibit the surface defects of the pipe, reduce the residual stress, improve the wall thickness uniformity and mechanical properties, and improve the production stability and product qualification rate.

Description

Technical Field

[0001] This invention belongs to the field of pipe manufacturing and intelligent control technology, and more specifically relates to a pipe forming system based on real-time defect feedback. Background Technology

[0002] In the field of pipe manufacturing, extrusion stretching and orientation molding is one of the most widely used core processing technologies. It has been widely applied in the pipe forming process of thermoplastic materials such as polyethylene, polyvinyl chloride, and random copolymer polypropylene. The basic process of this technology usually includes: first, heating the thermoplastic raw material to a molten state through an extruder, and then extruding it through a ring die to form a continuous tubular blank. Under the axial tension applied by the traction device, the high-temperature tube blank is stretched in a specific direction. At the same time, it is rapidly cooled and shaped by a cooling system, and finally, finished pipes with specific outer diameter, wall thickness and mechanical properties are obtained. This technology is widely used in municipal heating, building heating and centralized cooling projects due to its advantages such as relatively simple equipment structure, high production efficiency and controllable cost.

[0003] However, although the process is relatively mature in industrial applications, during the processing, longitudinal cracks, irregular scratches, or pitting resembling ramen noodles may occasionally appear on the outer surface of the tube.

[0004] Pipes exhibiting this condition not only suffer from poor appearance consistency, but are also highly susceptible to brittle fracture, cracking, and leakage during subsequent installation, construction, and long-term use due to external impacts, medium pressure, environmental corrosion, and other factors. This significantly shortens the designed service life of the pipe and may even lead to safety hazards. Summary of the Invention

[0005] To address the shortcomings of existing technologies, the present invention aims to provide a pipe forming system and control method based on real-time defect feedback. By introducing a three-zone structure of cooling, constant temperature, and re-cooling into the extrusion stretching orientation forming process, and combining it with multimodal online defect detection, the system achieves decoupled control of thermal stress and mechanical tensile force, thereby effectively suppressing the generation of surface defects in the pipe, improving the mechanical properties of the pipe, and significantly increasing the pass rate and production stability.

[0006] A pipe forming system based on real-time defect feedback includes: an actuator, a central control device, and a multimodal sensing device; the temperature control device includes: a constant temperature section set between the initial cooling section and the final cooling section, and the multimodal sensing device is arranged around the outlet of the initial cooling section, the outlet of the constant temperature section, and the outlet of the final cooling section.

[0007] The central control unit is equipped with central control commands, including:

[0008] S1: Real-time acquisition of multi-data of the pipe material through a multi-modal sensing device to generate acquisition information;

[0009] S2: Construct a defect feature map of the pipe based on the collected information;

[0010] S3: Based on the defect feature map, retrieve the thermo-coupling control instruction set from the pre-built process knowledge base;

[0011] S4: Execute the thermodynamic coupling control instruction set;

[0012] S5: A feedback compensation strategy is configured. The feedback compensation strategy is used to generate source tracing analysis results and generate compensation parameters based on the source tracing analysis results. The compensation parameters are used to correct the thermo-coupling control instruction set in the process knowledge base.

[0013] Multivariate data includes images of the pipe's outer surface, wall thickness profile data, and infrared temperature data;

[0014] The construction of the pipe defect feature map includes: extracting the spatial frequency features of appearance defects; generating temperature gradient vector fields along the axial and circumferential directions of the pipe; extracting the low-frequency fluctuation components of the wall thickness profile data as wall thickness non-uniformity indicators; fusing the spatial frequency features, temperature gradient vector fields, and wall thickness non-uniformity indicators into a defect characterization vector; and inputting the defect characterization vector into a pre-trained lightweight convolutional neural network model to output the defect feature map.

[0015] The defect feature map includes defect classification labels and corresponding confidence scores;

[0016] The thermodynamic coupling control instruction set includes an optimized control instruction set and an emergency control instruction set. When the confidence score is higher than the confidence threshold, the optimized control instruction set is retrieved based on the defect classification label; when the confidence score is lower than the confidence threshold, the emergency control instruction set is retrieved.

[0017] The S4 also includes a parameter optimization sub-strategy. The thermo-coupling control instruction set includes joint control parameters corresponding to different execution devices. The parameter optimization sub-strategy is used to optimize the joint control parameters in the optimized control instruction set. The parameter optimization sub-strategy configures corresponding quality optimization weights and matches corresponding defect classification labels according to different time windows, and generates update factors through a preset multi-objective optimization algorithm to update the corresponding joint control parameters.

[0018] S4 also includes an emergency response sub-strategy, which responds to the execution of the emergency control instruction set and includes returning to step S1 until the confidence score meets the preset stable achievement conditions.

[0019] The S5 further includes comparing the collected data corresponding to the multimodal sensing devices with preset relationships to generate feedback results, and then matching them with preset influence rules in the process knowledge base to generate source tracing analysis results.

[0020] The upstream multimodal sensing device is a multimodal sensing device installed at the outlet of the primary cooling section and the outlet of the constant temperature section, and the downstream multimodal sensing device is a multimodal sensing device installed at the outlet of the final cooling section.

[0021] All multimodal sensing devices acquire data synchronously based on a unified timestamp, and a mapping relationship between acquisition time and pipe axial position is established with pipe production speed as the benchmark. Feedback results are generated by calculating the feature similarity of upstream and downstream defect characterization vectors and the rate of change of defect severity. The rate of change of defect severity is quantified by the reduction ratio of longitudinal crack density, the reduction magnitude of temperature gradient standard deviation, and the change in wall thickness fluctuation variance. Based on the feedback results, the weighting coefficients are adjusted, the effective control parameter adjustment magnitude is increased proportionally, or the parameter combination is redesigned.

[0022] The actuator includes a thermostatic heating element arrayed inside the thermostatic cavity;

[0023] The thermodynamic coupling control instruction set includes thermostatic device control instructions, which include axial heating sub-instructions and circumferential heating sub-instructions. The axial heating sub-instructions are used to control the relationship between the heating power of thermostatic heating elements located on the same axis, and the circumferential heating sub-instructions are used to control the relationship between the heating power of thermostatic heating elements located on the same circumference.

[0024] The final cooling section adopts a gradient cooling structure, including a first-stage atomized spray zone, a second-stage air-mist mixing zone, and a third-stage immersion water tank zone. The cooling rate of each zone satisfies a piecewise function relationship along the pipe axis, and the cooling coefficient of each zone decreases sequentially. Based on the real-time wall thickness profile data collected by the multi-modal sensing device, the cooling intensity ratio of the first-stage atomized spray zone to the third-stage immersion water tank zone is dynamically adjusted.

[0025] The execution device includes a spraying compensation device, and the corresponding thermo-coupling control instruction set includes a spraying compensation sub-instruction. The spraying compensation sub-instruction is used to control the operation of the spraying compensation device. The spraying compensation sub-instruction is configured with a trigger condition. When the compensation trigger condition is met, the spraying compensation sub-instruction is output. The compensation trigger condition is that the spatial frequency characteristics of the pipe appearance defects collected by the multi-modal sensing device at the outlet of the final cooling section are greater than a threshold, and the traction speed of the traction device is lower than the maximum speed allowed for spraying.

[0026] The beneficial effects of this invention are:

[0027] The system features a three-zone temperature control structure consisting of an initial cooling zone, a constant temperature zone, and a final cooling zone. Combined with the surrounding deployment of multi-modal sensors at the outlets of each zone, it enables real-time acquisition of diverse data throughout the entire pipe forming process. This allows for the accurate capture of defect characteristics at different forming stages, providing comprehensive data support for subsequent control.

[0028] By using a lightweight convolutional neural network model to construct a feature map of pipe defects, the defect classification labels and confidence scores are clearly defined. Combined with a pre-set process knowledge base, the system can accurately retrieve the optimized control command set and the emergency control command set. Differentiated control strategies are adopted for different defect severity levels, which effectively improves the targeting and efficiency of defect handling.

[0029] The constant temperature section is equipped with an array of constant temperature heating elements. The heating power is precisely controlled by axial heating sub-commands and circumferential heating sub-commands. Combined with the gradient cooling structure of the first-stage atomized spray zone, the second-stage air-mist mixing zone, and the third-stage immersion water tank zone in the final cooling section, as well as the dynamic cooling intensity adjustment, the decoupling control of thermal stress and mechanical tensile force is achieved, and the generation of defects on the pipe surface is suppressed.

[0030] The parameter optimization sub-strategy dynamically adjusts the quality optimization weights according to different time windows and updates the joint control parameters through a multi-objective optimization algorithm. Meanwhile, the emergency response sub-strategy locks key parameters and activates buffer cooling intensity when the confidence score fails to meet the standard, ensuring process stability and product consistency.

[0031] The feedback compensation strategy compares the data collected by upstream and downstream multimodal sensing devices, quantifies the similarity of defect features and the rate of change of severity, and generates source analysis results and compensation parameters by combining the influence rules of the process knowledge base. This enables dynamic correction of the thermo-coupling control instruction set and continuous optimization of process parameter matching. Attached Figure Description

[0032] Figure 1 This is a flowchart of a pipe forming system based on real-time defect feedback.

[0033] Figure 2 This is a structural diagram of a pipe forming system based on real-time defect feedback.

[0034] Figure 3 This is a cross-sectional view of a pipe forming system based on real-time defect feedback.

[0035] In the diagram: 11. Initial cooling section, 12. Constant temperature section, 13. Final cooling section, 131. First-stage atomizing spray zone, 132. Second-stage aerosol mixing zone, 133. Third-stage immersion water tank zone, 2. Extrusion device, 51. Atomizing sprayer I, 52. Constant temperature heating element, 521. Electric heating wire, 522. Electric heating ring, 53. Atomizing sprayer II, 54. Aerosol sprayer, 55. Immersion tank, 6. Multimodal sensing device, 7. Traction device. Detailed Implementation

[0036] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. 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.

[0037] It should be noted that when a component is described as "fixed to" another component, it can be directly on the other component or may have a component in between. When a component is considered "connected to" another component, it can be directly connected to the other component or may have a component in between. When a component is considered "set on" another component, it can be directly set on the other component or may have a component in between. The terms "vertical," "horizontal," "left," "right," and similar expressions used in this document are for illustrative purposes only.

[0038] Unless otherwise defined, all technical and scientific terms used herein have the same meaning as commonly understood by one of ordinary skill in the art to which this invention pertains. The terminology used herein in the description of the invention is for the purpose of describing particular embodiments only and is not intended to be limiting of the invention. The term "and / or" as used herein includes any and all combinations of one or more of the associated listed items.

[0039] Regarding the occasional appearance of ramen-like longitudinal cracks, irregular scratches, or pitting on the outer surface of the tube during processing, this application, through research, discovered that the cause of this problem is as follows: When the high-temperature molten tube blank is extruded from the die head, it immediately enters the cooling zone and undergoes a rapid cooling process. During this process, due to the different order of contact between the inner and outer walls of the tube and the cooling medium and the different heat conduction rates, a significant radial temperature gradient is created. The outer layer cools and shrinks rapidly, while the inner layer remains at a higher temperature and has not yet fully solidified. This results in uneven thermal shrinkage from the outside to the inside, forming a complex thermal stress distribution. At the same time, the traction device continuously applies axial tensile force to control the outer diameter and wall thickness uniformity of the tube. This mechanical tensile action and thermal stress highly overlap in time and space, forcing the material to solidify in a non-equilibrium state before stress relaxation is completed, which easily leads to the formation of stress concentration areas, i.e., thermal coupling imbalance.

[0040] Under the combined effect of high stretching rate and rapid cooling, due to the excessive orientation of molecular chains in the stretching direction in local areas and the failure to relax effectively, longitudinal microcracks or fine scratches resembling ramen will appear on the surface of the pipe. If the cooling rate is too fast and the material shrinks unevenly, pits will form on the surface of the pipe.

[0041] Based on the above-mentioned problems, the present invention provides a pipe forming system based on real-time defect feedback to solve these problems;

[0042] The pipe forming system includes: an actuator, an extrusion device 2, a temperature control device, a spraying compensation device, a traction device 7, and a central control device;

[0043] The temperature control device of the pipe forming system includes: an initial cooling section 11, a constant temperature section 12, and a final cooling section 13 arranged sequentially between the extrusion device 2 and the traction device 7;

[0044] The pipe forming system also includes: a multimodal sensing device 6, which is arranged around the outlet of the primary cooling section 11, the outlet of the constant temperature section 12 and the outlet of the final cooling section 13.

[0045] The extrusion unit 2 is used to continuously extrude molten thermoplastic material to form a high-temperature tubular preform. The primary cooling section 11 is located downstream of the extrusion unit 2, and the actuator includes an atomizing sprayer I51, which is used to rapidly and initially cool the freshly extruded tubular preform to suppress surface defects and preliminarily shape it. The constant temperature section 12 is located downstream of the primary cooling section 11, and the actuator also includes a constant temperature heating element 52. In this application, the constant temperature heating element consists of an axially distributed electric heating wire 521 and circumferentially partitioned electric heating rings 522, used to control the temperature of the preform. The process achieves molecular chain relaxation and stress release under the following conditions: The final cooling section 13 is located downstream of the constant temperature section 12 and includes a first-stage atomizing spray zone 131, a second-stage aerosol mixing zone 132, and a third-stage immersion water tank zone 133. The actuator also includes corresponding atomizing sprayer II 53, aerosol sprayer 54, and an adjustable water level immersion tank 55, which are used to complete the final shaping of the pipe and suppress the accumulation of residual stress. The traction device 7 is located at the outlet of the final cooling section and has programmable speed control and tension feedback functions, which are used to control the stretching rate and tension of the pipe.

[0046] Multimodal sensing devices 6 are arranged around the outlets of the primary cooling section 11, the constant temperature section 12, and the final cooling section 13 to collect real-time information on the surface morphology, temperature field distribution, and geometric dimensions of the pipe. In this embodiment, the multimodal sensing device 6 includes a high-speed line scan camera, an infrared thermal imager, and a laser rangefinder, which are used to acquire surface texture images, radial temperature field distribution, and wall thickness profile data of the pipe, respectively.

[0047] The final cooling section 13 employs a three-stage gradient cooling structure to form a controlled gradient cooling curve, thereby suppressing residual stress accumulation. The first-stage atomizing spray zone 131 sprays fine water mist through atomizer II 53, achieving rapid but controllable initial cooling. The second-stage air-mist mixing zone 132 introduces compressed air through atomizer 54 to mix with the water mist, reducing the cooling intensity. The third-stage immersion water tank zone 133 completely immerses the pipe in the circulating cooling water medium through immersion tank 55 to complete the final shaping. Cooling rate It satisfies a piecewise function along the axial direction, and the calculation formula is:

[0048]

[0049] , , These are the cooling coefficients for the three cooling zones; the higher the value, the faster the cooling speed. In this application... , It is 8k / s. 4k / s The design intensity is 1.5 k / s. The rationale for this design is as follows: the pipe is still in a highly elastic state immediately after leaving the constant temperature zone 12. Rapid cooling of the surface layer is required to fix the outer diameter and prevent deformation caused by subsequent stretching. The high cooling intensity allows the surface of the pipe to solidify quickly without causing a sudden drop in the core temperature, leaving sufficient space for subsequent stress relaxation. After the surface layer solidifies, the core of the pipe still has a relatively high temperature and residual stress. The moderate cooling intensity allows the molecular chains to move slowly, releasing most of the internal stress and avoiding cracking caused by stress concentration. When the temperature is close to room temperature, the low cooling intensity allows the pipe to cool slowly, making the temperature of the inner and outer walls of the pipe uniform and avoiding secondary thermal stress caused by temperature differences, thus ensuring the final dimensional stability.

[0050] The ambient temperature is set at room temperature, eliminating the need for additional heating or cooling and reducing energy consumption. The temperature of the aerosol mixing zone is slightly lower than the ambient temperature, which ensures moderate cooling intensity without causing the aerosol to condense into water droplets due to excessively low temperature, thus affecting the surface quality of the pipe. The water bath temperature is lower than the aerosol mixing zone temperature to ensure the final setting effect. In this application, the ambient room temperature is used as the reference. =25℃, =20℃, =15℃, ensuring the pipe material gradually cools to room temperature to avoid secondary thermal shock; when the laser ranging sensor detects that the wall thickness of a certain axial section is too thin, the system dynamically reduces the corresponding temperature in that area. The value, thereby reducing the atomized spray pressure, slows down the initial shrinkage, and simultaneously improves... By increasing the immersion depth, the final shaping is accelerated, achieving synergistic optimization of dimensional accuracy and internal stress.

[0051] The central control unit is equipped with central control commands, including:

[0052] S1: During the pipe extrusion molding process, the multi-modal sensing device 6 collects multi-data of the pipe in real time to generate collected information. The multi-data includes the morphological characteristics of the outer surface and inner wall of the pipe, temperature distribution and wall thickness fluctuation information.

[0053] S2: Based on the collected information, construct a defect feature map of the pipe material to identify the current defect type, spatial location and severity;

[0054] S3: Based on the defect feature map, retrieve the thermo-coupling control instruction set from the pre-built process knowledge base. The thermo-coupling control instruction set includes the temperature set value of the constant temperature section 12, the traction speed of the traction device 7, and the joint adjustment parameters of the cooling intensity of the initial cooling section 11 and the final cooling section 13.

[0055] The process knowledge base stores basic thresholds and limit parameters, including defect allowable thresholds for different production stages and materials, equipment physical limits, and process safety boundaries, which serve as standards for judgment and adjustment. It also records historical optimization cases and mapping relationships, stores successful case experiences from past production, from identifying defect types to adjusting parameters to the adjustment effect, and mapping rules for adjusting parameter amplitude based on defect characteristics, thereby reducing the calculation time of the decision-making layer and ensuring the accuracy of adjustment.

[0056] S4: Execute the thermo-coupling control instruction set; dynamically adjust the heating power of the constant temperature heating body in the constant temperature section 12, the stretching rate of the traction device 7, and the fluid heat transfer parameters of the cooling fluid in the initial cooling section 11 and the final cooling section 13, so that the pipe can achieve stress relaxation and orderly rearrangement of molecular chains in the constant temperature section 12, and form a controlled gradient cooling curve regulated by the fluid heat transfer parameters in the final cooling section 13 to suppress the accumulation of residual stress.

[0057] S5: A feedback compensation strategy is configured. The feedback compensation strategy is used to generate source tracing analysis results and generate compensation parameters based on the source tracing analysis results. The compensation parameters are used to correct the thermo-coupling control instruction set in the process knowledge base.

[0058] S11: After the molten polymer is extruded through the extrusion device 2, it forms a high-temperature tubular billet, which then enters the primary cooling section 11. At this time, the high-speed line scan camera, infrared thermal imager, and laser rangefinder at the outlet of the primary cooling section 11 simultaneously collect data: the high-speed line scan camera acquires high-resolution texture images of the outer surface of the pipe, which are used to detect surface defects such as longitudinal cracks, pits, and scratches; the infrared thermal imager captures the radial temperature field distribution of the pipe, which is used to reflect the cooling uniformity; and the laser rangefinder obtains the inner and outer wall contours of the pipe by measurement, and then calculates the real-time wall thickness.

[0059] S21: The construction of the defect feature map includes the following steps: performing wavelet denoising and edge enhancement processing on the surface image to extract the spatial frequency features of longitudinal cracks, pits, and scratches; performing two-dimensional interpolation reconstruction on the infrared temperature data to generate a temperature gradient vector field along the axial and circumferential directions of the pipe; performing Fourier transform on the wall thickness profile to extract low-frequency fluctuation components as wall thickness unevenness indicators; fusing the above three types of features into a three-dimensional defect representation vector, and inputting the defect representation vector into a pre-trained lightweight convolutional neural network model to output the defect feature map, wherein the defect feature map includes defect classification labels and corresponding confidence scores.

[0060] S31: The thermo-coupling control instruction set includes an optimized control instruction set and an emergency control instruction set. When the confidence score is higher than the confidence threshold, the optimized control instruction set is retrieved according to the defect classification label; when the confidence score is lower than the confidence threshold, the emergency control instruction set is retrieved.

[0061] Confidence score The calculation formula is:

[0062]

[0063] in, This is the surface defect density normalization factor, i.e., the deduction item for appearance defects. =1 / (1+Dobs), where Dobs is the measured defect density. The more cracks and pits there are per meter of pipe, the more defects there are, and the smaller the value will be. This is a deduction item for radial thermal stress. The L2 norm of the radial temperature gradient is calculated by collecting measurement points along the radial direction of the pipe using an infrared thermal imager. A larger value indicates more drastic radial temperature changes across the entire circumference, reflecting a higher risk of thermal stress, and resulting in a lower score. This is an empirical coefficient, ranging from 0.05 to 0.1. This range is obtained through experience in engineering practice. For example, PP pipes generally have low thermal conductivity, and temperature differences have a significant impact on cracking. Taking 0.1, PE pipes have high thermal conductivity. Take 0.05; For wall thickness variance, For nominal wall thickness, the more uniform the wall thickness, the smaller the wall thickness variance and the higher the score; conversely, the larger the wall thickness deviation, the lower the score.

[0064] S41: S4 further includes a parameter optimization sub-strategy. The thermo-coupling control instruction set includes joint control parameters corresponding to different actuators, namely, the heating power of the heating element in the constant temperature section 12, the stretching rate of the traction device 7, and the fluid heat transfer parameters of the cooling fluid in the initial cooling section 11 and the final cooling section 13. The parameter optimization sub-strategy is used to optimize the joint control parameters in the optimization control instruction set. The parameter optimization sub-strategy configures corresponding quality optimization weights and matches corresponding defect classification labels according to different time windows, and generates update factors through a preset multi-objective optimization algorithm to update the corresponding joint control parameters. The calculation formula for the thermo-coupling control instruction set is:

[0065]

[0066] in, To control the variable vector, , These are the set temperatures for the three independent temperature control zones within the 12 constant temperature ranges. The traction speed of traction device 7. The atomized spray flow rate of the initial cooling section 11, The liquid level depth in the immersion tank 55 of the final cooling section 13. To predict the surface defect density under the current control strategy, the defect density under the current control strategy is quickly predicted based on the correlation between real-time surface defect characteristics and historical parameters. The residual stress level inside the pipe is calculated in real time using thermo-coupled finite element simulation, information collected by multimodal sensing device 6, and process temperature. The wall thickness fluctuation variance is calculated in real time from laser ranging data. , , These are weighting coefficients, dynamically adjusted based on the current time window. As constraints, the process safety boundaries of the heating pipe are obtained from the preset process knowledge base, including the physical limits of the equipment such as the maximum heating power and the minimum traction speed, as well as the heating temperature of the pipe. These constraints then constrain the temperature range of the constant temperature section 12, the traction speed range of the traction device 7, and the spray flow range of the initial cooling section 11.

[0067] In this application, the time window includes: the initial start-up stage, the initial transition to steady state stage, and the steady state stage. In the initial start-up stage, the proportion of [missing information] is the highest, and the appearance quality is given priority. In the initial transition to steady state stage, the proportion of [missing information] is equal to that of [missing information], and the consistency of appearance quality and mechanical properties is guaranteed. After entering steady state production, the proportion of [missing information] is the highest, and the consistency of mechanical properties is given priority.

[0068] Specifically, the first 5 minutes of system startup are defined as the initial startup phase. , , The weighting ratio of the three is 6:3:1;

[0069] High-speed line scanning camera, infrared thermal imager and laser rangefinder sensor collect pipeline information, output defect classification labels and confidence scores through defect feature map, and determine defect priority according to weight ratio when the confidence score is higher than the preset confidence threshold.

[0070] If the high-speed linear scanning camera detects that the longitudinal crack density exceeds the threshold, in this stage, longitudinal cracks and surface appearance defects such as scratches are the main priority defects. The parameter mapping rules are matched according to the process knowledge base. As recorded in the process knowledge base, in the startup stage, the corresponding control scheme for PPR pipes with excessive longitudinal cracks is as follows: the temperature of the constant temperature section 12 is increased by turning on the electric heating wires 521 in the first, second and third zones of the constant temperature section 12, and the spray flow rate of the initial cooling section 11 is reduced. That is, the temperature of the constant temperature section 12 is increased according to the excessive value of crack density, and the flow rate of the atomizing sprayer I51 in the initial cooling section 11 is reduced according to the excessive value of radial temperature gradient. Then, the parameters are updated by the parameter optimization sub-strategy in the thermo-coupling control instruction set and the corresponding convergence threshold recorded in the process knowledge base to update the corresponding joint control parameters.

[0071] The adjustment is based on the following: reducing the spray flow rate of the initial cooling section 11 to slow down the cooling rate, thereby reducing the temperature difference between the inner and outer walls of the pipe and alleviating thermal stress; increasing the temperature of the constant temperature section 12 to extend the stress relaxation time of the material and repair microcracks; and not adjusting the traction speed to avoid the superposition of mechanical tensile force and thermal stress, and to prevent the cracks from worsening.

[0072] Specifically, the period from the 5th to the 15th minute of system startup is defined as the startup-to-steady-state phase. , , The weighting ratio of the three is 4:4:2;

[0073] If the high-speed linear scanning camera detects that the spot density exceeds the threshold range, the infrared thermal imager detects that the circumferential temperature gradient exceeds the threshold range, or the laser rangefinder detects that the wall thickness variance is below the threshold range.

[0074] During this stage, surface defects such as longitudinal cracks, pitting and scratches, as well as thermal stress problems caused by uneven circumferential temperature, are the main priority defects, while thin wall thickness is a secondary defect.

[0075] Based on the process knowledge base, parameter mapping rules are matched. For example, as recorded in the process knowledge base, during the start-up and steady-state transition phase, the control scheme for pitting on PPR pipes is to reduce the initial cooling intensity of the first-stage atomizing spray zone 131 in the final cooling section 13; the control scheme for uneven circumferential temperature is to activate the electric heating ring 522 in the corresponding pitted areas in the second and third zones along the pipe travel direction in the constant temperature section 12 to perform circumferential zone heating; the control scheme for thinner wall thickness is to raise the water level in the immersion tank 55 of the final cooling section 13; then, through the parameter optimization sub-strategy in the thermo-coupling control instruction set, the parameters are updated according to the corresponding convergence threshold recorded in the process knowledge base to update the corresponding joint control parameters.

[0076] Adjustment basis: By circumferentially heating the corresponding area of ​​the pipe, the insufficient temperature in the corresponding area of ​​the pipe is compensated, thereby alleviating the difference in circumferential thermal shrinkage and eliminating pitting; by reducing the cooling intensity of the first-stage atomizing spray zone 131 of the final cooling section 13, the cooling rate of the first-stage atomizing spray zone 131 is slowed down, avoiding uneven shrinkage of the pipe surface; by increasing the water level of the immersion tank 55, the immersion depth of the pipe is increased, thereby accelerating the shaping of the pipe and correcting the problem of thin wall thickness.

[0077] Specifically, the 15th minute after system startup is defined as the steady-state phase. , , The weighting ratio of the three is 2:6:2;

[0078] The infrared thermal imager detected that the radial temperature gradient L2 norm was higher than the threshold range for this stage, while the laser rangefinder detected that the wall thickness variance was lower than the threshold range for this stage.

[0079] During this stage, the residual stress risk caused by excessive radial temperature gradient is the primary priority defect, while thin wall thickness is a secondary defect.

[0080] The parameter mapping rules are matched based on the process knowledge base. For example, as recorded in the process knowledge base, during the steady-state phase...

[0081] The control scheme corresponding to the excessive radial temperature gradient is to increase the temperature of the constant temperature section 12 and reduce the cooling intensity of the first-stage atomizing spray zone 131 of the final cooling section 13. The control scheme corresponding to the thin wall thickness is to reduce the traction speed of the traction device 7 and increase the water level of the immersion tank 55 of the final cooling section 13. Then, through the parameter optimization sub-strategy in the thermo-coupling control instruction set, the parameters are updated by updating the corresponding joint control parameters through the corresponding convergence threshold recorded in the process knowledge base.

[0082] Adjustment basis: By increasing the temperature of the constant temperature section 12 and reducing the traction speed of the traction device 7, the stress relaxation time is extended, thereby reducing residual stress; by reducing the cooling intensity of the first-stage atomizing spray zone 131 of the final cooling section, secondary thermal shock is avoided, thereby suppressing the accumulation of residual stress; by increasing the water level in the water tank, the cooling and shaping effect is enhanced, thereby correcting the problem of thin wall thickness.

[0083] S4 also includes an emergency response sub-strategy, which responds to the execution of the emergency control instruction set and includes returning to step S1 until the confidence score meets the preset stable achievement conditions.

[0084] When the confidence score is lower than the preset confidence threshold, the emergency response sub-strategy limits the rate of change of the traction speed of the traction device 7 according to the mapping rules in the process knowledge base, and freezes the set values ​​of each temperature zone in the constant temperature section to prevent the process from running out of control due to large parameter fluctuations. The traction speed is fixed at the historical best value, and the entire constant temperature section 12 is uniformly set to the annealing temperature of the pipe material. For example, if the pipe is made of PP material, the annealing temperature is 150℃. At the same time, the final cooling section 13 activates the maximum buffer cooling intensity, that is, reduces the cooling coefficient of the first-stage atomized spray zone 131 in the final cooling section and increases the cooling coefficient of the immersion tank 55. The system continues to operate in this mode until the process confidence score is higher than the preset confidence threshold for three consecutive samples, at which point the lockout is released.

[0085] The S4 also includes an emergency response sub-strategy. When the confidence score remains below a preset threshold for more than a preset time, the emergency response sub-strategy responds to the execution of the emergency control command set. Under the emergency response sub-strategy, the traction speed is fixed at the historical best value, and the constant temperature section is uniformly set to the annealing temperature of the pipe material. For example, if the pipe is made of PP material, the annealing temperature is 150℃. At the same time, the final cooling section activates the maximum buffer cooling intensity, that is, reducing the cooling coefficient of the first-stage atomizing spray zone 131 and increasing the cooling coefficient of the third-stage immersion water tank zone 133. In this application... Reduced to 6k / s Increase the speed to 3k / s, and continue running the system until the process confidence score is higher than the preset confidence threshold for three consecutive samples. Then, unlock the system and return to step one.

[0086] S51: The feature map constructed from the data collected by the upstream multimodal sensing device 6 includes the defect characterization vector collected by the upstream multimodal sensing device 6. When the pipeline reaches the collection area of ​​the downstream multimodal sensing device 6, a defect characterization vector is formed based on the data collected by the downstream multimodal sensing device 6. The two defect characterization vectors are compared to generate feedback results. According to the preset influence rules in the process knowledge base, the cause is deduced in reverse and compensation parameters are generated. The compensation parameters are used to correct the thermo-coupling control instruction set in the process knowledge base.

[0087] The upstream multimodal sensing device 6 collects pipe data at the outlet of the initial cooling section 11 and the outlet of the constant temperature section 12, respectively. The downstream multimodal sensing device 6 completes data collection at the outlet of the final cooling section. All collection actions are executed synchronously according to a unified timestamp. For each pipe segment, upstream and downstream defect characterization vectors are extracted. The vectors contain three categories: spatial frequency characteristics of appearance defects, temperature gradient vector field characteristics, and wall thickness unevenness indicators. Based on the pipe production speed, a mapping relationship between the collection time and the axial position of the pipe is established. The defect characterization vector of the pipe at a specific upstream position is accurately matched with the defect characterization vector when the pipe flows to the downstream detection point, avoiding comparison deviations caused by traction speed fluctuations or detection time sequence differences.

[0088] The actual effect of the control parameters is quantified by calculating the feature similarity of upstream and downstream defect characterization vectors and the rate of change of defect severity. Feature similarity is used to judge the degree of agreement between defect types. If there are longitudinal crack features upstream, the higher the matching degree of the downstream feature, the more it indicates that the control has not triggered new defect types. The rate of change of defect severity is calculated by the numerical changes of various defect indicators, such as the decrease in longitudinal crack density, the decrease in temperature gradient standard deviation, and the change in wall thickness fluctuation variance. Based on the calculation results, the feedback results are divided into three categories: the first category is effective control, that is, the feature similarity of downstream defect characterization vectors is high and the defect severity has decreased significantly; the second category is insufficient control, that is, the defect severity has improved but has not reached the preset standard, or some defect indicators have not changed significantly; the third category is ineffective control, that is, the defect severity has not improved or has even worsened, or new defect features have appeared.

[0089] The system retrieves the pre-defined association rules between defect type, control parameter, and influencing factors from the process knowledge base. It then matches these rules with the current feedback results and the actual control parameters to generate a source tracing analysis result. For example, if a longitudinal crack is detected upstream and the control parameter is to increase the temperature in the isothermal section, the crack density downstream decreases but does not meet the standard. In this case, the system matches the association rules between longitudinal crack, isothermal temperature, and stress relaxation time in the knowledge base. The initial determination is that the cause may be insufficient increase in the isothermal section temperature, resulting in inadequate stress relaxation of the material. For cases with multiple coupled effects, the system decomposes the weight of each control parameter to identify the core influencing factors. For example, when the isothermal section temperature and the initial cooling section spray flow rate are adjusted simultaneously, the system analyzes the influence coefficients of the two types of parameters on the defect indicators to determine which parameter's adjustment effect did not meet expectations.

[0090] Different compensation parameter generation strategies are adopted for different feedback result types. If the control is effective, the compensation parameters are mainly fine-tuned and optimized, and the weight coefficients in the multi-objective optimization algorithm are appropriately adjusted to consolidate the current control effect. For example, the weight of residual stress control is further increased in the steady-state production stage. If the control is insufficient, the adjustment range of the effective control parameters is increased proportionally according to the core causes identified by the source analysis, or auxiliary control measures are added. For example, if the temperature increase in the constant temperature section is insufficient, it is increased on the original basis, and the success of the circumferential zone heating is checked to ensure comprehensive temperature compensation. If the control fails, the current control strategy is overturned, and the parameter combination is redesigned based on the source results. For example, if the defect is aggravated due to the mismatch between traction speed and cooling intensity, the adjustment range of traction speed and the ratio of cooling coefficients in each area of ​​the final cooling section are replanned. The generation of all compensation parameters must strictly follow the equipment physical limits and process safety boundaries in the process knowledge base to ensure that parameter adjustments will not cause process runaway or equipment damage.

[0091] Compensation parameters, source analysis results, and feedback results are bound and stored, and added as new cases to the historical optimization case library of the knowledge base to enrich the case reserves of the knowledge base. For cases where control is effective or insufficient, the mapping rules of the control parameters corresponding to the original defect features are corrected, and the recommended range of parameter adjustment range is adjusted to improve the control accuracy of similar defects in the future. For cases where control fails, the problems of the current control strategy are recorded, the failure rules in the knowledge base are updated, and similar parameter combinations are avoided in the future. At the same time, based on the actual application effect of the compensation parameters, the constraints and convergence thresholds of the multi-objective optimization algorithm are continuously optimized, so that the process knowledge base can dynamically adapt to the pipe production needs of different production stages and different material types, and realize the continuous optimization of control strategies.

[0092] The execution device also includes a spraying compensation device located between the outlet of the final cooling section 13 and the traction device 7. The spraying compensation device includes an annular heating nozzle and an atomizing nozzle. The corresponding thermo-coupling control instruction set includes a spraying compensation sub-instruction. The spraying compensation sub-instruction is used to control the operation of the spraying compensation device. The spraying compensation sub-instruction is configured with triggering conditions. The triggering conditions of this application are: the collected data corresponding to the multimodal sensing device 6 with a preset relationship are compared to generate feedback results. When the feedback results show that there are still appearance defects on the pipe surface, that is, the spatial frequency characteristics of the appearance defects of the pipe collected by the multimodal sensing device at the outlet of the final cooling section are greater than the threshold, and the traction speed of the traction device is lower than the maximum speed allowed for spraying, the spraying compensation sub-instruction is output. The defect location is instantaneously and locally heated by the annular heating nozzle, and then a thermosetting polymer precursor, such as an amine curing agent mixture, is sprayed. The residual heat of the pipe is used to complete the in-situ cross-linking and curing, forming a reinforcing layer on the pipe wall. To ensure the uniformity of spraying, otherwise it is marked as a defective product.

[0093] In summary, this invention achieves decoupled control of thermal stress and mechanical tensile force through the deep integration of a three-zone structure of initial cooling, constant temperature, and subsequent cooling with multimodal real-time defect feedback. This not only effectively suppresses the generation of surface defects but also significantly improves the consistency of the mechanical properties of the pipe and the production stability, making it suitable for continuous manufacturing scenarios of plastic pipes.

[0094] The above are merely preferred embodiments of the present invention. The scope of protection of the present invention is not limited to the above embodiments. All technical solutions falling within the scope of the present invention's concept are within the scope of protection of the present invention. It should be noted that for those skilled in the art, any improvements and modifications made without departing from the principle of the present invention should also be considered within the scope of protection of the present invention.

Claims

1. A pipe forming system based on real-time defect feedback, comprising: An actuator, a central control device, and a multimodal sensing device; characterized in that the temperature control device includes: a constant temperature section disposed between the initial cooling section and the final cooling section, and the multimodal sensing devices arranged around the outlet of the initial cooling section, the outlet of the constant temperature section, and the outlet of the final cooling section. The central control unit is equipped with central control commands, including: S1: Real-time acquisition of multi-data of the pipe material through a multi-modal sensing device to generate acquisition information; S2: Construct a defect feature map of the pipe based on the collected information; S3: Based on the defect feature map, retrieve the thermo-coupling control instruction set from the pre-built process knowledge base; S4: Execute the thermodynamic coupling control instruction set; S5: A feedback compensation strategy is configured. The feedback compensation strategy is used to generate source analysis results and generate compensation parameters based on the source analysis results. The compensation parameters are used to correct the thermo-coupling control instruction set in the process knowledge base. Multivariate data includes images of the pipe's outer surface, wall thickness profile data, and infrared temperature data; The construction of the pipe defect feature map includes: extracting the spatial frequency features of appearance defects, generating temperature gradient vector fields along the axial and circumferential directions of the pipe, extracting the low-frequency fluctuation components of the wall thickness profile data as wall thickness unevenness indicators; fusing the spatial frequency features, temperature gradient vector fields, and wall thickness unevenness indicators into a defect characterization vector, and inputting the defect characterization vector into a pre-trained lightweight convolutional neural network model to output the defect feature map.

2. The pipe forming system based on real-time defect feedback according to claim 1, characterized in that, The defect feature map includes defect classification labels and corresponding confidence scores; The thermodynamic coupling control instruction set includes an optimized control instruction set and an emergency control instruction set. When the confidence score is higher than the confidence threshold, the optimized control instruction set is retrieved based on the defect classification label; when the confidence score is lower than the confidence threshold, the emergency control instruction set is retrieved.

3. The pipe forming system based on real-time defect feedback according to claim 2, characterized in that, The S4 also includes a parameter optimization sub-strategy. The thermo-coupling control instruction set includes joint control parameters corresponding to different execution devices. The parameter optimization sub-strategy is used to optimize the joint control parameters in the optimized control instruction set. The parameter optimization sub-strategy configures corresponding quality optimization weights and matches corresponding defect classification labels according to different time windows, and generates update factors through a preset multi-objective optimization algorithm to update the corresponding joint control parameters.

4. The pipe forming system based on real-time defect feedback according to claim 3, characterized in that, S4 also includes an emergency response sub-strategy, which responds to the execution of the emergency control instruction set and includes returning to step S1 until the confidence score meets the preset stable achievement conditions.

5. A pipe forming system based on real-time defect feedback according to claim 4, characterized in that, The S5 further includes comparing the collected data corresponding to the multimodal sensing devices with preset relationships to generate feedback results, and then matching them with preset influence rules in the process knowledge base to generate source tracing analysis results.

6. A pipe forming system based on real-time defect feedback according to claim 5, characterized in that, The upstream multimodal sensing device is installed at the outlet of the initial cooling section and the outlet of the constant temperature section; the downstream multimodal sensing device is installed at the outlet of the final cooling section. All multimodal sensing devices are synchronously executed based on a unified timestamp, and a mapping relationship between the acquisition time and the axial position of the pipe is established based on the pipe production speed. The feedback results are generated by calculating the feature similarity of the upstream and downstream defect characterization vectors and the rate of change of defect severity. The rate of change of defect severity is quantified by the longitudinal crack density reduction ratio, the reduction magnitude of the temperature gradient standard deviation, and the change in wall thickness fluctuation variance. Adjust the weighting coefficients based on the feedback results, or proportionally increase the effective control parameter adjustment range, or redesign the parameter combination.

7. A pipe forming system based on real-time defect feedback according to claim 1, characterized in that, The actuator includes a thermostatic heating element arrayed inside the thermostatic cavity; The thermodynamic coupling control instruction set includes thermostatic device control instructions, which include axial heating sub-instructions and circumferential heating sub-instructions. The axial heating sub-instructions are used to control the relationship between the heating power of thermostatic heating elements located on the same axis, and the circumferential heating sub-instructions are used to control the relationship between the heating power of thermostatic heating elements located on the same circumference.

8. A pipe forming system based on real-time defect feedback according to claim 1, characterized in that, The final cooling section adopts a gradient cooling structure, including a first-stage atomized spray zone, a second-stage air-mist mixing zone, and a third-stage immersion water tank zone. The cooling rate of each zone satisfies a piecewise function relationship along the pipe axis, and the cooling coefficient of each zone decreases sequentially. Based on the real-time wall thickness profile data collected by the multi-modal sensing device, the cooling intensity ratio of the first-stage atomized spray zone to the third-stage immersion water tank zone is dynamically adjusted.

9. A pipe forming system based on real-time defect feedback according to claim 1, characterized in that, The execution device includes a spraying compensation device, and the corresponding thermo-coupling control instruction set includes a spraying compensation sub-instruction. The spraying compensation sub-instruction is used to control the operation of the spraying compensation device. The spraying compensation sub-instruction is configured with a trigger condition. When the compensation trigger condition is met, the spraying compensation sub-instruction is output. The compensation trigger condition is a value, and the execution is performed when the traction speed of the traction device is lower than the maximum speed allowed for spraying.

Images