HDPE geomembrane welding process and quality detection method

Abstract

The invention relates to the technical field of geomembrane welding, and discloses an HDPE geomembrane welding process and a quality detection method, and the HDPE geomembrane welding process comprises a welding process dynamic control step, a multi-mode quality detection step and a quality judgment and positioning step. According to the method, through real-time monitoring and dynamic closed-loop control in the welding process, the uniformity and consistency of welding seam formation are improved from the source, and defects are effectively prevented; by synchronously executing air pressure detection and machine vision detection and performing intelligent judgment based on a fusion rule, comprehensive, efficient and objective evaluation on the airtightness and the appearance quality of the welding seam is realized, and the accuracy and the reliability of detection are improved; meanwhile, technological parameter self-optimization and defect accurate positioning promote continuous improvement of the welding technology, powerful support is provided for engineering quality tracing and efficient repairing, and the intelligent level of construction and the guarantee capacity of engineering quality are overall improved.

Description

Technical Field

[0001] This invention relates to the field of geomembrane welding technology, specifically a welding process and quality inspection method for HDPE geomembrane. Background Technology

[0002] In water conservancy projects and environmental seepage prevention, the welding quality of HDPE geomembranes directly affects the reliability of the entire seepage prevention system. Currently, the welding construction and quality inspection of HDPE geomembranes are typically independent processes with significant disconnect. The welding process relies heavily on operator experience and pre-set parameters, lacking real-time tracking and dynamic adjustment of key process parameters such as welding temperature and speed. This leads to hidden defects such as incomplete welds and missed welds. In the quality inspection stage, manual visual inspection combined with a separate air-filling method for post-inspection is commonly used. This method is inefficient, highly dependent on subjective factors, and fails to provide a comprehensive, objective, rapid, and accurate assessment of weld quality. Furthermore, the fixed welding process parameters cannot be self-optimized based on environmental changes and historical data, hindering continuous improvement in welding quality and the level of intelligent project management. Therefore, an integrated solution that deeply integrates intelligent control of the welding process with multimodal quality inspection is urgently needed. Summary of the Invention

[0003] The purpose of this invention is to provide a welding process and quality inspection method for HDPE geomembrane to solve the technical problems mentioned in the background art.

[0004] To achieve the above objectives, the present invention discloses the following technical solution: a welding process and quality inspection method for HDPE geomembrane, comprising: The dynamic control steps of the welding process include: real-time monitoring of welding temperature and welding speed during the welding process; comparing the monitored welding temperature and welding speed with preset target ranges for welding temperature and welding speed, respectively; and dynamically adjusting the heating power and travel speed of the welding machine based on the comparison results through a preset control algorithm. The multimodal quality inspection steps include: after welding is completed, simultaneously performing weld sealing inspection and weld appearance visual inspection; wherein, the weld sealing inspection includes filling and pressurizing the closed cavity formed by the double welds and monitoring its pressure change value, and the weld appearance visual inspection includes acquiring weld images and using image recognition algorithms to identify appearance defect features. The quality judgment and positioning steps include: based on the pressure change value and the appearance defect characteristics, outputting the weld quality level according to the preset fusion judgment rules; when the weld is judged to be unqualified, recording the geographical location information or the mileage information relative to the weld start point of the unqualified weld.

[0005] Optionally, in the dynamic control step of the welding process: The control algorithm is a PID control algorithm; The real-time monitoring of the welding temperature is achieved through a non-contact infrared temperature sensor. The real-time monitoring of the welding speed is achieved by a rotary encoder installed on the drive wheel shaft of the welding machine.

[0006] Optionally, in the visual inspection of the weld appearance, the image recognition algorithm is an image recognition model based on a convolutional neural network.

[0007] Optionally, the construction and training of the image recognition model includes: Data preparation: Collect a large number of HDPE geomembrane weld images containing qualified welds and various typical defects to establish a sample library. The typical defects include incomplete welds, missing welds, burns, and wrinkles. Model construction: Convolutional neural network structure is built, which includes at least an input layer, multiple alternating convolutional and pooling layers, a fully connected layer, and an output layer; Model training: The constructed convolutional neural network structure is trained using the sample library, and the network weight parameters are optimized through the backpropagation algorithm until the model loss function converges, thereby obtaining the image recognition model that can automatically extract features from the weld image and identify the defect type.

[0008] Optionally, when acquiring the weld seam image, the equipped ring LED light source is enabled to provide uniform illumination so that the quality of the acquired weld seam image is not affected by ambient light.

[0009] Optionally, the method further includes: The process parameter optimization steps include: after each welding process, storing in association the ambient temperature, ambient humidity, target welding temperature range, target welding speed range, and the final determined weld quality grade collected in real time by environmental sensors to form a process database; using ambient temperature and humidity as input features, and the target welding temperature range and target welding speed range that can obtain the optimal weld quality grade as output targets, training a machine learning algorithm using historical data in the process database to obtain a welding parameter recommendation model; for a new welding task, the welding parameter recommendation model outputs initial suggested values ​​for the optimized target welding temperature range and target welding speed range based on the ambient temperature and humidity collected in real time during the welding process corresponding to the welding task.

[0010] Optionally, the process parameter optimization step further includes: setting a trigger condition, such that when the number of newly added welding process records in the process database reaches a predetermined threshold or the model prediction accuracy is lower than a preset standard, the retraining of the welding parameter recommendation model is initiated to update the model.

[0011] Optionally, in the quality comprehensive judgment step, the fusion judgment rule includes: During the pressure holding time: if the gas pressure stability of the sealed cavity meets the first preset standard, or the total number of appearance defect features identified based on the weld image is zero, the weld quality level is determined to be excellent; if the gas pressure stability of the sealed cavity meets the first preset standard, or the total number of appearance defect features identified based on the weld image is greater than zero, and all identified defect features belong to the preset list of permissible defect types, the weld quality level is determined to be good; if the gas pressure stability of the sealed cavity does not meet the second preset standard, or at least one of the appearance defect features identified based on the weld image does not belong to the list of permissible defect types, the weld quality level is determined to be unqualified, and an audible and visual alarm is triggered.

[0012] Optionally, the first preset standard includes: after the sealed cavity is inflated to an initial pressure of 0.15~0.2MPa and timing begins, during a pressure holding period of 1~5 minutes, the absolute value of the pressure drop between the real-time pressure value of the sealed cavity and the initial pressure is not greater than 0.02MPa. The second preset standard includes: after the sealed cavity is inflated to an initial pressure of 0.15~0.2MPa and timing begins, during a pressure holding period of 1~5 minutes, the absolute value of the pressure drop between the real-time pressure value of the sealed cavity and the initial pressure is not less than 0.05MPa.

[0013] Optionally, the method further includes: Pre-welding pretreatment steps include: cleaning the welding joint area of ​​the HDPE geomembrane to ensure it is clean and dry before welding begins, and using a leveling roller to roll the overlapping area of ​​the geomembrane to eliminate wrinkles.

[0014] Beneficial Effects: The HDPE geomembrane welding process and quality inspection method of this invention improves the uniformity and consistency of weld formation from the source through real-time monitoring and dynamic closed-loop control of the welding process, effectively preventing defects. By simultaneously executing air pressure testing and machine vision inspection, and performing intelligent judgment based on fusion rules, a comprehensive, efficient, and objective assessment of weld sealing and appearance quality is achieved, improving the accuracy and reliability of inspection. Simultaneously, self-optimization of process parameters and precise defect location promote continuous improvement of the welding process and provide strong support for engineering quality traceability and efficient repair, thereby comprehensively improving the level of intelligent construction and the ability to ensure engineering quality. Attached Figure Description

[0015] To more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0016] Figure 1 This is a flowchart illustrating the welding process and quality inspection method for HDPE geomembrane provided in an embodiment of the present invention. Detailed Implementation

[0017] To facilitate understanding of the technical solutions provided in the embodiments of this application, the background technology involved in the embodiments of this application will be described below.

[0018] HDPE geomembrane (high-density polyethylene geomembrane), as an engineering material with excellent impermeability, chemical stability, and anti-aging properties, has been widely used in key areas such as water conservancy projects (e.g., reservoir dam seepage prevention, river ecological restoration seepage prevention layers) and environmental protection projects (e.g., municipal solid waste landfill seepage prevention systems, hazardous waste disposal site linings, and sewage treatment plant sedimentation tank seepage prevention). In these applications, HDPE geomembrane is not laid alone but needs to be welded to form a continuous and complete seepage prevention system. The quality of the weld directly determines the reliability of the entire seepage prevention system. Defects in the weld can lead to water leakage and pollutant leakage, which not only damages the surrounding ecological environment and wastes water resources but can also cause structural safety hazards in severe cases (e.g., piping caused by dam seepage and soil pollution caused by landfill leachate leakage). Therefore, the welding quality control of HDPE geomembrane is a core aspect of engineering construction.

[0019] However, the welding construction and quality inspection of HDPE geomembranes in the industry are generally separated, making it difficult to form a closed loop for quality control. The specific shortcomings are reflected in the following three aspects: First, the welding process relies on experience and lacks dynamic control capabilities. In current welding construction, the setting of process parameters (such as welding temperature and welding speed) largely relies on the operator's past experience. Typically, fixed parameters are preset before construction based on general environmental conditions and then continuously executed, lacking a real-time monitoring and dynamic adjustment mechanism for key parameters during the welding process. For example, welding temperature control often relies on the theoretical temperature setting built into the welding machine, rather than the actual temperature applied to the geomembrane welding joint. When ambient temperature fluctuates (e.g., high temperatures at noon in summer, low temperatures in winter), geomembrane thickness changes, or the welding machine's heating components age, the preset fixed temperature may be too high, leading to geomembrane burns and excessive melting, or too low, resulting in insufficient weld fusion and incomplete welds. Simultaneously, the welding speed is manually controlled by the operator's walking mechanism, which is prone to uneven weld width and reduced bonding strength due to human error (e.g., inconsistent walking speed). This experience-driven static parameter control mode cannot respond promptly to dynamic changes during construction, becoming a major source of hidden defects such as incomplete welds, missed welds, and burns. If these defects are not detected in subsequent inspections, they will become long-term hidden dangers to the seepage prevention system.

[0020] Second, quality inspection is inefficient and highly subjective, making it difficult to achieve a comprehensive assessment. In the quality inspection stage after welding, existing technologies mostly adopt a post-weld sampling inspection mode of manual visual inspection plus separate gas inflation and pressure holding test, which has obvious limitations: The shortcomings of manual visual inspection: It relies on the naked eye of the inspector to judge the appearance of the weld (such as whether there are wrinkles, undercut, or missing weld marks). It is greatly affected by subjective factors such as the inspector's experience, visual fatigue, and ambient light. For minor appearance defects (such as localized incomplete weld marks less than 1mm wide or slight surface wrinkles), it is easy to miss or misjudge, and the objectivity and consistency of the inspection results cannot be guaranteed. The shortcomings of the single inflation and pressure holding test: This method requires inflation and pressure holding of each closed cavity formed by the double weld seams, which takes a long time for a single test (usually 5-10 minutes per segment). Therefore, in actual projects, sampling inspection is often used (the sampling ratio is usually only 5%-10%), which cannot cover all weld seams and there is a risk of hidden defects in the unsampled segments. At the same time, this test can only determine whether there is leakage (sealing) in the weld seam, and cannot simultaneously assess the appearance quality. It needs to be performed separately from visual inspection, resulting in low overall inspection efficiency and difficulty in meeting the time requirements of large-scale projects.

[0021] This phased, incomplete, and subjective testing mode cannot achieve a comprehensive assessment of weld sealing and appearance quality, nor can it meet the engineering requirements for testing efficiency and accuracy.

[0022] Third, the inability of process parameters to self-optimize restricts the level of intelligent management. In existing technologies, the setting and adjustment of welding process parameters lack data support. During construction, a database linking environmental parameters, process parameters, and weld quality has not been established. This prevents the optimization of parameters based on dynamic changes in the actual construction environment (such as variations in ambient temperature and humidity), and also hinders iterative improvement based on historical welding data (such as weld pass rate under a specific temperature / speed combination). For example, when ambient humidity increases from 30% to 70%, existing technologies still use the original welding temperature and speed, failing to recognize that high humidity may cause moisture to adhere to the geomembrane surface, necessitating an appropriate increase in temperature to ensure welding effectiveness. Furthermore, the lack of accurate recording of the locations of substandard welds in project management (e.g., failure to link them to geographical location or mileage information) necessitates re-inspection during later maintenance, further reducing project management efficiency.

[0023] In summary, existing HDPE geomembrane welding and testing technologies suffer from drawbacks such as lack of dynamic process control, lack of comprehensive testing, and lack of parameter self-optimization, resulting in poor weld quality stability, low testing efficiency, and insufficient engineering intelligence.

[0024] Because of these problems, this embodiment provides an integrated solution that can deeply integrate dynamic control of the welding process, multimodal quality detection, intelligent judgment and parameter optimization, so as to realize the transformation of welding process from experience-driven to data-driven, and comprehensively improve the welding quality of HDPE geomembrane and the efficiency of project management.

[0025] To enable those skilled in the art to better understand the present application, the technical solutions in the embodiments of the present application will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present application, and not all embodiments. Based on the embodiments in the present application, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present application. Secondly, in this document, the term "comprising" is intended to cover non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements includes not only those elements, but also other elements not expressly listed, or elements inherent to such a process, method, article, or apparatus. Without further limitations, an element defined by the phrase "comprising..." does not exclude the presence of other identical elements in the process, method, article, or apparatus that includes the element.

[0026] This embodiment provides a welding process and quality inspection method for HDPE geomembrane, such as Figure 1 As shown, the steps include, in sequence: dynamic control of the welding process, multimodal quality inspection, and quality judgment and positioning, as detailed below.

[0027] The dynamic control steps of the welding process include: real-time monitoring of welding temperature and welding speed during the welding process; comparing the monitored welding temperature and welding speed with preset target ranges for welding temperature and welding speed, respectively; and dynamically adjusting the heating power and travel speed of the welding machine based on the comparison results through a preset control algorithm.

[0028] In practice, real-time monitoring of welding temperature is achieved by acquiring the actual temperature signal of the weld area using a non-contact infrared temperature sensor. The sensor is installed below the heating element of the welding machine, 5-10 cm from the weld surface, with a sampling frequency of 10Hz to avoid direct contact with the high-temperature components and ensure real-time temperature acquisition. Welding speed is acquired using a rotary encoder mounted on the drive wheel shaft of the welding machine. The encoder is fixed to the drive wheel shaft via a coupling, outputting a fixed number of pulses per rotation. The travel distance is calculated based on the drive wheel diameter, and the welding speed is then calculated using the time difference. The sampling frequency is set to 20Hz to ensure accurate speed detection. The preset target welding temperature and welding speed ranges need to be determined based on the thickness of the HDPE geomembrane (e.g., 1.0-3.0 mm), material type, and engineering design requirements. For example, for a 2.0 mm thick HDPE geomembrane, the target welding temperature range can be set to 200-220℃, and the target welding speed range can be set to 2-3 m / min.

[0029] Secondly, the control algorithm needs to have real-time responsiveness and be able to quickly output adjustment commands based on parameter deviations. Therefore, the control algorithm in this embodiment is a PID control algorithm. The proportional coefficient, integral time constant, and derivative time constant of the PID algorithm need to be determined by debugging according to the welding machine model and geomembrane characteristics. For example, the proportional coefficient is set to 1.0-2.0, the integral time constant is set to 0.5-1.0s, and the derivative time constant is set to 0.1-0.3s to achieve stability and speed of parameter adjustment.

[0030] The multimodal quality inspection steps include: after welding is completed, simultaneously performing weld sealing inspection and weld appearance visual inspection; wherein, weld sealing inspection includes filling and pressurizing the closed cavity formed by the double welds and monitoring its pressure change value, and weld appearance visual inspection includes acquiring weld images and using image recognition algorithms to identify appearance defect features.

[0031] In practice, inflation can be achieved using a miniature air pump, and pressure monitoring is conducted using a high-precision pressure sensor. The sealed cavity must be properly sealed at both ends beforehand. Weld seam images are captured using a high-definition industrial camera mounted at the tail of the welding machine, which moves synchronously with the machine.

[0032] The quality assessment and positioning steps include: based on the pressure change value and appearance defect characteristics, outputting the weld quality level according to the preset fusion assessment rules; when the weld is determined to be unqualified, recording the geographical location information or the mileage information relative to the weld start point of the unqualified weld.

[0033] In practice, quality grades are divided into three categories: excellent, good, and unqualified. The integrated judgment rules need to comprehensively consider the influence weight of airtightness and appearance defects. Geographical location information can be collected through a GPS positioning module, and the mileage information relative to the weld start point is calculated by accumulating the welding machine's travel distance.

[0034] Based on the above, the HDPE geomembrane welding process and quality inspection method of this embodiment constructs an integrated process of dynamic control, multimodal detection, and intelligent judgment, forming a closed-loop management from welding process to quality assessment. During the welding process, the welding temperature and speed are monitored in real time and compared with preset ranges. The parameters are dynamically adjusted using a control algorithm. The principle is based on closed-loop control theory, which corrects parameter deviations in a timely manner and avoids defects such as incomplete welding and burns caused by excessively high / low temperatures or excessively fast / slow speeds, thereby improving the uniformity and consistency of weld formation. Multimodal detection simultaneously performs airtightness testing and visual inspection. The principle is to combine the complementarity of physical pressure testing (reflecting seepage prevention performance) and machine vision testing (reflecting appearance quality), breaking through the limitations of a single detection method and achieving comprehensive coverage of weld quality. Quality judgment adopts a fusion rule, which integrates the output levels of the two types of test data. When non-conforming, the location information is accurately recorded. The principle is based on multi-source data fusion decision-making, ensuring that the judgment results are objective and reliable, and providing accurate positioning basis for subsequent repairs. This effectively solves the problems of process loss of control, one-sided detection, and subjective judgment in traditional processes, and improves the overall welding quality and engineering management efficiency.

[0035] As an optional implementation method in this embodiment, the image recognition algorithm in the visual inspection of weld appearance is an image recognition model based on a convolutional neural network. This image recognition model needs to be customized for the texture features and defect types of HDPE geomembrane welds to ensure sensitivity in identifying typical defects such as incomplete welds and leaks. The model input is the pixel matrix of the weld image, and the output is the recognition result and confidence level of the defect features.

[0036] Based on the above, the HDPE geomembrane welding process and quality inspection method of this embodiment utilizes a convolutional neural network-based image recognition model for weld appearance inspection. The technical principle is to automatically extract deep features (such as grayscale distribution, texture changes, and contour anomalies) from weld images using a convolutional neural network, eliminating the need for manually designed feature extraction operators. Compared to traditional image recognition algorithms (such as threshold segmentation and edge detection), it better adapts to the complex backgrounds and diverse defects of weld images. Through a customized model, typical defect features such as incomplete welds, missing welds, and burns can be accurately identified, effectively avoiding the subjectivity and missed detection problems of manual visual inspection, while improving the automation and efficiency of appearance inspection. This model can quickly process acquired weld images and output defect identification results in real time, providing timely and reliable appearance data support for subsequent quality judgment, further enhancing the comprehensiveness and accuracy of multimodal detection.

[0037] Based on the aforementioned limitations of the image recognition model, as a further optional implementation method of this embodiment, the construction and training of the image recognition model includes: Data preparation: Collect a large number of HDPE geomembrane weld images containing qualified welds and various typical defects to establish a sample library. Typical defects include incomplete welds, missing welds, burns, and wrinkles. Model construction: Convolutional neural network structure is built. The convolutional neural network structure includes at least an input layer, multiple alternating convolutional and pooling layers, a fully connected layer, and an output layer. Model training: The constructed convolutional neural network structure is trained using a sample library. The network weight parameters are optimized through backpropagation algorithm until the model loss function converges, resulting in an image recognition model that can automatically extract features from weld images and identify defect types.

[0038] In practice, the image acquisition in the sample library needs to cover weld samples under different lighting conditions and welding parameters, and the number of samples for each defect type should not be less than 500, to ensure the diversity and representativeness of the sample library.

[0039] Secondly, in the convolutional neural network structure, the convolutional layer uses a 3×3 convolutional kernel, the pooling layer uses a 2×2 max pooling kernel, the fully connected layer has 2-3 layers, and the output layer uses the softmax activation function to output the probability distribution of defect types.

[0040] Furthermore, the network weight parameters are optimized using stochastic gradient descent, with a learning rate of 0.001-0.01 and 100-200 iterations. The applicable model loss function is the cross-entropy loss function, and the convergence criterion is a loss value below 0.05.

[0041] Based on the above, the HDPE geomembrane welding process and quality inspection method in this embodiment ensures that the image recognition model has high recognition accuracy and stability through systematic sample preparation, structural design, and training optimization. The data preparation stage collects diverse samples to avoid model overfitting and adapt it to weld images under different engineering scenarios. The model construction employs alternating convolutional and pooling layers. The convolutional layers are responsible for extracting local features of the weld image, the pooling layers achieve feature dimensionality reduction and retain key information, and the fully connected layers map the extracted features to defect types. This structural design can efficiently capture the essential characteristics of weld defects. During training, the weight parameters are optimized through the backpropagation algorithm, causing the loss function to gradually converge and ensuring that the model can accurately learn the differences between qualified and defective welds. The finally trained image recognition model can automatically and accurately identify multiple defect types, providing core algorithmic support for weld appearance visual inspection and improving the intelligence and reliability of appearance inspection.

[0042] Based on the aforementioned image recognition model, as another further optional implementation of this embodiment, when acquiring weld seam images, the equipped ring LED light source is enabled to provide uniform illumination so that the quality of the acquired weld seam images is not affected by ambient light.

[0043] In practice, the ring-shaped LED light source is installed around the lens of the high-definition industrial camera. The power of the light source is set to 10-20W, and the light intensity can be adaptively adjusted according to the ambient light intensity to ensure uniform lighting in the weld area without obvious shadows or reflections.

[0044] Based on the above, the HDPE geomembrane welding process and quality inspection method of this embodiment solves the problem of ambient light interference in weld image acquisition by configuring a ring LED light source. The technical principle is to utilize the surround illumination characteristics of the ring light source to provide uniform and stable illumination to the weld area, eliminating reflections caused by direct strong light and shadows in low-light environments, ensuring uniform grayscale distribution and clear details in the weld image. In actual engineering, changes in ambient light (such as strong light on sunny days, weak light on cloudy days, and insufficient lighting at night) can severely affect image quality, causing subsequent recognition models to be unable to accurately extract defect features. The ring LED light source, by adaptively adjusting the light intensity, ensures that the acquired weld images maintain consistent high quality under different ambient light conditions, providing reliable input data for the image recognition model, thereby improving the accuracy and stability of defect identification, avoiding missed detections and misjudgments due to poor image quality, and ensuring the effectiveness of visual inspection.

[0045] As an optional implementation of this embodiment, the HDPE geomembrane welding process and quality inspection method of this embodiment further includes: The process parameter optimization steps include: after each welding process, the ambient temperature, ambient humidity, target welding temperature range, target welding speed range, and final weld quality grade, all collected in real time by environmental sensors, are associated and stored to form a process database; using ambient temperature and humidity as input features, and the target welding temperature range and target welding speed range that achieve the optimal weld quality grade as output targets, a machine learning algorithm is trained using historical data from the process database to obtain a welding parameter recommendation model; for a new welding task, the welding parameter recommendation model outputs initial suggested values ​​for the optimized target welding temperature range and target welding speed range based on the ambient temperature and humidity collected in real time during the welding process corresponding to that welding task.

[0046] In practice, ambient temperature is collected using a thermocouple temperature sensor, with a measurement range of -10℃ to 50℃ and an accuracy of ±0.5℃. Ambient humidity is collected using a capacitive humidity sensor, with a measurement range of 0% to 100%RH and an accuracy of ±3%RH.

[0047] Secondly, the process database adopts a relational database, and each record contains fields such as timestamp, environmental parameters, process parameters, and quality grade, which facilitates data query and retrieval.

[0048] In addition, machine learning algorithms such as random forests and neural networks can be used. During training, the data is divided into training and test sets in a 7:3 ratio.

[0049] Based on the above, the HDPE geomembrane welding process and quality inspection method of this embodiment achieves adaptive optimization of welding parameters through a process parameter optimization step. Its technical principle is based on the correlation between environmental parameters, process parameters, and quality levels. Historical data is used to train a machine learning model, enabling the model to learn the optimal combination of process parameters under different environmental conditions. Environmental sensors accurately collect temperature and humidity data to ensure the accuracy of input features; the process database stores complete historical records, providing sufficient data support for model training; the machine learning algorithm establishes a mapping relationship between the environment and optimal parameters by mining potential patterns in the data. For new welding tasks, the model can quickly output optimized initial parameter suggestions based on real-time environmental parameters, avoiding the limitations of relying on experience-based preset parameters in traditional processes. This allows process parameters to dynamically adapt to environmental changes, improving the stability and consistency of weld quality from the source. Simultaneously, this step promotes the transformation of welding processes from experience-driven to data-driven, providing data support and technical assurance for continuous process improvement.

[0050] Based on the aforementioned process parameter optimization steps, as an optional implementation method of this embodiment, the process parameter optimization steps further include: setting trigger conditions, when the number of newly added welding process records in the process database reaches a predetermined threshold or the model prediction accuracy is lower than a preset standard, starting the retraining of the welding parameter recommendation model to update the model.

[0051] In practice, the preset threshold for newly added welding process records in the process database is set according to the project scale and data requirements, such as 1000 records, to ensure that the new data can provide effective information for model updates. The preset accuracy standard for the model prediction accuracy is set according to the project quality requirements, such as 90%, to ensure that the parameter recommendations output by the model are reliable.

[0052] Based on the above, the HDPE geomembrane welding process and quality inspection method in this embodiment ensures the long-term effectiveness and adaptability of the welding parameter recommendation model by setting trigger conditions to initiate model retraining. The technical principle is that as engineering data accumulates and environmental conditions change, the prediction accuracy of the original model may decrease. By setting reasonable trigger conditions, the model parameters are updated in a timely manner using newly added data. When the number of new records reaches a predetermined threshold, retraining can incorporate more diverse environmental and process combinations, enabling the model to adapt to the needs of different engineering scenarios. When the model's prediction accuracy is lower than a preset standard, retraining can correct the model's deviation and improve prediction reliability. This dynamic update mechanism avoids the problem of inaccurate parameter recommendations caused by model solidification, ensuring that the model can always output optimal parameter recommendations that match the current environment and process, continuously providing support for improving welding quality and further enhancing the scientific nature and timeliness of process parameter optimization.

[0053] As an optional implementation of this embodiment, the fusion judgment rules in the quality comprehensive judgment step include: During the pressure holding time: if the gas pressure stability of the closed cavity meets the first preset standard, or the total number of appearance defect features identified based on the weld image is zero, the weld quality level is determined to be excellent; if the gas pressure stability of the closed cavity meets the first preset standard, or the total number of appearance defect features identified based on the weld image is greater than zero, and all identified defect features belong to the preset list of permissible defect types, the weld quality level is determined to be good; if the gas pressure stability of the closed cavity does not meet the second preset standard, or at least one of the appearance defect features identified based on the weld image is a defect feature that does not belong to the list of permissible defect types, the weld quality level is determined to be unqualified, and an audible and visual alarm is triggered.

[0054] In practice, the list of defect types is allowed to be formulated according to engineering specifications and seepage prevention requirements, such as tiny wrinkles less than 0.5mm wide and less than 5mm long that do not affect the seepage prevention performance of the weld. An audible and visual alarm is installed on the welding machine's control panel, with an alarm sound intensity of no less than 80dB and a flashing red light to ensure that operators can detect the alarm promptly.

[0055] Based on the above, the HDPE geomembrane welding process and quality inspection method of this embodiment achieves precise grading of weld quality through clear fusion judgment rules. Its technical principle is to comprehensively assess the weld's sealing performance (reflecting the core seepage prevention performance) and appearance quality (reflecting the level of construction technology) and categorize it into three levels: excellent, good, and unqualified. A list of permissible defect types is also established to balance actual project needs with quality standards. For the excellent level, sealing performance must meet standards and there must be no appearance defects, ensuring the weld possesses optimal seepage prevention performance and appearance quality. For the good level, minor appearance defects that do not affect seepage prevention performance are allowed, adapting to the actual situation of a small number of non-fatal defects in projects. For the unqualified level, if sealing performance fails to meet standards or serious appearance defects exist, immediate judgment and alarm are triggered to prevent unqualified welds from being put into use. This grading judgment method ensures the comprehensiveness and objectivity of quality assessment and provides timely early warning of unqualified defects through audible and visual alarms, effectively improving the pertinence and effectiveness of quality control and providing a clear judgment basis for project quality control.

[0056] Based on the aforementioned fusion determination rules, as a further optional implementation method of this embodiment, the first preset standard includes: after the closed cavity is inflated to an initial pressure of 0.15~0.2MPa and timing begins, during the pressure holding period of 1~5 minutes, the absolute value of the pressure drop between the real-time pressure value of the closed cavity and the initial pressure is not greater than 0.02MPa. The second preset standard includes: after the sealed cavity is inflated to an initial pressure of 0.15~0.2MPa and the timing is started, during the pressure holding period of 1~5 minutes, the absolute value of the pressure drop between the real-time pressure value of the sealed cavity and the initial pressure is not less than 0.05MPa.

[0057] In practice, the initial pressure selection must balance the detection sensitivity and the bearing capacity of the HDPE geomembrane, avoiding excessive pressure that could damage the membrane and insufficient pressure that would lead to inaccurate detection. Secondly, the holding time should be set according to the project's testing efficiency requirements, balancing detection accuracy and construction progress.

[0058] Based on the above, the HDPE geomembrane welding process and quality inspection method of this embodiment provides a quantitative basis for weld tightness assessment by clarifying the pressure standard for tightness testing. Its technical principle is to objectively judge the sealing performance of the weld by setting specific initial pressure, holding time, and pressure change thresholds. The initial pressure is set at 0.15-0.2 MPa, ensuring that the closed cavity produces a significant pressure change to reflect leakage, without damaging the geomembrane due to excessive pressure. The holding time is set at 1-5 minutes, allowing the pressure to stabilize and accurately capture the pressure drop caused by leakage, without affecting construction efficiency due to excessive time. The first preset standard ensures that the tightness of excellent and good grade welds meets the standard through a strict pressure drop threshold (≤0.02 MPa), while the second preset standard quickly identifies serious leakage defects through a clear non-compliance threshold (≥0.05 MPa). This quantitative standard avoids the bias of subjective judgment in traditional tightness testing, ensures the consistency and reliability of results under different testing scenarios, provides accurate tightness data support for quality fusion judgment, and effectively protects the core performance of the seepage prevention system.

[0059] As an optional implementation of this embodiment, the HDPE geomembrane welding process and quality inspection method of this embodiment further includes: Pre-welding pretreatment steps include: cleaning the welding joint area of ​​the HDPE geomembrane to ensure it is clean and dry before welding begins, and using a leveling roller to roll the overlapping area of ​​the geomembrane to eliminate wrinkles.

[0060] In practice, cleaning methods include using a brush to remove surface dust and impurities, drying the surface with a hairdryer, and wiping away oil stains with alcohol if necessary, ensuring that there are no foreign objects in the joint area that could affect welding quality. Secondly, the leveling roller is made of rubber and can have a diameter of 100-150mm. The rolling pressure can be controlled by adjusting the roller's weight to ensure that wrinkles are completely smoothed without damaging the geomembrane. Based on the above, the HDPE geomembrane welding process and quality inspection method of this embodiment lays the foundation for high-quality welding through pre-welding pretreatment steps. The technical principle is to eliminate impurities, moisture, and wrinkles in the welding joint area, preventing these factors from causing defects during welding. Cleaning the joint area removes dust, oil, and moisture, preventing impurities from being trapped in the weld and forming air bubbles during welding. Moisture evaporation causes porosity in the weld, thus avoiding defects such as incomplete welds and missed welds. The leveling roller rolling eliminates wrinkles in the overlapping area, ensuring a smooth geomembrane overlap, uniform heating and consistent pressure during welding, avoiding problems such as uneven weld width and insufficient bonding strength caused by wrinkles. Pre-welding pretreatment reduces the causes of defects from the source, providing a good foundation for dynamic control of the subsequent welding process, further improving the quality of weld formation, ensuring the effectiveness of subsequent welding and inspection steps, and overall guaranteeing the reliability of the seepage prevention system.

[0061] In the embodiments provided by this invention, it should be understood that the embodiments described herein can be implemented in hardware, software, firmware, middleware, code, or any suitable combination thereof. For hardware implementation, the processor can be implemented in one or more of the following: application-specific integrated circuits (ASICs), digital signal processors (DSPs), digital signal processing devices (DSPDs), programmable logic devices (PLDs), field-programmable gate arrays (FPGAs), processors, controllers, microcontrollers, microprocessors, other electronic units designed to implement the functions described herein, or combinations thereof. For software implementation, some or all of the processes of the embodiments can be performed by a computer program instructing the associated hardware. During implementation, the program can be stored in a computer-readable storage medium or transmitted as one or more instructions or code on a computer-readable storage medium. Computer-readable storage media include computer storage media and communication media, wherein communication media include any medium that facilitates the transmission of a computer program from one place to another. Storage media can be any available medium accessible to a computer. Computer-readable storage media can include, but are not limited to, RAM, ROM, EEPROM, CD-ROM or other optical disk storage, magnetic disk storage media or other magnetic storage devices, or any other medium capable of carrying or storing desired program code having the form of instructions or data structures and accessible to a computer.

[0062] Finally, it should be noted that the above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. Although the present invention has been described in detail with reference to the foregoing embodiments, those skilled in the art can still modify the technical solutions described in the foregoing embodiments or make equivalent substitutions for some of the technical features. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of the present invention should be included within the protection scope of the present invention.

Claims

1. A method for HDPE geomembrane welding process and quality detection, characterized in that, The method comprises: a welding process dynamic control step, comprising: monitoring welding temperature and welding speed in real time during the welding process; comparing the monitored welding temperature and welding speed with preset welding temperature target interval and welding speed target interval respectively, and dynamically adjusting the heating power and walking speed of the welding machine through a preset control algorithm based on the comparison result; a multi-modal quality detection step, comprising: synchronously performing weld tightness detection and weld appearance visual detection after the welding is completed; wherein the weld tightness detection comprises inflating and maintaining pressure of a closed cavity formed by double welds and monitoring pressure change value thereof, and the weld appearance visual detection comprises collecting weld images and identifying appearance defect features by using an image recognition algorithm; a quality determination and positioning step, comprising: outputting a weld quality grade according to a preset fusion determination rule based on the pressure change value and the appearance defect features; and recording geographical position information or mileage information relative to a weld starting point of the unqualified weld when the unqualified weld is determined.

2. The HDPE geomembrane welding process and quality detection method according to claim 1, characterized in that, In the welding process dynamic control step: the control algorithm is a PID control algorithm; the real-time monitoring of the welding temperature is realized by using a non-contact infrared temperature sensor; the real-time monitoring of the welding speed is realized by using a rotary encoder installed on a welding machine driving wheel shaft.

3. The HDPE geomembrane welding process and quality detection method according to claim 1, characterized in that, In the weld appearance visual detection, the image recognition algorithm is an image recognition model based on a convolutional neural network.

4. The HDPE geomembrane welding process and quality detection method according to claim 3, characterized in that, Construction and training of the image recognition model comprises: data preparation: collecting a large number of HDPE geomembrane weld images containing qualified welds and various typical defects, establishing a sample library, and the typical defects include virtual welding, missing welding, scalding and wrinkling; model construction: building a convolutional neural network structure, the convolutional neural network structure at least comprising an input layer, a plurality of alternately arranged convolutional layers and pooling layers, a fully connected layer and an output layer; model training: training the constructed convolutional neural network structure using the sample library, optimizing network weight parameters through a back propagation algorithm until the model loss function converges, and obtaining the image recognition model capable of automatically extracting features and identifying defect types from the weld images.

5. The HDPE geomembrane welding process and quality detection method according to claim 3, characterized in that, When the weld images are collected, the equipped ring-shaped LED light source is enabled to provide uniform illumination, so that the quality of the collected weld images is not disturbed by ambient light.

6. The HDPE geomembrane welding process and quality detection method according to claim 1, characterized in that, The method further comprises: The process parameter optimization step includes: after each welding process, the environmental temperature collected in real time by the environmental sensor, the environmental humidity collected in real time, the welding temperature target interval used in welding, the welding speed target interval used in welding, and the final determined weld quality grade are associatedly stored to form a process database; the environmental temperature and the environmental humidity are taken as input characteristics, the welding temperature target interval and the welding speed target interval capable of obtaining the optimal weld quality grade are taken as output targets, historical data in the process database is used to train a machine learning algorithm to obtain a welding parameter recommendation model; for a new welding task, the welding parameter recommendation model outputs initial recommended values of the optimized welding temperature target interval and the welding speed target interval according to the environmental temperature and the environmental humidity collected in real time in the welding process corresponding to the welding task.

7. The HDPE geomembrane welding process and quality detection method according to claim 6, characterized in that, The process parameter optimization step further includes: setting a trigger condition, when the number of newly added welding process records in the process database reaches a predetermined threshold or the model prediction accuracy is lower than a preset standard, retraining the welding parameter recommendation model to update the model.

8. The HDPE geomembrane welding process and quality detection method according to claim 1, characterized in that, In the quality comprehensive determination step, the fusion determination rule includes: During the holding time: if the gas pressure stability of the closed cavity meets the first preset standard, or the total number of appearance defect features identified based on the weld image is zero, the weld quality grade is determined to be excellent; if the gas pressure stability of the closed cavity meets the first preset standard, or the total number of appearance defect features identified based on the weld image is greater than zero, and all identified defect features belong to the list of preset allowed defect types, the weld quality grade is determined to be good; if the gas pressure stability of the closed cavity does not meet the second preset standard, or there is at least one defect feature in the appearance defect features identified based on the weld image that does not belong to the list of allowed defect types, the weld quality grade is determined to be unqualified, and an audible and light alarm is triggered.

9. The HDPE geomembrane welding process and quality detection method according to claim 8, characterized in that, The first preset standard includes: after the closed cavity is inflated to an initial pressure of 0.15-0.2 MPa and the timing starts, the absolute value of the pressure drop between the real-time pressure value of the closed cavity and the initial pressure is not greater than 0.02 MPa within a holding time period of 1-5 minutes. The second preset standard includes: after the closed cavity is inflated to an initial pressure of 0.15-0.2 MPa and the timing starts, the absolute value of the pressure drop between the real-time pressure value of the closed cavity and the initial pressure is not less than 0.05 MPa within a holding time period of 1-5 minutes.

10. The HDPE geomembrane welding process and quality detection method according to claim 1, characterized in that, The method further includes: The pre-welding pretreatment step includes: before welding starts, cleaning the welding joint area of the HDPE geomembrane to ensure cleanliness and dryness, and using a flattening roller to roll the lapped area of the geomembrane to eliminate wrinkles.

Images