Welding quality analysis method and system for welding seam detection of engineering vehicle

Abstract

The invention relates to a welding quality analysis method and system for engineering vehicle welding seam detection, and relates to the field of quality analysis. Carrying out vehicle specification quality influence analysis, railway track detection quality influence analysis and driving vibration analysis according to the plurality of predicted deformations; obtaining the probability of track specification abnormity of a track of a target railway section corresponding to the railway engineering vehicle, calculating to obtain a track abnormity rate, and correcting the track detection influence parameter to obtain a corrected track detection influence parameter; and calculating according to the vehicle specification influence parameters, the corrected track detection influence parameters and the track detection vibration parameters to obtain a welding quality detection result of the railway engineering vehicle. The technical problem that potential safety hazards are caused by insufficient detection quality due to low efficiency and difficulty in accurately judging internal defects and stress deformation conditions of welding seams in welding seam quality detection is solved.

Description

Technical Field

[0001] The present invention relates to the field of quality analysis, and particularly to a welding quality analysis method and system for detecting welds of engineering vehicles. Background Art

[0002] During the manufacturing and maintenance of railway engineering vehicles, the welding quality of welds is a key factor in ensuring the structural integrity and safety of the vehicles. Most traditional weld quality detection methods rely on manual visual inspection or simple measuring tools. These methods are not only inefficient but also difficult to accurately judge the internal defects and stress deformation conditions of the welds. With the rapid development of computer vision and artificial intelligence technologies, image processing and machine learning algorithms have been widely applied in the field of weld quality detection. By collecting images of the weld area and using advanced algorithms to analyze and process the images, a rapid and accurate assessment of the weld quality can be achieved. However, existing weld quality detection methods often only focus on the defect conditions of the welds themselves. During the operation of railway engineering vehicles, the weld area is subjected to various complex forces, resulting in stress deformation of the welds. If the stress deformation of the welds exceeds a certain allowable range, it will not only affect the structural strength and stability of the vehicle but also have an adverse impact on the track detection accuracy and driving safety. Summary of the Invention

[0003] In view of the technical problem that the existing weld quality detection has low efficiency and is difficult to accurately judge the internal defects and stress deformation conditions of the welds, resulting in insufficient detection quality and potential safety hazards, the present invention provides a welding quality analysis method and system for detecting welds of engineering vehicles to solve this problem.

[0004] The technical solution of the present invention to solve the above technical problems is as follows:

[0005] In a first aspect, the present invention provides a welding quality analysis method for detecting welds of engineering vehicles. The method includes: after the welding of a railway engineering vehicle is completed, collecting weld images of multiple weld areas, predicting welding stress deformation, and obtaining multiple predicted deformation amounts; according to the multiple predicted deformation amounts, analyzing the influence of the vehicle specifications of the railway engineering vehicle to obtain vehicle specification influence parameters, and analyzing the influence on railway track detection quality and driving vibration to obtain track detection influence parameters and track detection vibration parameters; obtaining the probability of track specification anomalies occurring in the target railway section corresponding to the railway engineering vehicle, calculating to obtain the track anomaly rate, and correcting the track detection influence parameters to obtain corrected track detection influence parameters; calculating the welding quality detection result of the railway engineering vehicle according to the vehicle specification influence parameters, corrected track detection influence parameters, and track detection vibration parameters.

[0006] Second aspect, the present invention provides a welding quality analysis system for weld detection of engineering vehicles, the system comprising: an image acquisition module, configured to, after the welding of a railway engineering vehicle is completed, acquire weld images of a plurality of weld regions, perform prediction of welding stress deformation, and obtain a plurality of predicted deformation amounts; a quality analysis module, configured to perform analysis on the influence of the vehicle specifications quality of the railway engineering vehicle according to the plurality of predicted deformation amounts to obtain vehicle specification influence parameters, and perform analysis on the influence of railway track detection quality and driving vibration analysis to obtain track detection influence parameters and track detection vibration parameters; a probability calculation module, configured to obtain the probability of track specification anomalies occurring in the target railway section corresponding to the railway engineering vehicle, calculate and obtain a track anomaly rate, and correct the track detection influence parameters to obtain corrected track detection influence parameters; and a result acquisition module, configured to calculate and obtain a weld quality detection result of the railway engineering vehicle according to the vehicle specification influence parameters, the corrected track detection influence parameters, and the track detection vibration parameters.

[0007] The beneficial effects of the present invention are as follows: By acquiring weld images to perform prediction of welding stress deformation, and combining multi-dimensional analysis such as vehicle specifications, track detection quality, and driving vibration, and at the same time considering the track anomaly rate of the target railway section for parameter correction, it is finally possible to comprehensively and accurately evaluate the welding quality of railway engineering vehicles and improve the efficiency and accuracy of weld detection. Description of the Drawings

[0008] Figure 1 It is a schematic flow chart of a welding quality analysis method for weld detection of an engineering vehicle provided by the present invention.

[0009] Figure 2 It is a schematic structural diagram of a welding quality analysis system for weld detection of an engineering vehicle provided by the present invention.

[0010] Description of the reference numerals: Image acquisition module 11, quality analysis module 12, probability calculation module 13, result acquisition module 14. Detailed Embodiments

[0011] Next, the technical solutions in the embodiments of the present invention will be clearly and completely described in conjunction with the accompanying drawings in the embodiments of the present invention. Obviously, the described embodiments are only a part of the embodiments of the present invention, rather than all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative efforts fall within the protection scope of the present invention.

[0012] In the description of the present invention, the terms "first" and "second" are only used for descriptive purposes and cannot be construed as indicating or implying relative importance or implicitly specifying the quantity of the indicated technical features. Thus, the features defined with "first" and "second" may explicitly or implicitly include one or more of the said features. In the description of the present invention, "a plurality of" means two or more unless otherwise specifically defined.

[0013] In the description of the present invention, the term "for example" is used to mean "serving as an example, illustration, or explanation". Any embodiment described as "for example" in the present invention is not necessarily construed as being more preferred or advantageous than other embodiments. The following description is given to enable any person skilled in the art to implement and use the present invention. In the following description, details are set forth for purposes of explanation. It should be understood that those of ordinary skill in the art can recognize that the present invention can be implemented without these specific details. In other instances, well-known structures and processes are not described in detail to avoid unnecessary details from obscuring the description of the present invention. Therefore, the present invention is not intended to be limited to the embodiments shown, but is to be accorded the widest scope consistent with the principles and features disclosed herein.

[0014] Embodiment 1:

[0015] As Figure 1 shown, the embodiment of the present invention provides a welding quality analysis method for weld detection of an engineering vehicle, and the method includes:

[0016] S10: After the welding of the railway engineering vehicle is completed, collect weld images of a plurality of weld regions, perform welding stress deformation prediction, and obtain a plurality of predicted deformation amounts.

[0017] S20: According to the plurality of predicted deformation amounts, perform vehicle specification quality impact analysis on the railway engineering vehicle to obtain vehicle specification impact parameters, and perform railway track detection quality impact analysis and driving vibration analysis to obtain track detection impact parameters and track detection vibration parameters.

[0018] S30: Obtain the probability that the track of the target railway section corresponding to the railway engineering vehicle has track specification anomalies, calculate to obtain the track anomaly rate, and correct the track detection impact parameters to obtain corrected track detection impact parameters.

[0019] S40: Calculate the welding quality detection result of the railway engineering vehicle according to the vehicle specification impact parameters, the corrected track detection impact parameters, and the track detection vibration parameters.

[0020] Exemplarily, after the railway engineering vehicle completes the welding operation, images of multiple weld areas are collected. This step aims to accurately record the actual conditions of each weld through high-resolution image capture technology. These weld images not only contain the appearance characteristics of the welds but also imply various quality information that may occur during the welding process. Furthermore, image processing and data analysis techniques are used to analyze these weld images, and various defects that may exist in the welds are concerned during this process, such as excessive weld reinforcement, insufficient weld leg, etc. These defects are often the main causes of stress concentration during the welding process, and they will cause varying degrees of deformation of the weld after cooling or during the actual use of the railway engineering vehicle due to the release of internal stress. In order to predict these potential deformation situations in advance, based on the defect information in the weld images, combined with various information such as welding process parameters, material characteristics, and historical deformation data, complex mathematical models are used to predict the welding stress and deformation. This step can generate multiple predicted deformation amounts, and these deformation amounts reflect the degree and direction of deformation that the weld may undergo under different conditions. In this way, the manufacturers and maintenance personnel of railway engineering vehicles can understand in advance the possible problems existing in the welds, and thus take corresponding measures for prevention or repair. This can not only improve the overall quality of railway engineering vehicles but also effectively extend their service life, ensuring the safety and stability of railway transportation.

[0021] Optionally, when the railway track deforms, this change not only directly affects the specification quality and driving performance of the vehicle, but also affects the accuracy of track detection and the vibration characteristics of vehicle driving. Collect and record multiple predicted deformation amounts of the railway track, which may include geometric deviations, unevenness, twist, or lateral displacement of the track, etc. These are all key indicators for evaluating the track condition. Then, conduct an analysis of the impact on vehicle specification quality based on these predicted deformation amount data. This step aims to understand how track deformation changes the interaction between the vehicle and the track, thereby affecting the stability, controllability, and riding comfort of the vehicle. For example, track irregularities may lead to abnormal wear between the wheels and the track, accelerating the aging of vehicle components and thus shortening the service life of the vehicle. Through quantitative analysis, vehicle specification impact parameters can be extracted, which reflect the degree of vehicle performance degradation under different deformation levels. At the same time, considering that the railway engineering vehicle itself undertakes the task of detecting track quality, track deformation will also in turn affect the accuracy and reliability of the detection equipment. Deformation may cause a change in the relative position between the detection equipment and the track, introducing measurement errors and thus affecting the accurate judgment of the track condition. Therefore, conducting an analysis of the impact on railway track detection quality can evaluate the degree of influence of deformation on the detection results, and then determine track detection impact parameters, which help to correct the detection data and improve the accuracy of the detection results. In addition, track deformation also affects the vibration characteristics during vehicle driving. An uneven track will exacerbate the bumping and vibration of the vehicle, which not only reduces the riding comfort but may also impose additional burdens on key components such as the vehicle's suspension system and bogie. Through driving vibration analysis, the vibration amplitude, frequency, and distribution characteristics caused by track deformation can be quantified to obtain track detection vibration parameters, which are crucial for evaluating the safety and durability of vehicle driving and also provide a scientific basis for subsequent vibration reduction measures. In summary, the deformation of the railway track is not only related to the specification quality and driving performance of the vehicle itself, but also affects the accuracy of track detection and the vibration characteristics of vehicle driving. Through comprehensive analysis of the impact on vehicle specification quality, railway track detection quality, and driving vibration, the multi-faceted impacts of track deformation can be comprehensively evaluated, providing strong data support and decision-making basis for railway maintenance and safety monitoring.

[0022] Furthermore, collect the historical track inspection data of the target railway section. These data record the track conditions at different time points, including track geometric dimensions, material conditions, deformation situations, etc. Conduct statistical analysis on these data to identify cases of abnormal track specifications, such as deformations, damages, or wear that exceed the specified tolerance ranges. Calculate the frequency of abnormal track specifications in the target railway section within a specific time period based on the number of cases of abnormal track specifications in the historical data. This frequency reflects the universality and severity of track problems in the target railway section. Compare the track abnormality frequency of the target railway section with the average track abnormality frequency of the entire railway network or railway sections under similar conditions. Calculate a relative value, namely the track abnormality rate, through this comparison. The higher the track abnormality rate, the greater the probability of abnormal track specifications in the target railway section compared to the average level. Moreover, track deformations, such as unevenness and twist, will directly affect the measurement accuracy of the detection equipment on railway engineering vehicles. The greater the degree of deformation, the greater the detection error may be, thus affecting the accurate judgment of the track condition. Therefore, correct the initially obtained track inspection influence parameters based on the track abnormality rate. The principle of correction is: the higher the track abnormality rate of the target railway section, the greater the impact of vehicle deformation on the detection accuracy. Therefore, the corrected track inspection influence parameter should be increased accordingly. This can be achieved by applying the track abnormality rate as a multiplier factor to the initially obtained track inspection influence parameter. Using the corrected track inspection influence parameter can more accurately evaluate the actual detection accuracy of railway engineering vehicles when performing inspection tasks in the target railway section.

[0023] Finally, comprehensively consider the vehicle specification influence parameters, calibration track detection influence parameters, and track detection vibration parameters to comprehensively evaluate the welding quality. On the one hand, according to the specific application scenarios, usage requirements, and safety standards of railway engineering vehicles, corresponding weights are assigned to the vehicle specification influence parameters, calibration track detection influence parameters, and track detection vibration parameters respectively. The magnitude of the weight is usually determined based on the importance of the parameter's overall impact on welding quality. For example, if the vehicle specification quality is the primary consideration factor, the weight of the vehicle specification influence parameter may be relatively high. Another method is to calculate the mean of these three parameters to obtain a comprehensive welding quality evaluation value, which is applicable when the influence of each parameter on welding quality is relatively balanced. If the weighted calculation method is adopted, each parameter is multiplied by its corresponding weight, and then the obtained products are added together to obtain the final welding quality detection result. If the mean calculation method is adopted, the numerical values of the three parameters are directly added and then divided by 3 to obtain the average evaluation value of welding quality. The welding quality detection result obtained by weighted calculation or mean calculation can intuitively understand the overall quality status of the welds of railway engineering vehicles. If the detection result is high, it indicates that the weld quality is good, and the impact on vehicle specification quality, track detection accuracy, and driving vibration characteristics is small. If the detection result is low, it indicates that there may be significant quality problems with the welds, and further detection, repair, or optimization measures need to be taken. Based on the welding quality detection result, the manufacturer or maintenance team of railway engineering vehicles can formulate corresponding follow-up measures, which may include improving the monitoring of welds, optimizing the welding process, enhancing material quality, or strengthening employee training. In summary, by comprehensively evaluating the vehicle specification influence parameters, calibration track detection influence parameters, and track detection vibration parameters, the welding quality detection result of railway engineering vehicles can be obtained comprehensively and accurately, providing a strong guarantee for the safe operation of railway engineering vehicles.

[0024] In a preferred embodiment, after the welding of the railway engineering vehicle is completed, weld images of multiple weld areas are collected, welding stress deformation prediction is performed, and multiple predicted deformation amounts are obtained, including: after the welding of the railway engineering vehicle is completed, weld images of multiple weld areas are collected, where the multiple weld areas include a frame area and a detection equipment support area; a convolutional neural network is used to train a welding stress deformation predictor; multiple weld images are input into the welding stress deformation predictor, and multiple predicted deformation amounts of multiple weld areas are obtained by prediction output.

[0025] Specifically, the stress and deformation conditions of the weld seam after welding are directly related to the overall performance and safety of the vehicle. The stress and deformation of the weld can be predicted by collecting weld images and using a convolutional neural network. After the railway engineering vehicle completes the welding operation, it enters the weld quality assessment stage. At this time, it is necessary to clarify the weld areas to be concerned about, which usually include the frame area and the detection equipment support area, because these parts are crucial for the stability and detection accuracy of the vehicle. Use high-resolution imaging equipment, such as industrial cameras or dedicated weld detection equipment, to collect images of multiple weld areas. Ensure that each weld area can be clearly and completely recorded for subsequent analysis. Preprocess the collected weld images, including operations such as denoising, enhancing contrast, and adjusting the size, to improve the image quality for subsequent neural network processing. Select a convolutional neural network model suitable for processing image data, such as ResNet, VGG, etc. These models perform well in tasks such as image classification and object detection and are suitable for predicting the stress and deformation of weld images. Use a large number of weld images with known stress and deformation conditions as training data to train the CNN. During the training process, the CNN will learn the features in the weld images, such as the weld morphology, material texture, etc., and the correlation between these features and the stress and deformation. Input the preprocessed multiple weld images into the trained CNN model. The model will extract features and classify each weld image, and predict and output the predicted deformation amounts of the corresponding weld areas. These deformation amounts may include different types of deformations such as tensile, compressive, and torsional deformations of the weld. The CNN model will output one or more predicted deformation amounts for each weld area, which reflect the stress and deformation conditions of the weld in different directions. By comparing and analyzing these predicted deformation amounts, the stress and deformation degree of the weld can be evaluated, and then the quality of the welding can be judged. In summary, by collecting weld images and using a convolutional neural network to predict the welding stress and deformation, a rapid and accurate assessment of the weld quality of railway engineering vehicles can be achieved, providing strong support for subsequent maintenance and improvement.

[0026] In a preferred embodiment, a convolutional neural network is used to train a welding stress and deformation predictor, including: according to the welding data of the railway engineering vehicle within the historical time, collecting a set of multiple sample weld images of multiple weld areas, and collecting the stress and deformation amounts after the use of different sample weld images after welding, which are marked as a set of multiple sample deformation amounts; using a convolutional neural network to construct a welding stress and deformation predictor, wherein the welding stress and deformation predictor includes multiple welding deformation prediction branches corresponding to multiple weld areas; respectively using the set of multiple sample weld images and the set of multiple sample deformation amounts to perform supervised training and testing on the multiple welding deformation prediction branches until the requirements are met, and obtaining a welding stress and deformation predictor.

[0027] Preferably, the process of training the welding stress and deformation predictor using a convolutional neural network (CNN) is as follows: First, collect welding data of multiple weld areas from the historical welding records of railway engineering vehicles. These weld areas include the frame and the parts of the detection equipment support. For each weld area, collect images of multiple sample welds, and these images should be able to clearly reflect the characteristics of the weld morphology, material texture, etc. After welding is completed, use professional measuring tools or equipment to measure the stress and deformation amounts of each sample weld. These deformation amounts include various types such as tension, compression, and torsion. Associate the measured stress and deformation amounts with the corresponding sample weld images to form multiple sample deformation amount sets, and these sets will serve as the data basis for supervised training. Use a convolutional neural network (CNN) as the basic framework of the prediction model. According to the different weld areas, construct multiple welding deformation prediction branches. Each branch is responsible for predicting the stress and deformation amounts of the corresponding weld area. The design of the CNN model should take into account the requirements of tasks such as feature extraction, classification, and regression of weld images. Ensure that the model can accurately capture the key information in the weld images by selecting appropriate parameters such as the number of network layers, the size of the convolutional kernel, and the activation function. Input multiple sample weld image sets and the corresponding multiple sample deformation amount sets as training data into the CNN model. During the training process, use the method of supervised learning to continuously adjust the model parameters by minimizing the error between the predicted deformation amount and the actual deformation amount. At the same time, in order to verify the performance of the model, some data can be used as a test set to test the trained model. Evaluate the accuracy and generalization ability of the model by comparing the predicted results with the actual results on the test set. Subsequently, perform iterative optimization on the CNN model according to the test results, including means such as adjusting the network structure, increasing the training data, and using data augmentation techniques. Repeat the above training and testing process until the performance of the model on the test set meets the preset requirements or reaches a convergence state. After multiple iterative optimizations, obtain a welding stress and deformation predictor with stable performance and accurate prediction. This predictor can predict the corresponding stress and deformation amounts based on the image input of different weld areas. Apply the trained welding stress and deformation predictor to the actual production of railway engineering vehicles. After welding is completed, the stress and deformation conditions of the weld can be quickly and accurately evaluated by collecting the weld image and inputting it into the predictor. In summary, by using a convolutional neural network to train the welding stress and deformation predictor, the rapid and accurate prediction of the weld stress and deformation of railway engineering vehicles can be achieved.

[0028] In a preferred embodiment, vehicle specification quality impact analysis is performed based on the multiple predicted deformation amounts to obtain vehicle specification impact parameters, including: obtaining multiple allowable deformation tolerances for the multiple weld regions during the welding process of the railway engineering vehicle; respectively taking the multiple allowable deformation tolerances as a reference, calculating the abnormal amplitude of the multiple predicted deformation amounts exceeding the multiple allowable deformation tolerances to obtain multiple deformation abnormal coefficients; and calculating and obtaining vehicle specification impact parameters based on the multiple deformation abnormal coefficients.

[0029] Specifically, during the design and manufacturing stages of the railway engineering vehicle, an allowable deformation tolerance is set for each weld region. This tolerance is comprehensively determined based on factors such as the vehicle's usage requirements, material properties, and welding process, and is used to measure whether the weld deformation is within an acceptable range. The predicted deformation amounts of multiple weld regions have been obtained through the prediction model, and these deformation amounts reflect the actual deformation conditions of the welds after welding. Comparing the predicted deformation amount of each weld region with the corresponding allowable deformation tolerance, if the predicted deformation amount exceeds the allowable deformation tolerance, calculate the exceeded amplitude, that is, the abnormal amplitude. This amplitude reflects the potential impact degree of the weld deformation on the vehicle specification quality. To quantify the impact of weld deformation on vehicle specification quality, the abnormal amplitude can be converted into a deformation abnormal coefficient. This coefficient can be the ratio of the abnormal amplitude to the allowable deformation tolerance, or other indicators that can reflect the severity of the deformation. Corresponding deformation abnormal coefficients are obtained for all weld regions, and these coefficients reflect the comprehensive impact of the deformation of different weld regions on vehicle specification quality. Subsequently, a vehicle specification impact parameter can be calculated based on the deformation abnormal coefficients of all weld regions. This parameter can be the average value, maximum value, weighted sum, or other indicators that can comprehensively reflect the impact on vehicle specification quality. The magnitude of the vehicle specification impact parameter directly reflects the overall impact of welding quality on vehicle specification quality. If the parameter value is high, it indicates that the weld deformation is large and the impact on vehicle specification quality is also large, and corresponding measures need to be taken for repair or optimization. In summary, by comparing the predicted deformation amount with the allowable deformation tolerance to calculate the deformation abnormal coefficient and finally obtaining the vehicle specification impact parameter, the impact of welding quality on the vehicle specification quality of the railway engineering vehicle can be comprehensively and accurately evaluated.

[0030] In a preferred embodiment, railway track detection quality impact analysis and driving vibration analysis are performed, including: training a track detection impact analyzer based on the recorded data of the railway engineering vehicle during track detection on the railway track, where the track detection impact analyzer includes a detection impact analysis path and a detection vibration analysis path; combining the multiple predicted deformation amounts and respectively inputting them into the detection impact analysis path and the detection vibration analysis path in the track detection impact analyzer, and outputting to obtain a track detection impact parameter and a track detection vibration parameter.

[0031] Optionally, ensuring the quality of railway tracks requires attention to the geometric state of the tracks, material integrity, and its impact on train operation. To analyze these factors in depth, data collected by a railway engineering vehicle during track inspection can be utilized to train an analyzer for the impact analysis of track inspection in combination with the predicted weld deformation obtained. Specialized inspection equipment on the railway engineering vehicle is used to conduct a comprehensive inspection of items such as the geometric state of the tracks and material defects on the railway. These inspections will generate a large amount of data, including but not limited to the geometric deviation of the tracks, material wear conditions, weld status, etc. The predicted weld deformations obtained through the prediction model are integrated. These deformations reflect the deformation conditions of the welds during the welding process and are an important part of the overall quality assessment of the tracks. An analyzer for the impact analysis of track inspection is trained using the collected track inspection data and the integrated predicted deformations. This analyzer consists of two main paths: the detection impact analysis path and the detection vibration analysis path. The former focuses on evaluating the impact of factors such as the geometric state of the tracks and material integrity on the safety of train operation; the latter focuses on analyzing how these factors affect the vibration characteristics of train travel. The integrated predicted deformations and other track inspection data are separately input into the two paths of the analyzer for the impact analysis of track inspection, which ensures that the analysis can comprehensively consider the impact of weld deformation and other track states on train operation. The detection impact analysis path will output a series of track inspection impact parameters, which reflect the specific impact of the track state on the safety of train operation, such as the impact of track geometric deviation on train stability and the impact of material wear on train braking distance. At the same time, the detection vibration analysis path will output track inspection vibration parameters, which describe the vibration characteristics of the train when traveling on the tracks, such as vibration frequency and amplitude. These parameters are of great significance for evaluating train ride comfort, reducing track wear, and predicting potential failures. In summary, by training the analyzer for the impact analysis of track inspection and using it to analyze track inspection data and weld predicted deformations, the impact of track state on train operation can be deeply understood, providing a scientific basis for the maintenance and operation of the railway system.

[0032] In a preferred embodiment, an analyzer for analyzing the impact of track detection is trained based on the recorded data of a railway engineering vehicle during track detection on a railway track, including: collecting a set of sample deformation quantity combinations according to the recorded data of the railway engineering vehicle during track detection on the railway track, and collecting the average error ratio that occurs during railway track detection under different sample deformation quantity combinations, which is labeled as a set of sample track detection impact parameters; collecting the increase ratio of the vehicle vibration amplitude during railway track detection under different sample deformation quantity combinations, which is labeled as a set of sample track detection vibration parameters; using machine learning to construct a detection impact analysis path and a detection vibration analysis path; using the set of sample deformation quantity combinations as input features, and respectively using the set of sample track detection impact parameters and the set of sample track detection vibration parameters as output features, to perform supervised training and testing on the detection impact analysis path and the detection vibration analysis path until the requirements are met, and obtaining an analyzer for analyzing the impact of track detection.

[0033] Exemplarily, a large amount of track detection data is collected, which is sourced from the records of railway engineering vehicles during the detection of items such as track geometric conditions and material integrity on the railway. Pay attention to the deformation conditions of key parts such as welds in this data, and collect sample data under different deformation quantity combinations. This set of sample deformation quantity combinations represents various deformation situations that may be encountered on the track. Next, railway track detection experiments are carried out according to these sample deformation quantity combinations, and the average error ratio that occurs during the detection process is recorded. This ratio reflects the impact of track deformation on the detection accuracy, and it is labeled as a set of sample track detection impact parameters. At the same time, the increase ratio of the vehicle vibration amplitude during track detection under different sample deformation quantity combinations is collected. This ratio reveals the impact of track deformation on the smooth running of the train, and it is labeled as a set of sample track detection vibration parameters. With these labeled data, the analyzer for analyzing the impact of track detection is started to be constructed. Specifically, machine learning techniques are used to construct a detection impact analysis path and a detection vibration analysis path respectively. These two paths are used to analyze the impact of track deformation on the detection accuracy and the smooth running of the train respectively. During the training process, the set of sample deformation quantity combinations is used as the input feature, and the set of sample track detection impact parameters and the set of sample track detection vibration parameters are used as the output features respectively to perform supervised training on the detection impact analysis path and the detection vibration analysis path. By continuously adjusting the model parameters and optimizing the model performance until the preset training requirements are met, finally an analyzer for analyzing the impact of track detection is obtained. This analyzer can predict the impact degree on the detection accuracy and the smooth running of the train according to the input track deformation quantity combination, providing strong data support for the maintenance and optimization of the railway system. Through this method, the track state can be evaluated more scientifically and accurately, ensuring the safety and comfort of train operation.

[0034] In a preferred embodiment, the probability that the track of the target railway section corresponding to the railway engineering vehicle has an abnormal track specification is obtained, the track abnormality rate is calculated, and the track detection influence parameter is corrected to obtain a corrected track detection influence parameter, including: obtaining the frequency of the track of the target railway section corresponding to the railway engineering vehicle having an abnormal track specification as the target track abnormality frequency; obtaining the average frequency of the tracks of multiple sample railway sections having abnormal track specifications as the average track abnormality frequency; calculating the ratio of the target track abnormality frequency to the average track abnormality frequency as the track abnormality rate; and using the track abnormality rate to perform a correction calculation on the track detection influence parameter to obtain a corrected track detection influence parameter.

[0035] Preferably, focus on the target railway section corresponding to the railway engineering vehicle, and obtain the frequency of the track of this section having an abnormal track specification, that is, the target track abnormality frequency, through historical data and real-time monitoring. This frequency reflects the probability of the track of this section having specification problems within a specific time period. Next, in order to more comprehensively understand the situation of abnormal track specifications, it is also necessary to collect the track specification abnormality frequencies of multiple sample railway sections (these sample sections should be representative and able to reflect the general situation of the entire railway system), and calculate the average value of these frequencies to obtain the average track abnormality frequency. This average value provides a benchmark for comparing the risk levels of abnormal track specifications in different railway sections. Then, the track abnormality rate is obtained by calculating the ratio of the target track abnormality frequency to the average track abnormality frequency. This ratio reveals the level of risk of abnormal track specifications of the target railway section relative to the average level of the entire railway system. If the track abnormality rate is higher than 1, it indicates that the risk of abnormal track specifications of the target railway section is higher than the average level; if it is lower than 1, the risk is lower than the average level. Finally, use the track abnormality rate to correct the previously obtained track detection influence parameter. The purpose of the correction is to make these parameters more capable of reflecting the actual risk of abnormal track specifications of the target railway section. Specifically, the track abnormality rate can be used as a weighting factor and multiplied by the original track detection influence parameter to obtain the corrected track detection influence parameter. In this way, the corrected parameter not only considers the influence of track deformation on detection accuracy but also incorporates the risk factor of abnormal track specifications, providing more accurate and comprehensive information support for the maintenance and management of the railway system.

[0036] In a preferred embodiment, according to the vehicle specification influence parameter, the corrected track detection influence parameter, and the track detection vibration parameter, the welding quality detection result of the railway engineering vehicle is calculated, including: calculating a welding quality influence parameter according to the vehicle specification influence parameter, the corrected track detection influence parameter, and the track detection vibration parameter; and using the welding quality influence parameter as the welding quality detection result of the railway engineering vehicle.

[0037] Specifically, the vehicle specification influence parameter, the calibration track detection influence parameter, and the track detection vibration parameter respectively reflect the influence of welding quality on the railway engineering vehicle from three aspects: vehicle specification compliance, calibration of track detection accuracy, and train running vibration characteristics. To obtain a comprehensive welding quality detection result, the method of weighted calculation or mean calculation is adopted. Assuming that the influence degrees of these three parameters on welding quality are equivalent, the mean value of them is selected as the welding quality influence parameter. Specifically, the vehicle specification influence parameter, the calibration track detection influence parameter, and the track detection vibration parameter are denoted as A, B, and C respectively, and then their arithmetic mean value is calculated, that is, the welding quality influence parameter D = (A + B + C) / 3. This welding quality influence parameter D comprehensively reflects the influence degree of welding quality on the overall performance of the railway engineering vehicle. It takes into account the influence of welding deformation on vehicle specification compliance, the calibration requirement for track detection accuracy, and the change in train running vibration characteristics. Finally, this welding quality influence parameter D is used as the welding quality detection result of the railway engineering vehicle. This result not only provides direct feedback on welding quality for vehicle manufacturers but also provides an important reference for the maintenance and management of the railway system. Assuming that the vehicle specification influence parameter A is 0.8 (indicating that the influence of welding deformation on vehicle specifications is small), the calibration track detection influence parameter B is 0.9 (indicating that the influence of welding deformation on track detection accuracy is moderate), and the track detection vibration parameter C is 0.7 (indicating that the influence of welding deformation on train running vibration characteristics is large). Then, the welding quality influence parameter D = (0.8 + 0.9 + 0.7) / 3 = 0.8. This result indicates that the overall welding quality of the railway engineering vehicle is good, but the influence of welding deformation on train running vibration characteristics still needs to be concerned. Through such a comprehensive calculation method, the welding quality of the railway engineering vehicle can be evaluated more comprehensively and accurately, providing a strong guarantee for the safe and efficient operation of the railway system.

[0038] The welding quality analysis method for weld detection of an engineering vehicle provided by an embodiment of the present invention has at least the following technical effects:

[0039] 1. By collecting weld images of multiple weld regions and training a welding stress and deformation predictor using a convolutional neural network, the stress and deformation amount of the weld can be accurately predicted, which not only improves the accuracy of prediction but also greatly shortens the detection time, realizing the rapid evaluation of weld quality. In addition, by constructing a welding deformation prediction branch corresponding to multiple weld regions, the predictor can perform customized prediction for different types of welds, further improving the accuracy and practicality of prediction.

[0040] 2. The solution not only considers the impact of weld deformation on the vehicle specification quality, but also deeply analyzes the impact of weld deformation on the railway track detection quality and the train running vibration. By training the track detection impact analyzer and obtaining the track detection impact parameters and track detection vibration parameters respectively, it is possible to comprehensively evaluate the impact of weld deformation on the overall performance of the railway system, providing more accurate and comprehensive data support for the maintenance and management of the railway system, and helping to detect and repair potential safety hazards in a timely manner.

[0041] 3. By calculating the ratio of the frequency of track specification anomalies in the target railway section to the average anomaly frequency of multiple sample railway sections, the track anomaly rate is obtained, and based on this, the track detection impact parameters are corrected, fully considering the differences in track quality in different railway sections, making the corrected track detection impact parameters better reflect the actual situation of the target railway section, not only improving the accuracy of welding quality assessment, but also providing a scientific basis for the personalized maintenance and management of the railway system.

[0042] Embodiment 2:

[0043] As Figure 2 shown, based on the same inventive concept as the welding quality analysis method for the weld detection of an engineering vehicle provided in Embodiment 1, the present invention embodiment also provides a welding quality analysis system for the weld detection of an engineering vehicle, and the system includes:

[0044] An image acquisition module 11, configured to collect weld images of multiple weld areas after the welding of the railway engineering vehicle is completed, perform welding stress deformation prediction, and obtain multiple predicted deformation amounts.

[0045] A quality analysis module 12, configured to perform vehicle specification quality impact analysis on the railway engineering vehicle according to the multiple predicted deformation amounts to obtain vehicle specification impact parameters, and perform railway track detection quality impact analysis and running vibration analysis to obtain track detection impact parameters and track detection vibration parameters.

[0046] A probability calculation module 13, configured to obtain the probability of track specification anomalies occurring in the target railway section corresponding to the railway engineering vehicle, calculate the track anomaly rate, and correct the track detection impact parameters to obtain corrected track detection impact parameters.

[0047] A result acquisition module 14, configured to calculate the welding quality detection result of the railway engineering vehicle according to the vehicle specification impact parameters, the corrected track detection impact parameters, and the track detection vibration parameters.

[0048] Furthermore, the image acquisition module 11 is further configured to perform the following steps:

[0049] After the welding of a railway engineering vehicle is completed, weld images of multiple weld areas are collected, where the multiple weld areas include a frame area and a bracket area for a detection device. A convolutional neural network is used to train a welding stress and deformation predictor. The multiple weld images are input into the welding stress and deformation predictor, and multiple predicted deformation amounts of the multiple weld areas are obtained through prediction output.

[0050] Furthermore, the image acquisition module 11 is further configured to perform the following steps:

[0051] According to the welding data of the railway engineering vehicle within a historical time, multiple sets of sample weld images of multiple weld areas are collected, and the stress and deformation amounts after the completion of the welding of different sample weld images are collected and labeled as multiple sets of sample deformation amounts. A convolutional neural network is used to construct a welding stress and deformation predictor, where the welding stress and deformation predictor includes multiple welding deformation prediction branches corresponding to the multiple weld areas. The multiple sets of sample weld images and multiple sets of sample deformation amounts are respectively used to perform supervised training and testing on the multiple welding deformation prediction branches until the requirements are met, and a welding stress and deformation predictor is obtained.

[0052] Furthermore, the quality analysis module 12 is further configured to perform the following steps:

[0053] Obtain the multiple allowable deformation tolerances of the multiple weld areas during the welding of the railway engineering vehicle. Respectively taking the multiple allowable deformation tolerances as a reference, calculate the abnormal amplitudes of the multiple predicted deformation amounts exceeding the multiple allowable deformation tolerances to obtain multiple deformation abnormal coefficients. According to the multiple deformation abnormal coefficients, calculate and obtain a vehicle specification influence parameter.

[0054] Furthermore, the quality analysis module 12 is further configured to perform the following steps:

[0055] According to the recorded data of the railway engineering vehicle performing track detection on a railway track, train a track detection influence analyzer, where the track detection influence analyzer includes a detection influence analysis path and a detection vibration analysis path. Combine the multiple predicted deformation amounts and input them into the detection influence analysis path and the detection vibration analysis path in the track detection influence analyzer respectively, and output to obtain a track detection influence parameter and a track detection vibration parameter.

[0056] Furthermore, the quality analysis module 12 is further configured to perform the following steps:

[0057] According to the recorded data of the railway engineering vehicle for track detection on the railway track, collect the sample deformation quantity combination set, and collect the average error ratio that appears during railway track detection under different sample deformation quantity combinations, and label it as the sample track detection influence parameter set; collect the vehicle vibration amplitude increase ratio that appears during railway track detection under different sample deformation quantity combinations, and label it as the sample track detection vibration parameter set; use machine learning to construct a detection influence analysis path and a detection vibration analysis path; use the sample deformation quantity combination set as the input feature, and use the sample track detection influence parameter set and the sample track detection vibration parameter set as the output features respectively, and perform supervised training and testing on the detection influence analysis path and the detection vibration analysis path until the requirements are met to obtain a track detection influence analyzer.

[0058] Furthermore, the probability calculation module 13 is further configured to perform the following steps:

[0059] Obtain the frequency of track specification anomalies occurring on the track of the target railway section corresponding to the railway engineering vehicle as the target track anomaly frequency; obtain the average frequency of track specification anomalies occurring on the tracks of multiple sample railway sections as the average track anomaly frequency; calculate the ratio of the target track anomaly frequency to the average track anomaly frequency as the track anomaly rate; use the track anomaly rate to perform a correction calculation on the track detection influence parameter to obtain a corrected track detection influence parameter.

[0060] Furthermore, the result acquisition module 14 is further configured to perform the following steps:

[0061] Calculate and obtain a welding quality influence parameter according to the vehicle specification influence parameter, the corrected track detection influence parameter, and the track detection vibration parameter; use the welding quality influence parameter as the welding quality detection result of the railway engineering vehicle.

[0062] Through the foregoing detailed description of a welding quality analysis method for weld detection of an engineering vehicle in this specification, those skilled in the art can clearly know a welding quality analysis system for weld detection of an engineering vehicle in this embodiment. For the system disclosed in the embodiment, since it corresponds to the method disclosed in the embodiment, the description is relatively simple, and for the relevant parts, reference may be made to the description in the method part.

[0063] The above description of the disclosed embodiments enables those skilled in the art to implement or use the present application. Various modifications to these embodiments will be apparent to those skilled in the art, and the general principles defined herein can be implemented in other embodiments without departing from the spirit or scope of the present application. Therefore, the present application will not be limited to these embodiments shown herein, but will be accorded the widest scope consistent with the principles and novel features disclosed herein.

Claims

1. A method for analyzing the welding quality of weld seam detection of engineering vehicles, characterized in that, The method comprises: After the welding of railway engineering vehicles is completed, weld images of multiple weld areas are collected to predict welding stress and deformation and obtain multiple predicted deformation amounts; Based on the multiple predicted deformations, performing a vehicle specification quality impact analysis of the railway engineering vehicle to obtain vehicle specification impact parameters, and performing a railway track inspection quality impact analysis and a running vibration analysis to obtain track inspection impact parameters and track inspection vibration parameters; Obtaining a probability of track specification abnormality occurring on the track of the target railway section corresponding to the railway engineering vehicle, calculating a track abnormality rate, and correcting the track detection influencing parameters to obtain corrected track detection influencing parameters; The welding quality inspection result of the railway engineering vehicle is obtained by calculation according to the vehicle specification influencing parameters, the corrected track detection influencing parameters and the track detection vibration parameters.

2. The welding quality analysis method for the weld seam detection of an engineering vehicle according to claim 1, wherein After the welding of railway engineering vehicles is completed, weld images of multiple weld areas are collected to predict welding stress and deformation, and multiple predicted deformation values are obtained, including: After the welding of the railway engineering vehicle is completed, weld images of multiple weld areas are collected, wherein the multiple weld areas include the vehicle frame area and the detection equipment bracket area; Use convolutional neural network to train welding stress and deformation predictor; A plurality of weld seam images are input into the welding stress and deformation predictor, and a plurality of predicted deformation amounts of a plurality of weld seam regions are obtained through prediction output.

3. The welding quality analysis method for the weld seam detection of engineering vehicles according to claim 2, characterized in that, A convolutional neural network is used to train a welding stress and deformation predictor, including: Based on the welding data of railway engineering vehicles in the historical period, multiple sample weld image sets of multiple weld areas are collected, and the stress and deformation of different sample weld images after welding are collected and marked as multiple sample deformation sets; A convolutional neural network is used to construct a welding stress and deformation predictor, wherein the welding stress and deformation predictor includes multiple welding deformation prediction branches corresponding to multiple weld regions; The plurality of sample weld image sets and the plurality of sample deformation amount sets are respectively used to perform supervised training and testing on the plurality of welding deformation prediction branches until the requirements are met, thereby obtaining a welding stress deformation predictor.

4. The welding quality analysis method for the weld detection of engineering vehicles according to claim 1, characterized in that Based on the multiple predicted deformations, a vehicle specification quality impact analysis of the railway engineering vehicle is performed to obtain vehicle specification impact parameters, including: Obtaining a plurality of allowable deformation tolerances of the plurality of weld areas during a railway engineering vehicle welding process; Taking the plurality of allowable deformation tolerances as references, respectively, calculating abnormal amplitudes by which the plurality of predicted deformation amounts exceed the plurality of allowable deformation tolerances, and obtaining a plurality of deformation abnormality coefficients; The vehicle specification influencing parameters are calculated based on the multiple deformation abnormality coefficients.

5. The welding quality analysis method for the weld detection of engineering vehicles according to claim 1, characterized in that Conduct railway track inspection quality impact analysis and running vibration analysis, including: Training a track detection impact analyzer based on recorded data of track detection performed by a railway engineering vehicle on a railway track, wherein the track detection impact analyzer includes a detection impact analysis path and a detection vibration analysis path; The multiple predicted deformation amounts are combined and input into the detection impact analysis path and the detection vibration analysis path in the track detection impact analyzer respectively, and the track detection impact parameters and the track detection vibration parameters are output.

6. The welding quality analysis method for the weld detection of engineering vehicles according to claim 5, characterized in that Training an analyzer for detecting the impact of track inspection based on the recorded data of a railway engineering vehicle during track inspection on a railway track, including: Collecting a set of combined sample deformation amounts based on the recorded data of the railway engineering vehicle during track inspection on the railway track, and collecting the average error ratio that occurs during railway track inspection under different combined sample deformation amounts, which is labeled as a set of sample track inspection impact parameters; Collecting the increased ratio of the vehicle vibration amplitude during railway track inspection under different combined sample deformation amounts, which is labeled as a set of sample track inspection vibration parameters; Using machine learning to construct an analysis path for inspection impact and an analysis path for inspection vibration; Using the set of combined sample deformation amounts as input features, and respectively using the set of sample track inspection impact parameters and the set of sample track inspection vibration parameters as output features to perform supervised training and testing on the analysis path for inspection impact and the analysis path for inspection vibration until the requirements are met, and obtaining an analyzer for detecting the impact of track inspection.

7. The welding quality analysis method for the weld detection of an engineering vehicle according to claim 1, characterized in that, Obtaining the probability that the track of the target railway section corresponding to the railway engineering vehicle has an abnormal track specification, calculating to obtain a track abnormality rate, and correcting the track inspection impact parameters to obtain corrected track inspection impact parameters, including: Obtaining the frequency of the track of the target railway section corresponding to the railway engineering vehicle having an abnormal track specification, which is used as the target track abnormality frequency; Obtaining the average frequency of the tracks of multiple sample railway sections having an abnormal track specification, which is used as the average track abnormality frequency; Calculating the ratio of the target track abnormality frequency to the average track abnormality frequency, which is used as the track abnormality rate; Using the track abnormality rate to perform a correction calculation on the track inspection impact parameters to obtain corrected track inspection impact parameters.

8. The welding quality analysis method for the weld seam detection of engineering vehicles according to claim 1, characterized in that, Calculating the welding quality detection result of the railway engineering vehicle based on the vehicle specification impact parameters, the corrected track inspection impact parameters, and the track inspection vibration parameters, including: Calculating a welding quality impact parameter based on the vehicle specification impact parameters, the corrected track inspection impact parameters, and the track inspection vibration parameters; Taking the welding quality impact parameter as the welding quality detection result of the railway engineering vehicle.

9. A welding quality analysis system for weld detection of engineering vehicles, characterized in that, A welding quality analysis method for implementing the weld inspection of an engineering vehicle according to any one of claims 1-8, the system includes: An image acquisition module, which is used to collect weld images of multiple weld areas after the welding of the railway engineering vehicle is completed, perform prediction on welding stress deformation, and obtain multiple predicted deformation amounts; A quality analysis module, which is used to perform an analysis on the impact of the vehicle specification quality of the railway engineering vehicle based on the multiple predicted deformation amounts to obtain vehicle specification impact parameters, and perform an analysis on the impact of railway track inspection quality and driving vibration analysis to obtain track inspection impact parameters and track inspection vibration parameters; A probability calculation module, which is used to obtain the probability that the track of the target railway section corresponding to the railway engineering vehicle has an abnormal track specification, calculate to obtain a track abnormality rate, and correct the track inspection impact parameters to obtain corrected track inspection impact parameters; A result acquisition module, which is used to calculate the welding quality detection result of the railway engineering vehicle based on the vehicle specification impact parameters, the corrected track inspection impact parameters, and the track inspection vibration parameters.

Images