Railway bridge crack classification and identification system and method

Abstract

The invention discloses a railway bridge crack classification and identification system and method. The system comprises a data acquisition module, a crack feature analysis module, a crack risk assessment module, a climate correlation trend prediction module, a data fusion and decision module and a remote monitoring and alarm module. According to the method, the crack propagation trend prediction model is constructed by combining the geometrical characteristics of the crack and the climate conditions, so that the physical characteristics of the crack and the propagation behaviors of the crack under different climate conditions can be comprehensively analyzed, the prediction precision of the system on the crack propagation trend is improved, early warning can be performed on potential dangerous cracks, and the prediction accuracy of the crack propagation trend is improved. And the accuracy and reliability of crack classification and risk assessment are greatly improved.

Description

Technical Field

[0001] The present invention relates to the technical field of railway bridges, and particularly to a railway bridge crack classification and recognition system and method. Background Art

[0002] With the rapid development of the transportation industry, railway bridges, as important transportation hubs connecting various places, their structural safety and durability are directly related to the safety of people's lives and property and the smoothness of transportation. However, during long-term use, due to the influence of various factors such as environmental factors (such as temperature changes, humidity fluctuations, chemical erosion, etc.), load effects (such as vehicle traffic, wind loads, etc.), and material aging, the railway bridge structure often shows varying degrees of damage, among which cracks are one of the most common and easily observable damage forms.

[0003] Traditionally, the detection and management of railway bridge cracks mainly rely on manual inspections. This method is not only inefficient but also vulnerable to human factors, making it difficult to achieve comprehensive and accurate monitoring and assessment of cracks. In recent years, with the rapid development of technologies such as computer vision, machine learning, and big data analysis, automated crack detection and classification systems based on image processing have gradually become a research hotspot. However, most of the existing technologies focus on the automatic identification and preliminary classification of cracks, and the ability to predict the crack propagation trend and risk assessment is still insufficient.

[0004] Specifically, current crack detection systems mostly extract the geometric features of cracks (such as length, width, direction, etc.) through image processing algorithms, but these systems often ignore the complex relationship between crack propagation and environmental factors. In fact, the propagation speed, direction, and morphology of cracks are not only affected by their own geometric characteristics but also closely related to the climate conditions (such as extreme temperatures, rainfall frequency, humidity changes, etc.) in which they are located. Therefore, the crack classification and assessment method based solely on geometric features is difficult to comprehensively reflect the actual situation and potential risks of cracks. Summary of the Invention

[0005] The purpose of this part is to outline some aspects of the embodiments of the present invention and briefly introduce some preferred embodiments. Some simplifications or omissions may be made in this part, as well as in the abstract and title of the present application, to avoid obscuring the purpose of this part, the abstract, and the title, and such simplifications or omissions shall not be used to limit the scope of the present invention.

[0006] To solve the above technical problems, the present invention provides the following technical solution: A railway bridge crack classification and recognition system mainly includes:

[0007] A data acquisition module, which integrates a variety of sensors and can collect multi-dimensional data of bridge cracks and the climate conditions of the environment where the bridge is located;

[0008] The crack feature analysis module receives the crack geometric feature data transmitted by the data acquisition module and uses deep learning algorithms to extract and classify the features of the cracks;

[0009] The crack risk assessment module evaluates the potential risk of the cracks according to the crack geometric features combined with the stress analysis model;

[0010] The climate-related trend prediction module is used to construct a crack propagation trend prediction model by combining climate conditions and geometric features;

[0011] The data fusion and decision-making module is used to fuse the results of the crack feature analysis module and the climate-related trend prediction module, and through the interactive analysis of multi-dimensional data, give the final classification and risk assessment results of the cracks;

[0012] The remote monitoring and alarm module can transmit the results of the comprehensive analysis to the bridge monitoring center through the wireless communication network, and automatically trigger an alarm signal when a dangerous crack propagation trend is detected.

[0013] As a preferred solution of the railway bridge crack classification and identification system described in the present invention, wherein: the sensors include high-definition cameras, laser scanners, depth sensors, thermal imagers and meteorological sensors.

[0014] As a preferred solution of the railway bridge crack classification and identification system described in the present invention, wherein: the multi-dimensional data includes the width, depth, orientation and density of the cracks.

[0015] As a preferred solution of the railway bridge crack classification and identification system described in the present invention, wherein: the steps of extracting and classifying the features of the cracks are as follows:

[0016] S11: Preprocess the collected crack data;

[0017] S12: Use a deep learning model to extract the geometric features of the cracks. The specific steps are as follows:

[0018] Calculate the maximum width, average width and width change rate of the cracks through the width sensor data as the width features for classification;

[0019] Obtain the maximum depth, average depth and gradual change of the depth of the cracks through the depth sensor data as the depth features for classification

[0020] Combine the width and depth of the cracks to generate a two-dimensional feature vector for subsequent classification;

[0021] S13: Define the criteria for crack classification according to the width and depth features of the cracks;

[0022] S14: Classify the cracks using a machine learning classification algorithm based on the crack width and depth features. The specific steps are as follows:

[0023] Utilize known crack data to construct a training dataset that includes width, depth, and crack classification results;

[0024] Use a convolutional neural network classification model to train the model according to the training data;

[0025] Adjust the model parameters by inputting the preprocessed crack width and depth features to optimize the classification accuracy;

[0026] According to the output results of the classification model, classify the cracks into minor cracks, moderate cracks, and severe cracks, and provide an evaluation result of the crack risk level;

[0027] S15: The system evaluates the output results of the classification model and adjusts the model according to the actual classification situation, which specifically includes:

[0028] Verify the accuracy of the classification model through historical crack data to ensure that the crack classification results conform to the actual situation;

[0029] According to the evaluation results, adjust the weight parameters of width and depth in the classification model to improve the classification accuracy;

[0030] S16: The system generates a report on the classification results and the corresponding crack geometric features.

[0031] As a preferred solution of the railway bridge crack classification and recognition system described in the present invention, wherein: the criteria for crack classification are as follows:

[0032] Crack width < 0.1mm and crack depth < 2mm are regarded as minor cracks;

[0033] Crack width between 0.1 - 0.3mm and depth between 2 - 5mm are regarded as moderate cracks, and such cracks need to be monitored key;

[0034] Crack width > 0.3mm and depth > 5mm are regarded as severe cracks, and such cracks pose a structural danger and need to be repaired immediately.

[0035] As a preferred solution of the railway bridge crack classification and recognition system described in the present invention, wherein: the steps for constructing the crack propagation trend prediction model include the following:

[0036] S21: Collect multi-dimensional data related to crack propagation through integrating multiple types of sensors, and then transmit the multi-dimensional data in real time through the data acquisition module. The sensors include but are not limited to:

[0037] A crack geometry parameter sensor is used to monitor the width W(t) and depth D(t) of bridge cracks in real time;

[0038] An environmental climate sensor is used to collect the climate conditions of the bridge environment, such as climate factors V of humidity H(t), temperature, and wind speed i (t);

[0039] A stress distribution sensor is used to obtain the dynamic change data S(t) of the stress distribution in the bridge structure and the influence of its different stress factors C i (t);

[0040] S22: Extract features and normalize the collected crack geometry data;

[0041] S23: Combine the stress distribution S(t) of the bridge structure and various stress factors C i (t) to establish a relationship model between stress and crack propagation;

[0042] S24: Combine the crack geometry features, stress distribution, and climate influence to construct a crack propagation trend prediction model;

[0043] S25: Dynamically calibrate the constructed crack propagation prediction model, optimize the model through historical monitoring data, and make it better adapt to the crack propagation under different bridge structures and different climate conditions.

[0044] As a preferred solution of the railway bridge crack classification and recognition system described in the present invention, wherein: the formula of the crack propagation trend prediction model is as follows:

[0045]

[0046] Wherein:

[0047] E(t) represents the predicted value of crack propagation at time t;

[0048] t is a time variable and changes with time;

[0049] W(t) represents the width of the crack at time t; W(t) represents the crack width data, which is dynamically updated with time, and the data is collected in real time through the sensor;

[0050] D(t) represents the depth of the crack at time t; D(t) represents the crack depth parameter, and the depth affects the danger of the crack, and is collected through the sensor or other detection devices;

[0051] H(t) represents the environmental humidity at time t, and the influence of humidity on crack propagation is reflected by an exponential function;

[0052] S(t) represents the stress distribution in the bridge structure at time t, and the stress distribution is calculated through a stress analysis model;

[0053] C i (t) is the effect of the i-th stress influencing factor at time t; C i (t) represents the contribution of the i-th stress in the stress analysis model to crack propagation during the crack propagation process;

[0054] V i (t) represents the climate condition parameters related to crack propagation at time t, indicating the influence of climate conditions on the crack propagation speed;

[0055] n represents the non-linear influence coefficient of crack width on the propagation trend, n > 1, reflecting the non-linear relationship between crack width and propagation trend. The larger the width, the higher the propagation risk index;

[0056] α, β, and γ are adjustment parameters in the model, used to control the contribution degrees of crack geometric characteristics, climate conditions, and stress distribution to the crack propagation trend;

[0057] α is the comprehensive influence coefficient of width and depth on crack propagation;

[0058] β is the attenuation adjustment coefficient in the humidity exponential function;

[0059] γ is the sensitivity coefficient of humidity to the crack propagation trend;

[0060] W(t) n ·D(t) is the geometric feature part of the crack, including width W(t) and depth D(t). The influence of width is adjusted by the non-linear exponent n, and D(t) directly affects the crack propagation risk. The larger the geometric feature, the higher the propagation risk;

[0061] S(t)·C i (t) is the product of the stress distribution S(t) and the i-th stress factor C i (t), representing the combined action of various different types of stresses on crack propagation under the stress distribution;

[0062] 1 + βe -γH(t) When the humidity H(t) is relatively large, the value of the exponential decay function is relatively small, indicating that the inhibitory effect of humidity on crack propagation gradually weakens; when the humidity H(t) is relatively small, the value of the exponential function is relatively large, indicating that crack propagation is more significantly affected by humidity;

[0063] Express the relationship between the geometric characteristics of cracks and the influence of climate factors; humidity regulates the crack propagation trend through the exponential function in the denominator, and geometric characteristics affect the crack propagation risk through the numerator. Overall, the greater the humidity, the larger the value of the denominator, resulting in a smaller value of the entire fraction, that is, the crack propagation speed slows down when the humidity is high; the larger the geometric characteristics, the larger the numerator, and the stronger the crack propagation trend.

[0064] Represents the mutual influence of stress and climate conditions. The denominator V i (t) represents climate-related dynamic factors such as temperature and wind speed; the numerator S(t)·C i (t) represents the contribution of stress. The more severe the climate conditions, the larger the value of the denominator, resulting in a decrease in the overall value of the fraction, that is, the climate conditions play an inhibitory role in crack propagation.

[0065] Range meaning:

[0066] The value range of E(t) is [0, ∞). The larger the value, the higher the risk of crack propagation.

[0067] As a preferred solution of the railway bridge crack classification and identification system described in the present invention, where:

[0068] When E(t) ≤ 10, at this time the crack width and depth are small, and the influence of the environmental climate on crack propagation is weak. The stress borne by the bridge is within the normal range, and the crack propagation trend is relatively stable. Regular inspections can be carried out, and no immediate maintenance actions are required.

[0069] When 10 < E(t) ≤ 30, at this time the geometric dimensions and stress factors of the cracks begin to significantly affect the propagation. Especially in an environment with high humidity and large temperature differences, the crack propagation speeds up. It is recommended to strengthen the monitoring and carry out local repairs in a timely manner to avoid further deterioration of the cracks.

[0070] When 30 < E(t) ≤ 50, the stress and climate conditions have a greater impact, and the crack propagation trend is rapid, which may cause damage to local areas of the bridge structure. It is recommended to arrange crack repairs as soon as possible to avoid further expansion and strengthen the stress management of the structure.

[0071] When E(t) > 50, the cracks have developed to a serious degree, which may pose a threat to the stability of the entire bridge structure. Immediate emergency measures need to be taken for comprehensive repairs and the overall safety of the bridge needs to be evaluated.

[0072] As a preferred solution of the railway bridge crack classification and identification system described in the present invention, where: the system further includes a crack self-healing analysis module for analyzing the natural healing ability of cracks under specific stress states and environmental conditions.

[0073] The recognition method using the above-mentioned railway bridge crack classification and recognition system includes the following steps:

[0074] Step 1: Use an integrated high-precision sensor to monitor bridge cracks in real time, collect geometric feature data of the cracks, and collect environmental data in combination with the climate conditions in the area where the bridge is located.

[0075] Step 2: Use a deep learning algorithm to process the crack collection data, extract geometric features, and perform a preliminary classification according to the changes in the crack width W(t) and depth D(t) to obtain the initial classification value of the crack.

[0076] In combination with the bridge structure model, use a stress sensor to monitor the stress distribution of the bridge, calculate the stress influence at the crack, extract the parameter S(t) related to crack propagation according to the geometric features of the crack and the distribution of the stress field, and form stress influence data.

[0077] Step 3: According to the crack geometric features and stress data in Step 2, construct a comprehensive risk assessment model for cracks, use the formula to calculate the propagation risk E(t) of each crack, and determine the risk level of the crack according to its value.

[0078] Step 4: Transmit the crack classification result and the risk assessment result to the remote bridge monitoring system in real time through a wireless communication network.

[0079] Advantages of the present invention:

[0080] In the present invention, the crack propagation trend prediction model constructed by combining crack geometric features and climate conditions can comprehensively analyze the physical properties of cracks and their propagation behavior under different climate conditions. This not only improves the prediction accuracy of the system for crack propagation trends but also can provide early warnings for potential dangerous cracks, greatly improving the accuracy and reliability of crack classification and risk assessment. Description of the drawings

[0081] In order to more clearly illustrate the technical solutions of the embodiments of the present invention, the following will briefly introduce the drawings required for the description of the embodiments. Obviously, the following drawings are only some embodiments of the present invention. For those of ordinary skill in the art, other drawings can be obtained based on these drawings without creative efforts. Among them:

[0082] Figure 1 It is a flow block diagram of a railway bridge crack classification and recognition system of the present invention.

[0083] Figure 2 It is a flow block diagram of the recognition method of a railway bridge crack classification and recognition system of the present invention. Specific embodiments

[0084] In order to make the above - mentioned objects, features, and advantages of the present invention more obvious and understandable, the following will provide a detailed description of the specific embodiments of the present invention in conjunction with the accompanying drawings of the specification.

[0085] In the following description, many specific details are set forth in order to fully understand the present invention. However, the present invention can also be implemented in other ways different from those described herein. Those skilled in the art can make similar generalizations without departing from the connotation of the present invention. Therefore, the present invention is not limited by the specific embodiments disclosed below.

[0086] Secondly, the so - called "one embodiment" or "embodiment" herein refers to a specific feature, structure, or characteristic that can be included in at least one implementation manner of the present invention. The phrase "in one embodiment" appearing in different places in this specification does not necessarily refer to the same embodiment, nor is it an independent or alternative embodiment that mutually excludes other embodiments.

[0087] Thirdly, the present invention is described in detail in conjunction with schematic diagrams. When detailing the embodiments of the present invention, for the sake of convenience of explanation, the cross - sectional views showing the device structure will be enlarged locally out of the general proportion, and the schematic diagrams are only examples and should not limit the scope of protection of the present invention herein. In addition, in actual production, three - dimensional spatial dimensions including length, width, and depth should be included.

[0088] Referring to Figure 1 , for an embodiment of the present invention, a railway bridge crack classification and recognition system is provided, which mainly includes:

[0089] A data acquisition module, which integrates a variety of sensors, including a high - definition camera, a laser scanner, a depth sensor, a thermal imager, and a meteorological sensor. This module can collect multi - dimensional data of bridge cracks, specifically including geometric feature information such as the width, depth, orientation, and density of cracks, as well as climate data (such as temperature, humidity, wind speed, etc.) of the environment where the bridge is located.

[0090] A crack feature analysis module, which receives the crack geometric feature data transmitted by the data acquisition module and uses deep - learning algorithms to extract and classify the features of the cracks.

[0091] A crack risk assessment module, which uses a stress analysis model to evaluate the potential risk of the cracks. The system calculates the relationship between the crack depth, width, and the stress condition of the bridge structure, and combines the fatigue life model of the bridge material to classify and evaluate the risk level of the cracks.

[0092] Climate-related trend prediction module. The climate-related trend prediction module constructs a crack propagation trend prediction model based on the climate conditions (such as temperature, humidity, wind speed, etc.) in the area where the bridge is located. This module analyzes the impact of climate conditions on crack propagation through machine learning algorithms and predicts the possible development trends of cracks under different climate conditions. Especially under extreme weather conditions, the system can predict the accelerated propagation trend of cracks in advance and provide early warnings for subsequent maintenance.

[0093] Data fusion and decision-making module. The data fusion and decision-making module fuses the results of the crack feature analysis module and the climate-related trend prediction module, and gives the final classification and risk assessment results of the cracks through the interactive analysis of multi-dimensional data. The system can dynamically adjust the crack propagation trend based on real-time monitoring data and historical monitoring data and provide real-time warnings.

[0094] Remote monitoring and alarm module. The remote monitoring and alarm module can transmit the results of comprehensive analysis to the bridge monitoring center through a wireless communication network and automatically trigger an alarm signal when a dangerous crack propagation trend is detected.

[0095] Specifically, the steps for feature extraction and classification of cracks are as follows:

[0096] S11: Preprocess the collected crack data. The steps include:

[0097] Denoising: Eliminate the noise in the data caused by factors such as the acquisition environment and sensor errors;

[0098] Normalization: Normalize data with different units and ranges (such as crack width, depth, etc.) so that they can be uniformly processed during feature extraction and classification;

[0099] Data alignment: Align the data such as the width, depth, and orientation of the cracks in terms of time and space for subsequent analysis.

[0100] S12: Use a deep learning model to extract the geometric features of the cracks. The specific steps are as follows:

[0101] Calculate the maximum width, average width, and width change rate of the cracks through width sensing data as the width features for classification;

[0102] Obtain the maximum depth, average depth, and gradual change of the depth of the cracks through depth sensing data as the depth features for classification

[0103] Combine the width and depth of the cracks to generate a two-dimensional feature vector (width, depth) for subsequent classification.

[0104] S13: Define the criteria for crack classification based on the width and depth features of the cracks. Specifically:

[0105] When the crack width < 0.1 mm and the crack depth < 2 mm, it is regarded as a minor crack;

[0106] When the crack width is between 0.1 - 0.3 mm and the depth is between 2 - 5 mm, it is regarded as a moderate crack, and such cracks need to be monitored key - point;

[0107] When the crack width > 0.3 mm and the depth > 5 mm, it is regarded as a severe crack. Such cracks pose a structural danger and need to be repaired immediately.

[0108] S14: Based on the crack width and depth characteristics, use a machine - learning classification algorithm to classify the cracks. The specific steps are as follows:

[0109] Use the known crack data to construct a training data set containing width, depth, and crack classification results;

[0110] Use classification models such as Support Vector Machine (SVM), Random Forest, or Convolutional Neural Network (CNN) to train the model according to the training data;

[0111] By inputting the pre - processed crack width and depth characteristics, adjust the model parameters to optimize the classification accuracy;

[0112] According to the output results of the classification model, classify the cracks into minor cracks, moderate cracks, and severe cracks, and provide an evaluation result of the crack risk level.

[0113] S15: The system evaluates the output results of the classification model and adjusts the model according to the actual classification situation. Specifically, it includes:

[0114] Verify the accuracy of the classification model through historical crack data to ensure that the crack classification results conform to the actual situation;

[0115] According to the evaluation results, adjust the weight parameters of width and depth in the classification model to improve the classification accuracy. For example, when both the crack width and depth are relatively large, the classification model will determine the crack as a high - risk crack.

[0116] S16: The system generates a report on the classification results and the corresponding crack geometric features. The report content includes:

[0117] The geometric features of the crack (width, depth, orientation, etc.);

[0118] The classification results of the crack (minor, moderate, severe);

[0119] The risk level evaluation of the crack (whether immediate repair is required);

[0120] The prediction of the crack development trend (predict the possibility of crack expansion according to climate conditions, etc.).

[0121] In this system, the classification of cracks is comprehensively evaluated based on the width and depth characteristics of the cracks. The larger the width and depth, the higher the risk level of the cracks. For cracks with a small width but a large depth, the system will evaluate their potential hazards and determine whether preventive measures need to be taken in combination with the stress analysis model.

[0122] Specifically, the construction of the crack propagation trend prediction model includes the following steps:

[0123] S21: Collect multi-dimensional data related to crack propagation through integrating multiple types of sensors, and then transmit the multi-dimensional data in real time through the data acquisition module. The sensors include but are not limited to:

[0124] Crack geometry parameter sensors, which are used to monitor the width W(t) and depth D(t) of the bridge cracks in real time;

[0125] Environmental climate sensors, which are used to collect the climate conditions of the environment where the bridge is located, such as humidity H(t), temperature, and climate factors such as wind speed V i (t);

[0126] Stress distribution sensors, which are used to obtain the dynamic change data S(t) of the stress distribution in the bridge structure and the influence of its different stress factors C i (t);

[0127] S22: Extract features and normalize the collected crack geometry data (width W(t) and depth D(t)); the influence of the width W(t) is adjusted by the non-linear exponential parameter n to enhance the sensitivity of the model to cracks with different widths.

[0128] The crack geometry features are processed through the following formula:

[0129]

[0130] S23: Combine the stress distribution S(t) of the bridge structure and various stress factors C i (t) to establish a relationship model between stress and crack propagation. The model is calculated through the following formula:

[0131]

[0132] S24: Combine the crack geometry features, stress distribution, and climate influence to construct a crack propagation trend prediction model, and the model is as follows:

[0133]

[0134] The formula integrates the comprehensive effects of the crack width W(t), depth D(t), stress distribution S(t), and climate conditions Vi(t) and H(t). High-precision prediction of crack propagation under complex environments is achieved through the non-linear adjustment factor n and the climate index adjustment factors β and γ. Additionally, the model uses the time integral function ∫ to represent the global prediction of the crack propagation trend over continuous time periods, enabling better prediction of the future crack propagation trend, especially the crack propagation speed under extreme climate conditions.

[0135] S25: Dynamically calibrate the constructed crack propagation prediction model, optimize the model through historical monitoring data to better adapt to crack propagation under different bridge structures and different climate conditions. The calibration process includes adjusting the parameters α, β, γ, and the exponent n to ensure the accuracy of the model. The calibrated model can provide real-time and accurate early warnings for bridge crack propagation.

[0136] In the above, the formula for the crack propagation trend prediction model is as follows:

[0137]

[0138] Where:

[0139] E(t) represents the predicted value of crack propagation at time t;

[0140] t is the time variable, which changes over time;

[0141] W(t) represents the width of the crack at time t; W(t) represents the crack width data, which is dynamically updated over time and the data is collected in real-time through sensors;

[0142] D(t) represents the depth of the crack at time t; D(t) represents the crack depth parameter, and the depth affects the danger of the crack, which is collected through sensors or other detection devices;

[0143] H(t) represents the environmental humidity at time t, and the influence of humidity on crack propagation is reflected through an exponential function;

[0144] S(t) represents the stress distribution in the bridge structure at time t, and the stress distribution is calculated through a stress analysis model;

[0145] C i (t) is the effect of the i-th stress influencing factor at time t; C i (t) represents the contribution of the i-th stress in the stress analysis model to crack propagation during the crack propagation process;

[0146] V i (t) represents the climate condition parameter related to crack propagation at time t, indicating the influence of climate conditions on the crack propagation speed;

[0147] n represents the non - linear influence coefficient of crack width on the expansion trend, where n > 1, reflecting the non - linear relationship between crack width and expansion trend. The larger the width, the higher the expansion risk index;

[0148] α, β, and γ are adjustment parameters in the model, used to control the contribution degrees of crack geometric characteristics, climate conditions, and stress distribution to the crack expansion trend;

[0149] α is the comprehensive influence coefficient of width and depth on crack expansion;

[0150] β is the attenuation adjustment coefficient in the humidity exponential function;

[0151] γ is the sensitivity coefficient of humidity to the crack expansion trend;

[0152] W(t) n ·D(t) is the geometric characteristic part of the crack, including width W(t) and depth D(t). The influence of width is adjusted by the non - linear exponent n, and D(t) directly affects the risk of crack expansion. The larger the geometric characteristics, the higher the expansion risk;

[0153] S(t)·C i (t) is the product of stress distribution S(t) and the i - th stress factor C i (t), representing the combined action of various different types of stresses on crack expansion under stress distribution;

[0154] 1 + βe -γH(t) When the humidity H(t) is relatively large, the value of the exponential decay function is small, indicating that the inhibitory effect of humidity on crack expansion gradually weakens; when the humidity H(t) is relatively small, the value of the exponential function is large, indicating that the crack expansion is more significantly affected by humidity;

[0155] Expresses the influence relationship between the geometric characteristics (width and depth) of the crack and the climate factor (humidity); humidity adjusts the crack expansion trend through the exponential function in the denominator, and geometric characteristics affect the crack expansion risk through the numerator. Overall, the larger the humidity, the larger the value of the denominator, resulting in the smaller value of the whole fraction, that is, the crack expansion speed slows down when the humidity is high; the larger the geometric characteristics, the larger the numerator, and the stronger the crack expansion trend;

[0156] Represents the mutual influence of stress and climate conditions. The denominator V i (t) represents climate - related dynamic factors such as temperature and wind speed; the numerator S(t)·C i (t) represents the contribution of stress. The more severe the climate conditions, the larger the value of the denominator, resulting in the reduction of the overall value of the fraction, that is, the climate conditions play an inhibitory role in crack expansion;

[0157] Range meaning:

[0158] The value range of E(t) is [0, ∞), and the larger the value, the higher the risk of crack propagation.

[0159] Specifically, when E(t) ≤ 10, the crack width and depth are small at this time, and the influence of environmental climate on crack propagation is weak. The stress borne by the bridge is within the normal range, and the crack propagation trend is relatively stable. Regular inspections are sufficient, and no immediate repair actions are required.

[0160] When 10 < E(t) ≤ 30, the geometric size and stress factors of the crack begin to significantly affect the propagation. Especially in an environment with high humidity and large temperature differences, the crack propagation accelerates. It is recommended to strengthen monitoring and conduct local repairs in a timely manner to avoid further deterioration of the crack.

[0161] When 30 < E(t) ≤ 50, the stress and climate conditions have a greater impact, and the crack propagation trend is rapid, which may cause damage to local areas of the bridge structure. It is recommended to arrange crack repairs as soon as possible to avoid further expansion and strengthen the stress management of the structure.

[0162] When E(t) > 50, the crack has developed to a serious degree, which may pose a threat to the stability of the entire bridge structure. Immediate emergency measures need to be taken for comprehensive repair, and the overall safety of the bridge needs to be evaluated.

[0163] By setting the above thresholds, it is possible to classify the risk level of bridge crack propagation based on the result E(t) predicted by the model, so as to achieve accurate early warning and risk assessment.

[0164] Specifically, the system also includes a crack self-healing analysis module for analyzing the natural healing ability of cracks under specific stress states and environmental conditions. By analyzing the climate conditions and changes in material properties, the possible self-healing behavior of cracks is predicted, providing a reference for auxiliary decision-making to help determine whether artificial intervention repair is required.

[0165] Refer to Figure 2 , and use the recognition method of the above-mentioned railway bridge crack classification and recognition system, including the following steps:

[0166] Step 1: Use an integrated high-precision sensor to monitor the bridge cracks in real time, collect the geometric feature data of the cracks, and combine the climate conditions in the area where the bridge is located to collect environmental data.

[0167] Step 2: Use a deep learning algorithm to process the crack acquisition data, extract geometric features, and according to the crack width W ( t ) and depth D ( t )Conduct a preliminary classification of the change situation to obtain the initial classification value of the crack.

[0168] Combined with the bridge structure model, use stress sensors to monitor the stress distribution of the bridge, calculate the stress influence at the crack, and extract the parameter S related to crack propagation according to the geometric characteristics of the crack and the distribution of the stress field. ( t ) to form stress influence data.

[0169] Step 3: According to the crack geometric characteristics and stress data in Step 2, construct a comprehensive risk assessment model for the crack, and use the formula to calculate the propagation risk E of each crack. ( t ) and determine the risk level of the crack according to its value.

[0170] Step 4: Transmit the crack classification result and the risk assessment result to the remote bridge monitoring system in real time through the wireless communication network.

[0171] In summary, the crack propagation trend prediction model constructed in the present invention by combining the crack geometric characteristics and climate conditions can comprehensively analyze the physical properties of the crack and its propagation behavior under different climate conditions. This not only improves the prediction accuracy of the system for the crack propagation trend, but also can early warn of potential dangerous cracks, greatly improving the accuracy and reliability of crack classification and risk assessment.

[0172] It should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention and not to limit them. Although the present invention has been described in detail with reference to the preferred embodiments, those of ordinary skill in the art should understand that the technical solutions of the present invention can be modified or equivalently replaced without departing from the spirit and scope of the technical solutions of the present invention, and they should all be covered within the scope of the claims of the present invention.

Claims

1. A railway bridge crack classification and recognition system, characterized in that, Including: A data acquisition module, which integrates a variety of sensors and can collect multi-dimensional data of bridge cracks and the climate conditions of the environment where the bridge is located; A crack feature analysis module, which receives the crack geometric feature data transmitted by the data acquisition module and uses deep learning algorithms to extract and classify the features of the cracks; A crack risk assessment module, which assesses the potential risk of cracks according to the crack geometric features combined with the stress analysis model; A climate-related trend prediction module, which is used to construct a crack propagation trend prediction model by combining climate conditions and geometric features; A data fusion and decision-making module, which is used to fuse the results of the crack feature analysis module and the climate-related trend prediction module, and through the interactive analysis of multi-dimensional data, give the final classification and risk assessment results of the cracks; A remote monitoring and alarm module, which can transmit the results of comprehensive analysis to the bridge monitoring center through a wireless communication network, and automatically trigger an alarm signal when a dangerous crack propagation trend is detected.

2. The railway bridge crack classification and recognition system according to claim 1, characterized in that: The sensors include high-definition cameras, laser scanners, depth sensors, thermal imagers and meteorological sensors.

3. The railway bridge crack classification and recognition system according to claim 1, characterized in that: The multi-dimensional data includes the width, depth, orientation and density of the cracks.

4. The railway bridge crack classification and recognition system according to claim 1, characterized in that: The steps of extracting and classifying the features of the cracks are as follows: S11: Preprocess the collected crack data; S12: Use a deep learning model to extract the geometric features of the cracks. The specific steps are as follows: Calculate the maximum width, average width and width change rate of the cracks through width sensing data as the width features for classification; Obtain the maximum depth, average depth and gradual change of the depth of the cracks through depth sensing data as the depth features for classification; Combine the width and depth of the cracks to generate a two-dimensional feature vector for subsequent classification; S13: Define the criteria for crack classification according to the width and depth features of the cracks; S14: Based on the width and depth features of the cracks, use a machine learning classification algorithm to classify the cracks. The specific steps are as follows: Use the known crack data to construct a training data set containing width, depth and crack classification results; Use a convolutional neural network classification model to train the model according to the training data; Adjust the model parameters by inputting the preprocessed crack width and depth features to optimize the classification accuracy; According to the output results of the classification model, classify the cracks into minor cracks, moderate cracks and severe cracks, and provide an assessment result of the crack risk; S15: The system evaluates the output results of the classification model and adjusts the model according to the actual classification situation, specifically including: Verify the accuracy of the classification model through historical crack data to ensure that the classification results of the cracks conform to the actual situation; According to the evaluation results, adjust the weight parameters of width and depth in the classification model to improve the classification accuracy; S16: The system generates a report on the classification results and the corresponding crack geometric features.

5. The railway bridge crack classification and recognition system according to claim 4, characterized in that: The criteria for the crack classification are as follows: Crack width < 0.1mm, crack depth < 2mm, regarded as minor cracks; The crack width is between 0.1 - 0.3 mm and the depth is between 2 - 5 mm, which is regarded as a moderate crack. Such cracks need to be monitored key points; The crack width > 0.3 mm and the depth > 5 mm are regarded as severe cracks. Such cracks pose a structural danger and need to be repaired immediately.

6. The railway bridge crack classification and recognition system according to claim 1, characterized in that: The construction of the crack propagation trend prediction model includes the following steps: S21: Collect multi-dimensional data related to crack propagation through an integrated multi-type sensor, and then transmit the multi-dimensional data in real time through a data acquisition module. The sensors include but are not limited to: A crack geometric parameter sensor for real-time monitoring of the width W(t) and depth D(t) of the bridge crack; An environmental climate sensor is used to collect the climate conditions of the environment where the bridge is located, such as climate factors V of humidity H(t), temperature, and wind speed i (t); A stress distribution sensor is used to obtain the dynamic change data S(t) of the stress distribution in a bridge structure and the influence of its different stress factors C i (t); S22: Extract features and normalize the collected crack geometric data; S23: Combine the stress distribution S(t) of the bridge structure and various stress factors C i (t) to establish a relationship model between stress and crack propagation; S24: Combine the crack geometric features, stress distribution, and climate impact to construct a crack propagation trend prediction model; S25: Dynamically calibrate the constructed crack propagation prediction model, and optimize the model through historical monitoring data to better adapt to the crack propagation under different bridge structures and different climate conditions.

7. The railway bridge crack classification and recognition system according to claim 6, characterized in that: The formula of the crack propagation trend prediction model is as follows: Where: E(t) represents the predicted value of crack propagation at time t; t is a time variable that changes with time; W(t) represents the width of the crack at time t; W(t) represents the crack width data, which is dynamically updated with time, and the data is collected in real time through sensors; D(t) represents the depth of the crack at time t; D(t) represents the crack depth parameter, and the depth affects the danger of the crack, which is collected through sensors or other detection devices; H(t) represents the environmental humidity at time t, and the impact of humidity on crack propagation is reflected by an exponential function; S(t) represents the stress distribution in the bridge structure at time t, and the stress distribution is calculated through a stress analysis model; C i (t) The effect of the i-th stress influencing factor at time t; C i (t) represents the contribution of the i-th stress in the stress analysis model during the crack propagation process to the crack propagation; V i (t) represents the climate condition parameter related to crack propagation at time t, indicating the influence of climate conditions on the crack propagation speed; n represents the non-linear influence coefficient of the crack width on the propagation trend, n > 1, which reflects the non-linear relationship between the crack width and the propagation trend. The larger the width, the higher the risk index of propagation; α, β, γ are adjustment parameters in the model, which are used to control the contribution of crack geometric features, climate conditions, and stress distribution to the crack propagation trend; α is the comprehensive influence coefficient of width and depth on crack propagation; β is the attenuation adjustment coefficient in the humidity exponential function; γ is the sensitivity coefficient of humidity to the crack propagation trend; W(t) n · D(t) is the geometric feature part of the crack, including the width W(t) and the depth D(t). The influence of the width is adjusted by the non-linear exponent n, while D(t) directly affects the risk of crack propagation. The larger the geometric feature, the higher the propagation risk; S(t)·C i (t) is the product of the stress distribution S(t) and the i-th stress factor C i (t), representing the combined effect of various types of stresses on crack propagation under the stress distribution; 1 + βe -γH(t) When the medium humidity H(t) is relatively large, the value of the exponential decay function is relatively small, indicating that the inhibitory effect of humidity on crack propagation gradually weakens; when the humidity H(t) is relatively small, the value of the exponential function is relatively large, indicating that the crack propagation is more significantly affected by humidity; Express the relationship between the geometric characteristics of cracks and the influence of climate factors; humidity regulates the crack propagation trend through the exponential function in the denominator, and the geometric characteristics affect the crack propagation risk through the numerator. Overall, the greater the humidity, the larger the value of the denominator, resulting in a smaller value of the entire fraction, that is, the crack propagation speed slows down when the humidity is high; the larger the geometric characteristics, the larger the numerator, and the stronger the crack propagation trend; represents the interaction between stress and climatic conditions, and the denominator V i (t) represents climate-related dynamic factors such as temperature and wind speed; the numerator S(t)·C i (t) represents the contribution of stress. The worse the climatic conditions, the larger the value of the denominator, resulting in a decrease in the overall value of the fraction, that is, the climatic conditions play an inhibitory role in crack propagation; Range meaning: The value range of E(t) is [0, ∞), and the larger the value, the higher the risk of crack propagation.

8. The railway bridge crack classification and identification system according to claim 7, characterized in that: When E(t) ≤ 10, at this time the crack width and depth are small, and the environmental climate has a weak impact on crack propagation. The stress borne by the bridge is within the normal range, and the crack propagation trend is relatively stable. Regular inspections are sufficient, and there is no need to take immediate repair actions; When 10 < E(t) ≤ 30, at this time the geometric size and stress factors of the crack begin to significantly affect the propagation. Especially in an environment with high humidity and large temperature differences, the crack propagation accelerates. It is recommended to strengthen monitoring and perform local repairs in a timely manner to avoid further deterioration of the crack; When 30 < E(t) ≤ 50, stress and climatic conditions have a greater impact, the crack propagation trend is rapid, which may cause damage to local areas of the bridge structure. It is recommended to arrange crack repair as soon as possible to avoid further expansion and strengthen the stress management of the structure; When E(t) > 50, the cracks have developed to a serious level, which may pose a threat to the stability of the entire bridge structure. Immediate emergency measures need to be taken for comprehensive repair and the overall safety of the bridge needs to be evaluated.

9. The railway bridge crack classification and recognition system according to claim 1, characterized in that: The system also includes a crack self-healing analysis module for analyzing the natural healing ability of cracks under specific stress states and environmental conditions.

10. The recognition method of a railway bridge crack classification and recognition system according to any one of claims 1-9, characterized in that, It includes the following steps: Step 1: Use integrated high-precision sensors to monitor bridge cracks in real time, collect geometric feature data of the cracks, and collect environmental data in combination with the climatic conditions of the area where the bridge is located; Step 2: Use deep learning algorithms to process the crack collection data, extract geometric features, and conduct a preliminary classification according to the changes in crack width W(t) and depth D(t) to obtain the initial classification value of the cracks; Combined with the bridge structure model, use stress sensors to monitor the stress distribution of the bridge, calculate the stress influence at the cracks, and extract the parameters S(t) related to crack propagation according to the geometric features of the cracks and the distribution of the stress field to form stress influence data; Step 3: According to the crack geometric features and stress data in Step 2, construct a comprehensive risk assessment model for cracks, calculate the propagation risk E(t) of each crack using the formula, and determine the risk level of the cracks based on its value; Step 4: Transmit the crack classification results and risk assessment results to the remote bridge monitoring system in real time through the wireless communication network.

Images