Bridge health monitoring method based on unmanned aerial vehicle

By using multi-drone collaborative patrols and intelligent data analysis, the problem of comprehensive coverage and real-time monitoring in bridge health monitoring has been solved, enabling efficient and accurate assessment and dynamic optimization of bridge structures.

CN118936561APending Publication Date: 2024-11-12SHANGHAI RESEARCH INSTITUTE OF BUILDING SCIENCES CO LTD
View PDF 0 Cites 19 Cited by

Patent Information

Application Number
CN202411031252.0
Authority / Receiving Office
CN · China
Patent Type
Applications(China)
Current Assignee / Owner
Filing Date
2024-07-30
Publication Date
2024-11-12

AI Technical Summary

Technical Problem

Existing bridge health monitoring methods rely on manual inspection and fixed sensors, which are time-consuming and labor-intensive, and cannot achieve comprehensive and real-time monitoring. Furthermore, the path planning and data processing of multi-drone collaborative operations have a low level of intelligence.

Method used

The system employs multi-UAV collaborative cruise path planning, equipped with high-resolution cameras and ultrasonic sensors. It transmits data in real time via wireless communication, and uses CNN and LSTM models to analyze image and acoustic signals. It dynamically adjusts the cruise path and data acquisition strategy to achieve comprehensive coverage and accurate assessment of the bridge.

Benefits of technology

It improves the efficiency and accuracy of bridge monitoring, ensures detailed monitoring of key parts, achieves comprehensive coverage and real-time feedback of bridge structures, and optimizes the flexibility and adaptability of monitoring schemes.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN118936561A_ABST
    Figure CN118936561A_ABST
Patent Text Reader

Abstract

The invention relates to the technical field of bridge monitoring, in particular to an unmanned aerial vehicle-based bridge health monitoring method, which comprises the following steps of: planning a cooperative cruise path of a plurality of unmanned aerial vehicles to ensure that an unmanned aerial vehicle group covers all key parts of a bridge; each unmanned aerial vehicle carries a camera and an ultrasonic sensor which are respectively responsible for acquiring bridge image information and bridge structure internal sound wave information; analyzing the bridge image information to obtain bridge apparent defects; sound wave information in the bridge structure is analyzed, and cracks and stress distribution hidden dangers in the bridge are obtained; performing fusion and deep analysis on bridge appearance defects, bridge internal cracks and stress distribution hidden dangers to obtain health state information of a bridge structure; and performing dynamic feedback and path optimization. According to the real-time monitoring data and the analysis result, the monitoring strategy is flexibly adjusted, accurate monitoring of important areas is ensured, and the adaptability and the continuous optimization capability of the monitoring scheme are improved.
Need to check novelty before this filing date? Find Prior Art

Description

Technical Field

[0001] The present invention relates to the technical field of bridge monitoring, and in particular to a bridge health monitoring method based on an unmanned aerial vehicle. Background Art

[0002] As a key infrastructure in the transportation network, the structural health of bridges is directly related to public safety and economic development. However, traditional bridge health monitoring methods mainly rely on manual inspection and fixed sensors, which is not only time-consuming and labor-intensive, but also often fails to timely and comprehensively detect structural hazards due to the subjectivity and limitations of manual inspection. Although fixed sensors can provide some real-time data, their installation and maintenance costs are high, and their coverage is limited, making it difficult to achieve all-round monitoring of large-span bridges.

[0003] In recent years, the development of drone technology has provided a new means for bridge health monitoring. Drones have the advantages of high efficiency, flexibility, and safety, and can quickly reach various parts of the bridge for data collection. However, the existing drone-based bridge health monitoring methods still have some problems, such as incomplete data collection, complex processing, and low degree of automation. In addition, the path planning of multi-drone collaborative work, the real-time and stability of data transmission, and the intelligent degree of data processing also need to be improved.

[0004] In the existing technology, most bridge health monitoring methods are limited to a single data collection method or lack intelligent data analysis and feedback mechanisms, making it difficult to achieve comprehensive monitoring and accurate evaluation of bridge structures. Summary of the invention

[0005] Based on the above objectives, the present invention provides a bridge health monitoring method based on drones.

[0006] The UAV-based bridge health monitoring method includes the following steps:

[0007] S1, multi-UAV collaborative cruise path planning: determine the key parts to be monitored according to the structural characteristics and monitoring requirements of the bridge, accurately plan the collaborative cruise paths of multiple UAVs to ensure that the UAV swarm covers all key parts of the bridge; UAVs communicate with each other in real time through wireless networks to achieve interconnection of flight status and position, and dynamically adjust flight paths to avoid collisions and optimize coverage;

[0008] S2, data collection: Each drone is equipped with a high-resolution camera and an ultrasonic sensor, which are responsible for collecting bridge image information and sound wave information inside the bridge structure respectively. The multi-channel wireless communication technology is used to transmit the collected bridge image information and internal information of the bridge structure to the ground control center in real time;

[0009] S3, health status information analysis: analyze the bridge image information to obtain the apparent defects of the bridge; analyze the internal sound wave information of the bridge structure to obtain the hidden dangers of internal cracks and stress distribution of the bridge; integrate and deeply analyze the apparent defects of the bridge, the internal cracks of the bridge, and the hidden dangers of stress distribution to obtain the health status information of the bridge structure;

[0010] S4, dynamic feedback and path optimization: According to the analysis results of health status information, the UAV cruising path and data collection strategy are dynamically adjusted to achieve continuous optimization of the monitoring plan.

[0011] Optionally, the multi-UAV collaborative cruising path planning in S1 specifically includes:

[0012] S11, Analysis of bridge structure characteristics and monitoring requirements:

[0013] Identify the geometry of bridges, including decks, piers, towers, cables, and trusses;

[0014] Determine the material type of the bridge, including concrete, steel, and composite materials, to determine the monitoring requirements of different materials;

[0015] Analyze the environment in which the bridge will be used, including climate conditions, traffic flow, and vibration frequency;

[0016] Key areas include bridge connections, load-bearing areas, stress concentration areas, and areas that have had problems in the past;

[0017] S12, using GIS technology to obtain the three-dimensional model of the bridge, generate a digital map of the bridge, and mark the key parts. The key parts are represented as a node set {N1, N2, ..., N k}, where each node N i Represents a key part of the bridge; after generating the digital map, determine the flight area A and flight altitude H f , the key point set P in the flight area = {P1,P2,...,P m}, where each point P j Indicates a location that the drone needs to cover, ensuring that all parts that need to be monitored are covered; determines the initial position and target position of the drone, plans the flight path of the drone, and uses the Dijkstra algorithm to calculate the optimal path of the drone:

[0018] S13, cruise path planning:

[0019] Plan the take-off point, landing point and cruise route of each drone to ensure that each drone covers its assigned monitoring area. Consider the relative position and flight trajectory between drones, and realize real-time position sharing and path coordination between drones through wireless communication network to avoid path overlap and collision. Drones share position information L = {(x1, y1, z1), (x2, y2, z2), ...} in real time through wireless network, and adjust the flight path according to the real-time position to avoid collision, which is expressed as: L t =L t-1 +ΔL, where L t represents the position of the UAV at time t, ΔL is the position change, and combined with the structural characteristics of the bridge, the dwelling time of the UAV at key locations and the frequency of data collection are determined to ensure detailed monitoring of important locations.

[0020] Optionally, the residence time calculation includes:

[0021] Drones in key locations i The residence time T i Determined by collection requirements and environmental impact, the calculation formula is: Among them, D i Indicates the amount of data, R i represents the data collection rate, and F is the environmental factor adjustment coefficient;

[0022] The data acquisition frequency is calculated as: data acquisition frequency f i Determined according to monitoring needs and drone capabilities, the calculation is:

[0023] Optionally, the S2 specifically includes:

[0024] Camera collects bridge image information: The drone uses GPS positioning technology to locate the key parts of the bridge, adjusts the flight altitude and angle, and aims the high-resolution camera at the target area. The camera captures the image of the bridge at a fixed time or according to preset conditions (such as reaching a specified location or detecting surface abnormalities) during the flight of the drone, covering the bridge deck, piers, and slings.

[0025] Ultrasonic sensors collect internal information of bridge structures: The ultrasonic sensors carried by the drones are aimed at the surface of the bridge and emit ultrasonic signals. The ultrasonic sensors transmit ultrasonic signals. The ultrasonic waves penetrate the bridge structure and are reflected back when encountering interfaces or defects of different materials. The reflected ultrasonic signals are received and the reflection time and intensity are recorded to form ultrasonic echo data.

[0026] Optionally, the analyzing the bridge image information in S3 to obtain the apparent defects of the bridge specifically includes:

[0027] S31, image preprocessing: denoising the bridge image captured by the camera to remove the noise in the image, and using histogram equalization to enhance the details and edge information in the image to make the apparent defects more obvious;

[0028] S32, feature extraction: Use the Sobel operator to extract edge information in the image, identify cracks, corrosion, and spalling defects on the bridge surface, and process the extracted edges through morphological closing operations to eliminate noise and discontinuous edges to obtain a complete defect outline;

[0029] S33, Defect Identification and Classification: Build and train a CNN model (convolutional neural network model) using a bridge image dataset to enable the network to automatically extract image features and identify defects. The training process includes:

[0030] Build CNN model: Design CNN architecture, including convolutional layer, pooling layer, and fully connected layer;

[0031] Data annotation: Manually annotate bridge images to determine the type and location of defects;

[0032] Model training: Use the labeled data set to train CNN, adjust the network parameters through the back propagation algorithm to improve the recognition accuracy of the model, test the model performance on the validation set, adjust the hyperparameters and network structure, optimize the recognition effect, input the preprocessed image into the trained CNN model, automatically identify the defects in the image, determine the type and location of the defects, and classify the defects into cracks, corrosion, and spalling types according to the output results of the CNN model, and record the detailed information of each defect;

[0033] S34, Defect quantification and evaluation: Quantify the identified defects, measure their length, width, and area, evaluate the severity of the defects, count the number and distribution of defects, and analyze the distribution pattern and concentrated areas of defects on the bridge surface.

[0034] Optionally, the defect quantification and evaluation includes setting the boundary points of the defect as {(x1,y1),(x2,y2),...,(x n ,y n )}, then the calculation formula of length L and width W is:

[0035]

[0036]

[0037] Count the number and distribution of defects, draw a defect distribution map, and analyze the distribution patterns and concentrated areas of defects on the bridge surface.

[0038] Optionally, the analyzing of the internal acoustic wave information of the bridge structure in S3 to obtain the hidden dangers of cracks and stress distribution inside the bridge specifically includes:

[0039] S34, denoising the sound wave signal collected by the ultrasonic sensor to remove environmental noise and sensor noise, and using wavelet transform to enhance detail information in the signal to make internal defects more obvious;

[0040] S35, feature extraction: perform time domain and frequency domain analysis on the received ultrasonic echo signal, extract feature parameters, including echo amplitude, echo time, and spectrum features, use continuous wavelet transform (CWT) to perform time-frequency analysis on the echo signal, extract time-frequency features, and identify the location and features of internal defects;

[0041] S36, defect identification and classification: By building an LSTM (Long Short-Term Memory) model, an LSTM network architecture for processing time series data is designed, including an input layer, an LSTM layer, and an output layer. The acoustic wave signal is manually labeled to determine the location of cracks and stress distribution hazards. The LSTM is trained using the labeled data set. The pre-processed acoustic wave signal is input into the trained LSTM model to automatically identify internal defects in the signal and determine the location of cracks and stress distribution hazards. According to the output results of the LSTM model, the defects are classified into cracks and stress concentration types, and detailed information of each defect is recorded. Defect identification is represented as:

[0042] Among them, W o is the output weight matrix, h t is the hidden state of LSTM, b o is the output bias, σ is the activation function;

[0043] S37, defect quantification and evaluation: quantify the identified cracks, measure their length, width, and depth, and evaluate the severity of the cracks. Suppose the boundary points of the cracks are {(t1,A1),(t2,A2),...,(t n ,A n )}, then the length L is calculated as:

[0044]

[0045] Evaluate the identified stress distribution hazards, measure the stress value and distribution of the stress concentration area, and assume that the stress value is σ and the position is (x, y, z). The stress distribution is expressed as:

[0046] σ(x,y,z)=f(x,y,z), where f is the stress distribution function.

[0047] Optionally, the acquisition of the health status information of the bridge structure specifically includes:

[0048] Using the 3D model of the bridge, the surface defect data and the internal crack and stress distribution data are spatially registered to ensure that different data types are integrated in the same spatial coordinate system;

[0049] Feature combination: Combine the apparent defect features with the internal defect features to form a complete defect feature vector: F = [Fapparent, Finternal], where Fapparent and Finternal represent the feature vectors of apparent defects and internal defects, respectively.

[0050] Optionally, the drone cruising path in S4 specifically includes:

[0051] Extract the bridge health assessment results, including apparent defects, internal cracks, stress distribution hazards and their location and severity, determine the monitoring priority according to the type, location and severity of the defects, generate a priority list, determine the key areas according to the order of the priority list (for example, take the top 20% as the key areas according to the overall percentage of the list), and re-plan the drone's cruise path to give priority to covering the identified key areas;

[0052] The path adjustment formula is as follows: Among them, d(P i , P i+1 ) represents the distance between adjacent points in the path, R j is the key area, w j is the weight of the key area, and λ is the adjustment coefficient.

[0053] Optionally, the dynamic adjustment of the data collection strategy in S4 specifically includes:

[0054] Adjustment of data collection frequency: According to the health status information, the data collection frequency of different areas is adjusted. The data collection frequency is increased for key areas and reduced for areas in good condition. The adjustment of data collection frequency is expressed as: f i =f base ×(1+k·r i ), where f base is the basic acquisition frequency, k is the adjustment coefficient, r i The regional risk level.

[0055] Beneficial effects of the present invention:

[0056] The present invention utilizes GIS and GPS technologies, combined with the geometric shape, material type and use environment of the bridge, to accurately plan the collaborative cruising paths of multiple UAVs. Through an intelligent optimization algorithm, it ensures that the UAV swarm can cover all key parts of the bridge, avoid path overlap and collision, and improve cruising efficiency. The path planning method can greatly improve the monitoring efficiency of UAVs, reduce flight time and energy consumption, achieve comprehensive coverage of bridge structures, and ensure the comprehensiveness and accuracy of data collection.

[0057] In the present invention, each drone is equipped with a high-resolution camera and an ultrasonic sensor, which are responsible for collecting the apparent defects and internal structure information of the bridge respectively. Through multi-channel wireless communication technology, the collected data is transmitted to the ground control center in real time, and the convolutional neural network (CNN) is used for image analysis, and the long short-term memory network (LSTM) is used for acoustic signal analysis to identify and classify the apparent defects, internal cracks and stress distribution hazards of the bridge, thereby achieving comprehensive detection and identification of surface and internal defects of the bridge, and improving the accuracy and processing efficiency of the data. Using CNN and LSTM for in-depth analysis can automatically extract features and accurately classify defects, providing reliable data support for the health status assessment of the bridge.

[0058] The present invention dynamically adjusts the cruise path and data collection strategy of the drone according to the analysis results of the health status information. Through real-time monitoring and feedback mechanisms, the data collection frequency and sensor configuration are optimized to ensure the monitoring accuracy of key areas. The calculation formulas and algorithms for path adjustment and collection strategy optimization further improve the flexibility and effectiveness of the monitoring scheme, and can flexibly adjust the monitoring strategy according to real-time monitoring data and analysis results to ensure accurate monitoring of important areas, and improve the adaptability and continuous optimization capabilities of the monitoring scheme. BRIEF DESCRIPTION OF THE DRAWINGS

[0059] In order to more clearly illustrate the technical solutions in the present invention or the prior art, the drawings required for use in the embodiments or the description of the prior art will be briefly introduced below. Obviously, the drawings in the following description are only for the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying creative work.

[0060] Figure 1 A schematic diagram of a monitoring method flow chart of an embodiment of the present invention;

[0061] Figure 2 The figure is a schematic diagram of the dynamic feedback and path optimization process according to an embodiment of the present invention. DETAILED DESCRIPTION

[0062] The present invention is described in detail below in conjunction with the accompanying drawings and specific embodiments. At the same time, it is explained here that in order to make the embodiments more detailed, the following embodiments are the best and preferred embodiments, and those skilled in the art may also adopt other alternatives to implement some known technologies; and the accompanying drawings are only for more specific description of the embodiments, and are not intended to specifically limit the present invention.

[0063] It should be noted that the references to "one embodiment", "an embodiment", "an exemplary embodiment", "some embodiments" and the like in the specification indicate that the embodiments described may include specific features, structures or characteristics, but not every embodiment may include the specific features, structures or characteristics. In addition, when a specific feature, structure or characteristic is described in conjunction with an embodiment, it should be within the knowledge of a person skilled in the art to implement such feature, structure or characteristic in conjunction with other embodiments (whether or not explicitly described).

[0064] In general, a term can be understood, at least in part, from its use in context. For example, depending, at least in part, on the context, the term "one or more" as used herein can be used to describe any feature, structure, or characteristic in the singular sense, or can be used to describe a combination of features, structures, or characteristics in the plural sense. Additionally, the term "based on" can be understood as not necessarily intended to convey an exclusive set of factors, but can instead, depending, at least in part, on the context, allow for the presence of other factors that are not necessarily explicitly described.

[0065] like Figure 1-Figure 2 As shown, the UAV-based bridge health monitoring method includes the following steps:

[0066] S1, multi-UAV collaborative cruise path planning: determine the key parts to be monitored according to the structural characteristics and monitoring requirements of the bridge, accurately plan the collaborative cruise paths of multiple UAVs to ensure that the UAV swarm covers all key parts of the bridge; UAVs communicate with each other in real time through wireless networks to achieve interconnection of flight status and position, and dynamically adjust flight paths to avoid collisions and optimize coverage;

[0067] S2, data collection: Each drone is equipped with a high-resolution camera and an ultrasonic sensor, which are responsible for collecting bridge image information and sound wave information inside the bridge structure respectively. The multi-channel wireless communication technology is used to transmit the collected bridge image information and internal information of the bridge structure to the ground control center in real time;

[0068] S3, health status information analysis: analyze the bridge image information to obtain the apparent defects of the bridge; analyze the internal sound wave information of the bridge structure to obtain the hidden dangers of internal cracks and stress distribution of the bridge; integrate and deeply analyze the apparent defects of the bridge, the internal cracks of the bridge, and the hidden dangers of stress distribution to obtain the health status information of the bridge structure;

[0069] S4, dynamic feedback and path optimization: According to the analysis results of health status information, the UAV cruising path and data collection strategy are dynamically adjusted to achieve continuous optimization of the monitoring plan.

[0070] The multi-UAV collaborative cruise path planning in S1 specifically includes:

[0071] S11, Analysis of bridge structure characteristics and monitoring requirements:

[0072] Identify the geometry of bridges, including decks, piers, towers, cables, and trusses;

[0073] Determine the material type of the bridge, including concrete, steel, and composite materials, to determine the monitoring requirements of different materials;

[0074] Analyze the environment in which the bridge will be used, including climate conditions, traffic flow, and vibration frequency;

[0075] Key areas include bridge connections, load-bearing areas, stress concentration areas, and areas that have had problems in the past;

[0076] S12, using GIS technology to obtain the three-dimensional model of the bridge, generate a digital map of the bridge, and mark the key parts. The key parts are represented as a node set {N1, N2, ..., N k}, where each node N i Represents a key part of the bridge; after generating the digital map, determine the flight area A and flight altitude H f , the key point set P in the flight area = {P1,P2,...,P m}, where each point P j Indicates a location that the drone needs to cover, ensuring that all parts that need to be monitored are covered; determines the initial position and target position of the drone, plans the flight path of the drone, and uses the Dijkstra algorithm to calculate the optimal path of the drone:

[0077] Among them, d(P i ,P i+1 ) represents the adjacent point P in the path i and P i+1 The distance between the initial position P0 and the target position P t , accurate navigation is achieved through GPS;

[0078] S13, cruise path planning:

[0079] Plan the take-off point, landing point and cruise route of each drone to ensure that each drone covers its assigned monitoring area. iPlan take-off point S i , Landing point D i and cruise route R i , where the cruise route R i Represented as a node sequence {P i1 ,P i2 ,...,P in};

[0080] Genetic algorithm is used for path optimization. The fitness function of genetic algorithm is:

[0081] Among them, c(U j ) indicates drone U j Considering the relative position and flight trajectory between UAVs, the real-time position sharing and path coordination between UAVs are realized through the wireless communication network to avoid path overlap and collision. UAVs share the position information L = {(x1, y1, z1), (x2, y2, z2), ...} in real time through the wireless network, and adjust the flight path according to the real-time position to avoid collision, which is expressed as: L t =L t-1 +ΔL, where L t represents the position of the UAV at time t, ΔL is the position change, and combined with the structural characteristics of the bridge, the dwelling time of the UAV at key locations and the frequency of data collection are determined to ensure detailed monitoring of important locations.

[0082] The dwell time calculation includes:

[0083] Drones in key locations i The residence time T i Determined by collection requirements and environmental impact, the calculation formula is: Among them, D i Indicates the amount of data, R i represents the data collection rate, and F is the environmental factor adjustment coefficient;

[0084] The data collection frequency is calculated as: Data collection frequency f i Determined according to monitoring needs and drone capabilities, the calculation is:

[0085] S2 specifically includes:

[0086] Camera collects bridge image information: The drone uses GPS positioning technology to locate the key parts of the bridge, adjusts the flight altitude and angle, and aims the high-resolution camera at the target area. The camera captures the image of the bridge at a fixed time or according to preset conditions (such as reaching a specified location or detecting surface abnormalities) during the flight of the drone, covering the bridge deck, piers, and slings.

[0087] Ultrasonic sensors collect internal information of bridge structures: The ultrasonic sensors carried by the drones are aimed at the surface of the bridge and emit ultrasonic signals. The ultrasonic sensors transmit ultrasonic signals. The ultrasonic waves penetrate the bridge structure and are reflected back when encountering interfaces or defects of different materials. The reflected ultrasonic signals are received and the reflection time and intensity are recorded to form ultrasonic echo data.

[0088] In S3, the bridge image information is analyzed to obtain the apparent defects of the bridge, including:

[0089] S31, image preprocessing: denoising the bridge image captured by the camera to remove the noise in the image, and using histogram equalization to enhance the details and edge information in the image to make the apparent defects more obvious;

[0090] S32, feature extraction: Use the Sobel operator to extract edge information in the image, identify cracks, corrosion, and spalling defects on the bridge surface, and process the extracted edges through morphological closing operations to eliminate noise and discontinuous edges to obtain a complete defect outline;

[0091] S33, Defect Identification and Classification: Build and train a CNN model (convolutional neural network model) using a bridge image dataset to enable the network to automatically extract image features and identify defects. The training process includes:

[0092] Build CNN model: Design CNN architecture, including convolutional layer, pooling layer, and fully connected layer;

[0093] Data annotation: Manually annotate bridge images to determine the type and location of defects;

[0094] Model training: Use the labeled data set to train CNN, adjust the network parameters through the back propagation algorithm to improve the recognition accuracy of the model, test the model performance on the validation set, adjust the hyperparameters and network structure, optimize the recognition effect, input the preprocessed image into the trained CNN model, automatically identify the defects in the image, determine the type and location of the defects, and classify the defects into cracks, corrosion, and spalling types according to the output results of the CNN model, and record the detailed information of each defect;

[0095] S34, Defect quantification and evaluation: Quantify the identified defects, measure their length, width, and area, evaluate the severity of the defects, count the number and distribution of defects, and analyze the distribution pattern and concentrated areas of defects on the bridge surface.

[0096] Design a CNN architecture, including convolutional layer, pooling layer, and fully connected layer. Suppose the input image is I and the convolution kernel is K, then the convolution operation formula is:

[0097] Among them, (x, y) is the output position, m and n are the sizes of the convolution kernel;

[0098] Data annotation: Manually annotate bridge images to determine the type and location of defects;

[0099] Model training: Use the labeled data set to train CNN, and the loss function is the cross entropy loss function:

[0100] Where N is the number of samples, y i is the true label, is the predicted value.

[0101] The model is tested on the validation set. The preprocessed images are input into the trained CNN model to automatically identify defects in the image and determine the type and location of the defects. The calculation formula for defect recognition is: Among them, W is the weight matrix, X is the input feature, b is the bias, and softmax is the normalized activation function:

[0102] Among them, C is the number of categories, z i is the score of the i-th category. According to the output results of the CNN model, the defects are classified into different types: cracks, corrosion, and spalling.

[0103] Defect quantification and evaluation include setting the boundary points of the defect as {(x1,y1),(x2,y2),...,(x n ,y n )}, then the calculation formula of length L and width W is:

[0104]

[0105] Count the number and distribution of defects, draw a defect distribution map, and analyze the distribution patterns and concentrated areas of defects on the bridge surface.

[0106] S3 analyzes the internal acoustic wave information of the bridge structure to obtain the hidden dangers of cracks and stress distribution inside the bridge, including:

[0107] S34, denoising the sound wave signal collected by the ultrasonic sensor to remove environmental noise and sensor noise, and using wavelet transform to enhance detail information in the signal to make internal defects more obvious;

[0108] S35, feature extraction: perform time domain and frequency domain analysis on the received ultrasonic echo signal, extract feature parameters, including echo amplitude, echo time, and spectrum features, use continuous wavelet transform (CWT) to perform time-frequency analysis on the echo signal, extract time-frequency features, and identify the location and features of internal defects;

[0109] S36, defect identification and classification: By building an LSTM (Long Short-Term Memory) model, an LSTM network architecture for processing time series data is designed, including an input layer, an LSTM layer, and an output layer. The acoustic wave signal is manually labeled to determine the location of cracks and stress distribution hazards. The LSTM is trained using the labeled data set. The pre-processed acoustic wave signal is input into the trained LSTM model to automatically identify internal defects in the signal and determine the location of cracks and stress distribution hazards. According to the output results of the LSTM model, the defects are classified into cracks and stress concentration types, and detailed information of each defect is recorded. Defect identification is represented as:

[0110] Among them, W o is the output weight matrix, h t is the hidden state of LSTM, b o is the output bias, σ is the activation function;

[0111] S37, defect quantification and evaluation: quantify the identified cracks, measure their length, width, and depth, and evaluate the severity of the cracks. Suppose the boundary points of the cracks are {(t1,A1),(t2,A2),...,(t n ,A n )}, then the length L is calculated as:

[0112]

[0113] Evaluate the identified stress distribution hazards, measure the stress value and distribution of the stress concentration area, and assume that the stress value is σ and the position is (x, y, z). The stress distribution is expressed as:

[0114] σ(x,y,z)=f(x,y,z), where f is the stress distribution function.

[0115] LSTM network architecture, including input layer, LSTM layer and output layer. Let the input signal be S and the input of LSTM unit be x t , the hidden state is h t , then LSTM is expressed as:

[0116] f t =σ(W f ·[h t-1 ,x t ]+b f );

[0117] i t =σ(W i ·[h t-1 ,x t ]+b i );

[0118] o t =σ(W o ·[h t-1 ,x t ]+b o );

[0119]

[0120] h t =o t *tanh(C t );

[0121] Among them, f t For the forget gate, i t is the input gate, o t is the output gate, C t is the cell state, h t In hidden state.

[0122] Use the labeled dataset to train LSTM, and the training is based on the mean square error loss function: Where N is the number of samples, y i is the true value, is the predicted value.

[0123] The acquisition of health status information of bridge structure specifically includes:

[0124] Using the 3D model of the bridge, the surface defect data and the internal crack and stress distribution data are spatially registered to ensure that different data types are integrated in the same spatial coordinate system;

[0125] Feature combination: Combine the apparent defect features with the internal defect features to form a complete defect feature vector: F = [Fapparent, Finternal], where Fapparent and Finternal represent the feature vectors of apparent defects and internal defects, respectively.

[0126] The specific cruising paths of drones in S4 include:

[0127] Extract the bridge health assessment results, including apparent defects, internal cracks, stress distribution hazards and their location and severity, determine the monitoring priority according to the type, location and severity of the defects, generate a priority list, determine the key areas according to the order of the priority list (for example, take the top 20% as the key areas according to the overall percentage of the list), and re-plan the drone's cruise path to give priority to covering the identified key areas;

[0128] The path adjustment formula is as follows: Among them, d(P i , P i+1) represents the distance between adjacent points in the path, R j is the key area, w j is the weight of the key area, and λ is the adjustment coefficient.

[0129] The dynamic adjustment of the data collection strategy in S4 specifically includes:

[0130] Adjustment of data collection frequency: According to the health status information, the data collection frequency of different areas is adjusted. The data collection frequency is increased for key areas and reduced for areas in good condition. The adjustment of data collection frequency is expressed as: f i =f base ×(1+k·r i ), where f base is the basic acquisition frequency, k is the adjustment coefficient, r i The regional risk level.

[0131] Adjust the configuration and use strategy of drone sensors according to the defect type and location. For example, increase the use frequency of ultrasonic sensors in stress concentration areas and increase the use frequency of high-resolution cameras in apparent defect concentration areas.

[0132] The present invention covers any substitution, modification, equivalent method and scheme made on the essence and scope of the present invention. In order to make the public have a thorough understanding of the present invention, specific details are described in detail in the following preferred embodiments of the present invention, but those skilled in the art can fully understand the present invention without the description of these details. In addition, in order to avoid unnecessary confusion about the essence of the present invention, well-known methods, processes, procedures, components and circuits are not described in detail.

[0133] The above is only a preferred embodiment of the present invention. It should be pointed out that for ordinary technicians in this technical field, several improvements and modifications can be made without departing from the principle of the present invention. These improvements and modifications should also be regarded as the scope of protection of the present invention.

Claims

1. The bridge health monitoring method based on drone is characterized by: The following steps are involved: S1, multi-UAV collaborative cruise path planning: According to the structural characteristics and monitoring requirements of the bridge, the key parts to be monitored are determined, and the collaborative cruise paths of multiple UAVs are planned to ensure that the UAV swarm covers all key parts of the bridge; UAVs communicate with each other in real time through wireless networks to achieve interconnection of flight status and position, and dynamically adjust flight paths to avoid collisions and optimize coverage; S2, data collection: Each drone is equipped with a camera and an ultrasonic sensor, which are responsible for collecting bridge image information and sound wave information inside the bridge structure respectively. The multi-channel wireless communication technology is used to transmit the collected bridge image information and internal information of the bridge structure to the ground control center in real time; S3, health status information analysis: analyze the bridge image information to obtain the apparent defects of the bridge; analyze the internal sound wave information of the bridge structure to obtain the hidden dangers of internal cracks and stress distribution of the bridge; integrate and deeply analyze the apparent defects of the bridge, the internal cracks of the bridge, and the hidden dangers of stress distribution to obtain the health status information of the bridge structure; S4, dynamic feedback and path optimization: According to the analysis results of health status information, the UAV cruising path and data collection strategy are dynamically adjusted to achieve continuous optimization of the monitoring plan.

2. The method for monitoring bridge health based on drone according to claim 1, characterized in that: The multi-UAV collaborative cruising path planning in S1 specifically includes: S11, Analysis of bridge structure characteristics and monitoring requirements: Identify the geometry of bridges, including decks, piers, towers, cables, and trusses; Determine the material type of the bridge, including concrete, steel, and composite materials, to determine the monitoring requirements of different materials; Analyze the environment in which the bridge will be used, including climate conditions, traffic flow, and vibration frequency; Key areas include bridge connections, load-bearing areas, stress concentration areas, and areas that have had problems in the past; S12, using GIS technology to obtain the three-dimensional model of the bridge, generate a digital map of the bridge, and mark the key parts. The key parts are represented as a node set {N1, N2, ..., N k }, where each node N i Represents a key part of the bridge; after generating the digital map, determine the flight area A and flight altitude H f , the key point set P in the flight area = {P1,P2,...,P m }, where each point P j Indicates a location that the drone needs to cover, ensuring that all areas that need to be monitored are covered; determines the initial and target locations of the drone, and plans the flight path of the drone; S13, cruise path planning: Plan the take-off point, landing point and cruise route of each drone to ensure that each drone covers its assigned monitoring area. Consider the relative position and flight trajectory between drones, and realize real-time position sharing and path coordination between drones through wireless communication network to avoid path overlap and collision. Drones share position information L = {(x1, y1, z1), (x2, y2, z2), ...} in real time through wireless network, and adjust the flight path according to the real-time position to avoid collision, which is expressed as: L t =L t-1 +ΔL, where L t represents the position of the UAV at time t, ΔL is the position change, and combined with the structural characteristics of the bridge, the dwelling time of the UAV at key locations and the frequency of data collection are determined to ensure detailed monitoring of important locations.

3. The method for bridge health monitoring based on drone according to claim 2, characterized in that: The residence time calculation includes: Drones in key locations i The residence time T i Determined by collection requirements and environmental impact, the calculation formula is: Among them, D i Indicates the amount of data, R i represents the data collection rate, and F is the environmental factor adjustment coefficient; The data acquisition frequency is calculated as: data acquisition frequency f i Determined according to monitoring needs and drone capabilities, the calculation is:

4. The method for monitoring bridge health based on drone according to claim 1, characterized in that: The S2 specifically includes: Camera collects bridge image information: The drone uses GPS positioning technology to locate the key parts of the bridge, adjusts the flight altitude and angle, and aims the high-resolution camera at the target area. The camera captures the image of the bridge at a fixed time or according to preset conditions during the flight of the drone; Ultrasonic sensors collect internal information of bridge structures: The ultrasonic sensors carried by the drones are aimed at the surface of the bridge and emit ultrasonic signals. The ultrasonic sensors transmit ultrasonic signals. The ultrasonic waves penetrate the bridge structure and are reflected back when encountering interfaces or defects of different materials. The reflected ultrasonic signals are received and the reflection time and intensity are recorded to form ultrasonic echo data.

5. The method for bridge health monitoring based on drone according to claim 1, characterized in that: The analyzing of the bridge image information in S3 to obtain the apparent defects of the bridge specifically includes: S31, image preprocessing: denoising the bridge image captured by the camera to remove the noise in the image, and using histogram equalization to enhance the details and edge information in the image to make the apparent defects more obvious; S32, feature extraction: Use the Sobel operator to extract edge information in the image, identify cracks, corrosion, and spalling defects on the bridge surface, and process the extracted edges through morphological closing operations to eliminate noise and discontinuous edges to obtain a complete defect outline; S33, Defect Identification and Classification: Build and train a CNN model, using a bridge image dataset for training, so that the network can automatically extract image features and identify defects. The training process includes: Build CNN model: Design CNN architecture, including convolutional layer, pooling layer, and fully connected layer; Data annotation: Manually annotate bridge images to determine the type and location of defects; Model training: Use the labeled data set to train CNN, test the model performance on the validation set, adjust the hyperparameters and network structure, optimize the recognition effect, input the preprocessed image into the trained CNN model, automatically identify the defects in the image, determine the type and location of the defects, and classify the defects into cracks, corrosion, and spalling types based on the output results of the CNN model; S34, Defect quantification and evaluation: Quantify the identified defects, measure their length, width, and area, evaluate the severity of the defects, count the number and distribution of defects, and analyze the distribution pattern and concentrated areas of defects on the bridge surface.

6. The method for bridge health monitoring based on drone according to claim 1, characterized in that: The defect quantification and evaluation includes setting the boundary points of the defect as {(x1,y1),(x2,y2),...,(x n ,y n )}, then the calculation formula of length L and width W is: Count the number and distribution of defects, draw a defect distribution map, and analyze the distribution patterns and concentrated areas of defects on the bridge surface.

7. The method for bridge health monitoring based on drone according to claim 6, characterized in that: The analysis of the acoustic wave information inside the bridge structure in S3 to obtain the hidden dangers of cracks and stress distribution inside the bridge specifically includes: S34, denoising the sound wave signal collected by the ultrasonic sensor to remove environmental noise and sensor noise, and using wavelet transform to enhance detail information in the signal to make internal defects more obvious; S35, feature extraction: perform time domain and frequency domain analysis on the received ultrasonic echo signal, extract feature parameters, including echo amplitude, echo time, and spectrum features, use continuous wavelet transform to perform time-frequency analysis on the echo signal, extract time-frequency features, and identify the location and features of internal defects; S36, defect identification and classification: By building an LSTM model, an LSTM network architecture for processing time series data is designed, including an input layer, an LSTM layer, and an output layer. The acoustic wave signal is manually labeled to determine the location of cracks and stress distribution hazards. The labeled data set is used to train the LSTM. The preprocessed acoustic wave signal is input into the trained LSTM model to automatically identify internal defects in the signal and determine the location of cracks and stress distribution hazards. According to the output results of the LSTM model, the defects are classified into cracks and stress concentration types, and detailed information of each defect is recorded. Defect identification is represented as: Among them, W o is the output weight matrix, h t is the hidden state of LSTM, b o is the output bias, σ is the activation function; S37, defect quantification and evaluation: quantify the identified cracks, measure their length, width, and depth, and evaluate the severity of the cracks. Suppose the boundary points of the cracks are {(t1,A1),(t2,A2),...,(t n ,A n )}, the length L is calculated as: Evaluate the identified stress distribution hazards, measure the stress value and distribution of the stress concentration area, and assume that the stress value is σ and the position is (x, y, z). The stress distribution is expressed as: σ(x,y,z)=f(x,y,z), where f is the stress distribution function.

8. The method for bridge health monitoring based on drone according to claim 7, characterized in that: The acquisition of the health status information of the bridge structure specifically includes: Using the 3D model of the bridge, the surface defect data and the internal crack and stress distribution data are spatially registered to ensure that different data types are integrated in the same spatial coordinate system; Feature combination: Combine the apparent defect features with the internal defect features to form a complete defect feature vector: F = [Fapparent, Finternal], where Fapparent and Finternal represent the feature vectors of apparent defects and internal defects, respectively.

9. The method for monitoring bridge health based on drone according to claim 2, characterized in that: The UAV cruising path in S4 specifically includes: Extract the bridge health assessment results, including surface defects, internal cracks, stress distribution hazards and their locations and severity, determine the monitoring priority according to the type, location and severity of the defects, generate a priority list, determine the key areas according to the order of the priority list, and re-plan the drone's cruise path to give priority to covering the identified key areas; The path adjustment formula is as follows: Among them, d(P i , P i+1 ) represents the distance between adjacent points in the path, R j is the key area, w j is the weight of the key area, and λ is the adjustment coefficient.

10. The method for bridge health monitoring based on drone according to claim 9, characterized in that: The dynamic adjustment of the data collection strategy in S4 specifically includes: Adjustment of data collection frequency: According to the health status information, the data collection frequency of different areas is adjusted. The data collection frequency is increased for key areas and reduced for areas in good condition. The adjustment of data collection frequency is expressed as: f i =f base ×(1+k·r i ), where f base is the basic acquisition frequency, k is the adjustment coefficient, r i The regional risk level.

Citation Information

Cited By

  • Multi-source data fusion pier collision damage detection device and method

    CN119309633A

  • Building material quality detection method and system

    CN119395237A

  • Bridge structure detection method based on unmanned aerial vehicle, product, medium and equipment

    CN119510419A

  • Intelligent bridge support detection and maintenance management system and method

    CN119714433A

  • Monitoring and early warning system and method for bridge pier column

    CN119714747A