Expressway collapse monitoring and early warning system and method

Abstract

The invention belongs to the technical field of road traffic safety, and discloses an expressway collapse monitoring and early warning system and method, and the method comprises the steps: a route planning module sets a flight route and cruise parameters of an unmanned plane; the unmanned aerial vehicle performs cruise monitoring according to a set route; the pavement defect recognition module analyzes abnormal conditions of the images in real time; a radar detector timely recognizes potential collapse risk conditions of the cracks and the holes; the collapse anomaly analysis module comprehensively analyzes the collapse risk of the highway pavement in combination with the recognition results of the radar detector and the pavement defect recognition module; when the collapse risk is detected, the early warning module triggers the alarm; the unmanned aerial vehicle hovers in front of a road section with the collapse hidden danger, and warning characters are displayed through a red light LED lamp on the warning lamp panel assembly. According to the invention, the expressway pavement is monitored in all directions in an unmanned aerial vehicle cruise monitoring mode, and the landslide risk of the highway is comprehensively analyzed in combination with the recognition results of the radar detector and the pavement defect recognition module.

Description

Technical Field

[0001] This application relates to the field of road traffic safety technology, and more specifically, to a highway landslide monitoring and early warning system and method. Background Technology

[0002] Existing technology publication CN108510797A discloses a highway early warning system and method based on radar detection. The system includes several radar detection devices installed along the highway side. These devices transmit radar microwave signals to their coverage area at predetermined time intervals to obtain the speed, straight-line distance, and azimuth information of moving targets within the coverage area. This information, combined with the device installation location information, yields vehicle driving status data. A sudden event detection server receives the vehicle driving status data detected by the radar detection devices, performs real-time detection of sudden events, and determines the type of sudden event. Several early warning devices installed along the highway side automatically issue corresponding early warning information based on the sudden event signals emitted by the sudden event detection server. This invention can extract potential sudden events and issue early warnings by real-time detection of vehicle driving status, and is unaffected by weather and lighting conditions.

[0003] While the existing technical solutions described above can achieve the relevant beneficial effects through their structure, they still have the following drawbacks: 1. They cannot quickly and comprehensively inspect and monitor highways; they cannot promptly detect potential landslide hazards on the road. 2. They cannot take timely measures to address sections with potential landslide hazards, and cannot prevent vehicles from continuing to drive over these sections, thus accelerating their collapse. 3. Monitoring landslide hazards solely through radar detection equipment lacks sufficient accuracy.

[0004] In view of this, we propose a landslide monitoring and early warning system and method for highways. Summary of the Invention

[0005] 1. Technical problems to be solved

[0006] The purpose of this application is to provide a landslide monitoring and early warning system and method for highways, which solves the technical problems mentioned in the background art. It realizes the ability to monitor the highway surface in all directions through drone patrol monitoring, and to conduct a comprehensive analysis of the landslide risk of the highway by combining the identification results of radar detectors and road surface defect identification modules. It combines the road surface conditions with the radar monitoring results for comprehensive analysis and timely identification of road sections with the possibility of landslides.

[0007] 2. Technical Solution

[0008] This application provides a landslide monitoring and early warning system for highways, including: a drone, a weather sensor, a radar detector, a high-definition camera, a route planning module, a road surface defect identification module, a PLC control module, a landslide anomaly analysis module, and an early warning module; the drone is preferably a waterproof drone, which can conduct patrol monitoring in rainy weather.

[0009] The drone is equipped with a fixed radar detector, a high-definition camera, and a weather sensor.

[0010] High-definition images of the road surface are captured by a high-definition camera; night vision and infrared shooting are supported to ensure clear images can be obtained under various lighting conditions.

[0011] Radar detectors are used to identify defects such as cracks and holes in road surfaces. These detectors are characterized by high resolution and high penetration, and are used to detect potential landslide risks such as cracks and holes in road surfaces.

[0012] Weather sensors are used to monitor weather conditions in the flight area in real time, including parameters such as temperature, humidity, wind speed, wind direction, and rainfall. Weather sensors are essential for timely understanding of local weather conditions, with particular emphasis on monitoring rainy days; they also provide real-time weather information, especially rainfall amount and predicted rainfall trends.

[0013] Route planning module: Based on highway route maps and topographic maps, the software algorithm plans the optimal flight route for the drone, including the starting point, destination, and alternate charging point, to ensure that the drone can complete the patrol mission efficiently and safely.

[0014] Road surface defect recognition module: Based on the collected images, and using image processing and machine learning technologies, the module performs real-time analysis on the images captured by the high-definition camera to identify abnormalities such as cracks, potholes, and bumps on the road surface.

[0015] The collapse anomaly analysis module combines the identification results from radar detectors and the road surface defect recognition module to comprehensively analyze the collapse risk of highway pavements and promptly identify road sections with a potential collapse possibility. It integrates road surface conditions with radar monitoring results for comprehensive analysis. If road surface anomalies coincide with anomalies in underground structures detected by radar, especially in areas showing instability across multiple indicators, a collapse hazard should be highly suspected. Special attention should be paid to road sections located near mountains or slopes, as these areas are more susceptible to geological movements.

[0016] The early warning module includes an alarm and warning light panel assembly. When a landslide risk is detected, the alarm sounds an audible alert, and the warning light panel displays a warning message. Simultaneously, the warning information is transmitted in real-time to the highway traffic control center via a communication module, and passing vehicles and relevant departments are notified through broadcasts and SMS messages. The alarm sounds an audible alert, and the warning light panel displays "Landslide risk ahead, please stop and drive carefully" in red. The potential landslide information is also transmitted wirelessly to the highway traffic control center, and broadcasts a warning that there is a landslide hazard on this section of road, advising vehicles to detour. At the same time, the highway bureau will arrange traffic control and emergency repairs for the potentially landslide-prone section as soon as possible.

[0017] PLC control module: Network-connected with collapse anomaly analysis module, early warning module, drone, weather sensor, radar detector, high-definition camera and route planning module.

[0018] Using the above technical solution, before takeoff, the drone's flight route and cruise parameters are set via a route planning module. The drone cruises along the set route while simultaneously monitoring weather conditions in real time using weather sensors. A high-definition camera continuously captures images of the road surface, transmitting the collected images to a road defect identification module for processing. This module analyzes the images in real time, identifying anomalies such as cracks, potholes, and bumps on the road surface. A radar detector scans the road surface, identifying potential landslide risks such as cracks and holes. A landslide anomaly analysis module combines the identification results from the radar detector and the road defect identification module to comprehensively analyze the landslide risk of the road surface. When a landslide risk is detected, the early warning module triggers the alarm and warning light panel components, and sends the warning information to the highway traffic control center and relevant departments via a communication module. Based on the received warning information, the highway traffic control center promptly takes traffic control and repair measures to ensure safe and smooth road traffic.

[0019] As an optional embodiment of the present invention, the radar detector identifies defects such as cracks and holes in the road surface; specifically, it includes signal processing, image analysis, and defect identification.

[0020] Signal processing: After receiving reflected signals from the road surface, radar detectors often contain a significant amount of noise and interference. This noise may originate from the environment (such as electromagnetic interference, weather conditions, etc.) or the radar system itself (such as hardware noise, system errors, etc.). Therefore, signal processing is the primary step for radar detectors to identify defects.

[0021] The primary goal of signal processing is to improve the signal-to-noise ratio (SNR) and image quality. This typically involves the following steps:

[0022] Filtering: By designing appropriate filters, such as low-pass filters, high-pass filters, and band-pass filters, high-frequency noise or interference at specific frequencies can be removed from a signal.

[0023] Enhancement: Increasing the amplitude or energy of a useful signal to improve its visibility against a noisy background. This can be achieved through various enhancement algorithms, such as contrast enhancement and brightness enhancement.

[0024] Signal reconstruction: After filtering and enhancement, the signal may need to be reconstructed or resampled to better match the needs of subsequent image analysis.

[0025] Image analysis: After signal processing, radar images reveal different layers and characteristics of the road surface structure. Cracks and holes typically appear as specific reflection patterns or anomalous areas on the image. These anomalous areas may differ from the surrounding road surface structure in terms of reflection intensity, reflection pattern, or geometry.

[0026] To extract these anomalous regions, various image analysis techniques can be used, including:

[0027] Thresholding segmentation: By setting one or more thresholds, an image is divided into different regions or categories. In radar images, features such as reflection intensity or reflection pattern can be used as the basis for thresholding segmentation.

[0028] Edge detection: This technique uses edge information in an image to identify defects such as cracks and holes. Edge detection algorithms can detect locations where there are abrupt changes in reflection intensity or reflection pattern in an image, thereby determining the boundaries of cracks and holes.

[0029] Morphological processing: This involves using morphological operations (such as erosion, dilation, opening, and closing operations) to enhance or remove specific structures or features in an image. These operations can help highlight the shape and size characteristics of defects such as cracks and holes.

[0030] Defect identification: After extracting the abnormal areas, further defect identification is required. This includes several steps:

[0031] Feature extraction: Representative and discriminative features, such as reflection intensity, reflection pattern, shape, and size, are extracted from the abnormal regions. These features will be used for subsequent defect classification and identification.

[0032] Classifier Design: Based on the extracted features, design a suitable classifier to distinguish different types of defects (such as cracks, holes, etc.). Classifiers can be statistical (such as support vector machines, Naive Bayes, etc.), rule-based (such as decision trees, random forests, etc.), or learning-based (such as neural networks, deep learning, etc.).

[0033] Model validation: The classifier is trained and validated using sample data with known labels to evaluate its performance and accuracy. This typically includes methods such as cross-validation and hold-out validation.

[0034] Decision Support: Once the classifier is designed and validated, it can be applied to real radar images to automatically identify and quantify defects such as cracks and holes. These results can provide decision support for road maintenance and repair, helping to determine areas requiring repair and appropriate repair methods.

[0035] As an optional solution of the present invention, when planning the optimal route for the UAV to perform a cruise mission on the highway, the route planning module will integrate the highway route map, topographic map and real-time environmental data, and through a series of complex software algorithms, ensure that the UAV can complete the predetermined task efficiently and safely. Specifically, it includes the following steps: data collection and preprocessing, starting point and ending point setting, path optimization and verification, task allocation and monitoring, alternate landing charging point planning and path planning.

[0036] Data collection and preprocessing: Collect data such as highway route maps, topographic maps, real-time environmental data, and UAV performance parameters;

[0037] Highway route map: Get the latest highway network layout map, including information such as road directions, entrance and exit locations, and service area locations.

[0038] Topographic maps: Acquire high-precision topographic data, including elevation, slope, and landform type (such as mountains, plains, rivers, etc.).

[0039] Real-time environmental data: Real-time meteorological information such as wind speed, wind direction, temperature, and humidity are obtained through meteorological services and environmental monitoring stations, as well as real-time road condition information such as possible traffic congestion and construction sections.

[0040] Drone performance parameters: These parameters include the drone's flight speed, maximum flight altitude, maximum range, endurance, and payload capacity.

[0041] Start and End Point Settings: Based on the requirements of the cruise mission, set the takeoff point (start) and landing point (end) for the drone. Ensure these points are located in suitable positions along the highway to facilitate drone takeoff and landing.

[0042] Alternate charging point planning: Based on the drone's endurance and the length of the cruise mission, select suitable service areas or safe areas along the highway as alternate charging points. Consider the distance between alternate charging points to ensure that the drone can safely reach the next alternate charging point before its battery runs out.

[0043] Path planning: Using advanced path planning algorithms (such as Dijkstra's algorithm, A* algorithm, genetic algorithm, etc.), combined with highway route maps, topographic maps, and real-time environmental data, the optimal flight path from the starting point to the destination is calculated, taking into account alternate landing and charging points. During path planning, multiple factors must be comprehensively considered, such as flight distance, flight time, flight safety (avoiding no-fly zones, terrain obstacles, etc.), and flight efficiency.

[0044] Path optimization and validation: The initially planned path is further optimized, such as by adjusting parameters like flight altitude and speed, to improve flight efficiency and safety. The feasibility and effectiveness of the planned path are validated through simulated flight or actual test flights, and necessary adjustments are made based on actual conditions.

[0045] Task allocation and monitoring: The planned flight path and mission instructions are sent to the UAV to ensure it executes the patrol mission according to the planned path. During the mission, the UAV's position, status, battery level, and other information are monitored in real time through a monitoring system to ensure successful mission completion. Considering potential emergencies (such as severe weather, equipment failure, etc.), corresponding emergency response mechanisms are established, such as activating backup flight paths and emergency landing, to ensure the safety of the UAV and personnel.

[0046] Through the above technical solutions, the route planning module can ensure that the UAV can efficiently and safely complete its predetermined tasks when conducting patrol missions on highways.

[0047] As an optional embodiment of the present invention, the road surface defect recognition module is used to analyze road surface images captured by a high-definition camera in real time and accurately identify abnormalities such as cracks, potholes, and bumps. Specifically, it includes:

[0048] Image preprocessing: The acquired raw images are preprocessed, including noise removal, contrast enhancement, and brightness adjustment, to improve image clarity and detail. Images may be cropped or scaled as needed to better suit subsequent processing algorithms.

[0049] Feature extraction: Image processing techniques are used to extract features related to road surface defects from the preprocessed image. These features include color, texture, shape, edges, etc., which can reflect the unique properties of road surface defects.

[0050] Dataset preparation: Prepare a labeled dataset containing sample images of various road surface defects. The dataset should cover road surface defects of different types, sizes, and severity to ensure the model's generalization ability.

[0051] Support Vector Machine (SVM) model training: An SVM model is trained using a labeled dataset to accurately identify different types of road surface defects. During training, the model's performance is optimized and recognition accuracy is improved by adjusting SVM parameters (such as kernel function, penalty factor, etc.).

[0052] Real-time identification: The pre-processed real-time images are input into the trained SVM model for real-time identification of road surface defects. The model analyzes the features in the image and compares them with previously learned defect features to determine whether there are abnormalities such as cracks, dents, or bumps.

[0053] Model Updates and Optimization: As new road surface image data is continuously generated, the SVM model is regularly updated and optimized to improve its recognition accuracy and generalization ability. The model can be fine-tuned using new labeled image datasets to adapt to the needs of road defect recognition in different road sections and environments.

[0054] As an optional embodiment of the present invention, the collapse anomaly analysis module comprehensively analyzes the risk of road collapse by combining the identification results from radar detectors, road surface defect identification modules, and other data sources (such as weather data). This allows for the timely identification of road sections with a high probability of collapse. Specifically, this includes:

[0055] Data source integration: Integrate image data analysis results (from the road surface defect identification module), weather data, and radar monitoring results.

[0056] Weighting: Each data source is assigned a weight (w1, w2, w3) based on its importance in predicting road collapses. These weights reflect the relative contribution of different data sources to the prediction.

[0057] Data fusion:

[0058] A weighted fusion method is used to combine information from different data sources.

[0059] Data fusion weight calculation:

[0060] The overall collapse risk level W = w1×T1 + w2×T2 + w3×T3;

[0061] Where T1 is the quantized value of the image data analysis result; T2 is the quantized value of the weather data; T3 is the quantized value of the radar monitoring result; w1, w2, and w3 are the weights of the image data, weather conditions, and radar data, respectively, and w1+w2+w3=1.

[0062] Data partitioning: The fused data is divided into a test set and a validation set for subsequent model training and validation.

[0063] Road Collapse Probability Prediction: Using a logistic regression model to predict the probability P of a road collapse. 塌方 .

[0064] P 塌方 =1 / [1+e -(β0+β1×W+β2×A) ];

[0065] Where β0, β1, and β2 are the parameters of the model; A represents other relevant factors.

[0066] β0 is the constant term in the logistic regression equation, also known as the intercept or bias term. When all eigenvalues ​​(i.e., W and other relevant factors) are equal to 0, β0 determines the baseline value or probability of the predicted outcome. In practical applications, since eigenvalues ​​are rarely simultaneously zero, the explanatory power of β0 is relatively weak, but it remains an important parameter in the model.

[0067] β1 (Weighting Coefficient 1): β1 is the weighting coefficient of the overall collapse risk W. It represents the degree of influence of the overall collapse risk W on the probability of highway collapse. Specifically, when W increases by one unit, the logarithmic odds (log-odds) of the collapse probability will increase by β1 units. The sign of β1 determines the positive or negative correlation between W and the collapse probability. If β1 is positive, the larger W is, the higher the probability of collapse; if β1 is negative, the larger W is, the lower the probability of collapse (but in reality, since W is the overall collapse risk, we usually expect its value to be larger, as the risk of collapse is higher, so β1 is usually positive).

[0068] β2 (Weighting Coefficient 2): β2 is the weighting coefficient for "Other Relevant Factors." "Other relevant factors" include variables other than W that affect road collapses, such as rainfall, geological conditions, and traffic flow. The value of β2 also indicates the degree of influence of these factors on the probability of road collapses. The specific interpretation is similar to β1.

[0069] The values ​​of these parameters (β0, β1, β2) are estimated by maximizing the log-likelihood function of the logistic regression model, typically using optimization algorithms such as gradient descent or Newton's method. During model training, the values ​​of these parameters are continuously adjusted to minimize prediction error and optimize the model's predictive performance.

[0070] Build a predictive model: Based on historical data and expert knowledge, build a model to predict the probability and timing of road collapses. Use machine learning algorithms (such as logistic regression, decision trees, random forests, etc.) or deep learning models (such as neural networks).

[0071] Model training and validation: Train the model using a historical test set. Use the validation set to evaluate the model's performance and adjust parameters to optimize prediction results.

[0072] Predictive analytics: Inputting the current data into the trained model.

[0073] The model outputs predictions, including the probability of a road collapse.

[0074] Results Presentation and Decision Support: Visualize the forecast results for decision-makers. Provide decision support, such as suggesting measures to prevent or mitigate the impact of road collapses.

[0075] As an optional embodiment of the present invention, the warning light panel assembly includes a drone, a motor, a warning panel A, a warning panel B, and a warning panel C;

[0076] The motor is fixedly mounted on the drone;

[0077] Warning plate A is rotatably mounted on the lower part of the drone; warning plate B is slidably mounted on warning plate A; warning plate C is slidably mounted on warning plate B.

[0078] A spring is fixedly connected between warning board A and warning board B; a spring is fixedly connected between warning board B and warning board C.

[0079] Several magnetic blocks are fixedly installed at the end of warning plate B closest to warning plate A; several electromagnets are fixedly installed at the end of warning plate A furthest from warning plate B. Each electromagnet corresponds to one of the magnetic blocks.

[0080] Several magnetic blocks are fixedly installed at the end of warning plate C near warning plate B; several electromagnets are fixedly installed at the end of warning plate B4 away from warning plate C.

[0081] The drone is equipped with a fixed speaker that can emit audible warnings.

[0082] Several red LEDs are fixedly installed on both sides of warning panels A, B, and C. Warning messages can be displayed using these red LEDs. The red LEDs on both sides of warning panels A, B, and C are connected in parallel; red LEDs on the same side can also work individually to display warning messages.

[0083] In this technical solution, the drone hovers in front of a road section with potential for landslides (leaving sufficient safe braking distance), starts the motor to rotate warning plate A to a vertical position, and warning plate A drives warning plates B and C to rotate. At the same time, the electromagnet is de-energized, and under the action of the spring, warning plates B and C extend outward. The red LED lights on warning plates A, B, and C display warning words to remind drivers on the road to stop in time and pay attention to safety.

[0084] This invention provides a method for monitoring and early warning of highway landslides, comprising the following steps:

[0085] S1. Before the drone takes off, the flight route and cruise parameters are set through the route planning module.

[0086] S2. The drone cruises along a set route while using weather sensors to monitor weather conditions in real time; it continuously photographs the road surface with a high-definition camera and identifies defects such as cracks and holes in the road surface using a radar detector.

[0087] S3, the road surface defect recognition module performs real-time analysis of the collected images to promptly identify abnormalities such as cracks, potholes, and bumps on the road surface.

[0088] S4. Radar detectors scan the road surface to promptly identify potential landslide risks such as cracks and holes.

[0089] S5, the collapse anomaly analysis module, combines the identification results of the radar detector and the road surface defect identification module to conduct a comprehensive analysis of the collapse risk of the highway road surface.

[0090] S6. When a landslide risk is detected, the early warning module triggers the alarm and warning light panel components, and sends the early warning information to the highway traffic control center and relevant departments via the communication module. Based on the received early warning information, the highway traffic control center promptly takes traffic diversion and emergency repair measures to ensure safe and smooth road traffic.

[0091] S7. The drone hovers in front of a road section with potential landslide hazards, starts the motor to rotate warning plate A to a vertical position, and warning plate A drives warning plates B and C to rotate. At the same time, the electromagnet is de-energized, and under the action of the spring, warning plates B and C extend outward. The red LED lights on warning plates A, B and C display warning words to remind drivers on the road to stop in time and pay attention to safety.

[0092] 3. Beneficial effects

[0093] One or more technical solutions provided in this application have at least the following technical effects or advantages:

[0094] 1. This invention enables comprehensive monitoring of highway surfaces through drone patrol monitoring, eliminating the need for numerous monitoring devices on the road, thus reducing monitoring costs and improving work efficiency.

[0095] 2. The road surface defect identification module performs real-time analysis of the acquired images to promptly identify abnormalities such as cracks, potholes, and bumps on the road surface. A radar detector scans the highway surface to promptly identify potential landslide risks such as cracks and holes. Then, by combining the identification results from the radar detector and the road surface defect identification module, a comprehensive analysis of the road surface landslide risk is conducted, integrating road conditions with radar monitoring results to identify road sections with a high probability of landslides.

[0096] 3. When a landslide risk is detected, the early warning module triggers the alarm and warning light panel components, and sends the early warning information to the highway traffic control center and relevant departments via the communication module. Based on the received early warning information, the highway traffic control center promptly takes traffic control and emergency repair measures to ensure safe and smooth road traffic.

[0097] 4. A drone can be hovered in front of a road section with potential landslide hazards. The motor is started to rotate warning sign A to a vertical position. The red LED lights on warning signs A, B and C display warning words to remind drivers on the highway to stop in time and pay attention to safety. Even vehicles far away can see the warning signs hovering in the air, allowing drivers to take safety braking measures in advance. Attached Figure Description

[0098] Figure 1 This is an overall schematic diagram of a highway landslide monitoring and early warning system disclosed in a preferred embodiment of this application;

[0099] Figure 2 This is a flowchart illustrating a preferred embodiment of the highway landslide monitoring and early warning method disclosed in this application.

[0100] Figure 3 This is a schematic diagram of a warning light panel assembly of a highway landslide monitoring and early warning system disclosed in a preferred embodiment of this application;

[0101] Figure 4 This is a schematic diagram showing the coordination of warning boards, warning board B, and warning board C in a preferred embodiment of the highway landslide monitoring and early warning system disclosed in this application.

[0102] Figure 5 for Figure 4 Enlarged view of section A.

[0103] Figure label:

[0104] 1. Drone; 2. Motor; 3. Warning board A; 4. Warning board B; 5. Warning board C; 6. Spring; 7. Magnetic block. Detailed Implementation

[0105] The present application will be further described in detail below with reference to the accompanying drawings.

[0106] Reference Figure 1 and Figure 2 This application provides a highway landslide monitoring and early warning system, including: a drone, a weather sensor, a radar detector, a high-definition camera, a route planning module, a road surface defect identification module, a PLC control module, a landslide anomaly analysis module, and an early warning module;

[0107] The drone is equipped with a weather sensor, radar detector, high-definition camera, and LED lights.

[0108] The drone is designed for long-duration, stable, and reliable operation. It is equipped with key components such as a flight control system, navigation and positioning module, communication module, and power module; these are existing technologies and will not be elaborated upon here. Waterproof drones are preferred, allowing for monitoring and patrolling in rainy weather.

[0109] The drone is equipped with a fixed radar detector, a high-definition camera, and a weather sensor.

[0110] High-definition images of the road surface are captured by a high-definition camera; night vision and infrared shooting are supported to ensure clear images can be obtained under various lighting conditions.

[0111] Radar detectors are used to identify defects such as cracks and holes in road surfaces. These detectors are characterized by high resolution and high penetration, and are used to detect potential landslide risks such as cracks and holes in road surfaces.

[0112] Weather sensors are used to monitor weather conditions in the flight area in real time, including parameters such as temperature, humidity, wind speed, wind direction, and rainfall. Weather sensors are essential for timely understanding of local weather conditions, with particular emphasis on monitoring rainy days; they also provide real-time weather information, especially rainfall amount and predicted rainfall trends.

[0113] Route planning module: Based on highway route maps and topographic maps, the software algorithm plans the optimal flight route for the drone, including the starting point, destination, and alternate charging point, to ensure that the drone can complete the patrol mission efficiently and safely.

[0114] Road surface defect recognition module: Based on the collected images, and using image processing and machine learning technologies, the module performs real-time analysis on the images captured by the high-definition camera to identify abnormalities such as cracks, potholes, and bumps on the road surface.

[0115] The collapse anomaly analysis module combines the identification results from radar detectors and the road surface defect recognition module to comprehensively analyze the collapse risk of highway pavements and promptly identify road sections with a potential collapse possibility. It integrates road surface conditions with radar monitoring results for comprehensive analysis. If road surface anomalies coincide with anomalies in underground structures detected by radar, especially in areas showing instability across multiple indicators, a collapse hazard should be highly suspected. Special attention should be paid to road sections located near mountains or slopes, as these areas are more susceptible to geological movements.

[0116] The early warning module includes an alarm and warning light panel assembly. When a landslide risk is detected, the alarm sounds an audible alert, and the warning light panel displays a warning message. Simultaneously, the warning information is transmitted in real-time to the highway traffic control center via a communication module, and passing vehicles and relevant departments are notified through broadcasts and SMS messages. The alarm sounds an audible alert, and the warning light panel displays "Landslide risk ahead, please stop and drive carefully" in red. The potential landslide information is also transmitted wirelessly to the highway traffic control center, and broadcasts a warning that there is a landslide hazard on this section of road, advising vehicles to detour. At the same time, the highway bureau will arrange traffic control and emergency repairs for the potentially landslide-prone section as soon as possible.

[0117] PLC control module: Network-connected with collapse anomaly analysis module, early warning module, drone, weather sensor, radar detector, high-definition camera and route planning module.

[0118] In this technical solution, before takeoff, the drone sets its flight route and cruise parameters via a route planning module. The drone cruises along the set route while simultaneously monitoring weather conditions in real time using weather sensors. A high-definition camera continuously captures images of the road surface, transmitting the collected images to a road defect identification module for processing. This module analyzes the images in real time, identifying anomalies such as cracks, potholes, and bumps on the road surface. A radar detector scans the road surface, identifying potential landslide risks such as cracks and holes. A landslide anomaly analysis module combines the results from the radar detector and the road defect identification module to comprehensively analyze the landslide risk of the road surface. When a landslide risk is detected, the early warning module triggers an alarm and warning light panel assembly, and sends the warning information to the highway traffic control center and relevant departments via a communication module. Based on the received warning information, the highway traffic control center promptly takes traffic control and repair measures to ensure safe and smooth road traffic.

[0119] Furthermore, radar detectors identify defects such as cracks and holes in highway pavements; specifically, this includes signal processing, image analysis, and defect identification.

[0120] Signal processing: After receiving reflected signals from the road surface, radar detectors often contain a significant amount of noise and interference. This noise may originate from the environment (such as electromagnetic interference, weather conditions, etc.) or the radar system itself (such as hardware noise, system errors, etc.). Therefore, signal processing is the primary step for radar detectors to identify defects.

[0121] The primary goal of signal processing is to improve the signal-to-noise ratio (SNR) and image quality. This typically involves the following steps:

[0122] Filtering: By designing appropriate filters, such as low-pass filters, high-pass filters, and band-pass filters, high-frequency noise or interference at specific frequencies can be removed from a signal.

[0123] Enhancement: Increasing the amplitude or energy of a useful signal to improve its visibility against a noisy background. This can be achieved through various enhancement algorithms, such as contrast enhancement and brightness enhancement.

[0124] Signal reconstruction: After filtering and enhancement, the signal may need to be reconstructed or resampled to better match the needs of subsequent image analysis.

[0125] Image analysis: After signal processing, radar images reveal different layers and characteristics of the road surface structure. Cracks and holes typically appear as specific reflection patterns or anomalous areas on the image. These anomalous areas may differ from the surrounding road surface structure in terms of reflection intensity, reflection pattern, or geometry.

[0126] To extract these anomalous regions, various image analysis techniques can be used, including:

[0127] Thresholding segmentation: By setting one or more thresholds, an image is divided into different regions or categories. In radar images, features such as reflection intensity or reflection pattern can be used as the basis for thresholding segmentation.

[0128] Edge detection: This technique uses edge information in an image to identify defects such as cracks and holes. Edge detection algorithms can detect locations where there are abrupt changes in reflection intensity or reflection pattern in an image, thereby determining the boundaries of cracks and holes.

[0129] Morphological processing: This involves using morphological operations (such as erosion, dilation, opening, and closing operations) to enhance or remove specific structures or features in an image. These operations can help highlight the shape and size characteristics of defects such as cracks and holes.

[0130] Defect identification: After extracting the abnormal areas, further defect identification is required. This includes several steps:

[0131] Feature extraction: Representative and discriminative features, such as reflection intensity, reflection pattern, shape, and size, are extracted from the abnormal regions. These features will be used for subsequent defect classification and identification.

[0132] Classifier Design: Based on the extracted features, design a suitable classifier to distinguish different types of defects (such as cracks, holes, etc.). Classifiers can be statistical (such as support vector machines, Naive Bayes, etc.), rule-based (such as decision trees, random forests, etc.), or learning-based (such as neural networks, deep learning, etc.).

[0133] Model validation: The classifier is trained and validated using sample data with known labels to evaluate its performance and accuracy. This typically includes methods such as cross-validation and hold-out validation.

[0134] Decision Support: Once the classifier is designed and validated, it can be applied to real radar images to automatically identify and quantify defects such as cracks and holes. These results can provide decision support for road maintenance and repair, helping to determine areas requiring repair and appropriate repair methods.

[0135] Furthermore, when planning the optimal route for the UAV to cruise on the highway, the route planning module will integrate the highway route map, topographic map and real-time environmental data, and through a series of complex software algorithms, ensure that the UAV can complete the predetermined task efficiently and safely. Specifically, it includes the following steps: data collection and preprocessing, starting point and ending point setting, path optimization and verification, task allocation and monitoring, alternate landing and charging point planning and path planning.

[0136] Data collection and preprocessing: Collect data such as highway route maps, topographic maps, real-time environmental data, and UAV performance parameters;

[0137] Highway route map: Get the latest highway network layout map, including information such as road directions, entrance and exit locations, and service area locations.

[0138] Topographic maps: Acquire high-precision topographic data, including elevation, slope, and landform type (such as mountains, plains, rivers, etc.).

[0139] Real-time environmental data: Real-time meteorological information such as wind speed, wind direction, temperature, and humidity are obtained through meteorological services and environmental monitoring stations, as well as real-time road condition information such as possible traffic congestion and construction sections.

[0140] Drone performance parameters: These parameters include the drone's flight speed, maximum flight altitude, maximum range, endurance, and payload capacity.

[0141] Start and End Point Settings: Based on the requirements of the cruise mission, set the takeoff point (start) and landing point (end) for the drone. Ensure these points are located in suitable positions along the highway to facilitate drone takeoff and landing.

[0142] Alternate charging point planning: Based on the drone's endurance and the length of the cruise mission, select suitable service areas or safe areas along the highway as alternate charging points. Consider the distance between alternate charging points to ensure that the drone can safely reach the next alternate charging point before its battery runs out.

[0143] Path planning: Using advanced path planning algorithms (such as Dijkstra's algorithm, A* algorithm, genetic algorithm, etc.), combined with highway route maps, topographic maps, and real-time environmental data, the optimal flight path from the starting point to the destination is calculated, taking into account alternate landing and charging points. During path planning, multiple factors must be comprehensively considered, such as flight distance, flight time, flight safety (avoiding no-fly zones, terrain obstacles, etc.), and flight efficiency.

[0144] Path optimization and validation: The initially planned path is further optimized, such as by adjusting parameters like flight altitude and speed, to improve flight efficiency and safety. The feasibility and effectiveness of the planned path are validated through simulated flight or actual test flights, and necessary adjustments are made based on actual conditions.

[0145] Task allocation and monitoring: The planned flight path and mission instructions are sent to the UAV to ensure it executes the patrol mission according to the planned path. During the mission, the UAV's position, status, battery level, and other information are monitored in real time through a monitoring system to ensure successful mission completion. Considering potential emergencies (such as severe weather, equipment failure, etc.), corresponding emergency response mechanisms are established, such as activating backup flight paths and emergency landing, to ensure the safety of the UAV and personnel.

[0146] By following the steps and considering the factors mentioned above, the route planning module can ensure that the drone can efficiently and safely complete its mission when conducting patrol missions on highways.

[0147] Furthermore, the road surface defect recognition module is used to analyze road surface images captured by high-definition cameras in real time and accurately identify anomalies such as cracks, potholes, and bumps. Specifically, it includes:

[0148] Image preprocessing: The acquired raw images are preprocessed, including noise removal, contrast enhancement, and brightness adjustment, to improve image clarity and detail. Images may be cropped or scaled as needed to better suit subsequent processing algorithms.

[0149] Feature extraction: Image processing techniques are used to extract features related to road surface defects from the preprocessed image. These features include color, texture, shape, edges, etc., which can reflect the unique properties of road surface defects.

[0150] Dataset preparation: Prepare a labeled dataset containing sample images of various road surface defects. The dataset should cover road surface defects of different types, sizes, and severity to ensure the model's generalization ability.

[0151] Support Vector Machine (SVM) model training: An SVM model is trained using a labeled dataset to accurately identify different types of road surface defects. During training, the model's performance is optimized and recognition accuracy is improved by adjusting SVM parameters (such as kernel function, penalty factor, etc.).

[0152] Real-time identification: The pre-processed real-time images are input into the trained SVM model for real-time identification of road surface defects. The model analyzes the features in the image and compares them with previously learned defect features to determine whether there are abnormalities such as cracks, dents, or bumps.

[0153] Model Updates and Optimization: As new road surface image data is continuously generated, the SVM model is regularly updated and optimized to improve its recognition accuracy and generalization ability. The model can be fine-tuned using new labeled image datasets to adapt to the needs of road defect recognition in different road sections and environments.

[0154] Furthermore, the collapse anomaly analysis module combines the identification results from radar detectors, pavement defect identification modules, and other data sources (such as weather data) to comprehensively analyze the risk of road collapses. This allows for the timely identification of road sections with a high probability of collapse. Specifically, this includes:

[0155] Data source integration: Integrate image data analysis results (from the road surface defect identification module), weather data, and radar monitoring results.

[0156] Weighting: Each data source is assigned a weight (w1, w2, w3) based on its importance in predicting road collapses. These weights reflect the relative contribution of different data sources to the prediction.

[0157] Data fusion:

[0158] A weighted fusion method is used to combine information from different data sources.

[0159] Data fusion weight calculation:

[0160] The overall collapse risk level W = w1×T1 + w2×T2 + w3×T3;

[0161] Where T1 is the quantized value of the image data analysis result; T2 is the quantized value of the weather data; T3 is the quantized value of the radar monitoring result; w1, w2, and w3 are the weights of the image data, weather conditions, and radar data, respectively, and w1+w2+w3=1.

[0162] Data partitioning: The fused data is divided into a test set and a validation set for subsequent model training and validation.

[0163] Road Collapse Probability Prediction: Using a logistic regression model to predict the probability P of a road collapse. 塌方 .

[0164] P 塌方 =1 / [1+e -(β0+β1×W+β2×A) ];

[0165] Where β0, β1, and β2 are the parameters of the model; A represents other relevant factors.

[0166] β0 is the constant term in the logistic regression equation, also known as the intercept or bias term. When all eigenvalues ​​(i.e., W and other relevant factors) are equal to 0, β0 determines the baseline value or probability of the predicted outcome. In practical applications, since eigenvalues ​​are rarely simultaneously zero, the explanatory power of β0 is relatively weak, but it remains an important parameter in the model.

[0167] β1 (Weighting Coefficient 1): β1 is the weighting coefficient of the overall collapse risk W. It represents the degree of influence of the overall collapse risk W on the probability of highway collapse. Specifically, when W increases by one unit, the logarithmic odds (log-odds) of the collapse probability will increase by β1 units. The sign of β1 determines the positive or negative correlation between W and the collapse probability. If β1 is positive, the larger W is, the higher the probability of collapse; if β1 is negative, the larger W is, the lower the probability of collapse (but in reality, since W is the overall collapse risk, we usually expect its value to be larger, as the risk of collapse is higher, so β1 is usually positive).

[0168] β2 (Weighting Coefficient 2): β2 is the weighting coefficient for "Other Relevant Factors." "Other relevant factors" include variables other than W that affect road collapses, such as rainfall, geological conditions, and traffic flow. The value of β2 also indicates the degree of influence of these factors on the probability of road collapses. The specific interpretation is similar to β1.

[0169] The values ​​of these parameters (β0, β1, β2) are estimated by maximizing the log-likelihood function of the logistic regression model, typically using optimization algorithms such as gradient descent or Newton's method. During model training, the values ​​of these parameters are continuously adjusted to minimize prediction error and optimize the model's predictive performance.

[0170] Build a predictive model: Based on historical data and expert knowledge, build a model to predict the probability and timing of road collapses. Use machine learning algorithms (such as logistic regression, decision trees, random forests, etc.) or deep learning models (such as neural networks).

[0171] Model training and validation: Train the model using a historical test set. Use the validation set to evaluate the model's performance and adjust parameters to optimize prediction results.

[0172] Predictive analytics: Inputting the current data into the trained model.

[0173] The model outputs predictions, including the probability of a road collapse.

[0174] Results Presentation and Decision Support: Visualize the forecast results for decision-makers. Provide decision support, such as suggesting measures to prevent or mitigate the impact of road collapses.

[0175] Reference Figure 3 , Figure 4 and Figure 5 The warning light panel assembly includes a drone 1, a motor 2, a warning panel A3, a warning panel B4, and a warning panel C5;

[0176] Motor 2 is fixedly mounted on UAV 1;

[0177] A warning plate A3 is rotatably mounted on the lower end of the drone 1;

[0178] Warning panel B4 is slidably mounted on warning panel A3;

[0179] Warning panel C5 is slidably mounted on warning panel B4;

[0180] A spring 6 is fixedly connected between warning plate A3 and warning plate B4; a spring 6 is fixedly connected between warning plate B4 and warning plate C5.

[0181] Several magnetic blocks 7 are fixedly installed at one end of warning plate B4 near warning plate A3; several electromagnets are fixedly installed at the other end of warning plate A3 away from warning plate B4. Each electromagnet corresponds to one of the magnetic blocks 7.

[0182] Several magnetic blocks 7 are fixedly installed at one end of warning plate C5 near warning plate B4; several electromagnets are fixedly installed at the other end of warning plate B4 away from warning plate C5.

[0183] The drone 1 is equipped with a speaker that can emit audible warnings.

[0184] Several red LEDs are fixedly installed on both sides of warning panels A3, B4, and C5. Warning messages can be displayed using these red LEDs. The red LEDs on both sides of warning panels A3, B4, and C5 are connected in parallel; red LEDs on the same side can also work individually to display warning messages.

[0185] In this technical solution, the drone 1 hovers in front of the road section with potential for landslides (leaving sufficient safe braking distance), and the starter motor 2 drives the warning plate A3 to rotate to a vertical position. The warning plate A3 then drives the warning plates B4 and C5 to rotate. At the same time, the electromagnet is de-energized, and under the action of the spring 6, the warning plates B4 and C5 extend outward. The red LED lights on the warning plates A3, B4, and C5 display warning messages to remind drivers on the road to stop in time and pay attention to safety.

[0186] This invention provides a method for monitoring and early warning of highway landslides, comprising the following steps:

[0187] S1. Before the drone takes off, the flight route and cruise parameters are set through the route planning module.

[0188] S2. The drone cruises along a set route while using weather sensors to monitor weather conditions in real time; it continuously photographs the road surface with a high-definition camera and identifies defects such as cracks and holes in the road surface using a radar detector.

[0189] S3, the road surface defect recognition module performs real-time analysis of the collected images to promptly identify abnormalities such as cracks, potholes, and bumps on the road surface.

[0190] S4. Radar detectors scan the road surface to promptly identify potential landslide risks such as cracks and holes.

[0191] S5, the collapse anomaly analysis module, combines the identification results of the radar detector and the road surface defect identification module to conduct a comprehensive analysis of the collapse risk of the highway road surface.

[0192] S6. When a landslide risk is detected, the early warning module triggers the alarm and warning light panel components, and sends the early warning information to the highway traffic control center and relevant departments via the communication module. Based on the received early warning information, the highway traffic control center promptly takes traffic diversion and emergency repair measures to ensure safe and smooth road traffic.

[0193] S7. Drone 1 hovers in front of a road section with potential landslide hazards (leaving sufficient safe braking distance), starts motor 2 to rotate warning plate A3 to a vertical position, and warning plate A3 drives warning plates B4 and C5 to rotate. At the same time, the electromagnet is de-energized, and under the action of spring 6, warning plates B4 and C5 extend outward. The red LED lights on warning plates A3, B4 and C5 display warning words to remind drivers on the road to stop in time and pay attention to safety.

[0194] The working principle of this invention's highway landslide monitoring and early warning system is as follows: Before takeoff, the drone sets its flight route and cruise parameters through a route planning module. The drone cruises along the set route while simultaneously monitoring weather conditions in real time using weather sensors; it continuously photographs the highway surface using a high-definition camera and identifies defects such as cracks and holes using a radar detector; the road surface defect identification module analyzes the collected images in real time, promptly identifying anomalies such as cracks, potholes, and protrusions on the road surface. The radar detector scans the highway surface, promptly identifying potential landslide risks such as cracks and holes. The landslide anomaly analysis module combines the identification results from the radar detector and the road surface defect identification module to comprehensively analyze the landslide risk of the highway surface. When a landslide risk is detected, the early warning module triggers the alarm and warning light panel components and sends the early warning information to the highway traffic control center and relevant departments through the communication module. Based on the received early warning information, the highway traffic control center promptly takes traffic diversion and repair measures to ensure road safety and smooth traffic. Drone 1 hovers in front of a road section with a potential landslide hazard, leaving sufficient safe braking distance. Motor 2 starts, causing warning plate A3 to rotate to a vertical position. Warning plate A3 then rotates warning plates B4 and C5. Simultaneously, the electromagnet is de-energized, causing warning plates B4 and C5 to extend outwards under the action of spring 6. Red LED lights on warning plates A3, B4, and C5 display warning messages, reminding drivers on the road to stop and pay attention to safety, ensuring the safety of vehicles and their occupants.

[0195] This invention utilizes unmanned aerial vehicle (UAV) patrol monitoring to provide comprehensive oversight of highway surfaces, reducing the need for extensive on-road monitoring equipment, thus lowering costs and increasing efficiency. A road surface defect identification module analyzes collected images in real time, promptly identifying anomalies such as cracks, potholes, and bumps. A radar detector scans the road surface, identifying potential landslide risks such as cracks and holes. By combining the results from both the radar detector and the road surface defect identification module, a comprehensive analysis of the road surface landslide risk is performed, integrating road conditions with radar monitoring results to identify sections with potential landslide risks. When a landslide risk is detected, the early warning module triggers alarms and warning lights, and transmits the warning information to the highway traffic control center and relevant departments via a communication module. Based on the received warning information, the highway traffic control center promptly implements traffic management and repair measures to ensure safe and smooth traffic flow. The drone 1 can be hovered in front of a road section with potential landslide risks. The motor 2 is started to rotate the warning board A3 to a vertical position. The red LED lights on the warning boards A3, B4 and C5 display warning words to remind drivers on the highway to stop in time and pay attention to safety.

[0196] The above-described embodiments are only used to illustrate the technical solutions of the present invention, and are not intended to limit it. Although the present invention has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that modifications can still be made to the technical solutions described in the foregoing embodiments, or equivalent substitutions can be made to some of the technical features. Such modifications or substitutions do not cause the essence of the corresponding technical solutions to deviate from the spirit and scope of the technical solutions of the embodiments of the present invention.

Claims

1. A method for monitoring and early warning of highway landslides, characterized in that, Includes the following steps: S1, the route planning module sets the drone's flight route and cruise parameters; S2. The drone cruises along a set route, with weather sensors monitoring weather conditions in real time; a high-definition camera captures images, and a radar detector identifies cracks and holes. S3, the road surface defect identification module can promptly identify abnormal conditions on the road surface; S4. Radar detectors scan the road surface to identify cracks and holes that could lead to potential landslides. S5, the collapse anomaly analysis module, conducts a comprehensive analysis of the collapse risk of highway pavement; S6. When a risk of landslide is detected, the early warning module triggers an alarm. S7. When the drone hovers in front of a section of road with a risk of landslide, the motor drives the warning board A to rotate to a vertical position, and the red LED lights on the warning light board assembly display warning words to remind the driver to stop in time.

2. The method for monitoring and early warning of highway landslides according to claim 1, characterized in that: Step S3 specifically includes the following: Image preprocessing: Preprocessing the acquired raw images; Feature extraction: Extracting features related to road surface defects, including color, texture, and shape features; Dataset preparation: Prepare a labeled dataset containing sample images of various road surface defects; Support Vector Machine (SVM) model training: Train the SVM model using a labeled dataset and adjust the SVM parameters to optimize the model's performance; Real-time identification: The pre-processed real-time image is input into the trained SVM model to identify road surface defects in real time; Model updates and optimizations: Regularly update and optimize the SVM model.

3. The method for monitoring and early warning of highway landslides according to claim 1, characterized in that: Step S1 includes the following: Data collection: Collect route maps, topographic maps, real-time environmental data, and UAV performance parameters; Start and End Point Settings: Set the takeoff and landing points for the drone; Alternate charging point planning: Select suitable service areas or safe areas as alternate charging points; Path planning: The optimal flight path is planned using Dijkstra's path planning algorithm; Path optimization and validation: Further optimize and validate the initially planned path; Task assignment and monitoring: Flight paths and task instructions are sent to the UAV; the UAV then monitors the mission.

4. The method for monitoring and early warning of highway landslides according to claim 2, characterized in that: Step S5 includes the following: Data source integration: Integrating image data analysis results, weather data, and radar monitoring results; Weighting: Assign weights based on the importance of each data source for predicting road collapses; Data fusion: Using weighted fusion methods to combine information from different data sources; Data partitioning: The merged data is divided into a test set and a validation set; Road Collapse Probability Prediction: Using a logistic regression model to predict the probability of a road collapse; Establish a predictive model: Develop a model to predict the probability and timing of highway landslides; Model training and validation: Use the test set to train the model; use the validation set to evaluate the model's performance. Predictive analytics: Inputting current data into a trained model; The model outputs predictions, including the probability of a road collapse; Results presentation and decision support: Present the prediction results in a visual way.

5. The method for monitoring and early warning of highway landslides according to claim 4, characterized in that: The formula for calculating the data fusion weight is as follows: Comprehensive collapse risk W = w1×T1 + w2×T2 + w3×T3; where T1 is the quantified value of the image data analysis result; T2 is the quantified value of the weather data; T3 is the quantified value of the radar monitoring result; w1, w2, and w3 are the weights of the image data, weather data, and radar data, respectively, and w1 + w2 + w3 = 1.

6. The method for monitoring and early warning of highway landslides according to claim 4, characterized in that: Predict the probability P of a road collapse using a logistic regression model. 塌方 The formula is as follows: P 塌方 =1 / [1+e -(β0+β1×W+β2×A) ]; where β0, β1, and β2 are the parameters of the model; A represents other relevant factors; β0 is the constant term in the logistic regression equation; β1 is the weighting coefficient of the comprehensive collapse risk W, representing the degree of influence of the comprehensive collapse risk W on the probability of highway collapse; β2 is the weighting coefficient of other relevant factors; other relevant factors include variables other than W that affect highway collapse.

7. The method for monitoring and early warning of highway landslides according to claim 6, characterized in that: The values ​​of parameters β0, β1, and β2 are estimated by maximizing the log-likelihood function of the logistic regression model, and the solution is obtained using the gradient descent optimization algorithm.

8. The method for monitoring and early warning of highway landslides according to claim 1, characterized in that: The warning light panel assembly includes a drone, a motor, warning panel A, warning panel B, and warning panel C; the motor is fixedly mounted on the drone; warning panel A is rotatably mounted on the lower end of the drone; warning panel B is slidably mounted on warning panel A; warning panel C is slidably mounted on warning panel B; a spring is fixedly connected between warning panel A and warning panel B; a spring is fixedly connected between warning panel B and warning panel C; several electromagnets are fixedly mounted on one end of warning panel B near warning panel A and the other end of warning panel A away from warning panel B.

9. The method for monitoring and early warning of highway landslides according to claim 8, characterized in that: Several red LED lights are fixedly installed on both sides of warning board A, warning board B and warning board C. The warning words are displayed by the red LED lights on warning board A, warning board B and warning board C. The several red LED lights on both sides of warning board A, warning board B and warning board C are connected in parallel.

10. A highway landslide monitoring and early warning system, comprising: The system comprises a drone, a weather sensor, a radar detector, a high-definition camera, a route planning module, a road surface defect identification module, a PLC control module, a subsidence anomaly analysis module, and an early warning module. Its features include: the drone equipped with a radar detector, a high-definition camera, and a weather sensor; the route planning module planning the optimal flight route for the drone; the road surface defect identification module performing real-time image analysis to identify abnormal conditions on the road surface; the subsidence anomaly analysis module comprehensively analyzing the risk of road surface collapse and identifying sections of road with a high probability of collapse; the early warning module including an alarm and warning light panel components, whereby the alarm sounds an alarm and the warning light panel components display warning information when a collapse risk is detected; and the PLC control module being network-connected to the subsidence anomaly analysis module, the early warning module, the drone, the weather sensor, the radar detector, the high-definition camera, and the route planning module.

Images