Geological disaster grading early warning method and device based on unmanned aerial vehicle and optical fiber monitoring

Abstract

The invention discloses a geological disaster grading early warning method and device based on an unmanned aerial vehicle and optical fiber monitoring. The method comprises the following steps: acquiring optical fiber signal data and image data acquired by an unmanned aerial vehicle; effective signals conforming to a disaster characteristic mode are reserved, and an optical fiber characteristic vector is obtained; feature indexes are extracted from the image data, and image feature vectors including road cracks, settlement and slope anomalies are formed; outputting a confidence probability through a deep learning model; weighting the confidence probability, the environmental factor index and the position index to obtain a second certainty degree; based on the optical fiber feature vector and the image feature vector, first weight and second weight are respectively weighted to calculate a comprehensive disaster index; and based on the comprehensive disaster index, determining a disaster grade, and generating an emergency response measure. The method realizes adaptive weighted fusion of monitoring data, has real-time performance, intelligence and expandability, and can effectively improve the accuracy and reliability of traffic linear engineering geological disaster monitoring and early warning.

Description

Technical Field

[0001] This invention belongs to the field of traffic engineering and geological disaster monitoring technology, specifically relating to a geological disaster classification and early warning method and device based on UAV and fiber optic monitoring. Background Technology

[0002] Currently, existing technologies for monitoring geological hazards in transportation engineering mainly focus on the following two categories: (1) Monitoring method based on distributed optical fiber sensing This type of technology, by burying optical fibers in roadbeds, slopes, or tunnel surrounding rock, acquires strain, vibration, or temperature signals in real time. It can cover long-distance lines and has the advantages of continuous monitoring and maintenance-free operation. However, this method has two prominent problems in practical applications: optical fiber signals are highly sensitive to non-disaster factors such as vehicle loads, pedestrian traffic, or animal trampling, which can easily generate interference signals; and the spatial resolution of optical fiber positioning is limited, making it difficult to provide sufficiently intuitive criteria when the disaster occurs close to the road surface.

[0003] (2) Monitoring methods based on UAV inspection Drones can be equipped with visible light, infrared, or laser sensors to quickly acquire surface images and identify cracks, landslides, and subsidence, providing a highly intuitive understanding. However, this method also has significant limitations: drone monitoring results are greatly affected by weather, visibility, and flight conditions; image recognition relies on algorithm confidence and is easily affected by factors such as lighting and background complexity; and inspections are intermittent and cannot provide continuous real-time monitoring.

[0004] In summary, solutions relying solely on fiber optics or drones have limitations: fiber optic monitoring is prone to oversensitivity, and drone inspections are susceptible to distortion. Existing multi-source fusion research often involves simple superposition, lacking a mechanism for dynamically adjusting weights based on environmental conditions and data reliability.

[0005] Therefore, in linear transportation engineering (highways, railways, urban rail transit), there is an urgent need for a method that can combine fiber optic and UAV monitoring results to achieve disaster identification and graded early warning, and has high monitoring accuracy, robustness and feasibility. Summary of the Invention

[0006] To address the issues that existing distributed fiber optic sensing monitoring is susceptible to interference from vehicle loads and pedestrian trampling, and that drone inspections are limited by weather and visibility, this application provides a method and device for graded early warning of geological disasters based on drone and fiber optic monitoring.

[0007] Firstly, this application provides a method for graded early warning of geological disasters based on UAV and fiber optic monitoring, including: Acquire fiber optic signal data; The features of the optical fiber signal data are extracted to obtain the initial optical fiber feature vector; the low-frequency energy ratio and first confidence level of the optical fiber signal data are calculated to remove noise signals and retain effective signals that conform to the disaster characteristic pattern to obtain the optical fiber feature vector. If the fiber optic signal is a non-noise signal, then perform drone inspection and photography to obtain image data; Based on image data, the confidence probability is output using the YOLO model. The second degree of confidence is obtained by weighting the confidence probability, environmental factor indicators, and location indicators. The comprehensive disaster index is calculated by weighting the fiber optic feature vector and the image feature vector with a first weight and a second weight, respectively; the first weight is determined by the first confidence level and the second confidence level of the effective signal. Based on the comprehensive disaster index, the disaster level is determined and emergency response measures are generated.

[0008] In one possible implementation, the fiber optic signal data includes fiber optic strain, vibration, and displacement signal data of the geological disaster area; the image data includes visible light images and thermal imaging data.

[0009] In one possible implementation, the formula for the first degree of certainty is:

[0010] in, For the first degree of certainty, Smoothing control function, For threshold; , Indicates the smoothing threshold. These are vectors representing the various dimensions of the fiber's feature vectors. a i The feature weights are for the smoothing threshold. Ω 0 is the normalization constant, and d represents the dimension of the fiber eigenvector; Methods for determining valid signals based on a first degree of confidence include:

[0011] in Indicates the signal type result. , This is a dynamic threshold.

[0012] In one possible implementation, the step of outputting confidence probabilities based on image data using a YOLO model includes: The YOLO model was trained and validated using a training set of three types of labeled disaster images to obtain the trained YOLO model; the three types of disaster images include road cracks, road surface collapses, and slope landslides. Image data is input into a pre-trained YOLO model to perform multi-scale detection on the image, and outputs the bounding box position, target class, and class probability of each candidate target; the maximum class probability among all candidate targets is taken as the confidence probability.

[0013] In one possible implementation, the formula for the second degree of certainty is:

[0014] Wherein, η1, η2, and η3 are the weighting coefficients for the second degree of certainty; The confidence probability output by the deep learning model; As an environmental factor indicator, ;in As a correction factor, ; For image clarity indicators, For visibility, Flight index; These are the weighting coefficients of the correction factor; For location indicators, ,in The positional difference between the actual detection by the drone and the fiber optic signal. To allow for matching space tolerance.

[0015] In one possible implementation, the formula for calculating the first weight is:

[0016] in, As the first weight, , The weight coefficient for the first weight. Environmental factors.

[0017] In one possible implementation, the formula for the comprehensive disaster index is:

[0018] in, As a comprehensive disaster index, , This represents the smoothing threshold.

[0019] Secondly, this application provides a geological disaster classification and early warning device based on UAV and fiber optic monitoring, including: Acquisition device, used to acquire fiber optic signal data; The first extraction module is used to remove noise signals by calculating the low-frequency energy ratio and first confidence level of the optical fiber signal data, retaining effective signals that conform to the disaster characteristic pattern, and obtaining the optical fiber feature vector. The second extraction module is used to perform drone inspection and photography to obtain image data if the fiber optic signal is a non-noise signal; based on the image data, it outputs the confidence probability through the YOLO model; and weights the confidence probability, environmental factor indicators, and location indicators to obtain the second degree of confidence. The fusion module is used to calculate the comprehensive disaster index by weighting the fiber optic feature vector and the image feature vector with a first weight and a second weight, respectively; the first weight is determined by the first confidence level and the second confidence level of the effective signal. The output module is used to determine the disaster level based on the comprehensive disaster index and generate emergency response measures.

[0020] Thirdly, this application provides an electronic device, including a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor executes the program to implement the geological disaster classification and early warning method based on UAV and fiber optic monitoring as described in the first aspect.

[0021] Fifthly, this application provides a non-transitory computer-readable storage medium having a computer program stored thereon, wherein the computer program, when executed by a processor, implements the geological disaster classification and early warning method based on UAV and fiber optic monitoring as described in the first aspect.

[0022] The beneficial effects of this application are as follows: 1. The method provided in this application extracts the time-domain, frequency-domain, and spatial characteristics of optical fiber signals, and combines them with environmental correction and spatial consistency constraints of UAV detection results to effectively eliminate non-disaster signals, thereby improving the reliability and applicability of monitoring results.

[0023] 2. The method provided in this application establishes a mathematical model to adaptively weight the confidence levels of optical fibers and UAVs, dynamically adjusting the contribution ratio of the two data sources under different environments and operating conditions. By constructing a fused disaster index, it achieves quantitative characterization of disaster risk and classifies disasters into levels I–IV based on set thresholds, overcoming the shortcomings of existing methods that simply superimpose multi-source information and lack dynamic adjustment.

[0024] 3. The method provided in this application establishes a tiered emergency response system corresponding to traffic management measures based on disaster level determination. When the disaster index is at different levels, the system can automatically trigger corresponding control measures, such as speed limit reminders, priority inspection by drones, partial or full road closures, and coordinate with ground rescue teams to achieve an end-to-end closed loop from monitoring and early warning to emergency response, thereby improving the safety and emergency response efficiency of traffic engineering and possessing high monitoring accuracy, robustness, and feasibility. Attached Figure Description

[0025] Figure 1 A flowchart illustrating the geological disaster classification and early warning method based on UAV and fiber optic monitoring provided in this application embodiment; Figure 2 These are schematic diagrams of three categories of images in the embodiments of this application; the left image shows a road crack, the middle image shows a road surface collapse, and the right image shows a slope landslide. Figure 3 This is a schematic diagram of the fiber optic and UAV deployment provided in Embodiment 1 of this application; Figure 4 This is a schematic diagram of the fiber optic cable and UAV deployment provided in Embodiment 2 of this application; in, Figure 3-4 In the middle, 100-road slope, 200-optical fiber, 201-slope foot optical fiber, 202-road surface optical fiber, 203-track pre-embedded optical fiber, 300-UAV device, 301-UAV, 302-UAV base station.

[0026] Figure 5 A schematic diagram of the geological disaster classification and early warning device based on UAV and fiber optic monitoring provided in this application embodiment; Figure 6 This is a schematic diagram of the structure of the electronic device provided in the embodiments of this application. Detailed Implementation

[0027] The technical solutions of the embodiments of this application will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of this application, and not all embodiments. Based on the embodiments of this application, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of this application.

[0028] In the description of this application, the terms "first" and "second" are used for descriptive purposes only and should not be construed as indicating or implying relative importance or implicitly specifying the number of indicated technical features. Thus, a feature defined as "first" or "second" may explicitly or implicitly include one or more of the stated features. In the description of this application, "multiple" means two or more, unless otherwise explicitly specified.

[0029] In the description of this application, it should also be noted that, unless otherwise expressly specified and limited, the terms "set up," "install," "connect," and "link" should be interpreted broadly. For example, they can refer to a fixed connection, a detachable connection, or an integral connection; they can refer to a mechanical connection or an electrical connection; they can refer to a direct connection or an indirect connection through an intermediate medium; and they can refer to the internal connection of two components. Those skilled in the art can understand the specific meaning of the above terms in this technology based on the specific circumstances.

[0030] In the description of this application, spatial relation terms such as "below," "under," "below," "below," "above," "over," etc., are used herein to describe the relationship between one element or feature shown in the figures and other elements or features. It should be understood that, in addition to the orientation shown in the figures, spatial relation terms also include different orientations of the device in use and operation. For example, if the device in the figures is flipped, an element or feature described as "below" or "under" or "below" of other elements or features will be oriented "above" other elements or features. Therefore, the exemplary terms "below" and "under" can include both upper and lower orientations. Furthermore, the device may also include other orientations (e.g., rotated 90 degrees or other orientations), and the spatial descriptive terms used herein are interpreted accordingly.

[0031] In the description of this application, the term "for example" is used to mean "used as an example, illustration, or description." Any embodiment described as "for example" in this application is not necessarily to be construed as being more preferred or advantageous than other embodiments. The following description is provided to enable any person skilled in the art to make and use this application. Details are set forth in the following description for purposes of explanation. It should be understood that those skilled in the art will recognize that this application can be made without using these specific details. In other instances, well-known structures and processes will not be described in detail to avoid unnecessarily obscuring the description of this application. Therefore, this application is not intended to be limited to the embodiments shown, but is consistent with the broadest scope of the principles and features disclosed in this application.

[0032] To address the issues that existing distributed fiber optic sensing monitoring is susceptible to interference from vehicle loads and pedestrian trampling, and that drone inspections are limited by weather and visibility, this application provides a method and device for graded early warning of geological disasters based on drone and fiber optic monitoring.

[0033] See Figure 1 The geological disaster classification and early warning method based on UAV and fiber optic monitoring provided in this application includes: S101. Acquire fiber optic signal data.

[0034] In one possible implementation, the fiber optic signal data includes fiber optic strain, vibration, and displacement signal data of the geological disaster area.

[0035] Furthermore, a distributed optical fiber sensor network is deployed along the highway in the landslide hazard zone of S101.

[0036] Specifically, a ring-shaped arrangement is adopted, for example, a long-distance distributed optical fiber is laid at the toe of the slope on the side of the highway facing the mountain and at the roadbed, forming a "monitoring ring" to monitor both slope deformation data and roadbed defects and anomalies. The optical cable is buried at a depth of 0.3-1m at the limp point and at a depth of 1-3m in the roadbed.

[0037] Furthermore, in section S101, a drone base station is deployed every 5 km along roads in areas prone to geological disasters (reduced to 3 km in areas with greater disaster risks), equipped with drones capable of autonomous flight, damage detection, and data processing and transmission. This allows the drones to operate in gale-force winds (level 8) and in rain or snow, carrying trained algorithms for damage identification and uploading data to a data terminal via a data transmission module.

[0038] S102. Extract the features of the optical fiber signal data to obtain the initial optical fiber feature vector; by calculating the low-frequency energy ratio and first confidence level of the initial optical fiber feature vector to remove noise signals and retain the effective signals that conform to the disaster characteristic pattern, the optical fiber feature vector is obtained.

[0039] In one possible implementation, the initial fiber eigenvector is Λ, Λ=[λ1,λ2,λ3,λ4,λ5,λ6,λ7,λ8], where λ1 is the peak strain, λ2 is the strain rate, λ3 is the duration, λ4 is the spatial extension, λ5 is the low-frequency energy, λ6 is the high-frequency energy, λ7 is the spatial correlation, and λ8 is the temperature correction.

[0040] λ1 represents the maximum strain amplitude detected by the distributed optical fiber under external stress within the monitoring time window. It reflects the extreme response of the soil or roadbed structure under tension, compression, or bending, and is used to characterize the severity of structural deformation. λ2 represents the change in optical fiber strain value per unit time, with units of microstrain per second (με / s). It is used to characterize the evolution rate of structural deformation to distinguish between slow creep and sudden deformation processes. λ3 represents the duration of abnormal strain or abnormal vibration signals, used to describe the stability of high-strain zones or abnormal vibration phenomena in the time dimension. In the embodiments of this application, λ3 specifically refers to the duration of continuous abnormal strain exceeding a preset strain threshold, used to distinguish between instantaneous disturbances and persistent structural deformations with disaster evolution characteristics. λ4 represents the continuous spatial distribution scale of abnormal physical quantities along the fiber length direction. It reflects the spatial expansion range of high-strain zones, abnormal vibration zones, or potential slip zones, with units of meters (m), and is used to distinguish between local disturbances and regional structural instability. λ5 represents the signal energy integral calculated within a defined low-frequency range after spectral analysis of the original time-domain signal. It primarily reflects the low-frequency response characteristics caused by slow deformation of the geological body, soil creep, or overall slippage. λ6 represents the signal energy integral calculated within a defined high-frequency range, used to characterize the high-frequency vibration characteristics caused by rapid events such as sudden failure, rock and soil cracking, rockfall, or structural fracture. λ7 is a statistical characteristic quantity used to quantify the similarity of monitoring signals at different spatial locations along the optical fiber in the time dimension. Its value reflects whether the abnormal signal has spatial synchronicity or propagation characteristics, thus distinguishing between local noise and systemic structural response. λ8 represents the temperature compensation quantity, used to eliminate the influence of environmental temperature changes on the optical fiber strain measurement results. It is obtained through independent temperature measurement or historical temperature-strain models and is used to correct the original strain signal during the data processing stage.

[0041] Furthermore, in S102, the method for extracting optical fiber signal data features to obtain an initial optical fiber feature vector includes: using an optical fiber signal wave processing instrument to extract the data features of the optical signal to obtain an initial optical fiber feature vector.

[0042] In one possible implementation, in S102, the low-to-high frequency energy ratio The formula is as follows:

[0043] in φ 0 is the stability constant. If If the signal is large, the signal tends to indicate a geological anomaly. If the interference is small, it is determined to be interference from vehicles or pedestrians.

[0044] It is understandable that when the energy of low and high frequencies is relatively large, the signal characteristics tend to be abnormal; when the energy is relatively small, the signal may be caused by vehicles, people or animals stepping on the ground.

[0045] In one possible implementation, in S102, the formula for the first degree of certainty is:

[0046] in, As the first degree of certainty, For smooth control function, For threshold; , Indicates the smoothing threshold. These are vectors representing the various dimensions of the fiber's feature vectors. Ω 0 is the normalization constant, and d represents the dimension of the fiber eigenvector; α i The feature weights are used for the smoothing threshold.

[0047] Methods for determining valid signals based on a first degree of confidence include:

[0048] in Indicates the signal type result. , This is a dynamic threshold.

[0049] further, , , The dynamic threshold is adaptively determined based on historical operating data, noise statistics, and current environmental conditions of the fiber optic monitoring system. It is suitable for classifying and identifying abnormal signals under different operating conditions. For example, in the embodiments of this application, Take 0.18, Take 0.4, Take 0.7.

[0050] Furthermore, the smoothing control function The expression is:

[0051] S103. If the fiber optic signal is a non-noise signal, then perform drone inspection and photography to obtain image data.

[0052] In one possible approach, the drone inspection behavior is controlled in a tiered manner: when Determined as noise ( If the current signal is determined to be environmental noise or a non-abnormal disturbance, the drone takeoff inspection will not be triggered; if D(Ψ) is determined to be pending review ( If the current signal is deemed to have a potential anomaly risk, the drone is triggered to take off and inspect and photograph the corresponding area; when D(Ψ) is determined to be a valid anomaly ( The system determines that the current signal is a highly reliable abnormal signal, triggering the drone to take off and conduct key inspections and photography of the corresponding area.

[0053] After the drone takes off, it uses its onboard imaging equipment to inspect and photograph the corresponding area, acquiring images for subsequent confidence probability calculations based on the YOLO model.

[0054] Specifically, the image data includes visible light images and thermal imaging data.

[0055] S104. Based on image data, output confidence probability using the YOLO model.

[0056] In one possible implementation, S104 includes: S104a. The YOLO model is trained and validated using a training set of three types of labeled disaster images to obtain the trained YOLO model; the three types of disaster images include road cracks, road surface collapses, and slope landslides.

[0057] It should be noted that road cracks usually appear as linear or network-like discontinuous areas extending along the road surface; road collapses appear as localized depressions, damage, or structural defects; and slope landslides appear as large-area displacement of the slope surface, exposed soil, or areas where vegetation has been destroyed. Figure 2 The diagrams show three categories of images.

[0058] S104b: Input the image data into the pre-trained YOLO model, perform multi-scale detection on the image, and output the bounding box position, target category, and category probability of each candidate target; take the maximum category probability among all candidate targets as the confidence probability.

[0059] In one possible implementation, in S104b, the confidence probability is expressed as:

[0060] Among them, the class probability among all candidate targets is taken as the confidence probability of the model output; This indicates the probability that a target exists within the final detection bounding box. This represents the conditional probability that the target belongs to the k-th type of disaster, and the disaster category includes road cracks, road surface collapse, and slope landslide.

[0061] It is understandable that the confidence level mentioned The maximum value among all candidate targets (including targets of three categories) is the maximum value of the category probability, which is the maximum value of the final score.

[0062] Specifically, the YOLO model is a general model, such as YOLOv8 to v12.

[0063] S105. The confidence probability, environmental factor indicators and location indicators are weighted to obtain the second confidence level.

[0064] In one possible implementation, in S105, the formula for the second degree of certainty is:

[0065] Wherein, η1, η2, and η3 are the weighting coefficients for the second degree of certainty; The confidence probability output by the deep learning model; As an environmental factor indicator, ;in As a correction factor, ; For image clarity indicators, For visibility, Flight index; The coefficient of the correction factor; For location indicators, ,in The positional difference between the actual detection by the drone and the fiber optic signal. To allow for matching space tolerance.

[0066] It should be noted that the weighting coefficients for the second degree of confidence are assigned in a graded manner based on the degree of influence of each factor on the accuracy of UAV detection. In the embodiments of this application, η1 ranges from 0.4 to 0.6, η2 ranges from 0.2 to 0.4, and η3 ranges from 0.1 to 0.3.

[0067] The image clarity index reflects the degree to which target edges, textures, and structural information are discernible in images acquired by the UAV, characterizing the impact of image quality on defect identification results. Values ​​range from 0.8-1.0 for clear, 0.4-0.7 for moderate, and 0.0-0.3 for blurry. The visibility index reflects the degree to which environmental conditions affect UAV image acquisition, including factors such as fog, rain, nighttime, or strong backlighting. Values ​​range from 0.0-0.2 for clear daytime, 0.3-0.6 for light rain / fog, and 0.7-1.0 for heavy fog / nighttime.

[0068] In one possible implementation, Calculations are made based on the drone's flight status parameters, including the magnitude of attitude changes, flight speed fluctuations, or track deviations. Smooth cruise: 0.1–0.3; slight jitter: 0.4–0.6; severe disturbance: 0.7–1.0.

[0069] The correction factors λ1, λ2, and λ3 reflect the relative influence of image quality, environmental visibility, and flight status on the reliability of UAV detection results, respectively. In this embodiment, λ1 is set to 1.0, λ2 to 0.8, and λ3 to 0.6.

[0070] The three-dimensional coordinates of the abnormal location point of the UAV. The three-dimensional coordinates of the abnormal location point of the optical fiber.

[0071] The difference between the actual location detected by the drone and the location generated by the fiber optic signal: the difference between the location detected by the drone and the location generated by the fiber optic signal. For example, the coordinates of the location of the road crack detected by the drone are... =(x0,y0,z0), the location where the optical fiber signal is generated is... =(x1,y1,z1), where the position difference is the absolute value of the difference between the two coordinates.

[0072] Allowable matching space tolerance: σ x This represents the permissible spatial deviation range between the UAV's detected location and the fiber optic anomaly location, used to accommodate positioning errors from both monitoring methods. Since UAVs rely on visual ranging for positioning, while distributed fiber optic sensing systems use optical time-domain reflectometry for spatial positioning, inherent differences exist in spatial resolution, coordinate reference, and time synchronization accuracy. Therefore, a spatial tolerance parameter σ is needed. x To avoid misclassifying the same physical anomaly as different events.

[0073] S106. Based on the fiber optic feature vector and the image feature vector, calculate the comprehensive disaster index by weighting the first weight and the second weight respectively; the first weight is determined by the first confidence level and the second confidence level of the effective signal.

[0074] In one possible implementation, the formula for calculating the first weight is:

[0075] in, As the first weight, , The weight coefficient for the first weight. Environmental factors.

[0076] It should be noted that ρ1 is determined through offline training or simulation analysis of historical samples, so that when the credibility of drones or optical fibers is dominant in known disaster samples, their corresponding weights can be significantly improved, with an engineering recommended range of [5,15]. ρ2 is used to adjust the influence of environmental factors on the weight calculation, with a recommended range of [1,8]. To characterize the overall impact of current environmental conditions on the reliability of UAV monitoring, each environmental factor The data is mapped to the interval [0,1] based on real-time sensor data or system status information, where 0 indicates no significant impact on the monitoring results and 1 indicates a significant adverse impact on the monitoring results.

[0077] In one possible implementation, the sum of the first weight and the second weight is 1.

[0078] In one possible implementation, the formula for the comprehensive disaster index is:

[0079] in, As a comprehensive disaster index, , This represents the smoothing threshold.

[0080] S107. Based on the comprehensive disaster index, determine the disaster level and generate emergency response measures.

[0081] In one possible implementation, the formula for determining the comprehensive disaster index is:

[0082] Level I is for emergency closure, and Level IV is for safety observation. Different levels correspond to different traffic management and emergency response measures. The values ​​are 0.3, 0.5, and 0.7, respectively, determined based on historical sample statistics, engineering experience, and the principle of safety redundancy, and support on-site calibration and dynamic adjustment. The physical meaning of the fusion index Ω is 0 → extremely low risk (normal operating state), 1 → extremely high risk (disaster has occurred or is highly imminent).

[0083] In one possible implementation, methods for generating emergency response measures include: (1) Level IV: Traffic management measures include displaying a message on navigation and variable information signs (VMS) that “slow down is required on the road ahead” and suggesting a speed limit of 80 km / h. Drivers are advised to take note. Emergency response measures include increasing the data sampling frequency, having staff review the data, and having drones perform low-priority inspections.

[0084] (2) Level III: Traffic management measures include a speed limit of 60 km / h, displaying warning information at road entrances, and providing warnings to drivers via navigation; emergency response measures include notifying on-duty personnel to enter the "yellow warning" state; the system automatically saves data and arranges ground patrol teams to go to the hazard points, and drones conduct high-priority inspections; (3) Level II: Traffic management measures include allowing only one lane to pass on highways with a speed limit of 30 km / h, and informing drivers to quickly exit highways and reduce train speeds to safe low speeds on railways; emergency response measures include joint confirmation by drones and ground teams, and the emergency command center receiving an “orange warning” signal, recommending that the transportation department temporarily control traffic flow and prepare for road closures.

[0085] (4) Level I: The system immediately triggers the road closure / track closure order, prohibits vehicles from entering the highway, automatically notifies the navigation system to detour, the railway immediately stops operation, and the train stops at the nearest station; the emergency response measures are to activate the "red alert", the emergency department immediately evacuates the crowd and dispatches rescue, the drone continuously conducts high-frequency inspections and transmits high-definition images, all drones of adjacent base stations are dispatched to cooperate with the ground team, and a geological disaster disposal expert group is dispatched for on-site assessment.

[0086] The geological disaster classification and early warning method based on UAV and fiber optic monitoring provided in this application first obtains fiber optic signals and image data collected by UAVs. Next, noise signals are filtered out from the initial fiber optic feature vector based on the low-to-high frequency energy ratio and a first confidence level to obtain effective signals that conform to disaster characteristic patterns. This step significantly improves the signal-to-noise ratio of the fiber optic signal, effectively reduces the frequency of noise processing, and enhances the effectiveness of the early warning. Then, feature indicators are extracted from the image data to form image features. A confidence probability is output based on a deep learning model. The confidence probability, environmental factor indicators, and location indicators are weighted to obtain a second confidence level. This step fully considers the influence of environmental and location factors, making the calculation of the second confidence level more accurate and closer to reality. Finally, the fusion weights are determined using the first and second confidence levels, and a comprehensive disaster index is further calculated using weighted fusion. Emergency response measures are then generated. This step fuses the two types of data to achieve multi-dimensional, dynamic, and comprehensive disaster level classification and provides targeted emergency response measures. In summary, the method provided in this application embodiment can realize an end-to-end closed loop from monitoring and early warning to emergency response, improve the safety and emergency response efficiency of traffic engineering, and has high monitoring accuracy, robustness and feasibility.

[0087] The following detailed embodiments illustrate this point.

[0088] Example 1 like Figure 3 As shown, a monitoring system was deployed on a secondary highway in a mountainous area. One distributed optical fiber was buried at the toe of the highway slope and another at the roadbed. The fiber at the toe of the slope was buried at a depth of 0.6 m, and the fiber at the roadbed was buried at a depth of 2.5 m, forming a "monitoring ring". A drone base station was set up every 5 km for take-off inspection.

[0089] Initial fiber optic feature values ​​were extracted from the fiber optic signal: Λ=[λ1=1,λ2=5,λ3=3,λ4=0.5,λ5=2,λ6=50,λ7=0.02,λ8=0]. Ω0=150, φ0=0.1, =0.18. The decision region is... 1 = 0.4 2 = 0.7. a =[0.25,0.1,0.1,0.05,0.05,0.3,0.1,0.05] (1) Solve for the low-frequency ratio:

[0090] What is sought It is very small. High frequencies dominate in the signal output, consistent with typical impulse / load interference characteristics.

[0091] (2) Calculation of the filter scoring function:

[0092] (3) Confidence mapping:

[0093] Calculations revealed that: Therefore, it was determined to be noise. The output signal was discarded. The event was short in duration, had a small spatial range, and a predominantly high-frequency spectrum, consistent with typical characteristics of vehicle wheel pressure or animal trampling. Therefore, this event was excluded to avoid false alarms in the future. The data can be subsequently compiled into a database, and machine learning can be used with a large model to filter out this noise signal.

[0094] Example 2 like Figure 4 As shown, a railway tunnel exit is located near the toe of a slope. The fiber optic cable at the toe is buried at a depth of 0.5 m, while the fiber optic cable in the roadbed is 2.5 m deep, providing a monitoring coverage of 2 km. A drone base station is located at the tunnel entrance, capable of taking off in rainy and windy conditions. The operating condition is based on several days of continuous rainfall. After the rain stops, a large-scale slope anomaly is detected, and the drone captures obvious cracks. The initial fiber optic characteristic values ​​are: Λ=[λ1=3000,λ2=200,λ3=300,λ4=100,λ5=800,λ6=50,λ7=0.9,λ8=0.3]. Ω0=150, φ0=0.1, =0.18. The decision region is... 1 = 0.4 2 = 0.7. a =[0.25,0.1,0.1,0.05,0.05,0.3,0.1,0.05]. According to the same method as in Example 1 of the above formula, we can obtain:

[0095]

[0096]

[0097] If the result is not found, it is considered a valid anomaly, the data is retained, and the calculation is performed using the formula.

[0098] The drone detection results are as follows:

[0099] Image / Environment / Flight Stability Metrics: , , .and .

[0100]

[0101]

[0102] Indicators of spatial location:

[0103] The final confidence level of the drone is: =0.3,

[0104]

[0105] but:

[0106]

[0107]

[0108] The comprehensive disaster index is:

[0109]

[0110] This determines the final disaster level to be Level I, and Level I emergency response measures are implemented. When the visual evidence from the drone is strong and the operational strategy prioritizes visual verification, the system can increase the drone's weight to expedite decision-making, thereby triggering a higher-level emergency response more quickly.

[0111] Example 3 The mountain highway slope has a total length of 5km. Fiber optic cables are laid at a depth of 0.6 m at the slope toe and 2.2 m in the roadbed, with a total length of 2km. Drone base stations are deployed every 5km. The operating condition is after a heavy rain; the fiber optic cables detect a continuous, large-scale anomaly, but the drones cannot operate due to low visibility caused by heavy fog, resulting in unclear images.

[0112] The initial fiber characteristics are: Λ=[λ1=2200,λ2=150,λ3=200,λ4=80,λ5=600,λ6=40,λ7=0.8,λ8=0.2]. Ω0=150, φ0=0.1, =0.18. The decision region is... 1 = 0.4 2 = 0.7. a =[0.25,0.1,0.1,0.05,0.05,0.3,0.1,0.05]. Low-to-high frequency ratio:

[0113]

[0114] Signal abnormal.

[0115]

[0116] Pick , , .

[0117]

[0118]

[0119]

[0120]

[0121]

[0122]

[0123] The comprehensive disaster index is:

[0124]

[0125] Therefore, the final disaster level is determined to be Level IV. In this example, the weather conditions were poor, and heavy fog reduced visibility. When the comprehensive disaster index is determined to be Level IV, Level IV emergency response measures should be implemented.

[0126] The geological disaster classification and early warning device based on UAV and fiber optic monitoring provided in this application is described below. The geological disaster classification and early warning device based on UAV and fiber optic monitoring described below can be referred to in correspondence with the geological disaster classification and early warning method based on UAV and fiber optic monitoring described above.

[0127] Figure 5 This is a schematic diagram of the geological disaster classification and early warning device based on UAV and fiber optic monitoring provided in this application embodiment, as shown below. Figure 5 As shown, it includes: an acquisition device 51, a first extraction module 52, a second extraction module 53, a fusion module 54, and an output module 55, wherein: Acquisition device 51 is used to acquire fiber optic signal data; The first extraction module 52 is used to remove noise signals by calculating the low-frequency energy ratio and first confidence of the optical fiber signal data, retain the effective signals that conform to the disaster characteristic pattern, and obtain the optical fiber feature vector. The second extraction module 53 is used to perform drone inspection and photography to obtain image data if the fiber optic signal is a non-noise signal; based on the image data, output the confidence probability through the YOLO model; and weight the confidence probability, environmental factor indicators and location indicators to obtain the second confidence level. The fusion module 54 is used to calculate the comprehensive disaster index by weighting the fiber optic feature vector and the image feature vector with a first weight and a second weight, respectively; the first weight is determined by the first confidence level and the second confidence level of the effective signal. Output module 55 is used to determine the disaster level and generate emergency response measures based on the comprehensive disaster index.

[0128] Figure 6 An example is a schematic diagram of the physical structure of an electronic device, such as... Figure 6 As shown, the electronic device may include a processor 610, a communications interface 620, a memory 630, and a communications bus 640. The processor 610, communications interface 620, and memory 630 communicate with each other via the communications bus 640. The processor 610 can call logical instructions from the memory 630 to execute a geological disaster classification and early warning method based on UAV and fiber optic monitoring.

[0129] Furthermore, the logical instructions in the aforementioned memory 630 can be implemented as software functional units and, when sold or used as independent products, can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of this application, in essence, or the part that contributes to the prior art, or a part of the technical solution, can be embodied in the form of a software product. This computer software product is stored in a storage medium and includes several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) to execute all or part of the steps of the methods described in the various embodiments of this application. The aforementioned storage medium includes various media capable of storing program code, such as USB flash drives, portable hard drives, read-only memory (ROM), random access memory (RAM), magnetic disks, or optical disks.

[0130] On the other hand, this application also provides a non-transitory computer-readable storage medium storing a computer program thereon, which, when executed by a processor, is implemented to perform the geological disaster classification and early warning method based on UAV and fiber optic monitoring provided by the above methods.

[0131] The device embodiments described above are merely illustrative. The units described as separate components may or may not be physically separate. The components shown as units may or may not be physical units; that is, they may be located in one place or distributed across multiple network units. Some or all of the modules can be selected to achieve the purpose of this embodiment according to actual needs. Those skilled in the art can understand and implement this without any creative effort.

[0132] Through the above description of the embodiments, those skilled in the art can clearly understand that each embodiment can be implemented by means of software plus necessary general-purpose hardware platforms, and of course, it can also be implemented by hardware. Based on this understanding, the above technical solutions, in essence or the part that contributes to the prior art, can be embodied in the form of a software product. This computer software product can be stored in a computer-readable storage medium, such as ROM / RAM, magnetic disk, optical disk, etc., and includes several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) to execute the methods described in the various embodiments or some parts of the embodiments.

[0133] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of this application, and are not intended to limit them. Although this application 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 this application.

Claims

1. A method for graded early warning of geological disasters based on UAV and fiber optic monitoring, characterized in that, include: Acquire fiber optic signal data; Extract the features of the optical fiber signal data to obtain the initial optical fiber feature vector; By calculating the low-frequency energy ratio and first confidence level of the fiber optic signal data to remove noise signals and retain effective signals that conform to the disaster characteristic pattern, the fiber optic feature vector is obtained. If the fiber optic signal is a non-noise signal, then perform drone inspection and photography to obtain image data; Based on image data, the confidence probability is output using the YOLO model. The second degree of confidence is obtained by weighting the confidence probability, environmental factor indicators, and location indicators. The comprehensive disaster index is calculated by weighting the fiber optic feature vector and the image feature vector with a first weight and a second weight, respectively; the first weight is determined by the first confidence level and the second confidence level of the effective signal. Based on the comprehensive disaster index, the disaster level is determined and emergency response measures are generated.

2. The geological disaster classification and early warning method based on UAV and fiber optic monitoring according to claim 1, characterized in that, The fiber optic signal data includes fiber optic strain, vibration, and displacement signal data of the geological disaster area; the image data includes visible light images and thermal imaging data.

3. The geological disaster classification and early warning method based on UAV and fiber optic monitoring according to claim 1, characterized in that, The formula for the first degree of confidence is: in, For the first degree of certainty, For smooth control function, For threshold; , Indicates the smoothing threshold. These are vectors representing the various dimensions of the fiber's feature vectors. a i The feature weights are for the smoothing threshold. Ω 0 is the normalization constant, and d represents the dimension of the fiber eigenvector; Methods for determining valid signals based on a first degree of confidence include: in Indicates the signal type result. , This is a dynamic threshold.

4. The geological disaster classification and early warning method based on UAV and fiber optic monitoring according to claim 1, characterized in that, The process of outputting confidence probabilities based on image data using a YOLO model includes: The YOLO model was trained and validated using a training set of three types of labeled disaster images to obtain the trained YOLO model; the three types of disaster images include road cracks, road surface collapses, and slope landslides. Image data is input into a pre-trained YOLO model to perform multi-scale detection on the image, and outputs the bounding box position, target class, and class probability of each candidate target; the maximum class probability among all candidate targets is taken as the confidence probability.

5. The geological disaster classification and early warning method based on UAV and fiber optic monitoring according to claim 1, characterized in that, The formula for the second level of certainty is: Wherein, η1, η2, and η3 are the weighting coefficients for the second degree of certainty; The confidence probability output by the YOLO model; As an environmental factor indicator, ;in As a correction factor, ; For image clarity indicators, For visibility, Flight index; These are the weighting coefficients of the correction factor; For location indicators, ,in The positional difference between the actual detection by the drone and the fiber optic signal. To allow for matching space tolerance.

6. The geological disaster classification and early warning method based on UAV and fiber optic monitoring according to claim 1, characterized in that, The formula for calculating the first weight is: in, As the first weight, , The weight coefficient for the first weight. Environmental factors.

7. The geological disaster classification and early warning method based on UAV and fiber optic monitoring according to claim 1, characterized in that, The formula for the comprehensive disaster index is: in, As a comprehensive disaster index, , This represents the smoothing threshold.

8. A geological disaster classification and early warning device based on UAV and fiber optic monitoring, characterized in that, include: Acquisition device, used to acquire fiber optic signal data; The first extraction module is used to remove noise signals by calculating the low-frequency energy ratio and first confidence of the optical fiber signal data, retaining effective signals that conform to the disaster characteristic pattern, and obtaining the optical fiber feature vector. The second extraction module is used to perform drone inspection and photography to obtain image data if the fiber optic signal is a non-noise signal; based on the image data, it outputs the confidence probability through the YOLO model; and weights the confidence probability, environmental factor indicators, and location indicators to obtain the second degree of confidence. The fusion module is used to calculate the comprehensive disaster index by weighting the fiber optic feature vector and the image feature vector with a first weight and a second weight, respectively; the first weight is determined by the first confidence level and the second confidence level of the effective signal. The output module is used to determine the disaster level based on the comprehensive disaster index and generate emergency response measures.

9. An electronic device comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, characterized in that, When the processor executes the program, it implements the geological disaster classification and early warning method based on UAV and fiber optic monitoring as described in any one of claims 1 to 7.

10. A non-transitory computer-readable storage medium having a computer program stored thereon, characterized in that, When the computer program is executed by the processor, it implements the geological disaster classification and early warning method based on UAV and fiber optic monitoring as described in any one of claims 1 to 7.

Images