Highway subgrade safety monitoring system and risk identification method

Abstract

The invention relates to the field of highway safety monitoring, and discloses a highway subgrade safety monitoring system and a risk identification method, and the system comprises a distributed optical fiber sensing subsystem and a video image capturing equipment subsystem which are respectively connected with a rear-end server; the distributed optical fiber sensing subsystem comprises distributed Rayleigh scattering optical fibers and / or distributed Brillouin scattering optical fibers which are arranged on a highway subgrade according to risk levels and is used for collecting dynamic information related to the highway subgrade, and the dynamic information represents environment and state changes of the highway subgrade; the video image capture device subsystem comprises a plurality of AI video image capture devices; and after receiving the dynamic information, the back-end server inputs the dynamic information into a preset calculation model to obtain a deformation index parameter output by the preset calculation model, determines a response level of the AI video image capture device according to the deformation index parameter, and configures an operation parameter of the AI video image capture device based on the response level. According to the system, real-time, accurate and comprehensive monitoring of a road is realized through cooperation of the distributed optical fiber, the AI video image capturing equipment and the back-end server.

Description

Technical Field

[0001] The present application relates to the field of highway safety monitoring, and more specifically, to a highway roadbed safety monitoring system and a risk identification method. Background Art

[0002] Since the mid-1980s, my country's highway construction has experienced a rapid development stage. By the end of 2022, the mileage of highways in operation has reached 177,000 kilometers, greatly promoting the rapid development of the country's economy and society. Especially in recent years, driven by the national development strategies of a strong transportation country and smart transportation, the integration and innovation of new generation information technologies such as artificial intelligence, big data, and cloud computing with the transportation industry have been promoted to achieve digital transformation and high-quality development of highway transportation.

[0003] Despite the remarkable achievements, some problems and challenges have also emerged with the rapid development of my country's highway network and digitalization. From the perspective of the highway traffic safety facility monitoring system, the specific problems mainly include:

[0004] (1) Due to the long length of the highway, the large scale of the project and the linear engineering characteristics, the geographical environment, geological environment, meteorological and hydrological conditions along the route are complex and changeable. In addition, the slope risk assessment system is still imperfect, the slope risk assessment cannot be accurately quantified, and the risk zoning is relatively rough. It is difficult to achieve all-round and accurate control under limited resource conditions;

[0005] (2) Under the condition of limited human, financial and material resources, the current highway embankment slopes are still mainly inspected and monitored manually. The point sensors used can only measure a limited number of discrete points, and some are greatly affected by the environment. It is impossible to achieve continuous detection in the entire space, resulting in a high probability of missed detection and misjudgment.

[0006] (3) Although the risk control of highway embankment slope safety has received increasing attention from operating management units and the industry in recent years, considering that most key slopes are monitored by point sensors, the construction cost of the slope monitoring system is relatively high, and the number of automated monitoring points is limited, slope safety cannot be fully controlled.

[0007] In 2024, a slope collapse accident occurred on the Chayang section of the Mei-Da Expressway in Guangdong, which triggered a deep reflection on the safety of highway operations in the whole society, and also issued a warning to further improve the safety management and risk prevention work of highways.

[0008] In order to meet the above challenges, further improve the construction of highway operation and roadbed safety prevention and control systems, effectively reduce the probability of highway accidents, timely discover potential risks, and improve the level of intelligent management and control of inspection and maintenance, it is urgent to solve the current problems of omissions, poor real-time performance and inaccuracy in highway roadbed safety monitoring. Summary of the invention

[0009] In view of this, the embodiments of the present application are dedicated to providing a highway roadbed safety monitoring system and a risk identification method to solve the problems of omissions, poor real-time performance and inaccuracy in current highway roadbed safety monitoring.

[0010] In a first aspect, this specification provides a highway roadbed safety monitoring system, comprising: a distributed optical fiber sensing subsystem and a video image capturing device subsystem respectively connected to a back-end server;

[0011] The distributed optical fiber sensing subsystem includes distributed Rayleigh scattering optical fibers and / or distributed Brillouin scattering optical fibers arranged on the roadbed according to the risk level, and is used to collect dynamic information related to the roadbed, and the dynamic information represents the environment and state changes of the roadbed;

[0012] The video image capture device subsystem includes a plurality of AI video image capture devices;

[0013] After receiving the dynamic information, the back-end server inputs the dynamic information into the preset calculation model, obtains the deformation index parameters output by the preset calculation model, determines the response level of the AI ​​video image capture device according to the deformation index parameters, and configures the operating parameters of the AI ​​video image capture device based on the response level.

[0014] According to the first aspect, in a possible implementation manner, a method of constructing a preset calculation model includes:

[0015] Preprocess the dynamic information to obtain statistical features;

[0016] The statistical features are fused and normalized to obtain a normalized composite feature vector;

[0017] Based on the clustering algorithm, the normalized composite feature vectors are analyzed to form clusters;

[0018] The clusters are evaluated to obtain deformation index parameters.

[0019] According to the first aspect, in a possible implementation, the distributed Rayleigh scattering optical fiber and / or the distributed Brillouin scattering optical fiber arranged on the roadbed according to the risk level includes:

[0020] For the first-level risk sections, distributed Rayleigh scattering optical fibers are laid on the surface of the highway guardrail structure;

[0021] For the secondary risk sections, distributed Brillouin scattering optical fibers are laid on the surface of the highway guardrail structure;

[0022] For the third-level risk sections, distributed Brillouin scattering optical fiber is laid on the surface of the highway guardrail structure, and distributed Rayleigh scattering optical fiber is laid on the shoulder of the roadbed side.

[0023] According to the first aspect, in one possible implementation, the AI ​​video image capture device includes an AI camera, a drone, and a lidar.

[0024] According to the first aspect, in a possible implementation manner, the dynamic information includes light intensity, temperature, and strain.

[0025] According to the first aspect, in a possible implementation, the clustering algorithm includes a density-based clustering algorithm.

[0026] According to the first aspect, in a possible implementation, the deformation index parameters include: segment length, segment maximum displacement, segment average strain rate, segment average displacement rate, segment maximum light intensity attenuation, and segment average light intensity attenuation rate.

[0027] In a second aspect, this specification provides a risk identification method using any of the above systems, including:

[0028] According to the characteristics of the dynamic information, the segment maximum displacement threshold, the segment average displacement rate threshold, the segment maximum light intensity attenuation threshold and the segment average light intensity attenuation rate threshold are determined respectively;

[0029] Compare the maximum displacement of the section, the average displacement rate of the section, the maximum light intensity attenuation of the section, and the average light intensity attenuation rate of the section with the corresponding thresholds to determine whether a large deformation has occurred in the monitored section;

[0030] According to the characteristics of the dynamic information, a segment length threshold is set for the segment length;

[0031] The normalized over-limit ratio of the maximum displacement of the segment, the normalized over-limit ratio of the average displacement rate of the segment, the normalized over-limit ratio of the maximum light intensity attenuation of the segment, the normalized over-limit ratio of the average light intensity attenuation rate of the segment and the normalized over-limit ratio of the segment length are calculated respectively. The normalized over-limit ratio represents the ratio of the difference between the actual value of the deformation index parameter and the threshold value of the deformation index parameter to the threshold value of the deformation index parameter.

[0032] According to the characteristics of dynamic information, the maximum displacement weight of the segment, the average displacement rate weight of the segment, the maximum light intensity attenuation weight of the segment, the average light intensity attenuation rate weight of the segment and the segment length weight are determined respectively;

[0033] The comprehensive deformation index is calculated based on the normalized over-limit ratio and weight of the maximum displacement of the segment, the average displacement rate of the segment, the maximum light intensity attenuation of the segment, the average light intensity attenuation rate of the segment and the segment length;

[0034] Based on the comprehensive deformation index, different levels of response based on AI camera functions are triggered.

[0035] According to the second aspect, in one possible implementation, the AI ​​video image capture device is an AI camera, and based on the comprehensive deformation index, different levels of responses based on the AI ​​camera function are triggered, including:

[0036] If D th <D k ≤D th +ΔD, the first-level response is triggered, the pan / tilt control function of the AI ​​image capture device is activated, the camera angle is immediately adjusted to review and monitor the section, and a preliminary abnormal cause analysis is performed, and an alarm is issued;

[0037] If D th +ΔD<D k , then the second level response is triggered, and based on the first level response, it immediately guides the manual on-site investigation and review work.

[0038] Among them, D k is the comprehensive deformation index, D th is the warning value of the comprehensive deformation index, and ΔD is the preset increment.

[0039] In a third aspect, this specification provides an electronic device comprising: at least one processor; and a memory communicatively connected to the at least one processor; wherein the memory stores instructions executable by the at least one processor, and the instructions are executed by the at least one processor to enable the at least one processor to perform each step in the above method.

[0040] In a fourth aspect, the present specification provides a computer-readable storage medium, the computer-readable storage medium including a stored program, wherein the program executes each step in the above method.

[0041] Compared with the prior art, the beneficial effects of the present invention are:

[0042] Different from the prior art, the highway roadbed safety monitoring system provided by the present invention includes: a distributed optical fiber sensing subsystem and a video image capture device subsystem respectively connected to the back-end server; the distributed optical fiber sensing subsystem includes distributed Rayleigh scattering optical fiber and / or distributed Brillouin scattering optical fiber arranged on the highway roadbed according to the risk level, which is used to collect dynamic information related to the highway roadbed, and the dynamic information characterizes the environment and state changes of the highway roadbed; the video image capture device subsystem includes multiple AI video image capture devices; after receiving the dynamic information, the back-end server inputs the dynamic information into the preset calculation model, obtains the deformation index parameters output by the preset calculation model, and determines the response level of the AI ​​video image capture device according to the deformation index parameters, and configures the operating parameters of the AI ​​video image capture device based on the response level. The present invention realizes real-time, accurate and comprehensive monitoring of the highway through the coordinated cooperation between the distributed optical fiber, the AI ​​video image capture device and the back-end server. BRIEF DESCRIPTION OF THE DRAWINGS

[0043] In order to more clearly illustrate the embodiments of the present application or the technical solutions in the prior art, the drawings required for use in the embodiments or the description of the prior art will be briefly introduced below. Obviously, the drawings described below are only embodiments of the present application. For ordinary technicians in this field, other drawings can be obtained based on the provided drawings without paying any creative work.

[0044] Figure 1 A schematic diagram of the structure of a highway roadbed safety monitoring system provided in an embodiment of the present application;

[0045] Figure 2 A flow chart of a risk identification method provided in an embodiment of the present application;

[0046] Figure 3 A schematic diagram of the structure of an electronic device provided in an embodiment of the present application. DETAILED DESCRIPTION

[0047] Unless otherwise defined, the technical terms or scientific terms used in the embodiments of this specification shall have the common meanings understood by persons with ordinary skills in the field to which this specification belongs. The words "first", "second" and similar words used in the embodiments of this specification do not indicate any order, quantity or importance, but are only used to avoid confusion of constituent elements.

[0048] Unless the context requires otherwise, throughout the specification, "plurality" means "at least two", and "including" is interpreted as an open, inclusive meaning, that is, "including, but not limited to". In the description of the specification, the terms "one embodiment", "some embodiments", "exemplary embodiments", "examples", "specific examples" or "some examples" are intended to indicate that a particular feature, structure, material or characteristic associated with the embodiment or example is included in at least one embodiment or example of the specification. The schematic representation of the above terms does not necessarily refer to the same embodiment or example.

[0049] The following will be combined with the drawings in the embodiments of this specification to clearly and completely describe the technical solutions in the embodiments of this specification. Obviously, the described embodiments are only part of the embodiments of this specification, not all of the embodiments. Based on the embodiments in this specification, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of this specification.

[0050] Traditional roadbed safety monitoring mainly relies on manual inspections and fixed-point instrument monitoring, which has problems such as low monitoring density, poor real-time performance, and low data utilization. It is difficult to meet the needs of roadbed safety monitoring in the increasingly complex highway operating environment.

[0051] In recent years, distributed fiber optic sensing technology has been widely used in the field of structural health monitoring due to its advantages of continuity, distribution, and high sensitivity. At the same time, the development of artificial intelligence technology has also provided a new way for the intelligent analysis of video surveillance data. However, there are relatively few studies on combining distributed fiber optic sensing with AI video image analysis for monitoring and identifying highway roadbed safety risks.

[0052] Therefore, the present invention proposes a highway subgrade safety monitoring system and risk identification method based on the coordination of distributed optical fiber sensing and AI video image analysis technology. The system establishes a comprehensive, all-weather, high-precision monitoring, early warning and event classification verification system for highway subgrade safety risks by deploying a distributed optical fiber sensing subsystem and a video image capture device subsystem for coordinated linkage control operation, effectively making up for the shortcomings of existing technologies and providing strong support for improving the level of highway operation safety management.

[0053] like Figure 1 FIG. 1 is a schematic diagram of the structure of a highway roadbed safety monitoring system provided by an embodiment of the present invention, wherein the system includes a distributed optical fiber sensing subsystem and a video image capturing device subsystem respectively connected to a back-end server;

[0054] The distributed optical fiber sensing subsystem includes distributed Rayleigh scattering optical fibers and / or distributed Brillouin scattering optical fibers arranged on the roadbed according to the risk level, and is used to collect dynamic information related to the roadbed, and the dynamic information represents the environment and state changes of the roadbed;

[0055] The video image capture device subsystem includes a plurality of AI video image capture devices;

[0056] After receiving the dynamic information, the back-end server inputs the dynamic information into the preset calculation model, obtains the deformation index parameters output by the preset calculation model, determines the response level of the AI ​​video image capture device according to the deformation index parameters, and configures the operating parameters of the AI ​​video image capture device based on the response level.

[0057] The system can obtain preliminary sensing data through a distributed fiber optic sensing subsystem. After receiving the data, the back-end server determines whether to trigger the AI ​​video image capture device to perform a response action. If the response level is reached, the operating parameters of the AI ​​video image capture device are configured for response processing.

[0058] It can be understood that the information feedback of the distributed fiber optic sensing subsystem can be triggered by a variety of events, such as landslides, vehicle collisions, man-made damage, etc. The system realizes all-round, all-weather, high-precision monitoring and early warning of highway roadbed safety risks and event classification verification through the coordinated cooperation between distributed optical fiber, AI video image capture equipment and back-end servers, thereby achieving real-time, accurate and comprehensive monitoring of highways.

[0059] The "roadbed" here refers to the basic structure of the highway, including embankments, slope protection, side slopes, etc. Whether the distributed optical fiber is buried in the roadbed, attached to the surface of the guardrail, or laid on the shoulder, the ultimate goal is the same, that is, to use distributed optical fiber to achieve continuous monitoring of the longitudinal deformation of the highway roadbed. Different layout methods only differ in the way of implementation, but do not affect the effectiveness of monitoring.

[0060] Dynamic information may include information characterizing the highway subgrade environment (such as temperature) and information on changes in highway subgrade conditions (such as light intensity, strain, etc.).

[0061] The distributed fiber optic sensing subsystem adopts two types of distributed fiber optic sensing technologies: distributed Rayleigh scattering fiber optic sensing and distributed Brillouin scattering fiber optic sensing.

[0062] (1) Distributed Rayleigh scattering fiber sensing

[0063] The optical fiber used is preferably a standard single-mode optical fiber (other types of optical fibers, such as special optical fibers, photonic crystal optical fibers, etc., may also be used), which is continuously buried along the roadbed. The monitoring principle is: using the Rayleigh scattering and Fresnel reflection effects in the optical fiber to achieve highly sensitive, continuous distributed monitoring of the intensity of the optical fiber optical signal generated by the environmental disturbance around the optical fiber, and the event space positioning accuracy can reach meter level. In the present invention, the distributed Rayleigh optical fiber optical loss measurement resolution is about 0.01dB, the measurement accuracy is about ±0.25dB, and the event space positioning measurement accuracy is about ±1m.

[0064] Standard single-mode optical fiber is the most widely used and mature optical fiber type in the field of optical fiber communication. It has a complete industrial chain, unified specifications, reliable quality and low cost. The use of standard single-mode optical fiber is conducive to the promotion and application of the present invention and the reliability of the solution.

[0065] The advantages of this distributed optical fiber are low cost, intrinsic safety, strong anti-electromagnetic interference ability, and suitability for large-scale coverage deployment. The disadvantage is that it can only measure changes in the light intensity of the sensing optical fiber and is used for semi-quantitative evaluation of the deformation state of the structure, and the data results are not very intuitive.

[0066] (2) Distributed Brillouin scattering fiber sensing

[0067] The optical fiber used is preferably a special single-mode 2-core optical fiber (1 core is used to measure temperature, and 1 core is used to measure strain), which is continuously buried along the roadbed. The monitoring principle is: using the sensitivity of the stimulated Brillouin scattering effect in the optical fiber to temperature and strain, the frequency shift between the scattered light and the incident light is linearly related to the temperature and strain. Through Brillouin Optical Time Domain Analysis (BOTDA), high-precision, long-distance distributed monitoring of the absolute temperature and strain along the optical fiber can be achieved, and the sampling accuracy can reach the centimeter level. In the present invention, the distributed Brillouin optical fiber temperature measurement accuracy is ±1°C, the strain measurement resolution is 1με, the strain measurement accuracy is ±15με, and the event space positioning measurement accuracy is about ±0.5m.

[0068] Sampling accuracy refers to the distance between the two closest sensing points that a distributed fiber optic sensing system can distinguish, which determines the spatial sampling density of the sensing system, that is, how many independent sensing data can be obtained per unit length of optical fiber. Event spatial positioning measurement accuracy refers to the accuracy of locating the location of an event (such as temperature change, strain change, etc.) when the fiber optic sensing system detects an event, which determines the system's ability to locate the event, that is, the specific location of the abnormality can be determined within a certain range.

[0069] The advantages of this distributed optical fiber are that it can measure absolute temperature and strain simultaneously, with high measurement accuracy and good long-term stability. The disadvantage is that the laying of the optical fiber and the cost of the instrument are relatively high.

[0070] By combining Rayleigh scattering and Brillouin scattering, two fiber optic sensing technologies, we can obtain dynamic information such as light intensity and strain caused by disturbances, so as to fully grasp the safety status of the roadbed, achieve universal coverage monitoring, and highlight potential risks.

[0071] AI video image capture devices can include AI cameras, drones, and laser radars, etc., as long as they can achieve video image capture and analysis. This solution prefers AI cameras, and is supplemented by drones and laser radars. When risks occur, one or more AI video image capture devices can be mobilized to respond.

[0072] When the AI ​​video image capture device uses an AI camera, the video image capture device subsystem can be arranged as follows.

[0073] High-definition AI cameras are deployed at equal intervals along the highway to achieve full video surveillance coverage of the roadbed. The AI ​​cameras use the latest intelligent visual analysis technology and have built-in high-performance AI processing chips to achieve functions such as target detection, behavior analysis, and abnormal alarms.

[0074] When the distributed fiber optic sensing subsystem detects an abnormal situation, it will promptly trigger the AI ​​camera to conduct key monitoring and analysis of the corresponding road section, identify the abnormal cause (such as landslides, vehicle collisions, human damage, etc.), and trigger an alarm to guide safety precautions and emergency response.

[0075] The advantage of AI cameras is that they can provide intuitive visual information, and when combined with fiber optic sensors, they can achieve more comprehensive and accurate abnormality analysis and diagnosis. The disadvantage is that they are relatively expensive and are affected by ambient lighting and occlusion.

[0076] For cost considerations, AI cameras may not be deployed on all roads, but only on high-risk roads. AI video image capture devices may not be deployed on low-risk roads, or low-cost AI video image capture devices may be deployed. When risks occur on roads where AI video image capture devices are not deployed, nearby AI video image capture devices (such as drones) can be called to respond.

[0077] In order to obtain deformation index parameters, a preset calculation model needs to be built in advance. The construction method of the preset calculation model includes:

[0078] Preprocess the dynamic information to obtain statistical features;

[0079] The statistical features are fused and normalized to obtain a normalized composite feature vector;

[0080] Based on the clustering algorithm, the normalized composite feature vectors are analyzed to form clusters;

[0081] The clusters are evaluated to obtain deformation index parameters.

[0082] The construction of the preset calculation model can be carried out through simulation in a simulation environment, data feedback in a small test field, data feedback in actual use, or a combination of the above. The deformation index parameters include the section length, section maximum displacement, section average strain rate, section average displacement rate, section maximum light intensity attenuation, and section average light intensity attenuation rate related to the risk section.

[0083] "Section" refers to a number of monitoring units divided along the highway, each covering a certain length of the road section. The purpose of segmentation is to facilitate the description of the deformation state of the highway in sections and to achieve zoning monitoring and management of the highway. Specifically, the entire highway can be divided into several sections based on factors such as the length, importance, and complexity of geological conditions.

[0084] The construction of the preset calculation model is introduced in detail below.

[0085] In order to simplify the model derivation process, three cases are now considered uniformly: only distributed Rayleigh scattering fiber is laid, only distributed Brillouin scattering fiber is laid, and both distributed Rayleigh scattering fiber and distributed Brillouin scattering fiber are laid.

[0086] According to the resolution of distributed Brillouin scattering fiber and distributed Rayleigh scattering fiber, the Brillouin sensing unit is defined as: a basic strain measurement unit in the Brillouin scattering fiber sensing system, with a temperature measurement accuracy of ±1°C, a strain measurement resolution of 1με, a strain measurement accuracy of ±15με, and an event space positioning measurement accuracy of approximately ±0.5m. The Rayleigh sensing unit is defined as: a basic optical loss measurement unit in the Rayleigh scattering fiber sensing system, with an optical loss measurement resolution of approximately 0.01dB, a measurement accuracy of approximately ±0.25dB, and an event space positioning measurement accuracy of approximately ±1m.

[0087] Assume that there are M Brillouin fiber sensing units and N Rayleigh fiber sensing units. For the i-th Brillouin unit and the j-th Rayleigh unit, the strain values ​​at time t are ε i (t), the light intensity value is I j (t).

[0088] 1. Data preprocessing

[0089] In order to sensitively reflect the changes in the sensor unit in the near period, a time sliding window is set, and the time window length T is set w , represents the number of historical time steps considered when calculating the statistical features of the sensor unit. Then at time t, the time window pt-T is extracted for each sensor unit w+1,t] time series statistical characteristics:

[0090] Brillouin unit i:

[0091] (1) Mean strain:

[0092] (2) Strain variance:

[0093] (3) Mean value of strain change rate:

[0094] (4) Strain change rate variance:

[0095] Rayleigh unit j:

[0096] (1) Mean light intensity:

[0097] (2) Light intensity variance:

[0098] (3) Mean value of light intensity change rate:

[0099] (4) Variance of light intensity change rate:

[0100] Where Δt is the sampling interval.

[0101] 2. Feature fusion and normalization

[0102] For each pair of adjacent Brillouin units i and Rayleigh units j, their statistical features are fused into a composite feature vector:

[0103]

[0104] For all composite eigenvectors f ij Perform Z-score normalization processing.

[0105] For the characteristic components of Brillouin fiber (such as strain mean, strain variance, etc.), their mean and standard deviation should only consider the number of Brillouin units M:

[0106]

[0107] Among them, f i (k) represents the kth eigencomponent of the i-th Brillouin unit.

[0108] For the characteristic components of Rayleigh fiber (such as light intensity mean, light intensity variance, etc.), their mean and standard deviation should only consider the number N of Rayleigh units:

[0109]

[0110] Among them, f j (k) represents the kth eigencomponent of the jth Rayleigh unit.

[0111] When performing Z-score normalization, you need to select the corresponding mean and standard deviation according to the type of feature component:

[0112]

[0113] Therefore, the normalized composite eigenvector is:

[0114]

[0115] Among them, the first four components belong to Brillouin fiber, and the last four components belong to Rayleigh fiber.

[0116] 3. DBSCAN Clustering

[0117] Since the density-based clustering algorithm can well adapt to the characteristics of highway roadbed deformation monitoring: (1) it can find clusters of any size and is suitable for processing the complex linear spatial distribution of roadbed deformation; (2) it has strong robustness to noise data and can effectively process outliers in the monitoring data; (3) it does not need to pre-specify the number of clusters and is suitable for situations where the number of deformed sections is unknown; therefore, this application prefers the density-based clustering algorithm (Density-Based Spatial Clustering of Applications with Noise, DBSCAN). It can be understood that other clustering algorithms can also be used. For example, when the number of deformed sections is roughly known, the K-means clustering algorithm can be used. When it is necessary to analyze the hierarchical relationship between deformed sections, a hierarchical-based clustering algorithm can be used. According to the DBSCAN clustering algorithm, the present invention makes the following assumptions:

[0118] For all normalized eigenvectors Perform DBSCAN clustering, set the neighborhood radius to ∈, and the minimum number of points in the neighborhood to MinPt.

[0119] For any two eigenvectors and The Euclidean distance between them is:

[0120]

[0121] For the feature vector Its ∈-neighborhood is defined as:

[0122]

[0123] if Then it is called A core object.

[0124] DBSCAN clustering starts from any core object and recursively finds all objects that are density-connected to it through density-reachable relationships to form a cluster.

[0125] The steps of the DBSCAN algorithm are as follows:

[0126] (1) Choose an unvisited object p (i.e., an unvisited feature vector Corresponding to the fusion feature of a Brillouin unit i and Rayleigh unit j), mark it as visited and calculate its ∈-neighborhood N ∈ (p).

[0127] (2) If |N ∈ (p)|<MinPts, then p is marked as a noise point; otherwise, a new cluster C is created, p is added to C, and N ∈ All points in (p) are added to the candidate set N.

[0128] (3) For each point q in N (i.e., in the ∈-neighborhood N ∈ (p) Corresponding to the fusion characteristics of a Brillouin unit i and Rayleigh unit j), calculate its ∈-neighborhood N ∈ (q):

[0129] 1) If |N ∈ (q)|≥MinPt, then N ∈ The unvisited points in (q) are added to N;

[0130] 2) If q does not belong to any cluster, add it to cluster C.

[0131] (4) Repeat step 3 until N is empty, that is, cluster C is expanded.

[0132] (5) Repeat steps (1)-(4) until all points have been visited.

[0133] Finally, we get several clusters and some noise points, each cluster corresponds to a deformation segment.

[0134] 4. Calculation of deformation index parameters

[0135] For each deformation segment (cluster) C k , calculate the following deformation index parameters:

[0136] (1) Segment length L k

[0137] Assume deformation section C k The set of Brillouin cell indices included is The Rayleigh unit index set is The elements are sorted from small to large according to the coordinates of the unit center.

[0138] The center coordinates of Brillouin cell i are defined as The center coordinates of Rayleigh unit j are The length of the Brillouin cell is Δx B , the length of the Rayleigh unit is Δx R .

[0139] Considering the deformation section C k The actual boundary point is located at the boundary of the outermost unit, not the unit center. Therefore, when calculating the segment length, it is necessary to add or subtract half the unit length based on the center coordinates of the outermost unit.

[0140] Then the deformation section C k The actual starting coordinates and end coordinates for:

[0141]

[0142] Therefore, the deformation section C k The length L k for:

[0143]

[0144] (2) Maximum displacement of the segment

[0145] Assuming that the displacement of each point of the Brillouin fiber in the initial state is 0, the calculated displacement change relative to the initial state is obtained. By making full use of the strain information of the Brillouin fiber, the displacement distribution of each position on the entire fiber is obtained, and then the displacement distribution curve is intercepted through the range of the deformation section to obtain the true displacement value of each position in the section, including the maximum displacement.

[0146] Assume that the spatial resolution of the Brillouin fiber (i.e. the distance between adjacent measurement points) is Δx B , then the fiber interval corresponding to the i-th Brillouin sensor unit is [(i-1)Δx B ,iΔx B ], whose center position is x i =(i-0.5)Δx B .

[0147] For the entire fiber, the displacement u(x) at any position x∈[0,L] can be calculated by the accumulated strain:

[0148]

[0149] Where m(x) = max{i|x i ≤x} represents the number of Brillouin cells before position x, is the average strain of the ith Brillouin unit.

[0150] By using the above formula, the displacement value at each position on the optical fiber can be calculated to form a displacement distribution curve u(x).

[0151] Then, for each deformation segment (cluster) C k , whose range is in:

[0152]

[0153] Using u(x), we can directly get the deformation section C k The displacement value of each position in is recorded as

[0154] Therefore, the deformation section C k Maximum displacement It can be expressed as:

[0155]

[0156] (3) Segment average strain rate

[0157] Based on the above analysis, the average strain rate of the segment can be obtained

[0158]

[0159] (4) Average displacement rate of the segment

[0160] Directly use the true displacement value of each position in the deformation section at two moments, and through the entire deformation section C k By averaging the displacement changes, we can get the average displacement rate of the entire section to reflect the overall trend of displacement changes in the section. However, this method requires obtaining a complete displacement distribution curve at two times, so it has high requirements for data collection and storage.

[0161] Assume that the two acquisition times are t1 and t2 respectively, and the cumulative strain formula has been applied at time t1 and t2:

[0162]

[0163] Where m(x) = max{i|x i ≤x} represents the number of Brillouin cells before position x, is the average strain of the ith Brillouin unit.

[0164] The displacement distribution curves u(x, t1) and u(x, t2) of each position on the entire optical fiber at time t1 and t2 are obtained respectively, and then the displacement distribution curves of each deformation segment (cluster) C can be directly obtained. k The displacement value at time t1 and t2 is denoted as u k (x,t1) and u k (x,t2), where x∈ Then the deformation section C k The displacement change of any position x within the sampling interval Δt=t2-t1 is:

[0165] Δu k (x) = u k (x,t2)-u k (x,t1)

[0166] By averaging the displacement changes of all positions in the segment, we can get the average displacement change of the segment:

[0167]

[0168] Then divide the average displacement change by the sampling interval to get the deformation section C k The average displacement rate is:

[0169]

[0170] (5) Maximum light intensity attenuation in the segment

[0171]

[0172] Among them, J k Cluster C k The set of Rayleigh cell indices contained in it, is the mean initial light intensity of Rayleigh unit j.

[0173] (6) Average light intensity decay rate in the segment

[0174] Based on the above analysis, the average light intensity attenuation rate of the segment can be obtained.

[0175]

[0176] In order to meet the monitoring needs and take economic applicability into consideration, specific designs are made for the deployment of distributed Rayleigh scattering optical fibers and / or distributed Brillouin scattering optical fibers according to the risk level, including:

[0177] For the first-level risk sections, distributed Rayleigh scattering optical fibers are laid on the surface of the highway guardrail structure;

[0178] For the secondary risk sections, distributed Brillouin scattering optical fibers are laid on the surface of the highway guardrail structure;

[0179] For level 3 risk sections, distributed Brillouin scattering optical fiber is laid on the surface of the highway guardrail structure, and distributed Rayleigh scattering optical fiber is laid on or inside the shoulder on the side of the roadbed.

[0180] For sections with different risk levels, a combination of distributed Rayleigh scattering fiber and distributed Brillouin scattering fiber is used to conduct graded and segmented monitoring, so that the monitoring method matches the risk level and takes into account both monitoring effect and cost:

[0181] (1) For sections with general risks, distributed Rayleigh scattering optical fibers are laid on the surface of the highway guardrail structure for continuous monitoring. By measuring the time domain changes of light intensity, the deformation state of the guardrail structure (deformation degree and length range) is preliminarily analyzed to achieve large-scale safety overall prevention and control;

[0182] (2) For key sections with certain risks, distributed Brillouin scattering optical fibers are laid on the surface of highway guardrail structures for continuous monitoring. By measuring the time domain changes of optical fiber strain, the stress state of the structure can be directly obtained with high precision, thus achieving refined monitoring of the structural safety of key areas;

[0183] (3) In local sections with high potential risk threats (fill, half-fill and half-excavation), on the basis of (2), distributed Rayleigh optical fibers are continuously laid on the shoulders of the roadbed to form a continuous coverage dual-parameter verification monitoring mechanism to improve the monitoring robustness and reliability of key areas.

[0184] The highway roadbed safety monitoring system can achieve the following beneficial effects:

[0185] 1. Utilize the low-cost advantage of distributed Rayleigh optical fiber to achieve full-line continuous coverage, making up for the shortcoming of Brillouin optical fiber that is difficult to lay out throughout the entire route due to cost constraints. While saving investment, it also eliminates monitoring blind spots to the greatest extent.

[0186] 2. Although the event space positioning accuracy of Rayleigh fiber is lower than that of Brillouin fiber, the macro deformation trend and abnormal events of the guardrail structure can still be effectively identified and located through the AI ​​video image capture device and the preset calculation model.

[0187] 3. Brillouin fiber can provide strain data with high accuracy (resolution 1με, accuracy ±15με), which can directly reflect the deformation of the guardrail structure without calibration. It is an important supplement to Rayleigh fiber and can be used for refined monitoring of key sections.

[0188] 4. The entire monitoring system adopts all-fiber sensing, which has the advantages of intrinsic safety, anti-electromagnetic interference, corrosion resistance, waterproof and dustproof, high and low temperature resistance, etc. It can adapt to the harsh environment of roads, is maintenance-free and has a long service life.

[0189] 5. Combine fiber optic sensing data with deployed AI video image capture equipment to build a complete highway roadbed safety risk monitoring system. When the deformation amount and range monitored by the fiber optic exceeds the preset threshold, the AI ​​video image capture equipment of the corresponding road section is immediately triggered to focus on the abnormal area and intelligently identify, classify and verify abnormal events that cause deformation (such as landslides, vehicle collisions, human interference, etc.).

[0190] like Figure 2 As shown, it is a flow chart of a risk identification method using the above highway roadbed safety monitoring system provided by the present application, and the method comprises:

[0191] According to the characteristics of the dynamic information, the segment maximum displacement threshold, the segment average displacement rate threshold, the segment maximum light intensity attenuation threshold and the segment average light intensity attenuation rate threshold are determined respectively;

[0192] Compare the maximum displacement of the section, the average displacement rate of the section, the maximum light intensity attenuation of the section, and the average light intensity attenuation rate of the section with the corresponding thresholds to determine whether a large deformation has occurred in the monitored section;

[0193] According to the characteristics of the dynamic information, a segment length threshold is set for the segment length;

[0194] The normalized over-limit ratio of the maximum displacement of the segment, the normalized over-limit ratio of the average displacement rate of the segment, the normalized over-limit ratio of the maximum light intensity attenuation of the segment, the normalized over-limit ratio of the average light intensity attenuation rate of the segment and the normalized over-limit ratio of the segment length are calculated respectively. The normalized over-limit ratio represents the ratio of the difference between the actual value of the deformation index parameter and the threshold value of the deformation index parameter to the threshold value of the deformation index parameter.

[0195] According to the characteristics of dynamic information, the maximum displacement weight of the segment, the average displacement rate weight of the segment, the maximum light intensity attenuation weight of the segment, the average light intensity attenuation rate weight of the segment and the segment length weight are determined respectively;

[0196] The comprehensive deformation index is calculated based on the normalized over-limit ratio and weight of the maximum displacement of the segment, the average displacement rate of the segment, the maximum light intensity attenuation of the segment, the average light intensity attenuation rate of the segment and the segment length;

[0197] Based on the comprehensive deformation index, different levels of response based on AI camera functions are triggered.

[0198] Specifically, for each deformation segment (cluster) C k , through the calculation model, the following deformation index parameters have been obtained:

[0199] (1) Maximum displacement of the segment

[0200] (2) Average displacement rate of the segment

[0201] (3) Maximum light intensity attenuation in the segment

[0202] (4) Average light intensity decay rate in the segment

[0203] (5) Segment range length

[0204] When the AI ​​video image capture device uses an AI camera, the specific implementation steps are as follows:

[0205] (1) Determine the thresholds of the first four indicators

[0206] According to the characteristics of historical data (such as the characteristics of dynamic information), thresholds are set for the first four indicators to indicate the criterion for a large deformation. th , ΔI th , They are the thresholds of the maximum displacement, average displacement rate, maximum light intensity attenuation and average light intensity attenuation rate of the segment. “Historical data characteristics” not only refer to the statistical characteristics obtained after data preprocessing, but also a general term for all kinds of data that have been obtained, that is, all kinds of monitoring data that are helpful to formulate more reasonable and reliable threshold standards.

[0207] (2) Determine whether a large deformation occurs

[0208] For each deformation segment (cluster) C k , if any of its first four indicators exceeds the corresponding threshold, that is: or or or

[0209] It is considered that the deformation section C k A large deformation has occurred and the severity of the deformation needs to be further assessed.

[0210] The purpose of determining whether a large deformation has occurred is mainly to distinguish between general deformation and large deformation, and to provide a basis for the subsequent multi-level response mechanism. Only sections that are determined to have large deformations will enter a more detailed comprehensive deformation index calculation and multi-level response process, which serves as a preliminary screening mechanism for triggering the response of the AI ​​video image capture device. By quickly determining whether a large deformation has occurred, it is possible to avoid starting the AI ​​video image capture device for all minor deformations, thereby improving system efficiency.

[0211] (3) Determine the threshold of the segment range length index

[0212] According to the characteristics of historical data (such as the characteristics of dynamic information), the segment range length indicator L k Set the threshold L th , as a reference when evaluating the severity of deformation.

[0213] (4) Calculate the normalized excess ratio of the five indicators

[0214] For the section C where a large deformation occurs k , calculate the difference between the actual value of the five indicators and the threshold, and then divide it by the threshold to get a dimensionless normalized over-limit ratio. Δu , r ΔI , r L are the maximum displacement, average displacement rate, maximum light intensity attenuation, average light intensity attenuation rate and the excess ratio of the range length, respectively:

[0215]

[0216] In general, the actual value of the deformation index parameter is greater than its threshold value. The above is the calculation of the normalized over-limit ratio of the deformation index in general. In special cases, if the actual value of a deformation index parameter is less than or equal to its threshold value, the corresponding over-limit ratio is 0. In a physical sense, in this case, the influence of the deformation index parameter on the structural deformation can be ignored. Therefore, when calculating the comprehensive deformation index in the subsequent calculation, it is not expected that this index will contribute. In order to achieve this mathematically, it is agreed that the over-limit ratio in this case is directly assigned to 0.

[0217] (5) Determine the weights of the five indicators

[0218] According to the characteristics of historical data (e.g., the characteristics of dynamic information), weights are assigned to the five indicators to indicate their relative importance to the severity of deformation. Δu , w ΔI , w Lare the weights of the maximum displacement, average displacement rate, maximum light intensity attenuation, average light intensity attenuation rate and range length of the segment, and they satisfy:

[0219]

[0220] (6) Calculation of comprehensive deformation index

[0221] For the section C where a large deformation occurs k , multiply the excess ratio of the five indicators by the weight, and then sum them up to get a comprehensive deformation index D k :

[0222]

[0223] Comprehensive deformation index D k Reflects the deformation section C k The overall deformation severity is calculated, and the contribution and importance of each indicator are taken into account.

[0224] (7) Establish a multi-level response trigger mechanism for AI cameras

[0225] According to the comprehensive deformation index D k The size of the trigger triggers different levels of response based on the AI ​​camera function:

[0226] 1) If D th <D k ≤D th +ΔD, the first-level response is triggered, the pan / tilt control function of the AI ​​image capture device is activated, the camera angle is immediately adjusted to review and monitor the section, and a preliminary abnormal cause analysis is performed, and an alarm is issued;

[0227] 2) If D th +ΔD <D k , then the second level response is triggered, and based on the first level response, it immediately guides the manual on-site investigation and review work.

[0228] Among them, D k is the comprehensive deformation index, D th is the warning value of the comprehensive deformation index, and ΔD is the preset increment, which is used to divide different response levels. In actual application, D needs to be determined by the road operation management unit or engineering technicians based on the engineering conditions and management requirements of different road sections. k , D th and the specific value of ΔD.

[0229] When the front-end video image capture devices include AI cameras, drones, and lidars, one or more front-end video image capture devices can be driven to perform response actions according to different response levels.

[0230] This method is not only applicable to the highway roadbed safety risk monitoring scenario in the present invention, but also to other similar engineering scenarios.

[0231] This risk identification method can achieve the following beneficial effects:

[0232] 1. Through the hierarchical deployment of distributed Rayleigh optical fiber and distributed Brillouin optical fiber, while eliminating monitoring blind spots on the entire road section, accurate diagnosis of key risk areas can be achieved, taking into account both monitoring effect and cost.

[0233] 2. Adaptively extract deformation sections through clustering algorithms, calculate multiple deformation indicators, and comprehensively quantify the spatiotemporal characteristics of roadbed deformation.

[0234] 3. Through the intelligent recognition and multimodal fusion of AI cameras, it provides intuitive and visual auxiliary diagnostic information to improve the reliability and interpretability of monitoring.

[0235] 4. Through multi-indicator comprehensive evaluation and graded response of AI video image capture equipment, while improving the sensitivity of early warning, avoid overreaction and waste of resources.

[0236] 5. Form an integrated technical chain of "monitoring-early warning-analysis-reporting" to achieve early detection, early warning and early disposal of roadbed safety risks.

[0237] like Figure 3 , which is a schematic diagram of the structure of an electronic device according to an embodiment of the present application. As shown in the figure, the electronic device 10 includes one or more processors 11 and a memory 12.

[0238] The processor 11 may be a central processing unit (CPU) or other forms of processing units having data processing capabilities and / or instruction execution capabilities, and may control other components in the electronic device 10 to perform desired functions.

[0239] The memory 12 may include one or more computer program products, and the memory may include various forms of computer-readable storage media, such as volatile memory and / or non-volatile memory. The volatile memory may include, for example, random access memory (RAM) and / or cache memory (cache), etc. The non-volatile memory may include, for example, read-only memory (ROM), hard disk, flash memory, etc. One or more computer program instructions may be stored on the computer-readable storage medium, and the processor 11 may run the program instructions to implement the risk identification method of each embodiment of the present application described above and / or other desired functions. Various contents such as input signals, signal components, noise components, etc. may also be stored in the computer-readable storage medium.

[0240] In one example, the electronic device 10 may further include: an input device 13 and an output device 14, and these components are interconnected via a bus system and / or other forms of connection mechanisms (not shown).

[0241] Of course, to simplify, Figure 3 Only some of the components related to the present application in the electronic device 10 are shown, and components such as a bus, an input / output interface, etc. are omitted. In addition, according to specific application situations, the electronic device 10 may also include any other appropriate components.

[0242] Based on the embodiments of the present application, the present application also provides a computer-readable storage medium, wherein computer instructions are used to enable a computer to execute each step in the risk identification method of the aforementioned technical solution.

[0243] Based on the embodiments of the present application, the present application also provides a processor, which is used to run a program, wherein the program executes each step of the risk identification method of the aforementioned technical solution when it is running.

[0244] It should be understood that each part of the present invention can be implemented by hardware, software, firmware or a combination thereof. In the described embodiment, a plurality of steps or methods can be implemented by software or firmware stored in a memory and executed by a suitable instruction execution system.

[0245] In addition, each functional unit in each embodiment of the present invention may be integrated into a processing module, or each unit may exist physically separately, or two or more units may be integrated into one module. The integrated module may be implemented in the form of hardware or in the form of a software functional module. If the integrated module is implemented in the form of a software functional module and sold or used as an independent product, it may also be stored in a computer-readable storage medium.

[0246] The above is only a specific implementation of this specification, but the protection scope of this specification is not limited to this. Any technician familiar with this technical field can easily think of changes or replacements within the technical scope disclosed in this specification, which should be included in the protection scope of this specification. Therefore, the protection scope of this specification should be based on the protection scope of the claims.

Claims

1. A highway roadbed safety monitoring system, characterized in that: include: a distributed optical fiber sensing subsystem and a video image capturing device subsystem respectively connected to the back-end server; The distributed optical fiber sensing subsystem includes distributed Rayleigh scattering optical fibers and / or distributed Brillouin scattering optical fibers arranged on the roadbed according to the risk level, and is used to collect dynamic information related to the roadbed, wherein the dynamic information represents the environment and state changes of the roadbed; The video image capture device subsystem includes a plurality of AI video image capture devices; After receiving the dynamic information, the back-end server inputs the dynamic information into a preset calculation model, obtains the deformation index parameters output by the preset calculation model, determines the response level of the AI ​​video image capture device according to the deformation index parameters, and configures the operating parameters of the AI ​​video image capture device based on the response level.

2. The system according to claim 1, characterized in that The method of constructing the preset calculation model includes: Preprocessing the dynamic information to obtain statistical features; fusing and normalizing the statistical features to obtain a normalized composite feature vector; Analyzing the normalized composite feature vector based on a clustering algorithm to form clusters; The cluster is calculated to obtain the deformation index parameter.

3. The system according to claim 1, characterized in that The distributed Rayleigh scattering optical fiber and / or distributed Brillouin scattering optical fiber arranged on the roadbed according to the risk level includes: For the first-level risk section, the distributed Rayleigh scattering optical fiber is laid on the surface of the highway guardrail structure; For the secondary risk road section, the distributed Brillouin scattering optical fiber is laid on the surface of the highway guardrail structure; For the third-level risk section, the distributed Brillouin scattering optical fiber is laid on the surface of the highway guardrail structure, and the distributed Rayleigh scattering optical fiber is laid on the shoulder of the roadbed side.

4. The system according to claim 1, characterized in that The AI ​​video image capturing device includes an AI camera, a drone and a lidar.

5. The system according to claim 1, characterized in that The dynamic information includes light intensity, temperature and strain.

6. The system according to claim 2, characterized in that The clustering algorithm includes a density-based clustering algorithm.

7. The system according to claim 1, characterized in that The deformation index parameters include: segment length, segment maximum displacement, segment average strain rate, segment average displacement rate, segment maximum light intensity attenuation and segment average light intensity attenuation rate.

8. A risk identification method using the system according to any one of claims 1 to 7, characterized in that: include: According to the characteristics of the dynamic information, respectively determine the segment maximum displacement threshold, the segment average displacement rate threshold, the segment maximum light intensity attenuation threshold and the segment average light intensity attenuation rate threshold; Respectively comparing the maximum displacement of the section, the average displacement rate of the section, the maximum light intensity attenuation of the section and the average light intensity attenuation rate of the section with corresponding thresholds to determine whether a large deformation occurs in the road section to be monitored; According to the characteristics of the dynamic information, a segment length threshold is set for the segment length; Calculate the normalized over-limit ratio of the maximum displacement of the segment, the normalized over-limit ratio of the average displacement rate of the segment, the normalized over-limit ratio of the maximum light intensity attenuation of the segment, the normalized over-limit ratio of the average light intensity attenuation rate of the segment and the normalized over-limit ratio of the segment length respectively, the normalized over-limit ratio represents the ratio of the difference between the actual value of the deformation index parameter and the threshold value of the deformation index parameter to the threshold value of the deformation index parameter; According to the characteristics of the dynamic information, respectively determine the segment maximum displacement weight, the segment average displacement rate weight, the segment maximum light intensity attenuation weight, the segment average light intensity attenuation rate weight and the segment length weight; Calculate a comprehensive deformation index according to the normalized over-limit ratio and weight of the maximum displacement of the segment, the average displacement rate of the segment, the maximum light intensity attenuation of the segment, the average light intensity attenuation rate of the segment and the length of the segment; Based on the comprehensive deformation index, different levels of response based on the AI ​​camera function are triggered.

9. An electronic device, characterized in that: include: at least one processor; and a memory communicatively connected to the at least one processor; wherein the memory stores instructions executable by the at least one processor, the instructions being executed by the at least one processor so that the at least one processor can execute the steps in the method of claim 8.

10. A computer-readable storage medium, characterized in that: The computer-readable storage medium includes a stored program, wherein the program executes each step of the method of claim 8 .

Images