Road crack monitoring method and system based on knowledge graph and disaster mechanism constraint

Abstract

This invention discloses a method and system for monitoring road cracks based on knowledge graphs and disaster-causing mechanisms. The method includes: obtaining a four-channel fused tensor based on a sequence of visible light images of the road surface and a synchronous infrared thermal image; inputting this tensor into a crack detection network to obtain a predicted crack probability map and crack category map; generating a candidate semantic fact set based on the crack probability map, crack category map, and basic information; generating a crack cause explanation chain based on the candidate semantic fact set and by calling a pre-constructed road crack disaster-causing mechanism attribute map; correcting and risk-assessing crack instances based on mechanism matching degree and mechanism consistency score, and updating the road crack disaster-causing mechanism attribute map; calculating the crack propagation rate based on the corrected crack instances; and calculating the crack risk level and maintenance priority based on crack information and crack propagation rate. This invention achieves closed-loop processing of road crack monitoring, mechanism verification, risk assessment, maintenance decision-making, and map updating.

Description

Technical Field

[0001] This invention relates to the field of road crack monitoring technology, and in particular to a road crack monitoring method, system, storage medium, and electronic device based on knowledge graph and disaster formation mechanism constraints. Background Technology

[0002] Road cracks are one of the most common early-stage defects in pavement structures during their service life. Once cracks form, rainwater infiltrates along them, temperature stress persists, and repeated traffic loads further weaken weak points in the base course and subgrade, eventually leading to crack expansion, network cracking, edge loosening, potholes, and localized instability. The accuracy of crack monitoring results directly impacts the timing of defect treatment and maintenance costs; therefore, road management departments continuously need an intelligent monitoring method that can operate reliably under complex conditions.

[0003] Existing crack monitoring schemes primarily rely on image texture features, edge features, and segmentation results from deep neural network outputs. While these schemes achieve good recognition results under standard lighting conditions, they frequently misidentify non-cracked targets as cracks in scenarios with shadows, water accumulation, repair marks, low light, tire wear, road marking interference, and localized reflections. The root cause of this problem lies in the fact that existing schemes mainly utilize superficial information for discrimination, failing to incorporate the road structure layer condition, environmental hazards, traffic loads, and historical damage evolution knowledge into the same judgment process.

[0004] Road cracks are not isolated surface textures, but rather the result of a combination of factors including material aging, base layer damage, water seepage, overloading, freeze-thaw cycles, and uneven settlement. The main morphology, propagation direction, and treatment measures for cracks vary significantly depending on the road grade, surface material, and structural layer combination. Relying solely on a visual model to output a crack label cannot answer engineering questions such as why the crack forms, whether it poses a risk of rapid propagation, and which maintenance measures should be prioritized.

[0005] While some existing systems incorporate disease knowledge bases, most of these bases only perform static query functions and do not participate in the screening and re-scoring of visual inspection results. They also fail to form a write-back loop between monitoring results and the knowledge base, and fail to transform time-series monitoring results into evolutionary relationships. This results in the system being able to output crack locations, but struggling to further output causal explanation chains, risk levels, and actionable maintenance priority lists. Summary of the Invention

[0006] This invention provides a road crack monitoring method, system, storage medium, and electronic device based on knowledge graphs and disaster formation mechanism constraints, which can solve the problems of lack of mechanism constraints in crack identification, many false detections in complex scenarios, insufficient risk interpretation, and unclear maintenance recommendations in the prior art.

[0007] This invention provides a road crack monitoring method based on knowledge graph and disaster formation mechanism constraints, including: Collect road surface visible light image sequences, synchronous infrared thermal images, basic road attribute data, environmental disaster-causing factor data, traffic load data, and spatiotemporal positioning data for the target road section within the current monitoring period; The visible light image sequence and the synchronous infrared thermal image of the road surface are preprocessed, and the preprocessed visible light image sequence and the preprocessed synchronous infrared thermal image are fused to obtain a four-channel fused tensor. The four-channel fusion tensor is input into the trained crack detection network to obtain the predicted crack probability map and crack category map. Crack information is extracted based on crack probability maps and crack category maps, and a candidate semantic fact set is generated based on road basic attribute data, environmental disaster-causing factor data, traffic load data, spatiotemporal positioning data, and crack information. Based on the candidate semantic fact set, and by calling the pre-constructed road crack disaster mechanism attribute map, a crack cause explanation chain is generated; Based on the mechanism matching degree and mechanism consistency score, the crack instances were corrected and risk assessments were performed, and the road crack disaster mechanism attribute map was updated. The crack propagation rate is calculated based on the corrected crack instance, and the crack risk level and maintenance priority are calculated based on the crack information and crack propagation rate.

[0008] Furthermore, according to the aforementioned road crack monitoring method based on knowledge graphs and disaster formation mechanisms, the visible light image sequence and synchronous infrared thermal image of the road surface are preprocessed, and the preprocessed visible light image sequence and the preprocessed synchronous infrared thermal image are fused to obtain a four-channel fusion tensor, including: Median filtering, contrast-limited adaptive histogram equalization, specular reflection suppression, and shadow correction were sequentially applied to the visible light image sequence of the road surface to obtain the preprocessed visible light image sequence of the road surface. Temperature normalization and geometric registration are performed on the synchronous infrared thermal image to obtain a preprocessed synchronous infrared thermal image. Then, perspective correction is used to unify the visible light image sequence of the road surface and the synchronous infrared thermal image to the same pixel coordinate system. The visible light image sequence of the road surface includes three visible light channels: red, green, and blue. The synchronous infrared thermal image includes a single-channel temperature matrix. The four-channel fusion tensor is obtained by superimposing the three visible light channels (red, green, and blue) and the single-channel temperature matrix.

[0009] Furthermore, according to the aforementioned road crack monitoring method based on knowledge graphs and disaster formation mechanisms, the crack detection network includes a backbone network, a feature pyramid network, a multi-head self-attention encoding layer, and a U-shaped decoding layer; the processing procedure of the crack detection network includes: The four-channel fusion tensor is input into the backbone network for multi-scale feature extraction to obtain shallow edge features and deep semantic features. Shallow edge features and deep semantic features of different scales are input into the feature pyramid network and fused to obtain a multi-scale feature map; The multi-scale feature map is input into a multi-head self-attention coding layer for feature flattening and multi-head self-attention calculation to obtain a global dependency representation. The global dependency representation is input into the U-shaped decoding layer to obtain the crack probability map and crack category map.

[0010] Furthermore, according to the above-mentioned road crack monitoring method based on knowledge graph and disaster formation mechanism constraints, the crack information includes crack location, crack length, crack width, crack direction, crack connectivity, thermal anomaly intensity, and visual detection confidence level. The candidate semantic fact set is a graph structure, including crack instance nodes, road segment nodes, lane nodes, structural layer nodes, environmental factor nodes, risk state nodes, and edges used to associate each node; the crack instance node corresponds to a crack instance, the road segment node corresponds to a road segment entity, the lane node corresponds to a lane entity, the structural layer node corresponds to a structural layer entity, the environmental factor node corresponds to the environmental factor action condition, and the risk state node corresponds to risk state information. The road basic attribute data is used to determine the road grade, structural layer combination and attributes, material type and service life of the road segment to which the crack instance belongs; the environmental disaster factor data is used to determine the conditions under which environmental factors act; the traffic load data is used to characterize the lane information of the lane entity; and the spatiotemporal positioning data is used to determine the station location, collection time and monitoring cycle number of the crack entity.

[0011] Furthermore, according to the above-mentioned road crack monitoring method based on knowledge graph and disaster formation mechanism constraints, the road crack disaster formation mechanism attribute graph uses a graph structure to store node attributes and relationship attributes. The node attributes include road entities, crack entities, crack attribute entities, disaster-causing factor entities, structural layer entities, risk state entities, and maintenance decision entities. The relationship attributes include located at, belong to, affected by, induced, aggravated, accompanied by, evolved into, corresponding risk level, recommended treatment measures, historical co-occurrence, and temporal precursor relationships. Based on a set of candidate semantic facts, and by invoking a pre-built attribute map of road crack formation mechanisms, an explanatory chain for crack formation causes is generated, including: Based on the road segment location, lane location, crack type, crack length, crack width, crack direction, thermal anomaly intensity, structural layer combination and attributes, environmental factors, traffic load and historical damage records of crack instances in the candidate semantic fact set, local subgraphs related to the current crack instance are retrieved from the road crack disaster mechanism attribute map. In the local subgraph, multiple preset disaster-causing rule paths are matched, and a chain of explanations for the causes of cracks is generated based on the successfully matched disaster-causing rule paths.

[0012] Furthermore, according to the aforementioned road crack monitoring method based on knowledge graphs and disaster-causing mechanisms, the crack instances are corrected and risk assessed based on mechanism matching degree and mechanism consistency score, and the road crack disaster-causing mechanism attribute map is updated, including: The computer calculates the matching degree of the mechanism and the consistency score of the mechanism. When the consistency score of the mechanism is less than the preset consistency threshold, the crack instance is removed, the category is corrected or the boundary is corrected. When the consistency score of the mechanism is greater than or equal to the preset consistency threshold, the crack instance is retained, and risk assessment and updating of the road crack disaster mechanism attribute map are performed based on the crack instance after removal of false detections, category correction or boundary correction.

[0013] Furthermore, according to the road crack monitoring method based on knowledge graphs and disaster-causing mechanisms described above, the mechanism matching degree is calculated using the following formula:

[0014] in, For mechanism matching degree, This represents the weight of the q-th disaster-causing rule. This indicates the satisfaction flag of the q-th disaster-causing rule. m This represents the total number of rules that cause a disaster. The consistency score of the mechanism is calculated using the following formula:

[0015] in, To score the consistency of the mechanism, Indicates the confidence level of visual detection. This indicates the support level for thermal anomalies.

[0016] This invention also provides a road crack monitoring system based on knowledge graphs and disaster formation mechanism constraints, comprising: The multi-source data acquisition module is used to collect road surface visible light image sequences, synchronous infrared thermal images, basic road attribute data, environmental disaster-causing factor data, traffic load data, and spatiotemporal positioning data of the target road section within the current monitoring period; The preprocessing and fusion module is used to preprocess the visible light image sequence and the synchronous infrared thermal image of the road surface, and then fuse the preprocessed visible light image sequence and the preprocessed synchronous infrared thermal image to obtain a four-channel fusion tensor. The crack detection module is used to input the four-channel fused tensor into the trained crack detection network to obtain the predicted crack probability map and crack category map. The candidate semantic fact set generation module is used to extract crack information based on crack probability map and crack category map, and generate candidate semantic fact set based on road basic attribute data, environmental disaster factor data, traffic load data and spatiotemporal positioning data and crack information. The crack formation explanation chain generation module is used to generate crack formation explanation chains based on a set of candidate semantic facts and by calling a pre-built road crack disaster mechanism attribute map. The crack instance correction and map write-back update module is used to correct and assess the risk of crack instances based on mechanism matching degree and mechanism consistency score, and update the road crack disaster mechanism attribute map. The crack risk level calculation and maintenance priority generation module is used to calculate the crack propagation rate based on crack information, and to calculate the crack risk level and maintenance priority based on crack information and crack propagation rate.

[0017] The present invention also provides a computer-readable storage medium storing a plurality of instructions adapted for loading by a processor to execute any of the above-described road crack monitoring methods based on knowledge graphs and disaster formation mechanism constraints.

[0018] The present invention also provides an electronic device, including a processor and a memory, wherein the processor is electrically connected to the memory, the memory is used to store instructions and data, and the processor is used in the steps of the road crack monitoring method based on knowledge graph and disaster formation mechanism constraints as described above.

[0019] This invention provides a road crack monitoring method, system, storage medium, and electronic device based on knowledge graphs and disaster-causing mechanism constraints. The core of this invention lies in converting the four-channel fusion detection results into a candidate semantic fact set, and using the disaster-causing rule paths in the road crack disaster-causing mechanism attribute graph to verify the mechanism consistency of candidate crack instances. The verification results are then used for false detection elimination, category correction, or boundary correction. The corrected crack instances then participate in risk assessment, maintenance priority generation, and graph write-back updates, thus forming a closed loop from detection and mechanism verification to knowledge increment updates. This invention has the following beneficial effects: First, this invention employs a fixed four-channel crack detection network and a fixed attribute graph reasoning framework (road crack disaster mechanism attribute graph), enabling stable integration between the detection and knowledge reasoning stages and avoiding the high false detection problem caused by relying solely on visual results. Second, this invention extends crack monitoring results from static identification to dynamic risk warning through temporal precursor relationships and expansion rate calculations. Furthermore, this invention outputs crack cause explanation chains, risk levels, and maintenance priorities, allowing the results to directly support inspection decisions and maintenance scheduling. Finally, after the monitoring results are written back to update the attribute graph, subsequent monitoring tasks can inherit existing knowledge and continuously update rule weights. Attached Figure Description

[0020] The technical solution and other beneficial effects of the present invention will become apparent from the following detailed description of specific embodiments of the invention, in conjunction with the accompanying drawings.

[0021] Figure 1 The flowchart illustrates a road crack monitoring method based on knowledge graphs and disaster formation mechanisms, as provided in this embodiment of the invention.

[0022] Figure 2 This is a flowchart for constructing and updating the road crack disaster mechanism attribute map provided in an embodiment of the present invention.

[0023] Figure 3 This is a schematic diagram of the road crack monitoring system based on knowledge graph and disaster formation mechanism constraints provided in an embodiment of the present invention.

[0024] Figure 4 This is a schematic diagram of the structure of an electronic device provided in an embodiment of the present invention. Detailed Implementation

[0025] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0026] This invention provides a road crack monitoring method, system, storage medium, and electronic device based on knowledge graphs and disaster formation mechanism constraints. The road crack monitoring system based on knowledge graphs and disaster formation mechanism constraints provided by this invention can be integrated into an electronic device, such as a terminal or server. The terminal can include a tablet computer, laptop computer, personal computer (PC), microprocessor box, or other devices.

[0027] Please see Figure 1 , Figure 1 The flowchart below illustrates a road crack monitoring method based on knowledge graphs and disaster formation mechanisms, provided in this embodiment of the invention. This method is applied in electronic devices. The following example, using a certain level highway asphalt pavement inspection scenario, details the method flow. The road crack monitoring method based on knowledge graphs and disaster formation mechanisms includes the following steps: S1 collects the road surface visible light image sequence, synchronous infrared thermal image, road basic attribute data, environmental disaster-causing factor data, traffic load data, and spatiotemporal positioning data of the target road section within the current monitoring period.

[0028] Multi-source monitoring data of the target road segment is collected within a monitoring cycle. The multi-source monitoring data includes visible light image sequences of the road surface, synchronous infrared thermal images, basic road attribute data, environmental disaster-causing factor data, traffic load data, and spatiotemporal positioning data. Among them, the basic road attribute data includes at least the road grade, surface material, base material, structural layer combination, construction year, and maintenance history; the environmental disaster-causing factor data includes at least rainfall, temperature, diurnal temperature range, humidity, freeze-thaw cycles, and groundwater seepage impact indicators; the traffic load data includes at least traffic volume, heavy load ratio, and axle load characteristics; and the spatiotemporal positioning data includes at least latitude and longitude, station number, collection time, and monitoring cycle number.

[0029] The specific implementation process of this step is as follows: A visible light camera, an infrared thermal imager, a positioning unit, and an edge computing terminal are fixedly installed on the inspection vehicle. The inspection vehicle continuously collects road surface images along the driving direction. Simultaneously, the edge computing terminal reads the road grade, structural layer information, repair records, and opening age corresponding to the current road segment from the road asset management database, and reads rainfall, temperature, humidity, freeze-thaw cycles, groundwater seepage, traffic volume, and heavy load ratio data for the corresponding time period from the meteorological and traffic interface. By uniformly binding the station number and latitude and longitude, the original input data package for the same crack monitoring task is generated. S2 preprocesses the visible light image sequence and the synchronous infrared thermal image of the road surface, and then fuses the preprocessed visible light image sequence and the preprocessed synchronous infrared thermal image to obtain a four-channel fused tensor.

[0030] In one embodiment, step S2 includes the following steps: S21, perform median filtering, contrast-limited adaptive histogram equalization, specular reflection suppression, and shadow correction sequentially on the visible light image sequence of the road surface to obtain the preprocessed visible light image sequence of the road surface.

[0031] S22, temperature normalization and geometric registration are performed on the synchronous infrared thermal image to obtain a preprocessed synchronous infrared thermal image, and perspective correction is used to unify the visible light image sequence of the road surface and the synchronous infrared thermal image into the same pixel coordinate system.

[0032] S23, the visible light image sequence of the road surface includes three visible light channels: red, green, and blue; the synchronous infrared thermal image includes a single-channel temperature matrix. The red, green, and blue visible light channels and the single-channel temperature matrix are superimposed to obtain a four-channel fused tensor.

[0033] Specifically, firstly, median filtering is used to remove high-frequency noise during the acquisition process, and then contrast-limited adaptive histogram equalization is used to enhance the grayscale differences of the cracks. For water accumulation or reflective areas, a specular reflection suppression method is used to reduce the occlusion effect of local bright spots on the crack texture. For backlit or locally shadowed areas, a shadow correction method based on the local brightness ratio is used to restore the crack texture details. After temperature normalization, the infrared thermal image is pixel-level perspective mapped to the visible light image through a pre-calibrated extrinsic parameter matrix, finally obtaining a fused input tensor with a size of 1024×1024×4.

[0034] S3. Input the four-channel fusion tensor into the trained crack detection network to obtain the predicted crack probability map and crack category map.

[0035] The crack detection network consists of a backbone network ResNet50, a feature pyramid network FPN, a multi-head self-attention coding layer, and a U-shaped decoding layer; the processing steps of the crack detection network include: S31, the four-channel fusion tensor is input into the backbone network for multi-scale feature extraction to obtain shallow edge features and deep semantic features; S32 inputs shallow edge features and deep semantic features of different scales into the feature pyramid network and fuses them to obtain a multi-scale feature map; S33, input the multi-scale feature map into the multi-head self-attention coding layer to perform feature flattening and multi-head self-attention calculation, and obtain the global dependency representation; S34, input the global dependency representation into the U-shaped decoding layer to obtain the crack probability map and crack category map.

[0036] The crack category map includes longitudinal cracks, transverse cracks, network cracks, reflective cracks, block cracks, edge cracks, and alligator cracks.

[0037] S4 extracts crack information based on crack probability map and crack category map, and generates a candidate semantic fact set based on road basic attribute data, environmental disaster-causing factor data, traffic load data, spatiotemporal positioning data and crack information.

[0038] The crack information includes crack location, crack length, crack width, crack direction, crack connectivity, thermal anomaly intensity, and visual inspection confidence level.

[0039] After threshold segmentation, skeleton refinement, and connected component filtering of the output crack probability map, the crack centerline length, maximum width, main direction angle, number of connected components, and thermal anomaly coverage ratio are calculated, and the visual detection confidence of each crack candidate region is obtained.

[0040] Specifically, the threshold segmentation steps are as follows: Select a suitable threshold (e.g., 0.5, or determine it using an adaptive method such as Otsu), and convert the crack probability map into a binary image. All pixels with a probability value greater than or equal to the threshold are marked as crack foreground (value 1), and the rest are background (value 0). This step yields a preliminary binary map of crack candidate regions.

[0041] The skeleton refinement process involves applying a morphological skeletonization algorithm (such as the Zhang-Suen parallel refinement algorithm) to the crack foreground region in the binary image of the crack candidate region. Boundary pixels are peeled off layer by layer, retaining the innermost single-pixel width of connected lines, thereby extracting the centerline skeleton of each crack region. The refined skeleton fully preserves the original crack's topological connectivity while eliminating width information.

[0042] The connected component selection process involves: labeling the connected components of the binary image (excluding the skeleton) using methods such as 4-neighborhood or 8-neighborhood to obtain several independent crack candidate regions. Based on preset area thresholds (such as minimum pixel count), shape factors (such as aspect ratio lower limit), or probability mean, isolated noise points or overly fragmented regions are removed, retaining the connected components corresponding to the actual cracks. Each retained connected component is considered an independent crack candidate region.

[0043] After completing the above preprocessing, the following geometric and thermal characteristics are calculated for each crack candidate region: Centerline length: Count the number of foreground pixels in the skeleton image of this region. If the spatial resolution of the image is known (mm / pixel or m / pixel), the physical length of the crack centerline (unit: mm or m) is obtained by multiplying the number of pixels by the actual length corresponding to each pixel. For curved cracks, this length is approximately the actual curve length.

[0044] Maximum width: Calculated using either a distance transformation method or a normal scan method. A common method is to perform an Euclidean distance transformation on the original binary crack region. The value of each foreground pixel is equal to its distance to the nearest background pixel. The maximum distance value within the region multiplied by 2 (or twice the distance radius according to the definition) is the maximum crack width. Alternatively, perpendicular lines can be drawn point by point along the centerline, and the distances between the two intersection points of the perpendicular lines and the crack boundary can be counted. The maximum value of all perpendicular line distances is then taken.

[0045] Principal orientation angle: Extract the coordinates of all foreground pixels in the crack region, calculate their covariance matrix, and perform principal component analysis (PCA). The direction of the eigenvector corresponding to the first principal component is the main direction of crack extension. The angle between this direction and the horizontal axis (x-axis) (usually between -90° and 90° or 0° and 180°) is the principal orientation angle. It can also be obtained by linear fitting based on the skeleton point set.

[0046] Thermal anomaly coverage ratio: This requires simultaneous access to thermal anomaly detection results aligned with the crack probability map (e.g., a binary image or continuous thermal image exceeding a specific temperature threshold in infrared thermography). For the current crack candidate area, the percentage of thermal anomaly pixels (or pixels exceeding a preset temperature threshold) within the area is calculated using the formula: Thermal anomaly coverage ratio = (Number of thermal anomaly pixels in the area / Total number of pixels in the area) × 100%. This indicator reflects the spatial overlap between the crack area and thermal anomaly phenomena, and is used to determine whether there are mechanistic supporting factors related to road defects in the crack area, such as moisture retention, material peeling, localized hollowing, abnormal moisture content in the base layer, or abnormal thermal response.

[0047] Visual detection confidence score: This indicates the credibility of the visual detection results for the candidate crack region. It is calculated based on a weighted average of multiple morphological and probabilistic features, such as: the average probability value of the crack region, the length-to-width ratio (slenderness), the continuity and number of branches of the skeleton (no breaks or excessive burrs), the region's convexity, and the proportion of thermal anomaly coverage (which can be used as a regularization term). Specifically, it can be implemented using a pre-trained logistic regression model or by setting rules (e.g., if the average probability is >0.7 and the aspect ratio is >5, the confidence score is 0.9; otherwise, it decays linearly) to obtain a confidence score between 0 and 1 for each region.

[0048] Finally, the total number of connected components after filtering is counted, which is the number of connected components. The above indicators (centerline length, maximum width, principal orientation angle, thermal anomaly coverage ratio, and visual detection confidence) are recorded in the attributes of each crack candidate region, and the crack information containing these parameters is finally output.

[0049] For each crack candidate region, a corresponding crack instance identifier is generated, and a mapping relationship is established between it and attributes such as station number, lane, crack type, length, width, direction, thermal anomaly intensity, and acquisition time. Then, it is written into the candidate semantic fact set.

[0050] The candidate semantic fact set is a graph structure, including crack instance nodes, road segment nodes, lane nodes, structural layer nodes, environmental factor nodes, risk state nodes, and edges used to connect each node; crack instance nodes correspond to crack instances, road segment nodes correspond to road segment entities, lane nodes correspond to lane entities, structural layer nodes correspond to structural layer entities, environmental factor nodes correspond to environmental factor conditions, and risk state nodes correspond to risk state information.

[0051] Road basic attribute data is used to determine the road grade, structural layer combination and attributes, material type and service life of the road segment to which the crack instance belongs; environmental disaster factor data is used to determine the conditions under which environmental factors act; traffic load data is used to characterize the lane information of the lane entity; spatiotemporal positioning data is used to determine the station location, collection time and monitoring cycle number of the crack entity.

[0052] S5 generates a chain of explanations for the causes of cracks based on a set of candidate semantic facts and by calling a pre-built attribute map of the road crack disaster mechanism.

[0053] Figure 2 The flowchart for constructing and updating the road crack disaster mechanism attribute map provided in this embodiment of the invention is as follows: The road crack disaster mechanism attribute map uses a graph structure to store node attributes and relationship attributes. The node attributes include road entity, crack entity, crack attribute entity, disaster-causing factor entity, structural layer entity, risk state entity, and maintenance decision entity. The relationship attributes include location, belonging to, affected by, induced, aggravated, accompanied, evolved into, corresponding risk level, recommended treatment measures, historical co-occurrence, and temporal precursor relationship.

[0054] Exemplary road crack formation mechanism attribute map facts include: (Road segment G1K102+300, belonging to, Class I highway); (Crack example C20260318-01, located, left lane wheel track zone); (Crack example C20260318-01, manifested as, longitudinal crack); (Heavy traffic, aggravated, longitudinal crack); (Groundwater seepage, aggravated, edge crack); (Crack example C20260311-01, temporally preceding crack example C20260318-01); (Severe risk, recommended treatment measure, milling and repaving).

[0055] The initial data sources for constructing the road crack disaster mechanism attribute map include road maintenance code provisions, historical inspection records, previous treatment records, on-site investigation records, expert rules, and candidate semantic fact sets generated during the current monitoring cycle. During construction, road segments, lanes, structural layers, cracks, disaster-causing factors, risk states, and maintenance measures are mapped as nodes, and the attributes of being located at, belonging to, affected by, induced, aggravated, accompanied by, evolved into, corresponding risk levels, and recommended treatment measures are mapped as edges. Time, space, structure, environment, and load attributes are written for nodes and edges.

[0056] In one embodiment, step S5 includes the following steps: S51, based on the road segment location, lane location, crack type, crack length, crack width, crack direction, thermal anomaly intensity, structural layer combination and attributes, environmental factors, traffic load and historical damage records of crack instances in the candidate semantic fact set, retrieve the local subgraphs related to the current crack instance from the road crack disaster mechanism attribute map.

[0057] S52, in the local subgraph, match according to multiple preset disaster-causing rule paths, and generate a crack cause explanation chain based on the successfully matched disaster-causing rule paths.

[0058] Among them, the preset multiple disaster-causing rule paths include at least "structural layer conditions - disaster-causing factors - crack morphology", "environmental conditions - thermal anomalies - crack propagation risk" and "historical diseases in the same location - temporal precursors - current crack status".

[0059] Disaster-causing rule paths can be generated from expert rules and historical co-occurrence relationships, including at least the following: semi-rigid base layers corresponding to reflective cracks due to temperature differences or rainfall; wheel tracks corresponding to high load ratios corresponding to longitudinal cracks; edge seepage and poor drainage corresponding to edge cracks; and historical defects in the same location and their temporal precursors corresponding to the current crack state. The weight of each rule path can be determined by expert assignment, normalization of historical co-occurrence frequencies, or adjustment based on feedback from retesting results. The rule satisfaction flag is set to 1 when all rule paths are satisfied, 0 when they are not satisfied, and when partially satisfied, the value can be determined based on the ratio of the number of satisfied conditions to the total number of conditions.

[0060] The explanation chain for crack formation includes at least four types of nodes: crack instances, main disaster-causing factors, risk status, and recommended remedial measures.

[0061] The specific implementation process of this step is as follows: When a candidate crack area is identified as a longitudinal crack located in the wheel track zone, and the corresponding road section has a high proportion of heavy traffic, a long service life of the surface layer, and local abnormal temperature rise in the infrared thermal image, the generated cause explanation chain can be expressed as: Crack instance C20260318-01 → Manifests as a longitudinal crack → Affected by heavy traffic → Accompanied by surface layer aging → Corresponds to moderate to severe risk of expansion → Recommended treatment measures are crack filling, local repair, or milling and repaving. For candidate results that clearly conflict with the map mechanism, such as repair edges, road marking remnants, stain shadows, and water reflections, they are marked as low-confidence visual anomalies and removed.

[0062] S6, based on mechanism matching degree and mechanism consistency score, corrects and assesses the risk of crack instances, and updates the disaster-causing mechanism attribute map of road cracks.

[0063] In one embodiment, step S6 includes: The computer calculates the matching degree of the mechanism and the consistency score of the mechanism. When the consistency score of the mechanism is less than the preset consistency threshold, the crack instance is removed, the category is corrected or the boundary is corrected. When the consistency score of the mechanism is greater than or equal to the preset consistency threshold, the crack instance is retained, and risk assessment and updating of the road crack disaster mechanism attribute map are performed based on the crack instance after removal of false detections, category correction or boundary correction.

[0064] The matching degree is calculated using the following formula:

[0065] in, For mechanism matching degree, This represents the weight of the q-th disaster-causing rule. This indicates the satisfaction flag of the q-th disaster-causing rule. m This represents the total number of rules that cause a disaster. The consistency score is calculated using the following formula:

[0066] in, To score the consistency of the mechanism, Indicates the confidence level of visual detection. This indicates the support level for thermal anomalies.

[0067] For ease of implementation and reference, this invention suggests the following mechanism verification methods: if the surface layer is aged and the construction period exceeds six years, increase the prior weight of longitudinal cracks and block cracks; if a semi-rigid base layer is superimposed with rainfall and temperature difference, increase the prior weight of reflective cracks; if there is water seepage at the edge and poor drainage at the shoulder, increase the prior weight of edge cracks; if cracks at the same historical location appear consecutively, increase the weight of propagation risk; if the infrared thermal anomaly area highly overlaps with the crack mask, increase the thermal anomaly support of the candidate target.

[0068] When the mechanism consistency score is lower than the preset consistency threshold, the following corrections can be made: When the visual detection confidence is higher than the preset confidence threshold, and the mechanism matching degree and thermal anomaly support degree are both lower than the corresponding mechanism matching degree threshold and thermal anomaly support degree threshold, and the candidate region is located in the marking, repair edge, or water accumulation reflective area, it is rejected as a false detection; when the visual crack category is inconsistent with the disaster-causing rule path with the highest matching degree in the map, the crack category is corrected to the crack category corresponding to the rule path; when the candidate crack mask locally overlaps with the thermal anomaly support area but the boundary expands outward, the crack boundary is corrected by shrinking or expanding according to the thermal anomaly support area, skeleton continuity, and connected domain boundary.

[0069] Furthermore, the road crack disaster mechanism attribute map is updated based on the crack correction results, risk status results, treatment suggestion results, and retest results, and the updated temporal precursor relationship and historical treatment records are called in subsequent monitoring cycles to participate in inference.

[0070] Specifically, the corrected crack instances, risk status, recommended measures, and retest times are written into the monitoring results database and simultaneously written back to the road crack disaster mechanism attribute map. For crack instances that reappear in subsequent inspections, existing temporal precursor relationships and historical treatment records are directly invoked to participate in the reasoning of the current cycle, so as to achieve continuous incremental updates of the map. Thus, this invention forms a closed-loop processing flow of "multi-source acquisition—visual inspection—mechanism verification—risk assessment—maintenance decision-making—map update".

[0071] S7. Calculate the crack propagation rate based on the corrected crack instance, and calculate the crack risk level and maintenance priority based on the crack information and crack propagation rate. The crack propagation rate is calculated using the following formula:

[0072] in, For crack propagation rate, This indicates the crack length during the current monitoring period. Δt represents the crack length in the previous monitoring cycle, and Δt represents the time interval between adjacent monitoring cycles.

[0073] Specifically, for crack instances in the same road segment, the same lane, and whose spatial overlap meets the threshold condition, they are considered to belong to the same temporal crack target, and the aforementioned crack propagation rate formula is used to calculate the propagation rate. For example, if the length of a longitudinal crack in the previous monitoring period was 1.20 meters, the length in the current monitoring period is 1.56 meters, and the monitoring interval is 30 days, then its expansion rate can be calculated to be 0.012 meters / day. If this expansion rate exceeds a preset expansion threshold, a new "high expansion risk" status node is added to the map, and corresponding temporal precursor relationships and risk association relationships are established.

[0074] The normalized value of crack propagation rate can be determined by the ratio of the current propagation rate to the preset maximum reference propagation rate, and the upper limit is limited to 1; the normalized value of crack width and the normalized value of crack length are determined by the ratio of the current width and current length to the preset maximum reference width and preset maximum reference length, respectively, and the upper limit is limited to 1.

[0075] The crack risk score is calculated using the following formula:

[0076] in, To score the risk of cracks, This represents the normalized value of the crack width. This represents the normalized value of the crack length. This represents the normalized value of the crack propagation rate. This represents the normalized value of the coupling strength between the environment and the load. This represents the normalized value of structural sensitivity.

[0077] Specifically, the crack width, length, propagation rate, environmental and traffic coupling strength, and structural sensitivity are normalized, and then the risk score R is calculated according to the aforementioned crack risk scoring formula. The normalized value of the environmental and load coupling strength can be obtained by weighting rainfall, freeze-thaw cycles, diurnal temperature range, groundwater seepage impact indicators, traffic volume, and heavy load ratio; the normalized value of structural sensitivity can be determined by road grade, structural layer combination, material type, service life, and maintenance history. When R is less than 0.35, a mild risk is output and crack sealing is recommended; when R is greater than or equal to 0.35 and less than 0.65, a moderate risk is output and sealing and local repairs are recommended; when R is greater than or equal to 0.65, a severe risk is output and milling and repaving, drainage improvement, and structural reinforcement are recommended. Under the same risk level, crack instances with high propagation rates, significant thermal anomalies, high road grades, or a high number of historical recurrences at the same location have higher maintenance priority.

[0078] The final output should include at least a crack distribution map, a crack attribute parameter table, a crack cause explanation chain, risk level results, a maintenance priority list, and early warning information; the map write-back content should include at least crack correction results, risk status results, treatment suggestion results, and retest results.

[0079] Based on the method described in the above embodiments, this embodiment will further describe the road crack monitoring system based on knowledge graph and disaster formation mechanism constraints. The road crack monitoring system based on knowledge graph and disaster formation mechanism constraints can be implemented as an independent entity or integrated into an electronic device. The electronic device can be a terminal, server, or other device. The terminal can include a tablet computer, laptop computer, personal computer (PC), microprocessor box, or other devices.

[0080] Please see Figure 3 , Figure 3 This invention specifically describes a road crack monitoring system based on knowledge graphs and disaster formation mechanisms, applied in electronic devices. The system may include: The multi-source data acquisition module is used to collect road surface visible light image sequences, synchronous infrared thermal images, basic road attribute data, environmental disaster-causing factor data, traffic load data, and spatiotemporal positioning data of the target road section within the current monitoring period; The preprocessing and fusion module is used to preprocess the visible light image sequence and the synchronous infrared thermal image of the road surface, and then fuse the preprocessed visible light image sequence and the preprocessed synchronous infrared thermal image to obtain a four-channel fusion tensor. The crack detection module is used to input the four-channel fused tensor into the trained crack detection network to obtain the predicted crack probability map and crack category map. The candidate semantic fact set generation module is used to extract crack information based on crack probability map and crack category map, and generate candidate semantic fact set based on road basic attribute data, environmental disaster factor data, traffic load data and spatiotemporal positioning data and crack information. The crack formation explanation chain generation module is used to generate crack formation explanation chains based on a set of candidate semantic facts and by calling a pre-built road crack disaster mechanism attribute map. The crack instance correction and map write-back update module is used to correct and assess the risk of crack instances based on mechanism matching degree and mechanism consistency score, and update the road crack disaster mechanism attribute map. The crack risk level calculation and maintenance priority generation module is used to calculate the crack propagation rate based on crack information, and to calculate the crack risk level and maintenance priority based on crack information and crack propagation rate.

[0081] In specific implementation, the above modules and / or units can be implemented as independent entities, or they can be arbitrarily combined and implemented as the same or several entities. For the specific implementation of the above modules and / or units, please refer to the previous method embodiments. For the specific beneficial effects that can be achieved, please also refer to the beneficial effects in the previous method embodiments, which will not be repeated here.

[0082] In addition, this embodiment of the invention also provides an electronic device, which may be a computer, tablet computer, or other similar device. This electronic device can implement the steps of any embodiment of the road crack monitoring method based on knowledge graph and disaster formation mechanism constraints provided in this embodiment of the invention. Therefore, it can achieve the beneficial effects that any road crack monitoring method based on knowledge graph and disaster formation mechanism constraints provided in this embodiment of the invention can achieve, as detailed in the preceding embodiments, and will not be repeated here.

[0083] Figure 4 A specific structural block diagram of an electronic device provided in an embodiment of the present invention is shown. This electronic device can be used to implement the road crack monitoring method based on knowledge graph and disaster formation mechanism constraints provided in the above embodiments. The electronic device 500 can be a terminal, server, or other device. The terminal can include a tablet computer, laptop computer, personal computer (PC), microprocessor box, or other devices.

[0084] The memory 520 can be used to store software programs and modules, such as the program instructions / modules corresponding to those in the above embodiments. The processor 580 executes various functional applications and data processing by running the software programs and modules stored in the memory 520. The memory 520 may include high-speed random access memory, and may also include non-volatile memory, such as one or more magnetic storage devices, flash memory, or other non-volatile solid-state memory. In some instances, the memory 520 may further include memory remotely located relative to the processor 580, and these remote memories can be connected to the electronic device 500 via a network. Examples of such networks include, but are not limited to, the Internet, corporate intranets, local area networks, mobile communication networks, and combinations thereof.

[0085] The input unit 530 can be used to receive input numeric or character information, and to generate a keyboard and mouse related to user settings and function control. Display unit 540 can be used to display information input by the user or information provided to the user, as well as various graphical user interfaces, which can be composed of graphics, text, icons, video, and any combination thereof. Display unit 540 may include display panel 541, which may optionally be configured in the form of LCD (Liquid Crystal Display), OLED (Organic Light-Emitting Diode), or other similar forms.

[0086] Electronic device 500, through transmission module 570 (e.g., Wi-Fi module), can help users receive requests, send information, etc., providing users with wireless broadband internet access. Although transmission module 570 is shown in the figure, it is understood that it is not an essential component of electronic device 500 and can be omitted as needed without changing the essence of the invention.

[0087] The processor 580 is the control center of the electronic device 500. It connects to various parts of the phone via various interfaces and lines, and performs various functions and processes data of the electronic device 500 by running or executing software programs and / or modules stored in the memory 520, and by calling data stored in the memory 520, thereby providing overall monitoring of the electronic device. Optionally, the processor 580 may include one or more processing cores; in some embodiments, the processor 580 may integrate an application processor and a modem processor, wherein the application processor mainly handles the operating system, user interface, and applications, and the modem processor mainly handles wireless communication. It is understood that the modem processor may also not be integrated into the processor 580.

[0088] Electronic device 500 also includes a power supply 590 (such as a battery) that supplies power to various components. In some embodiments, the power supply may be logically connected to processor 580 through a power management system, thereby enabling functions such as charging, discharging, and power consumption management through the power management system. The power supply 590 may also include one or more DC or AC power supplies, recharging systems, power fault detection circuits, power converters or inverters, power status indicators, and other arbitrary components.

[0089] Although not shown, the electronic device 500 also includes cameras (such as front-facing cameras and rear-facing cameras), Bluetooth modules, etc., which will not be described in detail here. Specifically, in this embodiment, the display unit of the electronic device is a touch screen display, and the mobile terminal also includes a memory and one or more programs, wherein one or more programs are stored in the memory and configured to be executed by one or more processors. One or more programs contain instructions for performing the following operations: Collect road surface visible light image sequences, synchronous infrared thermal images, basic road attribute data, environmental disaster-causing factor data, traffic load data, and spatiotemporal positioning data for the target road section within the current monitoring period; The visible light image sequence and the synchronous infrared thermal image of the road surface are preprocessed, and the preprocessed visible light image sequence and the preprocessed synchronous infrared thermal image are fused to obtain a four-channel fused tensor. The four-channel fusion tensor is input into the trained crack detection network to obtain the predicted crack probability map and crack category map. Crack information is extracted based on crack probability maps and crack category maps, and a candidate semantic fact set is generated based on road basic attribute data, environmental disaster-causing factor data, traffic load data, spatiotemporal positioning data, and crack information. Based on the candidate semantic fact set, and by calling the pre-constructed road crack disaster mechanism attribute map, a crack cause explanation chain is generated; Based on the mechanism matching degree and mechanism consistency score, the crack instances were corrected and risk assessments were performed, and the road crack disaster mechanism attribute map was updated. The crack propagation rate is calculated based on the corrected crack instance, and the crack risk level and maintenance priority are calculated based on the crack information and crack propagation rate.

[0090] In practice, the above modules can be implemented as independent entities or combined in any way to be implemented as the same or several entities. For the specific implementation of the above modules, please refer to the previous method implementation examples, which will not be repeated here.

[0091] Those skilled in the art will understand that all or part of the steps in the various methods of the above embodiments can be implemented by instructions, or by instructions controlling related hardware. These instructions can be stored in a computer-readable storage medium and loaded and executed by a processor. Therefore, embodiments of the present invention provide a storage medium storing multiple instructions that can be loaded by a processor to execute the steps of any embodiment of the road crack monitoring method based on knowledge graph and disaster formation mechanism constraints provided by the present invention.

[0092] The computer-readable storage medium may include: read-only memory (ROM), random access memory (RAM), disk or optical disk, etc.

[0093] Since the instructions stored in the storage medium can execute the steps in any embodiment of the road crack monitoring method based on knowledge graph and disaster formation mechanism constraints provided in the embodiments of the present invention, the beneficial effects that any road crack monitoring method based on knowledge graph and disaster formation mechanism constraints provided in the embodiments of the present invention can achieve can be realized. For details, please refer to the previous embodiments, which will not be repeated here.

[0094] The foregoing has provided a detailed description of a road crack monitoring method, system, storage medium, and electronic device based on knowledge graph and disaster formation mechanism constraints provided by embodiments of the present invention. Specific examples have been used to illustrate the principles and implementation methods of the present invention. The descriptions of the above embodiments are only for the purpose of helping to understand the method and core ideas of the present invention. At the same time, for those skilled in the art, there will be changes in the specific implementation methods and application scope based on the ideas of the present invention. Therefore, the content of this specification should not be construed as a limitation of the present invention.

Claims

1. A road crack monitoring method based on a knowledge graph and disaster mechanism constraint, characterized in that, include: Collect road surface visible light image sequences, synchronous infrared thermal images, basic road attribute data, environmental disaster-causing factor data, traffic load data, and spatiotemporal positioning data for the target road section within the current monitoring period; The visible light image sequence and the synchronous infrared thermal image of the road surface are preprocessed, and the preprocessed visible light image sequence and the preprocessed synchronous infrared thermal image are fused to obtain a four-channel fused tensor. The four-channel fusion tensor is input into the trained crack detection network to obtain the predicted crack probability map and crack category map. Crack information is extracted based on crack probability maps and crack category maps, and a candidate semantic fact set is generated based on road basic attribute data, environmental disaster-causing factor data, traffic load data, spatiotemporal positioning data, and crack information. Based on the candidate semantic fact set, and by calling the pre-constructed road crack disaster mechanism attribute map, a crack cause explanation chain is generated; Based on the mechanism matching degree and mechanism consistency score, the crack instances were corrected and risk assessments were performed, and the road crack disaster mechanism attribute map was updated. The crack propagation rate is calculated based on the corrected crack instance, and the crack risk level and maintenance priority are calculated based on the crack information and crack propagation rate.

2. The road crack monitoring method based on a knowledge graph and disaster mechanism constraint according to claim 1, characterized in that, The visible light image sequence and the synchronized infrared thermal image of the road surface are preprocessed, and then the preprocessed visible light image sequence and the preprocessed synchronized infrared thermal image are fused to obtain a four-channel fused tensor, including: Median filtering, contrast-limited adaptive histogram equalization, specular reflection suppression, and shadow correction were sequentially applied to the visible light image sequence of the road surface to obtain the preprocessed visible light image sequence of the road surface. Temperature normalization and geometric registration are performed on the synchronous infrared thermal image to obtain a preprocessed synchronous infrared thermal image. Then, perspective correction is used to unify the visible light image sequence of the road surface and the synchronous infrared thermal image to the same pixel coordinate system. The visible light image sequence of the road surface includes three visible light channels: red, green, and blue. The synchronous infrared thermal image includes a single-channel temperature matrix. The four-channel fusion tensor is obtained by superimposing the three visible light channels (red, green, and blue) and the single-channel temperature matrix.

3. The road crack monitoring method based on a knowledge graph and disaster mechanism constraint according to claim 1, characterized in that, The crack detection network comprises a backbone network, a feature pyramid network, a multi-head self-attention coding layer, and a U-shaped decoding layer; the processing procedure of the crack detection network includes: The four-channel fusion tensor is input into the backbone network for multi-scale feature extraction to obtain shallow edge features and deep semantic features. Shallow edge features and deep semantic features of different scales are input into the feature pyramid network and fused to obtain a multi-scale feature map; The multi-scale feature map is input into a multi-head self-attention coding layer for feature flattening and multi-head self-attention calculation to obtain a global dependency representation. The global dependency representation is input into the U-shaped decoding layer to obtain the crack probability map and crack category map.

4. The road crack monitoring method based on a knowledge graph and disaster mechanism constraint according to claim 1, characterized in that, The crack information includes crack location, crack length, crack width, crack direction, crack connectivity, thermal anomaly intensity, and visual detection confidence level. The candidate semantic fact set is a graph structure, including crack instance nodes, road segment nodes, lane nodes, structural layer nodes, environmental factor nodes, risk state nodes, and edges used to associate each node; the crack instance node corresponds to a crack instance, the road segment node corresponds to a road segment entity, the lane node corresponds to a lane entity, the structural layer node corresponds to a structural layer entity, the environmental factor node corresponds to the environmental factor action condition, and the risk state node corresponds to risk state information. The road basic attribute data is used to determine the road grade, structural layer combination and attributes, material type and service life of the road segment to which the crack instance belongs; the environmental disaster-causing factor data is used to determine the conditions under which environmental factors act. The traffic load data is used to characterize lane information of lane entities; The spatiotemporal positioning data is used to determine the station location, acquisition time, and monitoring cycle number of the crack entity.

5. The road crack monitoring method based on a knowledge graph and disaster mechanism constraint according to claim 4, characterized in that, The road crack disaster mechanism attribute map uses a graph structure to store node attributes and relationship attributes. The node attributes include road entity, crack entity, crack attribute entity, disaster-causing factor entity, structural layer entity, risk state entity, and maintenance decision entity. The relationship attributes include location, belonging to, affected by, induced, aggravated, accompanied by, evolved into, corresponding risk level, recommended treatment measures, historical co-occurrence, and temporal precursor relationship. Based on a set of candidate semantic facts, and by invoking a pre-built attribute map of road crack formation mechanisms, an explanatory chain for crack formation causes is generated, including: Based on the road segment location, lane location, crack type, crack length, crack width, crack direction, thermal anomaly intensity, structural layer combination and attributes, environmental factors, traffic load and historical damage records of crack instances in the candidate semantic fact set, local subgraphs related to the current crack instance are retrieved from the road crack disaster mechanism attribute map. In the local subgraph, multiple preset disaster-causing rule paths are matched, and a chain of explanations for the causes of cracks is generated based on the successfully matched disaster-causing rule paths.

6. The road crack monitoring method based on a knowledge graph and disaster mechanism constraint according to claim 1, characterized in that, Based on mechanism matching degree and mechanism consistency score, crack instances are corrected and risk assessments are conducted, and the road crack disaster mechanism attribute map is updated, including: The computer calculates the matching degree of the mechanism and the consistency score of the mechanism. When the consistency score of the mechanism is less than the preset consistency threshold, the crack instance is removed, the category is corrected or the boundary is corrected. When the consistency score of the mechanism is greater than or equal to the preset consistency threshold, the crack instance is retained, and risk assessment and updating of the road crack disaster mechanism attribute map are performed based on the crack instance after removal of false detections, category correction or boundary correction.

7. The road crack monitoring method based on a knowledge graph and disaster mechanism constraint according to claim 6, characterized in that, The mechanism matching degree is calculated using the following formula: wherein, is a mechanism matching degree, represents a weight of the qth disaster-causing rule, represents a satisfaction flag of the qth disaster-causing rule, m is a total number of disaster-causing rules; The consistency score of the mechanism is calculated using the following formula: in, To score the consistency of the mechanism, Indicates the confidence level of visual detection. This indicates the support level for thermal anomalies.

8. A road crack monitoring system based on a knowledge graph and disaster mechanism constraint, characterized in that, include: The multi-source data acquisition module is used to collect road surface visible light image sequences, synchronous infrared thermal images, basic road attribute data, environmental disaster-causing factor data, traffic load data, and spatiotemporal positioning data of the target road section within the current monitoring period; The preprocessing and fusion module is used to preprocess the visible light image sequence and the synchronous infrared thermal image of the road surface, and then fuse the preprocessed visible light image sequence and the preprocessed synchronous infrared thermal image to obtain a four-channel fusion tensor. The crack detection module is used to input the four-channel fused tensor into the trained crack detection network to obtain the predicted crack probability map and crack category map. The candidate semantic fact set generation module is used to extract crack information based on crack probability map and crack category map, and generate candidate semantic fact set based on road basic attribute data, environmental disaster factor data, traffic load data and spatiotemporal positioning data and crack information. The crack formation explanation chain generation module is used to generate crack formation explanation chains based on a set of candidate semantic facts and by calling a pre-built road crack disaster mechanism attribute map. The crack instance correction and map write-back update module is used to correct and assess the risk of crack instances based on mechanism matching degree and mechanism consistency score, and update the road crack disaster mechanism attribute map. The crack risk level calculation and maintenance priority generation module is used to calculate the crack propagation rate based on crack information, and to calculate the crack risk level and maintenance priority based on crack information and crack propagation rate.

9. A computer-readable storage medium, characterized in that, The computer-readable storage medium stores a plurality of instructions adapted to be loaded by a processor to execute the road crack monitoring method based on knowledge graph and disaster formation mechanism constraints as described in any one of claims 1 to 7.

10. An electronic device, comprising: The method includes a processor and a memory, the processor being electrically connected to the memory, the memory being used to store instructions and data, and the processor being used to execute the steps of the road crack monitoring method based on knowledge graph and disaster formation mechanism constraints as described in any one of claims 1 to 7.

Images