Mountainous road flood resilience margin early warning method based on multiple failure mechanisms

Abstract

The present application relates to a mountainous area highway rainstorm flood resilience margin early warning method based on multiple failure mechanisms, comprising: real-time acquisition of multi-source monitoring data, pre-processing to obtain the extended state vector at the current time; according to the pre-generated label, the resilience critical boundary is established respectively for different failure mechanisms to form a multi-failure mechanism resilience critical boundary model library; the extended state vector is mapped to the multi-failure mechanism resilience critical boundary model library, the mechanism component resilience margin is calculated and the comprehensive resilience margin and the current candidate dominant failure mechanism are obtained based on each mechanism component resilience margin; according to the mechanism component resilience margin and the comprehensive resilience margin, combined with the set early warning condition, the current early warning level and the final dominant failure mechanism are determined; the early warning information and the linkage control instruction corresponding to the current early warning level and the final dominant failure mechanism are output. Compared with the prior art, the present application has the advantages of strong interpretability, high advance, good stability and strong linkage targeting.

Description

Technical Field

[0001] This invention relates to the field of flood early warning, and in particular to an early warning method for the resilience margin of mountain roads during rainstorms and floods based on multiple failure mechanisms. Background Technology

[0002] Mountain roads serve multiple functions, including connecting mountain towns, transporting goods, facilitating tourism, and providing emergency rescue services. Due to the undulating terrain, short runoff paths, rapid surface runoff, high exposure of slopes and roadbeds, and complex bridge and culvert flow conditions, they are prone to disasters such as road surface water accumulation, slope instability, roadbed erosion, bridge and culvert blockage, structural damage, and traffic interruption under the influence of rainstorms and floods.

[0003] Existing technologies primarily follow two technical paths for research and application. First, focusing on pre-disaster scenario analysis, post-disaster assessment, or resilience evaluation, road system resilience is typically characterized based on functional loss-recovery processes, comprehensive scores, or statistical indicators. This approach is suitable for post-disaster summaries, scheme comparisons, or macro-level evaluations, but it usually relies on relatively complete disaster and recovery process information, making it difficult to directly use as an online early warning criterion during a disaster. Second, focusing on real-time monitoring and early warning during the disaster process, alarms are typically triggered by exceeding thresholds for single indicators such as rainfall, water level, displacement, moisture content, and slope safety factors. This approach offers some real-time capability, but it primarily targets local anomaly identification and struggles to comprehensively characterize the coupled evolutionary process between rainfall, water accumulation, slopes, erosion, bridges and culverts, and traffic operations. It also fails to differentiate between the two risk states: "short-term damage but still recoverable" and "approaching irreversible destruction."

[0004] Heavy rain and flooding on mountain roads typically involve multiple concurrent and propagating failure mechanisms. During the same rainfall event, water accumulation may initially reduce traffic capacity, subsequently triggering accelerated slope displacement, roadbed erosion, and bridge / culvert overflow failure. Furthermore, different failure mechanisms may amplify each other. Current technologies rarely model different failure mechanisms separately and map them uniformly into a single online discrimination framework, resulting in insufficient mechanistic targeting and interpretability of early warning results.

[0005] Whether a mountain road enters a high-risk state during a disaster depends not only on the current monitoring values, but also on whether traffic capacity can be restored within a limited time window, whether emergency resources are available, whether alternative routes are available, and whether irreversible damage such as roadbed or bridge and culvert washout has occurred. These factors essentially reflect the repairability constraints of the disaster state. Most existing methods do not incorporate these constraints into the real-time judgment process, thus failing to accurately reflect the actual connotation of resilience. Mountain torrential rain and flooding processes also exhibit significant path dependence and hysteresis characteristics. The same monitoring status may correspond to different risk meanings during the rainfall intensification phase and the receding recovery phase. Ignoring these phase memory characteristics can easily lead to inconsistent warnings, frequent level switching, and misjudgments.

[0006] Therefore, it is necessary to propose a new method for early warning of rainstorms and floods on mountain roads in order to achieve more reliable real-time early warning and coordinated control. Summary of the Invention

[0007] The purpose of this invention is to provide a method for early warning of rainstorm and flood resilience margin on mountain roads based on multiple failure mechanisms to improve the early warning lead time.

[0008] The objective of this invention can be achieved through the following technical solutions: A method for early warning of rainstorm and flood resilience margin of mountain roads based on multiple failure mechanisms includes the following steps: Real-time acquisition of multi-source monitoring data of mountain roads during the evolution of rainstorm and flood disasters, followed by preprocessing, yields the extended state vector at the current moment. ; Based on the labels pre-generated for each failure mechanism, toughness critical boundaries between recoverable and irreversible damage risk states are established for different failure mechanisms. This forms a library of toughness critical boundary models with multiple failure mechanisms, where subscripts... Number the failure mechanism; The extended state vector at the current moment Mapped to a multi-failure mechanism toughness critical boundary model library, the toughness margin of each mechanism component is calculated. Based on the toughness margin of each mechanism component, Obtain the overall toughness margin and the current candidate dominant failure mechanism ; Based on the aforementioned mechanism, the toughness margin is... Overall resilience margin and the current candidate dominant failure mechanism Based on the established warning conditions, determine the current warning level and the final dominant failure mechanism; Output warning information and linkage control commands corresponding to the current warning level and the final dominant failure mechanism.

[0009] Furthermore, the multi-source monitoring data includes at least three types of data from rainfall, water level or water accumulation, slope moisture content or displacement, roadbed scour, bridge and culvert conditions, and traffic operation status.

[0010] Furthermore, the extended state vector It contains a standardized set of state variables. hysteresis memory variable set and the set of restorative context variables ,in, For the first Each monitored state variable at time... Standardized state variables, for Time of the first A hysteresis memory variable, for Time of the first A restorative context variable, subscript To standardize the number of state variables, the subscript... The number of hysteresis variables, subscript The standardized set of state variables represents the number of restorative context variables. Each standardized state variable The expression is: , In the formula, For the first Each monitored state variable at time... The original value, and They represent the first The mean and standard deviation of each monitored state variable in the baseline sample set; The set of hysteresis memory variables Each hysteresis memory variable To characterize the path dependency features during the disaster loading and receding recovery phases, it is updated according to the following formula: , In the formula, As weight, For the previous sampling time, The loading phase function is related to increased rainfall, water accumulation, water content accumulation, accelerated displacement, or increased scour. It is a non-negative truncation function; The set of restorative context variables It should include at least one of the following: estimated repair time, amount of available emergency resources, resource accessibility, availability of alternative routes, cost of detours, and level of importance for maintaining traffic flow.

[0011] Furthermore, the failure mechanism includes at least two of the following: water accumulation causing failure of traffic function, slope instability mechanism, roadbed scouring mechanism, and bridge and culvert blockage or erosion mechanism.

[0012] Furthermore, the toughness critical boundary The steps to obtain it include: Historical disaster samples, mechanism calculation samples, and numerical simulation samples are obtained and uniformly mapped into extended state vector samples. Regarding the first The failure mechanism is calculated. The recovery rate and resource reachability of each extended state vector sample are expressed as follows: , In the formula, For the first Recovery rate of an extended state vector sample To preset the traffic capacity at the time of restoration, Based on pre-disaster baseline traffic capacity, For the first Resource reachability of a sample of extended state vectors. The amount of resources available for use. For the amount of resources required, To prevent positive numbers with a denominator of zero; Based on the aforementioned recovery rate and resource reachability, for the first Each failure mechanism generates a label for repairability constraints, denoted as: , In the formula, For the first Each failure mechanism is labeled with a value of 1, indicating a recoverable sample, and a value of -1, indicating a sample at risk of irreversible damage. The recovery rate threshold, For the duration of the closure, The threshold for continuous closure time. To estimate the repair time, As the expected repair time threshold, For resource accessibility threshold, To replace channel availability, To replace the channel availability threshold, This is an indicator of irreversible damage. Based on the extended state vector samples and their corresponding labels, construct the first... Training sample set for each failure mechanism , is represented as: , In the formula, For the first One extended state vector sample, To expand the number of state vector samples; Based on the training sample set The discriminant model is used to obtain the first... The toughness critical boundary of a failure mechanism .

[0013] Furthermore, the discriminant model includes one of the following: a support vector classifier, a kernel discriminant model, a probability boundary model, and an ensemble discriminant model, or a discriminant model constructed from combinations thereof, wherein when the support vector classifier is used, the first... Discriminant function for each failure mechanism for: , In the formula, For support vector coefficients, For kernel function, For bias terms; The corresponding toughness critical boundary for: .

[0014] Furthermore, the mechanistic component toughness margin The calculation formula is: , In the formula, For the first A discriminant function for a failure mechanism. For the current extended state vector to the th A signed shortest distance function for the toughness critical boundary of a failure mechanism; Among them, when When it is differential, the toughness margin of the mechanistic component The following formula is used for approximation: , In the formula, For the first The gradient of each failure mechanism discrimination function at the current extended state vector; The overall resilience margin Represented as: , In the formula, A set of failure mechanisms For the first Risk weights for each failure mechanism; The current candidate dominant failure mechanism Represented as: .

[0015] Furthermore, the steps for determining the current warning level and the ultimate dominant failure mechanism include: Based on the aforementioned mechanism component toughness margin and / or overall toughness margin Calculate the rate of change of toughness margin and the average rate of change of the sliding window, where the comprehensive toughness margin is used. During the calculation, the calculation expressions are as follows: , , In the formula, The rate of change of toughness margin, subscript The discrete sampling time number. The average rate of change of the sliding window. The length of the sliding window, subscript The sample number within the sliding window; The toughness margin of the aforementioned mechanism component Overall resilience margin The average rate of change of the sliding window and the duration of the warning level are compared with preset grading thresholds and release hysteresis conditions to obtain the current warning level and the final dominant failure mechanism. The final dominant failure mechanism is determined by the candidate dominant failure mechanisms within the duration corresponding to the current warning level. During the continuous period, the current candidate dominant failure mechanism that appears most frequently is determined as the final dominant failure mechanism. If the number of occurrences is the same, the failure mechanism with the smallest toughness margin of the mechanism component at the current moment is taken as the final dominant failure mechanism. The current warning level is determined according to any of the following constraints: (1) When satisfied and continue A yellow alert is triggered during each sampling period, among which... , For the grading threshold, The toughness margin is the mechanistic component corresponding to the final dominant failure mechanism. The yellow grading threshold corresponds to the final dominant failure mechanism; (2) When satisfied and continue An orange alert is triggered during each sampling period, among which... , For the grading threshold, The orange-level threshold corresponds to the final dominant failure mechanism; (3) When satisfied and continue A red alert is triggered during each sampling period, among which... , For the grading threshold, The red grading threshold corresponds to the final dominant failure mechanism; in, , All grading thresholds are jointly calibrated based on the statistical quantile values ​​of historical event samples, the early warning lead time target, the false alarm rate constraint, the missed alarm rate constraint, and the important traffic maintenance level. When the release hysteresis condition is met, it indicates that the warning can be lifted. The release hysteresis condition is as follows: and And continue Each sampling period, of which To allow for hysteresis, Release hysteresis for the rate of change.

[0016] Furthermore, the multi-failure mechanism toughness critical boundary model library is periodically modified based on newly added measured disaster samples. The modification methods include at least one of parameter retraining, boundary translation correction, threshold recalibration, and sample reweighting.

[0017] Furthermore, the linkage control command is generated based on the final dominant failure mechanism, and the specific steps are as follows: When the ultimate dominant failure mechanism is water accumulation leading to failure of the passage function, output speed limit, drainage, channelization detour or temporary road closure commands; When the ultimate dominant failure mechanism is slope instability, output instructions for slope inspection, danger zone control, advance deployment of emergency equipment, or personnel evacuation. When the ultimate dominant failure mechanism is the roadbed scour mechanism, output instructions for bridge approach slab inspection, scour protection, closure control, and emergency reinforcement. When the ultimate dominant failure mechanism is bridge or culvert blockage or erosion, output commands for dredging and clearing blockages, increasing bridge and culvert monitoring, detour diversion, and structural emergency repair.

[0018] Compared with the prior art, the present invention has the following beneficial effects: (1) By setting labels, this invention directly addresses the core issue of "whether it is still recoverable" in resilience determination, which can more accurately distinguish between recoverable attenuation and irreversible damage risk. Furthermore, by mapping the current real-time state to the mechanism component resilience margin and the comprehensive resilience margin, resilience is transformed from a post-disaster evaluation indicator into a continuously updated online discrimination quantity during a disaster, thereby improving the advance warning.

[0019] (2) Based on the repairability constraint, this invention introduces a multi-failure mechanism toughness critical boundary to address multiple failure mechanisms such as water accumulation, slope instability, roadbed scour, and bridge and culvert blockage or erosion. It can output the dominant failure mechanism, improving the interpretability of early warning results and the targeted nature of response. Simultaneously, it transforms the positional relationship between the current real-time state point and the boundary into a signed mechanistic component toughness margin. Thus, a comprehensive resilience margin is obtained through conservative integration. This transforms the resilience characterization of mountain roads during rainstorms and floods from a "post-event result" to an "online discriminant that can be continuously updated during the event," solving the problem of strong post-event dependence in existing resilience assessments.

[0020] (3) Compared with sending multi-source data directly into a general risk classifier, this invention further explicitly embeds the label definition of "whether the disaster is still within the recoverable domain" through repairability constraints, and identifies the dominant failure mechanism to correspond the early warning results with specific disposal measures, so that the early warning results not only have a level, but also explain "why it is dangerous", "what mechanism the main danger comes from" and "what actions should be taken".

[0021] (4) This invention incorporates multiple failure mechanisms, repairability constraints and path dependence features into a unified online discrimination framework, establishes a critical discrimination mechanism of "recoverable state - irreversible damage risk state", thereby achieving more reliable real-time early warning and linkage control. Furthermore, by transforming "recoverable state - irreversible damage risk state" into a continuously updatable online discrimination margin, it has the advantages of strong interpretability, high lead time, good stability and strong linkage targeting.

[0022] (5) This invention reduces the early warning jitter caused by single-moment fluctuations and improves early warning stability by introducing hysteresis memory variables, sliding window average rate of change, duration period, and release hysteresis conditions. It also distinguishes between the disaster loading stage and the recovery stage by using hysteresis memory variables and release hysteresis conditions, and uses the sliding window average rate of change and duration period to reduce the impact of instantaneous fluctuations on the early warning results, thereby improving early warning stability.

[0023] (6) This invention transforms the traditional resilience assessment that relies on the post-disaster recovery process into real-time judgment during the disaster through the overall mechanism of "multiple failure mechanism critical boundary + repairability constraint label + path-dependent hysteresis memory + comprehensive resilience margin evolution analysis", and further couples it with engineering linkage measures, thus making it more suitable for practical early warning applications in mountainous highway rainstorm and flood scenarios. Attached Figure Description

[0024] Figure 1 This is a schematic diagram of the method flow of the present invention; Figure 2 This is a schematic diagram of the critical boundary and real-time toughness margin of multiple failure mechanisms under the state-space dimensionality reduction projection of the present invention. Figure 3 This is a schematic diagram illustrating the combined triggering of the comprehensive toughness margin and graded early warning system of the present invention; Figure 4 This is a schematic diagram illustrating the repairability constraint label generation and boundary correction of the present invention; Figure 5 This is a system architecture diagram of the perception layer, analysis and decision-making layer, and application linkage layer of the present invention. Detailed Implementation

[0025] The present invention will now be described in detail with reference to the accompanying drawings and specific embodiments. These embodiments are based on the technical solution of the present invention and provide detailed implementation methods and specific operating procedures. However, the scope of protection of the present invention is not limited to the following embodiments.

[0026] Example 1 This embodiment provides a method for early warning of rainstorm and flood resilience margin of mountain roads based on multiple failure mechanisms. The method, as follows... Figure 1 As shown, the method includes the following steps: S1. Real-time multi-source monitoring data acquisition.

[0027] In this embodiment, the main failure mechanisms of mountain roads due to rainstorms and floods are denoted as follows: , in, A set of failure mechanisms This explains the mechanism by which water accumulation causes the passage function to fail. This describes the slope instability mechanism. This describes the mechanism of roadbed scour. This indicates the mechanism of bridge or culvert blockage or damage. In actual deployment, some mechanisms can be removed or additional mechanisms such as debris flow impact and rockfall obstruction can be added based on the characteristics of the road section.

[0028] Collect multi-source monitoring data on mountain roads during the evolution of rainstorm and flood disasters. The multi-source monitoring data should include at least three types of data from the following categories: rainfall, water level or water accumulation, slope moisture content or displacement, roadbed scour, bridge and culvert conditions, and traffic operation status. Recorded as: , in, It can be any one or more of the following: real-time rainfall intensity, previous cumulative rainfall, road surface water depth, water level, slope moisture content, displacement rate, pore water pressure, scour depth, bridge and culvert flow ratio, traffic flow, average vehicle speed, and capacity attenuation rate.

[0029] The multi-source monitoring data comes from at least one of the following: rain gauges, water level gauges, water depth sensors, slope moisture sensors, displacement gauges, pore water pressure gauges, scour monitoring devices, bridge and culvert health monitoring devices, video recognition equipment, drone inspection equipment, meteorological radar data interfaces, traffic detectors, and manual inspection terminals.

[0030] S2, Data preprocessing and extended state vector construction.

[0031] This step performs time synchronization, anomaly removal, missing data completion, and dimensional unification on the aforementioned multi-source monitoring data, constructing an extended state vector that includes standardized state variables, hysteresis memory variables, and restorative context variables. .

[0032] During data preprocessing, it is preferable to first synchronize the data from different sampling frequencies using a unified clock, and then use anomaly removal and missing data completion algorithms to correct abrupt changes and missing data points. The anomaly removal algorithm employs... The criteria, Hampel filtering, median filtering, or a combination thereof are used for processing, while missing data completion is handled by nearest neighbor completion, moving mean completion, Kalman filtering completion, or a combination thereof.

[0033] To eliminate the influence of different dimensions, the original state variables are standardized: , In the formula, For at any time No. A standardized state variable, and They represent the first The mean and standard deviation of a monitored state variable in a baseline sample set, which can consist of historical monitoring samples, simulated samples, or a mixture of both.

[0034] Considering the path-dependent characteristics of torrential rain and flooding processes in mountainous areas, this embodiment introduces hysteretic memory variables in addition to the standardized state variables. These variables are used to characterize the path-dependent characteristics between the disaster loading phase and the flood recovery phase, and are updated according to the following formula: , in, for Time of the first A hysteresis memory variable, It is preferably used to characterize loading behaviors such as increased rainfall, water accumulation, water content accumulation, accelerated displacement, or scour expansion. For the previous sampling time, It is a non-negative truncation function; when the rainfall stops and the water recedes, the hysteresis memory variable does not instantly return to zero, but gradually decays according to the recursive relationship, thus representing the path dependence characteristic that "the same state point has different meanings at different evolutionary stages".

[0035] In addition, to link toughness assessment with engineering repairability, this embodiment further introduces a repairability context variable: , in, This is a set of restorative context variables, which includes at least one of the following: estimated restoration time, available resources, resource reachability, alternative route availability, detour cost, and critical connectivity level. for Time of the first A restorative context variable, subscript This represents the number of restorative context variables. Therefore, the extended state vector at the current time step... The construction is as follows: , In the formula, For a standardized set of state variables, For the set of hysteresis memory variables, For a set of restorative context variables, subscript To standardize the number of state variables, the subscript... The number of hysteresis variables, subscript This represents the number of restorative context variables.

[0036] S3. Generation of recoverability constraint labels and construction of critical boundary for toughness in multiple failure mechanisms.

[0037] To construct the training sample set, it is preferable to uniformly map historical disaster samples, mechanism calculation samples, and numerical simulation samples into extended state vector samples. Historical disaster samples are derived from existing rainstorm monitoring data and post-disaster verification results; mechanism calculation samples are derived from mechanism models such as slope stability analysis, bridge and culvert flow analysis, and scour calculation; numerical simulation samples are derived from numerical simulations such as two-dimensional water accumulation evolution, finite element slope deformation, and scour development simulation.

[0038] The generation of this tag is as follows Figure 4 As shown: Regarding the first The first failure mechanism, the... The recovery rate and resource reachability optimization of each extended state vector sample are defined as follows: , in, For the first Recovery rate of an extended state vector sample Indicates the baseline traffic capacity before the disaster. This indicates the traffic capacity at the preset recovery time. For the first Resource reachability of a sample of extended state vectors. Indicates the amount of resources available for use. Indicates the amount of resources required. To prevent extremely small positive numbers with a denominator of zero.

[0039] No. The repairability constraint label corresponding to each failure mechanism is generated by the following formula: , in, For the first Each failure mechanism is labeled with a value of 1, indicating a recoverable sample, and a value of -1, indicating a sample at risk of irreversible damage. Indicates the duration of the closure. Indicates the estimated repair time. Indicates the availability of alternative channels. Indicates an irreversible damage indicator. The recovery rate threshold, The threshold for continuous closure time. As the expected repair time threshold, For resource accessibility threshold, To replace the access availability threshold; when situations such as roadbed erosion, overall slope instability, bridge and culvert structure erosion, damage to key components, or continuous closure exceeding the threshold occur, it is preferable to set... .

[0040] Based on the above tags, construct the first... Training sample set for each failure mechanism : , In the formula, For the first One extended state vector sample, To expand the number of state vector samples.

[0041] This embodiment can use support vector classifiers, kernel discriminant models, probabilistic boundary models, ensemble discriminant models, or combinations thereof to establish the resilience critical boundary. Among these, a support vector classifier is preferred. When using a support vector classifier, the first... The discriminant function for each failure mechanism is: , in, For support vector coefficients, For kernel function, This is the bias term. Therefore, we obtain the... The toughness critical boundary of a failure mechanism : .

[0042] Preferably, the kernel function used is a radial basis function kernel, and different weights are assigned to samples from different sources during training: higher weights are assigned to historical verification samples, medium weights to simulated samples, and weights corresponding to the credibility of critical samples derived from the mechanism. This strategy can maintain the engineering interpretability of the boundary when historical samples are insufficient, while avoiding the bias caused by relying entirely on simulated samples.

[0043] The above yields the first The toughness critical boundary of a failure mechanism Further, a multi-failure mechanism toughness critical boundary model library is formed. This toughness critical boundary model library is periodically modified based on newly added measured disaster samples. The modification methods include at least one of parameter retraining, boundary translation correction, threshold recalibration, and sample reweighting.

[0044] S4. Real-time state mapping and calculation of toughness margin of mechanistic components.

[0045] This step will use the current extended state vector Mapped to each toughness critical boundary in the multi-failure mechanism toughness critical boundary model library The mechanistic component toughness margin was obtained. : , In the formula, For the first A discriminant function for a failure mechanism. For the current extended state vector to the th A signed shortest distance function for the toughness critical boundary of a failure mechanism; when When differentiable, the following formula is preferred for fast approximation: , In the formula, For the first The gradient of each failure mechanism discrimination function at the current extended state vector; when When, it indicates that the current state is relative to the first state. The failure mechanism is still within the recoverable domain; when When, it indicates that the current state is at the th position. At a critical boundary; when When this occurs, it indicates that the failure mechanism has entered the risk domain of irreversible damage.

[0046] To ensure the conservatism of the overall early warning, this embodiment also obtains a comprehensive toughness margin based on the toughness margin of each mechanism component. : , in, For the first The risk weights for each failure mechanism can be determined based on the importance of the road segment, the hazard of the mechanism, historical susceptibility, and the need to maintain traffic flow. The resilience margin of the mechanism component corresponding to each failure mechanism is uniformly weighted with the risk weight, and the failure mechanism with the most unfavorable risk index after weighting is selected as the current candidate dominant failure mechanism. : .

[0047] The current candidate dominant failure mechanism is used for real-time preliminary judgment.

[0048] like Figure 2 The diagram shows the critical boundary and real-time toughness margin of multiple failure mechanisms under the state space dimensionality reduction projection. The diagram shows the recoverable domain, the irreversible damage risk domain, the critical boundary of each failure mechanism, the current state point, and the signed shortest distance from the state point to the corresponding critical boundary.

[0049] S5, Margin Evolution Analysis and Joint Trigger Determination.

[0050] This step first considers the overall resilience margin. and / or mechanism component toughness margin Calculate the rate of change of toughness margin and the average rate of change of the sliding window, obtain the duration information of the corresponding warning level, and then combine the toughness margin... Mechanistic component toughness margin The average rate of change and duration of the sliding window are compared with preset grading thresholds and release hysteresis conditions to determine the current warning level and the final dominant failure mechanism. Finally, warning information and linkage control commands corresponding to the current warning level and the final dominant failure mechanism are output. To avoid frequent switching of the dominant failure mechanism due to fluctuations at a single sampling time, this embodiment first determines the current candidate dominant failure mechanism, and then confirms the dominant mechanism by combining the duration of the corresponding warning level. Within the duration, if the same current candidate dominant failure mechanism appears most frequently, it is determined as the final dominant failure mechanism; if the number of occurrences is the same, the failure mechanism with the smallest resilience margin of the mechanism component at the current time is taken as the final dominant failure mechanism. This final dominant failure mechanism is used for warning issuance decisions. Specifically: When using comprehensive toughness margin In the calculation, to characterize the approximation rate of the overall toughness margin, the rate of change of the overall toughness margin and the average rate of change of the sliding window are defined as follows: , , In the formula, Indicates the length of the sliding window. The rate of change of toughness margin, subscript The discrete sampling time number. The average rate of change of the sliding window, subscript The sample number is within the sliding window.

[0051] By and When used in combination, it can issue early warnings based on the rapid deterioration trend of the overall resilience margin before it falls below a lower threshold.

[0052] The above warning level is determined according to any of the following conditions: When satisfied and continue A yellow alert is triggered during each sampling period, among which... , For the grading threshold, The toughness margin is the mechanistic component corresponding to the final dominant failure mechanism. The yellow grading threshold corresponds to the final dominant failure mechanism; When satisfied and continue An orange alert is triggered during each sampling period, among which... , For the grading threshold, The orange-level threshold corresponds to the final dominant failure mechanism; When satisfied and continue A red alert is triggered during each sampling period. , For the grading threshold, The red grading threshold corresponds to the final dominant failure mechanism; , .

[0053] To reduce warning jitter, this embodiment sets a release hysteresis condition: only when and And continue The current warning will be lifted only after one sampling period. To allow for hysteresis, This strategy releases hysteresis in the rate of change. It avoids frequent upgrades and downgrades due to short-term data bounces.

[0054] When a higher-level warning meets the release lag conditions but still meets the triggering conditions for a lower-level warning, the current warning level is downgraded to the corresponding lower-level warning; the warning is lifted only when no warning triggering conditions are met.

[0055] like Figure 3 The diagram shown illustrates the joint triggering of comprehensive resilience margin and tiered early warning. The diagram displays the evolution curve of comprehensive resilience margin over time, the yellow warning threshold, the orange warning threshold, the red warning threshold, and the sliding window average rate of change in the determination process for joint triggering of early warning.

[0056] Preferably, the above-mentioned classification thresholds and risk weights are jointly determined based on the statistical quantile values ​​of historical event samples, the early warning lead time target, the false alarm rate constraint, the missed alarm rate constraint, and the important traffic maintenance level.

[0057] The aforementioned linkage control commands are generated based on the ultimate dominant failure mechanism: when the ultimate dominant failure mechanism is water accumulation leading to traffic failure, commands for speed limit, drainage, channelization detour, or temporary road closure are output; when the ultimate dominant failure mechanism is slope instability, commands for slope inspection, danger zone closure, pre-positioning of emergency equipment, or personnel evacuation are output; when the ultimate dominant failure mechanism is roadbed scour, commands for bridge approach slab inspection, scour protection, closure control, and emergency reinforcement are output; when the ultimate dominant failure mechanism is bridge or culvert blockage or erosion, commands for dredging and unblocking, increased bridge and culvert monitoring, detour diversion, and structural repair are output.

[0058] After the actual disaster event has ended and post-disaster verification has been completed, newly added measured samples can be reinjected into the training sample set to perform periodic corrections on the multi-failure mechanism toughness critical boundary model library, including adjustments to... and The system involves retraining, recalibrating thresholds, and updating weights for different sample sources. As a result, the multi-failure-mechanism resilience critical boundary model library can gradually improve its adaptability as the road section environment changes and new disaster cases accumulate.

[0059] This embodiment uses the above method to achieve real-time early warning for multiple failure mechanisms in real mountain road sections. Taking a certain mountain road section as an example, this section runs along a river valley, with some sections featuring steep slopes, roadbeds near water, and multiple bridges and culverts. Historically, it has experienced multiple incidents of flooding, landslides, and erosion caused by torrential rains. Five automatic rain gauges, eight water level / flooding monitoring points, twelve slope moisture content / displacement monitoring points, four erosion monitoring points, three bridge and culvert status monitoring points, and four traffic operation monitoring points are deployed along the route, with a preferred sampling period of 5 minutes.

[0060] A primitive state vector is constructed using 12 basic state variables: real-time rainfall intensity, cumulative rainfall in the previous 1 hour, cumulative rainfall in the previous 6 hours, road surface water depth, ditch water level, slope moisture content, slope displacement rate, pore water pressure, roadbed scour depth, bridge-culvert flow ratio, traffic flow, and average vehicle speed. Four hysteretic memory variables are further constructed based on the formula (path-dependent update formula) to represent cumulative rainfall memory, water receding lag, displacement acceleration memory, and scour expansion memory, respectively. Simultaneously, three restorative context variables are introduced: estimated repair time, resource accessibility, and alternative route availability, forming a 19-dimensional extended state vector.

[0061] Regarding historical samples, 86 effective rainfall events and 17 disaster verification events from the past 10 years were collected for this road section; regarding mechanistic calculation samples, an infinite slope stability analysis model was used to supplement the sample for key slopes, and its safety factor can be expressed as: , in, For effective cohesion, The soil weight, The depth of the sliding surface. For the slope angle, Pore ​​water pressure, To determine the effective internal friction angle; in terms of numerical simulation samples, two-dimensional simulations of water accumulation, scouring, and flow capacity were carried out on typical water accumulation points, water-adjacent roadbeds, and bridge and culvert cross sections, ultimately forming 1620 mechanism sample windows.

[0062] When generating tags, it is preferable to set a recovery rate threshold for the slope instability mechanism. Continuous closure threshold Expected repair time threshold Resource accessibility threshold Alternative channel availability threshold For the subgrade scour mechanism, it is preferable to set stricter thresholds for continuous closure and repair time. Discriminant functions are trained for each of the four failure mechanisms. , , and .

[0063] After the warning threshold is calibrated, the comprehensive resilience margin threshold is preferably set. , , rate of change threshold , , Continuous period , , .

[0064] During a typical rainstorm, the overall resilience margin at 18:20 was... It is still above the yellow margin threshold, but the average rate of change of the sliding window meets the requirements. Furthermore, the situation continued to deteriorate for two consecutive sampling cycles, thus triggering a yellow alert in advance; at this point, the dominant failure mechanism was identified as slope instability mechanism.

[0065] At 19:05, the slope moisture content and displacement rate continued to rise, resulting in... , The conditions for an orange alert are met, and an alert message is output stating "The risk of instability of the high slope on the right side is increasing. Please limit speed and pre-deploy emergency equipment." At the same time, instructions for slope inspection and equipment deployment are sent to the maintenance management platform.

[0066] At 19:40, the margin of the slope mechanism component further decreased. The overall resilience margin fell below the red threshold, accompanied by a continuous increase in displacement acceleration, triggering a red alert and implementing coordinated measures such as lockdown, vehicle guidance in dangerous areas, emergency resource allocation, and downstream alert. Subsequently, as rainfall weakened and emergency measures were implemented, the overall resilience margin gradually recovered, but the system only lifted the alert after the release hysteresis conditions were met, avoiding frequent upgrades and downgrades caused by short-term rebounds.

[0067] This embodiment also utilizes the aforementioned method to provide early warnings for real-world scenarios when historical samples are insufficient. For a newly constructed mountainous highway test section, the historical disaster sample only includes 18 rainstorm events, and the post-disaster verification samples are limited. Directly training the boundary model makes it difficult to stably distinguish between recoverable and irreversible risk states. To address this, a mechanistic model and a numerical simulation model are first constructed using terrain, drainage facilities, and structural parameters.

[0068] To address the mechanism of traffic failure caused by water accumulation, a two-dimensional water accumulation evolution model is used to output the water depth, flow velocity, and duration of water accumulation on the road surface. For the slope instability mechanism, slope stability analysis under different rainfall patterns and soil parameter combinations is used to output the safety factor, displacement, and pore water pressure. For the roadbed scour mechanism, a scour development model is used to output the shoulder scour depth and propagation rate. For the bridge and culvert blockage or erosion mechanism, a combination of bridge and culvert flow capacity and blockage scenarios is used to extrapolate the flow ratio and overtopping probability. After parameter perturbation, a total of 1200 simulation sample windows are generated.

[0069] A training set is constructed by combining a limited number of measured verification samples with simulated samples, and the measured verification samples are assigned higher weights. After the initial training of the boundary model, when a new real disaster event occurs and verification is completed, the new event samples are weighted according to... Figure 4 The illustrated process recalculates the repairability label and achieves online iterative updates through boundary retraining and threshold correction. After three event updates, the model significantly increases the separation interval between recoverable samples and irreversible risk samples, reduces the false alarm rate, and the boundary position is closer to actual engineering experience.

[0070] In summary, this embodiment addresses risks such as waterlogging failure, slope instability, roadbed scour, and bridge / culvert blockage or erosion. It acquires multi-source monitoring data on rainfall, water accumulation, water level, displacement, pore water pressure, scour, bridge / culvert status, and traffic operation to construct an extended state vector. Based on historical disaster data, mechanism calculations, and numerical simulation samples, it constructs repairability constraint labels by combining recovery rate, closure duration, estimated repair time, resource accessibility, and irreversible damage indicators, establishing a toughness critical boundary for the failure mechanism. The state is mapped to the toughness critical boundary, and the theoretical component toughness margin and comprehensive toughness margin are calculated to identify the dominant failure mechanism. Yellow, orange, and red early warnings are determined by combining the duration period and release hysteresis conditions, outputting coordinated control commands for speed limits, diversion, closure, drainage, dredging, and emergency response. The core of this method lies in introducing a multi-failure mechanism toughness critical boundary based on repairability constraints. The positional relationship between the current real-time state point and the boundary is transformed into a signed mechanistic component toughness margin. Thus, a comprehensive resilience margin is obtained through conservative integration. This transforms the resilience characterization of mountain roads during torrential rains and floods from a "post-event outcome" to a "continuously updatable online metric during the event." Compared to directly feeding multi-source data into a general risk classifier, this method further explicitly embeds the definition of "whether the disaster is still within the recoverable domain" into a label through repairability constraints. It also identifies the dominant failure mechanism to correlate early warning results with specific response measures, ensuring that early warning results not only have levels but also explain "why it is dangerous," "which mechanism the main danger originates from," and "what actions should be taken." Furthermore, this method distinguishes between the disaster loading and recovery phases through hysteresis memory variables and hysteresis release conditions, and utilizes the sliding window average rate of change and duration period to reduce the impact of instantaneous fluctuations on early warning results, thereby improving early warning stability.

[0071] Example 2 This embodiment provides an early warning system for the resilience margin of mountain roads during rainstorms and floods based on multiple failure mechanisms, such as... Figure 5 As shown, the system adopts a collaborative architecture between the edge and the central end, including the edge end, the central end, and the application linkage layer.

[0072] The edge terminal is equipped with monitoring units for rainfall, water level, and water accumulation; monitoring units for slope moisture content and displacement; monitoring units for scour and bridges and culverts; monitoring units for traffic operation, video, and inspection; and an edge gateway. Each unit is used to collect multi-source monitoring data of mountain roads in real time during the evolution of rainstorm and flood disasters. The edge gateway includes a communication interface, a processor, and a memory. The processor is used to receive multi-source monitoring data uploaded by monitoring terminals along the route, perform time synchronization, anomaly removal, missing data completion, and initial standardization processing, and upload the processing results to the central terminal to execute the method described in Example 1.

[0073] When the communication link is normal, the edge device periodically uploads the preprocessed extended state vector or its intermediate results. After the central device generates a comprehensive early warning level and the dominant failure mechanism, it synchronously pushes the early warning information to VMS, broadcast, APP, vehicle-road cooperative platform and emergency dispatch system. When the communication link is restricted, the edge device can execute local primary early warning based on the most recently issued threshold and boundary simplification parameters to ensure the continuous operation capability of the system.

[0074] The central unit is responsible for expanding state vector construction, maintaining the boundary model library, calculating real-time resilience margin, identifying the dominant mechanism, making hierarchical early warning decisions, and generating coordinated control.

[0075] The application linkage layer is used to execute linkage control commands output from the central terminal. For different dominant failure mechanisms, the linkage strategy engine automatically calls the corresponding measure template. For example, when the dominant failure mechanism is water accumulation leading to traffic failure, the preferred linkage strategy is speed limit, opening drainage facilities, and temporary channelization; when the dominant failure mechanism is bridge or culvert blockage or erosion, the preferred linkage strategy is dredging, increased monitoring of bridge and culvert sections, upstream water inflow warning, and detour diversion; when the dominant failure mechanism is roadbed scour, the preferred linkage strategy is closing threatened lanes, reinforcement and emergency repair, and inspection of bridge approach slabs. Through the automatic mapping of dominant mechanisms and measures, the burden of manual judgment on complex disaster situations for dispatchers can be reduced.

[0076] The rest are as in Example 1.

[0077] In summary, this embodiment transforms the traditional resilience assessment that relies on the post-disaster recovery process into real-time assessment during a disaster through an overall mechanism of "multiple failure mechanism critical boundaries + repairability constraint labels + path-dependent hysteresis memory + comprehensive resilience margin evolution analysis". Furthermore, it is coupled with engineering linkage measures, making it more suitable for practical early warning applications in mountainous highway rainstorm and flood scenarios.

[0078] If the aforementioned functions are implemented as software functional units and sold or used as independent products, they can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of this invention, or the part that contributes to the prior art, or a part of the technical solution, can be embodied in the form of a software product. This computer software product is stored in a storage medium and includes several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) to execute all or part of the steps of the methods described in the various embodiments of this invention. The aforementioned storage medium includes various media capable of storing program code, such as USB flash drives, portable hard drives, read-only memory (ROM), random access memory (RAM), magnetic disks, or optical disks.

[0079] Those skilled in the art will understand that embodiments of the present invention can be provided as methods, systems, or computer program products. Therefore, the present invention can take the form of a completely hardware embodiment, a completely software embodiment, or an embodiment combining software and hardware aspects. Furthermore, the present invention can take the form of a computer program product implemented on one or more computer-usable storage media (including but not limited to disk storage, CD-ROM, optical storage, etc.) containing computer-usable program code. The solutions in the embodiments of the present invention can be implemented using various computer languages, such as the object-oriented programming language Java and the interpreted scripting language JavaScript.

[0080] This invention is described with reference to flowchart illustrations and / or block diagrams of methods, apparatus (systems), and computer program products according to embodiments of the invention. It will be understood that each block of the flowchart illustrations and / or block diagrams, and combinations of blocks in the flowchart illustrations and / or block diagrams, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, special-purpose computer, embedded processor, or other programmable data processing apparatus to produce a machine, such that the instructions, which execute via the processor of the computer or other programmable data processing apparatus, generate instructions for implementing the flowchart illustrations and / or block diagrams. Figure 1 One or more processes and / or boxes Figure 1 A device that provides the functions specified in one or more boxes.

[0081] These computer program instructions may also be stored in a computer-readable storage medium that can direct a computer or other programmable data processing device to function in a particular manner, such that the instructions stored in the computer-readable storage medium produce an article of manufacture including instruction means, which are implemented in a process Figure 1 One or more processes and / or boxes Figure 1 The function specified in one or more boxes.

[0082] These computer program instructions may also be loaded onto a computer or other programmable data processing equipment to cause a series of operational steps to be performed on the computer or other programmable equipment to produce a computer-implemented process, thereby providing instructions that execute on the computer or other programmable equipment for implementing the process. Figure 1 One or more processes and / or boxes Figure 1 The steps of the function specified in one or more boxes.

[0083] Although preferred embodiments of the invention have been described, those skilled in the art, upon learning the basic inventive concept, can make other changes and modifications to these embodiments. Therefore, the appended claims are intended to be interpreted as including the preferred embodiments as well as all changes and modifications falling within the scope of the invention.

[0084] Obviously, those skilled in the art can make various modifications and variations to this invention without departing from its spirit and scope. Therefore, if these modifications and variations fall within the scope of the claims of this invention and their equivalents, this invention also intends to include these modifications and variations.

Claims

1. A method for early warning of rainstorm and flood resilience margin of mountain roads based on multiple failure mechanisms, characterized in that, Includes the following steps: Real-time acquisition of multi-source monitoring data of mountain roads during the evolution of rainstorm and flood disasters, followed by preprocessing, yields the extended state vector at the current moment. ; Based on the labels pre-generated for each failure mechanism, toughness critical boundaries between recoverable and irreversible damage risk states are established for different failure mechanisms. This forms a library of toughness critical boundary models with multiple failure mechanisms, where subscripts... The failure mechanism is numbered, and the failure mechanism includes at least two of the following: water accumulation causing failure of traffic function, slope instability mechanism, roadbed scour mechanism, and bridge and culvert blockage or erosion mechanism. The extended state vector at the current moment Mapped to a multi-failure mechanism toughness critical boundary model library, the toughness margin of each mechanism component is calculated. Based on the toughness margin of each mechanism component, Obtain the overall toughness margin and the current candidate dominant failure mechanism The mechanistic component toughness margin The calculation formula is: ， In the formula, For the first A discriminant function for a failure mechanism. For the current extended state vector to the th A signed shortest distance function for the toughness critical boundary of a failure mechanism; Among them, when When it is differential, the toughness margin of the mechanistic component The following formula is used for approximation: ， In the formula, For the first The gradient of each failure mechanism discrimination function at the current extended state vector; The overall resilience margin Represented as: ， In the formula, A set of failure mechanisms For the first Risk weights for each failure mechanism; The current candidate dominant failure mechanism Represented as: ； Based on the aforementioned mechanism, the toughness margin is... Overall resilience margin and the current candidate dominant failure mechanism Based on the established warning conditions, determine the current warning level and the final dominant failure mechanism; Output warning information and linkage control commands corresponding to the current warning level and the final dominant failure mechanism.

2. The method for early warning of rainstorm and flood resilience margin of mountain roads based on multiple failure mechanisms according to claim 1, characterized in that, The multi-source monitoring data includes at least three types of data: rainfall, water level or water accumulation, slope moisture content or displacement, roadbed scour, bridge and culvert conditions, and traffic operation status.

3. The method for early warning of rainstorm and flood resilience margin of mountain roads based on multiple failure mechanisms according to claim 1, characterized in that, The extended state vector It contains a standardized set of state variables. hysteresis memory variable set and the set of restorative context variables ,in, For the first Each monitored state variable at time... Standardized state variables, for Time of the first A hysteresis memory variable, for Time of the first A restorative context variable, subscript To standardize the number of state variables, the subscript... The number of hysteresis variables, subscript The standardized set of state variables represents the number of restorative context variables. Each standardized state variable The expression is: ， In the formula, For the first Each monitored state variable at time... The original value, and They represent the first The mean and standard deviation of each monitored state variable in the baseline sample set; The set of hysteresis memory variables Each hysteresis memory variable To characterize the path dependency features during the disaster loading and receding recovery phases, it is updated according to the following formula: ， In the formula, As weight, For the previous sampling time, The loading phase function is related to increased rainfall, water accumulation, water content accumulation, accelerated displacement, or increased scour. It is a non-negative truncation function; The set of restorative context variables It should include at least one of the following: estimated repair time, amount of available emergency resources, resource accessibility, availability of alternative routes, cost of detours, and level of importance for maintaining traffic flow.

4. The method for early warning of rainstorm and flood resilience margin of mountain roads based on multiple failure mechanisms according to claim 1, characterized in that, The toughness critical boundary The steps to obtain it include: Historical disaster samples, mechanism calculation samples, and numerical simulation samples are obtained and uniformly mapped into extended state vector samples. Regarding the first The failure mechanism is calculated. The recovery rate and resource reachability of each extended state vector sample are expressed as follows: ， In the formula, For the first Recovery rate of an extended state vector sample To preset the traffic capacity at the time of restoration, Based on pre-disaster baseline traffic capacity, For the first Resource reachability of an extended state vector sample The amount of resources available for use. For the amount of resources required, To prevent positive numbers with a denominator of zero; Based on the aforementioned recovery rate and resource reachability, for the first Each failure mechanism generates a label for repairability constraints, denoted as: ， In the formula, For the first Each failure mechanism is labeled with a value of 1, indicating a recoverable sample, and a value of -1, indicating a sample at risk of irreversible damage. The recovery rate threshold, For the duration of the closure, The threshold for continuous closure time. To estimate the repair time, As the expected repair time threshold, For resource accessibility threshold, To replace channel availability, To replace the channel availability threshold, This is an indicator of irreversible damage. Based on the extended state vector samples and their corresponding labels, construct the first... Training sample set for each failure mechanism , represented as: ， In the formula, For the first One extended state vector sample, To expand the number of state vector samples; Based on the training sample set The discriminant model is used to obtain the first... The toughness critical boundary of a failure mechanism .

5. The method for early warning of rainstorm and flood resilience margin of mountain roads based on multiple failure mechanisms according to claim 4, characterized in that, The discriminant model includes one of the following: support vector classifier, kernel discriminant model, probability boundary model, ensemble discriminant model, or a combination thereof. When using the support vector classifier, the first... Discriminant function for each failure mechanism for: ， In the formula, For support vector coefficients, For kernel function, For bias terms; The corresponding toughness critical boundary for: 。 6. The method for early warning of rainstorm and flood resilience margin of mountain roads based on multiple failure mechanisms according to claim 3, characterized in that, The steps for determining the current warning level and the final dominant failure mechanism include: Based on the aforementioned mechanism component toughness margin and / or overall toughness margin Calculate the rate of change of toughness margin and the average rate of change of the sliding window, where the comprehensive toughness margin is used. During the calculation, the calculation expressions are as follows: ， ， In the formula, The rate of change of toughness margin, subscript This refers to the discrete sampling time sequence number. The average rate of change of the sliding window. The length of the sliding window, subscript The sample number within the sliding window; The toughness margin of the aforementioned mechanism component Overall resilience margin The average rate of change of the sliding window and the duration of the warning level are compared with preset grading thresholds and release hysteresis conditions to obtain the current warning level and the final dominant failure mechanism. The final dominant failure mechanism is determined by the candidate dominant failure mechanisms within the duration corresponding to the current warning level. During the continuous period, the current candidate dominant failure mechanism that appears most frequently is determined as the final dominant failure mechanism. If the number of occurrences is the same, the failure mechanism with the smallest toughness margin of the mechanism component at the current moment is taken as the final dominant failure mechanism. The current warning level is determined according to any of the following constraints: (1) When satisfied and continue A yellow alert is triggered during each sampling period, among which... , For the grading threshold, The toughness margin is the mechanistic component corresponding to the final dominant failure mechanism. The yellow grading threshold corresponds to the final dominant failure mechanism; (2) When satisfied and continue An orange alert is triggered during each sampling period, among which... , For the grading threshold, The orange-level threshold corresponds to the final dominant failure mechanism; (3) When satisfied and continue A red alert is triggered during each sampling period, among which... , For the grading threshold, The red grading threshold corresponds to the final dominant failure mechanism; in, , All grading thresholds are jointly calibrated based on the statistical quantile values ​​of historical event samples, the early warning lead time target, the false alarm rate constraint, the missed alarm rate constraint, and the important traffic maintenance level. When the release hysteresis condition is met, it indicates that the warning can be lifted. The release hysteresis condition is as follows: and And continue Each sampling period, of which To allow for hysteresis, Release hysteresis for the rate of change.

7. The method for early warning of rainstorm and flood resilience margin of mountain roads based on multiple failure mechanisms according to claim 1, characterized in that, The multi-failure mechanism toughness critical boundary model library performs periodic corrections based on newly added measured disaster samples. The correction methods include at least one of parameter retraining, boundary translation correction, threshold recalibration, and sample reweighting.

8. The method for early warning of rainstorm and flood resilience margin of mountain roads based on multiple failure mechanisms according to claim 1, characterized in that, The linkage control command is generated based on the final dominant failure mechanism, and the specific steps are as follows: When the ultimate dominant failure mechanism is water accumulation leading to failure of the passage function, output speed limit, drainage, channelization detour or temporary road closure commands; When the ultimate dominant failure mechanism is slope instability, output instructions for slope inspection, danger zone control, advance deployment of emergency equipment, or personnel evacuation. When the ultimate dominant failure mechanism is the roadbed scour mechanism, output instructions for bridge approach slab inspection, scour protection, closure control, and emergency reinforcement. When the ultimate dominant failure mechanism is bridge or culvert blockage or erosion, output commands for dredging and clearing blockages, increasing bridge and culvert monitoring, detour diversion, and structural emergency repair.

Images