Method and system for optimizing transportation path of canal oversize cargo considering conditional value at risk

Abstract

This invention relates to the field of transportation route optimization technology, and specifically discloses a method and system for optimizing canal transportation routes for oversized cargo considering conditional risk value. The method includes: acquiring multi-source data; dynamically correcting evaluation parameters in a preset risk model based on real-time dynamic data, and calculating multiple risk probability values ​​for each route segment; aggregating the multiple risk probability values ​​of each route segment to obtain a comprehensive risk probability value, and then using a conditional risk value model combined with the accident risk probability value of each route segment to calculate the expected risk of each route segment; constructing a multi-objective optimization model with the goal of minimizing the total route risk, total cost, and total time; obtaining the Pareto optimal route solution set through a multi-objective optimization algorithm; and determining and dynamically adjusting the target transportation route. This invention can accurately capture extreme risks in dynamic scenarios, achieve real-time dynamic quantification of risks, and improve the ability to identify high-risk routes.

Description

Technical Field

[0001] This invention relates to the field of transportation route optimization technology, and in particular to a method and system for optimizing the transportation routes of oversized cargo in canals, taking into account conditional risk value. Background Technology

[0002] In existing canal oversized cargo transportation route planning technologies, risk assessment relies on static VaR (Value at Risk) models, which cannot cope with extreme risks in dynamic scenarios such as rainstorms and traffic congestion; multi-objective optimization algorithms (such as NSGA-II) suffer from low efficiency and are prone to getting trapped in local optima; and the planning scheme is disconnected from the transportation execution process, lacking the ability to make real-time dynamic adjustments. Summary of the Invention

[0003] To overcome the problems of existing canal oversized cargo transportation route planning technologies, such as inability to handle multiple dynamic scenarios, low efficiency, and lack of dynamic adjustment capabilities, this invention provides a canal oversized cargo transportation route optimization method and system that considers conditional risk value.

[0004] In a first aspect, the present invention provides a method for optimizing the transportation route of oversized cargo in a canal, taking into account conditional risk value, comprising:

[0005] Acquire multi-source data; the multi-source data includes oversized item attribute data, static road network data, and real-time dynamic data during transportation. The evaluation parameters in the preset risk model are dynamically corrected based on the real-time dynamic data, and the risk probability values ​​of each road segment in the transportation network corresponding to various preset risk types are calculated based on the corrected preset risk model; the preset risk model includes a tunnel passage risk model, a bridge load-bearing risk model, a curve rollover risk model, and a curve collision risk model. After aggregating the multiple risk probability values ​​of each road segment to obtain the comprehensive risk probability value of each road segment, the expected risk of each road segment is calculated by combining the accident risk probability value of each road segment with the conditional value at risk model. A multi-objective optimization model is constructed with the objectives of minimizing the total risk of a transportation route, minimizing the total transportation cost, and minimizing the total transportation time. Here, a transportation route is a sequence of multiple road segments, and the total risk of a transportation route is determined by the expected risk of all road segments constituting the corresponding transportation route. The Pareto optimal path solution set is obtained by solving the multi-objective optimization model using a multi-objective optimization algorithm. The target transportation route is determined from the Pareto optimal path solution set, and the target transportation route is dynamically adjusted based on updated real-time dynamic data during transportation.

[0006] According to a specific implementation method, the above method calculates the expected risk of each road segment by combining the accident risk probability value of each road segment with the conditional value at risk model, specifically including: For each road segment, the conditional risk value of each road segment is determined based on the comprehensive risk probability value and accident risk probability value of each road segment under a preset confidence level, and the conditional risk value is used as the risk expectation of each road segment.

[0007] According to one specific implementation, in the above method, dynamically correcting the evaluation parameters of the preset risk model based on the real-time dynamic data includes: The bridge bearing capacity coefficient in the bridge bearing risk model is corrected based on real-time meteorological data. The critical rollover speed in the curve rollover risk model is corrected based on real-time road condition data. The curve collision risk threshold in the curve collision risk model is corrected based on real-time traffic flow data.

[0008] According to one specific implementation, the dynamic adjustment in the above method includes: Based on updated real-time dynamic data, the comprehensive risk probability value, real-time traffic flow data, and real-time weather data of the unreached sections on the target transportation route are continuously monitored. When the comprehensive risk probability value, real-time traffic flow data, and / or real-time weather data of any unreached road segment exceed the corresponding preset threshold, the subsequent routes of the target transportation route are locally re-optimized to generate an adjusted target transportation route.

[0009] According to a specific implementation, in the above method, the multi-objective optimization algorithm adopts the NSGA-II algorithm, and the NSGA-II algorithm uses road network constraint-guided initialization; solving the multi-objective optimization model through the multi-objective optimization algorithm includes: During the algorithm initialization phase, based on the static road network data and the attribute data of oversized items, feasible nodes that meet the physical traffic constraints are pre-screened from all road network nodes, and an initial population is generated based on the feasible nodes.

[0010] According to one specific implementation, the above method, which solves the multi-objective optimization model using a multi-objective optimization algorithm, further includes: During the algorithm iteration process, the crossover probability and / or mutation probability are dynamically adjusted based on the convergence of the current population.

[0011] According to one specific implementation, the above method, dynamically adjusting the crossover probability and / or mutation probability based on the convergence of the current population, includes: When the convergence of the current population is greater than a first preset threshold, the crossover probability and / or the mutation probability are mapped to a first set value range. When the convergence of the current population is less than the second preset threshold, the crossover probability and / or the mutation probability are mapped to a second set value range. The first set value range is greater than the second set value range.

[0012] According to one specific implementation, the above method, which solves the multi-objective optimization model using a multi-objective optimization algorithm, further includes: Differentiated weight coefficients are configured for the optimization objectives in the multi-objective optimization model according to different transportation scenarios; The transportation scenarios include major engineering transportation scenarios and commercial transportation scenarios.

[0013] Secondly, the present invention provides a canal oversized cargo transportation route optimization system that considers conditional risk value, comprising: The data acquisition module is used to acquire multi-source data, including oversized component attribute data, static road network data, and real-time dynamic data during transportation. The risk assessment module is used to dynamically correct the assessment parameters in the preset risk model based on the real-time dynamic data, and to calculate the risk probability value of each road segment in the transportation network corresponding to multiple preset risk types based on the corrected preset risk model; the preset risk model includes a tunnel passage risk model, a bridge load-bearing risk model, a curve rollover risk model, and a curve collision risk model; The risk aggregation module is used to aggregate multiple risk probability values ​​of each road segment to obtain the comprehensive risk probability value of each road segment, and then use the conditional value of risk model to calculate the expected risk of each road segment in combination with the accident risk probability value of each road segment. The model building module is used to construct a multi-objective optimization model with the objectives of minimizing the total risk of a transportation route, minimizing the total transportation cost, and minimizing the total transportation time. Here, a transportation route is a sequence of multiple road segments, and the total risk of a transportation route is determined by the expected risk of all road segments constituting the corresponding transportation route. The path solving module is used to solve the multi-objective optimization model using a multi-objective optimization algorithm to obtain the Pareto optimal path solution set; The route determination and adjustment module is used to determine the target transportation route from the Pareto optimal route solution set and dynamically adjust the target transportation route based on updated real-time dynamic data during transportation.

[0014] According to one specific implementation, the system further includes data acquisition hardware that is communicatively connected to the data acquisition module; The data acquisition hardware includes a Beidou positioning terminal for collecting real-time location and speed of transport vehicles, traffic monitoring equipment for collecting real-time traffic flow data, and meteorological sensing equipment for collecting real-time meteorological data.

[0015] Compared with the prior art, the beneficial effects of the present invention are as follows: This invention dynamically corrects the assessment parameters in the preset risk model based on real-time dynamic data, and calculates the risk probability values ​​of each road segment in the transportation network corresponding to multiple preset risk types based on the corrected preset risk model. Then, the values ​​are aggregated and the expected risk of each road segment is calculated using the conditional risk value model. This can accurately capture extreme risks in dynamic scenarios such as rainstorms, slippery conditions, and congestion, and realize the real-time dynamic quantification of transportation risks. This makes the risk assessment results closely related to the actual environmental changes in the transportation process and improves the ability to identify high-risk road segments. This invention constructs a three-dimensional multi-objective optimization model considering risk, cost, and time, enabling synergistic optimization among safety, economy, and timeliness, and providing multiple Pareto optimal path solutions to meet differentiated needs in different scenarios; This invention, through real-time monitoring and local re-optimization mechanisms during transportation, can dynamically adjust the original route in case of emergencies, thereby improving the transportation process's adaptability to dynamic environments. Attached Figure Description

[0016] Figure 1 This is a flowchart illustrating a method for optimizing the transportation route of oversized cargo in a canal, taking into account conditional risk value, as provided in an embodiment of the present invention. Detailed Implementation

[0017] The present invention will now be described in further detail with reference to specific embodiments. However, this should not be construed as limiting the scope of the present invention to the following embodiments; all technologies implemented based on the content of the present invention fall within the scope of the present invention.

[0018] Unless otherwise specified, the terms "upper," "lower," "left," "right," "center," "inner," and "outer," etc., used in the description of specific embodiments of the present invention to indicate orientation or positional relationships, are based on the orientation or positional relationships shown in the accompanying drawings, or the orientation or positional relationship in which the product / equipment / device is usually placed during use. These terms are merely for the purpose of facilitating the description of the present invention or simplifying the description in specific embodiments, and for enabling those skilled in the art to quickly understand the solution, and do not indicate or imply that a particular device / component / element must have a specific orientation, or be constructed and operated in a specific positional relationship. Therefore, they should not be construed as limitations on the present invention.

[0019] Furthermore, the use of terms such as "horizontal," "vertical," "suspended," "parallel," and "coaxial" does not imply that the corresponding device / component / element must be absolutely horizontal, vertical, suspended, parallel, or coaxial. Slight tilt or deviation is permissible, as long as it does not affect the normal function of the relevant component. For example, "horizontal" simply means that its direction is more horizontal relative to "vertical," not that the structure must be perfectly horizontal; a slight tilt is acceptable. "Coaxial" means that two components are arranged as coaxially as possible, allowing them to move coaxially or approximately coaxially when their relative positions change. Alternatively, it can be simplified to mean that the corresponding device / component / element, when arranged in "horizontal," "vertical," "suspended," "parallel," or "coaxial" directions, can have an error / deviation of ±10% relative to the corresponding direction, more preferably within ±8%, more preferably within ±6%, more preferably within ±5%, and more preferably within ±4%. For example, the deviation in the "coaxial" direction is controlled within 0.2-1mm, preferably within 0.2-0.5mm. As long as the corresponding device / component / element is within the error / deviation range, it can still achieve its function in the solution of the present invention.

[0020] Furthermore, the use of terms such as "first," "second," and "third" in terminology is merely for distinguishing descriptions of identical or similar components and should not be interpreted as emphasizing or implying the relative importance of a particular component.

[0021] Furthermore, in the description of the embodiments of the present invention, "several", "more than", and "a number of" represent at least two. The number can be any number, such as two, three, four, five, six, seven, eight, or nine, and can even exceed nine.

[0022] Furthermore, in the description of the technical solution of this invention, unless otherwise explicitly specified / limited / restricted, the terms "set up," "install," "connect," "link," "provided with," "laid out," and "arranged" should be interpreted broadly. For example, they can refer to fixed connections, detachable connections, or integral connections; they can refer to connection methods commonly used in the art, such as welding, riveting, bolting, and threaded connections. Such connections can be mechanical, electrical, or communication connections; they can be direct connections or indirect connections through an intermediate medium; and they can refer to the internal communication between two components.

[0023] As a crucial link in major engineering construction, the transportation of oversized and heavy-duty cargo is characterized by four core features: "oversized, overweight, high-value, and irreplaceable." Examples include oversized components (such as miter gates of ship locks and steel box girders for bridges) in major infrastructure projects like canal projects and cross-sea bridges. During the transportation of such cargo, accidents such as collisions, overturning, or bridge collapses can cause not only huge economic losses (with single accidents exceeding ten million yuan) but also delays in construction by months or even years. Therefore, the transportation routes must meet extremely high requirements for "physical feasibility, risk control, cost-effectiveness, and time efficiency."

[0024] In practical applications, existing technologies for optimizing the transportation routes of oversized and heavy equipment suffer from static risk assessments and insufficient capacity for managing extreme risks. They commonly employ static VaR (Value at Risk) models, calculating risk solely based on historical road network data (such as bridge design load capacity and tunnel clearance), neglecting the dynamic impact of real-time environmental changes on risk. Furthermore, VaR models only reflect risk levels under normal conditions and cannot cover low-probability, high-impact extreme events in the transportation of oversized and heavy equipment (such as tunnel collisions and curve rollovers), the consequences of which are often catastrophic.

[0025] Furthermore, multi-objective optimization is one-sided and lacks adaptability to engineering scenarios. Existing systems mostly use fixed multi-objective weights, failing to differentiate the priority requirements of different transportation scenarios—major engineering transportation requires "risk priority" to ensure safe delivery; commercial transportation requires "cost priority" to control transportation costs within budget; and ordinary industrial transportation requires "time priority" to ensure production progress. Their optimization algorithms are inefficient and lack solution set diversity. Existing NSGA-II algorithms for path optimization generally use random population initialization, failing to consider the strict physical constraints of transporting oversized and heavy items. Planning and execution are disconnected, lacking real-time dynamic adjustment capabilities. Existing systems can only output static path plans before transportation, without linking with real-time data during transportation execution. When unexpected situations occur during transportation, data needs to be collected manually and the path replanned, with response times generally exceeding 2 hours.

[0026] Based on this, this invention addresses the problem that traditional VaR models cannot cover extreme risks (such as sudden drop in bridge load-bearing capacity due to heavy rain or rollover on slippery roads). It uses a dynamic CVaR model (Conditional Value at Risk) to correct risk factors in real time, accurately quantify low-probability, high-impact events, and avoid safety accidents such as scrapes and bottoming out. It also solves the problems of blind initialization (high proportion of invalid individuals) and easy getting trapped in local optima in traditional algorithms. By improving the NSGA-II algorithm (road network constraint initialization and adaptive crossover mutation), it enhances the diversity of solution sets and convergence speed, balances risk, cost, and time objectives, and improves the efficiency and quality of multi-objective optimization.

[0027] The technical solution provided by the present invention will be described and explained in detail below with reference to the accompanying drawings.

[0028] Please refer to Figure 1 The diagram illustrates a flowchart of a canal oversized cargo transportation route optimization method considering conditional risk value, provided by an embodiment of the present invention, including: Step 1: Obtain multi-source data.

[0029] The multi-source data includes oversized cargo attribute data, static road network data, and real-time dynamic data during transportation. Specifically, static attribute data of oversized cargo, static road network parameters, and real-time dynamic data are collected. The data is cleaned, standardized, and spatiotemporally aligned, and then divided into training and testing sets. The real-time dynamic data includes the real-time location of transport vehicles, real-time traffic flow, road surface friction coefficient (wet / dry state), and short-term weather forecast data (heavy rain, strong winds, etc.) obtained from BeiDou positioning.

[0030] Step 2: Dynamically correct the evaluation parameters in the preset risk model based on the real-time dynamic data, and calculate the risk probability value of each road segment in the transportation network corresponding to multiple preset risk types based on the corrected preset risk model.

[0031] The preset risk models include a tunnel passage risk model, a bridge load-bearing risk model, a curve rollover risk model, and a curve collision risk model. It is understood that the various preset risk types include tunnel passage risk, bridge load-bearing risk, curve rollover risk, and curve collision risk, and the preset risk models include tunnel passage risk models, bridge load-bearing risk models, curve rollover risk models, and curve collision risk models, respectively used to calculate the various preset risk types.

[0032] The evaluation parameters of the preset risk model are dynamically adjusted based on the real-time dynamic data, including: The bridge bearing capacity coefficient in the bridge bearing risk model is corrected based on real-time meteorological data. The critical rollover speed in the curve rollover risk model is corrected based on real-time road condition data. The curve collision risk threshold in the curve collision risk model is corrected based on real-time traffic flow data.

[0033] In one possible implementation, the formula for calculating the tunnel's risk model is:

[0034] in, The total height of the oversized transport vehicle after loading, including the height of the vehicle itself and the height of the cargo; is the tunnel clearance height, which is the minimum vertical distance from the road surface to the top inside the tunnel; h is the safety margin, which is the minimum vertical distance reserved. This represents the probability value of the risk of passing through the tunnel.

[0035] The calculation formula for the bridge bearing capacity risk model is as follows:

[0036] The total distance for transporting oversized and heavy items is s, and the length of the transport section consisting of bridges is s. The bridge bearing capacity coefficient is ; This represents the probability value of bridge bearing capacity risk.

[0037] Specifically, the bridge load-bearing capacity correction in a rainstorm scenario is: α' = α × (1 - 0.02 × P), where P is the hourly rainfall (mm / h). For example, when P ≥ 50 mm / h, α' should be at least 0.7.

[0038] The curve rollover risk model is calculated by quantifying the rollover probability based on the balance relationship between centrifugal force and stabilizing torque. When centrifugal force... F c The resulting overturning moment M o Greater than the stabilizing torque generated by gravity M s ,Right now At this time, the vehicle may roll over, and the calculation formula is as follows:

[0039]

[0040]

[0041]

[0042] in, F c The centrifugal force of an oversized transport vehicle when it passes through a curve s; M o The overturning moment of an oversized transport vehicle when it passes through a curve s; M s The stabilizing moment of the oversized transport vehicle when it passes through curve s; m is the weight of the oversized item (kg); h is the height of the center of gravity of the oversized item (m). Δy The lateral offset of the center of gravity of the extra-large component (m); r s Let be the turning radius (m) of curve s; v sT represents the turning speed (m / s) of the oversized transport vehicle as it passes through curve s; T represents the wheelbase (m) of the oversized transport vehicle. This represents the probability value of rollover risk on a curve.

[0043] Specifically, the risk correction for rollover on wet, slippery road curves: critical rollover speed. Vs' = Vs × ,in f This represents the road surface friction coefficient. For example, on a dry road surface... f =0.8, in rainy conditions f =0.4, under icy and snowy road surface f =0.2.

[0044] The curve collision risk model calculates the collision probability based on the matching degree between the road width and the vehicle's turning radius. The formula is as follows:

[0045]

[0046]

[0047]

[0048]

[0049] in, The outer radius (m) of the oversized transport vehicle. The inner radius of the oversized transport vehicle (m); Minimum turning radius of the vehicle (m); This is the minimum road width (m) required for vehicles to turn. denoted as , where L is the probability value of a collision on a curve; a is the vehicle wheelbase (m); b is the vehicle front overhang length; T is the vehicle width; and T is the track width (m) for oversized transport vehicles. θ max This is the maximum deflection angle of the outer wheel on the front axle of the vehicle.

[0050] Specifically, real-time traffic congestion collision risk correction: collision risk threshold Where Q is the real-time traffic flow (vehicles / hour). For example, when Q ≥ 1000 vehicles / hour, Reduced to 0.08.

[0051] Step 3: After aggregating the multiple risk probability values ​​of each road segment to obtain the comprehensive risk probability value of each road segment, the expected risk of each road segment is calculated by combining the accident risk probability value of each road segment with the conditional value at risk model.

[0052] Specifically, for each road segment, the conditional risk value of each road segment is determined based on the comprehensive risk probability value and accident risk probability value of each road segment under a preset confidence level, and the conditional risk value is used as the risk expectation of each road segment.

[0053] In practical applications, the various risk probability values ​​calculated in step 2 can be regarded as multiple risk samples of the road segment. The basic risk sample of the road segment can be obtained by weighted summation, and this value can be used as the comprehensive risk probability value of the road segment.

[0054] In one possible implementation, for road segments ij The formula for calculating the overall risk probability value is:

[0055] In this embodiment of the application, each road segment is represented by a node. i To the node j The spaces between indicate, Indicates road segment ij The comprehensive risk probability value is used as the input sample for subsequent CvaR calculation. Indicates road segment ij The probability value of passing through the tunnel. Indicates road segment ij The probability value of bridge load-bearing risk. Indicates road segment ij The probability value of rollover risk on a curve. Indicates road segment ij The probability value of a collision on a curve. , , , These represent the weighting coefficients for different risk probability values.

[0056] The impact of accidents along each route is sorted in ascending order. . for The nth minimum value. Based on the kinetic energy of the oversized component and the geometric parameters of the road section, the formula for calculating the loss impact value after the accident is:

[0057] in, This represents the probability value of the accident risk. The road segment impact coefficient. M For the weight of extra-large items, The speed limit for the road segment. B, H These are the geometric parameters of the road segment.

[0058] In one possible implementation, for each road segment, its conditional risk value at a pre-set confidence level α is determined based on the segment's historical risk data or simulated loss distribution. Specifically, let road segment... ij Let X be the random variable of risk loss, and f(x) be its probability density function. Then, the value at risk VaRα is the quantile that satisfies P(X>VaRα)=1-α, and the conditional value at risk CVaRα is the conditional expected value of the loss exceeding VaRα. The formula is as follows:

[0059] To facilitate computational optimization, the Rockafellar & Uryasev method is used to transform CVaR into a solvable functional form:

[0060] in, , In the confidence interval The risk value, Risk expectations for each road section ij .

[0061] Step 4: Construct a multi-objective optimization model with the objectives of minimizing the total risk of the transportation route, minimizing the total transportation cost, and minimizing the total transportation time.

[0062] Specifically, in At a certain confidence level, the formula for minimizing the total risk of the transportation route is:

[0063]

[0064]

[0065] The formula for minimizing total transportation cost is:

[0066] The formula for minimizing the total transportation time is:

[0067] The definitions of each parameter in the above calculation formula are shown in Table 1.

[0068] Table 1 Parameter Illustration

[0069] Step 5: Solve the multi-objective optimization model using a multi-objective optimization algorithm to obtain the Pareto optimal path solution set.

[0070] The multi-objective optimization algorithm employs the NSGA-II algorithm, which uses road network constraint-guided initialization. Solving the multi-objective optimization model using the multi-objective optimization algorithm includes: During the algorithm initialization phase, based on the static road network data and the attribute data of oversized items, feasible nodes that meet the physical traffic constraints are pre-screened from all road network nodes, and an initial population is generated based on the feasible nodes.

[0071] Specifically, the physical access constraints include, but are not limited to: Height constraint: Total height of cargo after loading ≤ tunnel clearance height - safety margin; Width constraint: Cargo width ≤ road segment width limit; Weight constraint: Total vehicle weight ≤ design load-bearing capacity of bridges in the road section; Turning radius constraint: Minimum turning radius of vehicle ≤ curve radius of road segment; Length constraint: Total vehicle length ≤ maximum vehicle length allowed on the road segment.

[0072] Solving the multi-objective optimization model using a multi-objective optimization algorithm further includes: During the algorithm iteration process, the crossover probability and / or mutation probability are dynamically adjusted based on the convergence of the current population.

[0073] Specifically, when the convergence of the current population is greater than a first preset threshold, the crossover probability and / or the mutation probability are mapped to a first set value range, i.e. a higher probability range, in order to enhance the population's exploration ability. When the convergence of the current population is less than the second preset threshold, the crossover probability and / or the mutation probability are mapped to the second set value range, i.e., the lower probability range, in order to enhance the population development capability.

[0074] It is understandable that the first set value range is greater than the second set value range. Convergence is used to characterize the degree of clustering of a population in the objective function space. In this embodiment of the invention, convergence is defined as the ratio of the standard deviation of the top 20% of individuals in the current population (i.e., individuals at the first non-dominant ordination front) on each objective function to the standard deviation of the initial population, or as the average distance between the Pareto fronts of two adjacent generations of the population. A larger convergence value indicates lower population diversity; a smaller convergence value indicates a more dispersed population distribution.

[0075] Solving the multi-objective optimization model using a multi-objective optimization algorithm further includes: Differentiated weight coefficients are configured for the optimization objectives in the multi-objective optimization model according to different transportation scenarios; The transportation scenarios include major engineering transportation scenarios and commercial transportation scenarios.

[0076] In one specific implementation, the present invention employs an improved NSGA-II algorithm guided by road network constraints, consisting of population initialization, adaptive crossover mutation, and multi-attribute congestion ranking, to solve the Pareto optimal solution set of the multi-objective optimization model. The road network constraint-guided initialization avoids the generation of invalid populations by pre-screening feasible nodes that meet the physical limits of oversized components.

[0077] (1) Road network constraint-guided initialization: Traverse the road network nodes, remove nodes that do not meet the requirements of "cargo width ≤ road width limit, cargo height ≤ tunnel clearance, turning radius ≥ vehicle minimum turning radius", generate an initial population based only on feasible nodes, and the proportion of invalid individuals in the population is ≤5%; (2) Adaptive crossover and mutation: When the population convergence is ≥0.8 (convergence = difference rate of objective function values ​​of the top 20% of individuals), the crossover probability is increased to 1.0 and the mutation probability is increased to 0.1, which enhances the population diversity; when the convergence is ≤0.3, the crossover probability is reduced to 0.8 and the mutation probability is reduced to 0.001, which accelerates the convergence. (3) Multi-attribute crowding ranking: Based on the traditional crowding calculation, the risk fluctuation coefficient attribute is introduced to prioritize the retention of individuals with low risk fluctuation and avoid the failure of the optimal solution due to sudden risks.

[0078] Step 6: Determine the target transportation route from the Pareto optimal path solution set, and dynamically adjust the target transportation route based on updated real-time dynamic data during transportation.

[0079] Specifically, by combining project priorities (e.g., prioritizing transportation for major projects to ensure the lowest risk, and prioritizing commercial transportation to ensure the lowest cost), an evaluation system for solutions is established, outputting a set of optimal path candidates suitable for the scenario. The dynamic CVaR risk assessment module also includes an extreme scenario simulation unit: generating 1000 sets of extreme scenarios based on Monte Carlo simulation (e.g., heavy rain + traffic congestion, temporary bridge load restrictions), pre-calculating the risk probability value under each scenario, establishing a scenario-risk mapping library, and shortening the risk calculation time during real-time re-optimization.

[0080] The dynamic adjustment includes: Based on updated real-time dynamic data, the comprehensive risk probability value, real-time traffic flow data, and real-time weather data of the unreached sections on the target transportation route are continuously monitored. When the comprehensive risk probability value, real-time traffic flow data, and / or real-time weather data of any unreached road segment exceed the corresponding preset threshold, the subsequent routes of the target transportation route are locally re-optimized to generate an adjusted target transportation route.

[0081] It should be noted that, in this invention, a transportation path refers to an ordered set of multiple road segments connected sequentially from the starting point to the destination, according to the transportation direction. Each road segment has a definite directionality, pointing from the starting node to the ending node; multiple road segments are connected sequentially through shared nodes to form a complete reachable path from the starting point to the destination. In other words, a transportation path not only includes the set of road segments traversed but also implicitly contains the sequential connection order of these road segments, which is consistent with the actual transportation direction. In the dynamic adjustment of this invention, when performing local re-optimization on subsequent paths, the subsequent paths refer to the sequence of remaining road segments arranged sequentially according to the original direction after the current vehicle position.

[0082] Specifically, dynamic adjustment is mainly based on the real-time location and dynamic data feedback of transport vehicles using BeiDou positioning. If the risk probability value of a certain road segment exceeds a preset threshold, local route re-optimization is triggered, and the corrected real-time route is output. In one possible implementation, the above dynamic adjustment can also generate a two-dimensional risk-cost curve, marking the lowest-risk path under different budgets (e.g., when the budget increases by 10%, the risk can be reduced by 25%), providing decision-makers with a quantitative basis for trade-offs; and further, a heat map is used to display the path risk distribution, compare path schemes under different optimization objectives, and generate a decision report including road improvement suggestions.

[0083] Based on the above technical solution, the embodiments of the present invention dynamically correct the evaluation parameters in the preset risk model based on real-time dynamic data, and calculate the risk probability values ​​of each road segment in the transportation network corresponding to multiple preset risk types based on the corrected preset risk model. Then, the values ​​are aggregated and the risk expectation of each road segment is calculated using the conditional risk value model. This can accurately capture extreme risks in dynamic scenarios such as rainstorms, slippery conditions, and congestion, realize the real-time dynamic quantification of transportation risks, and closely link the risk assessment results with the actual environmental changes in the transportation process, thereby improving the ability to identify high-risk road segments. Furthermore, by constructing a three-dimensional multi-objective optimization model of risk, cost, and time, this invention can achieve synergistic optimization among safety, economy, and timeliness, and provide multiple Pareto optimal path solutions to meet the differentiated needs of different scenarios. Furthermore, through real-time monitoring and local re-optimization mechanisms during transportation, this invention can dynamically adjust the original route in case of emergencies, thereby improving the adaptability of the transportation process to dynamic environments.

[0084] On the other hand, embodiments of the present invention also provide a canal oversized cargo transportation route optimization system that considers conditional risk value, comprising: The data acquisition module is used to acquire multi-source data, including oversized component attribute data, static road network data, and real-time dynamic data during transportation. The risk assessment module is used to dynamically correct the assessment parameters in the preset risk model based on the real-time dynamic data, and to calculate the risk probability value of each road segment in the transportation network corresponding to multiple preset risk types based on the corrected preset risk model; the preset risk model includes a tunnel passage risk model, a bridge load-bearing risk model, a curve rollover risk model, and a curve collision risk model; The risk aggregation module is used to aggregate multiple risk probability values ​​of each road segment to obtain the comprehensive risk probability value of each road segment, and then use the conditional value of risk model to calculate the expected risk of each road segment in combination with the accident risk probability value of each road segment. The model building module is used to construct a multi-objective optimization model with the objectives of minimizing the total risk of a transportation route, minimizing the total transportation cost, and minimizing the total transportation time. Here, a transportation route is a sequence of multiple road segments, and the total risk of a transportation route is determined by the expected risk of all road segments constituting the corresponding transportation route. The path solving module is used to solve the multi-objective optimization model using a multi-objective optimization algorithm to obtain the Pareto optimal path solution set; The route determination and adjustment module is used to determine the target transportation route from the Pareto optimal route solution set and dynamically adjust the target transportation route based on updated real-time dynamic data during transportation.

[0085] According to one specific implementation, the system further includes data acquisition hardware that is communicatively connected to the data acquisition module; The data acquisition hardware includes a Beidou positioning terminal for collecting real-time location and speed of transport vehicles, traffic monitoring equipment for collecting real-time traffic flow data, and meteorological sensing equipment for collecting real-time meteorological data.

[0086] The technical solutions provided by the embodiments of the present invention will be described and explained in detail below with reference to specific implementation methods.

[0087] First, as described in step 1 above, a high-quality dataset is constructed through "layered collection - precise cleaning - standardized alignment" to solve the problems of traditional data fragmentation and large errors, providing reliable input for subsequent risk assessment and algorithm optimization. The specific operations are as follows: Step 101: Static Data Acquisition and Calibration Oversized component attribute data: Length, width, and height are collected using a high-precision laser dimension measuring instrument (accuracy ±0.5mm) (suitable for oversized components ranging from 10-50m); weight is obtained using an industrial-grade weight sensor (range 0-500t, error ≤0.1%); center of gravity height is determined using a center of gravity tester (measurement accuracy ±1mm), and the type of material (such as large transformers, tunnel boring machines) and the level of over-limit (Level 1 / Level 2 / Level 3 over-limit) are recorded simultaneously.

[0088] Road network static parameters: Import basic road network data from a Geographic Information System (GIS), including: Bridge parameters: design load-bearing capacity, bridge deck slope, height / width limit, bearing aging coefficient; Tunnel parameters: clearance dimensions, ventilation capacity, and lighting coverage; Curving parameters: turning radius, road surface friction coefficient, visibility obstruction level (no obstruction / partial obstruction / complete obstruction).

[0089] Step 102: Real-time dynamic data acquisition and synchronization The system uses a BeiDou-3 positioning terminal (positioning frequency 1Hz, horizontal accuracy ±1m) to obtain vehicle location, speed, and steering angle in real time, with a data upload interval of ≤3s. It also uses the public security traffic management platform API interface to obtain real-time traffic flow (vehicles / hour), average vehicle speed (km / h), and congestion level (smooth / slow / congested) for each road segment, with an update frequency of 5 minutes. Furthermore, it calls the National Meteorological Information Center API to obtain short-term forecasts (for the next 6 hours) along the transportation route, including rainfall (mm), wind force (level), and visibility (m). Abnormal weather (heavy rain, strong wind) triggers expedited data updates (1 minute).

[0090] Step 103: Data Preprocessing Operations Outliers (such as location jumps and sensor failure data) are removed using the "3σ principle". Missing values ​​are supplemented using the "adjacent time interpolation method" (dynamic data) or the "average value method of similar road segments" (static data). All data are uniformly converted to the "WGS84 coordinate system + timestamp (millisecond level)" format, and a four-dimensional data association table of "oversized items-vehicles-road network-environment" is established to ensure that the data accuracy is ≥99.2%.

[0091] Furthermore, as described in step 2 above, the extreme risks of transporting oversized and overweight items are accurately quantified through calculation using four types of risk models and dynamic CVaR aggregation. At the same time, real-time calculation efficiency is improved by relying on an extreme scenario mapping library. The specific operations are as follows: Step 201: Dynamic Adjustment of Risk Factors Based on real-time dynamic data, the core parameters of the four types of risk sub-models are modified according to specific scenarios, including the bridge bearing capacity coefficient, the critical speed for curve rollover, and the collision risk threshold for curve.

[0092] Bridge load-bearing capacity correction: Corrected coefficient = design load-bearing capacity × (1 - 0.05 × real-time rainfall coefficient) × (1 - 0.1 × vehicle overload rate), where the rainfall coefficient (0-1) is determined based on the actual rainfall and the bridge's impermeability grade; Curve rollover critical speed correction: Corrected critical speed = initial critical speed × (1 - 0.2 × road surface friction coefficient attenuation rate) × (1 - 0.1 × wind force level / 10), where the friction coefficient attenuation rate is calculated based on real-time rainfall. Curve collision risk threshold correction: Corrected threshold = initial threshold × (1 + 0.3 × road segment congestion level coefficient), where the congestion level coefficient (0-1) corresponds to "smooth traffic = 0.2 / slow traffic = 0.5 / congested = 0.8"; Construction section risk correction: Introduce a construction impact coefficient (0.8-1.5) and adjust the risk probability value according to the proportion of road width occupied by construction (1.0 for ≤30%, 1.2 for 30%-50%, and 1.5 for >50%).

[0093] Step 202: Four Core Risk Calculation Models (1) Tunnel Passage Risk Calculation Unit: Based on Formula

[0094] in, The total height of the oversized transport vehicle after loading, including the height of the vehicle itself and the height of the cargo; is the tunnel clearance height, which is the minimum vertical distance from the road surface to the top inside the tunnel; h is the safety margin, which is the minimum vertical distance reserved.

[0095] (2) Bridge bearing capacity risk calculation unit: based on formula

[0096] The total distance for transporting oversized and heavy items is s, and the length of the transport route consisting of bridges is s. l The bridge bearing capacity coefficient is α; (3) Cornering rollover risk calculation unit: Quantifies the rollover probability based on the balance relationship between centrifugal force and stabilizing torque. When the centrifugal force F c The resulting overturning moment M o The stabilizing torque M greater than that generated by gravity s Right now At this time, the vehicle may roll over, and the calculation formula is as follows:

[0097]

[0098]

[0099]

[0100] Among them, F c M represents the centrifugal force of an oversized transport vehicle as it passes through curve s; oM is the overturning moment of an oversized transport vehicle when it passes through a curve s; s The stabilizing moment of the oversized transport vehicle when it passes through curve s; m is the weight of the oversized item (kg); h is the height of the center of gravity of the oversized item (m); Δy is the lateral offset of the center of gravity of the oversized item (m); r s v is the turning radius of curve s (in meters); s T represents the turning speed (in m / s) of the oversized transport vehicle as it passes through curve s; T represents the wheelbase of the oversized transport vehicle (in m).

[0101] (4) Curve collision risk calculation unit: calculates the collision probability based on the matching degree between road width and vehicle turning radius.

[0102]

[0103]

[0104]

[0105]

[0106]

[0107] in, The outer radius (m) of the oversized transport vehicle. The inner radius of the oversized transport vehicle (m); Minimum turning radius of the vehicle (m); This is the minimum road width (m) required for vehicles to turn. For collision risk during vehicle turning; L is the vehicle wheelbase (m); a is the vehicle front overhang length; b is the vehicle width; T is the track width of the oversized transport vehicle (m); θ max This is the maximum deflection angle of the outer wheel on the front axle of the vehicle.

[0108] Step 203: Building the Extreme Scene Mapping Library Based on data from over 1200 major transport accidents in the past 5 years, a "risk factor-extreme scenario" mapping library was constructed, which includes 28 typical scenarios such as "overloaded bridges in heavy rain" and "cornering and overturning in strong winds". When the real-time risk factor combination matches the scenario in the library, the pre-calculated risk correction coefficient is directly called, which reduces the risk assessment time from the original 15s to ≤3s, adapting to the real-time decision-making needs during transportation.

[0109] Furthermore, as described in step 3 above, the specifics are as follows: Step 301: Dynamic CVaR Calculation Process Considering the "extreme risk priority" characteristic of oversized cargo transportation, a confidence level of α=0.99 was selected (i.e., only a 1% extreme risk probability is tolerated). When aggregating the four types of risk probability values ​​output in step 2, the weights are determined according to the transportation scenario (bridge-dense road sections: bridge bearing risk weight 0.4, others 0.2; mountainous curved road sections: rollover / collision risk weight 0.35 each, others 0.15), to obtain the comprehensive risk probability value.

[0110] Feasibility screening: Road segments with a comprehensive risk probability value ≥ 80 are eliminated (extremely high risk and infeasible), and the set of feasible road segments with a comprehensive risk probability value < 80 is retained. Then, according to step 3 above, combined with the accident risk probability value of each road segment, the conditional risk value of each road segment under the preset confidence level is determined, and the conditional risk value is used as the risk expectation of each road segment.

[0111] Step 302: Extreme Scenario Matching Operation Real-time risk factors (such as rainfall, traffic flow, and wind) of feasible road sections are feature-encoded to generate 12-dimensional feature vectors. The cosine similarity algorithm is used to calculate the similarity between the real-time feature vectors and the scenes in the extreme scene mapping library. When the similarity is ≥0.8, it is judged as a highly correlated scene. For road sections corresponding to highly correlated scenes, an additional risk buffer value of 10%-20% is added to avoid underestimating the risk in extreme scenes.

[0112] Furthermore, as described in step 4 above, this step aims to establish a multi-objective optimization framework that fits the actual transportation scenario, solving the problem of poor adaptability of traditional fixed-weight optimization. The specific implementation process includes: Risk objective function: Minimize the total dynamic risk probability value of the path (CVaR).

[0113]

[0114] Cost objective function: Minimize the total lifecycle transportation cost, encompassing vehicle fixed costs, overload fees, escort fees, road improvement costs, fuel costs, and contingency reserve costs. The formula is:

[0115]

[0116]

[0117] The time objective function is to minimize the total transportation time, differentiate driving speeds for ordinary road sections, tunnel sections, and bridge sections, and incorporate road reconstruction time and emergency waiting time. The formula is as follows:

[0118] Model parameter symbols and explanations

[0119] Furthermore, as described in step 5 above, this embodiment of the invention also provides a training process for a multi-objective optimization model, employing a four-fold cross-training method, as detailed below: Step 501: Training Dataset Design Eighteen typical oversized cargo transportation routes in China were selected (covering six scenarios including plains, mountains, and areas with dense bridges, including five cases of Class I oversized cargo, eight cases of Class II oversized cargo, and five cases of Class III oversized cargo); each route included actual values ​​of the three-dimensional targets of "risk-cost-time" (a total of 54 samples) and corresponding real-time dynamic data (output in step 1), ensuring that the data covered multiple trade-off scenarios such as "high risk-low cost-long time" and "low risk-high cost-short time".

[0120] Step 502: Four-fold cross-training process The 18 road segments were randomly divided into 4 subsets (3 subsets containing 5 road segments and 1 subset containing 3 road segments). For each training iteration, 3 subsets (13-14 road segments) were selected as training samples, and 1 subset was used as validation samples. This process was repeated 4 times to complete training on the full dataset. The goal was to minimize the mean absolute percentage error (MAPE, which measures the deviation between predicted and actual values) and the coefficient of determination (R²). 2 The optimization objective is to maximize the model fit; the Particle Swarm Optimization (PSO) algorithm is used to search for the optimal weight coefficients, with constraints applied during the iteration process.

[0121] Step 503: Training Results and Weight Determination After training, the model achieved a MAPE of 7.8% and an R² of 0.94, meeting the preset accuracy requirements of "R²≥0.93, MAPE≤8%", which can be used for subsequent algorithm optimization. For major engineering scenarios (such as the transportation of nuclear power equipment and ultra-high voltage transformers), the final weights were determined as follows: k1=0.5 (risk priority, ensuring transportation safety), k2=0.2 (cost control), and k3=0.3 (timeliness guarantee). The scenario fit rate of these weights in the validation set reached 92%.

[0122] Step 504: Improve the NSGA-II algorithm for optimization This step improves the algorithm's optimization efficiency and solution set quality through road network constraint-guided initialization, adaptive crossover and mutation, multi-attribute congestion ranking, and elite retention strategies. The specific operations are as follows: Feasible nodes that meet physical boundary constraints are pre-screened, and paths are represented using real-number encoding to generate an initial population of 200 individuals. The efficiency of the initial population is calculated, and the proportion of individuals including hub nodes is determined to meet the initialization requirements. The function values ​​of each individual under the three-dimensional objectives of risk, cost, and time are calculated, and the population is divided into different frontiers based on dominance relationships.

[0123] Convergence calculation and parameter adjustment: When iterating to the 50th iteration, the standard deviation ratio of the objective function values ​​of the first frontier individuals is calculated, and the convergence is 0.65 (0.3 < convergence < 0.8), the crossover probability remains at 0.9, and the mutation probability remains at 0.01; when iterating to the 70th iteration, the convergence is 0.82 (≥ 0.8), the crossover probability increases to 1.0, and the mutation probability increases to 0.1, thus enhancing population diversity; Calculate the function value of each individual under the three-dimensional objective of "risk-cost-time" (based on the objective function in step 4). According to the dominance relationship (if all objective values ​​of individual A are better than those of individual B, then A dominates B), divide the population into different frontiers. The first frontier is the current optimal solution set. Iteration to the 50th iteration: Calculate the standard deviation ratio of the objective function values ​​of the first frontier individuals, and obtain a convergence degree of 0.65 (in the range of 0.3-0.8, indicating stable population convergence), maintaining a crossover probability of 0.9 and a mutation probability of 0.01; After 70 iterations: convergence = 0.82 (≥0.8, insufficient population diversity), the crossover probability was increased to 1.0 and the mutation probability was increased to 0.1 to enhance the population's exploration capabilities. Adaptive crossover and mutation: Two parent individuals are randomly selected, and offspring individuals are generated according to the crossover probability to ensure that the offspring paths still satisfy node connectivity; the individual's encoding bits (node ​​indices) are mutated with a small probability (e.g., [1,3,5] mutates to [1,4,5]) to avoid the algorithm getting trapped in local optima. Multi-attribute crowding ranking: Calculate the crowding of the first frontier individual in the three dimensions of "risk, cost, and time" (the greater the distance, the less crowded, which represents better diversity).

[0124] Elite retention strategy: merge the parent population (200 individuals) and the offspring population (200 individuals), re-execute the non-dominated sorting and crowding calculation, and select the 200 best individuals to form the next generation population according to the "front priority (first front priority) + crowding priority (lowest crowding priority)".

[0125] The iteration process terminates after 80 iterations (or after 5 consecutive iterations where the first frontier solution set remains unchanged), outputting a Pareto optimal solution set (25-30 solutions). Each solution set includes a "path node sequence, three-dimensional objective values ​​(risk CVaR, total cost, and total time)". Furthermore, as described in step 6 above, the optimal route is output and dynamically adjusted in real time. This step forms a closed loop through "visualization - decision support - dynamic adjustment" to ensure that the route plan adapts to real-time changes during transportation. The specific operations are as follows: Step 601: Path Visualization and Decision Report Generation An interactive interface based on GIS maps is developed, marking nodes (such as starting point, toll stations, and destination) and road segments of the optimal route. A red (high risk) - yellow (medium risk) - green (low risk) heatmap is used to present the risk distribution, updating real-time vehicle trajectories every minute. Zooming and clicking to query road segment information (such as the CVaR value and estimated travel time for that segment) are supported. A PDF report is automatically generated, including: Risk assessment: Risk probability values ​​for each road segment, and suggestions for avoiding high-risk road segments; Cost-effectiveness: Total cost breakdown (e.g., fuel cost of 220,000 yuan, escort fee of 150,000 yuan), cost optimization potential (e.g., choosing nighttime transportation can reduce over-limit fees by 10%). Time planning: Estimated travel time for each road section and key time nodes; Road renovation recommendations: Location of the road section requiring renovation and renovation plan.

[0126] Step 602: Real-time dynamic adjustment mechanism Path re-optimization is triggered when the following conditions occur: The real-time risk probability value exceeds the threshold (dynamic CVaR ≥ 70). Sudden change in traffic flow (the congestion level of a road segment increases from "slow" to "congested"); Extreme weather warnings (such as blue or higher warnings for heavy rain and strong winds); Adjustment instruction push: The system pushes the adjustment report (including the new route plan and the reason for the adjustment) to the transportation command center through the system interface, and sends an SMS reminder to the fleet manager at the same time; After the fleet implements the adjustment, it transmits the "execution status" back through the Beidou terminal. The system updates the dataset based on the feedback data. If the risk of the new route still exceeds the standard, a second optimization is initiated (the algorithm optimization time is ≤5 minutes) to ensure that the transportation process is always in a state of "low risk - reasonable cost - controllable timeliness".

[0127] In this embodiment of the invention, the aforementioned data acquisition module is used to solve the problem of fragmented multi-source data, construct a high-quality "static + dynamic" dataset, and support the operation of subsequent modules. It interfaces with laser measuring instruments, sensors, and other equipment to acquire the dimensions / weight / center of gravity of oversized components and road network, bridge, and tunnel parameters, supporting data on oversized components ranging from 10-50m and national provincial-level or higher road networks. Through BeiDou terminals, traffic APIs, and meteorological APIs, it acquires vehicle locations (delay ≤ 3s), traffic flow (updated every 5 minutes), and short-term weather forecasts (updated every 1 minute for abnormal weather). Outliers are removed using the "3σ principle," missing values ​​are supplemented through interpolation, and the data format is standardized, achieving a processing accuracy of ≥ 99.2%. Data update delay is ≤ 3s, supporting 200 vehicles online simultaneously, and batch processing of 100,000 data entries takes ≤ 1 second per entry.

[0128] The risk assessment module is used to dynamically quantify extreme risks, solving the problem of poor generalization in static assessments. Based on real-time scenarios such as rainfall, congestion, and construction, it dynamically adjusts parameters such as bridge load-bearing capacity and critical speed of curves, with a correction response of ≤0.5s; it performs parallel calculations of tunnel, bridge, and curve rollover / collision risks, with a single-class calculation of ≤0.3s and a total time of ≤1s, achieving a risk level classification accuracy of ≥92%.

[0129] The risk aggregation module integrates various risks using CVaR, supports adaptive confidence levels (α=0.99 for major projects), and aggregates risk probability values ​​by weighting a library of 28 extreme scenarios. The total evaluation time is ≤3s, and the deviation is ≤8%. The extreme scenario recognition rate is ≥95%, and it is compatible with 8 types of dynamic scenarios such as rain, snow, and construction.

[0130] The model building module is used to establish a three-dimensional optimization framework of "risk-cost-time" to solve the problem of poor adaptability of fixed weights. It quantifies risk (minimizing total CVaR), cost (including 6 items such as fixed / over-limit / fuel), and time (distinguishing between road segment speed and modification / emergency time), with a calculation error of ≤3%; it sets physical limits, time, and cost constraints, supports dynamic updates (synchronization within 1 second for changes in road network / budget), and has a constraint satisfaction rate of ≥98%; the model building time is ≤5 minutes, and it supports user-defined scenario parameter templates.

[0131] The path finding module improves the efficiency and solution quality of path optimization, addressing the slow convergence issue of traditional algorithms. It pre-screens feasible nodes (removing road segments with a CVaR ≥ 80), generates 200 initial populations with real-valued encoding, achieving an efficiency ≥ 95% and initialization time ≤ 2 minutes. Adaptive crossover and mutation dynamically adjust parameters based on convergence (crossover 0.9 / mutation 0.01 for stability, 1.0 / 0.1 for insufficient diversity), achieving a progeny efficiency ≥ 98%. An elite retention strategy merges parent / progeny populations (400 individuals), selecting the 200 optimal individuals based on frontier + crowding, terminating after 80 iterations, outputting 25-30 Pareto solutions. A single optimization cycle takes ≤ 10 minutes (for a 500-node road network), achieving a solution set coverage ≥ 90% and a target value deviation ≤ 5%.

[0132] The route determination and adjustment module is used to achieve a closed loop of "visualization-decision-adjustment," solving the problem that fixed routes are difficult to handle in case of emergencies. It employs Pareto solution filtering, previews the solution set using a 3D scatter plot, and performs multi-dimensional filtering (risk / cost / time limits), recommending three optimal solutions based on scenario weights, with a processing time of ≤10 seconds. It outputs visualized results and reports, GIS map annotations of the route + risk heatmap (track updated every 1 minute), and automatically generates a PDF report (including risk / cost / modification suggestions), with report generation time ≤1 minute. It monitors risks / traffic / weather for 30 seconds, triggering re-optimization if limits are exceeded (≤3 minutes), and pushes adjustment instructions (SMS + system message), with a response time of ≤3 minutes. User decision-making efficiency is improved by 40%, instruction push success rate is ≥99%, and secondary optimization trigger rate is ≤5%.

[0133] This embodiment optimizes the transportation route for oversized items through the above steps. Compared with existing technologies, it achieves more precise risk management. It uses a CVaR model to quantify extreme risks, reducing the assessment error of tail loss by more than 40% compared to the VaR model, thus avoiding significant losses caused by black swan events. By simultaneously optimizing risk, cost, and time through the NSGA-II algorithm, the diversity of solution sets is increased by 50%, providing decision-makers with more flexible options. Dedicated models are designed for the physical constraints (size, weight) and road network characteristics (bridges / tunnels) of oversized items, which can reduce transportation delays by more than 30%. It has a high degree of intelligence: it automatically integrates multi-source data and generates modification suggestions, reducing the cost of manual decision-making and promoting the digital transformation of logistics and transportation.

[0134] It is understood that integrated module units, if implemented as software functional units and sold or used as independent products, can be stored in a readable storage medium. Based on this understanding, all or part of the processes in the methods of the above embodiments of the present invention can be implemented by a program instructing related hardware. The program can be stored in a readable storage medium, and when executed by a processor, it can implement the steps of the various method embodiments described above. The program includes program code, which can be in the form of source code, object code, executable files, or some intermediate form. A readable storage medium can include at least: any entity or device capable of carrying program code to a device / app, a recording medium, computer memory, read-only memory (ROM), random access memory (RAM), electrical carrier signals, telecommunication signals, and software distribution media. Examples include USB flash drives, portable hard drives, magnetic disks, or optical disks. In certain jurisdictions, depending on legislation and patent practice, a readable storage medium may not be an electrical carrier signal or a telecommunication signal.

[0135] Furthermore, various aspects or features of the present invention can be implemented as methods, apparatus, or articles of manufacture using standard programming and / or engineering techniques. The term "article of manufacture" as used herein encompasses a computer program accessible from any computer-readable device, carrier, or medium. For example, computer-readable media may include, but are not limited to: magnetic storage devices (e.g., hard disks, floppy disks, or magnetic tapes), optical discs (e.g., compact discs (CDs), digital versatile discs (DVDs), etc.), smart cards, and flash memory devices (e.g., erasable programmable read-only memory (EPROMs), cards, sticks, or key drives, etc.). Additionally, the various storage media described herein may represent one or more devices and / or other machine-readable media for storing information. The term "machine-readable medium" may include, but is not limited to, wireless channels and various other media capable of storing, containing, and / or carrying instructions and / or data.

[0136] Those skilled in the art will recognize that the units and algorithm steps of the various examples described in conjunction with the embodiments disclosed herein can be implemented in electronic hardware or a combination of computer software and electronic hardware. Whether these functions are implemented in hardware or software depends on the specific application and design constraints of the technical solution. Those skilled in the art can use different methods to implement the described functions for each specific application, but such implementations should not be considered beyond the scope of this invention.

[0137] In the embodiments provided by this invention, it should be understood that the disclosed apparatus / device and method can be implemented in other ways. For example, the apparatus / device embodiments described above are merely illustrative. For instance, the division of modules or units is only a logical functional division, and in actual implementation, there may be other division methods. For example, multiple units or components may be combined or integrated into another system, or some features may be ignored or not executed. Furthermore, the coupling or direct coupling or communication connection shown or discussed may be through some interfaces; the indirect coupling or communication connection between devices or units may be electrical, mechanical, or other forms.

[0138] The units described as separate components may or may not be physically separate. The components shown as units may or may not be physical units; that is, they may be located in one place or distributed across multiple network units. Some or all of the units can be selected to achieve the purpose of this embodiment according to actual needs.

[0139] The above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. Any modifications, equivalent substitutions, and improvements made within the spirit and principles of the present invention should be included within the protection scope of the present invention.

Claims

1. A method for optimizing canal transport routes for oversized cargo considering conditional risk value, characterized in that, include: Acquire multi-source data; the multi-source data includes oversized item attribute data, static road network data, and real-time dynamic data during transportation. The evaluation parameters in the preset risk model are dynamically corrected based on the real-time dynamic data, and the risk probability values ​​of each road segment in the transportation network corresponding to multiple preset risk types are calculated based on the corrected preset risk model. The preset risk models include a tunnel passage risk model, a bridge load-bearing risk model, a curve rollover risk model, and a curve collision risk model. After aggregating the multiple risk probability values ​​of each road segment to obtain the comprehensive risk probability value of each road segment, the expected risk of each road segment is calculated by combining the accident risk probability value of each road segment with the conditional value at risk model. A multi-objective optimization model is constructed with the objectives of minimizing the total risk of a transportation route, minimizing the total transportation cost, and minimizing the total transportation time. Here, a transportation route is a sequence of multiple road segments, and the total risk of a transportation route is determined by the expected risk of all road segments constituting the corresponding transportation route. The Pareto optimal path solution set is obtained by solving the multi-objective optimization model using a multi-objective optimization algorithm. The target transportation route is determined from the Pareto optimal path solution set, and the target transportation route is dynamically adjusted based on updated real-time dynamic data during transportation.

2. The method according to claim 1, characterized in that, The expected risk of each road segment is calculated using the conditional value-at-risk model combined with the accident risk probability value of each segment. Specifically, this includes: For each road segment, the conditional risk value of each road segment is determined based on the comprehensive risk probability value and accident risk probability value of each road segment under a preset confidence level, and the conditional risk value is used as the risk expectation of each road segment.

3. The method according to claim 1, characterized in that, The evaluation parameters of the preset risk model are dynamically adjusted based on the real-time dynamic data, including: The bridge bearing capacity coefficient in the bridge bearing risk model is corrected based on real-time meteorological data. The critical rollover speed in the curve rollover risk model is corrected based on real-time road condition data. The curve collision risk threshold in the curve collision risk model is corrected based on real-time traffic flow data.

4. The method according to claim 1, characterized in that, The dynamic adjustment includes: Based on updated real-time dynamic data, the comprehensive risk probability value, real-time traffic flow data, and real-time weather data of the unreached sections on the target transportation route are continuously monitored. When the comprehensive risk probability value, real-time traffic flow data, and / or real-time weather data of any unreached road segment exceed the corresponding preset threshold, the subsequent routes of the target transportation route are locally re-optimized to generate an adjusted target transportation route.

5. The method according to claim 1, characterized in that, The multi-objective optimization algorithm employs the NSGA-II algorithm, which uses road network constraint-guided initialization. Solving the multi-objective optimization model using the multi-objective optimization algorithm includes: During the algorithm initialization phase, based on the static road network data and the attribute data of oversized items, feasible nodes that meet the physical traffic constraints are pre-screened from all road network nodes, and an initial population is generated based on the feasible nodes.

6. The method according to claim 5, characterized in that, Solving the multi-objective optimization model using a multi-objective optimization algorithm further includes: During the algorithm iteration process, the crossover probability and / or mutation probability are dynamically adjusted based on the convergence of the current population.

7. The method according to claim 6, characterized in that, The crossover and / or mutation probabilities are dynamically adjusted based on the current population convergence, including: When the convergence of the current population is greater than a first preset threshold, the crossover probability and / or the mutation probability are mapped to a first set value range. When the convergence of the current population is less than the second preset threshold, the crossover probability and / or the mutation probability are mapped to a second set value range. The first set value range is greater than the second set value range.

8. The method according to claim 5, characterized in that, Solving the multi-objective optimization model using a multi-objective optimization algorithm further includes: Differentiated weight coefficients are configured for the optimization objectives in the multi-objective optimization model according to different transportation scenarios; The transportation scenarios include major engineering transportation scenarios and commercial transportation scenarios.

9. A canal oversized cargo transportation route optimization system considering conditional risk value, characterized in that, include: The data acquisition module is used to acquire multi-source data, including oversized component attribute data, static road network data, and real-time dynamic data during transportation. The risk assessment module is used to dynamically correct the assessment parameters in the preset risk model based on the real-time dynamic data, and to calculate the risk probability value of each road segment in the transportation network corresponding to multiple preset risk types based on the corrected preset risk model. The preset risk models include a tunnel passage risk model, a bridge load-bearing risk model, a curve rollover risk model, and a curve collision risk model. The risk aggregation module is used to aggregate multiple risk probability values ​​of each road segment to obtain the comprehensive risk probability value of each road segment, and then use the conditional value of risk model to calculate the expected risk of each road segment in combination with the accident risk probability value of each road segment. The model building module is used to construct a multi-objective optimization model with the objectives of minimizing the total risk of a transportation route, minimizing the total transportation cost, and minimizing the total transportation time. Here, a transportation route is a sequence of multiple road segments, and the total risk of a transportation route is determined by the expected risk of all road segments constituting the corresponding transportation route. The path solving module is used to solve the multi-objective optimization model using a multi-objective optimization algorithm to obtain the Pareto optimal path solution set; The route determination and adjustment module is used to determine the target transportation route from the Pareto optimal route solution set and dynamically adjust the target transportation route based on updated real-time dynamic data during transportation.

10. The system according to claim 9, characterized in that, The system also includes data acquisition hardware that is communicatively connected to the data acquisition module; The data acquisition hardware includes a Beidou positioning terminal for collecting real-time location and speed of transport vehicles, traffic monitoring equipment for collecting real-time traffic flow data, and meteorological sensing equipment for collecting real-time meteorological data.

Images