Chaotic annealing ant colony-based adversarial blocking timeliness emergency material transportation toughness intelligent decision-making method

Abstract

The invention discloses a chaos annealing ant colony-based adversarial blocking timeliness emergency material transportation toughness intelligent decision-making method, and relates to the technical field of intelligent emergency scheduling and optimization. The method comprises the steps of 1, extracting scene data and converting the scene data into model parameters; step 2, constructing a mathematical model for survivability optimization; 3, initializing algorithm parameters and generating an initial path; 4, dynamically adjusting the temperature and the pheromone volatilization rate; 5, generating an adversarial blocking perception path and implementing detour repair; 6, updating pheromone distribution based on an annealing criterion; and step 7, locking the critical path and outputting an optimal transportation scheme. According to the method, timely delivery of materials can be guaranteed in a complex road network blocking and emergency demand scene, the timeliness and toughness of an emergency transportation system are improved through a chaos annealing mechanism and an adversarial blocking simulation technology, the method can be widely applied to the fields of natural disaster rescue, public health event response and the like, and a scientific basis is provided for emergency decision making.

Description

Technical Field

[0001] The present invention belongs to the field of intelligent emergency logistics scheduling and optimization technology, and specifically relates to an intelligent decision-making method for emergency material transportation resilience against blocking timeliness based on chaotic annealing ant colonies. The method is particularly suitable for material transportation scenarios with extreme timeliness requirements in complex environments such as natural disaster relief and public health event response. Background Art

[0002] Traditional disaster relief logistics face multiple challenges in dynamic disaster environments. Optimizing a network of thousands of nodes based on fixed planning algorithms takes longer than the physical limit of the 72-hour prime rescue window. Furthermore, adaptive scheduling in the face of extreme disruptions such as power outages, power outages, and internet outages is inefficient. There is an urgent need to integrate various intelligent technologies to build a next-generation emergency logistics decision-making system with spatial and temporal flexibility and resilience.

[0003] From a technical perspective, existing methods and systems still have the following limitations: First, while traditional mathematical programming methods can accurately model constraints, they struggle to meet the demands of real-time response and high timeliness in large-scale, complex scenarios characterized by dynamic road network blockages and sudden increases in demand. Second, existing joint optimization methods for route planning and resource scheduling often focus on a single action space (e.g., discrete route selection) and fail to effectively address the spatiotemporal coupling of collaborative decision-making across multiple resources (vehicles, supplies, routes), leading to difficulties in strategy convergence. Third, most studies fail to quantify timeliness as a dynamic correlation between material delivery time and system performance, and lack robustness design for scheduling solutions in the most adverse scenarios, making it impossible to guarantee the completion rate of critical tasks in extreme environments. For example, when multiple road sections are suddenly blocked, existing algorithms often fall into local optima due to the strong coupling between route replanning and resource reallocation, resulting in a sharp drop in overall transportation efficiency. Summary of the Invention

[0004] In order to solve the problems existing in the background technology, the present invention provides an intelligent decision-making method for emergency material transportation resilience against blocking timeliness based on chaotic annealing ant colonies. This method integrates a hybrid heuristic framework of annealing mechanism and ant colony optimization, combines blocking simulation with chaotic perturbation strategy, and can generate a transportation path planning scheme with high resilience and timeliness in real time under complex road network blocking and sudden demand scenarios, significantly improving the timely delivery rate of high-priority materials. Specifically, this method controls the global search range through dynamic temperature adjustment and pheromone volatilization mechanism, embeds blocking simulation to evaluate path elasticity, and achieves the dual goals of optimizing timeliness and enhancing resilience while ensuring the feasibility of the scheme. To achieve the above objectives, the present invention provides an intelligent decision-making method for emergency material transportation resilience against blocking timeliness based on chaotic annealing ant colonies, comprising the following steps:

[0005] Step 1: Extract data and parameters related to the emergency transport scenario, collecting information on the road network topology, vehicle location, and demand distribution, and converting them into model parameters. By integrating road network data such as node coordinates, road connectivity, and blockage probability, combined with initial vehicle locations and material demand priorities, key parameters of the mathematical model are extracted, providing precise input for subsequent survivability optimization.

[0006] Step 2: Establish the objective function and key constraints for optimizing the resilience of time-sensitive emergency supplies. First, design the objective function to maximize the cumulative utility of high-priority supplies under the worst-case blockage scenario. This ensures that the timeliness requirements of critical supplies are prioritized while allowing for the blockage of up to Z edges. Furthermore, define the key constraint system to ensure the physical feasibility and logical consistency of the vehicle scheduling solution under complex road networks and dynamic interference.

[0007] Step 3: Initialize the algorithm parameters, initialize the pheromone concentration matrix τ, the initial temperature T0, and the chaotic perturbation coefficient. Generate the initial ant population and candidate path set based on the vehicle's initial position and task requirements, and set the sinusoidal fluctuation parameter sin(π·iter / MaxIter) in the temperature update formula to inject chaotic perturbations.

[0008] Step 4: Dynamically adjust the temperature to allow pheromones to volatilize adaptively. By exponentially lowering the temperature and superimposing chaotic perturbations, the global pheromone matrix decays proportionally (1-ρ). High temperatures accelerate exploration, while low temperatures preserve high-quality paths.

[0009] Step 5: Generate routes and perform detour repairs using a blockage-aware approach. Robust routes are dynamically generated by simulating potential blockage threats. First, ants select low-risk sections based on pheromone concentration and edge criticality (the impact of a blockage on the target), proactively avoiding vulnerable paths. If a route fails due to a blocked edge, a detour mechanism is triggered, prioritizing redundant sections with strong alternatives and minimal impact on timeliness, ensuring that high-priority supplies are delivered within the time window.

[0010] Step 6: Incrementally update pheromones based on the adversarial score and annealing acceptance criteria. Optimize pheromone distribution by dynamically evaluating the path's resilience. The adversarial score quantifies the actual delivery benefit of a path after being blocked, prioritizing the pheromone concentration of paths with high scores. The annealing mechanism probabilistically accepts locally inferior solutions to prevent premature convergence. Combined with temperature decay, it dynamically balances global exploration and local optimization, ultimately guiding the ant colony toward more resilient paths.

[0011] Step 7: Periodically protect the critical path. Terminate the algorithm after reaching the maximum number of iterations. Output the optimal path solution that meets the timeliness (time window H) and invulnerability requirements in the worst-case scenario of blocking Z edges.

[0012] In some embodiments, the present invention further includes the following technical features:

[0013] In step 1, the road network topology data, including node coordinates and edge weight distances, is collected to obtain the spatial distribution information of the vehicle initial position coordinates and material demand points, and the material priority weights and time window constraints are simultaneously integrated. Based on the above data, the key parameters of the mixed integer linear programming (MILP) model are further extracted: the vehicle capacity parameter C is set. ks To limit the maximum carrying capacity of each vehicle for material s, calculate the distribution point demand parameter Q ds The gap in the supply of material s at each demand point is quantified, the time window length H is clearly defined as the timeliness benchmark for task execution, the maximum allowable number of blocked edges Z is defined to simulate the worst-case scenario of road network damage, and the priority weight Wds is configured to distinguish the optimized weighting coefficients of high-urgency materials.

[0014] Preferably, the objective function for optimizing the resilience of emergency material dispatch in step 2 is:

[0015] Where W ds represents the priority weight of distribution point d for material s, m kds represents the number of materials s transported by vehicle k to the delivery point d, Q ds represents the total demand for material s at distribution point d, t kd represents the time it takes for vehicle k to arrive at delivery point d, H represents the total time window for task execution, and X represents all decision variables except the opponent variable z. This formula represents the weighted sum of the cumulative effect of different transport task completion degrees on time, based on their priority weights.

[0016] Optionally, the emergency material transportation scheduling constraints include flow balance, access and service relationship, service verification, demand limit constraints, and path connectivity constraints.

[0017] Furthermore, the flow balance constraint for each vehicle k and all non-start and end nodes (i.e., nodes other than the freight yard, distribution point, and vehicle starting point) is that the number of incoming edges of the vehicle is equal to the number of outgoing edges; this ensures path continuity and avoids "disconnection" of vehicles at intermediate nodes. The constraint expression is:

[0018]

[0019] Furthermore, the access and service relationship constraint indicates that if vehicle k transports materials (∑m kds >0), then the node (δ kd =1), and the path contains an edge entering d, the constraint expression is:

[0020]

[0021] Furthermore, the service verification constraint indicates that if vehicle k visits a delivery point d (the path includes an edge entering d), it must deliver materials to that point (∑m kds >0), the constraint expression is:

[0022]

[0023] Furthermore, the demand constraint indicates that the total amount of materials s received by the distribution point d must not exceed its demand quantity Q ds , the constraint expression is:

[0024]

[0025] Furthermore, the path connectivity constraint indicates that if there is a broken edge in the shortest alternative path of edge (i, j), the actual distance disij is set to the maximum value M (whether it is connected is marked by ξij)

[0026]

[0027] The step 3 of initializing the algorithm parameters includes the following sub-steps:

[0028] Step 3.1: Initialize the pheromone concentration matrix τ (the initial value of all paths is the same), initial temperature T0, chaotic perturbation coefficient, and maximum number of iterations;

[0029] Step 3.2: Based on the vehicle's initial position and task requirements, an initial ant population and a set of candidate paths are generated. Starting from the vehicle's current position, a greedy strategy is used to prioritize connecting to nearby high-priority demand points. Combined with random perturbations, a diverse initial path is generated. Each ant initializes its cargo capacity based on the task assignment results (material type and delivery volume), marks nodes with unmet demand, and constructs a preliminary set of candidate paths to cover key transportation tasks in different areas.

[0030] Step 3.3: Set the sinusoidal fluctuation parameter sin(π·iter / MaxIter) in the temperature update formula to inject chaotic perturbations. Superimpose periodic sinusoidal fluctuations on the basic cooling curve. Dynamically adjust the perturbation amplitude by the number of iterations iter to break the monotonic temperature decrease, increase the randomness and diversity of the path search, and prevent the algorithm from converging to a local optimal solution too early.

[0031] The step 4 of dynamic temperature adjustment and pheromone adaptive volatilization includes the following sub-steps:

[0032] Step 4.1: Update the temperature according to the exponential formula T = T0 × μ iterCooling down the temperature. On this basis, superimpose the chaotic perturbation T = T × (1 + chaos_factor sin(π iter / MaxIter)). The global search range is gradually narrowed by the exponential cooling formula. The superimposed chaotic perturbation breaks the temperature monotonicity. The perturbation amplitude is controlled by chaos_factor. The periodic fluctuation of the sine function prevents the algorithm from falling into repeated searches for the local optimal path.

[0033] Step 4.2: Control the volatilization of pheromones according to temperature, using the formula The evaporation rate is calculated, where κ is the evaporation intensity parameter. This ensures that the evaporation rate is high at high temperatures (accelerating the exploration of new paths) and low at low temperatures (stabilizing the pheromone content of high-quality paths). This achieves dynamic adjustment of the evaporation ratio. During the high-temperature phase, the pheromone concentration of inefficient paths is quickly reduced, promoting extensive exploration by the ant colony. During the low-temperature phase, the evaporation rate is slowed down to preserve the pheromone distribution of high-scoring paths, guiding the algorithm to converge towards the local optimum.

[0034] The step 5 of generating and repairing the anti-blocking perception path includes the following sub-steps:

[0035] Step 5.1: Path construction and criticality assessment: For each ant, starting from the initial position of the vehicle, traverse the set of adjacent nodes and calculate the blocking criticality index of each edge (u, v): criticality(u, v) = original path score - score after removing the edge, and generate the edge criticality inverse η = 1 / criticality(u, v); the path selection probability is generated by combining the pheromone concentration τ(u, v), the edge criticality inverse η and the temperature weight T. The next node is selected by roulette and the complete path is constructed iteratively.

[0036] Step 5.2: Update the material delivery status. When the path reaches the delivery node d, perform the material unloading operation according to the task allocation result: the actual delivery volume delivered = min(carried_load[s], unmet_demands[d][s]), update the vehicle cargo volume carried_load[s] - = delivered and the unmet demand unmet_demands[d][s] - = delivered. If the demand is fully met, mark node d as served.

[0037] Step 5.3: Perform a blocking simulation and adopt a flexible detour repair method. Traverse the current path edge set and calculate the objective function loss value loss(u,v) after removing each edge (u,v); select the first Z edges with the largest loss value and add them to the virtual blocking set blocked_edges. If the blocking causes the path to be broken (e.g., u→v is unreachable), call the shortest path algorithm (e.g., Dijkstra) to generate the detour path detour(u,v) and calculate the travel time after the detour. And re-evaluate the delivery volume of high-priority materials within the time window H to generate the post-blocking score adversarial .

[0038] Step 6, which uses adversarial scoring and annealing acceptance criteria to perform pheromone incremental update, includes the following sub-steps:

[0039] Step 6.1: Determine based on the return acceptance criteria. If the current path has an adversarial score, current >score best , directly accept the path and update the historical optimal value score best =score current ; Otherwise, calculate the score difference ΔScore = score current -score best , with probability P accept =e ΔScore / T Accept inferior solution paths to prevent the algorithm from falling into local optimality;

[0040] Step 6.2: Dynamically enhance the pheromone concentration and calculate the accepted path according to the formula Δτ(u,v)=Q·score current Calculate the pheromone increment, where Q is the pheromone intensity coefficient; traverse all edges (u, v) in the path and update their pheromone concentrations to τ(u, v) = τ(u, v) + Δτ(u, v), thus strengthening the pheromone distribution of high-scoring paths.

[0041] Step 6.3: Synchronously update the cargo capacity and demand status. Based on the distribution node d visited in the path, perform material unloading according to the task allocation result: the actual delivery volume delivered = min(carried_load[s], unmet_demands[d][s]), update the vehicle cargo capacity carried_load[s]-=delivered and the unmet demand unmet_demands[d][s]-=delivered. If unmet_demands[d][s]=0, mark the node demand as satisfied.

[0042] The step 7 of periodically protecting the critical path and terminating the iteration includes the following sub-steps:

[0043] Step 7.1: Filter the edge set that satisfies criticality(u,v)>θ from the current optimal path, where criticality(u,v)= Prioritize the path that high-priority materials must pass through, then set the pheromone concentration lock mark for the key edge (u, v), and set τ(u, v) = τ lock (fixed value) or prohibit it from participating in the volatilization operation to avoid chaotic disturbances that destroy the path stability;

[0044] Step 7.2: When the number of iterations iter ≥ Max, terminate the main loop and generate the optimal path plan, which includes the vehicle path sequence, material delivery sequence, and detour plan under blocking scenarios.

[0045] The technical solution provided by the present invention has the following beneficial effects:

[0046] (1) This invention addresses the core defect of traditional emergency logistics scheduling methods, which cannot balance timeliness and survivability in complex emergency scenarios. By constructing an optimization framework based on anti-blocking simulation, it jointly models the material utility time window and the worst-case scenario of road network damage for the first time, significantly improving the dynamic adaptability of the emergency transportation system. In actual disaster relief, this method can ensure that high-priority materials (such as first aid medicines and life detection equipment) are efficiently delivered within the golden rescue period, effectively reducing the risk of secondary disasters caused by material delays; at the same time, its survivability design enables the transportation plan to remain stable in extreme situations such as road blockages and vehicle failures, providing reliable decision-making support for emergency command.

[0047] (2) The present invention breaks through the limitations of a single algorithm through an innovative hybrid heuristic architecture: first, a chaotic annealing mechanism is introduced to embed periodic sinusoidal perturbations into the temperature decay curve, breaking the monotonic convergence mode of the traditional annealing algorithm, significantly improving the global search efficiency, and avoiding path planning from falling into local optimality; second, an anti-blocking scoring model is proposed to dynamically evaluate the resilience of the path under the worst damage scenario using minimax optimization logic, and based on this, the pheromone concentration is updated, so that the path generation process actively avoids highly vulnerable sections; in addition, the critical path protection mechanism is combined with the task-driven detour repair strategy to achieve a balance between computational efficiency and solution quality in large-scale road networks. BRIEF DESCRIPTION OF THE DRAWINGS

[0048] Figure 1 It is the overall flow chart of the method of the present invention;

[0049] Figure 2 This is the architecture diagram of the chaos annealing-ant colony hybrid optimization;

[0050] Figure 3Schematic diagram of the process of generating perception paths and simulating blocking to counteract blocking. DETAILED DESCRIPTION

[0051] The present invention is described in detail below with reference to the accompanying drawings and embodiments.

[0052] In response to the core defects of traditional emergency logistics, such as static path planning being unable to cope with dynamic damage to the road network, the disconnection between material priority and distribution sequence leading to delays in key materials, and the lack of anti-destruction design leading to insufficient mission continuity, the present invention proposes an intelligent decision-making method for the resilience of emergency material transportation against blocking timeliness based on chaotic annealing ant colonies. It adopts dynamic anti-blocking score-driven path optimization and chaotic perturbation-enhanced global search technology. This method greatly improves the delivery rate of high-priority materials within the time window and the system's anti-destruction performance, significantly shortens the response time to sudden demands, and greatly reduces the transportation failure rate caused by blocking.

[0053] like Figure 1 As shown, in one embodiment of the present invention, a method for intelligent decision-making based on chaotic annealing ant colony to resist blocking time-sensitive emergency material transportation resilience includes the following steps:

[0054] Step 1: Extract data and parameters related to emergency transportation scenarios, collect road network topology, vehicle location, and demand distribution information, and convert them into model parameters.

[0055] In this embodiment, the coordinates and connectivity of road nodes in the disaster area are obtained through a geographic information system (GIS), and the road damage risk is analyzed in combination with historical disaster data or real-time satellite imagery to form a dynamically updated road network blockage probability matrix. At the same time, the initial vehicle location information must be obtained in real time through on-board navigation devices or dispatch records from the emergency command center to ensure that the algorithm can perform route planning based on the actual vehicle distribution status.

[0056] Furthermore, according to the demand list provided by the post-disaster rescue command system, the demand quantity (Q ds ), and combined with the urgency of medical institutions in the disaster area or the density of the affected population, assign priority weights (W ds ). Preferably, the priority weight of emergency medicine is set to the highest level, while the weight of general daily necessities is appropriately reduced to reflect the core goal of the emergency task. At the same time, the vehicle capacity parameter (C ks ) must be determined comprehensively based on the vehicle model, cargo space, and material characteristics (such as volume and weight). Preferably, special rescue vehicles are dedicated to transporting large medical equipment, and their capacity parameters must be set separately from those of ordinary transport vehicles;

[0057] Finally, the above parameters are processed through data cleaning and normalization to eliminate dimensional differences and outlier interference, forming standardized inputs.

[0058] Step 2: Establish the objective function and key constraints of the time-sensitive emergency material resilience optimization problem.

[0059] Preferably, the specific method of defining the objective function of the emergency material transportation process is:

[0060]

[0061] Where W ds represents the priority weight of distribution point d for material s, m kds represents the number of materials s transported by vehicle k to the delivery point d, Q ds represents the total demand for material s at distribution point d, t kd represents the time it takes for vehicle k to arrive at delivery point d, H represents the total time window for task execution, and X represents all decision variables except the opponent variable z. This formula represents the weighted sum of the cumulative effect of different transport task completion degrees on time, based on their priority weights.

[0062] Furthermore, traffic balancing, access and service relationships, service verification, demand restriction constraints, and path connectivity constraints are constructed.

[0063] The flow balance constraint is that for each vehicle k and all non-start and end nodes (i.e., nodes other than freight yards, distribution points, and vehicle starting points), the number of incoming edges of the vehicle is equal to the number of outgoing edges; this ensures path continuity and avoids "disconnection" of vehicles at intermediate nodes. The constraint expression is:

[0064]

[0065] The access and service relationship constraint means that if vehicle k transports materials to delivery point d (∑m kds >0), then the node (δ kd =1), and the path contains an edge entering d, the constraint expression is:

[0066]

[0067] The service verification constraint means that if vehicle k visits a delivery point d (the path includes an edge entering d), it must deliver goods to that point (∑m kds >0), the constraint expression is:

[0068]

[0069] The demand constraint means that the total amount of materials s received by the distribution point d must not exceed its demand quantity Q ds , the constraint expression is:

[0070]

[0071] The path connectivity constraint means that if there is a broken edge in the shortest alternative path of edge (i, j), the actual distance dis ij is set to the maximum value M (by ξ ij Mark whether it is connected)

[0072]

[0073] Step 3: Initialization of the hybrid annealing ant colony algorithm is completed by the following sub-steps in the embodiment:

[0074] Step 3.1: Initialize the pheromone concentration matrix τ (with the same initial value for all paths), the initial temperature T0, the chaotic perturbation coefficient, and the maximum number of iterations. Preferably, the initial pheromone value for all paths is set to τ = 1 to ensure fairness in the initial search. The initial temperature parameter T0 is set to 1000, the chaotic perturbation coefficient is set to 0.2, and the maximum number of iterations is limited to 500, thereby balancing the algorithm's convergence speed with its global exploration capability. For example, 10 ambulances in the earthquake zone depart from a randomly distributed starting point and use a greedy strategy to prioritize connecting to nearby high-priority medical points (such as nodes B and E). This strategy also incorporates random perturbations (with a perturbation amplitude of ±15%) to generate a diverse set of candidate paths. For example, an ambulance's initial path is A→D→G→H, with a cargo capacity initialized to a full load of emergency medicine (500 kg). Unmet needs at nodes D and G are marked.

[0075] Step 3.2: Based on the vehicle's initial position and task requirements, an initial ant population and a set of candidate paths are generated. Starting from the vehicle's current position, a greedy strategy is used to prioritize connecting to nearby high-priority demand points. Combined with random perturbations, a diverse initial path is generated. Each ant initializes its cargo capacity based on the task assignment results (material type and delivery volume), marks nodes with unmet demand, and constructs a preliminary set of candidate paths to cover key transportation tasks in different areas.

[0076] Step 3.3: Set the sinusoidal fluctuation parameter sin(π·iter / MaxIter) in the temperature update formula to inject chaotic disturbance, superimpose periodic sinusoidal fluctuations on the basic cooling curve, and dynamically adjust the disturbance amplitude through the number of iterations iter to break the monotonous temperature drop, increase the randomness and diversity of the path search, and avoid the algorithm from converging to the local optimal solution too early. Preferably, in the embodiment, the basic cooling curve is T=1000×0.95 iter Attenuation, while superimposing the chaotic disturbance term ΔT = 0.2×sin(π×iter / 500).

[0077] Step 4: Dynamic temperature adjustment and pheromone adaptive volatilization, in this embodiment, is accomplished through the following sub-steps:

[0078] Step 4.1: The temperature update process uses the dual mechanism of exponential decay and chaotic perturbation to coordinately control the global search behavior. According to the exponential formula T = T0 × 0.95 iter Cooling down, the algorithm gradually narrows the search range to focus on high-quality paths, and at the same time superimposes the chaotic disturbance term T = T × (1 + chaos_factor sin (π iter / MaxIter)), breaking the monotonous downward trend of temperature through the periodic fluctuation of the sine function. The disturbance amplitude is dynamically adjusted by the chaos coefficient chaos_factor to avoid the algorithm falling into the trap of repeated search for the local optimal path due to a single temperature change, thereby ensuring path diversity. In an embodiment, preferably, based on the initial temperature T0 = 1000 and the chaotic disturbance coefficient chaos_factor = 0.2, the temperature update formula is T = 1000 × 0.95 iter ×(1+0.2·sin(π·iter / 500)), where the maximum number of iterations MaxIter=500. For example, when iter=100, the base temperature drops to 1000×0.95 100 ≈5.92. After superimposing the sinusoidal disturbance, the actual temperature T=5.92×(1+0.2·sin(0.2π))≈6.85, and the disturbance amplitude is 15.7%.

[0079] Step 4.2: By formula Calculate the volatilization rate. The volatilization rate is higher in the high temperature stage (initial stage of the algorithm), which accelerates the dissipation of pheromones in inefficient paths and forces the ant colony to explore new areas. In the embodiment, in the high temperature stage (such as iter = 0, T = 1000), preferably, the volatilization intensity parameter is 0.1, and the volatilization rate κ is The pheromone matrix decays quickly (retention rate 90%), which drives the ant colony to explore new paths extensively; in the low temperature stage (such as iter = 400, T ≈ 1000 × 0.95 400 ≈0.52), the evaporation rate dropped to The pheromone retention rate has increased to 99.3%, stabilizing the concentration distribution along high-quality routes. For example, along the ambulance route A→B→E, the initial pheromone concentration τ=1 drops to 0.9 after evaporation during the high-temperature phase, and only drops to 0.993 during the low-temperature phase, significantly reducing the risk of pheromone loss along critical routes.

[0080] Step 5, generating a blocking-resistant perception path and performing detour repair, is implemented in the embodiment according to the following steps:

[0081] Step 5.1: During the path construction and criticality assessment process, each ant starts from the initial position of the vehicle, traverses the set of adjacent nodes, and calculates its blocking criticality index by evaluating the impact of removing each road section on the overall score. It is defined as the difference between the original path score and the score after removing the edge, and takes the inverse to generate the edge criticality weight parameter. Combining the real-time pheromone concentration, edge criticality weight and temperature adjustment factor, a path selection probability model is constructed, and the next node is dynamically selected through a roulette mechanism to gradually generate a complete transport path. Among them, the path selection probability is calculated by normalizing the polynomial product of the pheromone concentration, edge criticality weight and temperature term to ensure that high-priority sections are explored first while retaining path diversity. In an embodiment, if Figure 3 As shown in the figure, the earthquake zone ambulance starts from node A as an example. First, the initial path A→B→E→H is generated, and its timeliness score is Σm / Q*(Ht current )=120, the critical blocking indexes of each edge are calculated as AB:50, BE:65, and EH:70.

[0082] Step 5.2: During the path execution process, when the vehicle arrives at the designated delivery node, the system performs the material unloading operation according to the task allocation results. The actual delivery volume is dynamically adjusted based on the current vehicle cargo capacity and the minimum value of the node's unmet demand, and the vehicle's remaining cargo capacity and node demand status are updated in real time. If all the demands of a node are met, it will be marked as served to avoid resource waste caused by repeated delivery. Through this mechanism, the system can accurately adapt to dynamic demand changes and ensure the timely delivery of high-priority materials. In this embodiment, the ambulance travels along the path A→B→E→H, unloading 200kg of first aid medicine at node B, reducing the cargo capacity from 500kg to 300kg, and the demand at node B is reset and marked as served; after arriving at node E, the remaining 300kg is unloaded, the cargo capacity returns to zero, and node E is marked as served. Finally, unloading is skipped at node H due to insufficient cargo capacity.

[0083] Step 5.3: Traverse all road segments in the vehicle's current path, perform edge removal simulations one by one, calculate the loss value of the objective function after removing each edge (i.e., the difference between the original path score and the score after blocking), and select the first Z edges with the largest loss value to add to the virtual blocking set, such as Figure 3 As shown. When the blocking operation causes the path to be broken, the Dijkstra shortest path algorithm is called to generate an alternative detour path based on the real-time road network topology and calculate the travel time after the detour. In the embodiment, for the ambulance path A→B→E→H, after blocking the most critical edge E→H (loss value 70), the system calls the Dijkstra algorithm to generate a detour path E→D→G→H, and the detour travel time increases to t detour , recalculate the delivery volume of high-priority materials within the time window, and the delivery score of node G due to detour delay drops to Σm / Q*(H–tdetour )=100, the total score retention rate after blocking is still 83.3% (original score 120), meeting the timeliness constraint.

[0084] Step 6: Using adversarial scoring and annealing acceptance criteria to perform pheromone incremental updates is achieved through the following sub-steps:

[0085] Step 6.1: Annealing acceptance criteria is determined based on the comparison between the current path adversarial score and the historical optimal value. If the current path adversarial score is current >score best , directly accept the path and update the historical optimal value score best =score current ; Otherwise, calculate the score difference ΔScore = score current -score best , with probability P accept =e ΔScore / T Accept the inferior solution path to avoid the algorithm falling into the local optimum; in the embodiment, the adversarial score of the ambulance path A→B→E→H is 1340.7, and the historical optimal value is 1320. Since the current score is higher, the optimal value is directly accepted and updated; if the score is inferior to the optimal value, the algorithm is updated with probability e (735-1340.7) / 6.85 Reject this path to ensure that the search direction converges to the high-efficiency solution.

[0086] Step 6.2: Dynamic enhancement of pheromone concentration For the accepted path, the increment Δτ(u,v) = Q score is calculated by multiplying the pheromone intensity coefficient Q by the current score. current , traverse all edges (u, v) in the path, and add increments to their pheromone concentration τ(u, v) = τ(u, v) + Δτ(u, v). Through this mechanism, the pheromone concentration of high-scoring paths is significantly enhanced, guiding subsequent ant populations to prioritize exploring high-quality paths while retaining the possibility of exploring low-frequency paths to maintain global search capabilities.

[0087] Step 6.3: Cargo Capacity and Demand Status Updated Synchronously Based on the route execution results, material unloading operations are performed on the visited delivery nodes according to the assigned quantity. The actual delivery quantity is dynamically determined by the minimum of the vehicle's remaining cargo capacity and the node's unmet demand. Vehicle cargo status and node demand data are updated in real time, and the node is marked as out of service when the demand is fully met.

[0088] Step 7: Periodic protection of the critical path and iteration termination are achieved through the following sub-steps:

[0089] Step 7.1: The critical path locking mechanism selects the edge set whose blocking criticality exceeds the preset threshold (θ = 50) from the current optimal path, where the blocking criticality is defined as the difference between the original path score and the blocked score (criticality(u,v) = score orig -score blocked ). For the selected key edges, a pheromone concentration lock flag is set to force their pheromone values ​​to remain fixed or prohibit them from participating in the volatilization operation, so as to avoid the destruction of the stability of the critical path by chaotic perturbations. In this embodiment, the edges B→E (criticality=65) and E→H (criticality=70) in the optimal path A→B→E→H for the ambulance are locked, and the pheromone concentrations are fixed to τ lock =100, prohibiting the volatilization and update operations in subsequent iterations to ensure the stability of the core rescue channel in the earthquake zone; the non-critical edge A→D (criticality=5) still participates in the dynamic optimization to retain the path flexibility.

[0090] Step 7.2: When the number of iterations reaches the preset upper limit, the main loop is terminated and the optimal path solution is output. The solution includes the vehicle path sequence, the material delivery time sequence, and a detour plan in the case of blockage. In the worst-case scenario where Z = 3 edges are blocked, the algorithm outputs the path A→D→G→H and its detour plan (e.g., detour D→E→G when D→G is blocked). Verification shows that the timeliness score after the detour is 100 (original score 120), with a retention rate of 83%, meeting the emergency requirement for high-priority material delivery within the time window H = 72 hours. It also provides delivery timestamps for each node, providing an executable anti-destruction transportation plan for rescue command.

Claims

1. A method for intelligent decision-making on emergency material transportation resilience against blocking timeliness based on chaotic annealing ant colony, characterized by: The following steps are involved: (1) Extract the road network topology data, vehicle initial location, and material demand distribution information of the emergency transportation scenario and convert them into mathematical model parameters; (2) Construct an objective function that maximizes the cumulative utility of high-priority materials under the most unfavorable blocking scenario and defines path connectivity and task timing constraints; (3) Initialize the pheromone concentration matrix, initial temperature and chaotic perturbation parameters to generate the initial ant population and candidate path set; (4) A dynamic temperature adjustment mechanism based on the superposition of exponential cooling and chaotic disturbances to control the global pheromone volatilization rate; (5) Adopting a path generation method that is resistant to blockage perception, combined with edge criticality assessment and roulette wheel selection strategy, a low-risk transportation path is constructed, and a detour repair mechanism is triggered after blockage; (6) Dynamically update the pheromone concentration according to the annealing acceptance criterion and the adversarial score to enhance the search probability of high-resistance paths; (7) Periodically lock the pheromone concentration of the critical path and iteratively output the optimal transportation plan that meets the requirements of timeliness and indestructibility.

2. The method according to claim 1, characterized in that The dynamic temperature adjustment mechanism in step (4) satisfies: The temperature update formula is T=T0×μ iter ×(1+chaos_factor·sin(π·iter / MaxIter)), where T0 is the initial temperature, μ is the cooling coefficient, chaos_factor is the chaos disturbance coefficient, iter is the current iteration number, and MaxIter is the maximum iteration number. The pheromone volatilization rate is dynamically adjusted according to the temperature, and the volatilization coefficient is calculated as Where κ is the volatility intensity parameter.

3. The method according to claim 1, characterized in that The method for generating an anti-blocking perception path in step (5) includes: (a) Calculate the blocking criticality of each edge in the path, defined as the difference between the original path score and the score after removing the edge; (b) The path selection probability is a combination of pheromone concentration, inverse edge criticality, and temperature weight, and is expressed as Where τ(u,v) is the pheromone concentration of edge (u,v), η(u,v) = 1 / criticality(u,v), α, β, γ are weight parameters; (c) If the path fails due to blocked edges, the shortest path algorithm is called to generate a detour path and the material delivery volume within the time window is re-evaluated.

4. The method according to claim 1, wherein The annealing acceptance criterion in step (6) is: if the adversarial score S of the current path current Better than the best historical rating S best , then directly accept the path; otherwise, P accept =e ΔScore / T Accept the inferior solution path and update the pheromone concentration along the path by pheromone increment, where Q is the pheromone intensity coefficient.

5. The method according to claim 1, wherein The critical path protection mechanism in step (7) includes: (a) Filter the set of edges whose blocking criticality exceeds the threshold from the optimal path; (b) Lock the pheromone concentration of the key edge and prohibit it from participating in volatilization or disturbance operations to ensure path stability.

6. The method according to claim 1, characterized in that The objective function is: Among them, W ds is the priority weight of distribution point d for material s, m kds is the number of materials s transported by vehicle k to distribution point d, Q ds is the total demand for material s at distribution point d, t kd is the time it takes for vehicle k to arrive at delivery point d, and H is the total time window length for task execution.

7. The method according to claim 1, characterized in that The method further includes: during the route construction process, dynamically updating the material delivery status based on the vehicle cargo capacity and the distribution point demand, ensuring that high-priority materials are unloaded first, and marking the nodes where the demand has been met.

8. The method according to claim 1, characterized in that The chaotic disturbance parameter is injected through a periodic sine function to break the monotonous temperature decrease law and enhance the global search diversity.

9. The method according to claim 1, characterized in that The pheromone matrix is ​​initialized using a uniform distribution strategy to ensure the fairness and exploration breadth of the initial path search.

10. The method according to claim 3, characterized in that In the detour repair mechanism, the travel time of the detour path is calculated as: Among them, detour is the detour path distance, v k is the speed of vehicle k, t current The current accumulated time.

Images