A snow removal vehicle path planning method considering demand splittability and road segment criticality

Abstract

The application discloses a snow removal vehicle path planning method considering demand divisibility and road section criticality, belongs to the technical field of expressway snow removal operation scheduling, and particularly relates to a snow removal vehicle path planning method considering demand divisibility and road section criticality. The application aims to solve the technical prejudice that a traditional limited-capacity arc path optimization model regards each road section as an indivisible whole service unit, the limitation that existing methods regard vehicle starting points as fixed parameters, the defect that existing critical road section identification methods rely on a single index, and the defect that existing path planning methods lack actual executability checking. The process is as follows: obtaining target road network data; calculating road section comprehensive criticality scores based on the target road network data; constructing a snow removal vehicle path planning model based on the comprehensive criticality scores of the road sections; solving the snow removal vehicle path planning model by adopting a tabu search algorithm based on a forest representation method, and outputting optimal starting point positions and vehicle paths.

Description

Technical Field

[0001] This invention belongs to the field of highway snow removal operation scheduling technology, specifically involving a snow removal vehicle route planning method that considers the decomposability of demand and the criticality of road sections. Background Technology

[0002] Winter snow removal is a crucial task for ensuring road traffic safety and maintaining urban operations. Northern Chinese cities experience frequent snowfall; if ice and snow are not cleared from roads promptly, the road surface friction coefficient will significantly decrease, easily leading to traffic accidents and congestion. Currently, urban road snow removal typically employs a "mechanical-based" approach, with commonly used equipment including snowplows, snow brushes, and snow spreaders. In practice, different cities have developed differentiated snow removal organization models. The "zoning and contracting" model allows each area to operate independently, simplifying management but making it difficult to balance the workload. The "centralized dispatch" model allows vehicles to depart from a unified assembly point, covering a wide area, but the empty driving distance is highly correlated with the location of the assembly point. Currently, the selection of assembly points largely relies on the experience and judgment of managers, lacking quantitative support for site selection.

[0003] In the field of snow removal path planning, the classic Capacity-Limited Arc Path Optimization (CARP) problem is a crucial theoretical foundation. However, existing methods generally treat each road segment as an indivisible, integral service unit, assuming that a vehicle can complete all snow removal work in a single pass. This assumption deviates from reality. Urban roads typically contain multiple lanes, and a snowplow can only clear one lane at a time. Clearing all lanes on the same road segment requires the same vehicle to pass multiple times or multiple vehicles to work collaboratively. Existing models, treating road segments as the smallest service unit, cannot handle scenarios involving multi-lane, segmented services.

[0004] In assessing the importance of road segments, existing research has explored various perspectives. For example, Tu et al. proposed a critical road segment identification method based on path flow weighted betweenness centrality; Cui et al. studied network repair sequence optimization based on resilience; Feng et al. studied emergency resource deployment considering road segment criticality under severe weather conditions; and Leng Junqiang et al. proposed the ISB-BPR function to characterize the impact of snow and ice conditions on road segment capacity. However, most of these methods are based on a single indicator dimension or rely solely on network topology, traffic flow load, or vulnerability scanning. A single indicator cannot comprehensively characterize the overall criticality of a road segment under snowfall disturbances: a high-topology hub segment may not be fully loaded in actual demand, a high-load segment may not be a bridge in the topology, and a highly vulnerable segment may also have the characteristics of low-topology traffic flow and no alternative paths. Existing methods lack an evaluation system that organically integrates the three dimensions of structure, function, and vulnerability.

[0005] Furthermore, most existing studies treat the vehicle's starting point as a given fixed parameter, neglecting the coupling relationship between starting point selection and path planning. Different starting points will lead to completely different optimal path structures. In actual operations, if the road sections traversed during a relocation trip have not yet been cleared of snow, vehicles will face the risk of being trapped or being run over again. However, existing models generally lack modeling and checking mechanisms for this engineering constraint. Summary of the Invention

[0006] The purpose of this invention is to address the technical bias of traditional limited-capacity arc path optimization models that treat each road segment as an indivisible whole service unit, the limitations of existing methods that treat the vehicle's starting point as a fixed parameter, the defects of existing critical road segment identification methods that rely on a single indicator, and the defects of existing path planning methods that lack practical feasibility checks. In response, this invention proposes a snow removal vehicle path planning method that considers demand decomposability and road segment criticality.

[0007] A snow removal vehicle route planning method that considers demand decomposability and road segment criticality is as follows:

[0008] Step 1: Obtain target road network data;

[0009] Step 2: Calculate the comprehensive criticality score of road segments based on the target road network data;

[0010] Step 3: Based on the comprehensive criticality score of the road segment, construct a snow removal vehicle route planning model;

[0011] Step 4: Use the tabu search algorithm based on forest representation to find the snow removal vehicle path planning model, and output the optimal starting point position and the path of each vehicle.

[0012] The beneficial effects of this invention are as follows:

[0013] To address the aforementioned technical deficiencies, this invention provides a snow removal vehicle route planning method that considers the decomposability of demand and the criticality of road segments: Snow removal demand for road segments is broken down into independent demand units based on the number of lanes for refined modeling; a comprehensive evaluation system for critical road segments is constructed from three dimensions: structure, function, and vulnerability; starting point selection and route planning are incorporated into a unified optimization framework; and an empty-run route legality check mechanism is introduced to ensure that vehicles run empty on already served road segments, guaranteeing the actual feasibility of the route plan.

[0014] In order to overcome the technical bias of traditional limited-capacity arc path optimization models that treat each road segment as an indivisible whole service unit, this invention establishes a lane-level demand-divisible modeling method, so that the model can truly reflect the actual situation of snow removal operations on multi-lane roads.

[0015] In order to overcome the limitations of existing methods that treat the vehicle's starting point as a fixed parameter, this invention incorporates starting point selection and path planning into a unified optimization framework to achieve coordinated optimization of strategic-level decision-making and tactical-level scheduling.

[0016] To overcome the shortcomings of existing critical road segment identification methods that rely on a single indicator, this invention constructs a comprehensive evaluation system from three dimensions: structure, function, and vulnerability, to achieve accurate positioning of critical road segments that restrict the performance of snowfall road networks.

[0017] To overcome the shortcomings of existing route planning methods that lack practical feasibility checks, this invention introduces an empty-run route legality check mechanism to ensure that vehicles run empty on already served road sections, thus guaranteeing the feasibility of the planning scheme in actual operation.

[0018] This invention discloses a snow removal vehicle route planning method that considers demand decomposability and road segment criticality. The method includes: acquiring road topology data, traffic demand data, and snowfall intensity data of the target road network, and obtaining a set of candidate starting points within the snow removal operation area; calculating the boundary number centrality of each road segment as a structural importance indicator based on the road topology data; calculating the functional importance indicator of road segments; when real-time road segment traffic flow is unavailable, calculating the traffic flow percentage of each road segment under snowfall conditions using a user-balanced allocation method based on the OD matrix; when real-time road segment traffic flow is available, directly using the real-time traffic flow percentage; reducing the capacity of the target road segment by a snowfall reduction factor and setting it to zero to simulate snow blockage, then reallocating traffic and calculating the relative increase in total network travel time as a vulnerability indicator to quantify road segment vulnerability. The performance degradation after the effect was measured; the three indicators were weighted and fused using the entropy weight method to obtain the comprehensive critical score for each road segment; the snow removal demand of each road segment was divided into multiple demand units according to the number of lanes, and a dual-objective optimization model was established with minimizing the weighted total travel time as the first objective and minimizing the weighted completion time as the second objective. Vehicle starting point selection and vehicle route planning were incorporated into a unified optimization framework. The model includes demand satisfaction constraints, starting point selection constraints, vehicle capacity constraints, and traffic balance constraints; the model was solved using a tabu search algorithm based on forest representation. The algorithm adopts a two-layer parallel framework of outer enumeration of candidate starting points and inner tabu search path optimization, integrates multiple neighborhood operators, and introduces empty route legality checks to ensure that vehicles do not pass through unserved road segments during empty driving; the optimal starting point position and the path of each vehicle are output. This invention overcomes the technical shortcomings of traditional arc path optimization, such as the indivisible road segments, fixed vehicle starting positions, and lack of comprehensive quantitative analysis of key road segments. It accurately locates key road segments through three-dimensional fusion identification, closely approximates actual snow removal operation scenarios through lane-level modeling, and ensures the actual feasibility of the path through legality checks. This invention can significantly reduce the empty driving time of snow removal vehicles and ensure priority service for key road segments. Attached Figure Description

[0019] Figure 1 This is a flowchart of the present invention;

[0020] Figure 2 This is a schematic diagram of the Sioux Falls road network topology in the embodiment;

[0021] Figure 3 A diagram showing the clustering results for hierarchical classification;

[0022] Figure 4 Score distribution and cluster center plot;

[0023] Figure 5 This is a map showing the critical distribution of road segments based on a multi-dimensional fusion strategy.

[0024] Figure 6 This is a diagram of the FTS-Snow algorithm framework.

[0025] Figure 7 The driving path of vehicle 1, whose starting point is node 8;

[0026] Figure 8 This is the driving path diagram for vehicle 2, whose starting point is node 8;

[0027] Figure 9 The driving path of vehicle 3, whose starting point is node 8;

[0028] Figure 10 The driving path of vehicle 4, whose starting point is node 8;

[0029] Figure 11 This is the driving path diagram for vehicle 5, whose starting point is node 8. Detailed Implementation

[0030] Specific implementation method one: Combining Figure 1 This embodiment describes a snow removal vehicle route planning method that considers both demand decomposability and road segment criticality. The specific process is as follows:

[0031] Step 1: Obtain target road network data;

[0032] Step 2: Calculate the comprehensive criticality score of road segments based on the target road network data;

[0033] Step 3: Based on the comprehensive criticality score of the road segment, construct a snow removal vehicle route planning model;

[0034] Step 4: Use the tabu search algorithm based on forest representation to find the snow removal vehicle path planning model, and output the optimal starting point position and the path of each vehicle.

[0035] Specific Implementation Method Two: This implementation method differs from Specific Implementation Method One in that: in step 1, the target road network data is acquired; the specific process is as follows:

[0036] Obtain the directed graph of the highway. ;

[0037] in, For a set of nodes, It is a set of directed edges;

[0038] The nodes are highway entrance and exit junctions; the directed edges are road segments between nodes.

[0039] Obtain the OD demand matrix for each node; obtain highway snowfall intensity data. ;

[0040] Obtain the set of candidate starting points within the highway snow removal operation area. , .

[0041] The other steps and parameters are the same as in Specific Implementation Method 1.

[0042] Specific Implementation Method Three: This implementation method differs from Specific Implementation Method One or Two in that: in step 2, the comprehensive criticality score of the road segment is calculated based on the target road network data; the specific process is as follows:

[0043] Step 21: Calculate each directed edge Structural importance index Functional importance indicators and vulnerability indicators ; ;

[0044] Step 22: Use the entropy weight method to weight and fuse the structural importance index, functional importance index, and vulnerability index of each directed edge to obtain the comprehensive criticality score of each directed edge;

[0045] Step 23: Based on the comprehensive criticality score of each directed edge, the K-Means clustering algorithm is used to divide all road segments into five criticality levels.

[0046] Other steps and parameters are the same as in specific implementation method one or two.

[0047] Specific Implementation Method Four: This implementation method differs from Specific Implementation Methods One to Three in that: in step 21, each directed edge is calculated... Structural importance index Functional importance indicators and vulnerability indicators ; The specific process is as follows:

[0048] Step 211: Calculate the structural importance index for each directed edge. The specific process is as follows:

[0049] The structural importance metric is measured by edge betweenness centrality, which is defined as the proportion of the number of shortest paths between all pairs of nodes in the network that pass through that edge.

[0050] (1)

[0051] In the formula, Indicates the structural importance index of each directed edge; Represents a node With nodes The total number of all shortest paths between them; Represents a node With nodes Among all shortest paths between them, the one that passes through the edge Quantity;

[0052] Step 212: Calculate each directed edge Functional importance indicators The specific process is as follows:

[0053] The functional importance index is the proportion of the traffic flow of this road segment to the total traffic flow of all road segments in the road network, which is used to reflect the actual traffic load borne by the road segment under snowfall conditions;

[0054] There are two ways to obtain functional importance indicators:

[0055] 1) When real-time traffic flow for road segments cannot be obtained, the user-balanced allocation method is used to calculate the traffic flow for each road segment based on the OD demand matrix of each node. Traffic flow under snowfall conditions ;

[0056] The user equalization allocation uses the ISB-BPR function as the road segment impedance function. The ISB-BPR function introduces a snowfall intensity parameter based on the classic BPR function. The form of the ISB-BPR function is:

[0057] (2)

[0058] In the formula, Data on snowfall intensity on highways is expressed in terms of snowfall amount. Indicates road segment Travel time under snowfall conditions; This function represents the effect of snowfall conditions on free travel time. ; This function represents the impact of snowfall conditions on the traffic capacity of a road segment. ; Indicates road section under normal conditions Free time for travel; Indicates road section under normal conditions Traffic flow; Indicates road segment Traffic capacity; , Indicates the parameters of the BPR function;

[0059] by road section Traffic flow under snowfall conditions The proportion of total traffic flow on all road segments in the road network under snowfall conditions is used as a functional importance indicator; expressed as:

[0060] (3)

[0061] In the formula, For road section Traffic flow under snowfall conditions; For road section Traffic flow under snowfall conditions;

[0062] It is the set of directed edges, that is, the set of all road segments in the road network;

[0063] 2) When real-time traffic flow for road segments is available, use the road segment directly. The proportion of real-time traffic flow under snowfall conditions to the total real-time traffic flow of all road segments in the road network under snowfall conditions is used as a functional importance indicator; expressed as:

[0064] (4)

[0065] Step 213: Calculate each directed edge Vulnerability indicators The specific process is as follows:

[0066] The vulnerability index employs a "road segment failure simulation" approach. Under snowfall conditions, the traffic capacity of the target road segment is completely reduced to zero (simulating snow blockage), and after a user rebalancing, the relative increase in total network travel time is calculated. This index directly reflects the marginal destructive power of a road segment; the higher the value, the more severe the impact of road segment failure on system performance, and the higher its priority in snowfall emergency management.

[0067] Vulnerability Indicators The calculation formula is:

[0068] (5)

[0069] In the formula, This represents the baseline total travel time for all road segments in the road network during snowfall.

[0070] For road section After being disturbed by snowfall, the road section With the capacity completely reset to zero, the total travel time of all road segments in the road network is obtained by re-evaluating the user-balanced allocation method based on the OD demand matrix of each node.

[0071] The other steps and parameters are the same as those in one of the specific implementation methods one to three.

[0072] Specific Implementation Method Five: This implementation method differs from Specific Implementation Methods One through Four in that: the baseline total travel time for all road segments in the road network under snowfall weather conditions is... The acquisition process is as follows:

[0073] (1): Initialization; the specific process is as follows:

[0074] Assuming the road network is completely idle and the initial flow of all road segments is 0, that is... ; For road section Traffic flow under snowfall conditions;

[0075] The initial impedance of each road segment is the free-flow time under snowfall conditions. ;by As a road obstacle, Dijkstra's algorithm is used to solve for the road segment between each node pair. The current shortest path between them;

[0076] in, Indicates the first One node; Indicates the first One node;

[0077] Based on the all-or-nothing approach, all node pairs are considered in the road segments. Total demand (Obtain the OD demand matrix for each node) Load all of them into the corresponding node pairs The initial traffic flow for all road segments under snowfall conditions is obtained by summing up the current shortest path (without taking any alternative paths). Set the current iteration number. ;

[0078] (2): Update the current travel time (impedance) of each segment; the specific process is as follows:

[0079] Number of iterations The obtained road section Traffic flow under snowfall conditions Substitute the values ​​into the ISB-BPR function and recalculate the actual travel time (impedance) of each road segment in the current road network according to formula (2). ;

[0080] (3): Generate auxiliary traffic flow, compare the traffic flow in (2) with the auxiliary traffic flow. If the two are exactly the same (or the difference is very small), it means that the road network is in equilibrium. Jump directly to (7) to output the result; if the two are different, execute (4).

[0081] The specific process is as follows:

[0082] The result calculated in (2) Treat it as a constant impedance and ignore the marginal effect of the increase in flow rate on the impedance;

[0083] With the latest impedance To address the road obstruction, Dijkstra's algorithm is used again to recalculate the road segments between each node pair in the current state. The shortest path; all node requirements All traffic is loaded onto the new shortest path, thus obtaining a new set of traffic flows for all road segments under snowfall conditions, denoted as auxiliary flow. ;

[0084] Compare and ;

[0085] If the two are exactly the same (or the difference is very small), it means that the road network is in equilibrium and directly jump to (7) to output the result;

[0086] If the two are not the same, execute (4);

[0087] (4): Now we have the current traffic and auxiliary traffic We need to determine from Towards How many steps should you take to move forward? This makes the next step of traffic The corresponding objective function decreases the most;

[0088] Find the optimal step size Make Beckmann objective function Take the minimum value; the specific process is as follows:

[0089] Constructing the Beckmann objective function ; indicates as:

[0090] (6)

[0091] in ;

[0092] Indicates the step size; , This represents the parameters of the BPR function; the default value is... , ; Represents variables, ;

[0093] right Find the derivative, set it to 0, and you will get the step size. For optimal step size ;

[0094] The univariate equation that needs to be solved is:

[0095] (7)

[0096] (5): Update traffic flow in the road segment; the specific process is as follows:

[0097] The exact optimal step size obtained from (4) Substituting into the traffic update formula below, calculate the segment traffic for the next iteration:

[0098] (8)

[0099] in: Indicates the number of iterations The obtained road section Traffic flow on road sections under snowfall conditions;

[0100] Indicates the number of iterations The obtained road section Traffic flow on road sections under snowfall conditions;

[0101] (6): Convergence test; the specific process is as follows:

[0102] It is necessary to quantitatively determine whether the current solution is close enough to the user's equilibrium state; in engineering, the relative gap index is used.

[0103] Calculate the actual total travel time :

[0104] (9)

[0105] in, Indicates will The road segment obtained by substituting into the ISB-BPR function Travel time under snowfall conditions;

[0106] Calculate the ideal shortest total time :

[0107] (10)

[0108] in, Based on the latest impedance The obtained node pairs of road segments Shortest path time;

[0109] Calculate the relative gap:

[0110] (11)

[0111] like ( The preset threshold is usually set to... If the algorithm converges, the iteration stops and the process moves to (7).

[0112] like ,make Return to (2) and continue iterating until... If the algorithm converges, the iteration stops and the process proceeds to (7).

[0113] (7): Output the final road segment flow rate and final road segment travel time. Based on the final road segment flow rate and final road segment travel time, derive the baseline total travel time for all road segments in the road network under snowfall weather. The specific process is as follows:

[0114] Output the final segment flow after convergence. ;

[0115] Output the final segment flow after convergence. Corresponding final route travel time ;

[0116] Utilize the final segment flow after convergence and the corresponding final route travel time Substituting into the formula for the total travel time of the entire network, we can obtain the baseline total travel time for all road segments in the road network under snowy weather conditions. ; indicates as:

[0117] (12).

[0118] The other steps and parameters are the same as in any of the specific implementation methods one to four.

[0119] Specific Implementation Method Six: This implementation method differs from Specific Implementation Methods One through Five in that the total travel time for all road segments in the road network is... The acquisition process is as follows:

[0120] (1): Initialization; the specific process is as follows:

[0121] Section After being disturbed by snowfall, the road section The traffic capacity is completely reduced to zero, and the road section is... Removed from the road network, the remaining road segments are completely idle, and their initial flow is 0. ; For road section Traffic flow under snowfall conditions;

[0122] At this point, the initial impedance of each road segment is the free-flow time under snowfall conditions. ;by As a road obstacle, Dijkstra's algorithm is used to solve for the road segment between each node pair. The current shortest path between them;

[0123] Based on the all-or-nothing approach, all node pairs are considered in the road segments. Total demand Load all to the corresponding node pairs The initial traffic flow for all road segments under snowfall conditions is obtained by summing up the current shortest path (without taking any alternative paths). Set the current iteration number. ;

[0124] (2): Update the current travel time (impedance) of each segment; the specific process is as follows:

[0125] Number of iterations The obtained road section Traffic flow under snowfall conditions Substitute the values ​​into the ISB-BPR function and recalculate the actual travel time (impedance) of each road segment in the current road network according to formula (2). ;

[0126] (3): Generate auxiliary traffic flow, compare the traffic flow in (2) with the auxiliary traffic flow. If the two are exactly the same (or the difference is very small), it means that the road network is in equilibrium. Jump directly to (7) to output the result; if the two are different, execute (4).

[0127] (4): Now we have the current traffic and auxiliary traffic We need to determine from Towards How many steps should you take to move forward? This makes the next step of traffic The corresponding objective function decreased the most.

[0128] Find the optimal step size Make Beckmann objective function Take the minimum value;

[0129] (5): Update traffic flow on road sections;

[0130] (6): Convergence test; the specific process is as follows:

[0131] It is necessary to quantitatively determine whether the current solution is close enough to the user's equilibrium state; in engineering, the relative gap index is used.

[0132] Calculate the actual total travel time according to formula (9). ; Calculate the ideal shortest total time according to formula (10) ;

[0133] Based on actual total travel time and the ideal shortest total time The relative gap is calculated according to formula (11). ;

[0134] like ( The preset threshold is usually set to... If the algorithm converges, the iteration stops and the process moves to (7).

[0135] like ,make Return to (2) and continue iterating until... If the algorithm converges, the iteration stops and the process proceeds to (7).

[0136] (7): Output the final road segment flow rate and final road segment travel time. Based on the final road segment flow rate and final road segment travel time, derive the baseline total travel time for all road segments in the road network under snowfall weather. The specific process is as follows:

[0137] Output the final segment flow after convergence. ;

[0138] Output the final segment flow after convergence. Corresponding final route travel time ;

[0139] The final converged traffic flow and the corresponding final route travel time Substituting into the formula for total network travel time, we can obtain the total network travel time. ; indicates as:

[0140] (13).

[0141] The other steps and parameters are the same as those in any of the specific implementation methods one to five.

[0142] Specific Implementation Method Seven: This implementation method differs from Specific Implementation Methods One through Six in that: in step 22, the entropy weight method is used to weight and fuse the structural importance index, functional importance index, and vulnerability index of each directed edge to obtain a comprehensive criticality score for each directed edge; the specific process is as follows:

[0143] Step 221: Normalize the structural importance index, functional importance index, and vulnerability index of each directed edge using the min-maximum method; expressed as:

[0144] (14)

[0145] in, Table 1 The first section of the road Normalized results of each indicator; Indicates the first The first section of the road Individual indicator values; Indicates the first of all road segments Individual indicator values; , Indicators representing structural importance Indicators representing functional importance Indicators of vulnerability; , Indicates the total number of road segments;

[0146] Step 222: Based on the normalized result of each directed edge, calculate the proportion value of each directed edge; expressed as:

[0147] (15)

[0148] in, Indicates the first The first section of the road The proportion values ​​of each indicator;

[0149] Step 223: Based on the proportion of each directed edge, calculate the entropy value of each index; expressed as:

[0150] (16)

[0151] in, Indicates the first The entropy value of each indicator;

[0152] Step 224: Calculate the weights of each indicator based on the entropy value; expressed as:

[0153] (17)

[0154] in, Indicates the first The weight of each indicator; Indicates the total number of indicators. ;

[0155] Step 225: Based on the weights of each indicator and the normalized results of each indicator for each road segment, calculate the comprehensive criticality score for each road segment. ; indicates as:

[0156] (18).

[0157] The other steps and parameters are the same as those in any of the specific implementation methods one to six.

[0158] Specific Implementation Method Eight: This implementation method differs from Specific Implementation Methods One through Seven in that: in step 3, a snow removal vehicle route planning model is constructed based on the comprehensive criticality score of the road segment; the specific process is as follows:

[0159] Each directed edge Snow removal requirements are broken down by the number of lanes. Each demand unit corresponds to one lane;

[0160] in, Indicates the first One node; Indicates the first One node;

[0161] Establish a snow removal vehicle route planning model with minimizing the weighted total travel time as the primary objective and minimizing the weighted completion time as the secondary objective.

[0162] The primary objective is to minimize the weighted total travel time.

[0163] The weighted total travel time is the sum of the weighted time of all vehicles performing snow removal operations on all road segments and the weighted time of all vehicles traveling empty through all road segments; expressed as:

[0164] (19)

[0165] (20)

[0166] The second objective is to minimize the weighted completion time;

[0167] The weighted completion time is the sum of the products of the overall criticality score for each road segment and the completion time when all lanes of the corresponding road segment have been cleared of snow; expressed as:

[0168] (twenty one)

[0169] in, The total number of vehicles; For nodes and nodes Intersection The overall key score; For nodes and nodes Intersection Travel time for snow removal operations; For empty driving through the node and nodes Intersection Travel time; For nodes and nodes Intersection Length; The speed at which a snowplow operates during snow removal; This refers to the empty speed of the snowplow; For vehicles At the node and nodes Intersection The number of service lanes on the road; For vehicles Passing through the node empty and nodes Intersection The number of times; For nodes and nodes Intersection The completion time is when all lanes have been cleared of snow.

[0170] The other steps and parameters are the same as those in any of the specific implementation methods one to seven.

[0171] Specific Implementation Method Nine: This implementation method differs from Specific Implementation Methods One through Eight in that the constraints of the snow removal vehicle path planning model include:

[0172] 1) Demand satisfaction constraint: All lanes on each directed edge are fully served, and each lane is served exactly once; expressed as:

[0173] (twenty two)

[0174] in, For vehicles At the node and nodes Intersection The number of service lanes on the road; For nodes and nodes Total number of lanes;

[0175] 2) Starting point selection constraint: From the set of candidate starting points We select exactly one node as the departure and final return point for all vehicles; Represents 0 and 1 variables, if candidate node If selected as a parking lot, then ;otherwise ; ; indicates as:

[0176] (twenty three)

[0177] 3) Vehicle Departure and Return Constraints: Each vehicle departs from the selected departure point only once and returns only to the corresponding departure point; represented as:

[0178] (twenty four)

[0179] (25)

[0180] in, For vehicles From node Departure to Node Number of empty runs; For fixed nodes From node Departure to Node ; For vehicles From node Departure to Node Number of empty runs;

[0181] 4) Flow balance constraint: The number of entries and exits at each node in each vehicle path is the same; expressed as:

[0182] (26)

[0183] in, For vehicles From node Departure to Node Number of empty runs; For vehicles From node Departure to Node Number of empty runs; For vehicles At the node and nodes Intersection The number of service lanes on the road; For vehicles At the node and nodes Intersection The number of service lanes on the road; For fixed nodes From node Departure to Node ; For fixed nodes From node Departure to Node ;

[0184] 5) Capacity-flow conservation constraint: The change in the number of remaining serviceable lanes after a vehicle has left a service lane satisfies the conservation relationship; expressed as:

[0185] (27)

[0186] in, Indicates vehicle Start working on the nodes and nodes Intersection Before providing service or empty runs, located at the node The number of remaining serviceable lanes at the time of the incident; Indicates vehicle Start working on the nodes and nodes Intersection Before providing service or empty runs, located at the node The number of remaining serviceable lanes at the time of the incident; For fixed nodes From node Departure to Node ;

[0187] 6) Total vehicle capacity constraint: The total number of lanes served by each vehicle during the entire operation trip shall not exceed the maximum number of lanes that the vehicle itself can serve. ; indicates as:

[0188] (28)

[0189] 7) Legality constraint of empty driving route: All road segments traversed by the vehicle on the empty driving route between any two services and on the empty driving route back to the starting point of the last service must be road segments that have been cleared of snow.

[0190] The other steps and parameters are the same as those in one of the specific implementation methods one to six.

[0191] Specific Implementation Method Ten: This implementation method differs from Specific Implementation Methods One through Nine in that: in step 4, a tabu search algorithm based on forest representation is used to solve the de-snow vehicle path planning model, outputting the optimal starting point position and the path of each vehicle; the specific process is as follows:

[0192] Step 41: Set the number of parking lots to the directed graph of the highway area to be cleared of snow. The number of nodes;

[0193] Step 42: Process each node to obtain the final path scheme for each node;

[0194] Step 43: Based on the final path scheme of each node, obtain the optimal starting point location and the path of each vehicle;

[0195] In step 42, each node is processed to obtain the final path scheme for each node; the specific process is as follows:

[0196] Step 421: Given a starting point (e.g., node 8), first generate a feasible initial solution; the specific process is as follows:

[0197] Step 4211: Establish a waiting pool: Put all lanes that require snowplow operation into a "waiting pool";

[0198] Step 4212: Create an empty vehicle: The empty vehicle departs from a given starting point (e.g., node 8), and the remaining number of lanes that the empty vehicle can serve is set to... ;

[0199] Step 4213: Greedy insertion: Randomly select a lane (corresponding to a directed edge, such as edge 15→10) from the waiting service pool; calculate the cost increment of inserting the randomly selected lane into all possible positions in the current vehicle's path;

[0200] Cost increment Including the increased operation time due to inserting into new lanes And the increased empty driving time due to the change in empty driving path caused by inserting into a new lane. ; indicates as:

[0201] (29)

[0202] Step 4214: Capacity Check: After inserting into a lane, check if the current vehicle's cumulative service lanes are less than or equal to... ;

[0203] If the current cumulative service lanes for vehicles are less than or equal to Then proceed with steps 4213-4214;

[0204] If the current vehicle's cumulative service lanes are greater than If the current vehicle is not assigned a lane, steps 4212-4214 are repeated until all lanes in the service pool are assigned a lane; the initial route plan is then obtained. ;

[0205] Step 422: Apply the initial vehicle path scheme obtained in step 421 The evaluation indicators are calculated; the specific process is as follows:

[0206] Step 4221: Calculate the initial principal objective ; indicates as:

[0207] (30)

[0208] Step 4222: Calculate the initial secondary objective ; indicates as:

[0209] (31)

[0210] Step 4223: Change the current initial path scheme Set as the globally historical best solution Save the primary objective value corresponding to the globally best historical solution. Secondary target value And the complete path sequence, and reset the continuous no-improvement counter to zero;

[0211] Step 423: Process the initial path scheme obtained in Step 421 to obtain the final path scheme (after all snow has been cleared from the area to be processed); the specific process is as follows:

[0212] Step 4231: Set the number of iterations ;

[0213] Step 4232: Randomly select one of the Relocate, Exchange, Swap(1,1) and One-Split operators to process the initial path scheme obtained in Step 421, and obtain a new path scheme. The specific process is as follows:

[0214] Relocate operator: Cut a lane from the path of car A and insert it into a certain position in the path of car B (or another position of car A itself).

[0215] Exchange operator: Swaps the positions of one lane of car A and one lane of car B;

[0216] The Swap(1,1) operator randomly selects an edge, for example, when going from node 1 to node 2. It scans all vehicle paths and finds the set of vehicles currently serving the edge from node 1 to node 2, such as car A and car B. It reads the service counts of car A and car B respectively (A=2, B=1). The Swap(1,1) operator moves one lane at a time. The candidate solution is to reduce car A by 1 lane and add car B by 1 lane. In car A's path solution, one lane in the edge from node 1 to node 2 is deleted, and the order of the remaining edges in car A's path solution remains unchanged. In car B's path solution, this lane is added to the original edge position from node 1 to node 2, and the order of the remaining edges remains unchanged. After the modification is performed, the current cumulative number of lanes for car A and car B is calculated. If it is found that the cumulative number of lanes for car B exceeds the capacity limit of 30, the path solution is restored to the candidate solution (car A reduces 1 lane and car B adds 1 lane), and the candidate solution is marked as invalid.

[0217] One-Split operator:

[0218] 1) Select the target road segment (e.g., with 3 lanes), and delete all vehicles A and B currently serving on the target road segment from the service tasks; at this time, the target road segment is in a no-vehicle service state, and the remaining service lanes of vehicles A and B increase, thus clarifying the current upper limit of the remaining service lanes for each vehicle.

[0219] 2) Calculate the incremental cost required to re-insert a lane of the target road segment into each vehicle's path; sort the vehicles in order of increasing cost increment, and prioritize assigning the lane to the vehicle with the smallest cost increment.

[0220] 3) Repeat step 2) until all lane assignments for vehicles without service status are completed;

[0221] Step 4233: Feasibility review; the specific process is as follows:

[0222] The new path scheme obtained in step 4232 If review criteria 1, 2, and 3 are not met simultaneously, the application will be discarded and not considered.

[0223] The new path scheme obtained in step 4232 If all three conditions (Review 1, Review 2, and Review 3) are met, the new path option will be retained. Execute step 4234; the specific process is as follows:

[0224] Review 1, Requirement Satisfaction Review: Check if the total number of times each edge is served is exactly equal to the number of its lanes;

[0225] Review 2, Capacity Limitation Review: Check whether the total number of lanes served by each vehicle is less than or equal to 30;

[0226] Review 3, Legality Review of Empty Driving: The program simulates a forward progression along the timeline, assuming car A needs to start from node [node name missing]. Arrive at the node empty Go to the next path and extract the nodes. With nodes Find the shortest path between two points, and check whether all segments on the shortest path have been served.

[0227] Step 4234: Calculate the objective function value; the specific process is as follows:

[0228] If step 4233 retains the new path scheme The feasibility review was passed, and the new primary target value was recalculated using the exact same calculation formula as in step 422. and new sub-target values ;

[0229] Step 4235: The new path scheme retained from step 4233 Filtering is performed to obtain valid path schemes that pass the tabu check; the specific process is as follows:

[0230] The taboo list is defined as follows: If an operation in a new path solution (e.g., "removing edge 43 from vehicle A") is recorded and not yet de-banned, then the new path solution is temporarily prohibited from adoption (to prevent the operation from being moved out and then moved back, resulting in a loop). For example, suppose that in the 100th iteration, the algorithm successfully removes edge 43 from vehicle A's path through an operation, the system will immediately write the number "100" at the (vehicle A, edge 43) position in the register, meaning that vehicle A can no longer perform snow removal services on edge 43.

[0231] Extract the new path scheme retained in step 4233 Find the corresponding unique feature code (e.g., the number of the moved edge + the number of the source vehicle + the number of the destination vehicle) and look it up in the taboo table.

[0232] If the signature is not in the taboo list, or the remaining taboo period is less than or equal to 0: the path scheme corresponding to the signature is allowed to proceed normally to step 4236;

[0233] If the signature is in the tabu list and the remaining tabu period is greater than 0, then compare... Compared to the current global historical best ;

[0234] like Execute the amnesty; the amnesty is to force the release of the path scheme corresponding to the feature code and proceed to step 4236, and remove the corresponding feature code from the taboo table;

[0235] To prevent vehicles from being blacklisted for moving back to the original lane, the "reverse movement feature code" (i.e., the record of moving an edge from the current vehicle back to the original vehicle) is added to the blacklist during the first iteration. The tabu list is empty at this stage. When the algorithm performs neighborhood operations (step 4232), the operation (e.g., moving road segment 43 from vehicle 1 to vehicle 2) is recorded. This operation is not in the tabu list, and this solution (i.e., the path scheme) can then be calculated in step 4236. After the first iteration, the algorithm enters step 4237, extracting the reverse feature code of the successful passage operation from the first iteration (i.e., moving road segment 43 from vehicle 2 to vehicle 1) and inserting it into the tabu list. At this point, the tabu list is no longer empty. A new tabu list is generated with each subsequent iteration.

[0236] like Execute tabu; executing tabu means forcibly rejecting the path scheme corresponding to the current feature code from entering step 4236, directly discarding the path scheme corresponding to the current feature code, jumping back to step 4232, randomly selecting the next operator again, and generating another brand new candidate solution.

[0237] Note that the iteration count of the main program is not incremented by 1 at this time. This is because when the path scheme is intercepted by taboo, it jumps directly to the entry point of the last iteration count judgment in the loop, skipping the increment of the iteration count. Therefore, the iteration count of the main program is not incremented by 1.

[0238] Step 4236: Grant the valid path scheme approved in step 4235 Compared with the current path scheme Begin by comparing the advantages and disadvantages; the specific process is as follows:

[0239] Current primary target value and secondary target value It is based on the current path scheme The calculation process is the same as step 422.

[0240] Current path scheme The source is divided into two stages: the initial generation stage and the iterative update stage;

[0241] Initial generation phase, This is equivalent to the initial path scheme obtained in step 421. ;

[0242] During the iterative update phase, step 4236 will be updated. The specific update process is as follows:

[0243] If the new primary target value Unconditionally accept the legal path scheme approved in step 4235, and set the current path scheme. And reset the continuously unimproved counter to zero; The legal path scheme approved in step 4235;

[0244] If the new primary target value The valid path scheme approved in step 4235 is rejected, and the current path scheme is maintained. constant;

[0245] If the new primary target value Compare the new secondary target values and ;like Set the path scheme before The counter that has not improved continuously will be reset to zero; if Maintain the current path scheme constant;

[0246] Step 4237: If the new route plan was accepted in step 4236 (Acceptance is in step 4236) Once the current path is updated to the new path, step 4237 (updating the taboo table) can be performed to extract the new path. The reverse operation signature of the neighborhood operator (e.g., the forward direction is from car A to car B, and the reverse direction is from car B to car A); insert the reverse signature into the tabu list, update the tabu list, and set the remaining tabu period, the length of the remaining tabu period is equal to the length of the initial tabu period, the length of the initial tabu period is set manually; traverse all existing old signatures in the tabu list, and decrement the remaining tabu period of all old signatures by 1;

[0247] The algorithm initially sets a forbidden period length as the initial forbidden period, which is manually set. The remaining forbidden period is calculated by traversing the tabu list after each iteration, decrementing the forbidden period value of all records by 1. For example, if an operation is written to the tabu list in the 10th iteration, its remaining forbidden period equals the initial forbidden period. The forbidden period length is simply a number.

[0248] If step 4236 does not accept the new path scheme No reverse operation signature is extracted, and the tabu list remains unchanged; all existing records with remaining tabu periods are decremented by 1.

[0249] Step 4238: Update the globally optimal path scheme; the specific process is as follows:

[0250] Compare the current path with the new path accepted in step 4236. Current primary target value Current secondary target value With the optimal principal objective value Optimal secondary objective value ;

[0251] If the current global historical best primary objective value is Then the globally historical optimal path scheme will be... And update the new main target simultaneously. New secondary target value ;

[0252] If the current global historical best The original plan will remain unchanged.

[0253] If the current global historical best Compare secondary target values and ;like The globally optimal historical path scheme ;like The original plan will remain unchanged.

[0254] Step 4239: If the optimal path scheme in step 4238 has been updated, reset the no-improvement counter to zero;

[0255] If the optimal path scheme is not updated in step 4238, increment the continuous no-improvement counter by one, and determine whether the counter has reached the threshold of fifty.

[0256] If the condition is met, a perturbation operation is performed, the process of which is as follows:

[0257] First, disable the query function of the tabu table. All subsequent operations will not be intercepted by the tabu table, nor will any new records be written to the tabu table.

[0258] Then, a lane is randomly cut from one vehicle's path and randomly inserted into the path of another vehicle, and this process is repeated cumulatively. (Five to ten) times;

[0259] Finally, the structure of the current path scheme is completely broken down, a new current path scheme is generated, and the counter is reset to zero;

[0260] If the target is not met, no disturbance operation will be performed, and the process will proceed to step 42310.

[0261] Step 42310: Determine if the current iteration count has reached the set maximum iteration count limit; if not, set the iteration count to zero. Repeat steps 4232 to 4239; if the target is reached, exit the loop and obtain the final path solution.

[0262] The other steps and parameters are the same as those in any of the specific implementation methods one to nine.

[0263] The beneficial effects of the present invention are verified using the following embodiments:

[0264] Example 1:

[0265] This embodiment uses the Sioux Falls standard road network as an example to describe the method of the present invention in detail. Figure 2 As shown, the road network consists of 24 nodes and 76 directed road segments. It is a classic benchmark network in the field of traffic assignment research, and the OD demand matrix of the nodes contains 528 OD pairs.

[0266] Step 1: Obtain target road network data.

[0267] Obtain the directed graph G=(V,A) of the Sioux Falls road network, where V contains 24 nodes and A contains 76 directed road segments. Obtain the network topology file, containing basic attribute information such as node number, road segment direction, road segment length, capacity, free-flow travel time, and number of lanes; obtain the OD demand matrix, containing travel demand data for 528 OD pairs between the 24 nodes. List all 24 nodes as candidate departure points.

[0268] The snowfall intensity was set to I = 4 mm (moderate snowfall). The ISB-BPR function was used to simulate the reduction effect of snowfall on road capacity and the increase effect on free-flow travel time. The parameters of the ISB-BPR function were set as follows: , .

[0269] Step 2: Calculate the overall criticality score of the road segment.

[0270] To address the blind spots in identification based on a single indicator, this invention employs a three-dimensional objective evaluation system of "structure-function-vulnerability".

[0271] Calculate the structural importance index: for each directed edge The edge betweenness centrality is calculated according to formula (1). For the Sioux Falls road network, the edge betweenness centrality is distributed in the interval [0.015, 0.080], with a mean of 0.040 and a standard deviation of 0.012.

[0272] Functional Importance Indicators: Since real-time traffic flow for road segments cannot be obtained in this embodiment, a user-balanced allocation based on the OD matrix is ​​used to calculate functional importance indicators. Following the steps in Implementation Method Five, the OD demand matrix is ​​allocated to the road network to obtain the traffic flow for each road segment under balanced conditions. Then, the traffic flow percentage of each road segment is calculated according to formula (3). For the Sioux Falls road network, the traffic flow of each road segment is mainly distributed in the low-to-medium load range of 5000-11000 pcu / h, and the traffic flow of some core arterial roads exceeds 21000 pcu / h.

[0273] Calculate the vulnerability index: for each directed edge Calculated according to specific implementation methods five and six and Then, the vulnerability index is calculated according to formula (5). For the Sioux Falls road network, The values ​​range from 0.02 to 0.65, with a mean of 0.25 and a standard deviation of 0.24. Approximately 70% of the road sections... Less than 0.25, for a few highly vulnerable road sections More than 0.49.

[0274] Entropy weighting fusion: According to specific implementation method seven, the minimum-maximum method is used to normalize the three indicators. The normalized indicator values ​​are converted into a proportion matrix. The entropy value of each indicator is calculated based on the proportion matrix. The weight of each indicator is calculated based on the entropy value of each indicator. The comprehensive score is calculated based on the weight of each indicator and the normalized indicator value. The mean of the comprehensive score for the entire network is 0.267, and the standard deviation is 0.212.

[0275] The K-Means clustering algorithm was used to divide the 76 road segments into five key levels. The clustering results are as follows: Figure 3 , Figure 4 As shown, low-level road sections (Level 1 and 2) account for approximately 66%, while high-criticality roads (Level 4 and 5) account for 13.16%. The top 10 critical road sections are shown in Table 1.

[0276]

[0277] Figure 5 This map shows the distribution of road segment classifications across the entire network. Level 5 road segments. The east-west main corridor, located in the core area of ​​the network, is the most vulnerable "lifeline" of the entire network, and its overall score is much higher than that of other sections.

[0278] Step 3: Construct a lane-level demand decomposable model.

[0279] The 76 directed road segments of the Sioux Falls road network were mapped to lanes of 1-4 based on capacity, totaling 143 lane demand units. Snow removal operation speed. Take 15 km / h as the empty speed. Set the speed to 30 km / h. Set the number of vehicles. Single vehicle capacity The decision variable selects the optimal one from 24 candidate starting points as the starting and return points for all vehicles.

[0280] Establish a dual-objective optimization model:

[0281] The first objective (minimizing the weighted total travel time) is to minimize the sum of the weighted time of all vehicles performing snow removal operations on all road segments and the weighted time of all vehicles traveling empty through all road segments.

[0282]

[0283] The second objective (weighted completion time minimization) is to minimize the sum of the products of the overall criticality score of each road segment and the completion time when all lanes of that road segment are cleared of snow.

[0284]

[0285] in The road segment calculated in step 2 Overall key score.

[0286] Step 4: Use the tabu search algorithm based on forest representation to find the snow removal vehicle path planning model, and output the optimal starting point position and the path of each vehicle.

[0287] The framework diagram of the tabu search algorithm based on forest representation is as follows: Figure 6 As shown in Table 2:

[0288]

[0289] The above solution process was executed for each of the 24 candidate starting points, and the calculation results are shown in Table 3:

[0290]

[0291] As shown in Table 3, node 8 has the lowest weighted total time of 14.29 hours and a weighted completion time of 118.88 hours, making it the optimal vehicle starting point in this example. Node 8 is located in the central-to-core area of ​​the Sioux Falls road network and has good accessibility to highly critical road sections and major work areas.

[0292] Table 4 shows the travel routes of the five snowplows when the starting point is set at the optimal starting point node 8. Each of the five snowplows is responsible for serving lanes 26-30, completing the clearing work for a total of 143 lane demand units, resulting in balanced vehicle load. Because lane-level demand modeling is used, the same road segment is allowed to appear multiple times in the same vehicle path, verifying that the demand decomposable model can more realistically reflect the actual snow removal operation process.

[0293]

[0294] Note: → indicates that snow removal vehicles are operating on this section of road; → indicates that snow removal vehicles are only passing through this section of road without operating on it.

[0295] Figures 7 to 11 The visualization results of the vehicle routes for each option starting from point 8 are shown. Figures 7 to 11 As can be seen, the five paths exhibit a distinct fan-shaped radial structure, with each vehicle responsible for a different direction of the road network area, and the paths intersect relatively little.

[0296] The present invention may have other embodiments. Without departing from the spirit and essence of the present invention, those skilled in the art can make various corresponding changes and modifications according to the present invention, but these corresponding changes and modifications should all fall within the protection scope of the appended claims.

Claims

1. A snow removal vehicle route planning method that considers demand decomposability and road segment criticality, characterized in that: The specific process of the method is as follows: Step 1: Obtain target road network data; Step 2: Calculate the comprehensive criticality score of road segments based on the target road network data; Step 3: Based on the comprehensive criticality score of the road segment, construct a snow removal vehicle route planning model; Step 4: Use the tabu search algorithm based on forest representation to find the snow removal vehicle path planning model, and output the optimal starting point position and the path of each vehicle.

2. The snow removal vehicle route planning method considering demand decomposability and road segment criticality as described in claim 1, characterized in that: In step 1, the target road network data is obtained; the specific process is as follows: Obtain the directed graph of the highway. ; in, For a set of nodes, It is a set of directed edges; The nodes are highway entrance and exit junctions; the directed edges are road segments between nodes. Obtain the OD demand matrix for each node; Obtaining data on snowfall intensity on highways ; Obtain the set of candidate starting points within the highway snow removal operation area. , .

3. The snow removal vehicle route planning method considering demand decomposability and road segment criticality as described in claim 2, characterized in that: In step 2, the comprehensive criticality score of the road segment is calculated based on the target road network data; The specific process is as follows: Step 21: Calculate each directed edge Structural importance index Functional importance indicators and vulnerability indicators ; ; Step 22: Use the entropy weight method to weight and fuse the structural importance index, functional importance index, and vulnerability index of each directed edge to obtain the comprehensive criticality score of each directed edge; Step 23: Based on the comprehensive criticality score of each directed edge, the K-Means clustering algorithm is used to divide all road segments into five criticality levels.

4. A snow removal vehicle route planning method considering demand decomposability and road segment criticality as described in claim 3, characterized in that: In step 21, each directed edge is calculated. Structural importance index Functional importance indicators and vulnerability indicators ; ; The specific process is as follows: Step 211: Calculate the structural importance index for each directed edge. The specific process is as follows: (1) In the formula, Indicates the structural importance index of each directed edge; Represents a node With nodes The total number of all shortest paths between them; Represents a node With nodes Among all shortest paths between them, the one that passes through the edge Quantity; Step 212: Calculate each directed edge Functional importance indicators The specific process is as follows: There are two ways to obtain functional importance indicators: 1) When real-time traffic flow for road segments cannot be obtained, the user-balanced allocation method is used to calculate the traffic flow for each road segment based on the OD demand matrix of each node. Traffic flow under snowfall conditions ; The user equalization allocation uses the ISB-BPR function as the segment impedance function, and the ISB-BPR function has the following form: (2) In the formula, This indicates data on snowfall intensity on highways; Indicates road segment Travel time under snowfall conditions; This function represents the effect of snowfall conditions on free travel time. ; This function represents the impact of snowfall conditions on the traffic capacity of a road segment. ; Indicates road section under normal conditions Free time for travel; Indicates road section under normal conditions Traffic flow; Indicates road segment Traffic capacity; , Indicates the parameters of the BPR function; by road section Traffic flow under snowfall conditions The proportion of total traffic flow on all road segments in the road network under snowfall conditions is used as a functional importance indicator; expressed as: (3) In the formula, For road section Traffic flow under snowfall conditions; For road section Traffic flow under snowfall conditions; It is the set of directed edges, that is, the set of all road segments in the road network; 2) When real-time traffic flow for road segments is available, use the road segment directly. The proportion of real-time traffic flow under snowfall conditions to the total real-time traffic flow of all road segments in the road network under snowfall conditions is used as a functional importance indicator; expressed as: (4) Step 213: Calculate each directed edge Vulnerability indicators The specific process is as follows: Vulnerability Indicators The calculation formula is: (5) In the formula, This represents the baseline total travel time for all road segments in the road network during snowfall. For road section After being disturbed by snowfall, the road section With the capacity completely set to zero, the total travel time of all road segments in the road network is obtained by re-allocating users using the user-balanced allocation method based on the OD demand matrix of each node.

5. A snow removal vehicle route planning method considering demand decomposability and road segment criticality as described in claim 4, characterized in that: The baseline total travel time for all road sections in the road network under snowfall weather conditions. The acquisition process is as follows: (1): Initialization; the specific process is as follows: Assuming the road network is completely idle and the initial flow of all road segments is 0, that is... ; For road section Traffic flow under snowfall conditions; The initial impedance of each road segment is the free-flow time under snowfall conditions. ; by As a road obstacle, Dijkstra's algorithm is used to solve for the road segment between each node pair. The current shortest path between them; in, Indicates the first One node; Indicates the first One node; Based on the all-or-nothing approach, all node pairs are considered in the road segments. Total demand Load all to the corresponding node pairs The initial traffic flow for all road segments under snowfall conditions is obtained by summing up the current shortest path. Set the current iteration number. ; (2): Update the current travel time for each route segment; the specific process is as follows: Number of iterations The obtained road section Traffic flow under snowfall conditions Substitute the values ​​into the ISB-BPR function and recalculate the actual travel time for each road segment in the current road network according to formula (2). ; (3): Generate auxiliary traffic flow, compare the traffic flow from (2) with the auxiliary traffic flow. If they are exactly the same, it means that the road network is in equilibrium, and jump directly to (7) to output the result; if they are not the same, execute (4); the specific process is as follows: Calculated by (2) Consider it as a constant impedance; With the latest impedance To address the road obstruction, Dijkstra's algorithm is used again to recalculate the road segments between each node pair in the current state. The shortest path; all node requirements All traffic is loaded onto the new shortest path, thus obtaining a new set of traffic flows for all road segments under snowfall conditions, denoted as auxiliary flow. ; Compare and ; If the two are exactly the same, it means that the road network is in a balanced state, and we can directly jump to (7) to output the result; If the two are not the same, execute (4); (4): Finding the optimal step size Make Beckmann objective function Take the minimum value; the specific process is as follows: Constructing the Beckmann objective function ; indicates as: (6) in ; Indicates the step size; , Indicates the parameters of the BPR function; Represents variables, ; right Find the derivative, set it to 0, and you will get the step size. For optimal step size ; The univariate equation that needs to be solved is: (7) (5): Update traffic flow in the road segment; the specific process is as follows: The exact optimal step size obtained from (4) Substituting into the traffic update formula below, calculate the segment traffic for the next iteration: (8) in: Indicates the number of iterations The obtained road section Traffic flow on road sections under snowfall conditions; Indicates the number of iterations The obtained road section Traffic flow on road sections under snowfall conditions; (6): Convergence test; the specific process is as follows: Calculate the actual total travel time : (9) in, Indicates will The road segment obtained by substituting into the ISB-BPR function Travel time under snowfall conditions; Calculate the ideal shortest total time : (10) in, Based on the latest impedance The obtained node pairs of road segments Shortest path time; Calculate the relative gap: (11) like If the algorithm converges, the iteration stops and the process proceeds to (7). like ,make Return to (2) and continue iterating until... If the algorithm converges, the iteration stops and the process proceeds to (7). (7): Output the final road segment flow rate and final road segment travel time. Based on the final road segment flow rate and final road segment travel time, derive the baseline total travel time for all road segments in the road network under snowfall weather. The specific process is as follows: Output the final segment flow after convergence. ; Output the final segment flow after convergence. Corresponding final route travel time ; Utilize the final segment flow after convergence and the corresponding final journey time Substituting into the formula for the total travel time of the entire network, we can obtain the baseline total travel time for all road segments in the road network under snowy weather conditions. ; indicates as: (12)。 6. A snow removal vehicle route planning method considering demand decomposability and road segment criticality as described in claim 5, characterized in that: Total travel time for all road segments in the aforementioned road network The acquisition process is as follows: (1): Initialization; the specific process is as follows: Section After being disturbed by snowfall, the road section The traffic capacity is completely reduced to zero, and the road section is... Removed from the road network, the remaining road segments are completely idle, and their initial flow is 0. ; For road section Traffic flow under snowfall conditions; At this point, the initial impedance of each road segment is the free-flow time under snowfall conditions. ;by As a road obstacle, Dijkstra's algorithm is used to solve for the road segment between each node pair. The current shortest path between them; All nodes between road sections Total demand Load all to the corresponding node pairs The initial traffic flow for all road segments under snowfall conditions is obtained by summing up the current shortest path. Set the current iteration number. ; (2): Update the current travel time for each route segment; The specific process is as follows: Number of iterations The obtained road section Traffic flow under snowfall conditions Substitute the values ​​into the ISB-BPR function and recalculate the actual travel time for each road segment in the current road network according to formula (2). ; (3): Generate auxiliary flow, compare the traffic flow in (2) with the auxiliary flow. If the two are exactly the same, it means that the road network is in equilibrium. Jump directly to (7) to output the result. If the two are not the same, execute (4); (4): Finding the optimal step size Make Beckmann objective function Take the minimum value; (5): Update traffic flow on road sections; (6): Convergence test; the specific process is as follows: Calculate the actual total travel time according to formula (9). ; Calculate the ideal shortest total time according to formula (10). ; Based on actual total travel time and the ideal shortest total time The relative gap is calculated according to formula (11). ; like If the algorithm converges, the iteration stops and the process proceeds to (7). like ,make Return to (2) and continue iterating until... If the algorithm converges, the iteration stops and the process proceeds to (7). (7): Output the final road segment flow and the final road segment travel time. Based on the final road segment flow and the final road segment travel time, the baseline total travel time of all road segments in the road network under snowfall weather is obtained. The specific process is as follows: Output the final segment flow after convergence. ; Output the final segment flow after convergence. Corresponding final route travel time ; The final converged traffic flow and the corresponding final journey time Substituting into the formula for total network travel time, we can obtain the total network travel time. ; indicates as: (13)。 7. A snow removal vehicle route planning method considering demand decomposability and road segment criticality as described in claim 6, characterized in that: In step 22, the entropy weight method is used to weight and fuse the structural importance index, functional importance index and vulnerability index of each directed edge to obtain the comprehensive criticality score of each directed edge. The specific process is as follows: Step 221: Normalize the structural importance index, functional importance index, and vulnerability index of each directed edge using the min-maximum method; expressed as: (14) in, Table 1 The first section of the road Normalized results of each indicator; Indicates the first The first section of the road Individual indicator values; Indicates the first of all road segments Individual indicator values; , Indicators representing structural importance Indicators representing functional importance Indicators of vulnerability; , Indicates the total number of road segments; Step 222: Based on the normalized result of each directed edge, calculate the proportion value of each directed edge; expressed as: (15) in, Indicates the first The first section of the road The proportion values ​​of each indicator; Step 223: Based on the proportion of each directed edge, calculate the entropy value of each index; expressed as: (16) in, Indicates the first The entropy value of each indicator; Step 224: Calculate the weights of each indicator based on the entropy value; expressed as: (17) in, Indicates the first The weight of each indicator; Indicates the total number of indicators. ; Step 225: Based on the weights of each indicator and the normalized results of each indicator for each road segment, calculate the comprehensive criticality score for each road segment. ; indicates as: (18)。 8. A snow removal vehicle route planning method considering demand decomposability and road segment criticality as described in claim 7, characterized in that: In step 3, a snow removal vehicle route planning model is constructed based on the comprehensive criticality score of the road segment; the specific process is as follows: Each directed edge Snow removal requirements are broken down by the number of lanes. Each demand unit corresponds to one lane; in, Indicates the first One node; Indicates the first One node; Establish a snow removal vehicle route planning model with minimizing the weighted total travel time as the primary objective and minimizing the weighted completion time as the secondary objective. The primary objective is to minimize the weighted total travel time. The weighted total travel time is the sum of the weighted time of all vehicles performing snow removal operations on all road segments and the weighted time of all vehicles traveling empty through all road segments; expressed as: (19) (20) The second objective is to minimize the weighted completion time; The weighted completion time is the sum of the products of the overall criticality score for each road segment and the completion time when all lanes of the corresponding road segment have been cleared of snow; expressed as: (21) in, The total number of vehicles; For nodes and nodes Intersection The overall key score; For nodes and nodes Intersection Travel time for snow removal operations; For empty driving through the node and nodes Intersection Travel time; For nodes and nodes Intersection Length; The speed at which a snowplow operates during snow removal; This refers to the empty speed of the snowplow; For vehicles At the node and nodes Intersection The number of service lanes on the road; For vehicles Passing through the node empty and nodes Intersection The number of times; For nodes and nodes Intersection The completion time is when all lanes have been cleared of snow.

9. A snow removal vehicle route planning method considering demand decomposability and road segment criticality as described in claim 8, characterized in that: The constraints of the snow removal vehicle path planning model include: 1) Demand satisfaction constraint: All lanes on each directed edge are fully served, and each lane is served exactly once; expressed as: (22) in, For vehicles At the node and nodes Intersection The number of service lanes on the road; For nodes and nodes Total number of lanes; 2) Starting point selection constraint: From the set of candidate starting points We select exactly one node as the departure and final return point for all vehicles; Represents 0 and 1 variables, if candidate node If selected as a parking lot, then ;otherwise ; ; indicates as: (23) 3) Vehicle Departure and Return Constraints: Each vehicle departs from the selected departure point only once and returns only to the corresponding departure point; represented as: (24) (25) in, For vehicles From node Departure to Node Number of empty runs; For fixed nodes From node Departure to Node ; For vehicles From node Departure to Node Number of empty runs; 4) Flow balance constraint: The number of entries and exits at each node in each vehicle path is the same; expressed as: (26) in, For vehicles From node Departure to Node Number of empty runs; For vehicles From node Departure to Node Number of empty runs; For vehicles At the node and nodes Intersection The number of service lanes on the road; For vehicles At the node and nodes Intersection The number of service lanes on the road; For fixed nodes From node Departure to Node ; For fixed nodes From node Departure to Node ; 5) Capacity-flow conservation constraint: The change in the number of remaining serviceable lanes after a vehicle has left a service lane satisfies the conservation relationship; expressed as: (27) in, Indicates vehicle Start working on the nodes and nodes Intersection Before providing service or empty runs, located at the node The number of remaining serviceable lanes at the time of the incident; Indicates vehicle Start working on the nodes and nodes Intersection Before providing service or empty runs, located at the node The number of remaining serviceable lanes at the time of the incident; For fixed nodes From node Departure to Node ; 6) Total vehicle capacity constraint: The total number of lanes served by each vehicle during the entire operation trip shall not exceed the maximum number of lanes that the vehicle itself can serve. ; indicates as: (28) 7) Legality constraint of empty driving route: All road segments traversed by the vehicle on the empty driving route between any two services and on the empty driving route back to the starting point of the last service must be road segments that have been cleared of snow.

10. A snow removal vehicle route planning method considering demand decomposability and road segment criticality as described in claim 9, characterized in that: In step 4, a tabu search algorithm based on forest representation is used to find the de-snow vehicle path planning model, outputting the optimal starting point position and the path for each vehicle; the specific process is as follows: Step 41: Set the number of parking lots to the directed graph of the highway area to be cleared of snow. The number of nodes; Step 42: Process each node to obtain the final path scheme for each node; Step 43: Based on the final path scheme of each node, obtain the optimal starting point location and the path of each vehicle; In step 42, each node is processed to obtain the final path scheme for each node; the specific process is as follows: Step 421: Given a starting point, first generate a feasible initial plan; The specific process is as follows: Step 4211: Establish a waiting pool: Put all lanes that require snowplow operation into a "waiting pool"; Step 4212: Create an empty vehicle: The empty vehicle departs from the given starting point, and the remaining number of lanes that the empty vehicle can serve is set to... ; Step 4213: Greedy Insertion: Randomly select a lane from the waiting service pool; calculate the cost increment of inserting the randomly selected lane into all possible positions in the current vehicle's path; Cost increment Including the increased operation time due to inserting into new lanes And the increased empty driving time due to the change in empty driving path caused by inserting into a new lane. ; indicates as: (29) Step 4214: Capacity Check: After inserting into a lane, check if the current vehicle's cumulative service lanes are less than or equal to... ; If the current cumulative service lanes for vehicles are less than or equal to Then proceed with steps 4213-4214; If the current vehicle's cumulative service lanes are greater than If so, the current vehicle will be impounded; Repeat steps 4212-4214 until all lanes in the service pool have been allocated. Obtain the initial path scheme ; Step 422: Apply the initial vehicle path scheme obtained in step 421 Calculate evaluation indicators; The specific process is as follows: Step 4221: Calculate the initial principal objective ; indicates as: (30) Step 4222: Calculate the initial secondary objective ; indicates as: (31) Step 4223: Change the current initial path scheme Set as the globally historical best solution Save the primary objective value corresponding to the globally best historical solution. Secondary target value And the complete path sequence, and reset the continuous no-improvement counter to zero; Step 423: Process the initial path scheme obtained in step 421 to obtain the final path scheme; The specific process is as follows: Step 4231: Set the number of iterations ; Step 4232: Randomly select one of the Relocate, Exchange, Swap(1,1) and One-Split operators to process the initial path scheme obtained in Step 421, and obtain a new path scheme. ; The specific process is as follows: Relocate operator: Cuts a lane from the path of car A and inserts it into a certain position in the path of car B; Exchange operator: Swaps the positions of one lane of car A and one lane of car B; The Swap(1,1) operator randomly selects an edge, for example, when going from node 1 to node 2. It scans all vehicle paths and finds the set of vehicles currently serving the edge from node 1 to node 2, such as car A and car B. It reads the service count of car A and car B respectively. The Swap(1,1) operator moves one lane at a time. The candidate solution is to reduce one lane for car A and add one lane for car B. In the path solution of car A, one lane is deleted from the edge from node 1 to node 2, and the order of the other edges in the path solution of car A remains unchanged. In the path solution of car B, this lane is added to the original edge position from node 1 to node 2, and the order of the other edges remains unchanged. After the modification is executed, calculate the current cumulative number of lanes for car A and car B. If it is found that the cumulative number of lanes for car B exceeds the capacity limit of 30, the route plan is restored to the candidate plan and the candidate plan is marked as illegal. One-Split operator: 1) Select the target road segment and delete all vehicles A and B currently serving on the target road segment from the service tasks; at this time, the target road segment is in a vehicle-free service state, and the number of remaining service lanes for vehicles A and B increases, thus clarifying the current upper limit of the number of remaining service lanes for each vehicle. 2) Calculate the incremental cost required to re-insert one lane of the target road segment into each vehicle path; Vehicles are sorted in order of increasing cost, and lanes are assigned to vehicles with the smallest cost increment first. 3) Repeat step 2) until all lane assignments for vehicles without service status are completed; Step 4233: Feasibility review; The specific process is as follows: The new path scheme obtained in step 4232 If review criteria 1, 2, and 3 are not met simultaneously, the application will be discarded and not considered. The new path scheme obtained in step 4232 If all three conditions (Review 1, Review 2, and Review 3) are met, the new path option will be retained. Proceed to step 4234; The specific process is as follows: Review 1: Check if the total number of times each edge is served is exactly equal to the number of its lanes; Review 2: Check whether the total number of lanes served by each vehicle is less than or equal to 30; Review 3: Assume car A needs to start from node... Arrive at the node empty Go to the next path and extract the nodes. With nodes Find the shortest path between two points, and check whether all segments on the shortest path have been served. Step 4234: Calculate the objective function value; the specific process is as follows: If step 4233 retains the new path scheme The feasibility review was passed, and the new primary target value was recalculated using the exact same calculation formula as in step 422. and new sub-target values ; Step 4235: The new path scheme retained from step 4233 Filter the data to obtain valid paths that pass the taboo check; The specific process is as follows: Tabu list definition: If an operation in a new path is recorded and has not yet been de-banned, then the new path is prohibited from being adopted; Extract the new path scheme retained in step 4233 The corresponding unique feature code is used to check the taboo table; If the signature is not in the taboo list, or the remaining taboo period is less than or equal to 0: the path scheme corresponding to the signature is allowed to proceed normally to step 4236; If the signature is in the tabu list and the remaining tabu period is greater than 0, then compare... Compared to the current global historical best ; like Execute the amnesty; the amnesty is to force the release of the path scheme corresponding to the feature code and proceed to step 4236, and remove the corresponding feature code from the taboo table; like Execute tabu; executing tabu means forcibly rejecting the path scheme corresponding to the current feature code and entering step 4236, directly discarding the path scheme corresponding to the current feature code, jumping back to step 4232, randomly selecting the next operator again, and generating another brand new candidate solution. Step 4236: Grant the valid path scheme approved in step 4235 Compared with the current path scheme Begin comparing the advantages and disadvantages; The specific process is as follows: Current primary target value and secondary target value It is based on the current path scheme Calculated; Current path scheme The origin of the data is divided into two stages: the initial generation stage and the iterative update stage; Initial generation phase, This is equivalent to the initial path scheme obtained in step 421. ; During the iterative update phase, step 4236 will be updated. The specific update process is as follows: If the new primary target value Unconditionally accept the legal path scheme approved in step 4235, and set the current path scheme. And reset the continuously unimproved counter to zero; The legal path scheme approved in step 4235; If the new primary target value The valid path scheme approved in step 4235 is rejected, and the current path scheme is maintained. constant; If the new primary target value Compare the new secondary target values and ;like Set the path scheme before The counter that has not improved continuously will be reset to zero; if Maintain the current path scheme constant; Step 4237: If the new route plan was accepted in step 4236 Extract to obtain a new path scheme The reverse operation signature of the neighborhood operator; insert the reverse signature into the tabu list, update the tabu list, and set the remaining tabu period, the length of the remaining tabu period is equal to the length of the initial tabu period, the length of the initial tabu period is set manually; traverse all existing old signatures in the tabu list, and decrement the remaining tabu period of all old signatures by 1; If step 4236 does not accept the new path scheme No reverse operation signature is extracted, and the tabu list remains unchanged; all existing records with remaining tabu periods are decremented by 1. Step 4238: Update the globally optimal path scheme; the specific process is as follows: Compare the current path with the new path accepted in step 4236. Current primary target value Current secondary target value With the optimal principal objective value Optimal secondary objective value ; If the current global historical best primary objective value is Then the globally historical optimal path scheme will be... And update the new main target simultaneously. New secondary target value ; If the current global historical best The original plan will remain unchanged. If the current global historical best Compare secondary target values and ;like The globally optimal historical path scheme ;like The original plan will remain unchanged. Step 4239: If the optimal path scheme in step 4238 has been updated, reset the no-improvement counter to zero; If the optimal path scheme is not updated in step 4238, increment the continuous no-improvement counter by one, and determine whether the counter has reached the threshold of fifty. If the condition is met, a perturbation operation is performed, the process of which is as follows: First, disable the taboo list query function; Then, a lane is randomly cut from one vehicle's path and randomly inserted into the path of another vehicle, and this process is repeated cumulatively. Second-rate; Generate a new current path scheme and reset the counter to zero; If the target is not met, no disturbance operation will be performed, and the process will proceed to step 42310. Step 42310: Determine whether the current iteration count has reached the set maximum iteration count limit; If not reached, let the number of iterations... Repeat steps 4232 to 4239; if the target is reached, exit the loop and obtain the final path solution.

Images