City distribution network adaptive path planning method and system based on space-time correlation prediction

Abstract

This application provides an adaptive path planning method and system for urban distribution networks based on spatiotemporal correlation prediction, relating to the field of urban logistics and distribution technology. First, it obtains a set of constraint parameters for delivery tasks and a set of basic traffic parameters for the urban road network structure. Then, it constructs an initial solution space set for delivery paths based on these parameters. Next, it analyzes the travel time dependency relationship of road segment allocation sequences, generating travel time correlation parameters and travel time impact propagation parameters. Finally, it performs structural redundancy identification and dynamic response integration processing based on these parameters to generate a path decision set responding to real-time traffic events. This invention can perceive traffic conditions in real time and dynamically optimize delivery paths, improving delivery efficiency, accuracy, and reliability, reducing delivery costs, and enhancing customer satisfaction.

Description

Technical Field

[0001] This application relates to the field of urban logistics and distribution technology, and more specifically, to an adaptive path planning method and system for urban distribution networks based on spatiotemporal correlation prediction. Background Technology

[0002] Traditional urban delivery route optimization methods are primarily based on static road network and delivery task information. They typically assume that road network conditions are fixed and only consider basic information such as pickup and delivery locations, and expected delivery times, using classic algorithms (such as shortest path algorithms and genetic algorithms) to plan delivery routes. However, urban road networks are complex dynamic systems, influenced by factors such as traffic flow, traffic accidents, and road construction, resulting in constantly changing road conditions. Static route optimization methods cannot respond promptly to these real-time changes, leading to planned routes that may not be optimal in actual delivery, and even causing delivery tasks to fail to be completed on time due to sudden changes in road conditions. Furthermore, existing methods often lack in-depth analysis of the complex relationships between delivery task constraints and road network traffic parameters, making it difficult to generate route decisions that can flexibly adapt to different road conditions and delivery needs. Summary of the Invention

[0003] In view of this, the purpose of this application is to provide an adaptive path planning method and system for urban distribution networks based on spatiotemporal correlation prediction.

[0004] According to a first aspect of this application, an adaptive path planning method for urban distribution networks based on spatiotemporal correlation prediction is provided, the method comprising: The system obtains a set of delivery task constraint parameters for urban delivery vehicles and a set of basic traffic parameters for the urban road network structure. The set of delivery task constraint parameters includes the geographical coordinates of the pickup location for each delivery order, the geographical coordinates of the delivery location for each delivery order, and the endpoint values ​​of the expected delivery time interval for each delivery order. The set of basic traffic parameters for the urban road network structure includes the physical length of each road segment, the vehicle type identifier allowed to pass through each road segment, the historical average travel time for each road segment at different time periods, and the historical travel time fluctuation range parameter for each road segment at different time periods. Based on the set of delivery task constraint parameters and the set of basic traffic parameters of the urban road network structure, an initial solution space construction operation for delivery paths is performed to obtain a set of initial solution spaces for delivery paths consisting of multiple delivery sequence sequences and road segment allocation sequences corresponding to each delivery sequence sequence. Extract the road segment allocation sequence corresponding to each delivery sequence from the initial solution space set of the delivery route, perform travel time dependency analysis on the extracted road segment allocation sequence, and generate travel time correlation parameters between adjacent road segments within each road segment allocation sequence and travel time influence transmission parameters between non-adjacent road segments within each road segment allocation sequence. Based on the travel time association parameters, the travel time impact transmission parameters, and the initial solution space set of the delivery route, structural redundancy identification and dynamic response integration processing are performed to generate a path decision set for responding to real-time traffic events. The path decision set includes a target equivalent path representative sequence and the execution time scheduling parameters corresponding to the target equivalent path representative sequence.

[0005] According to a second aspect of this application, an adaptive path planning system for urban distribution networks based on spatiotemporal correlation prediction is provided. The system includes a machine-readable storage medium and a processor. The machine-readable storage medium stores machine-executable instructions. When the processor executes the machine-executable instructions, the system implements the aforementioned adaptive path planning method for urban distribution networks based on spatiotemporal correlation prediction.

[0006] Based on any of the above aspects, the technical effect of this application is as follows: By acquiring the set of delivery task constraint parameters for urban delivery vehicles and the set of basic traffic parameters for the urban road network structure, an initial solution space set for delivery routes is constructed based on these parameter sets. This allows for the generation of multiple possible delivery sequence sequences and road segment allocation sequences. Traffic time dependency analysis is performed on the road segment allocation sequences to generate traffic time correlation parameters and traffic time impact transmission parameters, deeply exploring the intrinsic connections between road segments. Based on these parameters, structural redundancy identification and dynamic response integration processing are performed to generate a set of path decisions for responding to real-time traffic events. This enables real-time perception of traffic changes and dynamic adjustment of delivery routes, ensuring that delivery vehicles always choose the optimal route in complex and ever-changing urban road networks. This effectively solves the problem that traditional static path optimization methods cannot adapt to real-time traffic changes, improving the efficiency, accuracy, and reliability of urban delivery, reducing delivery costs, and enhancing customer satisfaction. Attached Figure Description

[0007] Figure 1 A flowchart illustrating the adaptive path planning method for urban distribution networks based on spatiotemporal correlation prediction provided in this application embodiment is shown. Figure 2 This paper illustrates a schematic diagram of the component structure of an adaptive path planning system for urban distribution networks based on spatiotemporal correlation prediction, as provided in an embodiment of this application. Detailed Implementation

[0008] Figure 1 This paper illustrates a flowchart of an adaptive path planning method and system for urban distribution networks based on spatiotemporal correlation prediction, as provided in an embodiment of this application. The detailed steps include: Step S110: Obtain the set of delivery task constraint parameters for urban delivery vehicles and the set of basic traffic parameters for the urban road network structure. The set of delivery task constraint parameters includes the geographical coordinates of the pickup location corresponding to each delivery order, the geographical coordinates of the delivery location corresponding to each delivery order, and the endpoint values ​​of the expected delivery time interval corresponding to each delivery order. The set of basic traffic parameters for the urban road network structure includes the physical length values ​​of each road segment, the vehicle type identifiers allowed to pass through each road segment, the historical average travel time values ​​of each road segment at different time periods, and the historical travel time fluctuation range parameters of each road segment at different time periods.

[0009] In this embodiment, in order to support the efficient route planning of cold chain delivery fleets for fresh food e-commerce within urban areas, it is first necessary to construct basic data describing the delivery tasks and the urban traffic environment.

[0010] The system communicates with the order management server via an interface to obtain information on all currently pending delivery orders, thereby generating a set of delivery task constraint parameters. This set of parameters is indexed by the order identifier, with one record corresponding to each order. For a delivery task with the first order identifier, the geographical coordinates of its pickup location are extracted from the order data. These coordinates are presented as a tuple consisting of a first longitude value and a first latitude value. Simultaneously, the geographical coordinates of the delivery location are extracted, presented as a tuple consisting of a second longitude value and a second latitude value. Furthermore, the expected delivery time interval for the order is extracted. This interval consists of two time points: the left endpoint and the right endpoint.

[0011] Simultaneously, by communicating with the city's traffic information center or a third-party map service provider through an interface, road network data covering the entire urban delivery area is obtained, thereby generating a set of basic traffic parameters for the urban road network structure. This set of basic traffic parameters for the urban road network structure is indexed by road segment identifiers, with each road segment corresponding to one record. For a traffic unit with the first road segment identifier, its basic traffic parameters include the following: physical length value, in meters; vehicle type identifier, which is an enumerated value used to identify the types of vehicles allowed to pass through this road segment, such as "small passenger car", "light truck", "large truck", etc.; historical average travel time value corresponding to different time periods, in seconds, with the index parameter representing different time periods, such as morning peak hours, off-peak hours, and evening peak hours; and historical travel time fluctuation range parameter corresponding to different time periods, which describes the dispersion of historical travel time, specifically represented as an interval value, that is, the historical average travel time value plus or minus a standard deviation value, which is calculated based on historical travel time sample data.

[0012] Step S120: Perform the initial solution space construction operation of delivery path according to the set of delivery task constraint parameters and the set of basic traffic parameters of the urban road network structure, and obtain the initial solution space set of delivery path composed of multiple delivery sequence sequences and the road segment allocation sequence corresponding to each delivery sequence sequence.

[0013] After obtaining the set of constraint parameters for the delivery task and the set of basic traffic parameters for the urban road network structure, the initial solution space construction operation for the delivery path is performed.

[0014] Step S121: Extract the geographical coordinates of the pickup location and the geographical coordinates of the delivery location corresponding to all delivery orders from the set of delivery task constraint parameters, and map the extracted geographical coordinates of the pickup location and the geographical coordinates of the delivery location to the set of network nodes corresponding to the urban road network structure, respectively, to generate the network node identifier of the pickup location and the network node identifier of the delivery location corresponding to each delivery order.

[0015] First, iterate through each order identifier in the set of delivery task constraint parameters. For each order identifier, extract its corresponding pickup location geographic coordinates, i.e., the first longitude and the first latitude, and the delivery location geographic coordinates, i.e., the second longitude and the second latitude.

[0016] Next, the aforementioned geographic coordinates are mapped to the network node set of the urban road network structure. The network node set of the urban road network structure consists of a series of nodes with precise latitude and longitude coordinates; these nodes typically represent road intersections or road endpoints. The mapping operation is achieved by calculating the Euclidean distance between the geographic coordinates and each node in the network node set, and selecting the node with the smallest distance as the mapping result. Specifically, for the first longitude and first latitude values ​​of the pickup location's geographic coordinates, the Euclidean distance between them and each node in the network node set is calculated, and the unique identifier of the node with the smallest Euclidean distance is used as the pickup location network node identifier for that order. Similarly, for the second longitude and second latitude values ​​of the delivery location's geographic coordinates, the Euclidean distance between them and each node in the network node set is calculated, and the unique identifier of the node with the smallest Euclidean distance is used as the delivery location network node identifier for that order.

[0017] Step S122: Based on the pickup location network node identifier and delivery location network node identifier corresponding to each delivery order, call the physical connection relationship of road segments contained in the basic traffic parameter set of the urban road network structure to generate a set of all non-repeating acyclic path sequences from the pickup location network node identifier to the delivery location network node identifier for each delivery order.

[0018] After obtaining the pickup location network node identifier and delivery location network node identifier for each order, for each order, the pickup location network node identifier is used as the starting node for path search, and the delivery location network node identifier is used as the ending node for path search.

[0019] The system retrieves the physical connectivity relationships of road segments from the basic traffic parameter set of the city's road network structure. This physical connectivity is stored in a graph structure, where network nodes are vertices and road segments are edges. Each edge connects two network nodes and records the corresponding road segment identifier. Based on this graph structure, a depth-first search algorithm is executed, starting from the starting node and traversing along the connecting edges, recording the sequence of network nodes visited during the traversal. When a visited network node is encountered during the search, the search on the current branch is stopped to avoid forming a cycle. When the destination node is successfully reached, the edge sequence traversed in this search (i.e., the road segment identifier sequence) is recorded as an acyclic path sequence from the pickup location network node identifier to the delivery location network node identifier. By exhaustively exploring all possible depth-first search branches, all acyclic path sequences from the pickup location network node identifier to the delivery location network node identifier that do not repeatedly pass through the same network node are finally obtained. These sequences are then collected to form the acyclic path sequence set corresponding to the order.

[0020] Step S123: For each acyclic path sequence in the set of acyclic path sequences corresponding to each delivery order, the historical average travel time value corresponding to the basic travel parameter set of the urban road network structure is called according to the road segment identifier contained in the acyclic path sequence to perform cumulative calculation, so as to obtain the basic travel time cumulative value corresponding to each acyclic path sequence.

[0021] After generating a set of acyclic path sequences for each order, each acyclic path sequence in the set is processed for each order. First, the acyclic path sequence is parsed to obtain a list of road segment identifiers arranged in the order of travel. For each road segment identifier in the list, the historical average travel time value corresponding to that road segment identifier is retrieved from the set of basic traffic parameters of the urban road network structure, based on the current time period (e.g., the time period of the expected departure time). Then, the historical average travel time values ​​corresponding to all road segment identifiers in the acyclic path sequence are summed, and the sum is the basic accumulated travel time value for that acyclic path sequence. This basic accumulated travel time value is used to evaluate the traffic efficiency of different paths in subsequent steps.

[0022] Step S124: Combine and arrange the set of acyclic path sequences corresponding to all delivery orders according to the network node identifiers of the pickup location and the network node identifiers of the delivery location of the delivery orders to generate an initial delivery order enumeration space composed of the combination of acyclic path sequences of all delivery orders. The initial delivery order enumeration space contains a variety of delivery order sequences that arrange the pickup and delivery operations of each delivery order in different orders.

[0023] After obtaining the set of acyclic path sequences for each order, the execution order among multiple orders needs to be further considered. The pickup and delivery operations of all current orders to be delivered are considered as a series of task points that need to be executed sequentially. Each order corresponds to two task points: a pickup point (identified by its pickup location network node identifier) ​​and a delivery point (identified by its delivery location network node identifier). All possible task execution order sequences are generated by permuting these task points. A constraint is imposed when generating the sequence: for any order, its pickup operation must be executed before its delivery operation. Under this constraint, all possible task sequences are enumerated using a permutation and combination algorithm; each task sequence is a delivery order sequence. For each generated delivery order sequence, each task point (pickup or delivery) in the delivery order sequence is further replaced with a set of selectable acyclic path sequences for the order corresponding to that task point. By combining the acyclic path sequence options corresponding to each task point, an initial delivery order enumeration space is generated, consisting of combinations of acyclic path sequences for all delivery orders. This initial delivery order enumeration space contains all possible and valid combinations of delivery sequences and corresponding path choices.

[0024] Step S125: For each delivery sequence in the initial delivery sequence enumeration space, based on the geographical coordinate information corresponding to two adjacent operation positions in the delivery sequence, call the physical connection relationship of road segments in the basic traffic parameter set of the urban road network structure to generate the road segment allocation sequence corresponding to the shortest physical distance path connecting the geographical coordinates of two adjacent operation positions.

[0025] Each element in the initial delivery order enumeration space is a delivery order sequence that specifies the order in which vehicles visit task points (pickup or delivery points). For each pair of adjacent task points in this delivery order sequence (e.g., from the first task point to the second), the geographic coordinates corresponding to these two task points are obtained. If the task point is a pickup operation, its geographic coordinates are the geographic coordinates of the pickup location of the order; if the task is a delivery operation, its geographic coordinates are the geographic coordinates of the delivery location of the order. Then, these two geographic coordinates are mapped to the network node set of the city road network structure to obtain the corresponding starting network node identifier and ending network node identifier. Next, the physical connection relationships of road segments in the basic traffic parameter set of the city road network structure are invoked, with the starting network node identifier as the starting point and the ending network node identifier as the ending point, and Dijkstra's algorithm is executed to calculate a shortest physical distance path connecting these two network nodes. This shortest physical distance path consists of a series of consecutive road segment identifiers, which are arranged in the driving order, thus forming the road segment subsequence corresponding to the shortest physical distance path connecting these two adjacent operation locations. The road segment subsequences corresponding to all adjacent operation positions in the delivery sequence are concatenated sequentially to form a road segment allocation sequence. This road segment allocation sequence describes the complete driving path of the vehicle starting from the first task point and completing all task points in sequence.

[0026] Step S126: Associate and store each delivery sequence in the initial delivery sequence enumeration space with the road segment allocation sequence corresponding to that delivery sequence, forming a set of initial solution spaces for delivery paths that includes delivery sequence sequences and road segment allocation sequences.

[0027] After completing the above steps, each delivery sequence in the initial delivery sequence enumeration space, along with the road segment allocation sequence generated for it in step S125, is stored in association. Specifically, a data record is created for each delivery sequence, containing two fields: one field stores the delivery sequence itself, and the other field stores its corresponding complete road segment allocation sequence. All of the above data records together constitute the initial solution space set for delivery paths.

[0028] Step S130: Extract the road segment allocation sequence corresponding to each delivery sequence from the initial solution space set of the delivery path, perform travel time dependency analysis on the extracted road segment allocation sequence, and generate travel time correlation parameters between adjacent road segments within each road segment allocation sequence and travel time influence transmission parameters between non-adjacent road segments within each road segment allocation sequence.

[0029] After obtaining the initial solution space set of delivery routes containing a large number of potential routes, it is necessary to conduct an in-depth analysis of the travel time dependencies between road segments within the aforementioned routes in order to understand the propagation effect of road condition changes on the routes.

[0030] Step S131: Extract the target road segment allocation sequence corresponding to the target delivery sequence from the initial solution space set of the delivery route. The target road segment allocation sequence consists of the first road segment identifier, the second road segment identifier, and so on up to the Nth road segment identifier arranged in the order of vehicle travel.

[0031] Select a delivery sequence to be analyzed from the initial solution space set of the delivery route, and denote it as the target delivery sequence. Extract the corresponding road segment allocation sequence from the data associated with the target delivery sequence, and denote it as the target road segment allocation sequence. The target road segment allocation sequence is an ordered list, in which the elements are road segment identifiers arranged in the order in which vehicles travel, denoted as the first road segment identifier, the second road segment identifier, and so on, up to the last road segment identifier in the sequence, i.e., the Nth road segment identifier.

[0032] Step S132: Perform travel time correlation analysis on any adjacent first road segment identifier and second road segment identifier in the target road segment allocation sequence, collect the vehicle exit time parameter corresponding to the first road segment identifier and the vehicle entry time parameter corresponding to the second road segment identifier, calculate the time difference between the vehicle exit time parameter corresponding to the first road segment identifier and the vehicle entry time parameter corresponding to the second road segment identifier, and use the time difference as the travel time correlation parameter between the first road segment identifier and the second road segment identifier. The travel time correlation parameter is used to characterize the waiting time and connection loss experienced by the vehicle from completing the passage of the first road segment to starting the passage of the second road segment.

[0033] For each pair of adjacent road segment identifiers in the target road segment allocation sequence, such as the first road segment identifier and the immediately following second road segment identifier, a travel time correlation analysis is performed. From the vehicle's historical travel trajectory data, the time point when the vehicle exits the end of the segment represented by the first road segment identifier is collected, denoted as the first exit time point parameter; simultaneously, the time point when the vehicle enters the beginning of the segment represented by the second road segment identifier is collected, denoted as the second entry time point parameter. The difference between the first exit time point parameter and the second entry time point parameter is calculated. This difference is the travel time correlation parameter between the first and second road segment identifiers. This travel time correlation parameter not only includes the physical travel time between the two segments but also implicitly includes potential losses due to waiting (such as red lights at intersections, yielding, etc.) and speed changes at the segment junctions.

[0034] Step S133: Perform a transitivity analysis of travel time impact on the non-adjacent road segment identifiers i and j in the target road segment allocation sequence, where i is less than j and the difference between j and i is greater than 1. Collect the first deviation between the actual travel time value corresponding to the i road segment identifier and the historical average travel time value corresponding to the i road segment identifier. Collect the second deviation between the actual travel time value corresponding to the (i+1)th road segment identifier and the historical average travel time value corresponding to the (i+1)th road segment identifier. Calculate the ratio of the first deviation to the second deviation. Multiply this ratio by the cumulative impact attenuation coefficient from the (i+1)th road segment identifier to the (j-1)th road segment identifier to obtain the travel time impact transmission parameter of the i road segment identifier on the j road segment identifier.

[0035] For non-adjacent road segment identifier pairs in the target road segment allocation sequence, such as the i-th road segment identifier with index i (i is a positive integer) and the j-th road segment identifier with index j (j is greater than i+1), a transitivity analysis of travel time impact is performed. First, the actual travel time value of the i-th road segment identifier is obtained from historical or real-time data, and the difference between it and the historical average travel time value of that segment is calculated, denoted as the first deviation. Then, the actual travel time value of the (i+1)-th road segment identifier is obtained, and the difference between it and the historical average travel time value of that segment is calculated, denoted as the second deviation. The ratio of the first deviation to the second deviation is calculated. Next, the cumulative impact attenuation coefficient from the (i+1)-th road segment identifier to the (j-1)-th road segment identifier is determined. This cumulative impact attenuation coefficient can be calculated using a preset attenuation function, for example, the impact decreases by a fixed proportion after each road segment. The cumulative impact attenuation coefficient is the product of the attenuation coefficients of each segment from the (i+1)-th segment to the (j-1)-th segment. Finally, the ratio of the first deviation to the second deviation is multiplied by the cumulative impact attenuation coefficient, and the resulting product is the transmission parameter of the travel time impact of the i-th road segment identifier on the j-th road segment identifier.

[0036] Step S134: Summarize the travel time association parameters between all adjacent road segment identifiers in the target road segment allocation sequence to form an adjacent road segment travel time association parameter set; summarize the travel time influence transmission parameters between all non-adjacent road segment identifiers in the target road segment allocation sequence to form a non-adjacent road segment travel time influence transmission parameter set.

[0037] After analyzing all adjacent and non-adjacent segment pairs in the target road segment allocation sequence, all travel time correlation parameters calculated in step S132 are organized and summarized according to their corresponding road segment identifier pairs to form a set, denoted as the adjacent road segment travel time correlation parameter set. Simultaneously, all travel time impact transmission parameters calculated in step S133 are organized and summarized according to their corresponding road segment identifier pairs to form another set, denoted as the non-adjacent road segment travel time impact transmission parameter set. These two sets together describe the complex travel time dependencies within the target road segment allocation sequence.

[0038] Step S135: After completing the analysis of the road segment allocation sequence corresponding to all delivery sequence sequences in the initial solution space set of the delivery path, the set of adjacent road segment travel time correlation parameters and the set of non-adjacent road segment travel time influence transmission parameters corresponding to each road segment allocation sequence are obtained.

[0039] Repeat steps S131 to S134, traversing each delivery sequence and its associated road segment allocation sequence in the initial solution space set of the delivery path. For each road segment allocation sequence, generate its corresponding set of parameters related to the travel time of adjacent road segments and a set of parameters related to the travel time impact of non-adjacent road segments. Finally, obtain the mapping relationship between all road segment allocation sequences and their corresponding two parameter sets.

[0040] Step S140: Based on the travel time association parameters, the travel time impact transmission parameters, and the initial solution space set of the delivery route, perform structural redundancy identification and dynamic response integration processing to generate a path decision set for responding to real-time traffic events. The path decision set includes a target equivalent path representative sequence and the execution time scheduling parameters corresponding to the target equivalent path representative sequence.

[0041] After obtaining all potential paths and their internal travel time dependency parameters, structural redundancy identification and dynamic response integration processing are performed to extract path groups with similar travel time characteristics from a large number of paths and prepare for responding to real-time traffic events.

[0042] Step S141: Perform structural redundancy identification processing on the initial solution space set of the delivery path according to the travel time association parameter and the travel time influence transmission parameter to obtain a set of path structural redundancy groups containing multiple paths with the same delivery sequence but with local differences in road segment allocation sequence.

[0043] The goal of structural redundancy identification is to find paths that are identical in the order of core tasks but have different choices at certain intermediate segments, and these different choices have a similar impact on the overall travel time.

[0044] Step S1411: Select multiple candidate road segment allocation sequences with the same delivery order sequence from the initial solution space set of the delivery path. Perform a bit-by-bit comparison of the selected multiple candidate road segment allocation sequences according to the road segment identifier sequences they contain, and identify the maximum consecutive prefix length value and the maximum consecutive suffix length value that are the same from the start road segment identifier in the multiple candidate road segment allocation sequences.

[0045] First, traverse the initial solution space of delivery paths, grouping all road segment allocation sequences with the same delivery order sequence into one group. For multiple candidate road segment allocation sequences within this group, compare them one by one. For example, take any two candidate road segment allocation sequences, starting from the first road segment identifier of the sequence, compare them bit by bit until the first different position is encountered, and record the length of the common prefix, which is the maximum prefix length value. Similarly, starting from the last road segment identifier of the sequence, compare them bit by bit from back to front until the first different position is encountered, and record the length of the common suffix, which is the maximum suffix length value.

[0046] Step S1412: Determine the start and end positions of the variable intermediate segments in multiple candidate road segment allocation sequences based on the maximum prefix length value and the maximum suffix length value, and extract the road segment identifier subsequence located between the start and end positions of the variable intermediate segments in each candidate road segment allocation sequence as the variable intermediate segment sequence of that candidate road segment allocation sequence.

[0047] Based on the determined maximum prefix length and maximum suffix length values, the common prefix and common suffix portions of these candidate road segment allocation sequences can be identified. The portion of the candidate road segment allocation sequence that follows the common prefix and precedes the common suffix constitutes the variable intermediate segment. Therefore, the starting position index of the variable intermediate segment is the maximum prefix length value plus 1, and the ending position index of the variable intermediate segment is the total sequence length minus the maximum suffix length value. For each candidate road segment allocation sequence within a group, the road segment identifier subsequence from the starting position to the ending position is extracted, which is the variable intermediate segment sequence of that sequence.

[0048] Step S1413: Obtain the set of adjacent road segment travel time association parameters and the set of non-adjacent road segment travel time influence transmission parameters corresponding to each of the multiple candidate road segment allocation sequences. Extract the first subset of travel time association parameters between adjacent road segment identifiers within each variable intermediate segment sequence from the set of adjacent road segment travel time association parameters. Extract the first subset of travel time influence transmission parameters between non-adjacent road segment identifiers within each variable intermediate segment sequence from the set of non-adjacent road segment travel time influence transmission parameters.

[0049] For each candidate road segment allocation sequence, from the set of travel time-related parameters for its corresponding adjacent road segments, parameters whose corresponding road segment identifier pairs are entirely located within the variable intermediate segment sequence of that sequence are selected, forming the first subset of travel time-related parameters. Similarly, from the set of travel time impact transmission parameters for its corresponding non-adjacent road segments, parameters whose corresponding road segment identifier pairs are entirely located within the variable intermediate segment sequence are selected, forming the first subset of travel time impact transmission parameters. These two subsets together describe the travel time dependency characteristics within the variable intermediate segment.

[0050] Step S1414: Calculate the overlap of parameter value ranges for the first travel time associated parameter subset and the first travel time influence transmission parameter subset corresponding to multiple candidate road segment allocation sequences. When the absolute value of the difference between all corresponding position parameter values ​​in the first travel time associated parameter subset corresponding to two candidate road segment allocation sequences is less than the preset tolerance threshold for the difference of associated parameters and the absolute value of the difference between all corresponding position parameter values ​​in the first travel time influence transmission parameter subset is less than the preset tolerance threshold for the difference of transmission parameters, mark the two candidate road segment allocation sequences as equivalent path structures with similar travel time dependency relationships.

[0051] For any two candidate road segment allocation sequences within a group, obtain their first subset of travel time-related parameters and first subset of travel time impact propagation parameters. Compare the parameters in these two subsets item by item. For the first subset of travel time-related parameters, compare the travel time-related parameter values ​​at corresponding positions in the two sequences (e.g., the first pair of adjacent segments within a variable intermediate segment), and calculate the absolute value of the difference between the two parameter values. If the absolute value of the difference at all corresponding positions is less than a preset threshold, denoted as the correlation parameter difference tolerance threshold, then they are considered similar in terms of correlation parameters. Similarly, for the first subset of travel time impact propagation parameters, perform item-by-item comparisons. If the absolute value of the difference at all corresponding positions is less than another preset threshold, denoted as the propagation parameter difference tolerance threshold, then they are also considered similar in terms of propagation parameters. When both conditions are met simultaneously, mark the two candidate road segment allocation sequences as equivalent path structures with similar travel time dependencies.

[0052] Step S1415: All candidate road segment allocation sequences marked as having similarity in travel time dependency are grouped into the same path structure redundancy group, and the corresponding delivery sequence, variable intermediate segment start position, variable intermediate segment end position, and identifiers of all candidate road segment allocation sequences in the path structure redundancy group are recorded for the path structure redundancy group.

[0053] All candidate road segment allocation sequences marked as similar to each other are grouped into the same group, denoted as a path structure redundancy group. The following key information is recorded for this group: their common delivery sequence; their common variable intermediate segment start and end positions (these positions are determined based on prefix-suffix alignment); and the unique identifiers of all candidate road segment allocation sequences within this group.

[0054] Step S1416: Repeat the above structural redundancy identification processing operation until the grouping processing of all candidate road segment allocation sequences with the same delivery order sequence in the initial solution space set of the delivery path is completed, and a path structural redundancy group set containing multiple path structural redundancy groups is obtained.

[0055] For each candidate road segment with the same delivery order sequence in the initial solution space set of the delivery route, assign a sequence group and repeat steps S1411 to S1415 until all the above groups have been analyzed and divided. Finally, a set of path structure redundancy groups is obtained, consisting of multiple path structure redundancy groups.

[0056] Step S142: Perform an equivalent path merging operation on each path structure redundancy group in the path structure redundancy group set to generate the equivalent path representative sequence and the travel time fluctuation tolerance interval parameter corresponding to each path structure redundancy group.

[0057] After identifying the structural redundancy group, it is necessary to merge multiple similar paths within the group into a representative path and calculate the tolerance range of the path in terms of travel time.

[0058] Step S1421: Select a target path structure redundancy group from the path structure redundancy group set, and extract the delivery sequence, variable intermediate segment start position, variable intermediate segment end position, and identifiers of all candidate road segment allocation sequences within the target path structure redundancy group.

[0059] First, select a group to be processed from the set of path structure redundancy groups, denoted as the target path structure redundancy group. Extract the key information recorded in this group: delivery sequence, variable intermediate segment start position, variable intermediate segment end position, and a list of identifiers for all candidate road segment allocation sequences within the group.

[0060] Step S1422: Based on the identifiers of all candidate road segment allocation sequences within the target path structure redundancy group, call the adjacent road segment travel time association parameter set and the non-adjacent road segment travel time influence transmission parameter set corresponding to each candidate road segment allocation sequence. Extract the second travel time association parameter subset between adjacent road segment identifiers within the first fixed prefix segment before the start position of the variable intermediate segment from the adjacent road segment travel time association parameter set. Extract the second travel time influence transmission parameter subset between non-adjacent road segment identifiers within the first fixed prefix segment from the non-adjacent road segment travel time influence transmission parameter set.

[0061] For each candidate road segment allocation sequence within the target path structure redundancy group, the corresponding set of adjacent road segment travel time association parameters and the set of non-adjacent road segment travel time impact transmission parameters are retrieved based on its identifier. Based on the variable intermediate segment start position, the portion of the sequence preceding this start position can be determined and denoted as the first fixed prefix segment. From the retrieved set of adjacent road segment travel time association parameters, parameters whose corresponding road segment identifier pairs are completely within the first fixed prefix segment are selected to form the second subset of travel time association parameters. From the retrieved set of non-adjacent road segment travel time impact transmission parameters, parameters whose corresponding road segment identifier pairs are completely within the first fixed prefix segment are selected to form the second subset of travel time impact transmission parameters.

[0062] Step S1423: Extract the third travel time association parameter subset between adjacent road segment identifiers within the second fixed suffix segment located after the end position of the variable intermediate segment from the adjacent road segment travel time association parameter set; extract the third travel time influence transmission parameter subset between non-adjacent road segment identifiers within the second fixed suffix segment from the non-adjacent road segment travel time influence transmission parameter set.

[0063] Similarly, based on the end position of the variable intermediate segment, the portion of the sequence following that end position is determined and denoted as the second fixed suffix segment. From the retrieved set of adjacent road segment travel time association parameters, parameters whose corresponding road segment identifier pairs are completely located within the second fixed suffix segment are selected to form the third subset of travel time association parameters. From the retrieved set of non-adjacent road segment travel time influence transmission parameters, parameters whose corresponding road segment identifier pairs are completely located within the second fixed suffix segment are selected to form the third subset of travel time influence transmission parameters.

[0064] Step S1424: Concatenate the delivery sequence corresponding to the target path structure redundancy group, the road segment identifier sequence corresponding to the fixed prefix segment before the start position of the variable intermediate segment, and the road segment identifier sequence corresponding to the fixed suffix segment after the end position of the variable intermediate segment to generate the equivalent path representative sequence corresponding to the target path structure redundancy group.

[0065] In all candidate road segment allocation sequences within the group, their first fixed prefix segment and second fixed suffix segment are identical (this is determined by the property of the greatest common prefix and suffix). Therefore, the road segment identifier sequence corresponding to the first fixed prefix segment and the road segment identifier sequence corresponding to the second fixed suffix segment of any candidate sequence are taken and concatenated with the delivery order sequence corresponding to the target path structure redundancy group. The concatenation method is as follows: the first fixed prefix segment sequence, a placeholder (representing that the variable intermediate segment has been abstracted), and the second fixed suffix segment sequence are combined in sequence to form the equivalent path representative sequence corresponding to the path structure redundancy group. This equivalent path representative sequence retains the core framework of the path while abstracting the internal variable local road segments.

[0066] Step S1425: For each road segment location in the equivalent path representative sequence corresponding to the target path structure redundancy group, based on the road segment identifiers that may appear at that location according to the allocation sequence of all candidate road segments in the target path structure redundancy group, obtain the historical average travel time value and historical travel time fluctuation range parameter of the corresponding road segment identifier from the basic travel parameter set of the urban road network structure. Calculate the travel time fluctuation tolerance interval for that road segment location based on the obtained historical average travel time value and historical travel time fluctuation range parameter. Combine the travel time fluctuation tolerance intervals of all road segment locations according to the order of the road segments in the equivalent path representative sequence to form the travel time fluctuation tolerance interval parameter corresponding to the equivalent path representative sequence.

[0067] For each location in the equivalent path representation sequence (e.g., a location in the first fixed prefix segment, an abstract variable intermediate segment location, or a location in the second fixed suffix segment), it is necessary to calculate the tolerance range for travel time fluctuations. For locations in the first and second fixed prefix segments, since their corresponding road segment identifiers are fixed, the historical average travel time value and historical travel time fluctuation range parameter of the road segment identifier are directly obtained from the basic travel parameter set of the urban road network structure. The fluctuation range parameter is directly used as the tolerance range for travel time fluctuations at that location. For abstract variable intermediate segment locations, it is necessary to comprehensively consider the travel time characteristics of the actual road segment identifiers corresponding to that location in all candidate road segment allocation sequences within the group. Collect all road segment identifiers appearing at the corresponding locations within the variable intermediate segment (e.g., the first location of the variable intermediate segment) of all candidate sequences within the group, and obtain their historical average travel time values ​​and historical travel time fluctuation range parameters from the basic travel parameter set. Then, the minimum and maximum values ​​of all historical average travel times are taken to form the average travel time interval for that location; the minimum lower limit and maximum upper limit of all historical travel time fluctuation range parameters are taken to form the fluctuation range interval for that location. These two intervals are combined (e.g., by taking the union or weighted average of the average and fluctuation intervals) to obtain a comprehensive travel time fluctuation tolerance interval for that abstract location. The travel time fluctuation tolerance intervals calculated for all locations in the equivalent path representation sequence are organized according to the sequence order to form a parameter list, which represents the travel time fluctuation tolerance interval parameters for that equivalent path representation sequence.

[0068] Step S1426: Associate and store the equivalent path representative sequence with the travel time fluctuation tolerance interval parameter, and repeat the above equivalent path merging operation until the processing of all path structure redundancy groups in the path structure redundancy group set is completed, so as to obtain the equivalent path representative sequence and travel time fluctuation tolerance interval parameter corresponding to each path structure redundancy group.

[0069] The equivalent path representative sequence generated in step S1424 is associated and stored with the travel time fluctuation tolerance interval parameter generated in step S1425. Then, steps S1421 to S1425 are repeated to process the next path structure redundancy group in the path structure redundancy group set until all path structure redundancy groups have been processed. Finally, the equivalent path representative sequence and travel time fluctuation tolerance interval parameter corresponding to each path structure redundancy group are obtained.

[0070] Step S143: Obtain a set of real-time traffic event trigger signals, which includes parameters such as the instantaneous traffic speed reduction ratio for each road segment, the lane closure duration for each road segment, and the impact range of temporary traffic control for each road segment.

[0071] Step S1431: The roadside sensing terminal cluster deployed in the urban road network structure collects vehicle trajectory point data passing through each road section according to a preset sampling time interval. The vehicle trajectory point data includes the vehicle's unique identifier, the identifier of the road section currently traversed by the vehicle, the vehicle's current timestamp, and the vehicle's current instantaneous speed value.

[0072] Roadside sensing terminals in urban road networks, such as millimeter-wave radar and high-definition cameras, continuously collect trajectory data of vehicles passing through their monitoring range at fixed sampling intervals, such as once every three seconds. Each trajectory data point is recorded as a data tuple, which contains the following fields: a globally unique identifier for the vehicle, used to track the location of the same vehicle at different times; an identifier for the road segment where the vehicle is currently located, obtained by matching the vehicle's geographic coordinates with a road network map; the vehicle's current timestamp, accurate to the millisecond, recording the time of collection of the trajectory point; and the vehicle's current instantaneous speed value, obtained through radar speed measurement or image difference algorithms, in kilometers per hour.

[0073] Step S1432: The collected vehicle trajectory point data is grouped according to road segment identifiers. Within each group corresponding to the road segment identifier, the trajectory point sequence is recombined according to the vehicle's unique identifier to generate the traffic speed change curve for each vehicle's unique identifier on each road segment identifier.

[0074] The collected massive amounts of vehicle trajectory data are first grouped by road segment identifier, aggregating all trajectory data belonging to the same road segment. Then, within each road segment identifier group, secondary grouping is performed using the vehicle's unique identifier as the key, sorting all trajectory data of the same vehicle on that road segment in ascending order of timestamp, forming a trajectory sequence. This sequence is then smoothed, for example using moving average filtering, to eliminate instantaneous noise, thereby generating a speed change curve for the vehicle on that road segment. This speed change curve describes the instantaneous speed change trend of the vehicle over time from entering to leaving the segment.

[0075] Step S1433: Perform statistical analysis on the speed change curves corresponding to the unique identification codes of all vehicles on each road segment identifier, calculate the ratio between the average instantaneous speed of all vehicles within the current sampling time window for each road segment identifier and the historical average speed value converted from the historical average travel time value corresponding to that road segment identifier, and use the difference between the two as the instantaneous speed reduction ratio parameter corresponding to each road segment identifier.

[0076] For each road segment identifier, the speed change curve corresponding to the unique identifier of all vehicles passing through that segment within the current sampling time window (e.g., the past five minutes) is obtained. The arithmetic mean of all instantaneous speed values ​​on the curve is taken to obtain the current instantaneous average speed for that segment. Simultaneously, the historical average travel time for that road segment identifier in the current time period is obtained from the basic traffic parameter set of the urban road network structure, and converted into a historical average speed value using the physical length value. The ratio of the current instantaneous average speed to the historical average speed value is calculated. Then, this ratio is subtracted from the value 1, and the difference is the instantaneous speed reduction ratio parameter corresponding to that road segment identifier. The instantaneous speed reduction ratio parameter ranges from 0 to 1; a larger value indicates a more severe speed reduction.

[0077] Step S1434: Obtain the real-time traffic event broadcast data stream released by the city traffic control center, and parse out the set of starting road segment identifiers, the start timestamp of the event, the expected end timestamp of the event, and the event type code from the real-time traffic event broadcast data stream. Calculate the lane closure duration parameter corresponding to each event based on the time difference between the expected end timestamp and the start timestamp.

[0078] Through a data interface with the city's traffic control center, we continuously receive real-time traffic event broadcast data streams. For each broadcast event, we parse the following key fields from the data stream: the set of initial road segment identifiers affected by the event (which may contain one or more road segment identifiers); the start timestamp of the event (the time when the event began to be recorded); the estimated end timestamp of the event (the time when the traffic control center estimates the event's impact will be eliminated); and the event type code, such as "traffic accident," "road construction," or "large-scale event." For each parsed event, we calculate its lane closure duration parameter, which is obtained by subtracting the start timestamp of the event from its estimated end timestamp.

[0079] Step S1435: Based on the event type code, call the preset event impact range mapping rule library, extract the extended set of affected road segment identifiers that match the event type code from the event impact range mapping rule library, and use the number of road segment identifiers contained in the extended set of affected road segment identifiers as the temporary traffic control impact range parameter for each event.

[0080] The impact of an event often extends beyond the initial road segment, potentially affecting upstream and downstream road segments. Therefore, a pre-defined event impact range mapping rule base is established, storing the correspondence between different event type codes and extended rules for affected road segment identifiers. For example, the "traffic accident" type might correspond to a rule affecting the initial road segment identifier and one upstream and two downstream segments. Based on the parsed event type code, the corresponding extended rule is matched from the rule base, and an extended set of affected road segment identifiers is generated based on this rule and the event's initial road segment identifier set. The number of road segment identifiers included in this extended set is used as the temporary traffic control impact range parameter for the event. This temporary traffic control impact range parameter is an integer representing the number of affected road segments.

[0081] Step S1436: Associate and aggregate the instantaneous traffic speed reduction ratio parameter, lane closure duration parameter, and temporary traffic control impact range parameter corresponding to each road segment identifier according to the road segment identifier to form a set of real-time traffic event trigger signals indexed by the road segment identifier.

[0082] For each road segment identifier affected by a traffic event, the instantaneous speed reduction ratio parameter obtained in step S1433, the lane closure duration parameter obtained in steps S1434 and S1435, and the temporary traffic control impact range parameter are aggregated. If a road segment identifier is affected by multiple events simultaneously (e.g., congestion and an accident occur simultaneously), the maximum value of all relevant parameters is taken as the final parameter for that segment. Finally, a data mapping structure is generated with the road segment identifier as the key, and each key corresponds to a data tuple composed of the above three parameters. This mapping structure is the set of real-time traffic event trigger signals.

[0083] Step S144: Perform a dynamic switching decision operation on the delivery route based on the set of real-time traffic event trigger signals, the equivalent path representative sequence corresponding to each path structure redundancy group, and the tolerable interval parameter of the travel time fluctuation corresponding to the equivalent path representative sequence, to obtain the target equivalent path representative sequence to which the urban delivery vehicle should switch at the current time and the execution time scheduling parameter corresponding to the target equivalent path representative sequence.

[0084] After obtaining the set of real-time traffic event trigger signals and the pre-built set of route decisions, the final dynamic switching decision is executed to select the optimal route for the delivery vehicle and plan the detailed execution time.

[0085] Step S1441: Extract the instantaneous speed reduction ratio parameter, lane closure duration parameter, and temporary traffic control impact range parameter corresponding to the target road segment identifier from the real-time traffic event trigger signal set. Compare the instantaneous speed reduction ratio parameter corresponding to the target road segment identifier with a preset speed reduction ratio trigger threshold. When the instantaneous speed reduction ratio parameter exceeds the speed reduction ratio trigger threshold, mark the target road segment identifier as a road segment affected by speed reduction.

[0086] Iterate through each road segment identifier in the set of real-time traffic event trigger signals and obtain its corresponding three parameters. For each road segment identifier, first compare its instantaneous speed reduction percentage parameter with a preset trigger threshold, such as 0.3 (representing a speed reduction of more than 30%). If the parameter is greater than the threshold, then mark the road segment identifier as a road segment affected by the speed reduction.

[0087] Step S1442: Compare the lane closure duration parameter corresponding to the target road segment identifier with the preset duration trigger threshold. When the lane closure duration parameter exceeds the duration trigger threshold, mark the target road segment identifier as a road segment affected by lane closure.

[0088] Continue comparing the lane closure duration parameter of the same road segment identifier with another preset trigger threshold, such as 15 minutes. If the duration is greater than the trigger threshold, then the road segment identifier is marked as a road segment affected by lane closure.

[0089] Step S1443: Compare the temporary traffic control impact range parameter corresponding to the target road segment identifier with the preset impact range trigger threshold. When the temporary traffic control impact range parameter exceeds the impact range trigger threshold, mark the target road segment identifier as a road segment affected by the control range.

[0090] Finally, the temporary traffic control impact range parameter of the same road segment identifier is compared with another preset trigger threshold, such as three road segments. If the range parameter is greater than the threshold, the road segment identifier is marked as a road segment affected by the control range. A road segment may be marked as multiple types simultaneously.

[0091] Step S1444: Traverse all equivalent path representative sequences corresponding to the path structure redundancy groups, identify the road segment identifier positions that intersect with the road segments affected by speed reduction, road segments affected by lane closure, and road segments affected by control range from the road segment identifier sequences contained in each equivalent path representative sequence, and mark the identified road segment identifier positions as affected location points.

[0092] For each equivalent path representation sequence corresponding to a path structure redundancy group, parse its contained sequence of road segment identifiers (for abstract variable intermediate segment locations, it may be necessary to treat them as multiple possible specific segments). Check if the road segment identifier of each location (or each specific segment) in this sequence exists in the set of road segments affected by speed reduction, lane closure, or traffic control as marked in the above steps. If it exists, mark the index of that location in the equivalent path representation sequence as an affected location point.

[0093] Step S1445: For each equivalent path representative sequence, extract the travel time fluctuation tolerance interval parameters corresponding to all affected location points in the equivalent path representative sequence. Overlay the travel time fluctuation tolerance interval parameters corresponding to the affected location points with the instantaneous speed reduction ratio parameter, lane closure duration parameter, and temporary traffic control impact range parameter of the corresponding road segment identifier in the real-time traffic event trigger signal set, and calculate the expected travel time offset of the equivalent path representative sequence after being affected by the real-time traffic event.

[0094] For each equivalent path representative sequence, the travel time offset is calculated based on the marked affected location points.

[0095] Step S14451: Select the first affected location point from the target equivalent path representative sequence, obtain the first road segment identifier corresponding to the first affected location point, and extract the first lower limit value and the first upper limit value of the first road segment identifier from the travel time fluctuation tolerance interval parameter.

[0096] First, select the first affected point from the affected locations in the equivalent path representative sequence, and denote it as the first affected location point. Obtain the actual road segment identifier corresponding to this location point (if this location is an abstract variable intermediate segment, it may be necessary to select the most likely affected candidate segment based on real-time traffic conditions). Then, extract the lower limit and upper limit of travel time fluctuation corresponding to this location point from the travel time fluctuation tolerance interval parameters corresponding to the equivalent path representative sequence.

[0097] Step S14452: Extract the first instantaneous speed reduction ratio parameter, the first lane closure duration parameter, and the first temporary traffic control influence range parameter corresponding to the first road segment identifier from the real-time traffic event trigger signal set. Calculate the speed reduction factor corresponding to the first road segment identifier based on the first instantaneous speed reduction ratio parameter, calculate the capacity reduction factor corresponding to the first road segment identifier based on the first lane closure duration parameter, and calculate the detour factor corresponding to the first road segment identifier based on the first temporary traffic control influence range parameter.

[0098] From the set of real-time traffic event trigger signals, obtain the first instantaneous speed reduction ratio parameter, the first lane closure duration parameter, and the first temporary traffic control impact range parameter corresponding to the first road segment identifier. Based on the first instantaneous speed reduction ratio parameter, calculate a speed reduction factor; for example, this factor equals 1 divided by (1 minus the first instantaneous speed reduction ratio parameter). Based on the first lane closure duration parameter, calculate a capacity reduction factor; for example, this factor equals a base reduction value plus the product of the duration parameter and the unit-time reduction coefficient. Based on the first temporary traffic control impact range parameter, calculate a detour factor; for example, this factor equals 1 plus the product of the impact range parameter and the detour distance coefficient.

[0099] Step S14453: Multiply the first lower limit of travel time fluctuation with the travel speed reduction factor, the travel capacity reduction factor, and the travel path detour factor to obtain the first expected lower limit extended value of travel time for the first affected location point; multiply the first upper limit of travel time fluctuation with the travel speed reduction factor, the travel capacity reduction factor, and the travel path detour factor to obtain the first expected upper limit extended value of travel time for the first affected location point.

[0100] The first lower limit of travel time fluctuation obtained in step S14451 is multiplied by the travel speed reduction factor, capacity reduction factor, and travel route detour factor calculated in step S14452. The product is the expanded value of the first expected lower limit of travel time. Similarly, the first upper limit of travel time fluctuation is multiplied by these three factors to obtain the expanded value of the first expected upper limit of travel time. These three factors work together to amplify the original travel time fluctuation range to reflect the impact of deteriorating real-time traffic conditions.

[0101] Step S14454: Select the second affected location point from the target equivalent path representative sequence, obtain the second road segment identifier corresponding to the second affected location point, and obtain the second expected travel time lower limit extension value and the second expected travel time upper limit extension value of the second affected location point.

[0102] Repeat steps S14451 to S14453 to process the next affected position point on the equivalent path representative sequence, such as the second affected position point, to obtain its corresponding second expected travel time lower limit expansion value and second expected travel time upper limit expansion value.

[0103] Step S14455: Summate the lower bound extension values ​​of the expected travel time corresponding to all affected location points in the target equivalent path representative sequence to obtain the total lower bound extension value of the expected travel time, and sum the upper bound extension values ​​of the expected travel time corresponding to all affected location points in the target equivalent path representative sequence to obtain the total upper bound extension value of the expected travel time.

[0104] Traverse all affected locations in the equivalent path representation sequence, and sum the calculated lower bound expansion values ​​for each point to obtain the total lower bound expansion value for expected travel time. Similarly, sum all the upper bound expansion values ​​for expected travel time to obtain the total upper bound expansion value for expected travel time.

[0105] Step S14456: Calculate the first difference between the lower bound expansion value of the total expected travel time and the original cumulative travel time of the target equivalent path representative sequence at the affected location point in the unaffected state; calculate the second difference between the upper bound expansion value of the total expected travel time and the original cumulative travel time of the target equivalent path representative sequence at the affected location point in the unaffected state; and take the interval formed by the first difference and the second difference as the expected travel time offset of the equivalent path representative sequence after being affected by real-time traffic events.

[0106] In the unaffected state, the accumulated original travel time for all affected locations in the equivalent path representation sequence is calculated (i.e., the arithmetic mean of the original lower and upper limits of the travel time fluctuation tolerance range parameter, or directly the original historical average travel time). Then, the difference between the expanded lower limit of the total expected travel time and the accumulated original travel time is calculated as the first difference. The difference between the expanded upper limit of the total expected travel time and the accumulated original travel time is calculated as the second difference. The interval formed by the first and second differences represents the expected travel time offset of the equivalent path representation sequence after being affected by real-time traffic events.

[0107] Step S1446: Summate the expected travel time offset corresponding to each equivalent path representative sequence with the cumulative value of the basic travel time of the equivalent path representative sequence in the unaffected state to obtain the corrected travel time estimate of each equivalent path representative sequence after being affected by real-time traffic events.

[0108] For each equivalent path representative sequence, obtain its basic travel time accumulation value under unaffected conditions (e.g., by adding the historical average travel time of all fixed locations in the sequence to the average travel time of abstract locations). Summate this basic travel time accumulation value with the expected travel time offset (an interval value) calculated in step S1445 to obtain a new interval value, which is the corrected travel time estimate of the equivalent path representative sequence after being affected by real-time traffic events.

[0109] Step S1447: Select the equivalent path representative sequence with the minimum value from all the corrected travel time estimates corresponding to the equivalent path representative sequences as the target equivalent path representative sequence to which the urban delivery vehicle should switch at the current time. Generate the execution time scheduling parameters corresponding to the target equivalent path representative sequence based on the corrected travel time estimate corresponding to the target equivalent path representative sequence and the road segment identifier of the current location of the urban delivery vehicle. The execution time scheduling parameters include the expected departure time and expected arrival time of the urban delivery vehicle as it departs from its current location and passes through each road segment identifier in the target equivalent path representative sequence.

[0110] Among all the corrected travel time estimates (interval values) of the equivalent path representative sequences, the optimal one is selected through a preset decision strategy (e.g., taking the minimum midpoint of the interval, or taking the minimum upper limit of the interval, etc.), and its corresponding equivalent path representative sequence is taken as the target equivalent path representative sequence to which the urban delivery vehicle should switch at the current time.

[0111] For example, step S14471: Obtain the road segment identifier of the current location of the urban delivery vehicle as the starting road segment identifier, and obtain the first road segment identifier contained in the target equivalent path representative sequence as the starting road segment identifier of the target path.

[0112] The precise location of the delivery vehicle is obtained and mapped onto the urban road network using a map matching algorithm to obtain the corresponding road segment identifier, which is denoted as the starting road segment identifier. Simultaneously, the first road segment identifier is extracted from the selected target equivalent path representative sequence and denoted as the target path starting road segment identifier.

[0113] Step S14472: Compare the starting road segment identifier with the starting road segment identifier of the target path. When the starting road segment identifier is different from the starting road segment identifier of the target path, call the shortest connecting path sequence from the basic traffic parameter set of the urban road network structure that connects the location of the starting road segment identifier with the location of the starting road segment identifier of the target path. Concatenate the shortest connecting path sequence with the target equivalent path representative sequence to generate a complete execution path sequence.

[0114] Compare the starting road segment identifier with the starting road segment identifier of the target path. If they are the same, the sequence representing the target equivalent path is itself the execution path sequence. If they are different, a connecting path needs to be inserted between the current vehicle position and the starting point of the planned path. Using the basic traffic parameter set of the city road network structure, and employing the Dijkstra algorithm, a connecting path with the shortest physical distance or shortest travel time is calculated, starting from the end of the segment represented by the starting road segment identifier and ending at the beginning of the segment represented by the starting road segment identifier of the target path. This connecting path also consists of a series of road segment identifiers. This connecting path sequence is then concatenated with the sequence representing the target equivalent path to form a complete execution path sequence from the vehicle's current position to its final destination.

[0115] Step S14473: Extract the first estimated departure time point corresponding to the first road segment identifier from the complete execution path sequence, take the current time of the urban delivery vehicle as the first estimated departure time point, extract the k-th estimated departure time point and the k-th estimated arrival time point corresponding to the k-th road segment identifier from the complete execution path sequence, and determine the k-th estimated departure time point based on the connection waiting time parameter in the execution time scheduling parameter corresponding to the (k-1)-th road segment identifier and the (k-1)-th road segment identifier.

[0116] The current time is taken as the estimated departure time of the vehicle entering the first road segment identifier in the complete execution path sequence, and is denoted as the first estimated departure time. For the subsequent k-th road segment identifier, its estimated departure time is determined based on the time when the vehicle leaves the previous road segment identifier. Specifically, the time when the vehicle leaves the (k-1)-th road segment identifier is the (k-1)-th estimated arrival time. Between the end of the (k-1)-th segment and the beginning of the k-th segment, there may be a connection waiting time parameter, which can be estimated using the travel time correlation parameter calculated in step S132. Therefore, the estimated departure time of the k-th segment is equal to the (k-1)-th estimated arrival time plus the connection waiting time parameter.

[0117] Step S14474: Extract the segmented travel time estimates corresponding to each road segment identifier from the corrected travel time estimate corresponding to the target equivalent path representative sequence, associate the segmented travel time estimates with the expected arrival time of urban delivery vehicles at the starting port of each road segment identifier, and generate the expected entry time and expected exit time corresponding to each road segment identifier.

[0118] From the revised travel time estimate calculated in step S1446, the segmented travel time estimate for each affected location can be further decomposed. For unaffected fixed locations, their historical average travel time is used as the segmented travel time estimate. For each road segment identifier, its expected entry time (i.e., the expected departure time of the segment) is added to the segmented travel time estimate to obtain the expected exit time (i.e., the time of arrival at the end of the segment).

[0119] Step S14475: Combine the expected entry time and expected exit time corresponding to all road segment identifiers in the complete execution path sequence according to the arrangement order of the road segment identifiers in the complete execution path sequence to generate the execution time scheduling parameters corresponding to the target equivalent path representative sequence. The execution time scheduling parameters include the time node sequence of urban delivery vehicles passing through each road segment identifier in sequence and the duration interval of travel on each road segment identifier.

[0120] The estimated entry and exit times of each road segment identifier in the complete execution path sequence are organized into a time node sequence according to their order in the sequence. Simultaneously, for each road segment, the travel duration interval is the time period from the estimated entry time to the estimated exit time. This time node sequence and the travel duration interval list together constitute the execution time scheduling parameters corresponding to the target equivalent path representative sequence.

[0121] For example, the method may further include: step S210: constructing a cross-modal memory network of delivery tasks and road network status, wherein the cross-modal memory network includes a delivery task memory unit and a road network status memory unit, wherein the delivery task memory unit is used to store the pattern features of historical delivery task constraint parameter sets and corresponding execution path sequences, and the road network status memory unit is used to store the set of historical real-time road condition event trigger signals and their propagation evolution pattern features in the spatial and temporal dimensions.

[0122] To further enhance the intelligence and foresight of decision-making, a cross-modal memory network is constructed. This network consists of two core memory units: a delivery task memory unit and a road network state memory unit. The delivery task memory unit is built based on a key-value storage structure, where the key is a high-dimensional embedding vector of the set of historical delivery task constraint parameters, and the values ​​are the set of redundant path structures executed for the corresponding historical task, switching decision records, and the final path execution result feedback. The road network state memory unit is also based on a key-value storage structure, where the key is the feature vector of the set of historical real-time traffic event trigger signals in the spatiotemporal dimension, and the values ​​are the spatial propagation pattern (e.g., the speed and direction of congestion spread) and temporal evolution pattern (e.g., the rate of congestion dissipation).

[0123] Step S220: Before the current delivery task is executed, the set of delivery task constraint parameters and the set of basic traffic parameters of the current urban road network structure are input into the delivery task memory unit. The delivery task memory unit matches the historical delivery task patterns similar to the current delivery task through a key-value retrieval mechanism and outputs the path switching decision records of the historical similar delivery task patterns under multiple path structure redundancy groups.

[0124] Before starting route planning for a new delivery task, the current task's delivery constraint parameter set (order pickup and delivery points, expected time windows, etc.) and the current city road network's basic traffic parameter set are encoded to generate a query vector. This query vector is then input into the delivery task memory unit. The delivery task memory unit retrieves the most similar historical delivery task patterns by calculating the similarity between the query vector and the stored historical task key vectors (e.g., using cosine similarity). Then, it outputs the route switching decisions made by these historically similar delivery task patterns during execution, in response to various road condition events encountered, under the corresponding path structure redundancy group.

[0125] Step S230: After the path structure redundancy group set is generated, extract the spatial coverage contour corresponding to each path structure redundancy group from the initial solution space set of the delivery path. The spatial coverage contour is formed by the union boundary of the set of road segment geometry formed by all candidate road segment allocation sequences in the path structure redundancy group on the electronic map.

[0126] After generating the path structure redundancy group set in step S141, for each path structure redundancy group, its spatial coverage contour is extracted. This spatial coverage contour is generated as follows: the road segment identifiers contained in all candidate road segment allocation sequences within the group are mapped onto an electronic map as corresponding geometric shapes (i.e., lines representing these road segments on the map). Then, the union of all these geometric shape sets is calculated, and the boundary of this union is extracted. This boundary is the spatial coverage contour of the path structure redundancy group. This spatial coverage contour describes the entire area that the path structure redundancy group may occupy in space.

[0127] Step S240: Extract the spatial distribution field of the instantaneous traffic speed reduction ratio parameter of each road segment from the real-time traffic event trigger signal set, and perform feature extraction on the spatial distribution field to obtain the spatial hotspot area coordinates and spatial propagation direction vector of the real-time traffic conditions.

[0128] The instantaneous traffic speed reduction ratio of all road segments in the real-time traffic event trigger signal set is projected onto a city map to form a scalar field, namely the spatial distribution field. Gradient calculation and cluster analysis are performed on this spatial distribution field to identify local maxima where the parameter values ​​are significantly higher than those of the surrounding areas; the locations of these points are the coordinates of spatial hotspot areas. Simultaneously, the boundary change trend of the hotspot areas is analyzed, and the centroid movement direction of the hotspot area within the most recent time windows is calculated to obtain the spatial propagation direction vector. This spatial propagation direction vector describes the spatial diffusion trend of traffic events (such as congestion).

[0129] Step S250: Input the spatial coverage contour of each path structure redundancy group and the coordinates of the spatial hotspot area of ​​real-time traffic conditions into the road network state memory unit. The road network state memory unit predicts the expected traffic condition impact intensity parameters of the spatial hotspot area on the spatial coverage contour of each path structure redundancy group in the future preset time period under the influence of the current spatial propagation direction vector through a spatiotemporal matching retrieval mechanism.

[0130] The spatial coverage contour of each path structure redundancy group and the coordinates of the spatial hotspot area of ​​the current real-time traffic conditions are used as query conditions and input into the road network state memory unit. The road network state memory unit retrieves its stored historical event propagation evolution patterns to find spatiotemporal patterns similar to the current hotspot area coordinates, propagation direction vector, and query contour. Through spatiotemporal matching and pattern extrapolation, the degree of impact that the spatial hotspot area will have on the spatial coverage contour of each path structure redundancy group in the future (e.g., the next 30 minutes) under the influence of the current spatial propagation direction vector is predicted. This degree of impact is represented by the expected traffic condition impact intensity parameter, which can be a value between 0 and 1, with a larger value indicating a more severe impact.

[0131] Step S260: Based on the expected road condition impact intensity parameter, the travel time fluctuation tolerance interval parameter of the equivalent path representative sequence corresponding to each path structure redundancy group is prospectively corrected to generate prospectively corrected travel time tolerance interval parameters corresponding to each equivalent path representative sequence.

[0132] For each path structure redundancy group, based on the expected road condition impact intensity parameter predicted in step S250, the travel time fluctuation tolerance interval parameter of its equivalent path representative sequence is corrected. The correction method can be: for the entire equivalent path representative sequence, or for the portion whose spatial coverage overlaps with hotspot areas, both the upper and lower limits of its travel time fluctuation tolerance interval are amplified by a coefficient positively correlated with the expected road condition impact intensity parameter. For example, if the expected road condition impact intensity parameter is α, then the original lower and upper limits of travel time fluctuation are multiplied by (1+β) respectively. α), where β is a preset amplification factor, is used to obtain the forward-looking correction travel time tolerance interval parameter. This allows the travel time estimation of a route to predict the possibility of future road condition deterioration in advance.

[0133] Step S270: When making dynamic switching decisions for delivery routes, the forward-looking corrected travel time tolerance interval parameter is replaced with the original travel time fluctuation tolerance interval parameter, and superimposed with the current real-time traffic event trigger signal set for analysis to calculate the corrected travel time estimate of each equivalent path representative sequence after being affected by current and expected future traffic events.

[0134] When making dynamic route switching decisions in step S144, the forward-looking corrected travel time tolerance interval parameter generated in step S260 is used to replace the original travel time fluctuation tolerance interval parameter. Then, following the methods described in steps S1445 and S1446, the above-mentioned forward-looking corrected parameter is overlaid and analyzed with the current real-time traffic event trigger signal set. In this way, the calculated corrected travel time estimate not only reflects the impact of current traffic conditions but also incorporates predictions of future traffic condition evolution.

[0135] Step S280: After the decision is made, the set of delivery task constraint parameters, the set of generated path structure redundancy groups, the set of received real-time traffic event trigger signals, the dynamic switching decision operation of the delivery path, and the actual path execution results are fed back to the cross-modal memory network to update the mode features and evolution laws in the delivery task memory unit and the road network state memory unit.

[0136] After the delivery task is completed and the actual route execution results are obtained (e.g., actual travel time on each road segment, whether delivery was on time, etc.), the entire process data of this task is used as a new learning sample. The set of delivery task constraint parameters and the generated set of path structure redundancy groups, along with their execution results, are encoded and stored in the delivery task memory unit, updating their key-value pairs. Simultaneously, the set of real-time traffic event trigger signals received during this task (including their spatiotemporal evolution) and the final route switching decision results are encoded and stored in the road network state memory unit, updating its stored event propagation evolution pattern. Through this continuous learning and feedback mechanism, the cross-modal memory network can continuously optimize its matching and prediction capabilities, making future route decisions more accurate and efficient.

[0137] Figure 2This application illustrates an adaptive route planning system 100 for urban distribution networks based on spatiotemporal correlation prediction, comprising a processor 1001 and a memory 1003. The processor 1001 and memory 1003 are connected, for example, via a bus 1002. Optionally, the adaptive route planning system 100 may further include a transceiver 1004, which can be used for data interaction between this adaptive route planning system and other adaptive route planning systems based on spatiotemporal correlation prediction, such as sending and / or receiving data. It should be noted that in actual scheduling, the transceiver 1004 is not limited to one, and the structure of this adaptive route planning system 100 does not constitute a limitation on the embodiments of this application.

[0138] The memory 1003 is used to store program code for executing the embodiments of this application, and its execution is controlled by the processor 1001. The processor 1001 is used to execute the program code stored in the memory 1003 to implement the steps shown in the foregoing method embodiments.

[0139] The above description is only an optional implementation method for some implementation scenarios of this application. It should be noted that for those skilled in the art, other similar implementation methods based on the technical concept of this application, without departing from the technical concept of this application, also fall within the protection scope of the embodiments of this application.

Claims

1. An adaptive path planning method for urban distribution networks based on spatiotemporal correlation prediction, characterized in that, The method includes: The system obtains a set of delivery task constraint parameters for urban delivery vehicles and a set of basic traffic parameters for the urban road network structure. The set of delivery task constraint parameters includes the geographical coordinates of the pickup location for each delivery order, the geographical coordinates of the delivery location for each delivery order, and the endpoint values ​​of the expected delivery time interval for each delivery order. The set of basic traffic parameters for the urban road network structure includes the physical length of each road segment, the vehicle type identifier allowed to pass through each road segment, the historical average travel time for each road segment at different time periods, and the historical travel time fluctuation range parameter for each road segment at different time periods. Based on the set of delivery task constraint parameters and the set of basic traffic parameters of the urban road network structure, an initial solution space construction operation for delivery paths is performed to obtain a set of initial solution spaces for delivery paths consisting of multiple delivery sequence sequences and road segment allocation sequences corresponding to each delivery sequence sequence. Extract the road segment allocation sequence corresponding to each delivery sequence from the initial solution space set of the delivery route, perform travel time dependency analysis on the extracted road segment allocation sequence, and generate travel time correlation parameters between adjacent road segments within each road segment allocation sequence and travel time influence transmission parameters between non-adjacent road segments within each road segment allocation sequence. Based on the travel time association parameters, the travel time impact transmission parameters, and the initial solution space set of the delivery route, structural redundancy identification and dynamic response integration processing are performed to generate a path decision set for responding to real-time traffic events. The path decision set includes a target equivalent path representative sequence and the execution time scheduling parameters corresponding to the target equivalent path representative sequence.

2. The adaptive path planning method for urban distribution networks based on spatiotemporal correlation prediction according to claim 1, characterized in that, The step of performing an initial solution space construction operation for delivery routes based on the set of delivery task constraint parameters and the set of basic traffic parameters of the urban road network structure yields an initial solution space set for delivery routes consisting of multiple delivery sequence sequences and road segment allocation sequences corresponding to each delivery sequence sequence, including: Extract the geographical coordinates of the pickup location and the geographical coordinates of the delivery location corresponding to all delivery orders from the set of delivery task constraint parameters, and map the extracted geographical coordinates of the pickup location and the geographical coordinates of the delivery location to the set of network nodes corresponding to the urban road network structure, respectively, to generate the network node identifier of the pickup location and the network node identifier of the delivery location corresponding to each delivery order. Based on the pickup location network node identifier and delivery location network node identifier corresponding to each delivery order, the physical connection relationship of road segments contained in the basic traffic parameter set of the urban road network structure is invoked to generate a set of all non-repeating acyclic path sequences from the pickup location network node identifier to the delivery location network node identifier for each delivery order. For each acyclic path sequence in the set of acyclic path sequences corresponding to each delivery order, the historical average travel time value corresponding to the basic travel parameter set of the urban road network structure is called according to the road segment identifier contained in the acyclic path sequence and accumulated to obtain the basic travel time accumulation value corresponding to each acyclic path sequence; The set of acyclic path sequences corresponding to all delivery orders is combined and arranged according to the network node identifiers of the pickup location and the network node identifiers of the delivery location of the delivery order, to generate an initial delivery order enumeration space composed of the combination of acyclic path sequences of all delivery orders. The initial delivery order enumeration space contains a variety of delivery order sequences that arrange the pickup and delivery operations of each delivery order in different orders. For each delivery sequence in the initial delivery sequence enumeration space, the physical connection relationship of road segments in the basic traffic parameter set of the urban road network structure is called according to the geographical coordinate information of two adjacent operation positions in the delivery sequence, and the road segment allocation sequence corresponding to the shortest physical distance path connecting the geographical coordinates of two adjacent operation positions is generated. Each delivery sequence in the initial delivery sequence enumeration space is associated with and stored with the corresponding road segment allocation sequence, forming a set of initial solution spaces for delivery paths that includes delivery sequence sequences and road segment allocation sequences.

3. The adaptive path planning method for urban distribution networks based on spatiotemporal correlation prediction according to claim 1, characterized in that, The process involves extracting road segment allocation sequences corresponding to each delivery sequence from the initial solution space set of the delivery path, performing travel time dependency analysis on the extracted road segment allocation sequences, and generating travel time correlation parameters between adjacent road segments within each road segment allocation sequence and travel time influence transmission parameters between non-adjacent road segments within each road segment allocation sequence, including: Extract the target road segment allocation sequence corresponding to the target delivery sequence from the initial solution space set of the delivery route. The target road segment allocation sequence consists of the first road segment identifier, the second road segment identifier, and so on up to the Nth road segment identifier arranged in the order of vehicle travel. For any adjacent first road segment identifier and second road segment identifier in the target road segment allocation sequence, a travel time correlation analysis is performed. The vehicle exit time parameter corresponding to the first road segment identifier and the vehicle entry time parameter corresponding to the second road segment identifier are collected. The time difference between the vehicle exit time parameter corresponding to the first road segment identifier and the vehicle entry time parameter corresponding to the second road segment identifier is calculated. This time difference is used as the travel time correlation parameter between the first road segment identifier and the second road segment identifier. The travel time correlation parameter is used to characterize the waiting time and connection loss experienced by the vehicle from completing the passage of the first road segment to starting the passage of the second road segment. A transitivity analysis of the travel time impact is performed on the non-adjacent road segment identifiers i and j in the target road segment allocation sequence, where i is less than j and the difference between j and i is greater than 1. The first deviation between the actual travel time value corresponding to the i-th road segment identifier and the historical average travel time value corresponding to the i-th road segment identifier is collected. The second deviation between the actual travel time value corresponding to the (i+1)-th road segment identifier and the historical average travel time value corresponding to the (i+1)-th road segment identifier is collected. The ratio of the first deviation to the second deviation is calculated. This ratio is multiplied by the cumulative impact attenuation coefficient from the (i+1)-th to the (j-1)-th road segment identifier to obtain the travel time impact transitivity parameter of the i-th road segment identifier on the j-th road segment identifier. The travel time correlation parameters between all adjacent road segment identifiers in the target road segment allocation sequence are summarized to form an adjacent road segment travel time correlation parameter set, and the travel time influence transmission parameters between all non-adjacent road segment identifiers in the target road segment allocation sequence are summarized to form a non-adjacent road segment travel time influence transmission parameter set. After analyzing the road segment allocation sequences corresponding to all delivery sequence sequences in the initial solution space set of the delivery route, the set of parameters related to the travel time of adjacent road segments and the set of parameters for the transmission of the travel time influence of non-adjacent road segments corresponding to each road segment allocation sequence are obtained.

4. The adaptive path planning method for urban distribution networks based on spatiotemporal correlation prediction according to claim 1, characterized in that, The step of performing structural redundancy identification and dynamic response integration processing based on the travel time correlation parameters, the travel time impact transmission parameters, and the initial solution space set of the delivery route to generate a route decision set for responding to real-time traffic events includes: Based on the travel time correlation parameter and the travel time influence transmission parameter, the initial solution space set of the delivery route is subjected to structural redundancy identification processing to obtain a set of path structural redundancy groups containing multiple groups with the same delivery sequence but with local differences in road segment allocation sequence; Perform an equivalent path merging operation on each path structure redundancy group in the path structure redundancy group set to generate an equivalent path representative sequence and a travel time fluctuation tolerance interval parameter corresponding to each path structure redundancy group. Acquire a set of real-time traffic event trigger signals, which includes parameters such as the instantaneous traffic speed reduction ratio for each road segment, the lane closure duration for each road segment, and the impact range of temporary traffic control for each road segment. Based on the set of real-time traffic event trigger signals, the equivalent path representative sequence corresponding to each path structure redundancy group, and the travel time fluctuation tolerance interval parameter corresponding to the equivalent path representative sequence, a dynamic switching decision operation for delivery routes is performed to obtain the target equivalent path representative sequence to which urban delivery vehicles should switch at the current time and the execution time scheduling parameters corresponding to the target equivalent path representative sequence.

5. The adaptive path planning method for urban distribution networks based on spatiotemporal correlation prediction according to claim 4, characterized in that, The process of identifying structural redundancy in the initial solution space set of the delivery route based on the travel time correlation parameter and the travel time influence propagation parameter yields a set of path structural redundancy groups containing multiple routes with the same delivery sequence but local differences in road segment allocation sequences, including: In the initial solution space set of the delivery route, select multiple candidate road segment allocation sequences with the same delivery order sequence. Then, perform a bit-by-bit comparison of the selected multiple candidate road segment allocation sequences according to the road segment identifier sequences they contain, and identify the maximum consecutive prefix length value and the maximum consecutive suffix length value that are the same from the start road segment identifier in the multiple candidate road segment allocation sequences. The start and end positions of the variable intermediate segments in multiple candidate road segment allocation sequences are determined based on the maximum prefix length value and the maximum suffix length value. The road segment identifier subsequence located between the start and end positions of the variable intermediate segments in each candidate road segment allocation sequence is extracted as the variable intermediate segment sequence of that candidate road segment allocation sequence. Obtain the set of adjacent road segment travel time association parameters and the set of non-adjacent road segment travel time influence transmission parameters corresponding to each of the multiple candidate road segment allocation sequences. Extract the first subset of travel time association parameters between adjacent road segment identifiers within each variable intermediate segment sequence from the set of adjacent road segment travel time association parameters. Extract the first subset of travel time influence transmission parameters between non-adjacent road segment identifiers within each variable intermediate segment sequence from the set of non-adjacent road segment travel time influence transmission parameters. The parameter value range overlap of the first travel time associated parameter subset and the first travel time influence transmission parameter subset corresponding to multiple candidate road segment allocation sequences is calculated respectively. When the absolute value of the difference of all corresponding position parameter values ​​in the first travel time associated parameter subset corresponding to two candidate road segment allocation sequences is less than the preset association parameter difference tolerance threshold and the absolute value of the difference of all corresponding position parameter values ​​in the first travel time influence transmission parameter subset is less than the preset transmission parameter difference tolerance threshold, the two candidate road segment allocation sequences are marked as equivalent path structures with similar travel time dependency relationship. All candidate road segment allocation sequences marked as having similar travel time dependencies are grouped into the same path structure redundancy group, and the corresponding delivery sequence, variable intermediate segment start position, variable intermediate segment end position, and identifiers of all candidate road segment allocation sequences within the path structure redundancy group are recorded for the path structure redundancy group. Repeat the above structural redundancy identification and processing operation until the grouping processing of all candidate road segment allocation sequences with the same delivery order sequence in the initial solution space set of the delivery path is completed, resulting in a path structural redundancy group set containing multiple path structural redundancy groups.

6. The adaptive path planning method for urban distribution networks based on spatiotemporal correlation prediction according to claim 4, characterized in that, The step of performing an equivalent path merging operation on each path structure redundancy group in the set of path structure redundancy groups to generate an equivalent path representative sequence and a travel time fluctuation tolerance interval parameter corresponding to each path structure redundancy group includes: Select a target path structure redundancy group from the set of path structure redundancy groups, and extract the delivery sequence, variable intermediate segment start position, variable intermediate segment end position, and identifiers of all candidate road segment allocation sequences within the target path structure redundancy group. Based on the identifiers of all candidate road segment allocation sequences within the target path structure redundancy group, call the adjacent road segment travel time association parameter set and the non-adjacent road segment travel time influence transmission parameter set corresponding to each candidate road segment allocation sequence. Extract the second travel time association parameter subset between adjacent road segment identifiers within the first fixed prefix segment before the variable intermediate segment start position from the adjacent road segment travel time association parameter set. Extract the second travel time influence transmission parameter subset between non-adjacent road segment identifiers within the first fixed prefix segment from the non-adjacent road segment travel time influence transmission parameter set. Extract the third travel time association parameter subset between adjacent road segment identifiers within the second fixed suffix segment located after the end position of the variable intermediate segment from the set of travel time association parameters for adjacent road segments; and extract the third travel time influence transmission parameter subset between non-adjacent road segment identifiers within the second fixed suffix segment from the set of travel time influence transmission parameters for non-adjacent road segments. The delivery sequence corresponding to the target path structure redundancy group, the road segment identifier sequence corresponding to the fixed prefix segment before the start position of the variable intermediate segment, and the road segment identifier sequence corresponding to the fixed suffix segment after the end position of the variable intermediate segment are concatenated to generate the equivalent path representative sequence corresponding to the target path structure redundancy group. For each road segment location in the equivalent path representative sequence corresponding to the target path structure redundancy group, based on the road segment identifiers that may appear at that location according to the allocation sequence of all candidate road segments in the target path structure redundancy group, the historical average travel time value and historical travel time fluctuation range parameter of the corresponding road segment identifier are obtained from the basic travel parameter set of the urban road network structure. Based on the obtained historical average travel time value and historical travel time fluctuation range parameter, the travel time fluctuation tolerance interval of the road segment location is calculated. The travel time fluctuation tolerance intervals of all road segment locations are combined according to the order of the road segments in the equivalent path representative sequence to form the travel time fluctuation tolerance interval parameter corresponding to the equivalent path representative sequence. The equivalent path representative sequence is associated with the travel time fluctuation tolerance interval parameter and stored together. The above equivalent path merging operation is repeated until the processing of all path structure redundancy groups in the path structure redundancy group set is completed, and the equivalent path representative sequence and travel time fluctuation tolerance interval parameter corresponding to each path structure redundancy group are obtained.

7. The adaptive path planning method for urban distribution networks based on spatiotemporal correlation prediction according to claim 4, characterized in that, The acquisition of the real-time traffic event trigger signal set includes: The roadside sensing terminal cluster deployed in the urban road network structure collects vehicle trajectory point data passing through each road section according to a preset sampling time interval. The vehicle trajectory point data includes the vehicle's unique identifier, the identifier of the road section currently traversed by the vehicle, the vehicle's current timestamp, and the vehicle's current instantaneous speed value. The collected vehicle trajectory point data is grouped according to road segment identifiers. Within each group corresponding to the road segment identifier, the trajectory point sequence is recombined according to the vehicle's unique identifier to generate the speed change curve of each vehicle's unique identifier on each road segment identifier. Statistical analysis is performed on the speed change curves corresponding to the unique identification codes of all vehicles on each road segment identifier. The ratio between the average instantaneous speed of all vehicles on each road segment identifier within the current sampling time window and the historical average speed value converted from the historical average travel time value corresponding to that road segment identifier is calculated. The difference between this ratio and the value is used as the instantaneous speed reduction ratio parameter corresponding to each road segment identifier. Obtain the real-time traffic event broadcast data stream released by the city traffic control center, and parse out the set of starting road segment identifiers, the start timestamp of the event, the expected end timestamp of the event, and the event type code from the real-time traffic event broadcast data stream. Calculate the lane closure duration parameter corresponding to each event based on the time difference between the expected end timestamp and the start timestamp. The event type code is used to call the preset event impact range mapping rule library. The event impact range mapping rule library is used to extract the extended set of affected road segment identifiers that match the event type code. The number of road segment identifiers contained in the extended set of affected road segment identifiers is used as the temporary traffic control impact range parameter for each event. The instantaneous traffic speed reduction ratio, lane closure duration, and temporary traffic control impact range parameters corresponding to each road segment identifier are associated and aggregated according to the road segment identifier to form a set of real-time traffic event trigger signals indexed by the road segment identifier.

8. The adaptive path planning method for urban distribution networks based on spatiotemporal correlation prediction according to claim 4, characterized in that, The dynamic switching decision-making operation for delivery routes is performed based on the set of real-time traffic event trigger signals, the equivalent path representative sequences corresponding to each path structure redundancy group, and the tolerable travel time fluctuation interval parameters corresponding to the equivalent path representative sequences. This process yields the target equivalent path representative sequence to which urban delivery vehicles should switch at the current time, along with the corresponding execution time scheduling parameters. This includes: Extract the instantaneous speed reduction ratio parameter, lane closure duration parameter, and temporary traffic control impact range parameter corresponding to the target road segment identifier from the set of real-time traffic event trigger signals. Compare the instantaneous speed reduction ratio parameter corresponding to the target road segment identifier with a preset speed reduction ratio trigger threshold. When the instantaneous speed reduction ratio parameter exceeds the speed reduction ratio trigger threshold, mark the target road segment identifier as a road segment affected by speed reduction. The lane closure duration parameter corresponding to the target road segment identifier is compared with the preset duration trigger threshold. When the lane closure duration parameter exceeds the duration trigger threshold, the target road segment identifier is marked as a road segment affected by lane closure. The temporary traffic control impact range parameter corresponding to the target road segment identifier is compared with the preset impact range trigger threshold. When the temporary traffic control impact range parameter exceeds the impact range trigger threshold, the target road segment identifier is marked as a road segment affected by the control range. Traverse all equivalent path representative sequences corresponding to the path structure redundancy groups, identify the road segment identifier positions that intersect with the road segments affected by speed reduction, road segments affected by lane closure, and road segments affected by control range from the road segment identifier sequence contained in each equivalent path representative sequence, and mark the identified road segment identifier positions as affected location points; For each equivalent path representative sequence, extract the travel time fluctuation tolerance interval parameters corresponding to all affected location points in the equivalent path representative sequence. Then, overlay the travel time fluctuation tolerance interval parameters corresponding to the affected location points with the instantaneous speed reduction ratio parameter, lane closure duration parameter, and temporary traffic control impact range parameter of the corresponding road segment identifier in the real-time traffic event trigger signal set to calculate the expected travel time offset of the equivalent path representative sequence after being affected by the real-time traffic event. The expected travel time offset corresponding to each equivalent path representative sequence is summed with the cumulative value of the basic travel time of the equivalent path representative sequence under the unaffected state to obtain the corrected travel time estimate of each equivalent path representative sequence after being affected by real-time traffic events. Among all the estimated travel times corresponding to the equivalent path representative sequences, the equivalent path representative sequence with the minimum value is selected as the target equivalent path representative sequence to which the urban delivery vehicle should switch at the current time. The execution time scheduling parameters corresponding to the target equivalent path representative sequence are generated based on the estimated travel time corresponding to the target equivalent path representative sequence and the road segment identifier of the current location of the urban delivery vehicle. The execution time scheduling parameters include the expected departure time and expected arrival time of the urban delivery vehicle as it departs from its current location and passes through each road segment identifier in the target equivalent path representative sequence.

9. The adaptive path planning method for urban distribution networks based on spatiotemporal correlation prediction according to claim 8, characterized in that, For each equivalent path representative sequence, the tolerance range parameters for travel time fluctuations corresponding to all affected location points in the equivalent path representative sequence are extracted. These tolerance range parameters are then overlaid with parameters from the real-time traffic event trigger signal set, including the instantaneous speed reduction ratio, lane closure duration, and temporary traffic control impact range of the corresponding road segment identifier. The expected travel time offset of the equivalent path representative sequence after being affected by the real-time traffic event is calculated, including: Select the first affected location point from the target equivalent path representative sequence, obtain the first road segment identifier corresponding to the first affected location point, and extract the first lower limit value and the first upper limit value of the first road segment identifier from the traffic time fluctuation tolerance interval parameter. Extract the first instantaneous speed reduction ratio parameter, the first lane closure duration parameter, and the first temporary traffic control influence range parameter corresponding to the first road segment identifier from the set of real-time traffic event trigger signals. Calculate the speed reduction factor corresponding to the first road segment identifier based on the first instantaneous speed reduction ratio parameter, calculate the capacity reduction factor corresponding to the first road segment identifier based on the first lane closure duration parameter, and calculate the detour factor corresponding to the first road segment identifier based on the first temporary traffic control influence range parameter. The first expected lower limit of travel time fluctuation is obtained by multiplying the lower limit of travel time fluctuation with the travel speed reduction factor, the travel capacity reduction factor, and the travel route detour factor. The first expected lower limit of travel time fluctuation is obtained by multiplying the first upper limit of travel time fluctuation with the travel speed reduction factor, the travel capacity reduction factor, and the travel route detour factor. Select the second affected location point from the target equivalent path representative sequence, obtain the second road segment identifier corresponding to the second affected location point, and obtain the second expected travel time lower limit expansion value and the second expected travel time upper limit expansion value of the second affected location point; The total expected travel time lower bound expansion value is obtained by summing the expected travel time lower bound expansion values ​​corresponding to all affected location points in the target equivalent path representative sequence, and the total expected travel time upper bound expansion value is obtained by summing the expected travel time upper bound expansion values ​​corresponding to all affected location points in the target equivalent path representative sequence. Calculate the first difference between the lower bound of the total expected travel time extension value and the original cumulative travel time value of the target equivalent path representative sequence at the affected location points in the unaffected state. Calculate the second difference between the upper bound of the total expected travel time extension value and the original cumulative travel time value of the target equivalent path representative sequence at the affected location points in the unaffected state. Use the interval formed by the first difference and the second difference as the expected travel time offset of the equivalent path representative sequence after being affected by real-time traffic events.

10. An adaptive path planning system for urban distribution networks based on spatiotemporal correlation prediction, characterized in that, The method includes a processor and a computer-readable storage medium storing machine-executable instructions that, when executed by the processor, implement the urban distribution network adaptive path planning method based on spatiotemporal correlation prediction as described in any one of claims 1-9.

Images