Rail transit emergency plan recommendation method and system based on deep learning

Abstract

The application provides a kind of track traffic emergency plan recommendation method and system based on deep learning, it is related to track traffic operation management technical field, obtain line operation state data set after sudden operation event occurs, extract the section occupancy state characteristics of each operation section and the transmission path blocking characteristics of signal equipment transmission path, according to section occupancy state characteristics, construct line operation blocking situation description data representing the spatial continuous distribution form of blocked section;The blocked situation description data and the transmission path blocking characteristics are input into the pre-constructed emergency dispatching strategy generation network to model the correlation of the blocked section and generate the joint dispatching strategy, obtain track traffic emergency plan dispatching strategy data, thereby obtain the emergency dispatching instruction sequence for track traffic emergency dispatching system, significantly improve the matching accuracy of emergency plan and actual blocked situation, the consistency of multi-type dispatching strategy, and the timeliness and executability of emergency response decision.

Description

Technical Field

[0001] This invention relates to the field of rail transit operation management technology, and more specifically, to a method and system for recommending rail transit emergency plans based on deep learning. Background Technology

[0002] In the context of rail transit operation and management, the occurrence of sudden operational events (such as equipment failure, natural disasters, and large passenger flow surges) can lead to operational disruptions in some sections of the line topology, interruptions in signal equipment transmission paths, and abnormal passenger flow at transfer stations. It is urgent to quickly generate targeted emergency dispatch plans to restore line operation order and ensure passenger safety.

[0003] In conventional technologies for generating emergency plans for rail transit, a rule-based template matching method is typically used. This method pre-defines fixed plan templates for several typical event scenarios, and directly retrieves the corresponding template based on the event type when an event occurs. However, it fails to dynamically model the continuous spatial distribution of line obstruction caused by sudden events, nor does it consider the coupled impact of signal equipment transmission path obstruction on scheduling strategies, resulting in a lack of refined adaptability to the actual obstruction situation. Secondly, a scheduling method based on independent analysis of single sections can also be used, treating each section of the line as an independent unit and calculating scheduling schemes separately. This method fails to establish a modeling mechanism for the correlation between obstructed sections and ignores the linkage effect of spatial continuity constraints between adjacent sections on short-loop turnaround and train interval adjustments. Furthermore, an optimization method based on a single scheduling objective can be used, which only performs scheduling calculations with the single objective of minimizing train delay time or maximizing line capacity, without simultaneously encompassing the joint generation of three strategies: short-loop turnaround scheduling, train interval adjustment scheduling, and passenger flow guidance scheduling at affected transfer stations. Summary of the Invention

[0004] In view of the aforementioned problems, and in conjunction with the first aspect of the present invention, embodiments of the present invention provide a method for recommending emergency response plans for rail transit based on deep learning, the method comprising: The system acquires a set of line operation status data for a rail transit line after a sudden operational event occurs. The set of line operation status data includes real-time train operation location distribution data for each section of the line topology, signal equipment status data for each section, and event trigger location data of the sudden operational event in the line topology. The section occupancy status characteristics of each operating section in the line topology and the transmission path obstruction characteristics of the signal equipment transmission path in the line topology are extracted from the line operation status data set. Based on the section occupancy status characteristics of each operating section in the line topology, line operation obstruction status description data is constructed. The line operation obstruction status description data is used to characterize the spatial continuous distribution pattern of obstructed sections in the line topology caused by sudden operation events. The obstruction status description data of the line operation and the obstruction characteristics of the transmission path of the signal equipment are input into a pre-constructed emergency dispatch strategy generation network. The obstruction section correlation relationship modeling processing is performed on the obstruction section correlation relationship data to obtain obstruction section correlation relationship features. The obstruction section correlation relationship features and the transmission path obstruction features are then combined to generate a joint dispatch strategy to obtain rail transit emergency plan dispatch strategy data. The rail transit emergency plan dispatch strategy data includes small-loop turnaround dispatch strategy data for obstructed sections, train interval adjustment dispatch strategy data for adjacent unobstructed sections, and passenger flow guidance dispatch strategy data for affected transfer stations. The emergency dispatch strategy data of the rail transit emergency plan is parsed to obtain an emergency dispatch instruction sequence for the rail transit emergency dispatch system. The emergency dispatch instruction sequence includes short-route turnaround dispatch instructions, train interval adjustment dispatch instructions, and passenger flow guidance dispatch instructions.

[0005] Furthermore, embodiments of the present invention also provide a deep learning-based rail transit emergency plan recommendation system, comprising: A processor; a machine-readable storage medium for storing machine-executable instructions of the processor; wherein the processor is configured to execute the aforementioned deep learning-based rail transit emergency response plan recommendation method by executing the machine-executable instructions.

[0006] Based on the above, by acquiring a set of line operation status data including real-time train operation location distribution data, signal equipment status data, and event trigger location data for each section, section occupancy status features and signal equipment transmission path obstruction features are extracted. Then, based on the section occupancy status features, a line operation obstruction situation description data is constructed to characterize the spatial continuous distribution of obstructed sections. This allows the subsequent dispatching network to model relationships based on the spatial topological continuity of obstructed sections rather than isolated point information. On this basis, the line operation obstruction situation description data and transmission path obstruction features are input into a pre-constructed emergency dispatching strategy generation network to jointly generate dispatching strategies based on the obstructed section relationship features and transmission path obstruction features. The processing enables three types of scheduling strategies—the short-route turnaround scheduling strategy for obstructed sections, the train interval adjustment scheduling strategy for adjacent unobstructed sections, and the passenger flow guidance scheduling strategy for affected transfer stations—to be generated collaboratively within the same network framework rather than optimized independently. Furthermore, the generated rail transit emergency plan scheduling strategy data is parsed to obtain an emergency scheduling instruction sequence containing the three types of scheduling instructions. This allows for end-to-end intelligent emergency plan recommendation in sudden operational event scenarios, integrating continuous spatial modeling of obstructed situations, coupled analysis of signal path obstruction, and joint generation of multiple types of scheduling strategies. This significantly improves the matching accuracy between emergency plans and actual obstructed situations, the synergistic consistency of multiple types of scheduling strategies, and the timeliness and executability of emergency response decisions. Attached Figure Description

[0007] Figure 1 This is a schematic diagram of the execution flow of the rail transit emergency plan recommendation method based on deep learning provided in an embodiment of the present invention.

[0008] Figure 2 This is a schematic diagram of the scenario for analyzing the obstruction situation of rail transit lines and generating emergency dispatch strategies provided in the embodiments of this application.

[0009] Figure 3 This is a schematic diagram of the track circuit and turnout linkage control and the dynamic update scenario of obstructed section dissipation provided in the embodiments of this application.

[0010] Figure 4 This is a schematic diagram of the scenario of passenger flow guidance and turnaround section layer configuration and symmetrical turnaround strategy at the transfer station provided in the embodiments of this application. Detailed Implementation

[0011] Figure 1 This is a flowchart illustrating a deep learning-based method for recommending emergency response plans for rail transit, as provided in one embodiment of the present invention. A detailed description follows.

[0012] Step S110: Obtain a set of line operation status data of the rail transit line after the occurrence of a sudden operation event. The set of line operation status data includes real-time train operation location distribution data of each section in the line topology, signal equipment status data of each section, and event trigger location data of the sudden operation event in the line topology.

[0013] In this embodiment, the rail transit line comprises M consecutively arranged operating sections, each with a unique section number. After a sudden operational event occurs, the track occupancy detection device collects the train occupancy status of each section at fixed intervals, the signal equipment status acquisition device synchronously collects the signal equipment transmission status of each section, and the event trigger location acquisition device obtains the event trigger section number. The real-time train operation location distribution data uses a correspondence structure between the section number p and the train occupancy marker O(p), where O(p) takes a value of 0 or 1, where 0 indicates no train occupancy in section p and 1 indicates train occupancy in section p. The signal equipment status data uses a correspondence structure between the section number p and the transmission status marker T(p), where T(p) takes a value of 0 or 1, where 0 indicates the signal equipment transmission path in section p is interrupted and 1 indicates the signal equipment transmission path in section p is available. The event trigger location data uses a correspondence structure between the section number p and the event trigger marker E(p), where E(p) takes a value of 0 or 1, where 0 indicates section p is not an event trigger section and 1 indicates section p is an event trigger section. All three types of data are timestamped to ensure alignment to the same point in time. Data collection is limited to the operational status of the transportation lines and does not involve passengers' personal privacy information.

[0014] Step S120: Extract the segment occupancy status characteristics of each operating segment in the line topology and the transmission path obstruction characteristics of the signal equipment transmission path in the line topology from the line operation status data set, and construct line operation obstruction status description data based on the segment occupancy status characteristics of each operating segment in the line topology. The line operation obstruction status description data is used to characterize the spatial continuous distribution pattern of obstructed segments in the line topology caused by sudden operation events.

[0015] In this embodiment, train occupancy markers O(p) and train direction markers D(p) are extracted segment by segment from the real-time train location distribution data, and arranged in the order of line topology connection to form a segment occupancy status sequence Q. D(p) is determined according to the occupancy time sequence of adjacent segments, taking a value of 0 or 1, where 0 represents the up direction and 1 represents the down direction. The event triggering segment number p0 and its adjacent segment numbers are extracted from the event triggering location data, and the starting position of the obstructed segment and the direction of the obstructed segment are marked in Q to obtain the obstructed segment marker sequence B. The set of segments traversed by the transmission path of each signaling device and the transmission status marker T(p) of each segment are extracted from the signaling device status data. T(p) is associated with B, and the set of segment numbers with T(p) equal to 0 within the obstructed segment is extracted as interrupted segment marker data, and the set of segment numbers with T(p) equal to 1 outside the obstructed segment is extracted as available segment marker data. The two are concatenated to form the transmission path obstruction feature. Based on the spatially continuous distribution of obstructed sections in B, adjacent obstructed sections are merged into obstructed section groups. The first and last sections of each obstructed section group are marked as boundary sections, and the rest are marked as internal sections, resulting in obstructed section spatial distribution labeling data R. R and O(p) are concatenated according to the section number to obtain the line operation obstruction status description data H. Each section in H corresponds to a triplet, which includes the obstructed section group number, section role label, and train occupancy label.

[0016] like Figure 2 As shown, after a sudden operational event occurs, the acquired data set of line operation status is first used to construct a description of the line operation disruption situation. The figure shows the spatial relationship of the event trigger location (marked with an asterisk), the group of disrupted sections (diagonal texture), adjacent undisrupted sections, and the boundaries of the disrupted sections in the form of a line topology map. It also marks the transmission interruption locations in the signal equipment transmission path. The combination of these physical elements will serve as the core component of the subsequent description of the line operation disruption situation, fully reflecting the impact of the sudden operational event on the continuous spatial distribution of the line.

[0017] Step S121: Extract train occupancy marker data and train running direction marker data for each operating section in the line topology from the real-time train running position distribution data of each section in the line operation status data set.

[0018] In this embodiment, M operating sections are traversed, and the train occupancy marker O(p) for each section p is read from the real-time train position distribution data. The extraction rule for the train direction marker D(p) is as follows: obtain the occupancy status of section p at two adjacent collection times. If section p changes from idle to occupied and its upstream adjacent section was occupied at the previous time, then D(p) is 0; if section p changes from idle to occupied and its downstream adjacent section was occupied at the previous time, then D(p) is 1; if section p has no train occupancy or cannot be determined, then D(p) is 0. The O(p) and D(p) of all sections constitute a list of section occupancy pairs.

[0019] Step S122: Arrange the train occupancy marker data and train running direction marker data of each operating section in the line topology according to the actual connection order of each operating section in the line topology to obtain the section occupancy status sequence of the line topology.

[0020] In this embodiment, the line topology adjacency list structure G is obtained. Starting from the line's initial segment, each operating segment is visited sequentially along the adjacency relationship. The segment occupancy tuples of each segment p are arranged according to the visit order to form a segment occupancy state sequence Q. The q-th element in Q corresponds to the tuple of the q-th visited segment. Q is stored using a one-dimensional ordered list structure with a list length equal to M.

[0021] Step S123: Extract the triggering segment marker data and the adjacent segment marker data of the sudden operation event in the line topology from the event triggering location data of the sudden operation event in the line topology from the sudden operation event in the line operation status data set.

[0022] In this embodiment, the segment number where E(p) equals 1 is read from the event trigger location data as the trigger segment marker data p0. The list of directly adjacent segment numbers of p0 is queried according to G, and the adjacent marker value corresponding to each segment number in the list is set to 1, while the adjacent marker values ​​corresponding to the remaining segment numbers remain 0, thus obtaining the adjacent segment marker data of the trigger segment.

[0023] Step S124: Based on the trigger section marking data and the adjacent section marking data of the trigger section, mark the starting position of the blocked section and the direction of the blocked section in the section occupancy status sequence of the line topology to obtain the blocked section marking sequence.

[0024] In this embodiment, the index position corresponding to p0 is located in Q, and this index position is marked as the starting position of the obstructed section. The train occupancy flags corresponding to the preceding and following index positions of p0 in Q are read. If the train occupancy flag of the preceding index position is 1 and the train occupancy flag of the following index position is 0, then the obstructed section impact direction flag V is 0, indicating forward impact; if the train occupancy flag of the following index position is 1 and the train occupancy flag of the preceding index position is 0, then V is 1, indicating backward impact; if both the preceding and following index positions have train occupancy flags of 1, then V is 2, indicating bidirectional impact. An obstruction flag bit is added to each element of Q, with a value of 1 indicating that the section belongs to the obstructed section and a value of 0 indicating that it does not belong to the obstructed section, resulting in the obstructed section flag sequence B.

[0025] Step S125: Extract the set of segments through which the signal equipment transmission path passes and the transmission status marker data of the signal equipment transmission path in each segment from the signal equipment status data of each segment in the line operation status data set.

[0026] In this embodiment, a mapping table between signal equipment transmission paths and operating sections is pre-established. Each record contains the signal equipment transmission path number and a list of operating section numbers traversed by that path. For each signal equipment transmission path, the list of operating section numbers it traverses is extracted according to the mapping table. The T(p) value corresponding to each section in the list is read from the signal equipment status data, and the sections are arranged according to their connection order in the line topology to form a transmission status marker list for that signal equipment transmission path. The transmission status marker lists of all signal equipment transmission paths together constitute a transmission status marker data set.

[0027] Step S126: Associate the transmission status marker data of the signal equipment transmission path in each segment with the obstructed segment marker sequence, and extract the interrupted segment marker data of the signal equipment transmission path within the obstructed segment and the available segment marker data of the signal equipment transmission path outside the obstructed segment, as the transmission path obstruction feature of the signal equipment transmission path in the line topology.

[0028] In this embodiment, for each signal device transmission path transmission status marker list, each operating segment number is traversed one by one, and the obstruction marker bit corresponding to the segment number is queried in B. If the obstruction marker bit is 1 and the corresponding T(p) of the segment is 0, then the segment number is added to the interrupted segment marker data list A1. If the obstruction marker bit is 0 and the corresponding T(p) of the segment is 1, then the segment number is added to the available segment marker data list A2. A1 and A2 are concatenated to form the transmission path obstruction feature F, where the first half of F is the set of interrupted segment numbers and the second half is the set of available segment numbers.

[0029] Step S127: Based on the spatial continuous distribution pattern of the blocked sections in the blocked section marking sequence, adjacent blocked sections are merged into a blocked section group, and the boundary sections and internal sections of the blocked section group in the line topology are distinguished and marked to obtain the spatial distribution marking data of the blocked sections.

[0030] In this embodiment, the obstruction marker bits corresponding to each segment in B are traversed. When a segment with an obstruction marker bit of 1 is encountered, the segment number is recorded as the starting number of the current obstructed segment group. The traversal continues until a segment with an obstruction marker bit of 0 is encountered or the end of B is reached. The sequence of consecutive obstruction marker bits of 1 is then merged into one obstructed segment group. The first and last segments in the segment number list of each obstructed segment group are marked as boundary segments, and the rest are marked as internal segments. The obstructed segment spatial distribution marking data R adopts a corresponding structure of obstructed segment group number, segment role marker, and segment number. The segment role marker takes a value of 0 or 1, where 0 represents an internal segment and 1 represents a boundary segment.

[0031] Step S128: The spatial distribution marking data of the obstructed section is fused with the train occupancy marking data of each operating section in the line topology to obtain line operation obstruction status description data that includes the location of the obstructed section, the boundary of the obstructed section, and the train occupancy status inside the obstructed section.

[0032] In this embodiment, R and O(p) are aligned according to segment numbers. For each segment p, a blocked situation description tuple is generated, containing three fields: blocked segment group number, segment role flag, and train occupancy flag. If segment p does not belong to any blocked segment group, the blocked segment group number and segment role flag are set to null. All blocked situation description tuples for all segments are arranged in ascending order of segment number to form line operation blocked situation description data H. H is stored in a structured list with a list length equal to M.

[0033] Step S130: Input the line operation obstruction situation description data and the transmission path obstruction characteristics of the signal equipment transmission path into the pre-constructed emergency dispatch strategy generation network, perform obstruction section correlation modeling processing on the line operation obstruction situation description data to obtain obstruction section correlation characteristics, and perform joint dispatch strategy generation processing on the obstruction section correlation characteristics and the transmission path obstruction characteristics to obtain rail transit emergency plan dispatch strategy data. The rail transit emergency plan dispatch strategy data includes small-loop turnaround dispatch strategy data for obstructed sections, train interval adjustment dispatch strategy data for adjacent unobstructed sections, and passenger flow guidance dispatch strategy data for affected transfer stations.

[0034] In this embodiment, the emergency dispatch strategy generation network adopts a deep neural network structure, including a blocked section correlation encoding layer and a joint dispatch strategy generation layer. The blocked section correlation encoding layer adopts a graph attention network structure, and the joint dispatch strategy generation layer adopts a combination structure of multilayer perceptron and conditional random field. R from H is input into the blocked section correlation encoding layer, and adjacent section traversal, boundary extraction, and feature encoding are performed on each blocked section group to obtain the blocked section correlation feature C. C and F are input into the joint dispatch strategy generation layer to generate short-route turnaround dispatch strategy data S1, train interval adjustment dispatch strategy data S2, and passenger flow guidance dispatch strategy data S3. S1, S2, and S3 are then concatenated sequentially to form the rail transit emergency plan dispatch strategy data S.

[0035] like Figure 2 As shown, the data describing the obstruction of line operation and the characteristics of obstructed transmission paths of signal equipment are input into the emergency dispatch strategy generation network. The network structure in the figure consists of two layers: the upper layer is the obstructed section association encoding layer (a schematic diagram of the network structure composed of nodes and connections), used for topological modeling of the spatial distribution of obstructed section groups; the lower layer is the joint dispatch strategy generation layer, which outputs small-loop turnaround dispatch strategy data for obstructed sections, train interval adjustment dispatch strategy data for adjacent unobstructed sections, and passenger flow guidance dispatch strategy data for affected transfer stations. These three types of strategies are combined into rail transit emergency plan dispatch strategy data.

[0036] Step S131: Input the spatial distribution marking data of obstructed sections in the description data of the obstructed line operation status into the obstructed section association coding layer in the emergency dispatch strategy generation network. The obstructed section association coding layer performs adjacent section traversal processing on each obstructed section group in the spatial distribution marking data of obstructed sections to obtain the adjacent unobstructed section marking data corresponding to each obstructed section group.

[0037] In this embodiment, the blocked segment association encoding layer includes a node embedding layer, an adjacency aggregation layer, and a multi-head attention calculation layer. The node embedding layer concatenates the segment role markers and train occupancy markers of each segment within each blocked segment group into a dense input vector. This vector is then linearly transformed by the first fully connected sub-layer to obtain an initial node representation vector of dimension K, where K is a pre-defined embedding dimension. The adjacency aggregation layer searches for the directly adjacent unblocked segments of each blocked segment group boundary segment according to G, adds the unblocked segment number to the adjacent unblocked segment number list corresponding to that blocked segment group, and generates adjacency marker data for each adjacent unblocked segment, including the blocked segment group number, the adjacent unblocked segment number, and the adjacency direction marker. The multi-head attention calculation layer performs scaled dot product attention calculations on the initial node representation vector of each blocked segment group and the initial node representation vectors of its adjacent unblocked segments, outputting the adjacent unblocked segment marker data.

[0038] Step S132: The blocked segment association coding layer extracts the segment connection boundary marker data and segment connection direction marker data between each blocked segment group and the adjacent unblocked segment based on the adjacent unblocked segment marker data corresponding to each blocked segment group.

[0039] In this embodiment, for each blocked segment group, the adjacent unblocked segment marker data list is read one by one. Based on the adjacent unblocked segment number, the list of its directly adjacent segment numbers is queried in G. The segment number belonging to the blocked segment group is found in this list, and this segment number is used as the connection boundary segment number. The adjacent unblocked segment number is used as the outer segment number of the connection boundary, and the blocked segment group number is used as the inner blocked segment group number of the connection boundary. These are combined to form the segment connection boundary marker data. The segment connection direction marker data is directly taken from the adjacent marker data.

[0040] Step S133: The blocked segment association relationship encoding layer combines the segment connection order within each blocked segment group, the segment connection boundary marker data between each blocked segment group and the adjacent unblocked segment, and the segment connection direction marker data to obtain the blocked segment association relationship feature.

[0041] In this embodiment, for each blocked segment group, the internal segment number list is sequentially encoded into a segment order vector. The encoding method involves obtaining a segment embedding vector of dimension K by querying the learnable embedding matrix for each segment number. All segment embedding vectors are input into a Long Short-Term Memory (LSTM) network layer, and the segment order vector is output. The segment connection boundary label data and segment connection direction label data are encoded into connection boundary vectors and connection direction vectors, respectively. The connection boundary vector is obtained by concatenating the connection boundary segment number and the segment number outside the connection boundary after embedding query. The connection direction vector is obtained through one-hot encoding. The segment order vector, connection boundary vector, and connection direction vector are concatenated along the feature dimension to form a combined feature vector. The combined feature vectors of all blocked segment groups are arranged by group number to form the blocked segment association feature C.

[0042] Step S134: Input the obstructed section association characteristics and the transmission path obstruction characteristics of the signal equipment transmission path into the joint scheduling strategy generation layer in the emergency scheduling strategy generation network. The joint scheduling strategy generation layer performs turnaround section availability analysis processing on the section connection boundary marking data in the obstructed section association characteristics to obtain turnaround section marking data that can perform small-loop turnaround.

[0043] In this embodiment, the multilayer perceptron of the joint scheduling strategy generation layer includes an input layer, two fully connected hidden layers, and an output layer. The segment connection boundary label data in C is converted into a connection boundary segment feature matrix, where the number of rows represents the number of blocked segment groups, and the number of columns represents the number of connection boundary segments. The multilayer perceptron performs a nonlinear transformation on the connection boundary segment feature matrix, outputting an availability score for each connection boundary segment. The conditional random field (CRF) takes the availability score matrix as input and outputs an availability label for each connection boundary segment through Viterbi decoding. The label value is either 0 or 1, with 1 indicating that a short-loop reversal is possible. The connection boundary segment numbers with label 1 are collected as reversal segment label data L1.

[0044] Step S135: The joint scheduling strategy generation layer generates small-route turnaround scheduling strategy data containing the turnaround section location and turnaround direction based on the turnaround section marking data and the section connection direction marking data in the blocked section association characteristics.

[0045] In this embodiment, for each turnaround segment number in L1, the corresponding segment connection direction marker is queried in C. If the segment connection direction marker is 0, the turnaround running direction marker is 1; if the segment connection direction marker is 1, the turnaround running direction marker is 0. The turnaround segment number and the turnaround running direction marker are combined into a turnaround strategy tuple. All turnaround strategy tuples are summarized into short-route turnaround scheduling strategy data S1.

[0046] Step S136: The joint scheduling strategy generation layer performs train interval division processing on the available section marking data of the signal equipment transmission path outside the blocked section in the transmission path blocked feature of the signal equipment transmission path, and obtains the train interval adjustment scheduling strategy data of the adjacent unblocked section.

[0047] In this embodiment, available segment marker data A2 is read from F, and the segment numbers in A2 are sorted in the line topology according to the connection order. For each available segment number, the number of interval segments between it and the boundary segment of the nearest blocked segment group is calculated. The number of interval segments is defined as the absolute value of the difference between their numbers. Available segments with an interval segment number equal to a first distance threshold are classified into a first train interval level, available segments with an interval segment number equal to a second distance threshold are classified into a second train interval level, and available segments with an interval segment number greater than the second distance threshold are classified into a third train interval level. The train interval adjustment scheduling strategy data S2 adopts a correspondence structure between segment numbers and train interval level markers.

[0048] Step S137: The joint scheduling strategy generation layer performs passenger flow diversion path planning processing on the affected transfer stations based on the adjacent unobstructed section marking data in the obstructed section association relationship characteristics and the train occupancy status inside the obstructed section in the line operation obstruction status description data, to obtain passenger flow guidance scheduling strategy data for the affected transfer stations.

[0049] In this embodiment, adjacent unobstructed section marker data is read from C, and train occupancy markers within obstructed sections are read from H. The operating section number corresponding to each transfer station in the line topology is queried. If the operating section number belongs to any obstructed section group or the list of adjacent unobstructed sections of any obstructed section group, then the transfer station is marked as an affected transfer station. Passenger flow diversion path planning is completed in the passenger flow diversion path planning sub-layer of the joint scheduling strategy generation layer. This sub-layer adopts a graph attention network structure, taking the set of affected transfer stations, adjacent unobstructed section marker data, and train occupancy status within obstructed sections as input, and outputs a transfer direction reachability marker for each affected transfer station. The reachability marker value is 0 or 1. Passenger flow diversion path markers are generated for each affected transfer station based on the reachability markers, including the transfer station number and diversion path direction marker. The passenger flow diversion path markers of all affected transfer stations are summarized into passenger flow guidance scheduling strategy data S3.

[0050] Step S138: Combine the short-route turnaround scheduling strategy data, the train interval adjustment scheduling strategy data, and the passenger flow guidance scheduling strategy data to obtain the rail transit emergency plan scheduling strategy data.

[0051] In this embodiment, S1, S2, and S3 are concatenated in sequence to form the rail transit emergency plan dispatch strategy data S. S contains three fields: the first field is the short-route turnaround dispatch strategy data, the second field is the train interval adjustment dispatch strategy data, and the third field is the passenger flow guidance dispatch strategy data.

[0052] Step S140: Parse the emergency dispatch strategy data of the rail transit emergency plan to obtain an emergency dispatch instruction sequence for the rail transit emergency dispatch system. The emergency dispatch instruction sequence includes short-route turnaround dispatch instructions, train interval adjustment dispatch instructions, and passenger flow guidance dispatch instructions.

[0053] In this embodiment, strategy instructions are parsed for S and executed in three directions. The first direction extracts the turnaround section location marker and turnaround direction marker from S1, converting them into turnaround section number data and turnaround direction instruction code data. The second direction extracts the section number and train interval level marker from S2, converting them into train interval adjustment instruction code data. The third direction extracts the transfer station number and passenger flow diversion path marker from S3, converting them into passenger flow guidance instruction code data. The parsing results from the three directions are sorted according to instruction issuance priority, with the priority being turnaround instruction, train interval adjustment instruction, and passenger flow guidance instruction, to obtain the emergency dispatch instruction code sequence, which is then used to generate the emergency dispatch instruction sequence.

[0054] like Figure 2 As shown, the emergency response plan dispatch strategy data for rail transit is processed by the strategy instruction parsing module and transformed into specific instructions for the rail transit emergency dispatch system. The figure shows three types of instruction cards after parsing: short-route turnaround dispatch instructions, train interval adjustment dispatch instructions, and passenger flow guidance dispatch instructions. These instructions are combined into an emergency dispatch instruction sequence, which is directly issued to the system for execution.

[0055] Step S141: Extract the turnaround section location marker data and turnaround running direction marker data from the small-route turnaround scheduling strategy data in the rail transit emergency plan scheduling strategy data, convert the turnaround section location marker data into turnaround section number data in the line topology, and convert the turnaround running direction marker data into turnaround direction instruction code data.

[0056] In this embodiment, the turnaround segment number and turnaround direction marker of each turnaround strategy tuple in S1 are read. The turnaround segment number data and the turnaround segment position marker data use the same segment numbering system. The mapping relationship of the turnaround direction command encoding is as follows: when the turnaround direction marker is 0, the turnaround direction command encoding value is the first encoding value; when the turnaround direction marker is 1, the turnaround direction command encoding value is the second encoding value. Each turnaround strategy tuple is converted into a turnaround command encoding tuple, which contains the turnaround segment number and the turnaround direction command encoding.

[0057] Step S142: Extract the section number data of adjacent unobstructed sections and the train interval level mark data of each section from the train interval adjustment scheduling strategy data in the rail transit emergency plan scheduling strategy data, and convert the section number data and the train interval level mark data of each section into train interval adjustment instruction code data for the train operation control system.

[0058] In this embodiment, the correspondence between each segment number and the train interval level marker in S2 is read. The train interval adjustment instruction coding mapping relationship is as follows: when the level marker is 0, the code value is the first code value; when the level marker is 1, the code value is the second code value; and when the level marker is 2, the code value is the third code value. Each correspondence is converted into a train interval adjustment instruction code tuple, containing the segment number and the train interval adjustment instruction code.

[0059] Step S143: Extract the transfer station number data and passenger flow diversion path marking data of the affected transfer stations from the passenger flow guidance and scheduling strategy data in the rail transit emergency plan scheduling strategy data, and convert the transfer station number data and the passenger flow diversion path marking data into passenger flow guidance instruction code data for the station passenger information release system.

[0060] In this embodiment, the transfer station number and passenger flow diversion path marker of each affected transfer station in S3 are read. The passenger flow guidance instruction encoding mapping relationship is as follows: when the diversion path marker is 0, the encoding value is the first encoding value; when the diversion path marker is 1, the encoding value is the second encoding value. Each affected transfer station is converted into a passenger flow guidance instruction encoding tuple, containing the transfer station number and the passenger flow guidance instruction code.

[0061] Step S144: Sort the turnaround section number data, the turnaround direction instruction code data, the train interval adjustment instruction code data, the transfer station number data, and the passenger flow guidance instruction code data according to the instruction issuance priority to obtain the emergency dispatch instruction code sequence.

[0062] In this embodiment, the list of turnaround instruction code tuples is arranged at the beginning of the emergency dispatch instruction code sequence, the list of train interval adjustment instruction code tuples is arranged in the middle, and the list of passenger flow guidance instruction code tuples is arranged at the end. Within each level, the tuples are sorted in ascending order by section number or transfer station number.

[0063] Step S145: Generate short-route turnaround scheduling instructions, train interval adjustment scheduling instructions, and passenger flow guidance scheduling instructions for the rail transit emergency dispatch system according to the emergency dispatch instruction encoding sequence, and combine the short-route turnaround scheduling instructions, the train interval adjustment scheduling instructions, and the passenger flow guidance scheduling instructions into an emergency dispatch instruction sequence.

[0064] In this embodiment, each instruction code tuple in the emergency dispatch instruction encoding sequence is traversed and packetized according to the corresponding message format. The short-route turnaround dispatch instruction message includes an instruction type field, a turnaround section number field, and a turnaround direction instruction code field. The train interval adjustment dispatch instruction message includes an instruction type field, a section number field, and a train interval adjustment instruction code field. The passenger flow guidance dispatch instruction message includes an instruction type field, a transfer station number field, and a passenger flow guidance instruction code field. All the instructions obtained from the packetization are arranged in the order of the encoding sequence and combined into an emergency dispatch instruction sequence.

[0065] Step S210: Obtain the real-time train position change data set of the rail transit line after the occurrence of a sudden operational event. The real-time train position change data set includes the train position update data of each section in the line topology after the emergency dispatch instruction sequence is issued.

[0066] In this embodiment, after the emergency dispatch command sequence is issued, train position update data for each section is continuously collected. Each reported data includes a section number, a train occupancy marker, and a timestamp. The reported data from each section within the same time window are aggregated into a real-time train position change data set, represented by a correspondence between section numbers and train occupancy markers.

[0067] like Figure 3 As shown in the figure, the changing trend of the obstruction section's dissipation status is illustrated using a time axis and waveform curves. The dots on the curves mark the differences in train position changes for sections where short-route turnarounds have been implemented and the differences in train position changes for sections where train interval adjustments have been implemented. These data are input into the obstruction section dissipation trend analysis module, and after "obstruction section correlation update processing," updated obstruction section correlation characteristics are generated. Finally, these characteristics are input into the joint dispatching strategy generation layer, outputting an updated emergency dispatching instruction sequence.

[0068] Step S220: Compare the real-time train position change data set with the real-time train position distribution data of each section in the line operation status data set to obtain the difference marker data of train position change in each section.

[0069] In this embodiment, for each segment number, the train occupancy marker in the real-time train position change data set is compared with the train occupancy marker of the same segment in the line operation status data set. If the train occupancy markers for that segment are the same in both data sources, the difference marker value is 0; otherwise, the difference marker value is 1. The difference markers of all segments are arranged according to the segment number to form the train position change difference marker data for each segment.

[0070] Step S230: Extract the train position change difference marker data for sections that have performed short-route turnarounds and the train position change difference marker data for sections that have performed train interval adjustments based on the train position change difference marker data for each section.

[0071] In this embodiment, the difference markers of the corresponding sections are extracted from the train position change difference marker data of each section, combined with the list of turnaround section numbers and the list of train interval adjustment section numbers in the emergency dispatch instruction sequence, to form the train position change difference marker data of the sections that have performed small-route turnaround and the train position change difference marker data of the sections that have performed train interval adjustment.

[0072] Step S240: Input the difference marking data of train position changes in the sections where small-route turnaround has been performed and the difference marking data of train position changes in the sections where train interval adjustment has been performed into the emergency dispatch strategy generation network, and perform obstructed section dissipation trend analysis processing on the difference marking data of train position changes in the sections where small-route turnaround has been performed to obtain obstructed section dissipation status description data.

[0073] In this embodiment, the emergency dispatch strategy generation network also includes a sub-layer for analyzing the dissipation trend of obstructed sections. This sub-layer adopts a time-series convolutional network structure, including causal convolutional layers, dilated convolutional layers, and residual connection layers. The input consists of two types of differentially labeled data within multiple time windows arranged in chronological order. The causal convolutional layer performs convolution operations along the time dimension, the dilated convolutional layer extracts dissipation trend features, and the residual connection layer adds the input to the output of the dilated convolutional layer and then processes it through an activation function. The output is a description of the dissipation status of the obstructed section, using a correspondence structure between section numbers and dissipation status labels. The dissipation status label takes a value of 0 or 1, where 0 indicates that the obstruction is still present and 1 indicates that the obstruction has been resolved.

[0074] Step S250: Based on the description data of the dissipation state of the obstructed section, perform obstructed section association relationship update processing on the obstructed section association relationship features to obtain the updated obstructed section association relationship features.

[0075] In this embodiment, the blocked segment group corresponding to the segment marked as 1 in the dissipation state is found in C. The combined feature vector of the blocked segment group is removed from C, and the remaining combined feature vectors are rearranged according to the group number to obtain the updated blocked segment association features.

[0076] Step S260: Input the updated obstructed section association characteristics and the transmission path obstruction characteristics into the joint scheduling strategy generation layer in the emergency scheduling strategy generation network, and the joint scheduling strategy generation layer generates updated rail transit emergency plan scheduling strategy data.

[0077] In this embodiment, the updated obstructed section correlation characteristics are combined with the F input joint scheduling strategy generation layer, and steps S134 to S138 are executed to generate updated rail transit emergency plan scheduling strategy data.

[0078] Step S270: Parse the updated rail transit emergency plan dispatch strategy data to obtain an updated emergency dispatch instruction sequence. The updated emergency dispatch instruction sequence is used to adjust the short-route turnaround dispatch instruction and the train interval adjustment dispatch instruction.

[0079] In this embodiment, steps S141 to S145 are performed on the updated rail transit emergency plan dispatch strategy data to obtain the updated emergency dispatch instruction sequence.

[0080] Step S310: Obtain the track circuit status data set of each operating section in the line topology. The track circuit status data set includes the occupancy indication data of the track circuit in each section, the fault indication data of the track circuit in each section, and the transmission interruption indication data of the track circuit in each section.

[0081] In this embodiment, the track circuit status acquisition device collects the track circuit status of each section. Occupancy data uses a correspondence between section number and occupancy marker, with the occupancy marker value being 0 or 1. Fault data uses a correspondence between section number and fault marker, with the fault marker value being 0 or 1. Transmission interruption data uses a correspondence between section number and transmission interruption marker, with the transmission interruption marker value being 0 or 1. The acquisition process complies with the rail transit signaling equipment management regulations.

[0082] like Figure 3 As shown, in the track circuit status correlation analysis stage, the system established correlation markers between the track circuit and the signal equipment transmission path. The figure illustrates a track circuit fault representation (a diagonally textured rectangular block), with vertical dashed lines connecting the track circuit and signal equipment transmission path. The figure further uses arrows to show the track circuit fault-affected section marker data obtained from the fault-affected range analysis. Consistency analysis is performed between this track circuit fault-affected section marker data and the obstructed section marker data, ultimately generating supplementary obstructed section marker data.

[0083] Step S320: Perform track circuit signal correlation analysis processing on the track circuit status data set and the signal equipment status data of each section in the line operation status data set to obtain the correlation marker data of the track circuit and signal equipment transmission path.

[0084] In this embodiment, a mapping table relating track circuits and signal equipment transmission paths is pre-established. Each segment number in the track circuit status data set is traversed, and the corresponding signal equipment transmission path number list is queried from the mapping table. The corresponding transmission status flag is then read from the signal equipment status data. If the track circuit transmission interruption flag for that segment is 1 and the signal equipment transmission status flag is 0, then the association flag is set to 1; otherwise, the association flag is set to 0.

[0085] Step S330: Based on the association marker data of the track circuit and signal equipment transmission path, extract the track circuit fault indication data associated with the signal equipment transmission path in the obstructed section from the track circuit status data set.

[0086] In this embodiment, each segment number in A1 is traversed, and the associated tag corresponding to the segment number is queried in the associated tag data. If the associated tag is 1, the fault indication tag of the segment is read from the track circuit status data set, and the correspondence between the segment number and the fault indication tag is added to the track circuit fault indication data.

[0087] Step S340: Input the track circuit fault representation data associated with the signal equipment transmission path of the obstructed section into the emergency dispatch strategy generation network, perform fault impact range analysis on the track circuit fault representation data of each section in the track circuit fault representation data, and obtain track circuit fault impact section marking data.

[0088] In this embodiment, the emergency dispatch strategy generation network further includes a fault impact range analysis sublayer. This sublayer employs a graph convolutional network structure, comprising an input layer, a graph convolutional layer, a graph pooling layer, and an output layer. The input layer converts the track circuit fault representation data into a segment node feature matrix. The graph convolutional layer performs neighborhood aggregation operations based on G, using a symmetric normalized Laplacian operator. After multiple layers of graph convolutional layers are stacked, the graph pooling layer performs graph-level aggregation. The output layer outputs the fault impact marker for each segment. The segment numbers with a fault impact marker of 1 are collected as track circuit fault impact segment marker data.

[0089] Step S350: Perform a consistency analysis of the obstruction range based on the track circuit fault-affected section marking data and the obstruction section spatial distribution marking data in the line operation obstruction status description data to obtain obstruction range consistency analysis result data.

[0090] In this embodiment, each segment number in the track circuit fault-affected segment marking data is compared one by one with the segment number list of blocked segment groups in R. If a segment number does not appear in any blocked segment group, it is marked as an inconsistent segment and recorded in the inconsistent segment list in the consistency analysis result data; if it appears in a blocked segment group, it is marked as a consistent segment.

[0091] Step S360: Based on the consistency analysis results of the obstructed range, supplement the spatial distribution marking data of the obstructed section in the description data of the obstructed operation status of the line to obtain supplemented obstructed section marking data.

[0092] In this embodiment, for each segment number in the inconsistent segment list, its adjacency relationship with an existing blocked segment group is queried in G. If it is directly adjacent to the boundary segment of a certain blocked segment group, the segment number is added to the segment number list of the blocked segment group, the segment role label is updated to boundary segment, and the original boundary segment role label is updated to internal segment.

[0093] Step S370: Input the supplementary obstructed section marking data into the emergency dispatch strategy generation network, perform obstructed section correlation modeling processing on the supplementary obstructed section marking data, and obtain the supplementary obstructed section correlation features.

[0094] In this embodiment, the supplementary blocked segment marker data is input into the blocked segment association relationship encoding layer, and the processing steps S131 to S133 are performed to obtain the supplementary blocked segment association relationship features.

[0095] Step S380: Based on the characteristics of the correlation between the supplementary obstructed sections, the rail transit emergency plan scheduling strategy data is supplemented and generated to obtain supplementary rail transit emergency plan scheduling strategy data.

[0096] In this embodiment, the supplementary obstructed section correlation characteristics and F input are combined to generate a joint scheduling strategy layer, and steps S134 to S138 are executed to generate supplementary rail transit emergency plan scheduling strategy data.

[0097] Step S410: Obtain event type label data of sudden operational events in the line topology and train timetable data of each operating section in the line topology.

[0098] In this embodiment, the event type label data is reported by the sudden operational event detection device, using a correspondence structure between event type number and event type label. The event type label value is 0, 1, or 2, where 0 represents a track obstacle event, 1 represents a signal equipment failure event, and 2 represents a turnout failure event. Train timetable data is obtained from the operation timetable management system, using a correspondence structure between section number and planned departure and arrival times.

[0099] like Figure 4 As shown, the sequence of "trains diverting traffic in the short-route turnaround direction" and "trains diverting traffic in the continuing direction" is displayed using a timeline and train icon sequence. The turnaround execution sequence is supplemented by "train diversion sequence data" to obtain turnaround execution sequence marker data, and the short-route turnaround scheduling strategy data is further supplemented and generated.

[0100] Step S420: Extract the event-affected section range marker data corresponding to the event type marker data from the preset event type and obstructed section distribution relationship data according to the event type marker data.

[0101] In this embodiment, a data table relating event types to the distribution of obstructed sections is pre-established. Each record contains an event type marker and the distribution range of the obstructed section. The distribution range of the obstructed section is represented by the number of upstream and downstream offset sections based on the event triggering section. The corresponding obstructed section distribution range is queried based on the event type marker. The upper bound of the event-affected section range is obtained by adding the event triggering section number to the downstream offset section number, and the lower bound is obtained by subtracting the upstream offset section number. All section numbers between the upper and lower bounds are collected as event-affected section range marker data.

[0102] Step S430: Perform timetable segment coverage analysis on the event-affected segment range marking data and the train timetable data of each operating segment in the line topology to obtain the planned departure sequence marking data and planned arrival sequence marking data of each segment during the event-affected period.

[0103] In this embodiment, each segment number in the event-affected segment range marking data is traversed, and the planned departure time and planned arrival time corresponding to the segment number are queried in the train timetable data. The planned departure sequence marking data is formed by sorting the planned departure times from smallest to largest, and the planned arrival sequence marking data is formed by sorting the planned arrival times from smallest to largest.

[0104] Step S440: Based on the planned departure sequence marker data and planned arrival sequence marker data of each section within the event impact period, extract the on-the-way train marker data located within the event impact section and the on-the-way train marker data located outside the event impact section from the real-time train operation location distribution data of each section in the line operation status data set.

[0105] In this embodiment, each segment number in the real-time train operation location distribution data is traversed. If the segment number appears in the planned departure sequence marker data or the planned arrival sequence marker data and the train occupancy marker is 1, it is added to the list of on-the-way train marker data within the event-affected segment. If it does not appear and the train occupancy marker is 1, it is added to the list of on-the-way train marker data outside the event-affected segment.

[0106] Step S450: Input the on-the-way train marking data located within the event-affected section and the on-the-way train marking data located outside the event-affected section into the emergency dispatch strategy generation network, and perform evacuation direction division processing on the on-the-way train marking data located within the event-affected section to obtain evacuation train marking data in the short-route turnaround direction and evacuation train marking data in the continuing operation direction.

[0107] In this embodiment, the emergency dispatch strategy generation network further includes a deportation direction partitioning sublayer. This deportation direction partitioning sublayer adopts a sequence-to-sequence network structure, with the encoder employing a bidirectional long short-term memory network structure and the decoder employing a unidirectional long short-term memory network structure. The input is a list of two types of on-the-go train marker data. The encoder performs embedding encoding on each segment number and outputs forward and reverse hidden state sequences, which are concatenated into a context vector. The decoder generates deportation direction markers segment by segment based on the context vector. The marker value is 0 or 1, where 0 indicates deportation in the short-route turnaround direction and 1 indicates deportation in the continuing direction. Segment numbers marked with 0 are collected as train marker data for short-route turnaround direction deportation, and segment numbers marked with 1 are collected as train marker data for the continuing direction deportation.

[0108] Step S460: Based on the train evacuation marking data in the short-route turnaround direction and the train evacuation marking data in the continuing direction, and combined with the on-the-way train marking data located outside the event-affected section, generate train evacuation sequence data.

[0109] In this embodiment, the list of train markers for the short-route turnaround direction is arranged from near to far according to the upper boundary of the event-affected section range, the list of train markers for the continuing direction is arranged from near to far according to the lower boundary of the event-affected section range, and the list of train markers for trains on the way that are outside the event-affected section range is arranged from small to large according to the number of segments from the boundary of the event-affected section range. The three arrangement results are then concatenated to form the train evacuation sequence data.

[0110] Step S470: Perform turnaround execution order supplementation processing on the short-route turnaround scheduling strategy data according to the train dispersal sequence data to obtain short-route turnaround scheduling strategy data containing turnaround execution order marker data.

[0111] In this embodiment, each segment number in the train evacuation sequence data is traversed. If the segment number appears in the turnaround segment number list of S1, the position index corresponding to the segment number in the train evacuation sequence data is appended to the turnaround strategy tuple in S1 as a turnaround execution order marker.

[0112] Step S510: Obtain the real-time passenger flow distribution data set of each transfer station in the line topology. The real-time passenger flow distribution data set includes the platform passenger flow density marker data, transfer passage passenger flow density marker data, and entrance / exit passenger flow density marker data of each transfer station.

[0113] In this embodiment, the passenger flow density detection device collects passenger flow distribution data at each transfer station. The platform passenger flow density marker data uses a correspondence structure between the transfer station number and the platform passenger flow density level, with a level value of 0, 1, or 2, where 0 represents low density, 1 represents medium density, and 2 represents high density. The transfer passage passenger flow density marker data and the entrance / exit passenger flow density marker data use the same level value method. The passenger flow density level is obtained by converting passenger flow statistics through interval mapping rules. The collected data does not involve passengers' personal identification information.

[0114] like Figure 4 As shown in the figure, the top-down view of the transfer station illustrates the passenger flow distribution on the internal platforms. The figure marks the locations of "transfer direction blockages" and indicates "alternative transfer routes" using dashed detour lines. The system obtains the marked data for in-station diversion routes through "intra-station diversion route generation" and combines this with external transfer stations in the blocked section group to perform "transfer route diversion supplementation," ultimately generating "diversion passenger flow guidance marking data" and "transfer station coordinated passenger flow guidance and scheduling strategy data."

[0115] Step S520: Perform a transfer station obstruction correlation analysis on the real-time passenger flow distribution data set and the obstruction section spatial distribution marker data in the line operation obstruction situation description data to obtain transfer station marker data affected by obstruction sections and transfer station marker data not affected by obstruction sections.

[0116] In this embodiment, each transfer station number in the real-time passenger flow distribution data set is traversed. The corresponding operating segment number for each transfer station is queried in G, and whether the operating segment number belongs to any blocked segment group or the list of adjacent unblocked segments of any blocked segment group is queried in R. If it belongs, the transfer station number is added to the list of transfer stations affected by blocked segments; otherwise, it is added to the list of transfer stations unaffected by blocked segments.

[0117] Step S530: Input the transfer station marking data affected by the obstructed section and the real-time passenger flow distribution data set into the emergency dispatch strategy generation network, and perform transfer direction accessibility analysis processing on each affected transfer station in the transfer station marking data affected by the obstructed section to obtain the transfer direction accessibility marking data of each affected transfer station.

[0118] In this embodiment, the emergency dispatch strategy generation network also includes a transfer direction accessibility analysis sublayer. This sublayer adopts a graph attention network structure, comprising a node feature encoding layer, a graph attention layer, and an output classification layer. The node feature encoding layer concatenates the platform passenger flow density markers, transfer passage passenger flow density markers, and entrance / exit passenger flow density markers for each affected transfer station, and then performs a fully connected transformation to obtain the transfer station node feature vector. The graph attention layer constructs a transfer network graph with affected transfer stations as nodes and transfer passage connections as edges, performing neighborhood attention aggregation on each node. The output classification layer outputs an accessibility marker for each transfer direction, with a value of 0 or 1.

[0119] Step S540: Based on the accessibility marker data of the transfer direction of each affected transfer station and the platform passenger flow density marker data in the real-time passenger flow distribution data set, perform in-station diversion path generation processing on the affected transfer stations to obtain the in-station diversion path marker data of the affected transfer stations.

[0120] In this embodiment, for each affected transfer station, all transfer directions with an accessibility mark of 1 are read from its transfer direction accessibility mark list, and sorted from low to high according to the platform passenger flow density level corresponding to the transfer direction. The sorted transfer direction list is used as the in-station diversion path mark for the transfer station.

[0121] Step S550: Perform strategy comparison processing based on the in-station diversion path marking data of the affected transfer station and the passenger flow guidance and scheduling strategy data in the rail transit emergency plan scheduling strategy data to obtain supplementary marking data for passenger flow guidance and scheduling strategy.

[0122] In this embodiment, for each affected transfer station, the passenger flow diversion path marker in S3 is compared with the station's internal diversion path marker list. If the passenger flow diversion path marker is not included in the station's internal diversion path marker list, the transfer station number is added to the passenger flow guidance and scheduling strategy supplementary marker data list, and the diversion path marker in the station's internal diversion path marker list is supplemented.

[0123] Step S560: Supplement the passenger flow guidance and scheduling strategy data of the affected transfer station with the supplementary tagging data according to the passenger flow guidance and scheduling strategy, so as to obtain passenger flow guidance and scheduling strategy data containing the tagging data of the diversion path within the station.

[0124] In this embodiment, for each transfer station number in the supplementary tag data of the passenger flow guidance and scheduling strategy, each diversion path tag in the supplementary diversion path tag list is added to the passenger flow diversion path tag corresponding to the transfer station in S3.

[0125] Step S610: Obtain the turnout status data set of each operating section in the line topology. The turnout status data set includes the current open direction marking data, turnout locking status marking data, and turnout control mode marking data of each section's turnouts.

[0126] In this embodiment, the turnout status acquisition device collects the turnout status of each section. The current open direction marker data uses a correspondence between the turnout number and the open direction marker, with the open direction marker value being 0 or 1, where 0 represents the forward direction and 1 represents the lateral direction. The turnout locking status marker data uses a correspondence between the turnout number and the locking status marker, with the locking status marker value being 0 or 1, where 0 represents not locked and 1 represents locked. The turnout control mode marker data uses a correspondence between the turnout number and the control mode marker, with the control mode marker value being 0 or 1, where 0 represents automatic control mode and 1 represents manual control mode.

[0127] like Figure 3 As shown in the diagram, turnout positions are marked with diamond symbols, illustrating the relationship between track branching and reversing directions. Turnouts within the obstructed section are labeled "Locked Direction Maintained," while turnouts at the boundary of the obstructed section are labeled "Reversing Direction Open." The turnout linkage control sequence undergoes processing from "Turnout Linkage Control Sequence → Turnout Linkage Control Command Sequence," ultimately being embedded as a supplementary item into the short-route reversing scheduling command, forming a short-route reversing scheduling command that includes turnout linkage control commands.

[0128] Step S620: Perform turnout position correlation analysis on the turnout status data set and the spatial distribution marking data of the obstructed section in the line operation obstruction situation description data to obtain turnout marking data located inside the obstructed section and turnout marking data located at the boundary of the obstructed section.

[0129] In this embodiment, each turnout number in the turnout status data set is traversed. The corresponding operating section number for the turnout is queried in G, and the role label of the operating section number is queried in R. If the role label is an internal section, the turnout number is added to the turnout label data list located inside the obstructed section; if the role label is a boundary section, the turnout number is added to the turnout label data list located at the boundary of the obstructed section.

[0130] Step S630: Input the turnout marking data located inside the obstructed section and the turnout marking data located at the boundary of the obstructed section into the emergency dispatch strategy generation network. Based on the turnout locking status marking data in the turnout marking data located inside the obstructed section, perform locking direction maintenance processing on the turnouts inside the obstructed section to obtain turnout control command data inside the obstructed section.

[0131] In this embodiment, for each turnout number in the turnout marking data list located within the obstructed section, its locking status flag is read. If the locking status flag is 1, a locking direction maintenance command is generated; if the locking status flag is 0, a locking execution command is generated, locking the turnout in the currently open direction. All commands are summarized into turnout control command data within the obstructed section.

[0132] Step S640: Based on the turnout current opening direction marking data in the turnout marking data located at the boundary of the obstructed section and the turnout section position marking data in the short-distance turnout scheduling strategy data, perform turnout direction opening processing on the turnout located at the boundary of the obstructed section to obtain the obstructed section boundary turnout control command data.

[0133] In this embodiment, for each turnout number in the turnout marking data list located at the boundary of the obstructed section, the corresponding operating section number is queried in G, and the operating section number is checked in the turnaround section position marking data of S1 to see if the operating section number appears. If it appears, the current open direction marking and the turnaround running direction marking are read. If they are inconsistent, a turnaround direction opening command is generated; if they are consistent, a direction holding command is generated.

[0134] Step S650: Combine the turnout control command data inside the obstructed section and the turnout control command data at the boundary of the obstructed section according to the turnout linkage control sequence to obtain the turnout linkage control command sequence.

[0135] In this embodiment, the control commands for the turnouts at the boundary of the obstructed section are arranged first, followed by the control commands for the turnouts inside the obstructed section. The control commands at the boundary are arranged according to the execution order of the turnout section, and the control commands for the turnouts inside are arranged in ascending order of the turnout number, forming a sequence of turnout linkage control commands.

[0136] Step S660: Perform turnout linkage supplementation processing on the short-route turnaround scheduling instructions in the emergency dispatching instruction sequence according to the turnout linkage control instruction sequence to obtain a short-route turnaround scheduling instruction containing the turnout linkage control instructions.

[0137] In this embodiment, each turnout control instruction in the turnout linkage control instruction sequence is appended to the corresponding turnout strategy tuple in the short-route turnout scheduling instruction, so that each turnout strategy tuple has a turnout linkage control instruction field added.

[0138] For example, step S710: Obtain a set of line communication status data after a sudden operational event occurs. The set of line communication status data includes vehicle-to-ground communication link status marker data for each operating section in the line topology, communication node status marker data between each section, and communication link transmission delay marker data.

[0139] In this embodiment, the communication status monitoring device collects line communication status data. The vehicle-to-ground communication link status marker data uses a structure that maps segment numbers to vehicle-to-ground communication link status markers, with values ​​of 0 or 1, where 0 indicates interruption and 1 indicates normal operation. The communication node status marker data uses a structure that maps communication node numbers to communication node status markers, with values ​​of 0 or 1, where 0 indicates fault and 1 indicates normal operation. The communication link transmission delay marker data uses a structure that maps communication link numbers to transmission delay level markers, with values ​​of 0, 1, or 2.

[0140] like Figure 3 As shown in the figure, the "communication interruption section" is marked with a horizontal line and a cross, and the "emergency communication support path" is represented by a thick dashed arc-shaped detour path. This emergency communication support path bypasses the blocked area for the interruption section, ensuring the reliable transmission of emergency dispatch instructions.

[0141] Step S720: Perform communication link and signal equipment association analysis on the communication status data set of the line and the signal equipment status data of each section in the line operation status data set to obtain the association tag data of communication link and signal equipment.

[0142] In this embodiment, a mapping table between communication links and signaling equipment is pre-established. Each communication link number in the line communication status data set is traversed, and the list of signaling equipment numbers associated with that communication link is queried in the mapping table. The transmission status flag of the corresponding segment is read from the signaling equipment status data. If the transmission status flag of the segment where the signaling equipment is located is 0 and the vehicle-to-ground communication link status flag corresponding to that communication link is 0, then the association flag is set to 1; otherwise, the association flag is set to 0.

[0143] Step S730: Based on the association tag data between the communication link and the signal device, extract the communication link status tag data within the blocked section and the communication link status tag data outside the blocked section from the line communication status data set.

[0144] In this embodiment, each segment number in B is traversed. If the blocked flag bit is 1, the segment number and the corresponding communication link status flag are added to the communication link status flag data list within the blocked segment; if the blocked flag bit is 0, the segment number is added to the communication link status flag data list outside the blocked segment.

[0145] Step S740: Input the communication link status marker data within the blocked section and the communication link status marker data outside the blocked section into the emergency dispatch strategy generation network, and perform communication interruption section location processing on the communication link status marker data within the blocked section to obtain communication interruption section location marker data.

[0146] In this embodiment, the emergency dispatch strategy generation network also includes a communication interruption segment location sublayer, which adopts an autoencoder network structure. The encoder part contains multiple fully connected layers, and the decoder part contains multiple fully connected layers symmetrical to the encoder part. The input is a list of communication link status marker data within the blocked segment. The encoder compresses the input into a low-dimensional latent space vector, and the decoder reconstructs it into reconstructed data with the same dimension as the input. The reconstruction error between the input and the reconstructed data is calculated. Segment numbers with reconstruction errors greater than a preset threshold are marked as communication interruption segments. All communication interruption segment numbers are collected as communication interruption segment location marker data.

[0147] Step S750: Based on the communication interruption segment location marker data and the communication link status marker data outside the blocked segment, perform communication support path planning processing to obtain emergency communication support path marker data.

[0148] In this embodiment, the segment numbers with a vehicle-to-ground communication link status marker of 1 are filtered from the communication link status marker data outside the blocked segment, and arranged in the order of line topology connection as candidate communication support paths. For each communication interruption segment, the segment number on the candidate communication support path adjacent to the communication interruption segment is found in G, and the communication path from the adjacent segment number to the rail transit emergency dispatch system is taken as the emergency communication support path for the communication interruption segment, and summarized into emergency communication support path marker data.

[0149] Step S760: Based on the emergency communication support path marking data, perform communication support supplementary processing on the short-route turnaround scheduling instructions and train interval adjustment scheduling instructions in the emergency dispatch instruction sequence to obtain an emergency dispatch instruction sequence containing emergency communication support path marking data.

[0150] In this embodiment, each communication interruption segment number in the emergency communication support path marking data and the list of emergency communication support path segment numbers are appended to the instructions involving the communication interruption segment number in the short-route turnaround scheduling instructions and the train operation interval adjustment scheduling instructions.

[0151] Step S810: Perform boundary extension analysis on the spatial distribution marking data of obstructed sections in the description data of the obstructed operation status of the line, and extend the boundary section of each obstructed section group by one section in the direction of the adjacent unobstructed section to obtain the extended boundary marking data of the obstructed section group.

[0152] In this embodiment, for each blocked segment group, the unblocked segment directly adjacent to its boundary segment is found in G, the unblocked segment number is marked as an extended boundary segment and added to the segment number list of the blocked segment group, the extended boundary segment role is marked as a boundary segment, and the original boundary segment role is adjusted to an internal segment.

[0153] like Figure 4 As shown in the figure, the "extended boundary of the obstructed section group" is marked with a dashed box, and the "location of the outward turnaround section" is marked with a double line. When there is a "train occupancy conflict" in the outward turnaround section, the conflict location is replaced with the resolved outward turnaround section location through "turnaround section conflict resolution". The figure also shows the generation link from "terminal section turnaround supplement" to "terminal turnaround section" in the "terminal direction turnaround" module, as well as the process of generating available supplementary turnaround nodes using the turnaround nodes (diamond marks) within the obstructed section group, ultimately forming "small-route turnaround scheduling strategy data containing the main turnaround section and supplementary turnaround nodes".

[0154] Step S820: Perform outward matching processing on the outward matching of ...

[0155] In this embodiment, each turnaround segment number in S1 is traversed, and the role marker corresponding to the turnaround segment number is queried in the extended boundary marker data of the blocked segment group. If it is an internal segment, the extended boundary segment number of the blocked segment group where the turnaround segment number is located is used as the position marker of the outward turnaround segment; if it is still a boundary segment, the turnaround segment number is kept unchanged.

[0156] Step S830: Based on the location marking data of the outward turnaround section, perform outward turnaround section train occupancy conflict analysis processing on the train occupancy marking data of each operating section in the line topology to obtain outward turnaround section train occupancy conflict marking data.

[0157] In this embodiment, each outward turnaround section number in the outward turnaround section location marker data is traversed, and the train occupancy marker corresponding to the section number is queried in O(p). If the train occupancy marker is 1, it is determined that there is a train occupancy conflict in the outward turnaround section, and it is added to the conflict marker data list.

[0158] Step S840: Perform turnaround section conflict resolution processing on the outward turnaround section position marking data according to the outward turnaround section train occupancy conflict marking data to obtain the resolved outward turnaround section position marking data.

[0159] In this embodiment, for each segment number in the conflict marking data, the segment number is replaced in the outward expansion and reversal segment position marking data with the next adjacent boundary segment number in the obstructed segment group extension boundary marking data. If there is no adjacent next boundary segment number, the segment number is removed from the outward expansion and reversal segment position marking data.

[0160] Step S850: Perform a turnaround section replacement process on the resolved outward expansion turnaround section location marker data and the turnaround section location marker data in the short-route turnaround scheduling strategy data to obtain the replaced turnaround section location marker data.

[0161] In this embodiment, each turnaround segment number in S1 is traversed. If the replacement segment number corresponding to the turnaround segment number exists in the resolved outward expansion turnaround segment position marker data, it is replaced with the corresponding segment number; otherwise, the turnaround segment number is kept unchanged.

[0162] Step S860: Based on the replaced turnaround section location marker data, the turnaround direction marker data in the short-route turnaround scheduling strategy data is redirected to obtain the replaced turnaround direction marker data.

[0163] In this embodiment, for each segment number in the replaced turnaround segment position marker data, its positional relationship relative to the original turnaround segment number is queried in G. If the replaced segment number is located upstream of the original turnaround segment number, the turnaround running direction marker value is 0; if it is located downstream, the turnaround running direction marker value is 1.

[0164] Step S870: Merge the replaced turnaround section location marker data and the replaced turnaround running direction marker data into small-route turnaround scheduling strategy adjustment data, and add the small-route turnaround scheduling strategy adjustment data into the rail transit emergency plan scheduling strategy data.

[0165] In this embodiment, the replaced turnaround section location marker data and the replaced turnaround running direction marker data are merged into small-route turnaround scheduling strategy adjustment data according to the section number correspondence, and the small-route turnaround scheduling strategy adjustment data replaces the small-route turnaround scheduling strategy data in S as a whole.

[0166] Step S910: Perform connectivity analysis processing on the spatial distribution marker data of the obstructed sections in the description data of the obstructed operation status of the line, and extract the marker data of the section connectivity relationship between each operating section within each obstructed section group.

[0167] In this embodiment, for each blocked segment group, its segment number list is read, and G is used to query whether there is a direct adjacent relationship between any two segment numbers in the list. The segment number pairs with direct adjacent relationships are marked as connected segment pairs and summarized into segment connectivity label data.

[0168] Step S920: Based on the segment connectivity marker data, perform turnaround node identification processing on each operating segment within the obstructed segment group to obtain turnaround node marker data within the obstructed segment group.

[0169] In this embodiment, for each operating segment within each blocked segment group, the list of connected segment pairs corresponding to the operating segment number is queried in the segment connectivity labeling data. The number of times the operating segment number appears in the list of connected segment pairs is counted. If the number of occurrences is greater than or equal to a preset connectivity threshold, the operating segment number is marked as a reversible node.

[0170] Step S930: Perform a turnaround node supplement matching process on the turnaround node marker data within the blocked section group and the turnaround section location marker data in the short-route turnaround scheduling strategy data to obtain supplementary turnaround node marker data within the blocked section group.

[0171] In this embodiment, each reversible node segment number in the reversible node marking data within the blocked segment group is traversed. The reversible segment position marking data in S1 is checked to see if the segment number already exists. If it does not exist, the segment number is added to the supplementary reversible node marking data list.

[0172] Step S940: Based on the supplementary turnaround node marking data within the obstructed section group, perform turnaround node train occupancy status analysis and processing on the train occupancy marking data of each operating section within the obstructed section group to obtain supplementary turnaround node train occupancy status marking data.

[0173] In this embodiment, each segment number in the supplementary turnaround node marker data is traversed, and the train occupancy marker corresponding to the segment number is queried in O(p). The correspondence between the segment number and the train occupancy marker is added to the supplementary turnaround node train occupancy status marker data.

[0174] Step S950: Based on the supplementary turnaround node train occupancy status marking data, perform turnaround node availability filtering on the supplementary turnaround node marking data within the blocked section group to obtain available supplementary turnaround node marking data.

[0175] In this embodiment, each segment number in the supplementary turnaround node train occupancy status marker data is traversed. If the train occupancy marker corresponding to the segment number is 0, the segment number is added to the available supplementary turnaround node marker data list; if the train occupancy marker is 1, the segment number is removed.

[0176] Step S960: Perform hierarchical configuration processing of turnaround nodes on the small-route turnaround scheduling strategy data according to the available supplementary turnaround node marking data, to obtain small-route turnaround scheduling strategy data including main turnaround sections and supplementary turnaround nodes.

[0177] In this embodiment, the original turnaround segment numbers in S1 are marked as main turnaround segments, and each segment number in the available supplementary turnaround node marking data is marked as a supplementary turnaround node. The main turnaround segments and supplementary turnaround nodes are combined in the order of segment numbers to form a hierarchical turnaround segment list, which is then appended to S1.

[0178] Step S1010: Perform positional relationship analysis on the spatial distribution marking data of obstructed sections in the description data of obstructed line operation status to obtain the terminal section interval marking data between the obstructed section group and the terminal section of the line topology.

[0179] In this embodiment, for each blocked segment group, the number of segment intervals between its boundary segment and the line topology terminal segment is calculated. The number of segment intervals is defined as the total number of segments contained between the boundary segment number and the terminal segment number. The number of segment intervals is then appended to the tag data of the blocked segment group as a terminal segment interval marker.

[0180] Step S1020: Perform a terminal direction reversal feasibility analysis on the boundary section of the blocked section group based on the terminal section interval marking data to obtain terminal direction reversal feasibility marking data.

[0181] In this embodiment, for each blocked segment group, its terminal segment interval marker is read. If the terminal segment interval marker is less than the preset terminal direction turnaround interval threshold, the terminal direction turnaround feasibility marker is set to 1; if it is greater than or equal to the preset terminal direction turnaround interval threshold, the terminal direction turnaround feasibility marker is set to 0.

[0182] Step S1030: Based on the terminal direction turnaround feasibility marking data, perform terminal segment turnaround supplementation processing on the turnaround segment location marking data in the short-route turnaround scheduling strategy data to obtain terminal segment turnaround supplementation marking data.

[0183] In this embodiment, each turnaround segment number in S1 is traversed, and the terminal turnaround feasibility marker corresponding to the blocked segment group for that turnaround segment number is queried in the terminal turnaround feasibility marker data. If it is 1, the intermediate segment number list between the turnaround segment number and the terminal segment is queried in G, and the terminal segment turnaround supplementary marker data is added.

[0184] Step S1040: Based on the terminal section turnaround supplementary marking data, perform intermediate section train occupancy marking data analysis on the intermediate section train clearing sequence data between the blocked section group and the line topology terminal section to obtain intermediate section train clearing sequence marking data.

[0185] In this embodiment, for each intermediate segment number list in the terminal segment turnaround supplementary marking data, the train occupancy mark corresponding to each segment number is queried in O(p), and the segment numbers with train occupancy mark 1 are arranged from near to far from the terminal segment to form the intermediate segment train clearing order.

[0186] Step S1050: Based on the intermediate section train clearance sequence marking data, perform turnaround section hierarchical configuration processing on the turnaround section position marking data in the small-route turnaround scheduling strategy data to obtain small-route turnaround scheduling strategy data that includes terminal turnaround sections and intermediate turnaround sections.

[0187] In this embodiment, the original turnaround segment numbers in S1 are marked as intermediate turnaround segments, and the first segment number in the intermediate segment number list corresponding to each turnaround segment number in the terminal segment turnaround supplementary marking data is marked as the terminal turnaround segment. The terminal turnaround segments and intermediate turnaround segments are combined into a hierarchical turnaround segment list in the clearing order and appended to S1.

[0188] Step S1060: Based on the small-route turnaround scheduling strategy data including the terminal turnaround section and the intermediate turnaround section, perform train interval partitioning adjustment processing on the train interval adjustment scheduling strategy data of the adjacent unobstructed sections to obtain train interval adjustment scheduling strategy data including the terminal turnaround section and the intermediate turnaround section.

[0189] In this embodiment, each segment number in S2 is traversed, and the small-route turnaround scheduling strategy data containing terminal turnaround segments and intermediate turnaround segments is queried to see if the segment number belongs to a terminal turnaround segment or an intermediate turnaround segment. If it belongs to a terminal turnaround segment, the train interval level of the segment is adjusted to the first train interval level; if it belongs to an intermediate turnaround segment, it is adjusted to the second train interval level; if it does not belong to either segment, it remains unchanged.

[0190] Step S1110: Perform symmetry analysis on the spatial distribution marker data of obstructed sections in the description data of the obstructed operation status of the line, and extract the marker data of symmetrical section groups that are symmetrical with respect to the center point of the line topology.

[0191] In this embodiment, for each blocked section group, the symmetrical section number of each section number contained therein with respect to the center point of the line topology is calculated. The symmetrical section number is calculated by subtracting the current section number from the total number of operating sections in the line topology and then adding 1. All symmetrical section numbers are arranged in the order of section connection within the blocked section group to form symmetrical section group marking data.

[0192] like Figure 4 As shown, the "line topology center point" is marked with a vertical dashed line, and the "symmetrical section group" symmetrical about the center point is marked with a thick dashed box, indicating the train occupancy status within the symmetrical section group. The system obtains the "symmetrical turnaround section position" and "symmetrical turnaround running direction" through "symmetrical turnaround section matching", and merges them with the original turnaround strategy to generate "small-route turnaround scheduling strategy data containing symmetrical turnaround strategy".

[0193] Step S1120: Based on the symmetrical section group marking data, perform symmetrical section group train occupancy status analysis processing on the train occupancy marking data of each operating section in the line topology to obtain symmetrical section group train occupancy status marking data.

[0194] In this embodiment, the list of symmetrical segment numbers in the symmetrical segment group marking data is traversed, and the train occupancy mark corresponding to each symmetrical segment number is queried in O(p). The correspondence between the symmetrical segment number and the train occupancy mark is summarized into the symmetrical segment group train occupancy status marking data.

[0195] Step S1130: Perform symmetrical turnaround section matching processing on the turnaround section position marker data in the small-route turnaround scheduling strategy data according to the symmetrical section group train occupancy status marker data to obtain symmetrical turnaround section position marker data.

[0196] In this embodiment, each turnaround section number in S1 is traversed, and the symmetrical section number corresponding to the turnaround section number is queried in the symmetrical section group train occupancy status marker data. If the train occupancy marker corresponding to the symmetrical section number is 0, then the symmetrical section number is added to the symmetrical turnaround section position marker data list.

[0197] Step S1140: Based on the symmetrical turnaround section position marking data, perform symmetrical turnaround direction generation processing on the section connection direction marking data in the obstructed section association characteristics to obtain symmetrical turnaround running direction marking data.

[0198] In this embodiment, for each symmetrical turnaround segment number in the symmetrical turnaround segment position marking data, the segment connection direction mark corresponding to the symmetrical turnaround segment number is queried in C, and the reverse mark of the segment connection direction mark is used as the symmetrical turnaround running direction mark.

[0199] Step S1150: Combine the symmetrical turnaround section location marker data and the symmetrical turnaround running direction marker data with the turnaround section location marker data and the turnaround running direction marker data in the small-route turnaround scheduling strategy data to obtain small-route turnaround scheduling strategy data containing the symmetrical turnaround strategy.

[0200] In this embodiment, each symmetrical turnaround segment number in the symmetrical turnaround segment position marker data is combined with the corresponding symmetrical turnaround running direction marker in the symmetrical turnaround running direction marker data to form a symmetrical turnaround strategy tuple, and all symmetrical turnaround strategy tuples are appended to S1.

[0201] Step S1160: Based on the small-route turnaround scheduling strategy data containing the symmetrical turnaround strategy, perform symmetrical section train interval synchronization adjustment processing on the train interval adjustment scheduling strategy data to obtain train interval adjustment scheduling strategy data containing symmetrical section train interval synchronization adjustment marker data.

[0202] In this embodiment, each segment number in S2 is traversed, and the symmetrical turnaround segment number corresponding to the segment number is queried in the small-route turnaround scheduling strategy data containing the symmetrical turnaround strategy. If it exists, the train running interval level mark corresponding to the symmetrical turnaround segment number in S2 is adjusted to the train running interval level mark corresponding to the segment number.

[0203] Step S1210: Perform overlap analysis on the spatial distribution marking data of obstructed sections in the description data of the obstructed operation status of the line, and the location of transfer stations in the line topology, to obtain the transfer station marking data within the obstructed section group and the transfer station marking data outside the obstructed section group.

[0204] In this embodiment, each transfer station number in the transfer station location data is traversed. The corresponding operating segment number of the transfer station is queried in G, and the operating segment number is queried in R to see if the operating segment number belongs to any blocked segment group. If it belongs, it is added to the transfer station marking data list within the blocked segment group; if it does not belong, it is added to the transfer station marking data list outside the blocked segment group.

[0205] Step S1220: Based on the transfer station marking data within the obstructed section group, perform transfer direction blocking analysis processing on the transfer direction data of the transfer stations within the obstructed section group to obtain the transfer direction blocking marking data of the transfer stations within the obstructed section group.

[0206] In this embodiment, for each transfer station number in the transfer station marking data within the blocked section group, the target operating section number corresponding to each transfer direction in its transfer direction data is read, and the target operating section number is queried in R to see if the target operating section number belongs to the same blocked section group. If it does, the transfer direction is marked as a blocked transfer direction.

[0207] Step S1230: Based on the transfer direction blocking marker data of the transfer stations within the blocked section group, perform alternative transfer path search processing on the marker data of the transfer stations outside the blocked section group to obtain alternative transfer path marker data.

[0208] In this embodiment, for each transfer station number and its blocked transfer direction list in the transfer station blocking transfer direction marking data within the blocked section group, each transfer station number in the transfer station marking data outside the blocked section group is traversed, and the corresponding operating section number of the transfer station is queried in G. It is determined whether the operating section number is connected to the target operating section number corresponding to the blocked transfer direction through a non-blocked section. If they are connected, the transfer station number outside the blocked section group and the list of connecting path section numbers are used as alternative transfer path markings.

[0209] Step S1240: Based on the alternative transfer path marking data, perform transfer path redirection processing on the passenger flow guidance and scheduling strategy data of the transfer stations within the obstructed section group to obtain the redirected passenger flow guidance marking data of the transfer stations within the obstructed section group.

[0210] In this embodiment, for each blocked transfer station number in the alternative transfer path marking data, the passenger flow diversion path marking corresponding to the transfer station number is queried in S3, and the diversion path marking corresponding to the blocked transfer direction is replaced with the alternative transfer path segment number list.

[0211] Step S1250: Based on the redirection passenger flow guidance marker data of the transfer stations within the obstructed section group, perform transfer path diversion supplementation processing on the passenger flow guidance scheduling strategy data of the transfer stations outside the obstructed section group to obtain the diversion passenger flow guidance marker data of the transfer stations outside the obstructed section group.

[0212] In this embodiment, for each blocked transfer station number and its redirected diversion path mark in the passenger flow guidance mark data of the transfer station within the blocked section group, the number of the external transfer station involved in the blocked section group is searched in the redirected diversion path mark. In S3, the passenger flow diversion path mark corresponding to the external transfer station number of the blocked section group is queried, and the blocked transfer station number is attached as a diversion inlet mark to the passenger flow diversion path mark of the external transfer station of the blocked section group.

[0213] Step S1260: Merge the passenger flow redirection marker data of the transfer station within the obstructed section group and the passenger flow diversion marker data of the transfer station outside the obstructed section group into the transfer station linkage passenger flow guidance and scheduling strategy data, and add the transfer station linkage passenger flow guidance and scheduling strategy data into the rail transit emergency plan scheduling strategy data.

[0214] In this embodiment, the redirection passenger flow guidance marker data of transfer stations within the obstructed section group and the diversion passenger flow guidance marker data of transfer stations outside the obstructed section group are merged into transfer station linkage passenger flow guidance and scheduling strategy data according to the correspondence of transfer station numbers. The transfer station linkage passenger flow guidance and scheduling strategy data replaces the passenger flow guidance and scheduling strategy data in S as a whole.

[0215] In an exemplary embodiment, a deep learning-based rail transit emergency plan recommendation system is provided. This system can be a terminal, server, etc., and its internal structure includes a processor, memory, input / output interface, communication interface, display unit, and input device. The processor, memory, and input / output interface are connected via a system bus, and the communication interface, display unit, and input device are also connected to the system bus via the input / output interface. The processor provides computational and control capabilities. The memory includes a non-volatile storage medium and internal memory. The non-volatile storage medium stores an operating system and computer programs. The internal memory provides an environment for the operation of the operating system and computer programs in the non-volatile storage medium. The input / output interface is used for exchanging information between the processor and external devices. The communication interface is used for wired or wireless communication with external terminals; wireless communication can be achieved through Wi-Fi, mobile cellular networks, near-field communication, or other technologies. When the computer program is executed by the processor, it implements a deep learning-based rail transit emergency plan recommendation method. The display unit is used to form a visually visible image and can be a display screen, projection device, or virtual reality imaging device. The display screen can be an LCD screen or an e-ink screen. The input device can be a touch layer covering the display screen, or a button, trackball, or touchpad set on the shell of the deep learning-based rail transit emergency plan recommendation system, or an external keyboard, touchpad, or mouse, etc.

[0216] It should be noted that, in order to simplify the description of the present invention and thus help to understand one or more embodiments of the invention, multiple features may sometimes be grouped into one embodiment, drawing or description thereof in the foregoing description of the embodiments of the present invention.

Claims

1. A method for recommending emergency response plans for rail transit based on deep learning, characterized in that, The method includes: The system acquires a set of line operation status data for a rail transit line after a sudden operational event occurs. The set of line operation status data includes real-time train operation location distribution data for each section of the line topology, signal equipment status data for each section, and event trigger location data of the sudden operational event in the line topology. Extract the segment occupancy status characteristics of each operating segment in the line topology and the transmission path obstruction characteristics of the signal equipment transmission path in the line topology from the line operation status data set, and construct line operation obstruction status description data based on the segment occupancy status characteristics of each operating segment in the line topology. The obstruction status description data of the line operation and the obstruction characteristics of the transmission path of the signal equipment are input into a pre-constructed emergency dispatch strategy generation network. The obstruction section correlation relationship modeling processing is performed on the obstruction status description data of the line operation to obtain the obstruction section correlation relationship characteristics. The obstruction section correlation relationship characteristics and the transmission path obstruction characteristics are then combined to generate a joint dispatch strategy to obtain the rail transit emergency plan dispatch strategy data. The emergency dispatch strategy data of the rail transit emergency plan is parsed to obtain an emergency dispatch instruction sequence for the rail transit emergency dispatch system.

2. The method for recommending rail transit emergency plans based on deep learning according to claim 1, characterized in that, The step of extracting the segment occupancy status characteristics of each operating segment in the line topology and the transmission path obstruction characteristics of the signal equipment transmission paths in the line topology from the line operation status data set, and constructing line operation obstruction situation description data based on the segment occupancy status characteristics of each operating segment in the line topology, includes: Train occupancy marker data and train direction marker data for each operating section in the line topology are extracted from the real-time train position distribution data of each section in the line operation status data set. Arrange the train occupancy marker data and train running direction marker data of each operating section in the line topology according to the actual connection order of each operating section in the line topology to obtain the section occupancy status sequence of the line topology. Extract the trigger segment marker data and adjacent segment marker data of the sudden operation event in the line topology from the event trigger location data of the sudden operation event in the line topology from the event trigger location data of the sudden operation event in the line operation status data set; Based on the triggering section marking data and the adjacent section marking data of the triggering section, the starting position of the blocked section and the direction of the blocked section are marked in the section occupancy status sequence of the line topology to obtain the blocked section marking sequence; Extract the set of segments through which the signal equipment transmission path passes and the transmission status marker data of the signal equipment transmission path in each segment from the signal equipment status data of each segment in the line operation status data set. The transmission status marker data of the signal equipment transmission path in each segment is associated with the blocked segment marker sequence. The interrupted segment marker data of the signal equipment transmission path within the blocked segment and the available segment marker data of the signal equipment transmission path outside the blocked segment are extracted as the transmission path blocked feature of the signal equipment transmission path in the line topology. Based on the spatial continuous distribution pattern of the blocked sections in the blocked section marking sequence, adjacent blocked sections are merged into blocked section groups, and the boundary sections and internal sections of the blocked section groups in the line topology are distinguished and marked to obtain the spatial distribution marking data of blocked sections. The spatial distribution marking data of the obstructed sections is fused with the train occupancy marking data of each operating section in the line topology to obtain line operation obstruction status description data that includes the location of the obstructed sections, the boundaries of the obstructed sections, and the train occupancy status within the obstructed sections.

3. The method for recommending rail transit emergency plans based on deep learning according to claim 1, characterized in that, The process involves inputting the description data of the obstructed line operation status and the obstructed transmission path characteristics of the signal equipment transmission path into a pre-constructed emergency dispatch strategy generation network. This network performs obstructed section correlation modeling on the obstructed line operation obstruction status description data to obtain obstructed section correlation characteristics. Finally, it performs joint dispatch strategy generation processing on the obstructed section correlation characteristics and the obstructed transmission path characteristics to obtain rail transit emergency plan dispatch strategy data, including: The obstructed section spatial distribution marker data in the obstructed line operation status description data is input into the obstructed section association coding layer in the emergency dispatch strategy generation network. The obstructed section association coding layer performs adjacent section traversal processing on each obstructed section group in the obstructed section spatial distribution marker data to obtain the adjacent unobstructed section marker data corresponding to each obstructed section group. The blocked segment association coding layer extracts the segment connection boundary marker data and segment connection direction marker data between each blocked segment group and its adjacent unblocked segment based on the adjacent unblocked segment marker data corresponding to each blocked segment group. The blocked segment association relationship encoding layer combines the segment connection order within each blocked segment group, the segment connection boundary marker data between each blocked segment group and adjacent unblocked segments, and the segment connection direction marker data to obtain the blocked segment association relationship feature. The obstructed section association characteristics and the signal equipment transmission path obstruction characteristics are input into the joint scheduling strategy generation layer in the emergency scheduling strategy generation network. The joint scheduling strategy generation layer performs turnaround section availability analysis on the section connection boundary marking data in the obstructed section association characteristics to obtain turnaround section marking data that can perform small-loop turnaround. The joint scheduling strategy generation layer generates small-route turnaround scheduling strategy data containing the turnaround section location and turnaround direction based on the turnaround section marking data and the section connection direction marking data in the characteristics of the blocked section association. The joint scheduling strategy generation layer performs train interval division processing on the available section marking data of the signal equipment transmission path outside the blocked section in the transmission path blocked feature of the signal equipment transmission path, and obtains the train interval adjustment scheduling strategy data of the adjacent unblocked section. The joint scheduling strategy generation layer performs passenger flow diversion path planning processing on the affected transfer stations based on the adjacent unobstructed section marking data in the obstructed section association relationship characteristics and the train occupancy status within the obstructed section in the line operation obstruction situation description data, thereby obtaining passenger flow guidance scheduling strategy data for the affected transfer stations. The short-route turnaround scheduling strategy data, the train interval adjustment scheduling strategy data, and the passenger flow guidance scheduling strategy data are combined to obtain the rail transit emergency plan scheduling strategy data.

4. The method for recommending rail transit emergency plans based on deep learning according to claim 1, characterized in that, The process of parsing the emergency response plan dispatch strategy data for rail transit to obtain an emergency dispatch instruction sequence for the rail transit emergency dispatch system includes: Extract the turnaround section location marker data and turnaround running direction marker data from the small-route turnaround scheduling strategy data in the rail transit emergency plan scheduling strategy data, convert the turnaround section location marker data into turnaround section number data in the line topology, and convert the turnaround running direction marker data into turnaround direction instruction code data; Extract the section number data and the train interval adjustment scheduling strategy data of adjacent unobstructed sections from the train interval adjustment scheduling strategy data of the rail transit emergency plan scheduling strategy data, and convert the section number data and the train interval level marking data of each section into train interval adjustment instruction code data for the train operation control system. Extract the transfer station number data and passenger flow diversion path marking data of the affected transfer stations from the passenger flow guidance and scheduling strategy data in the rail transit emergency plan scheduling strategy data, and convert the transfer station number data and the passenger flow diversion path marking data into passenger flow guidance instruction code data for the station passenger information release system; The turnaround section number data, the turnaround direction instruction code data, the train interval adjustment instruction code data, the transfer station number data, and the passenger flow guidance instruction code data are sorted according to the instruction issuance priority to obtain the emergency dispatch instruction code sequence; Based on the emergency dispatch instruction encoding sequence, generate short-route turnaround dispatch instructions, train interval adjustment dispatch instructions, and passenger flow guidance dispatch instructions for the rail transit emergency dispatch system, and combine the short-route turnaround dispatch instructions, the train interval adjustment dispatch instructions, and the passenger flow guidance dispatch instructions into an emergency dispatch instruction sequence.

5. The method for recommending rail transit emergency plans based on deep learning according to claim 1, characterized in that, The method further includes: The real-time train position change data set of the rail transit line after the occurrence of a sudden operational event is obtained. The real-time train position change data set includes the train position update data of each section in the line topology after the emergency dispatch instruction sequence is issued. The difference comparison processing is performed between the real-time train position change data set and the real-time train position distribution data of each section in the line operation status data set to obtain the difference marker data of train position change in each section. Based on the train position change difference marker data of each section, extract the train position change difference marker data of the sections that have performed short-route turnaround and the train position change difference marker data of the sections that have performed train interval adjustment; The difference marking data of train position changes in the sections where small-route turnarounds have been performed and the difference marking data of train position changes in the sections where train interval adjustments have been performed are input into the emergency dispatch strategy generation network. The difference marking data of train position changes in the sections where small-route turnarounds have been performed are processed by the trend analysis of the dissipation of obstructed sections to obtain the dissipation status description data of obstructed sections. Based on the description data of the dissipation state of the obstructed section, the obstructed section association relationship feature is updated to obtain the updated obstructed section association relationship feature. The updated obstructed section association features and the transmission path obstruction features are input into the joint scheduling strategy generation layer in the emergency scheduling strategy generation network, and the joint scheduling strategy generation layer generates updated rail transit emergency plan scheduling strategy data. The updated rail transit emergency plan dispatch strategy data is parsed to obtain an updated emergency dispatch instruction sequence. The updated emergency dispatch instruction sequence is used to adjust the short-route turnaround dispatch instruction and the train interval adjustment dispatch instruction of the rail transit emergency dispatch system.

6. The method for recommending rail transit emergency plans based on deep learning according to claim 1, characterized in that, The method further includes: Obtain the track circuit status data set of each operating section in the line topology. The track circuit status data set includes the occupancy indication data of the track circuit in each section, the fault indication data of the track circuit in each section, and the transmission interruption indication data of the track circuit in each section. The track circuit status data set and the signal equipment status data of each section in the line operation status data set are subjected to track circuit signal correlation analysis to obtain the correlation marking data of the transmission path of track circuit and signal equipment. Based on the association marker data of the track circuit and signal equipment transmission path, track circuit fault indication data associated with the signal equipment transmission path in the obstructed section are extracted from the track circuit status data set; The track circuit fault representation data associated with the signal equipment transmission path of the obstructed section is input into the emergency dispatch strategy generation network. The fault impact range analysis and processing of the track circuit fault representation data of each section in the track circuit fault representation data is performed to obtain track circuit fault impact section marking data. Based on the track circuit fault-affected section marking data and the obstructed section spatial distribution marking data in the line operation obstruction status description data, an obstruction range consistency analysis is performed to obtain obstruction range consistency analysis result data. Based on the consistency analysis results of the obstructed range, the spatial distribution marking data of the obstructed section in the description data of the obstructed operation status of the line is supplemented to obtain supplemented obstructed section marking data. The supplementary obstructed section marker data is input into the emergency dispatch strategy generation network, and the obstructed section association relationship supplementation modeling processing is performed on the supplementary obstructed section marker data to obtain the supplementary obstructed section association relationship features. Based on the characteristics of the associated relationships of the additional obstructed sections, the emergency response plan scheduling strategy data for rail transit is supplemented and generated to obtain supplementary rail transit emergency response plan scheduling strategy data.

7. The method for recommending rail transit emergency plans based on deep learning according to claim 1, characterized in that, The method further includes: Acquire event type labeling data for sudden operational events in the line topology and train timetable data for each operating section in the line topology; Based on the event type labeling data, extract the event impact segment range labeling data corresponding to the event type labeling data from the preset event type and obstructed segment distribution relationship data; The event-affected area range marking data and the train timetable data of each operating section in the line topology are subjected to timetable section coverage analysis to obtain the planned departure sequence marking data and planned arrival sequence marking data of each section during the event-affected period; Based on the planned departure sequence marker data and planned arrival sequence marker data of each section during the event's impact period, the on-the-way train marker data located within the event's impact section and the on-the-way train marker data located outside the event's impact section are extracted from the real-time train operation location distribution data of each section in the line operation status data set. The on-the-way train marking data located within the event-affected section and the on-the-way train marking data located outside the event-affected section are input into the emergency dispatch strategy generation network. The on-the-way train marking data located within the event-affected section are processed to divide the evacuation direction, resulting in evacuation train marking data for the short-route turnaround direction and evacuation train marking data for the continuing operation direction. Based on the train marking data for the short-route turnaround direction and the train marking data for the continuing direction, combined with the on-the-way train marking data located outside the affected area of ​​the event, train evacuation sequence data is generated. Based on the train evacuation sequence data, the turnaround scheduling strategy data in the rail transit emergency plan scheduling strategy data is supplemented with turnaround execution order data to obtain small-route turnaround scheduling strategy data containing turnaround execution order marker data.

8. The method for recommending rail transit emergency plans based on deep learning according to claim 1, characterized in that, The method further includes: Obtain a set of real-time passenger flow distribution data for each transfer station in the line topology. The set of real-time passenger flow distribution data includes platform passenger flow density marker data, transfer passage passenger flow density marker data, and entrance / exit passenger flow density marker data for each transfer station. The real-time passenger flow distribution data set and the spatial distribution marker data of the obstructed sections in the description data of the obstructed operation status of the line are subjected to obstruction correlation analysis of transfer station to obtain the marker data of transfer stations affected by the obstructed sections and the marker data of transfer stations not affected by the obstructed sections. The set of transfer station marker data affected by the obstructed section and the real-time passenger flow distribution data are input into the emergency dispatch strategy generation network. The accessibility analysis of the transfer direction of each affected transfer station in the transfer station marker data affected by the obstructed section is performed to obtain the transfer direction accessibility marker data of each affected transfer station. Based on the accessibility marker data of the transfer direction of each affected transfer station and the platform passenger flow density marker data in the real-time passenger flow distribution data set, the internal diversion path generation process is performed on the affected transfer stations to obtain the internal diversion path marker data of the affected transfer stations. Based on the in-station diversion path marking data of the affected transfer stations, the passenger flow guidance and scheduling strategy data in the rail transit emergency plan scheduling strategy data are compared and processed to obtain supplementary marking data for passenger flow guidance and scheduling strategies. The passenger flow guidance and scheduling strategy data of the affected transfer stations is supplemented with supplementary marker data based on the passenger flow guidance and scheduling strategy, resulting in passenger flow guidance and scheduling strategy data that includes in-station diversion path marker data.

9. The method for recommending rail transit emergency plans based on deep learning according to claim 1, characterized in that, The method further includes: Obtain the turnout status data set of each operating section in the line topology. The turnout status data set includes the current open direction marking data, turnout locking status marking data, and turnout control mode marking data of each section's turnouts. The turnout status data set is correlated with the spatial distribution marking data of the obstructed section in the line operation obstruction situation description data by performing turnout position correlation analysis to obtain turnout marking data located inside the obstructed section and turnout marking data located at the boundary of the obstructed section. The turnout marking data located inside the obstructed section and the turnout marking data located at the boundary of the obstructed section are input into the emergency dispatch strategy generation network. Based on the turnout locking status marking data in the turnout marking data located inside the obstructed section, the locking direction maintenance processing of the turnouts inside the obstructed section is performed to obtain the turnout control command data inside the obstructed section. Based on the turnout current opening direction marking data in the turnout marking data located at the boundary of the obstructed section and the turnout section position marking data in the small-route turnout scheduling strategy data in the rail transit emergency plan scheduling strategy data, the turnout direction opening processing is performed on the turnout located at the boundary of the obstructed section to obtain the obstructed section boundary turnout control command data. The turnout control command data inside the obstructed section and the turnout control command data at the boundary of the obstructed section are combined according to the turnout linkage control sequence to obtain the turnout linkage control command sequence. Based on the turnout linkage control command sequence, the short-route turnaround scheduling command in the emergency dispatch command sequence is supplemented with turnout linkage to obtain a short-route turnaround scheduling command that includes turnout linkage control command.

10. A deep learning-based emergency response plan recommendation system for rail transit, characterized in that, include: processor; A machine-readable storage medium for storing machine-executable instructions of the processor; The processor is configured to execute the deep learning-based rail transit emergency response plan recommendation method according to any one of claims 1 to 9 by executing the machine-executable instructions.

Images