A single-line bus intelligent scheduling method based on double-layer constraint fusion

Abstract

The application relates to the technical field of bus management, in particular to a single-line bus intelligent scheduling method based on double-layer constraint fusion, which comprises the following steps: initializing scheduling parameters; simulating the operation process of each bus, constructing a departure event chain, and verifying the integrity of the departure event chain; taking the complete departure event chain of one bus as a row and the departure event of the same time at the platform as a column to construct a double-layer constraint matrix; performing constraint checking and adjustment on each departure event in the double-layer constraint matrix based on constraint priority; compressing or smoothing the abnormal time section; comprehensively checking the scheduling scheme after compression or smoothing, and generating a final scheduling scheme if all constraint conditions and transport capacity requirements are met. The application simultaneously considers the constraints of the two aspects of vehicle operation and platform departure, more reasonably schedules the buses, and realizes the collaborative optimization of vehicle operation and platform departure.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of bus management, in particular to a single-line bus intelligent scheduling method based on double-layer constraint fusion. BACKGROUND

[0002] At present, bus scheduling mainly adopts manual experience scheduling or single constraint automatic scheduling method. Manual experience scheduling has problems such as low efficiency, difficulty in optimization, and inability to adapt to complex scenarios. Although the single constraint automatic scheduling method can improve efficiency, it often only considers the single constraint of the vehicle layer or the platform layer, resulting in many problems in the actual execution of the scheduling result:

[0003] 1. Vehicle layer: unable to reasonably arrange vehicle turnover, resulting in too long or too short vehicle stay time at the site, affecting operational efficiency;

[0004] 2. Platform layer: uneven departure interval, insufficient departure density during peak hours, and excessive vehicles during off-peak hours, causing resource waste;

[0005] 3. Constraint conflict: the constraints of the vehicle layer and the platform layer often conflict, such as sacrificing vehicle reasonable turnover time to meet the departure interval, or causing uneven departure interval to ensure vehicle turnover.

[0006] Therefore, how to consider the constraints of both the vehicle operation and the platform departure, and more reasonably schedule the buses is a problem that needs to be solved by those skilled in the art. SUMMARY

[0007] Therefore, the present application provides a single-line bus intelligent scheduling method based on double-layer constraint fusion, which considers the constraints of both the vehicle operation and the platform departure, and more reasonably schedules the buses, achieving collaborative optimization of vehicle operation and platform departure.

[0008] In order to achieve the above purpose, the present application adopts the following technical solutions:

[0009] A single-line bus intelligent scheduling method based on double-layer constraint fusion, comprising the following steps:

[0010] S1, initializing the scheduling parameters of the buses under a certain line;

[0011] S2, assigning a first departure time to all vehicles under the line according to the first departure time and the departure interval requirement, starting from the first departure time, simulating the running process of each vehicle, constructing a scheduling event chain, and verifying the integrity of the scheduling event chain;

[0012] S3, taking a complete scheduling event chain of a vehicle as a row, and taking the scheduling events at the same time on the platform as a column, to construct a double-layer constraint matrix;

[0013] S4, dividing the whole day operation time into different time periods, extracting time period features of each time period; determining constraint priorities of each time period based on the time period features of different time periods, and performing constraint checking and adjustment on each dispatching event in the double-layer constraint matrix based on the constraint priorities;

[0014] S5, based on the time period features and operation requirements, identifying an abnormal time period, and performing compression or smoothing processing on the abnormal time period;

[0015] S6, performing comprehensive checking on the scheduling scheme after compression or smoothing processing, and if all constraint conditions and operation requirements are met, generating a final scheduling scheme.

[0016] Further, S1 includes:

[0017] S11, determining basic operation parameters according to line features, including first and last bus times and peak time period division;

[0018] S12, formulating differentiated headway requirements for different time periods, including maximum and minimum headways during peak period, flat peak period and first and last bus period;

[0019] S13, setting maximum and minimum working time lengths of a single vehicle, and minimum and maximum stop time length requirements in different time periods;

[0020] S14, dividing the whole day operation time into multiple time periods at fixed intervals, counting the running time of the vehicle at each time period, and establishing a time period-running time mapping relationship;

[0021] S15, based on the running time analysis result, calculating the single vehicle round trip cycle, and combining with the working time length limit to determine the theoretical dispatching frequency of each vehicle.

[0022] Further, S2 includes:

[0023] S21, according to the first bus time and headway requirement, assigning a first dispatching time to each vehicle to generate a first dispatching time sequence;

[0024] S22, constructing a SARIMAX model based on historical data and time period features, and predicting the baseline running time of different time periods based on the SARIMAX model;

[0025] S23, starting from the first dispatching time, simulating the running process of each vehicle under the constraint of the baseline running time of different time periods, including dispatching, arrival and turnover links;

[0026] S24, generating corresponding dispatching events for each link in the simulation process, including dispatching time, arrival time, running time and stop time, to obtain the dispatching event chain of each vehicle.

[0027] S25, integrity verification is performed on the departure event chain of each vehicle, and verification targets include that the departure time meets the first and last time range requirements, the vehicle completes the required number of departure trips, and it is ensured that the vehicle is received at the same site.

[0028] Further, S22 includes:

[0029] Anomaly detection and cleaning are performed on historical running time data, unreasonable data points are removed, and a SARIMAX model is constructed based on the processed historical running time;

[0030] The whole day operation time is divided into multiple time periods at fixed intervals, and for each time period, the historical average running time at the departure time of the time period is calculated;

[0031] For each planned departure event, the time period in which the departure event is located is located, the feature vector corresponding to the time period is extracted, including period features, weather features and passenger flow features; the historical average running time of each time period is corrected based on the SARIMAX model combined with the feature vector of each time period as the baseline running time of the planned departure event.

[0032] Further, S3 includes:

[0033] S31, initialize the matrix structure: create the basic matrix structure of the vehicle layer and the platform layer, take the vehicle layer as the row, which is used to represent the complete departure event chain of a vehicle, take the platform layer as the column, which is used to represent the departure event at the same time on the platform, and determine the matrix dimension and index relationship;

[0034] S32, map the departure event chain to the matrix, and the mapping process includes: a) fill in the row data according to the vehicle number and the departure order, establish the correspondence between the vehicle and the departure event; b) fill in the column data according to the time period and the departure interval requirement, ensure the continuity of the platform departure event;

[0035] Each element in the matrix is a departure event object, and each departure event object includes the following attributes: departure time, arrival time, running time, departure direction, departure site number, arrival site number and forced stop time;

[0036] S33, set constraints for the matrix, including stop time, departure interval and capacity demand constraints;

[0037] S34, construct a four-way linked list relationship between matrix elements, each departure event object establishes a bidirectional association with adjacent elements: a) up, indicating a link to the previous departure event in the same column, used to maintain the departure order of the platform; b) down, indicating a link to the next departure event in the same column, used to maintain the departure order of the platform; c) left, indicating a link to the previous departure event in the same row, used to track the vehicle running track; d) right, indicating a link to the next departure event in the same row, used to track the vehicle running track.

[0038] Further, S4 includes:

[0039] S41, divide the whole day operation time into multiple time periods, including the first bus, the morning peak, the flat peak, the evening peak and the last bus period;

[0040] S42, analyze the passenger flow characteristics and the demand characteristics of the operation capacity of each period, and determine the constraint priority of each period;

[0041] S43, traverse each departure event in the double-layer constraint matrix, and perform the following checks for each departure event:

[0042] Period constraint check: check whether the departure time meets the operation time requirements of the period; verify whether it meets the passenger flow and full load rate requirements of the period; confirm whether it meets the fixed-point departure rule;

[0043] Platform constraint check: verify whether the interval with the previous and next departure event meets the requirements; check whether it meets the minimum / maximum stop time length rule of the platform; for peak hours, check whether the departure density meets the requirements;

[0044] Vehicle constraint check: confirm whether the vehicle working time length is within the limit; verify whether the vehicle round trip turnaround time is reasonable; check whether it meets the vehicle rest and departure requirements;

[0045] S44, when a certain departure event violates a certain constraint condition, adjust the departure time according to the violated constraint type, and get the four-way associated elements of the current departure event, and process them in the order of up, down, left and right, and recheck the constraints for each associated element; if the adjustment leads to new constraint violation, recursively process the related elements.

[0046] Further, in S42, the passenger flow characteristics include: the maximum passenger flow of each period, the ratio of passenger flow to vehicle capacity, and the passenger flow change rate between adjacent periods;

[0047] The demand characteristics of the operation capacity include: the baseline running time, the baseline departure interval, the peak interval and the flat peak interval predicted by the SARIMAX model;

[0048] In peak hours, the minimum headway and full load rate requirements are prioritized; in flat peak hours, the balance of transport capacity and vehicle turnover is prioritized; in the first and last shift hours and rest shift hours, the operation rule requirements are prioritized.

[0049] Further, S5 includes:

[0050] S51, based on the headway deviation judgment, the section continuity judgment, the transport capacity distribution judgment and the constraint satisfaction degree judgment index, identify the abnormal time section that needs to be compressed or smoothed;

[0051] S52, evaluate the compressible range of each abnormal time section, considering vehicle operation constraints, station stop constraints and headway constraints, calculate the minimum compression unit, while not affecting subsequent and previous departure events as a prerequisite, determine the compression boundary;

[0052] S53, adopt a gradual compression strategy to compress the abnormal time section;

[0053] S54, determine the smoothing parameter according to the time period characteristics, and perform smoothing processing on the abnormal section according to the smoothing parameter; wherein the smoothing parameter includes a reference parameter and an adjustment parameter, the reference parameter includes: first bus time period reference interval, peak time period reference interval, flat peak time period reference interval and last bus time period reference interval; the adjustment parameter includes: adjustment time step, maximum iteration number and interval coefficient range.

[0054] Further, S53 includes:

[0055] S531, determine the start time point and end time point of the compression interval, which remain unchanged during the compression process; check whether the compression interval length meets the minimum compression requirement, and record the departure event information corresponding to the compression interval endpoints as the boundary constraint of the compression process;

[0056] S532, determine the target headway, and select a forward compression mode or a backward compression mode;

[0057] When the forward compression mode is selected, starting from the compression interval start time point, the new departure time is calculated in turn according to the target headway; the new time of each departure event is equal to the departure time of the previous departure event plus the target interval; the departure time of the last departure event must not exceed the termination time point of the compression interval;

[0058] When the backward compression mode is selected, starting from the compression interval termination point, the new departure time is calculated in reverse order according to the target headway; the new time of each departure event is equal to the departure time of the next event minus the target interval; the departure time of the first departure event must not be earlier than the start time point of the compression interval;

[0059] S533, performing a compression process:

[0060] Arranging the departure events in the compression interval in chronological order to establish a compression processing queue;

[0061] According to the selected compression mode, the departure time is gradually adjusted, and during the adjustment process, the single adjustment amount does not exceed the minimum compression unit, and after each adjustment, the constraint satisfaction condition is checked. If there is a constraint violation, it will be rolled back to the last feasible state;

[0062] For each adjusted departure event, its four-way associated events in the double-layer constraint matrix are obtained, and it is checked whether the adjustment affects the constraints of the associated events. If it does, the associated events are adjusted synchronously;

[0063] S534, compression result verification: verifying whether the standard deviation of the departure interval in the compressed interval meets the requirements; confirming whether all adjusted departure events meet the vehicle and platform constraints; verifying whether the connection between the compressed interval and the adjacent interval is smooth;

[0064] S535, abnormal situation processing:

[0065] When the compression interval length is insufficient to accommodate all departure events, the compression interval range is expanded, and the target departure interval is adjusted, or it is split into multiple sub-intervals for separate processing;

[0066] When there is a constraint conflict that cannot be met, the high-priority constraint is prioritized, the low-priority constraint is relaxed, and the unhandled constraint conflict is recorded for subsequent optimization.

[0067] Further, in S51, the abnormal time section determination method is:

[0068] Departure interval deviation determination: calculate the difference between the departure interval in the section and the target interval, and determine it as abnormal when the difference exceeds the threshold value, wherein the peak period threshold value is 5 minutes and the flat peak period threshold value is 15 minutes;

[0069] Section continuity determination: check whether the transition of adjacent sections is smooth;

[0070] Distribution of transport capacity determination: evaluate whether the distribution of transport capacity in the section is balanced;

[0071] Constraint satisfaction degree determination: count the number of constraint violations in the section, including the number of violations of the departure interval constraint and the stop time constraint.

[0072] According to the above technical solution, compared with the prior art, the present application has the following beneficial effects:

[0073] 1. The present application considers both vehicle operation and platform departure constraints, realizes a globally optimal scheduling scheme, and makes the final scheduling scheme more reasonable.

[0074] 2、The application automatically processes constraint conflicts by constraint fusion algorithm, reduces manual intervention, and realizes intelligent scheduling of each vehicle scheduling.

[0075] 3、The application divides the whole day operation time into multiple time periods, including peak period, flat peak period, first and last shift, etc., considers specific characteristics of different time periods, and finely controls peak period, special time period, etc. scene, improves the practicability of the scheduling scheme.

[0076] 4、The application improves the utilization efficiency of vehicle and human resources by identifying abnormal time sections and compressing and smoothing the abnormal sections.

[0077] 5、The application can flexibly cope with scheduling needs of different lines and different time periods, and has strong universality. BRIEF DESCRIPTION OF DRAWINGS

[0078] In order to more clearly illustrate the technical solutions in the embodiments of the present application or the prior art, the following will briefly introduce the drawings needed to be used in the embodiment or prior art description. Obviously, the drawings in the following description are only a part of the embodiments of the present application, and other drawings can be obtained by those skilled in the art without creative labor on the basis of the provided drawings.

[0079] Figure 1 The overall framework flow chart of the single line bus intelligent scheduling method based on double-layer constraint fusion provided by the present application is shown in the figure.

[0080] Figure 2 The detailed flow chart of the single line bus intelligent scheduling method based on double-layer constraint fusion provided by the present application is shown in the figure. DETAILED DESCRIPTION

[0081] The technical solutions in the embodiments of the present application will be described clearly and completely in the following with reference to the drawings in the embodiments of the present application. Obviously, the described embodiments are only a part of the embodiments of the present application, not all. Based on the embodiments in the present application, all other embodiments obtained by those skilled in the art without creative labor are within the scope of protection of the present application.

[0082] As shown in the figure, the embodiment of the present application discloses a single line bus intelligent scheduling method based on double-layer constraint fusion, comprising the following steps: Figure 1 S1, initialize the scheduling parameters of the bus under a certain line;

[0083]

[0084] ​S2, according to the first time and the departure interval requirements, assigning the first departure time for all vehicles under the line, starting from the first departure time, simulating the running process of each vehicle, constructing the departure event chain, and verifying the integrity of the departure event chain;

[0085] S3, taking the complete departure event chain of a vehicle as a row, and the departure event at the same time on the platform as a column, constructing a double-layer constraint matrix;

[0086] S4, dividing the whole day operation time into different time periods, extracting the time period characteristics of each time period, determining the constraint priority of each time period based on the time period characteristics of different time periods, and performing constraint checking and adjustment on each departure event in the double-layer constraint matrix based on the constraint priority;

[0087] S5, based on the time period characteristics and operation requirements, identifying the abnormal time period, and performing compression or smoothing processing on the abnormal time period;

[0088] S6, performing comprehensive checking on the scheduling scheme after compression or smoothing processing, and if all constraint conditions and operation requirements are met, generating the final scheduling scheme.

[0089] Next, combined with Figure 2 , the above steps are further described in detail.

[0090] S1, initializing the scheduling parameters of the buses under a certain line, specifically including:

[0091] S11, determining the basic operation parameters according to the line characteristics, including the first and last time and the division of peak period.

[0092] S12, setting the departure interval requirements: formulating differentiated departure interval requirements for different time periods, including the maximum and minimum departure interval during peak period, flat peak period and first and last period.

[0093] S13, defining vehicle working restrictions: setting the maximum and minimum working time of a single vehicle, and the minimum and maximum stop time requirements in different time periods.

[0094] S14, analyzing the running time: dividing the whole day operation time into multiple time periods according to fixed intervals, counting the running time of the vehicle at each time period, and establishing the time period-running time mapping relationship.

[0095] S15, calculating the theoretical departure times: based on the running time analysis results, calculating the single vehicle round trip cycle, and combining with the working time limit, determining the theoretical departure times of each vehicle.

[0096] In this step, the basic operation parameters are determined according to the actual operation characteristics of the line, including the determination of basic parameters such as the setting of the first and last bus time, the planning of the operation time, and the division of the peak period. In terms of setting the first and last bus time, according to historical operation data and actual demand, the optimal first bus departure time and last bus arrival time are determined, usually the first bus time is set between 5:30 and 6:30 in the morning, and the last bus time is set between 22:00 and 23:00 in the evening according to the passenger flow.

[0097] Differentiated departure interval requirements are formulated for different periods, especially during the morning peak (7:00-9:00) and evening peak (17:00-19:00), the departure interval is usually controlled at 3-5 minutes, during the flat peak, it can be appropriately relaxed to 8-15 minutes, and during the first and last bus period, it can be extended to 15-20 minutes.

[0098] In addition, it is also necessary to clearly stipulate the working restrictions of the vehicle, including the maximum working time of each vehicle not exceeding 8 hours, the minimum working time not less than 6 hours, and the stop time requirements in different periods, among which the minimum stop time in the peak period is not less than 3 minutes, and the minimum stop time in the flat peak period is not less than 5 minutes.

[0099] Through statistical analysis of the running time of each period, an accurate period-running time mapping relationship is established, recording the average running time, standard deviation and other statistical indicators of each time period (such as 15 minutes as a time slice), providing basic data support for subsequent turnover calculation.

[0100] Finally, based on the analysis results of the running time, combined with the single vehicle round trip cycle (usually including the uplink time, downlink time and stop time at both ends of the station) and the working time limit, the system will accurately calculate the theoretical departure times of each vehicle, ensuring the reasonable utilization of vehicles.

[0101] S2, constructing a departure event chain, specifically including:

[0102] S21, according to the first bus time and the departure interval requirement, assigning the first departure time to each vehicle, and generating the first departure time sequence.

[0103] S22, constructing a SARIMAX model based on historical data and period characteristics, predicting the baseline running time in different periods based on the SARIMAX model, specifically including:

[0104] S221, since the conditions of all vehicles on the same line are basically the same, only the relationship model between the departure time and the running time needs to be constructed, the construction process includes:

[0105] 1) Outlier detection and cleaning of historical running time data, eliminating unreasonable data points, and grouping by time period, based on the processed historical running time and time period characteristics to build a SARIMAX model.

[0106] 2) Feature engineering: build time period features (morning peak / flat peak / late peak), weather features, passenger flow features, and other exogenous variables.

[0107] 3) Model order determination: determine the model order (p, d, q) x (P, D, Q) s based on AIC and BIC criteria.

[0108] 4) Parameter estimation: estimate model parameters using maximum likelihood estimation method.

[0109] 5) Model diagnosis: verify model effectiveness through residual analysis.

[0110] S222, after the model is built, divide the whole day operation time into multiple time periods (such as 15 minutes for a time segment), for each time period, calculate the historical average running time at the departure time of that time period;

[0111] S223, for each planned departure event, locate the time period where the departure event occurs, extract the feature vector corresponding to the time period, including time period features, weather features and passenger flow features; based on the SARIMAX model combined with the feature vector of each time period, correct the historical average running time of the time period as the baseline running time of the planned departure event.

[0112] This prediction process is applicable to all vehicles on the same line, without considering individual differences, significantly simplifying the model complexity while ensuring prediction accuracy.

[0113] S23, simulate vehicle operation process: from the first departure time, under the constraint of baseline running time in different time periods, simulate the operation process of each vehicle, including departure, arrival and turnaround.

[0114] S24, generate departure events: in the simulation process, generate corresponding departure events for each link, including departure time, arrival time, running time and stop time, to get the departure event chain of each vehicle.

[0115] S25, verify the integrity of each vehicle's departure event chain, verify the following three aspects: departure time meets the first and last time range requirements, vehicle completes the required number of departure trips, and ensures that the vehicle is parked at the same station.

[0116] In this step, in the construction process of the departure event chain, first, according to the first departure time and the departure interval requirement, a reasonable first departure time is allocated for each vehicle. The allocation of the departure time needs to consider the matching relationship between the number of vehicles and the departure interval, so as to ensure that the stable departure interval can be maintained after the first bus departs.

[0117] Subsequently, based on historical data and time period characteristics, an accurate SARIMAX running time prediction model is constructed, and the model construction is completed through five key steps: first, data preprocessing, abnormal value detection and cleaning of historical running time data, and elimination of unreasonable data points; second, feature engineering, construction of external variables including time period characteristics (morning peak / flat peak / late peak), weather characteristics (sunny / rainy / temperature), passenger flow characteristics, etc.; third, model order determination, determination of the order (p, d, q) x (P, D, Q) s of the model through AIC (Akaike Information Criterion) and BIC (Bayesian Information Criterion); fourth, parameter estimation, estimation of model parameters using maximum likelihood estimation method; and finally, model diagnosis, verification of the effectiveness of the model through residual analysis.

[0118] In the prediction application stage, a day is divided into 96 15-minute time segments, a baseline running time model is established for each time segment t, the historical average running time μt of the time segment is calculated, and the baseline running time yt is predicted through the SARIMAX model combined with the feature vector xt of the current time segment. On this basis, the running process of each vehicle is simulated, including departure, arrival, turnover, etc., and the corresponding departure event is generated, and the key information of each event such as departure time, arrival time, running time, and stop time is recorded in detail. Finally, the integrity of the event chain is verified, mainly in three aspects: first, to ensure that all departure times are within the range of the first and last departure times; second, to verify whether each vehicle has completed the required number of departures; and finally, to ensure that all vehicles can be collected at the same station, ensuring the standardization of operation.

[0119] S3, construct a double-layer constraint matrix, specifically including:

[0120] S31, initialize the matrix structure: create the basic matrix structure of the vehicle layer and the platform layer, with the vehicle layer as the row, representing the complete departure event chain of a vehicle, and the platform layer as the column, representing the departure event at the same time on the platform, and determine the matrix dimension and index relationship.

[0121] S32, fill in the matrix: map the departure event chain to the matrix, the mapping process includes: a) fill in the row data according to the vehicle number and departure order, establish the correspondence between the vehicle and the departure event; b) fill in the column data according to the time period and departure interval requirement, ensure the continuity of the platform departure event;

[0122] Each element in the matrix is a departure event object, each departure event object includes the following attributes: a) departure_time: departure time; b) arrival_time: arrival time; c) runtime_minutes: running time; d) direction: departure direction (up / down); e) from_station: departure station number; f) to_station: arrival station number; g) force_stop_time: forced stop time.

[0123] S33, set constraints for the matrix, including stop time, departure interval and capacity demand constraints;

[0124] S34, establish matrix association: build a four-way linked list relationship between matrix elements, each departure event object is associated with adjacent elements in two directions: a) UP, indicating linking to the previous departure event in the same column, used to maintain the departure order of the platform; b) DOWN, indicating linking to the next departure event in the same column, used to maintain the departure order of the platform; c) LEFT, indicating linking to the previous departure event in the same row, used to track the vehicle running track; d) RIGHT, indicating linking to the next departure event in the same row, used to track the vehicle running track.

[0125] In the construction of the double-layer constraint matrix, a two-dimensional matrix structure containing vehicle layer and platform layer is created. In the horizontal axis (row) direction of the matrix, the vehicle dimension is represented, and each row records the departure event chain of a vehicle, including all events such as departure, arrival, turnaround, etc. of the vehicle throughout the day; in the vertical axis (column) direction, the platform dimension is represented, and each column records the departure event at the same time on the platform, used to manage and optimize the departure interval at the platform level. Each element in the matrix is a departure event (Departure) object, and a four-way linked list relationship is established between the matrix elements to achieve effective association and management of the departure event. This four-way linked list structure enables the system to quickly access and adjust the relevant information of any departure event, providing convenience for subsequent optimization and adjustment.

[0126] S4, perform constraint fusion optimization, the optimization process is based on historical operation data, and the following basic data needs to be provided by the user: a) passenger flow data: passenger flow distribution and trend in each period; b) vehicle data: full load rate, vehicle capacity and other capacity indicators; c) running data: historical turnaround time, interval running time, etc.; d) stop data: stop time distribution of each station. The specific optimization process includes:

[0127] S41, divide time sections: divide the whole day operation time into multiple time sections, including the first shift, morning peak, flat peak, evening peak and last shift.

[0128] S42. Extract time period characteristics: Analyze the passenger flow characteristics and capacity demand characteristics of each time period to determine the constraint priority of each time period;

[0129] Among them, passenger flow characteristics include: 1) Maximum cross-sectional passenger flow: the maximum passenger flow in each time period; 2) Load factor: the ratio of passenger flow to vehicle capacity, calculated as: (passenger flow / (number of departures * vehicle capacity)) * 100%; 3) Passenger flow trend: the rate of change of passenger flow between adjacent time periods.

[0130] The capacity demand characteristics include: 1) the baseline operating time predicted using the SARIMAX model; 2) the baseline departure interval: calculated based on operating time and number of departures, using the formula: operating time / (number of departures - 1); 3) the peak interval: 0.7 times the baseline interval; 4) the off-peak interval: 1.2 times the baseline interval.

[0131] Constraint attributes: 1) Stop duration constraint: Minimum and maximum stop duration requirements for each time period. 2) Departure interval constraint: Minimum and maximum departure interval requirements for each time period. 3) Fixed departure constraint: Whether departures are required at fixed times.

[0132] Priority settings: 1) During peak hours (7:00-9:00, 17:00-19:00), priority is given to ensuring the minimum departure interval and full load rate requirements; 2) During off-peak hours, priority is given to ensuring balanced transport capacity and vehicle turnover; 3) During the first and last bus times and rest and waiting periods, priority is given to meeting the requirements of the operation rules.

[0133] S43. Traverse each departure event in the double-layer constraint matrix and perform the following constraint checks on each departure event:

[0134] 1) Time period constraint verification: a) Departure time verification: Check whether the departure time meets the operating time requirements of the time period; b) Passenger flow full load verification: Verify whether it meets the passenger flow and full load rate requirements of the time period; c) Fixed-point departure verification: Confirm whether it meets the fixed-point departure rules.

[0135] 2) Platform constraint verification: a) Interval requirement verification: Verify whether the interval with the preceding and following departure events meets the requirements; b) Stop time verification: Check whether it complies with the minimum / maximum stop time rules of the platform; c) Peak density verification: For peak periods, verify whether the departure density meets the requirements.

[0136] 3) Vehicle constraint verification: a) Working hours verification: confirm whether the vehicle working hours are within the limit range; b) Turnover time verification: verify whether the vehicle round-trip turnover time is reasonable; c) Rest and standby verification: check whether the vehicle rest and standby requirements are met.

[0137] S44. When a departure event is found to violate a certain constraint, the following strategy shall be adopted for adjustment:

[0138] 1) Departure time adjustment: Calculate a reasonable adjustment range based on the type of constraint violated; find feasible new departure times within the adjustment range; prioritize the adjustment plan that has the least impact on other departure events.

[0139] 2) Linked list update: Get the four-way related elements of the current shift event, process them in the order of up, down, left, and right, and re-execute constraint verification for each related element;

[0140] 3) Conflict handling: If the adjustment results in a new constraint violation, the relevant elements are processed recursively; the adjustment path is recorded to avoid cyclic adjustments; if necessary, backtrack to alternative solutions.

[0141] S45. Global optimization is achieved through multiple iterations, specifically including:

[0142] 1) The departure time adjustment strategy is as follows: a) Forward adjustment: Attempt to advance the departure time by an adjustment amount of 1.2 times the difference in constraint violation, to avoid still violating the constraint after adjustment. b) Backward adjustment: Attempt to delay the departure time by an adjustment amount of 1.2 times the difference in constraint violation. c) Range search: Within the range of forward and backward adjustments, search with the minimum departure interval as the step size. d) Fluctuation adjustment: When the local search gets stuck, allow for a larger range of time adjustments to escape the local optimum.

[0143] 2) Feasibility Verification Methods: a) Constraint Satisfaction: Calculate the degree to which the adjusted plan satisfies various constraints. b) Chain Reaction: Assess the scope and extent of the impact of the adjustment on related departure events. c) Operational Indicators: Verify key indicators such as the adjusted load factor and departure interval. d) Boundary Conditions: Ensure that the adjustment will not lead to the exceeding of first and last shift times or working hours.

[0144] 3) Criteria for determining the optimal solution: a) Number of constraint violations: Count the number of constraint violations that still exist. b) Adjustment magnitude: Calculate the total time change caused by this round of adjustments. c) Indicator improvement: Assess the degree of improvement of key operating indicators. d) Stability: The trend of improvement magnitude over multiple rounds of adjustments.

[0145] 4) Iteration termination conditions: a) Reaching the maximum number of iterations (default 50 rounds). b) The improvement rate is less than 1 minute for 3 consecutive rounds. c) The number of constraint violations remains unchanged for 5 consecutive rounds. d) Oscillation of solutions occurs (the same solution appears repeatedly).

[0146] 5) Rollback mechanism: a) Record the complete solution for each iteration. b) Set checkpoints to save the best solution at each stage. c) Roll back when the new solution is inferior to the historical best solution. d) Try different adjustment strategies at the rollback point.

[0147] The constraint fusion optimization phase of this step first divides the entire day's operating hours into different time slots, including the first bus segment (5:30-7:00), morning peak segment (7:00-9:00), morning off-peak segment (9:00-17:00), evening peak segment (17:00-19:00), evening off-peak segment (19:00-22:00), and last bus segment (22:00-23:00). Passenger flow characteristics and capacity demand characteristics for each time slot are analyzed. Through comprehensive analysis of passenger flow data, vehicle data (load factor, vehicle capacity, etc.), operational data (historical turnaround time, interval travel time, etc.), and stop data (distribution of stop times at each station), the constraint priorities for different time slots are determined. Subsequently, constraint verification is performed on each departure event in the matrix. When a constraint violation is found, a gradual adjustment strategy is adopted. Next, global optimization is completed through multiple iterations. Each iteration includes steps such as departure time adjustment (forward adjustment, backward adjustment, interval search, fluctuation adjustment), feasibility verification (constraint satisfaction, chain effect, operational indicators, boundary conditions), and optimal solution determination (constraint violation number, adjustment range, indicator improvement, stability).

[0148] S5. Based on time period characteristics and operational needs, identify abnormal time segments and compress or smooth them, specifically including:

[0149] S51. Identify abnormal sections:

[0150] 1) Time period classification: The operating hours are divided into four basic time periods: first bus period (FIRST_TRIP), peak period (PEAK), off-peak period (OFF_PEAK) and last bus period (LAST_TRIP).

[0151] 2) Hierarchical segmentation: Construct a hierarchical time segment management structure (TimeSection) that supports parent-child relationship management, facilitating overall and local segment control.

[0152] 3) Anomaly Detection: Based on departure interval deviation, segment continuity, capacity distribution, and constraint satisfaction indicators, abnormal time segments requiring compression or smoothing are identified.

[0153] a) Departure interval deviation judgment: Calculate the difference between the departure interval and the target interval within the section. When the difference exceeds the threshold, it is judged as abnormal. The threshold is 5 minutes during peak hours and 15 minutes during off-peak hours.

[0154] b) Segment continuity determination: Check whether the transition between adjacent segments is smooth.

[0155] c) Determining the distribution of transport capacity: Assess whether the distribution of transport capacity within the segment is balanced.

[0156] d) Constraint satisfaction determination: Statistics on constraint violations within the segment, including the number of violations of departure interval constraints and station dwell time constraints.

[0157] S52. Calculate the compression space: Evaluate the compressible range of each abnormal time segment, considering vehicle operation constraints, station stopping constraints, and departure interval constraints, calculate the minimum compression unit, and determine the compression boundary on the premise that it does not affect subsequent departure events and preceding departure events.

[0158] The minimum compression unit refers to the smallest compressible time unit within a given segment. Its determination process includes: 1) Operational constraint calculation: Considering vehicle operation constraints, ensuring the compressed departure interval is not less than 3 minutes. 2) Stop constraint calculation: Considering station stop requirements, ensuring the compressed stop time is not less than the default minimum stop time of 5 minutes. 3) Interval constraint calculation: Considering departure interval restrictions for different time periods, ensuring the compressed interval meets the time period characteristics requirements.

[0159] The determination of compression boundaries includes: 1) Forward boundary calculation: Calculate the maximum adjustable time for compression towards the termination point to ensure that subsequent departure events are not affected. 2) Backward boundary calculation: Calculate the maximum adjustable time for compression towards the starting point to ensure that preceding departure events are not affected. 3) Scheme evaluation: Select the compression direction and magnitude, prioritizing schemes with a smaller impact on the overall departure schedule.

[0160] S53. Adopt a gradual compression strategy to compress abnormal time segments, specifically including:

[0161] S531. Compression interval determination: 1) Interval endpoint determination: Determine the start and end time points of the compression interval, which remain unchanged during the compression process; 2) Interval range verification: Check whether the length of the compression interval meets the minimum compression requirement; 3) Boundary condition recording: Record the departure event information corresponding to the endpoints of the compression interval as boundary constraints for the compression process.

[0162] S532. Compression Direction Selection: Determine the target departure interval and select either forward compression mode or backward compression mode;

[0163] 1) Target departure interval calculation: Based on the interval type (peak / off-peak) and interval length, calculate the ideal target departure interval: a) Peak interval: target interval = min(5 minutes, interval length / number of departure events); b) Off-peak interval: target interval = min(15 minutes, interval length / number of departure events).

[0164] 2) When forward compression mode is selected, i.e. compression towards the end point, the end point of the interval is fixed, and the departure events within the interval are pushed forward sequentially. Specifically, starting from the start time of the compressed interval, the new departure time is calculated sequentially according to the target departure interval; the new time of each departure event is equal to the departure time of the previous departure event plus the target interval; the departure time of the last departure event must not exceed the end time of the compressed interval.

[0165] 3) When selecting the backward compression mode, i.e., selecting compression towards the starting point, the starting point of the interval is fixed, and the departure events within the interval are pushed forward sequentially. Specifically, starting from the end point of the compressed interval, the new departure time is calculated in reverse order according to the target departure interval; the new time of each departure event is equal to the departure time of the next event minus the target interval; the departure time of the first departure event must not be earlier than the starting time of the compressed interval.

[0166] S533, Perform the compression process:

[0167] 1) Departure event sorting: Arrange the departure events within the compression interval in chronological order and establish a compression processing queue.

[0168] 2) Progressive compression execution: Adjust the departure time step by step according to the selected compression mode. During the adjustment process, the following must be met: a) The amount of adjustment in a single adjustment shall not exceed the minimum compression unit (default 2 minutes); b) Check the constraint satisfaction after each adjustment; c) If a constraint violation occurs, revert to the previous feasible state.

[0169] 3) Related event processing: For each adjusted departure event, obtain its four-way related events (UP, DOWN, LEFT, RIGHT) in the two-level constraint matrix, check whether the adjustment affects the constraints of the related events, and if so, adjust the related events synchronously.

[0170] S534. Verification of compression results mainly includes the following three aspects: 1) Interval uniformity check: verify whether the standard deviation of the departure interval in the compressed section meets the requirements; 2) Constraint satisfaction check: confirm whether all adjusted departure events meet the vehicle and platform constraints; 3) Boundary connection check: verify whether the connection between the compressed section and the adjacent section is smooth.

[0171] S535, Abnormal situation handling, mainly includes:

[0172] 1) Insufficient space handling: When the length of the compressed interval is insufficient to accommodate all departure events, expand the range of the compressed interval and adjust the target departure interval, or split it into multiple sub-intervals for separate processing.

[0173] 2) Constraint conflict handling: When a constraint conflict occurs that cannot be satisfied, prioritize the high-priority constraints (such as the minimum departure interval during peak hours), relax the low-priority constraints (such as the stop duration), and record the unresolved constraint conflicts for subsequent optimization.

[0174] S54. Determine the smoothing parameters based on the time period characteristics, and perform smoothing processing on abnormal segments according to the smoothing parameters;

[0175] The smoothing parameters include baseline parameters and adjustment parameters. The baseline parameters include: the baseline interval for the first bus (set to 10 minutes), the baseline interval for peak hours (set to 5 minutes), the baseline interval for off-peak hours (set to 15 minutes), and the baseline interval for the last bus (set to 15 minutes).

[0176] The adjustment parameters include: adjusting the time step (set to 2 minutes), the maximum number of iterations (10 rounds), and the range of interval coefficients (minimum value is 0.8, maximum value is 1.5), and setting the satisfaction threshold to 0.8.

[0177] After setting the smoothing parameters, perform the smoothing process: 1) Segment smoothing: a) Target interval calculation: Determine the ideal departure interval based on the time period type. b) Gradual adjustment execution: Adjust the departure time step by step with a minimum step size of 2 minutes. c) Constraint check execution: Ensure that the adjustment does not violate vehicle and platform constraints.

[0178] 2) Optimization Termination Criteria: a) Iteration Count Criteria: Terminate when the maximum number of iterations (10 rounds) is reached. b) Improvement Magnitude Criteria: Terminate when the improvement magnitude is less than the threshold. c) Satisfaction Criteria: Terminate when the satisfaction level reaches the target value of 0.8. d) Oscillation Criteria: Terminate when oscillation of the solution occurs.

[0179] In the segment compression and smoothing process, the operating time is first divided into four basic time periods by time period classification: first bus period (FIRST_TRIP), peak period (PEAK), off-peak period (OFF_PEAK), and last bus period (LAST_TRIP), and a hierarchical time segment management structure (TimeSection) is constructed to support parent-child relationship management.

[0180] Next, based on indicators such as departure interval deviation judgment (peak period threshold of 5 minutes, off-peak period threshold of 15 minutes), segment continuity judgment, capacity distribution judgment, and constraint satisfaction judgment, the time segments that need to be compressed or smoothed are identified.

[0181] Subsequently, the compressibility range of each abnormal segment is evaluated, and a compression scheme is planned by calculating the minimum compression unit (considering operational constraints, docking constraints, and interval constraints) and determining the compression boundaries (forward and backward boundaries). Finally, the compression process is executed, employing a progressive compression strategy when compressing segments.

[0182] The process includes steps such as departure event sequencing, target interval calculation, progressive compression execution, and related event handling. Within a segment, target interval calculation, progressive adjustment execution, and constraint checks are performed until the termination conditions are met.

[0183] S6. Conduct a comprehensive review of the compressed or smoothed scheduling plan, including:

[0184] 1) Perform constraint checks to comprehensively check whether the scheduling plan meets all constraints.

[0185] 2) Verify the time period coverage and confirm whether the departure arrangements for each time period meet the transportation capacity demand.

[0186] 3) Generate timetables, organize and optimize the scheduling plan, and generate detailed departure timetables.

[0187] When generating the final scheduling plan, a comprehensive check is performed to ensure that the plan meets all constraints, including vehicle-level constraints (such as working hours and turnaround time) and platform-level constraints (such as departure intervals and stop durations). Simultaneously, it is verified whether the departure arrangements for each time period meet capacity demands, particularly whether they can meet passenger flow needs during peak hours and whether there is any capacity waste during off-peak hours. If any constraints are found to be unmet, the departure intervals are fine-tuned. After confirming that all checked items meet the requirements, a detailed departure timetable is generated, containing complete information such as departure time, arrival time, and stop time for each vehicle, providing precise guidance for actual operations.

[0188] The various embodiments in this specification are described in a progressive manner, with each embodiment focusing on its differences from other embodiments. Similar or identical parts between embodiments can be referred to interchangeably. For the apparatus disclosed in the embodiments, since they correspond to the methods disclosed in the embodiments, the description is relatively simple; relevant parts can be referred to the method section.

[0189] The above description of the disclosed embodiments enables those skilled in the art to make or use the invention. Various modifications to these embodiments will be readily apparent to those skilled in the art, and the general principles defined herein may be implemented in other embodiments without departing from the spirit or scope of the invention. Therefore, the invention is not to be limited to the embodiments shown herein, but is to be accorded the widest scope consistent with the principles and novel features disclosed herein.

Claims

1. A method for intelligent scheduling of single-route buses based on two-layer constraint fusion, characterized in that, Includes the following steps: S1. Initialize the bus scheduling parameters for a specific route; S2. Based on the first departure time and departure interval requirements, assign the first departure time to all vehicles on this route. Starting from the first departure time, simulate the operation of each vehicle in sequence, construct the departure event chain, and verify the integrity of the departure event chain. S3. Construct a two-layer constraint matrix, using the complete departure event chain of a train as rows and departure events at the same time on the platform as columns; S3 includes: S31. Initialize the matrix structure: Create the basic matrix structure for the vehicle layer and the platform layer. The vehicle layer is used as the rows to represent the complete departure event chain of a vehicle, and the platform layer is used as the columns to represent the departure events at the platform at the same time. Determine the matrix dimensions and index relationships. S32. Map the departure event chain to the matrix. The mapping process includes: a) filling in row data according to vehicle number and departure order to establish the correspondence between vehicle and departure event; b) filling in column data according to time period and departure interval requirements to ensure the continuity of platform departure events. Each element in the matrix is ​​a departure event object, and each departure event object includes the following attributes: departure time, arrival time, running time, departure direction, departure station number, arrival station number, and forced stop duration. S33. Set constraints for the matrix, including stop duration, departure interval and capacity demand constraints; S34. Construct a four-way linked list relationship between matrix elements. Each departure event object establishes a bidirectional association with its adjacent elements: a) Upward, indicating a link to the previous departure event in the same column, used to maintain the platform departure order; b) Downward, indicating a link to the next departure event in the same column, used to maintain the platform departure order; c) Leftward, indicating a link to the previous departure event in the same row, used to track the vehicle's trajectory; d) Rightward, indicating a link to the next departure event in the same row, used to track the vehicle's trajectory. S4. Divide the daily operating hours into different time periods and extract the time period characteristics of each time period; determine the constraint priority of each time period based on the time period characteristics of different time periods, and perform constraint verification and adjustment on each departure event in the double-layer constraint matrix based on the constraint priority; S5. Based on time period characteristics and operational needs, identify abnormal time segments and compress or smooth them. S6. Conduct a comprehensive check on the compressed or smoothed scheduling plan. If all constraints and capacity requirements are met, generate the final scheduling plan.

2. The intelligent scheduling method for single-route buses based on dual-layer constraint fusion according to claim 1, characterized in that, S1 includes: S11. Determine basic operating parameters based on route characteristics, including first and last bus times and peak hour divisions; S12. Develop differentiated departure interval requirements for different time periods, including the maximum and minimum departure intervals during peak hours, off-peak hours, and the first and last bus periods; S13. Set the maximum and minimum working time for a single vehicle, as well as the minimum and maximum stopping time requirements for different time periods; S14. Divide the entire day's operating time into multiple time periods at fixed intervals, calculate the vehicle's running time when departing in each time period, and establish a time period-running time mapping relationship. S15. Based on the running time analysis results, calculate the round-trip cycle of a single vehicle, and in combination with the working time limit, determine the theoretical number of trips for each vehicle.

3. The intelligent scheduling method for single-route buses based on dual-layer constraint fusion according to claim 1, characterized in that, S2 include: S21. Based on the first departure time and departure interval requirements, assign the first departure time to each vehicle and generate the first departure time sequence; S22. Construct a SARIMAX model based on historical data and time period characteristics, and predict the baseline running time for different time periods based on the SARIMAX model; S23. Starting from the departure time, under the constraint of the baseline running time in different time periods, simulate the operation process of each vehicle in sequence, including departure, arrival and turnaround. S24. During the simulation, a corresponding departure event is generated for each link, including departure time, arrival time, running time and stop time, to obtain the departure event chain for each vehicle; S25. Perform integrity verification on the departure event chain for each vehicle. Verification objectives include: ensuring that the departure time meets the first and last departure time range requirements, that the vehicle completes the required number of departure trips, and ensuring that the vehicle returns to the same station.

4. The intelligent scheduling method for single-route buses based on dual-layer constraint fusion according to claim 3, characterized in that, S22 includes: Outlier detection and cleaning are performed on historical runtime data to remove unreasonable data points, and a SARIMAX model is constructed based on the processed historical runtime data. The entire day's operating hours are divided into multiple time periods at fixed intervals. For each time period, the historical average travel time at the time of departure during that time period is calculated. For each planned departure event, the time period in which the departure event occurs is located, and the corresponding feature vector for that time period is extracted, including time period features, weather features, and passenger flow features. Based on the SARIMAX model, the historical average running time corresponding to each time period is corrected by combining the feature vectors of each time period, and this is used as the benchmark running time for the planned departure event.

5. The intelligent scheduling method for single-route buses based on dual-layer constraint fusion according to claim 1, characterized in that, S4 include: S41. Divide the entire day's operating hours into multiple time periods, including the first bus, morning peak, off-peak, evening peak, and last bus. S42. Analyze the passenger flow characteristics and capacity demand characteristics of each time period to determine the constraint priority of each time period; S43. Traverse each departure event in the double-layer constraint matrix and perform the following checks on each departure event: Time-of-day constraint verification: Check whether the departure time meets the operating time requirements of the time period; verify whether it meets the passenger flow and load factor requirements of the time period; confirm whether it meets the fixed-point departure rules; Platform constraint verification: Verify whether the interval between the train and the preceding and following departure events meets the requirements; check whether it complies with the minimum / maximum dwell time rules of the platform; for peak hours, verify whether the departure density meets the requirements. Vehicle constraint verification: Confirm whether the vehicle's working hours are within the limit range; verify whether the vehicle's round-trip turnaround time is reasonable; check whether the vehicle's rest and standby requirements are met; S44. When a departure event is found to violate a certain constraint, the departure time is adjusted according to the type of constraint violation, and the four-way related elements of the current departure event are obtained and processed in the order of up, down, left, and right. Constraint verification is re-executed for each related element. If the adjustment causes a new constraint violation, the related elements are processed recursively.

6. The intelligent scheduling method for single-route buses based on dual-layer constraint fusion according to claim 1, characterized in that, In S42, passenger flow characteristics include: the maximum passenger flow in each time period, the ratio of passenger flow to vehicle capacity, and the rate of change in passenger flow between adjacent time periods; Capacity demand characteristics include: baseline operating time, baseline departure interval, peak interval, and off-peak interval as predicted using the SARIMAX model; During peak hours, priority is given to ensuring minimum departure intervals and full load capacity; during off-peak hours, priority is given to ensuring balanced transport capacity and vehicle turnover; during the first and last bus times and rest / waiting periods, priority is given to meeting operational rules.

7. The intelligent scheduling method for single-route buses based on dual-layer constraint fusion according to claim 1, characterized in that, S5 include: S51. Based on the departure interval deviation judgment, section continuity judgment, capacity distribution judgment and constraint satisfaction judgment indicators, abnormal time sections that need to be compressed or smoothed are identified. S52. Evaluate the compressible range of each abnormal time segment, taking into account vehicle operation constraints, station stopping constraints, and departure interval constraints, calculate the minimum compression unit, and determine the compression boundary on the premise that it does not affect subsequent departure events and preceding departure events. S53. Adopt a gradual compression strategy to compress abnormal time segments; S54. Determine the smoothing parameters based on the time period characteristics, and perform smoothing processing on abnormal segments according to the smoothing parameters; the smoothing parameters include baseline parameters and adjustment parameters. The baseline parameters include: baseline interval for the first bus time period, baseline interval for peak hours, baseline interval for off-peak hours, and baseline interval for the last bus time period; the adjustment parameters include: adjustment time step, maximum number of iterations, and range of interval coefficients.

8. The intelligent scheduling method for single-route buses based on dual-layer constraint fusion according to claim 7, characterized in that, S53 includes: S531. Determine the start and end times of the compression interval, which remain unchanged during the compression process; check whether the length of the compression interval meets the minimum compression requirement, and record the departure event information corresponding to the endpoints of the compression interval as boundary constraints for the compression process. S532. Determine the target departure interval and select either forward compression mode or backward compression mode; When forward compression mode is selected, the new departure time is calculated sequentially according to the target departure interval, starting from the beginning time of the compression interval; the new time of each departure event is equal to the departure time of the previous departure event plus the target interval; the departure time of the last departure event must not exceed the end time of the compression interval. When the backward compression mode is selected, the new departure time is calculated in reverse order from the end point of the compression interval according to the target departure interval; the new time of each departure event is equal to the departure time of the next event minus the target interval; the departure time of the first departure event must not be earlier than the start time of the compression interval. S533, Perform the compression process: Arrange the departure events within the compression interval in chronological order and establish a compression processing queue; According to the selected compression mode, the departure time is gradually adjusted. During the adjustment process, the amount of adjustment in a single instance shall not exceed the minimum compression unit. After each adjustment, the constraint satisfaction shall be checked. If a constraint violation occurs, the system shall revert to the previous feasible state. For each adjusted departure event, obtain its four-way associated events in the two-level constraint matrix, check whether the adjustment affects the constraints of the associated events, and if so, adjust the associated events synchronously. S534. Verification of Compression Results: Verify whether the standard deviation of the departure interval within the compressed section meets the requirements; confirm whether all adjusted departure events meet the vehicle and platform constraints; verify whether the connection between the compressed section and adjacent sections is smooth. S535, Abnormal Situation Handling: When the compressed interval length is insufficient to accommodate all departure events, the compressed interval range is expanded and the target departure interval is adjusted, or it is split into multiple sub-intervals for separate processing. When a constraint conflict occurs that cannot be satisfied, prioritize ensuring the high-priority constraint, relax the low-priority constraint, and record the unresolved constraint conflict for later optimization.

9. The intelligent scheduling method for single-route buses based on dual-layer constraint fusion according to claim 7, characterized in that, In S51, the method for determining abnormal time periods is as follows: Departure interval deviation judgment: Calculate the difference between the departure interval and the target interval within the section. When the difference exceeds the threshold, it is judged as abnormal. The threshold is 5 minutes during peak hours and 15 minutes during off-peak hours. Section continuity determination: Check whether the transition between adjacent sections is smooth; Capacity distribution assessment: Assess whether the capacity allocation within the segment is balanced; Constraint satisfaction determination: Statistics on constraint violations within the segment, including the number of violations of departure interval constraints and station dwell time constraints.

Images