Reservation decision and yard crane scheduling co-optimization method and system considering default disturbance

Abstract

This invention discloses a method and system for collaborative optimization of reservation decision-making and yard crane scheduling considering the disturbance of missed appointments. The method includes: collecting and preprocessing terminal and operation data; inputting the processed terminal and operation data into a static collaborative optimization model of external truck reservation and yard crane scheduling and solving the static collaborative optimization model to obtain a baseline scheduling scheme; executing external truck operation tasks according to the baseline scheduling scheme, and monitoring the missed appointment data of external trucks in real time; inputting the missed appointment data into a rescheduling model to adjust and optimize the baseline scheduling scheme in real time, and enabling external trucks to operate according to the optimized scheduling scheme. By establishing a static collaborative optimization model and a rescheduling model for external truck reservation and yard crane scheduling, this invention avoids the accumulation of errors in the scheduling scheme caused by continuous random missed appointment disturbances without disrupting the continuity of the original yard crane operation plan, thereby improving the continuity and executability of the scheduling scheme in the overall operation.

Description

Technical Field

[0001] This invention relates to the field of landside operation scheduling technology for container terminals, and in particular to a method and system for collaborative optimization of reservation decision-making and yard crane scheduling that takes into account the disturbance of missed appointments. Background Technology

[0002] With the continued growth of global container trade, the trend towards larger ships is becoming increasingly significant, with single-ship loading capacity exceeding 24,000 TEU. The berthing of large ships triggers a surge in container loading and unloading within a short period, placing immense instantaneous pressure on terminals in areas such as gate passage, yard operations, and crane dispatch. As the primary means of connecting ports with inland hinterlands, the disorderly arrival of external trucks can easily cause severe congestion in port areas. Furthermore, the existing truck reservation system suffers from two key flaws that limit its actual operational effectiveness.

[0003] First, the existing reservation system treats yard crane service capacity as a fixed constant when setting quotas, ignoring the dynamic occupancy of yard crane resources by ship loading operations. In actual operation of quay-side terminals, yard cranes must strictly adhere to the principle of prioritizing sea-side loading operations. Their capacity available to serve container trucks is essentially a dynamic surplus value after deducting the occupancy from loading operations, rather than a statically given constant. During peak loading periods, yard cranes may dedicate most of their capacity to loading, leaving their service capacity for container trucks close to zero. The existing reservation system fails to capture this dynamic change when issuing reservation quotas, resulting in a severe disconnect between quota settings and the actual available capacity of yard cranes. Although container trucks may pass through the gate smoothly, secondary queuing and congestion occur within the yard due to insufficient yard crane service capacity, significantly diminishing the effectiveness of the reservation mechanism.

[0004] Second, existing response mechanisms for truck delays are mostly single-time static triggers, lacking effective capabilities to cope with persistent random disturbances. In actual port operations, delays due to road congestion, vehicle malfunctions, documentation issues, and other reasons frequently occur within the scheduled time window, persisting throughout the entire planning period. These are persistent random disturbances rather than isolated events. Existing research largely relies on single-time static rescheduling. If a delay occurs again in subsequent periods, the error will continue to accumulate on the previously adjusted plan, resulting in a lack of system stability and a gradual decrease in the executability of the scheduling scheme over time. Summary of the Invention

[0005] To address the issues in existing technologies such as insufficient consideration of the dynamic occupation of yard crane resources by loading operations and the tendency for scheduling schemes to become less executable over time, this invention provides a collaborative optimization method and system for reservation decision-making and yard crane scheduling that takes into account the disturbance of missed appointments, thereby improving the sustainability and executability of scheduling schemes in the overall operation.

[0006] Therefore, the technical solution adopted by the present invention is as follows: A collaborative optimization method for reservation decision-making and yard bridge scheduling considering no-show disturbances is provided, the method comprising: Collect and preprocess terminal and operation data, input the processed terminal and operation data into the static collaborative optimization model of external truck reservation and yard crane scheduling, and solve the static collaborative optimization model to obtain the baseline scheduling scheme. The external container truck operation tasks are executed according to the baseline scheduling plan, and the data of missed tasks by the external container trucks are monitored in real time. The missed task data is input into the rescheduling model to adjust and optimize the baseline scheduling plan in real time, and the external container trucks are operated according to the optimized scheduling plan. Among them, the static collaborative optimization model of external truck reservation and yard bridge scheduling is established with minimizing yard bridge operating costs and cumulative queuing time as the objective function, and the optimal solution set of the scheduling scheme is solved, from which the compromise solution is selected as the benchmark scheduling scheme; the rescheduling model is specifically established with minimizing the cumulative impact of missed task return on truck waiting time in future periods as the objective function, and the optimal solution at this time is solved as the optimized scheduling scheme.

[0007] According to the above plan, the terminal and operation data include vessel berthing plans and port arrival time windows, total container loading demand of each vessel and target container area, loading operation volume of each container area in each time period, number of gate channels and service efficiency, number of available yard bridges and service efficiency, number of container yard areas and spatial layout parameters.

[0008] According to the above scheme, the preprocessing of terminal and operation data includes: dividing the process of external container trucks entering the port into a two-level serial queuing network consisting of the gate system and the yard system, and using the point-by-point fixed fluid approximation method to transform the non-stationary queuing process of external container trucks into time-by-time algebraic constraints; the non-stationary queuing process specifically refers to the arrival flow of external container trucks after leaving the gate and flowing through the port road network to form the yard.

[0009] According to the above scheme, the static collaborative optimization model of external truck reservation and yard crane scheduling satisfies the following constraints: reservation and loading task constraints, loading priority and remaining service capacity of yard cranes constraints, gate queuing constraints, yard queuing constraints, and yard crane location and operation status constraints.

[0010] According to the above scheme, the static collaborative optimization model is solved using an improved multi-objective evolutionary algorithm, which is specifically based on the NSGA-II algorithm and is as follows: The terminal and operational data are classified into yard crane scheduling segments and truck reservation quota segments, and the segments are coded in two layers. The total demand heat of each container area in each time period is calculated based on the encoded fragments, and the calculation results are sorted. The initial solution of the static collaborative optimization model is solved by combining the sorting results. The initial solution is iterated, and the initial solution generated by the iteration is optimized by combining the field bridge resource correction and the reservation adaptive flow adjustment. The iteration stops when the termination condition is met, and the Pareto optimal solution set is obtained. The compromise solution is selected as the optimal solution according to the Pareto optimal solution set.

[0011] According to the above scheme, the rescheduling model satisfies the following constraints: Rescheduling time constraints, rescheduling flow constraints, rescheduling quantity constraints, and rescheduling state constraints.

[0012] According to the above scheme, the rescheduling model is specifically solved using a heuristic greedy insertion algorithm, which is as follows: the feasible region of the ships to which the vacant trucks belong is screened to determine the set of future time periods in which the vacant trucks are allowed to be inserted; the congestion of each time period in the set of future time periods is evaluated and sorted, and the vacant trucks are inserted selectively; the congestion of the inserted time periods is re-evaluated until all vacant trucks are redistributed.

[0013] According to the above plan, the operating cost of the yard crane is specifically calculated based on the yard crane relocation cost, cross-regional transfer cost, and empty consumption cost; the cumulative queuing time is specifically calculated based on the queue length of the yard and the queue length of the gate channel; the cumulative impact of the return of missed tasks on the waiting time of container trucks in future periods is specifically calculated based on the cumulative queuing time of external container trucks in future periods.

[0014] A collaborative optimization system for reservation decision-making and yard bridge scheduling that considers the disturbance of non-reservation is also provided, the system comprising: The benchmark scheduling solution module is used to collect terminal and operation data and preprocess it. The processed terminal and operation data is then input into the static collaborative optimization model of external truck reservation and yard crane scheduling, and the static collaborative optimization model is solved to obtain the benchmark scheduling solution. The missed task optimization module is used to execute external truck operation tasks according to the baseline scheduling scheme, monitor the missed task data of external trucks in real time, input the missed task data into the rescheduling model, adjust and optimize the baseline scheduling scheme in real time, and enable external trucks to perform operations according to the optimized scheduling scheme. Among them, the static collaborative optimization model of external truck reservation and yard bridge scheduling is established with minimizing yard bridge operating costs and cumulative queuing time as the objective function, and the optimal solution set of the scheduling scheme is solved, from which the compromise solution is selected as the benchmark scheduling scheme; the rescheduling model is specifically established with minimizing the cumulative impact of missed task return on truck waiting time in future periods as the objective function, and the optimal solution at this time is solved as the optimized scheduling scheme.

[0015] A computer storage medium is also provided, which stores a computer program that can be executed by a processor, the computer program performing the aforementioned method for coordinated optimization of reservation decision and yard bridge scheduling.

[0016] The beneficial effects of this invention are as follows: By establishing a static collaborative optimization model and a rescheduling model for external truck reservation and yard crane scheduling, this invention generates a baseline scheduling plan using the static collaborative optimization model during the scheduling plan generation stage. During the execution stage, it monitors the data of unscheduled external truck tasks in real time and adjusts the baseline scheduling plan in real time using the rescheduling model. When a task is unscheduled while keeping the original yard crane scheduling plan locked, the unscheduled task can be reassigned. This avoids the accumulation of errors in the scheduling plan caused by continuous random unscheduled disturbances without disrupting the continuity of the original yard crane operation plan, thereby improving the continuity and executability of the scheduling plan in the overall operation.

[0017] Furthermore, by incorporating the remaining service capacity constraint of yard cranes into the static collaborative optimization model of external truck reservation and yard crane scheduling, this invention takes the dynamic occupancy of loading operations as a factor to be considered in the scheduling scheme generation process, breaking down the decision-making barriers between the reservation system and the yard crane scheduling system, reducing secondary congestion of external trucks in the yard during peak loading periods, and improving the executability and reliability of the generated scheduling scheme.

[0018] Furthermore, this invention designs a greedy initialization strategy based on job load heat in the improved multi-objective evolutionary algorithm, calculates and sorts the total demand heat of each container area in each time period, and solves the initial solution of the scheduling scheme based on the sorting results, so that the resource allocation of the initial population is consistent with the demand distribution in a macro trend, thereby improving the quality of the initial solution. Attached Figure Description

[0019] Figure 1 This is a schematic diagram of the method flow of the collaborative optimization method for reservation decision-making and yard bridge scheduling that considers the disturbance of missed appointments according to an embodiment of the present invention; Figure 2 This is a schematic diagram of a two-stage serial queuing network according to an embodiment of the present invention; Figure 3 This is a flowchart of the static collaborative optimization model solution process according to an embodiment of the present invention; Figure 4 This is a schematic diagram of the double-layer chromosome coding structure according to an embodiment of the present invention; Figure 5 This is a flowchart of the rescheduling model solution in an embodiment of the present invention; Figure 6 This is a schematic diagram comparing the performance of the static collaborative optimization model solving algorithm according to an embodiment of the present invention; Figure 7 This is an analysis chart of congestion periods for external trucks according to an embodiment of the present invention; Figure 8 This is a comparison of the effects of rescheduling before and after in an embodiment of the present invention. Figure 1 ; Figure 9 This is a comparison of the effects of rescheduling before and after in an embodiment of the present invention. Figure 2 ; Figure 10 This is a schematic diagram of the system structure of the collaborative optimization system for reservation decision-making and yard bridge scheduling that considers the disturbance of missed appointments in an embodiment of the present invention. Detailed Implementation

[0020] To make the objectives, technical solutions, and advantages of this invention clearer, the invention will be further described in detail below with reference to the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are merely illustrative and not intended to limit the invention.

[0021] To address the shortcomings of existing technologies that treat yard crane service capacity as a fixed constant, ignore the dynamic occupation of yard crane resources by ship loading operations, and whose existing response mechanisms for truck no-shows are mostly single-time static triggers lacking effective response to continuous random disturbances, this invention provides a collaborative optimization method for reservation decision-making and yard crane scheduling that considers no-show disturbances. Figure 1 As shown, the method includes: S1. Collect and preprocess terminal and operation data, input the processed terminal and operation data into the static collaborative optimization model of external truck reservation and yard crane scheduling, and solve the static collaborative optimization model to obtain the baseline scheduling scheme.

[0022] S2. Execute external container truck operation tasks according to the baseline scheduling scheme, monitor the missed task data of external container trucks in real time, input the missed task data into the rescheduling model, adjust and optimize the baseline scheduling scheme in real time, and enable external container trucks to perform operations according to the optimized scheduling scheme.

[0023] Specifically, the static collaborative optimization model for external truck reservation and yard bridge scheduling is established with the objective function of minimizing yard bridge operating costs and cumulative queuing time, and the baseline scheduling scheme including external truck reservation tasks and yard bridge scheduling tasks is solved; the rescheduling model is established with the objective function of minimizing the cumulative impact of missed task return on truck waiting time in future periods, and the adjusted and optimized scheduling scheme is solved.

[0024] Specifically, the preprocessing of terminal and operational data involves abstracting the process of container trucks entering the port into a form such as... Figure 2The diagram illustrates a two-tiered queuing network consisting of a gate system and a yard system. The gate system serves as the landside entrance, responsible for verifying the identity of external trucks and allocating tasks. The yard system is the core operational area, where external trucks wait for yard cranes to provide loading and unloading services. The two systems are connected in series; the outbound flow from the gates is transmitted through the port's road network to form the inbound flow from the yard. This embodiment employs a point-by-point fixed fluid approximation method to transform the aforementioned non-stationary queuing process into time-by-time algebraic constraints, thereby embedding a static collaborative optimization model.

[0025] The state evolution of the i-th channel of the gate system during time period t can be expressed as:

[0026]

[0027] in, The queue length of the i-th gate channel at the end of time period t; The number of external trucks arriving at the i-th channel within time period t; The number of external cards leaving the i-th channel within time period t; This represents the upper limit of the service capacity of a single gate channel. This indicates that the actual service volume is the smaller value between the current number of vehicles waiting to be served and the channel's service capacity; The state transition equation is for the queue leader at the gate.

[0028] Furthermore, for the yard system, the state evolution of container area j in time period t can be expressed as:

[0029]

[0030] in, The queue length of box area j at the end of time period t; The number of external trucks arriving at container area j within time period t; The number of external container trucks actually served in container area j within time period t; The effective remaining service capacity of container area j within time period t can be used to serve external container trucks.

[0031] Specifically, the static collaborative optimization model of external truck reservation and yard crane scheduling can be expressed as:

[0032]

[0033] In the formula, The operating cost of the yard crane consists of three parts: yard crane relocation cost, inter-regional relocation cost, and idle cost. This represents the set of reservation time slots within the planning period. Represents the set of available field bridges. This indicates the collection of export container areas within the yard; This represents the cost coefficient per unit travel distance of the yard crane. This represents the air consumption cost coefficient of the yard crane per unit time period; Indicates the yard bridge from the container area Move to container area Distance from the center; This represents the cost of a single turn when the yard bridge moves across the work lane. This indicates that if the yard bridge is from the container area Move to If it needs to cross a work lane, the value is 1; otherwise, it is 0. This represents a 0-1 variable; if the field bridge n is located in box zone j during time period t, it is 1; otherwise, it is 0. This represents a 0-1 variable. If the field bridge n is in an effective working state in box area j during time period t, it is 1; otherwise, it is 0. The system accumulates queue lengths, which are proportional to the total waiting time of trucks according to Little's Law, and use this as a proxy indicator of waiting time. Indicates the collection of dock gate channels; To introduce a generalized yard queue length with a nonlinear congestion penalty term; This represents the queue length of the i-th gate at the end of time period t.

[0034] in, The calculation method can be expressed as:

[0035] in Let j be the operational saturation level of container area j during time period t. ; This represents the total reservation quota for arrival at container area j during time period t; Represents the power of k. This represents the congestion risk reduction factor. With k=4 and Ω=360, the penalty term exhibits a nonlinear jump when the saturation approaches its upper limit, making the static co-optimization model sensitive to stochastic congestion risk and compensating for the shortcomings of deterministic fluid models in describing stochasticity.

[0036] Specifically, the static system optimization model satisfies the following constraints: reservation and loading task constraints, loading priority and remaining service capacity of yard cranes constraints, gate queuing constraints, yard queuing constraints, and yard crane location and operation status constraints.

[0037] The specific constraints on reservation and loading tasks are as follows: The sum of the reservation quotas for all target container areas for each vessel within the port arrival time window equals its planned total container volume: ; Container trucks can only book the target container area corresponding to the vessel: ; The reservation quota outside the port arrival time window is forcibly set at zero: ; Total arrival reservations for each container area at each time period: ; in, For vessel z, the reservation quota of container trucks for delivery to container area j is allocated during time period t. , These are the starting and ending points of the ship's arrival at the port. For the target container area set of the export containers of ship z; For the planned total container load of vessel z, Z represents the sum of the total vessel reservation quotas arriving at container area j during time period t, and Z represents the total number of vessels involved within the planning period.

[0038] The constraints of loading priority and remaining yard crane service capacity are as follows: The loading priority constraint means that, for any container area at any given time, the total capacity reserved by yard cranes for loading operations must not be less than the loading demand for that time period. ; The remaining service capacity constraint for yard cranes is defined as the dynamic remaining service capacity of container yard j available to serve external trucks in time period t, which is the remaining value after deducting the loading occupancy from the total service capacity of all yard cranes in that container yard: ; in, The yard crane n reserves capacity for loading operations in container area j during time period t. The loading task to be completed in container area j during time period t; N is the total number of available yard cranes. These two constraints directly incorporate the dynamic occupation of yard crane resources by loading operations into the external container truck reservation decision constraint system.

[0039] The specific queuing constraints at the gate are as follows: The total truck traffic arriving at the gate system during time period t is equal to the sum of the reservation quotas for all container areas during that time period: ; The total arrival flow is evenly distributed among the gate channels: ; in, Let J be the total number of external trucks arriving at the gate system during time period t, and J be the total number of container yard exit areas. Let be the number of external trucks arriving at the i-th gate channel during time period t; I is the total number of gate channels.

[0040] The specific queuing constraints in the storage yard are as follows: The truck traffic arriving at container yard j is equal to the portion of the total outbound traffic at the gate that belongs to the reserved proportion for that container yard:

[0041] in, The number of external trucks arriving at container area j within time period t. This represents the total number of trucks leaving the entire gate system during time period t. This constraint dynamically allocates the gate departure flow to each container yard according to the proportion of reservations made by each container area to the total reservations.

[0042] The specific constraints on the location and operational status of the yard bridge are as follows: An arbitrary field bridge must and can only be located in one physical box region at any given time: ; The number of yard cranes in the same container area at any given time shall not exceed the physical limit: ; The yard crane can only operate in a container area when it is physically located there: ; The yard crane can only undertake loading tasks when it is in an effective operational state: ; The yard crane will not perform redundant operations beyond actual needs: ; in, This represents the maximum number of yard cranes that a single container area can accommodate at the same time. The variable is 0-1. The field bridge n is 1 when the box area is in an effective working state during time period t, and 0 otherwise. This represents the total amount of ship loading operations that need to be completed in container area j during time period t.

[0043] Specifically, the improved multi-objective evolutionary algorithm is based on the NSGA-II algorithm and is an improvement upon it, such as... Figure 3 As shown, specifically: The terminal operation data is classified into yard crane scheduling segments and container truck reservation quota segments, and the segments are encoded using a two-layer chromosome encoding method. The total demand heat of each container area in each time period is calculated based on the encoded fragments, and the calculation results are sorted. The initial solution of the static collaborative optimization model is solved by combining the sorting results. The initial solution is iterated, and the initial solution generated by the iteration is optimized by combining the field bridge resource correction and the reservation adaptive flow adjustment. The iteration stops when the termination condition is met, and the Pareto optimal solution set is obtained. The compromise solution is selected as the optimal solution according to the Pareto optimal solution set.

[0044] The double-layer chromosome coding structure in this embodiment is as follows: Figure 4 As shown, the yard bridge scheduling segment is composed of Composed of gene loci, location index Corresponding to the Odaiba Bridge in The state of the time period, the gene value is the physical box area number where the bridge is located, and the value range is [value range missing]. Integers. This encoding naturally satisfies the physical constraint that a single yard crane can only be located in one container area at a time, compressing the search space; the truck reservation quota segment adopts sparse compression encoding, and the algorithm pre-scans the operating parameters of all ships, extracting only the valid ones that satisfy the time window constraint and container area constraint. By combining and constructing gene loci, the remaining positions are automatically set to zero, effectively filtering out invalid decision domains, and making chromosome length positively correlated with ship operation density rather than planning cycle length.

[0045] In addition, the sum of the loading task volume and the expected truck arrival volume of each container area in each time period is calculated as the total demand heat. The container areas are sorted from high to low heat and the yard cranes are allocated to high heat container areas first until the physical limit of a single container area is reached or the total number of yard cranes is exhausted. This ensures that the resource allocation and demand distribution of the initial population are consistent with the macro trend and improves the quality of the initial solution.

[0046] Specifically, such as Figure 3 As shown, when solving the algorithm, the initialization sets parameters such as population size, number of iterations, crossover probability, and mutation probability. The demand-driven heuristic initialization is called to generate the initial population. The initial population is decoded and two objective function values ​​are calculated. Fast non-dominated sorting and crowding distance calculation are then performed.

[0047] Subsequently, a binary tournament selection method was used to screen parent individuals; a simulated binary crossover operator was applied and rounded to generate offspring; enhanced mutation of the yard crane operation area was performed to identify bottleneck container areas in service capacity and to forcibly move idle or low-load yard cranes to bottleneck container areas; a translation mutation of the truck reservation time window was used to randomly transfer part of the time slot quota within the effective port arrival time window; and the repair operator was invoked to project infeasible individuals back to the feasible domain.

[0048] Finally, the offspring and parents are merged into a mixed population of size 2N; fast non-dominated sorting and crowding distance calculation are performed; the top N individuals are selected to form the next generation according to the elite retention principle of non-dominated level priority and larger crowding distance in the same level priority; the process terminates after reaching the maximum number of iterations and outputs the Pareto optimal solution set.

[0049] Preferably, the offspring individuals generated by genetic operations may violate loading priority constraints or service capacity constraints. In this embodiment, a repair operator based on remaining capacity is used to optimize the initial solution of the iteration, including yard crane resource correction: traversing all container areas throughout the entire time period, if the current yard crane configuration capacity is insufficient to cover the loading demand, idle yard cranes are prioritized for use; if there are no idle resources, yard cranes are drawn from low-load container areas to supplement until the loading priority constraint is met; it also includes adaptive adjustment of reservation flow: the calculated remaining service capacity is the upper limit, each effective time window is scanned, if the reservation quota for a certain time period exceeds the remaining service capacity of the yard cranes in that time period, the quota for that time period is reduced to the upper limit, and the reduced quota is redistributed according to a greedy strategy to the time period with the most abundant remaining service capacity within the same vessel's port arrival time window, ensuring the conservation of the total port arrival volume.

[0050] Specifically, during implementation, the system monitors the actual arrival status of external container trucks in real time and compares the actual arrival time with the reserved time window in the baseline scheduling plan. When the actual arrival time of an external container truck exceeds the end time of the reserved time window, it is determined that the external container truck has missed its appointment, and the corresponding task is recorded as a missed task.

[0051] Furthermore, the system employs a rolling time-domain control method to dynamically monitor missed tasks. When the number of missed tasks reaches a preset trigger threshold, a reactive rescheduling model is triggered. When rescheduling is triggered, the existing work plan currently being executed by the yard crane is locked, and only missed tasks in future time periods are reassigned to reduce the impact of missed task disturbances on subsequent scheduling schemes.

[0052] Specifically, regarding the random disturbances caused by the failure of external container trucks to deliver containers during operations according to the baseline scheduling scheme, the reactive rescheduling model based on rolling time-domain control constructed in this embodiment can be expressed as follows:

[0053] This indicates minimizing the cumulative impact of missed task return on truck waiting time in future periods.

[0054] Preferably, the rescheduling model satisfies the following constraints: rescheduling time constraint, rescheduling flow constraint, rescheduling quantity constraint, and rescheduling state constraint.

[0055] The rescheduling time constraint is as follows: All trucks that have missed their scheduled departures must be reallocated within the future timeframe of the vessel's port cut-off window, and the sum of the newly added traffic from the reallocation must be exactly equal to the total number of missed departures. ; in, Let be an integer variable, representing the number of vessels z trucks that will be rescheduled to a future time period t if they miss their appointment at time period t′ and their destination is container area j. The number of non-compliant container trucks belonging to vessel z in container area j during time period t′.

[0056] The specific constraints on rescheduling traffic are as follows: The total flow arriving at the gate in each time period in the future will be the sum of the original reservation volume and the newly added volume due to rescheduling: ; The specific constraints on the number of rescheduled events are as follows: The total number of external trucks leaving the entire gate system during time period t is the sum of the departure volumes from each channel: ; in, The reservation quota (with known parameters) for box area j in time period t in the original static optimal plan. Let t be the set of future reservation time slots, where t∈{t′+1,t′+2,...,T}.

[0057] The rescheduling state constraints can be specifically expressed as:

[0058] The above formula describes the available service capacity of yard cranes for external container trucks in the future period, calculated based on a locked yard crane assignment plan, with the portion occupied by loading vessels remaining unchanged. Among them, , These are the bridge assignment and ship loading reserve capacities in the original static optimal plan, respectively. During the rescheduling process, these are locked as known parameters and do not participate in the rescheduling decision.

[0059]

[0060]

[0061]

[0062] The above formula describes the time-by-time state evolution of the gate in the future time period after rescheduling, reflecting the Markov property of the rolling time domain.

[0063]

[0064] The above formula describes the dynamic allocation process of traffic flow from the gate to each container yard area after rescheduling. The allocation ratio is dynamically adjusted according to the updated total reservation volume, while taking into account the original planned reservation volume and the new volume added by rescheduling. This refers to the reservation quota for box area j in time period t in the original static optimal plan. This indicates the number of vessels and trucks (z) whose destination (container area j) are rescheduled to a future time period (t) after the missed time period (t′).

[0065]

[0066]

[0067] The above formula describes the time-by-time queuing state evolution of each container area in the yard after rescheduling. When t=t′+1, the initial queue length of the yard is determined by the real-time state parameters. Given, where, The actual queue length of container area j, obtained from real-time monitoring at the end of time period t′, serves as the initial boundary condition for rescheduling calculation.

[0068] Preferably, such as Figure 5 As shown, a heuristic greedy insertion algorithm is used to quickly solve the reactive rescheduling model. Specifically, for each batch of missed trucks belonging to ship z, with destination j and quantity S, feasible region screening, congestion assessment, optimal insertion and state update are performed in sequence. Among them, the feasible region screening determines the set of future time periods that the batch of trucks can be inserted into based on the port cut-off time window of vessel z. Crowding assessment For each candidate time period t, calculate the real-time congestion index of the target container area j, which is the ratio of the current total workload to the maximum service capacity of the yard crane, and calculate the remaining available container capacity, which is the remaining service capacity of the yard crane minus the existing reservations. The candidate time periods are ranked based primarily on the remaining available container capacity and secondarily on the congestion index. Time periods with a remaining available container capacity not less than the number of missed appointments S are prioritized to ensure that insertion does not cause new congestion. If no time periods meet the criteria, the time periods with the lowest congestion index are selected and inserted in batches. Status update: The number of missed appointments is added to the reservations of the selected time period, and the congestion index for that time period is updated. The congestion assessment, selection, and status update are repeated until all missed appointments in that batch are redistributed.

[0069] Specifically, an experiment was conducted using this method, with a typical scenario including a planning period of T=36 time periods (2 hours per time period, totaling 72 hours), 9 vessels arriving at the port (each vessel's arrival demand is 600~2000 TEU), 7 batches of loading tasks, 14 yard cranes, 6 container yard export areas, and 4 gate channels. The arrival time windows of each vessel and the loading task time windows are interspersed within the planning period to simulate the complex working condition of multiple vessels arriving at the port and multiple vessels loading in parallel.

[0070] Among them, the following table and Figure 6 As shown in the comparative experiment, the improved NSGA-II algorithm of this invention is compared with the standard NSGA-II and SPEA2 algorithms. The normalized hypervolume index HV is used to measure the solution set quality. The results show that the final convergence HV value of the improved algorithm is about 0.99, which is better than 0.92 of the standard NSGA-II and 0.91 of SPEA2. The Wilcoxon rank-sum test p-values ​​are all less than 0.01, and the performance improvement is statistically significant.

[0071] Table 1 Comparison of Solution Algorithm Performance

[0072] Furthermore, this invention incorporates the dynamic remaining service capacity of yard cranes into the reservation constraints, achieving dynamic matching between reservation quotas and the actual available capacity of yard cranes. In a case study based on a domestic container terminal, with a planning cycle of 3 days and involving 14 yard cranes and approximately 17,000 container operations, the maximum waiting time for a single external truck at the gate and yard is approximately 7.5 minutes. Figure 7 As shown, the queuing congestion in the yard caused by the occupancy of yard crane service capacity during peak loading periods was effectively suppressed, verifying the feasibility of the collaborative optimization scheme during periods of high loading activity.

[0073] Meanwhile, taking the 10th time period (5% no-show rate, 20 no-show trucks) as a typical disturbance scenario, the no-show vehicles were distributed to the 11th to 14th time periods by the system, with a 100% rescheduling success rate. Figure 8 As shown, the impact of the disturbance was concentrated in seven time periods from t=11 to t=17, with a peak waiting time of 8.5 minutes. The waiting time returned to the baseline level at t=17, and no secondary congestion occurred throughout the entire period.

[0074] Finally, the rolling time-domain rescheduling strategy proposed in this invention dynamically reallocates delayed tasks without changing the original yard crane operation plan, such as... Figure 9 As shown, under three no-show rates of 5%, 10%, and 15%, the average waiting time per vehicle in the rescheduling scenario increased from 4.40 minutes to 5.02 minutes, significantly better than the 4.89 minutes to 6.72 minutes in the non-rescheduling scenario. The difference in peak waiting time is even more pronounced. At a no-show rate of 15%, the peak waiting time with rescheduling is 15.31 minutes, while the peak waiting time without rescheduling reaches as high as 30.74 minutes, representing a peak improvement rate of approximately 50%.

[0075] Furthermore, embodiments of the present invention also provide a collaborative optimization system for reservation decision-making and yard bridge scheduling considering no-show disturbances, used to implement the collaborative optimization method for reservation decision-making and yard bridge scheduling considering no-show disturbances of the present invention, such as... Figure 10 As shown, the system includes: The benchmark scheduling solution module is used to collect terminal and operation data and preprocess it. The processed terminal and operation data is then input into the static collaborative optimization model of external truck reservation and yard crane scheduling, and the static collaborative optimization model is solved to obtain the benchmark scheduling solution. The missed task optimization module is used to execute external truck operation tasks according to the baseline scheduling scheme, monitor the missed task data of external trucks in real time, input the missed task data into the rescheduling model, adjust and optimize the baseline scheduling scheme in real time, and enable external trucks to perform operations according to the optimized scheduling scheme. Among them, the static collaborative optimization model of external truck reservation and yard bridge scheduling is established with minimizing yard bridge operating costs and cumulative queuing time as the objective function, and the optimal solution set of the scheduling scheme is solved, from which the compromise solution is selected as the benchmark scheduling scheme; the rescheduling model is specifically established with minimizing the cumulative impact of missed task return on truck waiting time in future periods as the objective function, and the optimal solution at this time is solved as the optimized scheduling scheme.

[0076] The various modules or mechanisms of the system are mainly used to implement the various steps of the above method embodiments, and will not be described in detail here.

[0077] Finally, this embodiment of the invention also provides a computer storage medium storing a computer program that can be executed by a processor, the computer program executing the aforementioned method for coordinated optimization of reservation decision and yard bridge scheduling.

[0078] This embodiment presents a method and system for collaborative optimization of reservation decision-making and yard crane scheduling, considering the impact of missed appointments. It proposes a dynamic remaining service capacity constraint for yard cranes, incorporating loading operations into the truck reservation decision-making system. Existing truck reservation systems treat yard crane service capacity as a fixed constant, ignoring the dynamic occupancy of loading operations. This invention uses the dynamic remaining service capacity of yard cranes in any container area at any time period as the upper limit of the truck reservation quota constraint, breaking down the decision-making barriers between the reservation system and the yard crane scheduling system, and eliminating secondary congestion caused by trucks in the yard outside of peak loading periods. Furthermore, it proposes a two-layer decision-making method combining static collaborative optimization and rolling time-domain rescheduling: the static layer improves the NSGA-II collaborative optimization of truck reservation quotas and yard crane location assignment, outputting a Pareto optimal baseline plan; the dynamic layer, based on rolling time-domain control, monitors the performance status in real time. When a missed appointment is triggered, it uses a greedy insertion strategy to roll-redistribute the missed task while locking the original yard crane assignment scheme, combining this with existing single-time static... Compared to triggering mechanisms, this method can continuously respond to random cancellations throughout the entire planning period, effectively preventing error accumulation. Finally, a two-segment hybrid integer coding and demand-driven repair mechanism for collaborative scheduling problems is proposed: to address the difficulty of coupled solution of discrete location decision-making for yard bridges and continuous flow decision-making for trucks, a two-segment hybrid integer chromosome structure is designed. The yard bridge scheduling segment coding naturally satisfies the unique location constraint of a single yard bridge; the truck reservation segment adopts sparse compression coding, extracting only the effective decision domain gene bits, compressing the search space, and combined with demand-driven heuristic initialization and a two-stage repair operator based on remaining capacity, significantly improving the feasible solution ratio and optimization efficiency of the algorithm under complex constraints.

[0079] It should be noted that, depending on the implementation needs, the various steps / components described in this application can be broken down into more steps / components, or two or more steps / components or parts of the operation of steps / components can be combined into new steps / components to achieve the purpose of this invention.

[0080] The order of the steps in the above embodiments does not imply the order of execution. The execution order of each process should be determined by its function and internal logic, and should not constitute any limitation on the implementation process of the embodiments of this application.

[0081] It should be understood that those skilled in the art can make improvements or modifications based on the above description, and all such improvements and modifications should fall within the protection scope of the appended claims.

Claims

1. A collaborative optimization method for reservation decision-making and yard bridge scheduling considering no-show disturbances, characterized in that, The method includes: Collect and preprocess terminal and operation data, input the processed terminal and operation data into the static collaborative optimization model of external truck reservation and yard crane scheduling, and solve the static collaborative optimization model to obtain the baseline scheduling scheme. The external container truck operation tasks are executed according to the baseline scheduling plan, and the data of missed tasks by the external container trucks are monitored in real time. The missed task data is input into the rescheduling model to adjust and optimize the baseline scheduling plan in real time, and the external container trucks are operated according to the optimized scheduling plan. Among them, the static collaborative optimization model is established with minimizing the operating cost of the yard bridge and the cumulative queuing time as the objective function, and solves the optimal solution set of the scheduling scheme, from which the compromise solution is selected as the benchmark scheduling scheme; the rescheduling model is specifically established with minimizing the cumulative impact of the return of missed tasks on the waiting time of trucks in future periods as the objective function, and solves the optimal solution at this time as the optimized scheduling scheme.

2. The collaborative optimization method for reservation decision-making and yard bridge scheduling considering no-show disturbances as described in claim 1, characterized in that, Terminal and operational data include vessel berthing plans and port arrival time windows, total container loading demand and target container areas for each vessel, loading volume for each container area at each time period, number and service efficiency of gate channels, number and service efficiency of available yard bridges, and number and spatial layout parameters of container yard areas.

3. The collaborative optimization method for reservation decision-making and yard bridge scheduling considering no-show disturbances as described in claim 2, characterized in that, Preprocessing of terminal and operational data includes: dividing the process of external trucks entering the port into a two-level serial queuing network consisting of a gate system and a yard system, and using a point-by-point fixed fluid approximation method to transform the non-stationary queuing process of external trucks into time-by-time algebraic constraints; the non-stationary queuing process specifically refers to the arrival flow of external trucks after leaving the gate and flowing through the port road network to form the yard.

4. The collaborative optimization method for reservation decision-making and yard bridge scheduling considering no-show disturbances as described in claim 1, characterized in that, The static collaborative optimization model for external truck reservation and yard crane scheduling satisfies the following constraints: reservation and loading task constraints, loading priority and remaining service capacity of yard cranes constraints, gate queuing constraints, yard queuing constraints, and yard crane location and operation status constraints.

5. The collaborative optimization method for reservation decision-making and yard bridge scheduling considering no-show disturbances as described in claim 3, characterized in that, The static collaborative optimization model is solved using an improved multi-objective evolutionary algorithm, which is specifically an improvement on the NSGA-II algorithm, as follows: The terminal and operational data are classified into yard crane scheduling segments and truck reservation quota segments, and the segments are coded in two layers. The total demand heat of each container area in each time period is calculated based on the encoded fragments, and the calculation results are sorted. The initial solution of the static collaborative optimization model is solved by combining the sorting results. The initial solution is iterated, and the initial solution generated by the iteration is optimized by combining the field bridge resource correction and the reservation adaptive flow adjustment. The iteration stops when the termination condition is met, and the Pareto optimal solution set is obtained. The compromise solution is selected as the optimal solution according to the Pareto optimal solution set.

6. The collaborative optimization method for reservation decision-making and yard bridge scheduling considering no-show disturbances as described in claim 1, characterized in that, The rescheduling model satisfies the following constraints: Rescheduling time constraints, rescheduling flow constraints, rescheduling quantity constraints, and rescheduling state constraints.

7. The collaborative optimization method for reservation decision-making and yard bridge scheduling considering no-show disturbances as described in claim 1, characterized in that, The rescheduling model is solved using a heuristic greedy insertion algorithm, which involves: filtering the feasible region of the vessels to which the vacant trucks belong to determine the set of future time periods into which the vacant trucks are allowed to be inserted; evaluating and sorting the congestion of each time period in the future time period set, and selectively inserting the vacant trucks; re-evaluating the congestion of the inserted time periods until all vacant trucks have been reassigned.

8. The collaborative optimization method for reservation decision-making and yard bridge scheduling considering no-show disturbances as described in claim 1, characterized in that, The operating cost of the yard crane is calculated based on the yard crane relocation cost, cross-regional transfer cost, and empty consumption cost; the cumulative queuing time is calculated based on the queue length of the yard and the queue length of the gate channel; the cumulative impact of the return of missed tasks on the waiting time of container trucks in future periods is calculated based on the cumulative queuing time of external container trucks in future periods.

9. A collaborative optimization system for reservation decision-making and yard bridge scheduling considering no-show disturbances, characterized in that, The system includes: The benchmark scheduling solution module is used to collect terminal and operation data and preprocess it. The processed terminal and operation data is then input into the static collaborative optimization model of external truck reservation and yard crane scheduling, and the static collaborative optimization model is solved to obtain the benchmark scheduling solution. The missed task optimization module is used to execute external truck operation tasks according to the baseline scheduling scheme, monitor the missed task data of external trucks in real time, input the missed task data into the rescheduling model, adjust and optimize the baseline scheduling scheme in real time, and enable external trucks to perform operations according to the optimized scheduling scheme. Among them, the static collaborative optimization model of external truck reservation and yard bridge scheduling is established with minimizing yard bridge operating costs and cumulative queuing time as the objective function, and the optimal solution set of the scheduling scheme is solved, from which the compromise solution is selected as the benchmark scheduling scheme; the rescheduling model is specifically established with minimizing the cumulative impact of missed task return on truck waiting time in future periods as the objective function, and the optimal solution at this time is solved as the optimized scheduling scheme.

10. A computer storage medium, characterized in that, It contains a computer program that can be executed by a processor, which performs the reservation decision and yard bridge scheduling collaborative optimization method as described in any one of claims 1-8.

Images