Flood season ship dispatching optimization method and system based on channel-ship lock coordination

Abstract

The invention discloses a flood season ship scheduling optimization method and system based on channel-ship lock coordination, and belongs to the technical field of water transportation traffic intelligent management, and the method comprises the steps: obtaining ship information, ship lock attribute information and flood season channel navigation requirement information of a water transportation junction; taking minimization of the total waiting time of the ship in the anchorage ground as an optimization target, constructing a ship scheduling optimization model, and establishing constraint conditions satisfied by ship lockage scheduling in the flood season; solving the ship scheduling optimization model based on a dynamic multi-population particle swarm optimization algorithm of recombination learning and mixed mutation to obtain an optimized flood season ship scheduling scheme; estimating the carbon emission of the ship, and verifying the low carbon and sustainability of the ship scheduling scheme in the flood season. According to the method, the ship off-anchor time and channel-ship lock coordination decision problem can be solved, the defect of shortage of a flood season scheduling method is overcome, and an effective reference is provided for reducing the ship navigation risk and improving the ship scheduling efficiency.

Description

Technical Field

[0001] The present invention belongs to the technical field of intelligent management of water transport, and more specifically, relates to a method and system for optimizing ship scheduling during flood season based on waterway-lock coordination. Background Art

[0002] Inland waterway transportation is an important way to connect coastal ports with inland areas, and has an important impact on economic growth and international trade development in river basins. In recent years, the role of inland waterway transportation in economic development has become increasingly obvious, which has led to the continuous expansion of inland waterway transportation such as the Yangtze River, resulting in a surge in demand for ship crossing dams. However, the capacity of inland waterways and locks is limited, which has become the main bottleneck restricting the development of Yangtze River waterway transportation. In addition, the ship locks and waterway scheduling of most water conservancy hubs are independently decided, which further limits the improvement of ship navigation efficiency. In view of this, the development of an efficient flood season ship scheduling optimization method and system can reduce the waiting time of ships while ensuring the safety of ship navigation by coordinating waterway and lock resources, which has important practical significance and application value.

[0003] At present, many scholars have studied the ship lock scheduling and waterway allocation problems, but there is a lack of research on the coordinated scheduling methods between ship locks and waterways. In addition, the research on ship lock scheduling and waterway allocation problems mainly focuses on optimization under normal conditions, such as good weather conditions and normal water level conditions. However, the change of water flow in the Three Gorges hub has an important impact on ship lock scheduling and waterway configuration, especially during the flood season, when the change of water flow will affect the departure time of ships, waterway allocation, lock chamber utilization and carbon emissions of ships. Therefore, how to fully expand the ship scheduling under normal conditions into the coordinated scheduling of waterways and ship locks under complex water flow conditions during the flood season is more realistic and innovative. Summary of the invention

[0004] In view of the above defects or improvement needs of the prior art, the present invention proposes a flood season ship scheduling optimization method and system based on waterway-lock coordination, which can better ensure the navigation safety of ships in the flood season and improve the navigation efficiency of ships.

[0005] To achieve the above object, according to one aspect of the present invention, a method for optimizing ship scheduling during flood season based on waterway-lock coordination is provided, comprising:

[0006] Obtain information on ships, lock properties and navigation requirements for flood season at water transport hubs;

[0007] Based on the ship information, lock attribute information and flood season navigation requirements of the water transport hub, a ship scheduling optimization model is constructed with the optimization goal of minimizing the total waiting time of ships at anchorage, and the constraint conditions satisfied by the ship lock scheduling during the flood season are established;

[0008] A dynamic multi-population particle swarm optimization algorithm based on recombination learning and hybrid mutation is used to solve the ship scheduling optimization model and obtain an optimized flood season ship scheduling plan;

[0009] Based on the optimized flood season ship scheduling plan, the carbon emissions of ships are estimated.

[0010] In some optional embodiments, the ship information includes: ship's arrival time, sailing speed, ship length, ship area, and the maximum water flow standard acceptable to the ship;

[0011] The lock attribute information includes: scheduling cycle, lock area, lock chamber capacity;

[0012] The flood season waterway navigation requirement information includes: when a ship approaches a lock, in order to ensure the safety of waterway passage, it is necessary to ensure that there is a time interval between two adjacent ships; when a ship is sailing, the current water flow standard of the waterway does not exceed the maximum acceptable value of the ship.

[0013] In some optional implementation schemes, the optimization objective is to minimize the total waiting time of ships at anchorages, construct a ship scheduling optimization model, and establish constraints that must be met for ship lock scheduling during flood seasons, including:

[0014] Calculate the total waiting time of the vessel at the anchorage;

[0015] According to the calculated total waiting time of the ship at the anchorage, the constraint conditions satisfied by the ship lock-passing scheduling during the flood season are constructed.

[0016] In some optional implementations, MinZ = ∑ i∈N (t bi -t ai ) calculates the total waiting time of the ship, where Z is the total waiting time of the ship, t ai is the time when the ship arrives at the waiting area, t bi is the time when the ship leaves the waiting area, and N represents the number of ships to be dispatched.

[0017] In some optional implementations, the constraints include:

[0018] Constraint 1: Among them, t s is the start time of the scheduling period;

[0019] Constraint 2: Among them, y ik ∈{0,1}, if the ship is transferred to the lock, then y ik =1, otherwise y ik =0, K represents the set of ship lock times, and M represents a constant;

[0020] Constraint 3: Among them, l i is the distance that the ship needs to pass through the channel to reach the lock, v i is the ship's speed through the channel, t ci The time when the ship leaves the channel and approaches the lock;

[0021] Constraint 4: Constraint 5: N; where the minimum time interval between two consecutive ships approaching the dam is △t, γ is a multiple of the ship length, and L i is the length of the ship;

[0022] Constraint 6: Among them, T k is the time required for a single gate, t di The time when the vessel leaves the lock;

[0023] Constraint 7: Among them, t e The end time of the scheduling period;

[0024] Constraint 8: z k If the gate is used, it is equal to 1, otherwise it is 0;

[0025] Constraint 9: Constraint 10: Among them, T l is the upper limit of the lock time, t k The start time of the ship lock;

[0026] Constraint 11: Constraint 12: Among them, σ represents the minimum time interval between consecutive gate times;

[0027] Constraint 13: Among them, z k ∈{0,1}, if the ship lock starts, then z k =1, otherwise z k =0;

[0028] Constraint 14: Among them, ξ it ∈{0,1}, if the ship enters the channel within the time step, then ξ it =1, otherwise ξ it =0, T is the set of time steps;

[0029] Constraint 15:

[0030] Constraint 16:

[0031] Constraint 17:

[0032] Among them, η ikt ∈{0,1}, if the ship enters the channel and passes through the lock within the time step, then η ikt =1, otherwise η ikt =0;

[0033] Constraint 18: Among them, f i The acceptable water flow standard for ships is is the actual water flow value of the channel within the time step;

[0034] Constraint 19: Among them, a i is the ship area, A k is the lock chamber area;

[0035] Constraint 20: Among them, q t Indicates the upper limit of the channel capacity within the time step.

[0036] In some optional implementation schemes, the dynamic multi-population particle swarm optimization algorithm based on recombination learning and hybrid mutation solves the ship scheduling optimization model to obtain an optimized flood season ship scheduling plan, including:

[0037] Determine the individual dimension N, where N represents the number of ships to be dispatched, the vector represents the departure time and order of ships in the waiting area, and the initial vector range [t s ,t e ], generating an initial solution based on a large language model, t s is the time when the scheduling period starts, t e The time when the scheduling period ends;

[0038] The total waiting time of the ship is taken as the fitness value, and the median of the fitness value is used as the boundary to divide the population into three sub-populations: the neighborhood domain, the alienation domain, and the recombination domain. Multiple populations search in different areas to determine the optimal ship scheduling plan, and within the specified time interval, a dynamic random recombination mechanism is used to continue the search process;

[0039] Reconstructed particles are generated by dynamically adjusting and mixing the reverse learning method and Gaussian mutation method, and the subspace information of the historical optimal position is retained when the particle velocity is updated.

[0040] In some optional implementation schemes, generating an initial solution based on a large language model includes:

[0041] In view of the ship dispatch problem during the flood season, the ship mission data is obtained, and the training samples are divided according to the ship mission data to obtain the training set, validation set and test set;

[0042] Select a large language model based on Transformer to understand natural language text, process structured data, and generate flood season ship scheduling plans;

[0043] Input the training set data into the model, generate the predicted flood season ship dispatch plan through forward propagation, calculate the loss function between the predicted and actual results, update the model parameters through the back propagation algorithm, and iterate the training until the model achieves satisfactory performance on the validation set;

[0044] Based on the meta-optimizer, the model with a passing validation result is fine-tuned using the ship mission data;

[0045] The performance of the model is evaluated using the test set. Once the evaluation passes, the model is deployed to actual applications to generate enough initial solutions to form the initial population.

[0046] In some optional embodiments, the generating of reconstructed particles by dynamically adjusting and mixing the reverse learning method and the Gaussian mutation method comprises:

[0047] Depend on Determine the number of dimensions that need to be mutated, where dim is the dimension to be mutated, D is the dimension, J is the number of iterations, and G is the number of iterations. best(d) is the optimal particle, G best(d) ′ is the new optimal particle generated, and fit() is the fitness function of the corresponding particle;

[0048] Determine whether the time-varying probability TP is less than the reverse learning probability OP. When TP < OP, perform the reverse learning method on the global optimal solution. When TP ≥ OP, perform the Gaussian mutation method on the global optimal solution.

[0049] After executing the reverse learning method, determine whether the dynamic boundary has been exceeded. If G best(d) <min(G best ) or G best(d) >max(G best ), randomize G best ,min(G best ) is the minimum value of the optimal particle, max(G best ) is the maximum value of the optimal particle;

[0050] Compare the fitness values ​​of the new individual with the original individual and output the reconstructed particles.

[0051] In some optional implementation schemes, estimating ship carbon emissions based on the optimized flood season ship scheduling plan includes:

[0052] Depend on Ship emission reduction = (T start -Z)·e estimates the carbon emissions of ships, where v i is the ship speed, w i is the ship load, ψ i is the weight of the ship, is the carbon emission coefficient, α, β, q are adjustment parameters, parameter α adjusts the ship's related characteristics and loading conditions, β adjusts the ship's fuel consumption at low speed, q is the correlation coefficient between fuel consumption and speed, T start is the total waiting time of the original ship scheduling plan, is the fuel consumption during the ship's voyage. is the fuel consumption of the ship during berthing at the anchorage, and e is the carbon emission per unit time.

[0053] According to another aspect of the present invention, a flood season ship scheduling optimization system based on waterway-lock coordination is provided, comprising:

[0054] The acquisition module is used to obtain the ship information, lock attribute information and navigation requirements of the waterway during the flood season at the water transport hub;

[0055] A construction module is used to construct a ship scheduling optimization model based on the ship information, lock attribute information and flood season waterway navigation requirement information of the water transport hub, with minimizing the total waiting time of ships at anchorage as the optimization goal, and establish constraint conditions satisfied by ship lock scheduling during the flood season;

[0056] An optimization solution module is used to solve the ship scheduling optimization model based on a dynamic multi-population particle swarm optimization algorithm based on recombination learning and hybrid mutation to obtain an optimized flood season ship scheduling plan;

[0057] A verification sustainability module is used to estimate the carbon emissions of ships based on the optimized flood season ship scheduling plan.

[0058] In general, the above technical solutions conceived by the present invention can achieve the following beneficial effects compared with the prior art:

[0059] The present invention provides a method and system for optimizing ship scheduling during flood season based on waterway-lock coordination, which innovatively expands the existing research on lock scheduling and establishes a ship scheduling optimization model based on lock-waterway coordination. It comprehensively considers factors such as the optimization of the time when the ship leaves the anchorage, the change of water flow in the waterway, and the coordinated configuration of the waterway and the lock. It establishes a ship scheduling optimization model with the minimum total waiting time for ships, promotes the coordination of waterways and locks in different scenarios, ensures the safety of ship navigation, and improves the efficiency of ship scheduling.

[0060] At the same time, the model proposed in this paper estimates the carbon emission reduction of ship scheduling schemes under different scenarios. By determining the optimal departure time for ships to leave the anchorage, effectively matching the ship channel operation time and the lock activation time, ensuring that the lock is available when the ship leaves the channel, effectively coordinating the lock and the channel, and reducing the delay time and carbon emissions of ships at anchorage. In addition, compared with normal conditions, considering the water flow during the flood season can ensure the navigation safety of ships and increase the feasibility of ship scheduling schemes. BRIEF DESCRIPTION OF THE DRAWINGS

[0061] Figure 1 It is a flow chart of steps of a flood season ship scheduling optimization method based on waterway-lock coordination proposed by the present invention;

[0062] Figure 2 It is a flow chart of the algorithm proposed by the present invention;

[0063] Figure 3 is a schematic diagram of reconstructed particle generation proposed by the present invention;

[0064] Figure 4 1 is a comparison diagram of different algorithms for solving the ship scheduling problem provided by the embodiments of the present invention, wherein (a) is a normal situation and (b) is a situation affected by floods;

[0065] Figure 5 It is a schematic diagram of carbon emission reduction rate under different situations provided by an embodiment of the present invention. DETAILED DESCRIPTION

[0066] In order to make the purpose, technical solutions and advantages of the present invention more clearly understood, the present invention is further described in detail below in conjunction with the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are only used to explain the present invention and are not intended to limit the present invention. In addition, the technical features involved in the various embodiments of the present invention described below can be combined with each other as long as they do not conflict with each other.

[0067] The technical solution of the present invention is: a method and system for optimizing ship scheduling during flood season based on waterway-lock coordination, such as Figure 1 As shown, the specific steps include:

[0068] Step S1: Obtaining ship information, dispatch information, and navigation requirements during flood season at the water transport hub;

[0069] In the embodiment of the present invention, the ship information includes: ship anchoring time, sailing speed, ship length, ship area, ship water flow standard and actual channel water flow;

[0070] The lock attribute information includes: scheduling cycle, lock area, and lock capacity;

[0071] The requirements for navigation during the flood season include: when ships pass through the lock, in order to ensure the safety of passage, there must be a time interval between adjacent ships; when the ship is sailing, the current water flow in the channel does not exceed the maximum acceptable value of the ship. For example, the safety interval of ships on the channel is set to 6 ship lengths, which is equal to 10 minutes.

[0072] Step S2: Taking minimizing the total waiting time of ships at the anchorage as the optimization goal, a ship scheduling optimization model is constructed to establish the constraint conditions that must be met for ship scheduling during flood season;

[0073] It should be noted that the optimization objective of minimizing the total waiting time of ships at anchorages is to construct a ship scheduling optimization model including:

[0074] Step S21: Calculate the total waiting time of the ship at the anchorage, the calculation formula is as follows:

[0075]

[0076] In the above formula, Z is the total waiting time of the ship at the anchorage, t ai is the time when the ship arrives at the anchorage, t bi is the time when the ship leaves the anchorage, and N represents a group of ships to be dispatched;

[0077] Step S22: According to the calculated total waiting time of the ship at the anchorage, the constraint conditions satisfied by the ship lock scheduling during the flood season are constructed, and the constraint conditions are as follows:

[0078] Constraint 1:

[0079] Constraint 1 means that each ship leaves the anchorage within the scheduling period, where t s is the start time of the scheduling period;

[0080] Constraint 2:

[0081] Constraint 2 states that each ship must be scheduled after its arrival. If the ship is assigned to a channel, the lock must be available, where y ik ∈{0,1}, if the ship is transferred to the lock, then y ik =1, otherwise yik =0, K and M represent the set of ship lock times and a sufficiently large constant respectively;

[0082] Constraint 3:

[0083] Constraint 3 defines the time when the ship approaches the lock, where l i is the distance that the ship needs to pass through the channel to reach the lock, v i is the ship's speed through the channel, t ci The time when the vessel leaves the channel and approaches the lock;

[0084] Constraint 4:

[0085] Constraint 5:

[0086] Constraints 4 and 5 define the minimum time interval between two consecutive ships approaching the dam as △t, where γ is a multiple of the ship length, L i is the length of the ship;

[0087] Constraint 6:

[0088] Constraint 6 defines the time when the ship leaves the lock, where T k is the time required for a single gate, t di The time when the vessel leaves the lock;

[0089] Constraint 7:

[0090] Constraint 7 ensures that each ship leaves the lock within the scheduling period, where t e The end time of the scheduling period;

[0091] Constraint 8:

[0092] Constraint 8 means that the ship must be transferred through the lock, z k To indicate that the gate is used, it is equal to 1, otherwise it is 0;

[0093] Constraint 9:

[0094] Constraint 10:

[0095] Constraints 9 and 10 ensure that the ship approaching the lock time is equal to the lock opening time, where T l is the upper limit of the lock time, t k The start time of the ship lock;

[0096] Constraint 11:

[0097] Constraint 12:

[0098] Constraints 11 and 12 represent the minimum time interval between the start times of two consecutive locks, and a new lock cannot be started before the ship approaches the lock, where σ represents the minimum time interval between consecutive locks;

[0099] Constraint 13:

[0100] Constraint 13 indicates that when a ship approaches a lock, the lock must perform a pass operation, where z k ∈{0,1}, if the ship lock starts, then z k =1, otherwise z k =0;

[0101] Constraint 14:

[0102] Constraint 14 represents the relationship between the time when the ship leaves the anchorage and the time when it leaves the channel and approaches the lock, where ξ it ∈{0,1}, if the ship enters the channel within the time step, then ξ it =1, otherwise ξ it =0, T is the set of time steps;

[0103] Constraint 15:

[0104] Constraint 15 means that when a ship enters the channel within the time step, the corresponding lock is valid;

[0105] Constraint 16:

[0106] Constraint 17:

[0107] Constraints 16 and 17 indicate whether the ship passes through the dam within the lock time step, where η ikt ∈{0,1}, if the ship enters the channel and passes through the lock within the time step, it is equal to η ikt =1, otherwise η ikt =0;

[0108] Constraint 18:

[0109] Constraint 18 indicates that the current water flow in the channel does not exceed the maximum acceptable value of the ship, and different types of ships have different water flow standards, where f i is the acceptable water flow rate for the ship, is the actual water flow in the waterway within the time step;

[0110] Constraint 19:

[0111] Constraint 19 states that the ship passing through the lock must not exceed the lock chamber area, where a i is the ship area, A k is the lock chamber area;

[0112] Constraint 20:

[0113] Constraint 20 states that the number of ships cannot exceed the channel capacity in the time step, where q t Indicates the upper limit of the channel capacity within the time step.

[0114] Step S3: solving the ship scheduling optimization model based on dynamic multi-population particle swarm optimization algorithm (DMPSO-RH) based on recombination learning and hybrid mutation to obtain an optimized flood season ship scheduling plan;

[0115] It should be noted that the dynamic multi-population particle swarm optimization algorithm (DMPSO-RH) based on recombination learning and hybrid mutation is used to solve the ship scheduling optimization model, including: population initialization, dynamic multi-population division strategy, probability-based hybrid mutation (MP) strategy, particle reconstruction and speed update strategy.

[0116] Step S31: Population initialization. Determine the individual dimension N, i.e., the number of ships, the vector representing the departure time and order of ships at the anchorage, and the initial vector range [t s ,t e ], generate initial solution based on large language model, t a is the time when the scheduling cycle starts, t e The time when the scheduling cycle ends;

[0117] Step S32: Dynamic multi-population division strategy. Using the median of the fitness value (i.e., the total waiting time of the ship) as the boundary, the population is divided into three sub-populations: the neighborhood domain, the alienation domain, and the recombination domain. Multiple populations search in different areas to determine the optimal ship scheduling solution; within the specified time interval, a dynamic random recombination mechanism (i.e., the movement strategy of the best individual) is used to continue the search process, improve the efficiency of information exchange between individuals, and prevent the algorithm from converging to a suboptimal solution;

[0118] Step S33: Probability-based hybrid mutation (MP) strategy. By dynamically adjusting and mixing the OBL strategy and Gaussian mutation (GM) strategy, the algorithm's ability to jump out of local minima is improved;

[0119] Step S34: Particle reconstruction and velocity update strategy. The introduction of reconstructed particles preserves the subspace information of the historical optimal position when updating the particle velocity, which helps to track elite learning and increase the diversity of the particle swarm from a multi-dimensional perspective.

[0120] Among them, the velocity update formula of the reconstructed particles is as follows:

[0121]

[0122] In the above formula, c1=1, c2=1.2, c3=0.3, To reconstruct particles during iteration, each dimension is randomly selected from a dimension corresponding to the current historical optimal position of each particle to ensure that multiple dimensions of the same particle will not be selected repeatedly. Figure 3 As shown, is the velocity of the dth dimension of the ith particle in the jth generation, ω is the inertia weight, r1 is a random number, is the position of the d-th dimension of the ith particle in the j-th generation, is the population individual of the d-dimensional i-th particle in the j-th generation, r2 is a random number, is the global extreme value of the dth dimension of the gth particle in the jth generation, and r3 is a random number.

[0123] Furthermore, the step S31 of generating an initial solution based on the large language model includes the following steps:

[0124] Step S311: Data collection: For the ship dispatch problem during the flood season, obtain ship mission data, such as the number of ships, average speed of ships, displacement, earliest and latest departure time of ships, etc., divide the training samples into training set, validation set and test set;

[0125] Step S312: Model architecture selection: select a large language model based on Transformer to understand natural language text, process structured data, and generate flood season ship scheduling plans;

[0126] Step S313: Model training: input the training set data into the model, generate the predicted flood season ship dispatch plan through forward propagation, calculate the loss function between the predicted and actual results, update the model parameters through the back propagation algorithm, and iterate the training until the model achieves satisfactory performance on the validation set;

[0127] Step S314: fine-tuning and iterative optimization: Based on the meta-optimizer, the model with a passed verification result is fine-tuned using the ship mission data to improve its accuracy on the task;

[0128] Step S315: Model evaluation and deployment: Use the test set to evaluate the performance of the model. When the evaluation passes, the model is deployed to actual applications to generate enough initial solutions to form the initial population.

[0129] Furthermore, the three subpopulations described in step S32, the neighborhood domain, the alienation domain, and the recombination domain, have the following specific meanings:

[0130] Neighborhood domain: It is composed of particles with smaller fitness values. This particle group inherits the location information of the optimal solution in the population for more detailed search;

[0131] Alienation domain: It is composed of particles close to the global extreme value, exploring new optimal solutions in the surrounding area and enhancing mining capabilities;

[0132] Reorganization domain: Dynamically adjust group diversity by absorbing individual experience and sharing global information.

[0133] Further, in step S33, the probability-based mixed mutation (MP) strategy process is as follows:

[0134] Step S331: Determine the number of dimensions that need to be mutated;

[0135]

[0136]

[0137] Among them, dim is the dimension of the selected mutation, D is the dimension, J is the number of iterations, G is best(d) is the optimal particle, G best(d) ′ is the new optimal particle, and fit() is the fitness function.

[0138] Step S332: Determine whether the time-varying probability (TP) is less than the reverse learning probability (OP), OP = 0.2. When TP < OP, the global optimal solution (G best ) implements the reverse learning strategy. When TP≥OP, best Implement Gaussian mutation strategy;

[0139] Step S333: After executing the reverse learning strategy, determine whether the dynamic boundary has been exceeded. best(d) <min(G best ) or G best(d) >max(G best ), randomize G best ,min(G best ) is the minimum value of the optimal particle, max(G best ) is the maximum value of the optimal particle;

[0140] Step S334: Compare the fitness values ​​of the new individual with the original individual and output the reconstructed particle Pmix .

[0141] Furthermore, in step S33, the Outward Bound Learning (OBL) strategy is used to calculate and evaluate the optimal solution and the anti-solution, and the optimal value is selected as the next generation global guidance to increase the probability of jumping out of the local optimum.

[0142] (1) Define the opposition point X i (′)=(x i1 (′), x i2 (′),...,x iD (′))

[0143] x′ q =a q +b q -x q , x q ∈(a q , b q )

[0144] Among them, X i =(x i1 , x i2 , ..., x iD ) is a point in D-dimensional space, x′ q is the reverse point, x q is a point in multidimensional space, a q is the dynamic lower boundary, b q is the dynamic lower boundary.

[0145] (2) Define the anti-solution of the optimal particle

[0146]

[0147] Among them, τ∈U(0,1) is the generalized coefficient, d a and d b are the dynamic lower and upper bounds of the search space, respectively, and d a =min(x),d b =max(x), x is the particle value in multidimensional space.

[0148] In addition, in the process of learning based on anti-solution, the solution value may exceed the dynamic boundary [d a , d b ], the calculation formula is as follows:

[0149]

[0150] Optionally, in step S33, the Gaussian mutation (GM) strategy is used to perturb the extreme value based on normal random distribution to obtain a new optimal particle, and the calculation formula is as follows:

[0151] G best(d) ′=G best(d) +r·(X max -X min )·Gaussian(μ,δ 2 )

[0152] G best(d) j′ =N(G best(d) j ,δ)

[0153]

[0154] Among them, X min and X max are the dynamic lower and upper bounds of the search space respectively, and δ is the best(d) uncertainty, θ1 and θ2 are radius parameters, whose values ​​are -1 and -3 respectively, r is a random number that follows a uniform distribution, μ is the mean of the Gaussian random number, J is the loop variable, G best(d) j is the multidimensional variable of the optimal particle, G best(d) j′ is the d-dimensional variable after optimal particle perturbation.

[0155] Step S4: Estimate the carbon emissions of ships and verify the low carbon and sustainability of the ship scheduling system.

[0156] Furthermore, the formula for estimating ship carbon emissions is as follows:

[0157]

[0158] Ship emission reduction = (T start -Z)·e

[0159] Among them, v i is the ship speed, w i is the ship load, ψ i is the weight of the ship, is the carbon emission coefficient, α, β, q are adjustment parameters, α>0, β≥0, q≥3, parameter α adjusts the ship's related characteristics and loading conditions, β adjusts the ship's fuel consumption at low speed, q is the correlation coefficient between fuel consumption and speed, T start is the total waiting time of the original ship scheduling plan, is the fuel consumption during the ship's voyage. is the fuel consumption during the ship's berthing process, and e is the carbon emissions per unit time.

[0160] Using data from the locks and nearby waterways of the Three Gorges Hydropower Project on the Yangtze River and the actual navigation data of ships, a ship dispatch plan is formulated and related tests are carried out with the goal of minimizing the total waiting time of ships at anchorage.

[0161] Figure 4 A comparison diagram of solving the ship scheduling problem provided by an embodiment of the present invention, wherein (a) is a normal situation, and (b) is a flood-affected situation. Figure 4 The results shown show that the DMPSO-RH algorithm can jump out of the local optimal solution and has better optimization capabilities than the PSO and DMPSO algorithms. Specifically, the optimization results of the DMPSO-RH algorithm under normal conditions are 2.04% and 0.35% better than the basic PSO and DMPSO algorithms, respectively; the optimization results of the DMPSO-RH algorithm under the flood season flood impact scenario are improved by 3.98% and 0.85%, respectively.

[0162] Figure 5 is a schematic diagram of carbon emission reduction rate under different conditions provided by an embodiment of the present invention, Figure 5 The results shown show that under normal conditions and flood-affected conditions, the carbon emissions of ships anchored at anchorages can be reduced by 29.1% and 28.6%, respectively. It should be noted that the carbon emission reduction rate of the ship decreased by 0.5% during the flood season. The main reason is that ships are restricted from entering the waterway at certain moments during the flood season, which changes the optimal departure time of the ship and makes it impossible for the ship to catch up with the current water flow time window, resulting in long waiting time and ship carbon emissions. The detailed results show that the proposed ship scheduling optimization method can promote the coordination between the waterway operation time and the lock activation time, thereby reducing the carbon emissions of ships waiting at anchorages, which is conducive to the sustainable development of inland waterway shipping.

[0163] In addition, an embodiment of the present invention further provides a flood season ship scheduling optimization system based on waterway-lock coordination, including:

[0164] The acquisition module is used to obtain the ship information, lock attribute information, and navigation requirements of the waterway during the flood season at the water transport hub;

[0165] The construction module is used to build a ship scheduling optimization model with the optimization goal of minimizing the total waiting time of ships at the anchorage, and establish the constraint conditions that ships must meet during the flood season;

[0166] The optimization solution module solves the ship scheduling optimization model based on the dynamic multi-population particle swarm optimization algorithm with recombinant learning and hybrid mutation (DMPSO-RH) to obtain the optimized flood season ship scheduling plan;

[0167] Verify the sustainability module, estimate the carbon emissions of ships, and verify the low carbon and sustainability of the ship scheduling system.

[0168] In summary, the embodiments of the present invention have the following technical effects:

[0169] (1) A mathematical model for optimizing ship scheduling in the Three Gorges Gezhouba Dam during the flood season was constructed. The characteristics of the ship scheduling problem during the flood season based on the coordination of the waterway and the ship lock were analyzed, and a mathematical model for optimizing ship scheduling during the flood season was constructed based on this. The model sets the optimization objective as minimizing the total waiting time of ships at the anchorage, and proposes several related constraints such as the ship water flow during the flood season and the coordination of the waterway and the ship lock.

[0170] (2) Research on improvement of solution algorithm. In order to solve the problem of ship scheduling optimization during flood season, the imbalance between the global exploration ability and the local development ability of the basic particle swarm algorithm will lead to insufficient improvement in overall performance, slow convergence speed, and easy to fall into local extreme values. This paper analyzes and improves the problems. A dynamic multi-population particle swarm optimization (DMPSO-RH) algorithm with reorganization learning and hybrid mutation is proposed. The population is divided into sub-populations and local PSO models of the same size, and different speed update strategies are implemented; effective particle reconstruction and speed update strategies are introduced to improve the local development ability of the population; a probability-based hybrid mutation strategy is proposed to update the adaptive dimension while maintaining elite learning and expand the global search range. It should be noted that the size of the serial number of each step in the above embodiment does not mean 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 embodiment of the present application.

[0171] (3) Model solution and example verification. Numerical experiments on the Three Gorges Dam show that the DMPSO-RH algorithm can obtain an approximate optimal solution to the actual large-scale ship scheduling problem within a reasonable time. The scheduling scheme generated by the DMPSO-RH algorithm optimizes the ship's departure time and departure order, effectively matches the channel operation time and the lock activation time, and enhances the coordination of the terminal channel and the lock under different scenarios. Through the above improvements, the total waiting time of the ship in the two cases was reduced by 29.08% and 28.59% respectively, and the carbon emissions of the ship were also reduced accordingly. This method can effectively avoid navigation risks and improve the efficiency of ship scheduling during the flood season. The research results are of great significance to reducing the waiting time of ships in the Three Gorges Reservoir area and improving environmental pollution.

[0172] It should be pointed out that, according to the needs of implementation, the various steps / components described in this application can be split into more steps / components, and two or more steps / components or partial operations of steps / components can be combined into new steps / components to achieve the purpose of the present invention.

[0173] It will be easily understood by those skilled in the art that the above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. Any modifications, equivalent substitutions and improvements made within the spirit and principles of the present invention should be included in the protection scope of the present invention.

Claims

1. A flood season ship scheduling optimization method based on waterway-lock coordination, characterized in that: include: Obtain information on ships, lock properties and navigation requirements for flood season at water transport hubs; Based on the ship information, lock attribute information and flood season navigation requirements of the water transport hub, a ship scheduling optimization model is constructed with the optimization goal of minimizing the total waiting time of ships at anchorage, and the constraint conditions satisfied by the ship lock scheduling during the flood season are established; A dynamic multi-population particle swarm optimization algorithm based on recombination learning and hybrid mutation is used to solve the ship scheduling optimization model and obtain an optimized flood season ship scheduling plan; Based on the optimized flood season ship scheduling plan, the carbon emissions of ships are estimated.

2. The method according to claim 1, characterized in that The ship information includes: ship's arrival time, sailing speed, ship length, ship area and the maximum water flow standard acceptable to the ship; The lock attribute information includes: scheduling cycle, lock area, lock chamber capacity; The flood season waterway navigation requirement information includes: when a ship approaches a lock, in order to ensure the safety of waterway passage, it is necessary to ensure that there is a time interval between two adjacent ships; when a ship is sailing, the current water flow standard of the waterway does not exceed the maximum acceptable value of the ship.

3. The method according to claim 1 or 2, characterized in that: The optimization objective is to minimize the total waiting time of ships at anchorages, build a ship scheduling optimization model, and establish constraints for ship scheduling during flood season, including: Calculate the total waiting time of the vessel at the anchorage; According to the calculated total waiting time of the ship at the anchorage, the constraint conditions satisfied by the ship lock-passing scheduling during the flood season are constructed.

4. The method according to claim 3, characterized in that By MinZ=∑ i∈N (t bi -t ai ) calculates the total waiting time of the ship at the anchorage, where Z is the total waiting time of the ship at the anchorage, t ai is the time when the ship arrives at the anchorage, t bi is the time when the ship leaves the anchorage, and N represents the number of ships to be dispatched.

5. The method according to claim 4, characterized in that The constraints include: Constraint 1: Among them, t s is the start time of the scheduling period; Constraint 2: Among them, y ik ∈{0, 1}, if the ship is transferred to the lock, then y ik =1, otherwise y ik =0, K represents the set of ship lock times, and M represents a constant; Constraint 3: Among them, l i is the distance that the ship needs to pass through the channel to reach the lock, v i is the ship's speed through the channel, t ci The time when the vessel leaves the channel and approaches the lock; Constraint 4: Constraint 5: Among them, the minimum time interval between two consecutive ships heading towards the dam is Δt, γ is the multiple of the ship length, and L i is the length of the ship; Constraint 6: Among them, T k is the time required for a single gate, t di The time when the vessel leaves the lock; Constraint 7: Among them, t e The end time of the scheduling period; Constraint 8: z k If the gate is used, it is equal to 1, otherwise it is 0; Constraint 9: Constraint 10: Among them, T l is the upper limit of the lock time, t k The start time of the ship lock; Constraint 11: Constraint 12: Among them, σ represents the minimum time interval between consecutive gate times; Constraint 13: Among them, z k ∈{0, 1}, if the ship lock starts, then z k =1, otherwise z k =0; Constraint 14: Among them, ξ it ∈{0, 1}, if the ship enters the channel within the time step, then ξ it =1, otherwise ξ it =0, T is the time step set; Constraint 15: Constraint 16: Constraint 17: Among them, η ikt ∈{0, 1}, if the ship enters the channel and passes through the lock within the time step, then η ikt =1, otherwise η ikt =0; Constraint 18: Among them, f i The acceptable water flow standard for ships is is the actual water flow value of the channel within the time step; Constraint 19: Among them, a i is the ship area, A k is the lock chamber area; Constraint 20: Among them, q t Indicates the upper limit of the channel capacity within the time step.

6. The method according to claim 4, characterized in that The dynamic multi-population particle swarm optimization algorithm based on recombination learning and hybrid mutation solves the ship scheduling optimization model to obtain an optimized flood season ship scheduling plan, including: Determine the individual dimension N, where N represents the number of ships to be dispatched, the vector represents the departure time and order of the ships at the anchorage, and the initial vector range [t s , t e ], generating an initial solution based on a large language model, t s is the time when the scheduling period starts, t e The time when the scheduling period ends; The total waiting time of the ship is taken as the fitness value, and the median of the fitness value is used as the boundary to divide the population into three sub-populations: the neighborhood domain, the alienation domain, and the recombination domain. Multiple populations search in different areas to determine the optimal ship scheduling plan, and within the specified time interval, a dynamic random recombination mechanism is used to continue the search process; Reconstructed particles are generated by dynamically adjusting and mixing the reverse learning method and Gaussian mutation method, and the subspace information of the historical optimal position is retained when the particle velocity is updated.

7. The method according to claim 6, characterized in that The generating an initial solution based on the large language model includes: In view of the ship dispatch problem during the flood season, the ship mission data is obtained, and the training samples are divided according to the ship mission data to obtain the training set, validation set and test set; Select a large language model based on Transformer to understand natural language text, process structured data, and generate flood season ship scheduling plans; Input the training set data into the model, generate the predicted flood season ship dispatch plan through forward propagation, calculate the loss function between the predicted and actual results, update the model parameters through the back propagation algorithm, and iterate the training until the model achieves satisfactory performance on the validation set; Based on the meta-optimizer, the model with a passing validation result is fine-tuned using the ship mission data; The performance of the model is evaluated using the test set. Once the evaluation passes, the model is deployed to actual applications to generate enough initial solutions to form the initial population.

8. The method according to claim 7, characterized in that The method of generating reconstructed particles by dynamically adjusting and mixing the reverse learning method and the Gaussian mutation method includes: Depend on Determine the number of dimensions that need to be mutated, where dim is the dimension to be mutated, D is the dimension, J is the number of iterations, and G is the number of iterations. best(d) is the optimal particle, G best(d) ′ is the new optimal particle generated, and fit() is the fitness function of the corresponding particle; Determine whether the time-varying probability TP is less than the reverse learning probability OP. When TP < OP, perform the reverse learning method on the global optimal solution. When TP ≥ OP, perform the Gaussian mutation method on the global optimal solution. After executing the reverse learning method, determine whether the dynamic boundary has been exceeded. If G best(d) <min(G best ) or G best(d) >max(G best ), randomize G best ,min(G best ) is the minimum value of the optimal particle, max(G best ) is the maximum value of the optimal particle; Compare the fitness values ​​of the new individual with the original individual and output the reconstructed particles.

9. The method according to claim 8, characterized in that The estimating of ship carbon emissions based on the optimized flood season ship scheduling plan includes: Depend on Ship emission reduction = (T start -Z)·e estimates the carbon emissions of ships, where v i is the ship speed, w i is the ship load, ψ i is the weight of the ship, is the carbon emission coefficient, α, β, q are adjustment parameters, parameter α adjusts the ship's related characteristics and loading conditions, β adjusts the ship's fuel consumption at low speed, q is the correlation coefficient between fuel consumption and speed, T start is the total waiting time of the original ship scheduling plan, is the fuel consumption during navigation, is the fuel consumption when berthing at the anchorage, and e is the carbon emission per unit time.

10. A flood season ship dispatch optimization system based on waterway-lock coordination, characterized in that: include: The acquisition module is used to obtain the ship information, lock attribute information and navigation requirements of the waterway during the flood season at the water transport hub; A construction module is used to construct a ship scheduling optimization model based on the ship information, lock attribute information and flood season waterway navigation requirement information of the water transport hub, with minimizing the total waiting time of ships at anchorage as the optimization goal, and establish constraint conditions satisfied by ship lock scheduling during the flood season; An optimization solution module is used to solve the ship scheduling optimization model based on a dynamic multi-population particle swarm optimization algorithm based on recombination learning and hybrid mutation to obtain an optimized flood season ship scheduling plan; A verification sustainability module is used to estimate the carbon emissions of ships based on the optimized flood season ship scheduling plan.

Images