Intelligent scheduling method for bulk and general cargo ships at port along river based on comprehensive cost optimization

Abstract

The invention discloses an intelligent scheduling method for bulk and general cargo ships at ports along rivers based on comprehensive cost optimization, and the method comprises the steps: forming an optimized berth distribution scheme in a manner of minimizing the sum of waiting time cost and berth cost through dynamic tracking of meteorological tide information, and achieving the intelligent scheduling of bulk and general cargo ships at ports along rivers. The parking cost comprises the delay cost or the sum of the in-plan berthing cost and the delay cost, and target function optimization is carried out by using a traditional genetic algorithm. According to the method, the phenomenon of ship overstock easily caused by a traditional pure first-arrival first-service mode is overcome, meanwhile, the first-arrival first-service requirement is also considered, and the lowest comprehensive cost of ship loading and unloading waiting and berthing is achieved.

Description

[0001] This application is a divisional application of the Chinese invention patent application with the application date of August 1, 2024, application number 202411044504.3, and invention name “A method for intelligent scheduling of bulk cargo loading and unloading ships in riverside ports”. Technical Field

[0002] The invention belongs to the technical field of ship loading and unloading, and in particular relates to an intelligent dispatching method for bulk cargo ships in riverside ports based on comprehensive cost optimization. Background Art

[0003] River ports have different loading and unloading requirements from seaports due to their great differences in hydrogeological conditions, port facilities, operation management, environmental protection, ship types, and transportation scope. For example, large bulk cargo ships need to enter and exit river ports during the tide, so real-time tracking of meteorological dynamics is required.

[0004] With the increase in port throughput, especially in ports with water-to-water transport, the phenomenon of ships being crowded at ports is becoming increasingly serious. The reasonable allocation and scheduling of berths can improve the operating efficiency of ports. The current terminal scheduling methods mainly include first-come, first-served, first-off, first-served, maximum equipment utilization, minimum process conflict, and demand plan priority. There are problems such as low loading and unloading efficiency, a single control and scheduling method for loading and unloading ships of various types and sizes, low matching between ship types and berths, low effective utilization of berths (such as long and deep berths serving small ships that arrive first), and lack of intelligent consideration of tidal effects. In particular, the traditional first-come, first-served model is prone to waste port resources and cause a serious backlog of ships.

[0005] The Chinese invention patent with publication number CN108298329A and name “A new type of river-sea transport automated container terminal loading and unloading system and plane layout” is aimed at the “river-sea transport” container terminal project with a “non-linear” terminal shoreline. The proportion of waterway collection and distribution containers in the total terminal throughput is generally large. If the automated loading and unloading system and layout scheme of “double trolley shore container loading and unloading bridge + AGV + automated rail gantry crane” are still adopted, the horizontal transportation operation for waterway collection and distribution operations in the port area will have the following disadvantages: first, the horizontal horizontal transportation distance of waterway collection and distribution containers increases, and the energy consumption is large; second, the vehicle turnover efficiency of the container horizontal transfer system is low. A complete loading and unloading system and plane layout scheme for a river-sea transport automated container terminal that introduces unmanned container trucks and is suitable for a non-linear shoreline layout is provided. On the basis of realizing full automation of all loading and unloading links in the port area in this type of container terminal, the equipment cost and supporting civil engineering cost of horizontal transportation operations can be reduced, the vehicle turnover efficiency in the port area can be effectively improved, and the energy consumption of equipment can be reduced.

[0006] The Chinese invention patent with publication number CN112573219B and name “A smart port based on an intermodal rail system” aims to solve the problem that the farther the inland area is from the port, the higher the transportation cost and the longer the time for port collection and transportation, which to a certain extent restricts the development of inland export industries far away from the port. The current intermodal transport mode and logistics transportation still have the problem of serving the last mile. It provides a smart port based on an intermodal rail transport system. The intermodal rail transport system consists of an intermodal rail system and new energy unmanned intermodal vehicles (hereinafter collectively referred to as high-speed intermodal vehicles) and / or unmanned intermodal high-speed rail logistics vehicles (hereinafter collectively referred to as high-speed rail logistics vehicles), which directly connects coastal ports or inland river terminals with inland ports or logistics centers hundreds or even thousands of kilometers away.

[0007] Looking at the existing technologies, they are mainly concentrated in river-sea transport, rail-water transport, rail transport, and intelligent control of loading and unloading, all of which are mainly based on container loading and unloading. Most of the relevant port equipment is the same and fixed, and can be loaded and unloaded in sequence, and basically no deployment operation is required. The difference with bulk cargo loading and unloading is that the tonnage of ships may vary greatly, and the cargoes are different. Different berths or different terminal equipment need to be deployed to meet the loading and unloading needs. Therefore, there is a problem of resource scheduling and matching. If the traditional first-come-first-served model is followed, it is very easy to cause a decrease in the utilization rate of port equipment and a serious backlog of ships. At the same time, the influence of environmental factors such as the tidal changes of the Yangtze River must be taken into account. Intelligent scheduling is extremely difficult. At present, there is a lack of targeted research in this field, and with the increasing freight traffic in rivers and ports, it is urgently needed to be solved. Summary of the invention

[0008] Aiming at the above-mentioned scheduling problem of bulk cargo loading and unloading ships in riverside ports, in order to reduce the operating costs of ports, the present invention takes minimizing the waiting time and berthing cost of ships as the goal, establishes a mathematical model, and adopts a genetic algorithm to solve the problem of terminal berth allocation and scheduling optimization. It performs precise scheduling adaptation with waiting time cost and demurrage cost as the core, and can realize low-cost and high-efficiency intelligent scheduling of port ships.

[0009] The present invention designs an intelligent dispatching method for bulk cargo ships in riverside ports based on comprehensive cost optimization. The goal of berth allocation and dispatching is to maximize the port's operating efficiency, minimize user costs, balance the interests of both ports and shipping, and improve the port's competitiveness. The optimization principle of the intelligent dispatching method includes the principle of minimizing comprehensive costs. The comprehensive costs include waiting time costs (referring to the waiting time from the arrival of the ship to the berthing) and berth demurrage costs, or include waiting time costs, berth demurrage costs and planned berthing costs. The optimization objective function of the ship intelligent dispatching method is:

[0010]

[0011] The first term in the formula is the waiting time cost, and the second term is the demurrage cost, and the third item is the planned berthing cost, and the sum of the second and third items is the berthing cost; the sum of the above three items is also called the comprehensive cost;

[0012] The intelligent scheduling method forms a berth allocation scheme by optimizing the objective function that satisfies the constraint conditions. The berth allocation scheme at least includes the time S for starting loading and unloading. kji , berth k, and ship j’s information.

[0013] in:

[0014] ω 1j is the waiting time cost weight;

[0015] k = 1, 2, ... M, is the berth set, M is the maximum number of berths;

[0016] j = 1, 2, ... N, is the set of ships, N is the number of ships involved in the optimization calculation;

[0017] i=1, 2, …P, is the service order of the ship, P is the maximum order number, P≤N;

[0018] S kji is the start time of loading and unloading of berth k and ship j, and its service order is the i-th. In principle, the serial number i may not be marked, but in order to make the optimization result more intuitive, this embodiment uses S kji Identification method;

[0019] A j is the time when the jth ship arrives;

[0020] f 1j is the waiting cost per unit time of different ships;

[0021] ω 2j is the demurrage cost weight;

[0022] ΔT kj is the demurrage time of different ships, ΔT kj =max{0,S kji +t u -t j}, t j is the planned departure time of the ship, t u is the time taken to load and unload a ship based on experience;

[0023] f 2kj It is the unit time demurrage cost corresponding to different ships and different berths;

[0024] ω 3jis the planned berthing cost weight, and the optimization objective function satisfies ω 1j +ω 2j +ω 3j =1;

[0025] F 3kj is the planned berthing cost of ships at different berths;

[0026] The optimization objective function satisfies tidal constraints, which include:

[0027] h ks For S kji The time corresponds to the water depth of the kth berth, d j, is the draft of the jth ship, ΔT 2j The distance to the next high tide when the water depth meets h ks ≥d j, Condition time.

[0028] Furthermore, the method obtains corresponding berth water depth information by dynamically tracking meteorological information (i.e., high and low tide forecast table).

[0029] Furthermore, the optimization objective function satisfies the berth length constraint condition that the berth length is greater than the ship length.

[0030] Furthermore, the optimization objective function satisfies the loading and unloading time constraint condition, which includes that the loading and unloading time period of the ship must be within the allowed time period, otherwise demurrage will be formed, that is, t j-start ≤S kji ≤S kji +t u ≤t j-end , where t j-start and t j-end are the agreed loading and unloading time ranges for the jth ship respectively.

[0031] Furthermore, the optimization objective function satisfies the single berth allocation constraint, that is, each berth can only be allocated to one ship in the same time period.

[0032] Further, the objective function optimization includes any one or any combination of initial optimization, new ship entry optimization, on-demand optimization and stage optimization; the initial optimization is the first optimization, and the objective function optimization is started according to the berth usage, the ships that have entered the port (not yet berthed) and the ships planned to enter the port (which can enter the port on time within the optimization time period and can be considered to be included in the optimization demand, or the ships that have not yet entered the port can be ignored); the new ship entry optimization includes starting the optimization when a new ship enters the port, including keeping the optimization results of the previous ships that have entered the port unchanged, optimizing the new ships that have entered the port alone (which is basically equivalent to the first-come, first-served mode), or re-optimizing the new ships that have entered the port together with the ships that have not yet berthed; the on-demand optimization includes restarting the optimization by manual intervention; the stage optimization includes automatically starting the objective function optimization according to a predetermined time interval or a limit on the number of new ships entering the port or a limit on the total number of ships to be berthed.

[0033] Furthermore, the optimization method of the objective function includes an optimization method based on a genetic algorithm, comprising the following steps:

[0034] S1. Establish objective function and constraints;

[0035] S2. Generate berth allocation plan by coding;

[0036] S3, initialize the population;

[0037] S4, calculating the objective function and fitness function;

[0038] S5. Select an operation;

[0039] S6, cross operation;

[0040] S7, mutation operation, generating a new population.

[0041] S8. Set genetic algorithm control parameters.

[0042] S9. Repeat steps S4 to S7 until an end condition is met, wherein the end condition includes that the number of iterations reaches a set maximum number of iterations or the objective function calculation result meets the set objective function accuracy requirement or the fitness function result meets the set value requirement.

[0043] Furthermore, the encoding in step S2 includes expressing the berth allocation scheme in integer code or binary code.

[0044] Furthermore, step S5 selects the operation using the fitness function and the berth utilization rate as a comprehensive optimization principle.

[0045] Furthermore, there is no order restriction for step S8 relative to steps S1 to S7.

[0046] Furthermore, after the iterative calculation is completed, ω is assigned1j ,ω 2j ,ω 3j Re-iterate the optimization with different combination values.

[0047] Furthermore, the weight parameter ω 1j ,ω 2j ,ω 3j The combined values ​​of are taken as the optimal search options, and comprehensive optimization is performed on different weight combinations and berth allocation schemes.

[0048] The advantages and beneficial effects of the present invention are as follows: the intelligent dispatching method for bulk cargo ships in riverside ports based on comprehensive cost optimization designed by the present invention dynamically tracks meteorological tidal information and minimizes comprehensive costs, wherein the comprehensive costs include waiting time costs and demurrage costs, and the planned berthing costs can also be considered, thereby forming an optimized berth allocation scheme; the optimization method does not aim to provide the best service for a single ship or a single berthing, but takes the principle of minimizing the comprehensive cost of all ships. On the one hand, berth allocation is more closely matched with ship demand (the cost of small ships berthing at large berths is higher and the berth utilization rate is low), and the berth utilization rate is also higher, thereby realizing intelligent decision-making affected by tides, while also taking into account the needs of first-come, first-served service (the cost is naturally low if the waiting time is short), thereby reducing the serious backlog of ships. BRIEF DESCRIPTION OF THE DRAWINGS

[0049] Figure 1 It is a flowchart of the steps of the intelligent dispatching method of bulk cargo ships in riverside ports based on comprehensive cost optimization. DETAILED DESCRIPTION

[0050] The specific implementation of the present invention is further described below in conjunction with the accompanying drawings and embodiments. The following embodiments are only used to more clearly illustrate the technical solution of the present invention, and cannot be used to limit the protection scope of the present invention.

[0051] Example 1

[0052] The present invention designs an intelligent dispatching method for bulk cargo ships in riverside ports based on comprehensive cost optimization. The goal of berth allocation dispatching is to maximize the operating efficiency of the port and minimize the user cost. The optimization principle of the intelligent dispatching method includes the principle of minimizing the comprehensive cost. The comprehensive cost includes the waiting time cost (referring to the waiting time from the arrival of the ship to the berthing) and the berth demurrage cost, or includes the waiting time cost, the berth demurrage cost and the planned berthing cost. The optimization objective function of the ship intelligent dispatching method is:

[0053]

[0054] The first term in the formula is the waiting time cost, and the second term is the demurrage cost, the third item is the berthing cost within the plan, and the sum of the second and third items is the berthing cost; the sum of the above three items is also called the comprehensive cost;

[0055] The intelligent scheduling method forms a berth allocation plan by optimizing the objective function that meets the constraint conditions. The berth allocation plan at least includes the start loading and unloading time S kji , information of berth k and ship j. In the actual engineering implementation of this embodiment, it also includes the service order i and the end loading and unloading departure time E kji information. The berth allocation plan generally allocates berths and berthing and unberthing times for all ships that need to berth but have not yet berthed at the port.

[0056] The result of the above optimization calculation is to obtain the berth allocation plan for the jth ship to start loading and unloading at the kth berth corresponding to the minimum comprehensive cost, and give the planned departure time. The comprehensive cost is the sum of the waiting time cost and the berthing cost. kji where:

[0057] ω

[0058] is the waiting time cost weight; 1j k = 1, 2,... M, is the berth set, and M is the maximum number of berths;

[0059] j = 1, 2,... N, is the ship set, and N is the number of ships participating in the optimization calculation;

[0060] i = 1, 2,... P, is the ship service order, P is the maximum sorting number, and P ≤ N; generally, there are 2 sorting methods. The first is that all berth services are sorted uniformly, then P = N. When there is an optimization result where different berths can start loading and unloading services simultaneously, for the ships that meet the optimization result, the sorting can be random, but their loading and unloading start time S

[0061] is the same, and the sorting value j does not affect the actual service order; the second is that the ships corresponding to different berth services are sorted separately. At this time, generally P < N, and there will be no sorting conflict phenomenon because a berth can only serve one ship at a time; this embodiment adopts the second sorting method; kji

[0062] S kji is the start loading and unloading time of berth k and ship j, and its service order is the ith. In principle, the serial number i can be not marked, but in order to make the optimization result more intuitive, this embodiment adopts the identification method of S kji j ;

[0063] A j is the arrival time of the jth ship;

[0064] f 1j It is the waiting cost per unit time of different ships, which can be measured in yuan / hour. It belongs to the empirical parameter series and is determined by factors such as ship tonnage, cargo type, anchorage fee, and transportation profit and loss of waiting time. In this embodiment, "waiting cost = anchorage cost + profit and loss cost" is set for calculation, wherein the anchorage cost is measured in accordance with the "Port Charge Calculation Method" revised by the Ministry of Transport and the National Development and Reform Commission, and the profit and loss cost is measured according to the average hourly income loss of the corresponding ship. This parameter can be calculated based on the total annual income of various types of ships divided by the total number of hours in a year, or it can be calculated using statistical data based on the current status of the industry. Generally, the ship type and cargo type should be distinguished. However, since various accurate statistical data are difficult to obtain, for those with certain data, they are calculated according to the certain data. For those without accurate data, for the sake of simplicity, this embodiment calculates the average hourly income of bulk cargo ships based on public information, and takes 3,000 yuan / hour for cost calculation.

[0065] ω 2j is the demurrage cost weight;

[0066] ΔT kj is the demurrage time of different ships. The demurrage time of ships at different berths and in different service orders is different. ΔT kj =max{0,S kji +t u -t j}, t j is the planned departure time of the ship, t u is the empirical loading and unloading time, which is related to factors such as the total amount of materials, the state of the material storage location, the loading and unloading equipment, and the selection of the process. This value is taken from historical experience data; t u ≥t r ,t r working hours for loading and unloading the boat for the reclaimer;

[0067] f 2kj It is the demurrage cost per unit time corresponding to different ships and different berths, which can be measured in RMB / hour. It is determined by the specific ship tonnage and the berth, and generally depends only on the berth. This embodiment is calculated according to the revised "Port Charge Calculation Method" by the Ministry of Transport and the Development and Reform Commission, or the default rate standard of the port and shipping parties; or the demurrage cost loss of both the port and shipping parties is considered at the same time, and a unified calculation is made according to a certain determination, such as 4,000 yuan / hour;

[0068] The calculation of demurrage cost, i.e. demurrage, is related to loading and unloading time and demurrage rate. The length of loading and unloading time and the level of demurrage rate are the result of negotiation between the lessor (port party) and the lessee (ship party) after comprehensive consideration of loading and unloading delay risk, operating cost, operating income and other factors, and the risk sharing is relatively balanced. Different routes, different goods, different ports, different ship parameters, and different shipping market situations will affect the negotiation results of the lessor and the lessee on loading and unloading time and demurrage rate.

[0069] ω 3j is the planned berthing cost weight, and the optimization objective function satisfies ω 1j +ω 2j +ω 3j =1, that is, the sum of the weights of the same group is 1; ω 1j ,ω 2j ,ω 3j Different ships can be assigned different weights, and thus identified by j. Alternatively, all ships can be assigned the same set of weights. In this embodiment, ω is selected. 1j =0.5,ω 2j =0.5,ω 3j =0, which means that the comprehensive cost does not take into account the planned berthing cost and does not vary from ship to ship;

[0070] F 3kj It is the planned berthing cost of ships at different berths. In general practical applications, the berthing cost agreed in the contract may be different for different ships, but it is generally no longer related to the berth selection, that is, for the same ship, the planned berthing cost of different berths is consistent. The port party selects the appropriate berth according to the actual situation for the specific berthing method. The intelligent scheduling method described in the present invention actually follows this principle, because unless there is no large ship waiting for scheduling, once a large berth is allocated to a significantly smaller ship, although the comprehensive cost of the small ship may be smaller, the comprehensive cost of the corresponding large ship will definitely increase; the planned berthing cost can also be calculated based on the revised "Port Charges and Calculation Methods" by the Ministry of Transport and the Development and Reform Commission and the agreed time period, or the allocated berth and the variable allocation time are considered at the same time. At this time, the planned berthing cost is a variable to be optimized;

[0071] The optimization objective function satisfies tidal constraints, which include:

[0072] h ks For S kji The time corresponds to the water depth of the kth berth, d j, is the draft of the jth ship, ΔT 2j The distance to the next high tide when the water depth meets h ks ≥d j, Condition time.

[0073] That is, only when S kji When the water depth of the kth berth is greater than or equal to the draft of the jth ship, S kji Only then can it meet the conditions for becoming a usable optimization result. The present invention converts the information of rising and falling tides into water depth conditions, making the method more universal. That is, if the port berth has a deep water depth, general ships can meet it and it is not affected by the rising and falling tides. The method of the present invention is still applicable because it only determines whether the water depth conditions meet the requirements. However, for most ports along the Yangtze River, the water depth is greatly affected by the rising and falling tides, and most of them do not meet the berthing conditions at low tide. The method of the present invention introduces meteorological tidal information, which can accurately predict the berth water depth at any time in the future, thereby making the ship berthing scheduling plan and plan more feasible. Both the port and the ship (also known as the navigation party) can arrange the work plan in advance, which improves the accuracy and executability of the plan, and also improves the work efficiency of all parties.

[0074] The demurrage cost refers to the additional demurrage fees that need to be levied when the ship exceeds the planned departure time due to special reasons, resulting in demurrage.

[0075] Regarding the planned berthing cost, for fixed ship types, the planned berthing cost is fixed or contractually agreed, so it is generally not necessary to consider it in the optimization process. However, considering that the normal berthing loading and unloading time cost may be related to the type and weight of the actual cargo loaded on the ship, and may also be related to the location, scale and loading and unloading operation content of the specific berth itself, this cost can also be considered in the optimization process; therefore, in the objective function of formula 1, the planned berthing cost parameter ω 3j Adjustment, when ω 3j >0, the planned berthing cost participates in the optimization, when ω 3j =0, the planned berthing cost does not participate in the optimization; or if the planned berthing cost does not need to be considered, the planned berthing cost only needs to be set to a fixed value, so that the parameter has no substantial impact on the optimization process, which is equivalent to not considering the normal berthing cost, which also reflects the strong adaptability of the method of the present invention.

[0076] The present invention adopts the user comprehensive cost minimization index to replace the traditional first-come-first-served model, but it can still take into account the principle of first-come-first-served, because if the ship waits too long, the cost will naturally increase. Although individual ships may not enjoy first-come-first-served service for a certain loading and unloading, their comprehensive cost for multiple berthing is still the lowest, or they can still enjoy the first-come-first-served service after comprehensive consideration, and the overall waiting time will not be too long. Because existing experience tells us that simple first-come-first-served service is likely to cause waste of terminal resources and backlog of ships. The backlog phenomenon will naturally cause more ships and a greater probability of long waiting time, resulting in excessively high comprehensive costs.

[0077] Preferably, the method obtains the corresponding berth water depth information by dynamically tracking meteorological information (i.e., the tide forecast table), and can also statistically predict the water depth at the corresponding time based on the tide time, seasonal tide changes and the measured water level. Generally, the water depths of different berths in the same port area are consistent. This embodiment also installs a water level monitor at the dock to continuously monitor the water level, and combined with the meteorological tide table, it can achieve a more accurate prediction of the water depth at the time of ship berthing.

[0078] Preferably, the optimization objective function satisfies the berth length constraint condition that the berth length is greater than the ship length, that is, S in Formula 1 that can become an effective optimization result kji The berth length l must be met k ≥l j , where l k is the berth length of berth k, l j is the ship length of ship j.

[0079] Preferably, the optimization objective function satisfies the loading and unloading time constraint condition, and the loading and unloading time constraint condition includes that the loading and unloading time period of the ship must be within the allowed time period, that is, t j-start ≤S kji ≤S kji +t u ≤t j-end , where t j-start and t j-end They are the time ranges for loading and unloading agreed upon for the jth ship, otherwise demurrage will be formed. This agreement is mainly agreed upon by the port and the ship through a contract or tacit agreement. Generally, the scope is larger, leaving room for the port to adjust its plans. On the one hand, this agreement is to satisfy the ship so that it does not have to wait too long, which is a constraint on the port. On the other hand, it is also to control the optimization process. It cannot be optimized indefinitely, but should be optimized within a certain range.

[0080] Preferably, the optimization objective function satisfies the single berth allocation constraint, that is, each berth can only be allocated to one ship in the same time period. Specifically, when setting the optimization scheme each time, a berth is continuously allocated in the available time period, and only one ship is allocated in the same time period. After all berths are allocated, the objective function value is calculated, and then the objective function values ​​of multiple allocation schemes are compared to select the lowest one, which of course satisfies all the constraints; or by adding several variables X kjt , berth is not allocated before initializing X kjt = 0, when the kth berth is allocated to the jth ship in time period t, kjt =1, or the time period attribute is additionally included, otherwise X kjt = 0. When calculating the objective function, first determine X kjtWhether it has been 1 during the optimization time period, if it is 1, the objective function value of the optimization solution is not calculated, otherwise if it is 0, then set X kjt Calculate the objective function after it is 1 or set X after calculating the objective function kjt is 1.

[0081] Preferably, the objective function optimization includes any one or any combination of initial optimization, optimization for new ships entering the port, on-demand optimization and stage optimization; the initial optimization is the first optimization, and the objective function optimization is started according to the berth usage, ships that have entered the port (have not yet berthed) and ships that are scheduled to enter the port (if they can enter the port on time within the optimization time period, they can be considered to be included in the optimization demand, or ships that have not yet entered the port can be ignored); the new ship entering the port optimization includes starting the optimization when a new ship enters the port, including keeping the optimization results of the previous ships that have entered the port unchanged, optimizing the new ships that have entered the port individually (which is basically equivalent to the first-come, first-served mode), or optimizing the new ships that have entered the port. The ship and the ships that have not yet berthed are re-optimized. This embodiment adopts the scheme of re-optimizing all the ships that have not berthed; the on-demand optimization includes manual intervention to restart the optimization, which is generally mainly based on the special needs of the port or the ship or the need to regenerate a new constraint optimization scheduling plan. If a ship fails to leave as planned after berthing, or leaves in advance, the above situations are all large discrepancies with the planned time, such as more than 10 minutes, and manual re-optimization can be organized; the stage optimization includes automatically starting the objective function optimization according to the established time interval or the limit on the number of new ships entering the port or the total limit on the number of ships to be berthed. After the initial optimization is completed, this embodiment usually adopts the mode of starting the stage optimization when there are more than 2 new ships entering the port, and optimizes all unberthed ships. Before starting the stage optimization according to the plan, if there is a berth vacancy or a berth vacancy is about to appear, and there is no corresponding berthing plan, the optimization is started immediately.

[0082] Preferably, the optimization method of the objective function includes an optimization method based on a genetic algorithm, comprising the following steps:

[0083] S1. Establish objective function and constraints;

[0084] S2. Generate berth allocation plan by coding;

[0085] S3, initializing the population, generating several berth allocation schemes in a random manner to initialize the population, and each berth allocation scheme must meet the requirements of various constraints;

[0086] S4. Calculate the objective function and fitness function. Calculate the objective function value according to the initialization population plan. Since the berth allocation problem is to minimize the sum of the waiting time cost and the berthing cost of the ship under constraints, this embodiment uses the exponential scaling method to evaluate the quality of each individual according to the objective function calculation result. The fitness function takes into account the minimization of the waiting time cost and the berthing cost. In berth scheduling optimization, the range of variation of the objective function value of different scheduling schemes varies greatly. If the linear transformation method or the power function transformation method is used, it is difficult to avoid the outstanding individual from affecting the global optimization performance. Therefore, the present invention selects the exponential transformation method as the fitness function.

[0087] f(x)=1 / (1+e 0.0001x )

[0088] Among them, x is the calculation result of the objective function.

[0089] S5, selection operation, according to the evaluation result of the fitness function, select individuals with higher fitness as the parents of the next generation population. In this embodiment, the selection operation adopts a roulette selection method;

[0090] S6, crossover operation, in this embodiment, a single-point crossover is performed by selecting partial berth allocation schemes of two parent individuals;

[0091] S7, mutation operation, generating a new population, that is, a number of berth allocation schemes. In this embodiment, the mutation operation is realized by exchanging the service order of two ships. A random transposition mutation operation can also be used to randomly generate two integer-coded positions, and then exchange them according to the two position values ​​to achieve the mutation purpose.

[0092] S8. Set the genetic algorithm loop control parameters. In this embodiment, the population size is set to 100, the evolutionary algebra is set to 30-50, the crossover probability is set to 0.6, and the mutation probability is set to 0.01. Generally, the population size should be determined according to the number of variables. For a small number of variables, the population size should be 20-50, and for a large number of variables, the population size should be 100-200. The evolutionary algebra can be dynamically adjusted according to the convergence situation.

[0093] S9. Repeat steps S4 to S7 until an end condition is met, wherein the end condition includes that the number of iterations reaches a set maximum number of iterations or the objective function calculation result meets the set objective function accuracy requirement or the fitness function result meets the set value requirement.

[0094] The specific execution process of the genetic algorithm can refer to the MATLAB genetic algorithm toolbox.

[0095] Preferably, the encoding in step S2 includes expressing the berth allocation scheme in integer code or binary code. For example, in this embodiment, 5 data are encoded to represent S kji(time of starting loading and unloading), berth k, vessel j, service order i and time of ending loading and unloading and leaving the berth E kji .

[0096] The integer code, such as a certain optimization, each ship in the optimization range is a gene, and the value corresponding to the gene is the ship number. For example, the first ship to enter the anchorage is numbered 1, and the second ship to enter the anchorage is numbered 2. Since there will be no ship with number 0, the number 0 is used as a separator for berths. 4, 1, 3, 5, 0, 2, 7, 11, 8, 12, 9, 0, 6, 10, represents that two 0s separate three berths. The first berth serves ships 4, 1, 3, 5 in order 1 to 4. The second berth serves ships 2, 7, 11, 8, 12, 9 in order 1 to 6. The third berth serves ships 6 and 10 in order 1 to 2.

[0097] Preferably, step S8 has no order restriction relative to steps S1 to S7, but should be completed before step S9.

[0098] Example 2

[0099] The difference from Example 1 is that, after the initial optimization is completed, this embodiment adopts a mode of starting the optimization phase when the number of new ships entering the port exceeds 5, and only performs optimization on the new ships entering the port, and the previous optimization results are retained, that is, the previous optimization results are not changed if there is no abnormality, and on-demand optimization is performed when there is an abnormality.

[0100] Example 3

[0101] The difference from Example 1 is that after the initial optimization is performed in this embodiment, the optimization is updated according to a comprehensive strategy of new ship arrival optimization and on-demand optimization, wherein each new ship entering the port is re-optimized together with the unberthed ships, and on-demand optimization is performed when there is an abnormality.

[0102] Example 4

[0103] The difference from Example 1 is that, in step S5 of this embodiment, the selection operation uses the fitness function and the berth utilization rate as the comprehensive optimization principle. While considering the traditional fitness function selection method of the genetic algorithm to select individuals with higher fitness as the parents of the next generation population, individuals with high berth utilization rate are preferred as the parents of the next generation population, that is, a certain number of individuals with both high fitness and high berth utilization rate are preferred as the parents of the next generation population. Specifically, a weight of 0.5 can be given to the fitness function calculation result and the berth utilization rate calculation result, and the individuals with the highest comprehensive sum of the two can be preferred; the berth utilization rate U j for

[0104]

[0105] The actual usage time is the sum of the planned usage time of all available berths in the berth allocation plan to be selected, and the total available time is the total available time of all available berths in the berth allocation plan to be selected, generally only deducting the unavailable time caused by tidal factors.

[0106] Example 5

[0107] The difference from Example 1 is that after the iterative calculation in step S9 of this embodiment is completed, ω is assigned 1j ,ω 2j ,ω 3j Re-iterate and optimize with different combination values. In this embodiment, ω is selected 1j =0.4,ω 2j =0.4,ω 3j =0.2, which is equivalent to taking the waiting time cost, berth demurrage cost and planned berthing cost into account in the comprehensive cost; comparing the optimization results of different weight combinations and selecting the best one.

[0108] Example 6

[0109] The difference from Example 1 is that in this embodiment, the weight parameter ω 1j ,ω 2j ,ω 3j The combined values ​​of are used as optimization options, and comprehensive optimization is performed on different weight combinations and berth allocation schemes. This method is mainly suitable for the initial application of the method of the present invention in ports. When there is a lack of experience in weight determination, several comprehensive optimizations involving weight combinations are first performed. Based on the results of multiple comprehensive optimizations, a group of suitable weight combinations are manually determined for subsequent normal berth allocation scheme optimization. Because the optimization calculation workload involving weight parameters is large and the efficiency is low, it is not worth doing frequently. It is simpler and more direct to directly optimize the berth allocation scheme based on empirical parameters, and the optimization results are also in line with the expectations of the port and the ship.

[0110] The above is only a part of the relatively systematic and comprehensive embodiments of the intelligent scheduling method for bulk cargo ships in river ports based on comprehensive cost optimization of the present invention. In fact, the combination of various preferred schemes, and the selection of other optimization algorithms such as neural network method, greedy algorithm, simulated return algorithm, ant colony algorithm, particle swarm algorithm, enumeration method, etc. to optimize the objective function, or the use of different parameter setting methods in various algorithms, such as setting different numbers of iterations in genetic algorithms, performing different selection, crossover or mutation operations, etc., these combinations or preferred schemes should also be regarded as the protection scope of the present invention, and they are not listed here one by one.

Claims

1. An intelligent dispatching method for bulk cargo ships in river ports based on comprehensive cost optimization, characterized in that: The optimization principle of the intelligent scheduling method includes the principle of minimizing the comprehensive cost. The comprehensive cost includes the waiting time cost and the demurrage cost, or includes the waiting time cost, the demurrage cost and the planned berthing cost. The optimization objective function of the ship intelligent scheduling method is: The intelligent scheduling method forms a berth allocation scheme by optimizing the objective function that satisfies the constraint conditions. The berth allocation scheme at least includes the time S for starting loading and unloading. kji , berth k, ship j information; in: ω 1j is the waiting time cost weight; k = 1, 2, ... M, is the berth set, M is the maximum number of berths; j = 1, 2, ... N, is the set of ships, N is the number of ships involved in the optimization calculation; i=1, 2, …P, is the service order of the ship, P is the maximum order number, P≤N; S kji is the time when loading and unloading starts at berth k and vessel j; A j is the time when the jth ship arrives; f 1j is the waiting cost per unit time of different ships; ω 2j is the demurrage cost weight; ΔT kj is the demurrage time of different ships, ΔT kj =max{0,S kji +t u -t j }, t j is the planned departure time of the ship, t u is the time taken to load and unload a ship based on experience; f 2kj It is the unit time demurrage cost corresponding to different ships and different berths; ω 3j is the planned berthing cost weight, and the optimization objective function satisfies oh 1j +oh 2j +oh 3j =1; F 3kj is the planned berthing cost of ships at different berths; The optimization objective function satisfies tidal constraints, which include: h ks For S kji The time corresponds to the water depth of the kth berth, d j, is the draft of the jth ship, ΔT 2j The distance to the next high tide when the water depth meets h ks ≥d j, The timing of the condition; The method obtains corresponding berth water depth information by dynamically tracking meteorological information; The optimization objective function satisfies the berth length constraint condition that the berth length is greater than the ship length; The optimization objective function satisfies the loading and unloading time constraint condition, which includes that the loading and unloading time period of the ship must be within the allowed time period, that is, t j-start ≤S kji ≤S kji +t u ≤t j-end , where t j-start and t j-end are the agreed loading and unloading time ranges for the jth ship respectively.

2. According to claim 1, a method for intelligent dispatching of bulk cargo ships in river ports based on comprehensive cost optimization is characterized in that: The optimization objective function satisfies the single berth allocation constraint, that is, each berth can only be allocated to one ship in the same time period.

3. According to the method of intelligent dispatching of bulk cargo ships in river ports based on comprehensive cost optimization according to claim 1, it is characterized in that: The objective function optimization includes any one or any combination of initial optimization, new ship entry optimization, on-demand optimization and stage optimization; the initial optimization starts the objective function optimization according to the berth usage and the ships that have entered the port; the new ship entry optimization includes starting the optimization when a new ship enters the port, including keeping the optimization results of the previous ships that have entered the port unchanged, optimizing the new ship that has entered the port alone, or re-optimizing the new ship that has entered the port together with the ships that have not yet berthed; the on-demand optimization includes restarting the optimization by manual intervention; the stage optimization includes automatically starting the objective function optimization according to a predetermined time interval or a limit on the number of new ships entering the port or a limit on the total number of ships to be berthed.

4. According to the method of intelligent dispatching of bulk cargo ships in river ports based on comprehensive cost optimization according to claim 1, it is characterized in that: The optimization method of the objective function includes an optimization method based on a genetic algorithm, comprising the following steps: S1. Establish objective function and constraints; S2. Generate berth allocation plan by coding; S3, initialize the population; S4, calculating the objective function and fitness function; S5. Select an operation; S6, cross operation; S7, mutation operation, generating a new population; S8, setting control parameters; S9. Repeat steps S4 to S7 until an end condition is met, wherein the end condition includes that the number of iterations reaches a set maximum number of iterations or the objective function calculation result meets the set objective function accuracy requirement or the fitness function result meets the set value requirement.

5. The intelligent dispatching method for bulk cargo ships in river ports based on comprehensive cost optimization according to claim 4 is characterized in that: The encoding in step S2 includes expressing the berth allocation scheme in integer code or binary code.

6. The intelligent dispatching method for bulk cargo ships in river ports based on comprehensive cost optimization according to claim 4 is characterized in that: Step S5 selects the operation using the fitness function and berth utilization rate as the comprehensive optimization principle.

7. The intelligent dispatching method for bulk cargo ships in river ports based on comprehensive cost optimization according to claim 4 is characterized in that: There is no order restriction for step S8 relative to steps S1 to S7.

8. The intelligent dispatching method for bulk cargo ships in river ports based on comprehensive cost optimization according to claim 4 is characterized in that: After the iterative calculation is completed, ω is assigned 1j ,ω 2j ,ω 3j Re-iterate the optimization with different combination values.

9. The intelligent dispatching method for bulk cargo ships in river ports based on comprehensive cost optimization according to claim 4 is characterized in that: The weight parameter ω 1j ,ω 2j ,ω 3j The combined values ​​of are taken as the optimal search options, and comprehensive optimization is performed on different weight combinations and berth allocation schemes.

Images