Intelligent scheduling method for wharf and port resources

Abstract

The invention discloses an intelligent scheduling method for wharf and port resources, and belongs to the technical field of port logistics optimization. An improved CSA algorithm is adopted, the problem of efficiency bottleneck of a traditional scheduling mode in a complex scene is solved, and the method comprises the steps of collecting a data set, setting constraint conditions, establishing the data set and a distribution matrix, establishing a multi-target fitness function, establishing a door machine scheduling rule, establishing a berth-door machine cooperative scheduling model and solving to obtain a scheduling scheme. The method mainly aims at minimizing the total waiting time of the ship and the equipment balance utilization rate, and meanwhile optimizes the berth utilization rate and the portal crane utilization rate. By constructing a multi-target scheduling model, a parking space allocation and portal crane scheduling scheme is dynamically planned, and efficient utilization of resources is ensured. The method has the advantages of optimizing the resource utilization rate and the like, and can be widely applied to the field of modern port logistics and management.

Description

Technical Field

[0001] The present invention relates to the technical field of port logistics optimization, in particular to an allocation method for joint resource scheduling, that is, an intelligent scheduling method for terminal port resources. Background Art

[0002] Dry bulk ports are important nodes in the maritime transportation system, and their ship loading operation efficiency directly affects the overall operation ability of the port and the dry bulk transportation level. In recent years, with the continuous development of the global economy, the cargo throughput of dry bulk ports has increased year by year, and the density of ships entering and leaving the port has also increased significantly. In this context, how to efficiently utilize the limited berth resources and gantry cranes in the port, reduce the ship's stay time in the port, and improve the overall operation efficiency has become the core issue in the management of dry bulk ports.

[0003] The scheduling mode of traditional ports usually relies on the experience of dispatchers to formulate berth allocation and ship loading operation plans. Dispatchers make manual decisions based on information such as ship forecasts, port resource status, and cargo demands. However, this experience-based scheduling mode has obvious limitations: (1) It is difficult to cope with complex port scenarios: When the number of incoming ships increases and the ship loading and unloading demands are diversified, the manual experience of dispatchers often cannot consider the optimal scheduling plan under multiple constraints, resulting in low resource utilization and long ship waiting times; (2) Lack of multi-device collaborative optimization ability: In actual ship loading operations, the collaborative efficiency of berths and gantry cranes directly determines the operation quality and time.

[0004] After retrieval, Chinese Patent Application No. 202010448803.9, with the application publication date of October 09, 2020, discloses a method, device, and medium for ship scheduling in a multi-berth port based on blockchain. The method includes: collecting information of ships to be operated; determining berth information in the blockchain; based on a preset intelligent contract, determining the berth for scheduling the ship to be operated according to the information of the ship to be operated and the berth information; writing the berth scheduling result into the blockchain. It is used to solve the problem that the management and scheduling of incoming ships in a multi-berth port lack orderliness, the incoming ships cannot complete operations in time, and are prone to accumulation, affecting the good management of the port. However, this method considers berth allocation or ship loading operations in isolation, ignores the collaborative needs among multiple devices, is difficult to ensure the overall operation efficiency, and does not comprehensively consider optimizing multiple objectives, easily causing unbalanced resource utilization.

[0005] Therefore, it is of great significance to design an efficient and flexible scheduling and allocation method for terminal berth resources. Summary of the Invention

[0006] 1. Problems to be Solved

[0007] The object of the present invention is to provide an intelligent scheduling method for terminal port resources, aiming to increase the dynamic cooperation requirements among multiple devices, effectively solve the limitations of existing methods, and improve resource utilization and scheduling efficiency.

[0008] 2. Technical solution

[0009] To solve the above problems, the technical solution adopted by the present invention is as follows:

[0010] An intelligent scheduling method for terminal port resources, which realizes the efficient allocation of berth and gantry crane resources by improving the CSA algorithm, reduces the waiting time of ships, and improves the utilization rate of berths and gantry cranes. It includes:

[0011] S1. Collect data sets, and collect basic berth data, basic weather information, gantry crane information, and basic information of incoming ships.

[0012] S2. Set constraint conditions, including ship length constraints, gantry crane allocation constraints, berth length constraints, ship operation constraints, weather constraints, and berth cargo type constraints.

[0013] S3. Establish data sets and allocation matrices, including ship sets, berth sets, gantry crane sets, berth allocation matrices, and gantry crane allocation matrices.

[0014] S4. Establish a multi-objective fitness function, which consists of ship waiting time, berth utilization rate, gantry crane utilization rate, conflict penalty, robustness, and dynamic weight; the role of the multi-objective fitness function is to calculate and compare in multiple iterations, and retain the minimum value in each iteration. The minimum value is a relatively optimal allocation scheme.

[0015] S5. Establish gantry crane scheduling rules.

[0016] S6. Establish a berth-gantry crane collaborative scheduling model with the total ship waiting time as the minimum objective function. After the berth-gantry crane collaborative scheduling model is solved, an allocation method is obtained. The solving process is as follows:

[0017] Input the data sets and allocation matrices in step S3 into the berth-gantry crane collaborative scheduling model. After ensuring that the solutions in the berth-gantry crane collaborative scheduling model meet the constraint conditions in step S2 and the gantry crane scheduling rules in step S5, use the multi-objective fitness function in step S4 to optimize the CSA algorithm, and then solve the berth-gantry crane collaborative scheduling model to obtain a scheduling plan.

[0018] As a possible implementation solution, in step S3, the data sets and allocation matrices include:

[0019] Ship set: S = {S1, S2, …, S N}, where S iDenote the \(i\)-th ship, where \(i\) is a positive integer less than or equal to \(N\);

[0020] Berth set: \(B=\{B_1,B_2,\ldots,B M \}\), where \(B j denotes the \(j\)-th berth, and \(j\) is a positive integer less than or equal to \(M\);

[0021] Gantry crane set: \(C = \{C_1,C_2,\ldots,C K \}\), where \(C k denotes the \(k\)-th gantry crane, and \(k\) is a positive integer less than or equal to \(K\);

[0022] Berth allocation matrix:

[0023] Gantry crane allocation matrix:

[0024] As a possible implementation solution, in step S6, the berth-gantry crane collaborative scheduling model meets the following assumptions:

[0025] (1) Each gantry crane continuously executes the material handling task for the same ship;

[0026] (2) Ship requirements: size, weight, quantity, arrival time, operation time;

[0027] (3) After the ship loading operation is completed, it immediately applies for departure.

[0028] As a possible implementation solution, in step S6, the solving process includes:

[0029] S61. Population initialization: Encode the berth and gantry crane scheduling scheme as a solution vector \(X=\{x_1,x_2,\ldots,x d \}\), and each component in the solution vector \(X\) includes a berth allocation matrix and a gantry crane allocation matrix;

[0030] Randomly generate an initial population \(P_0=\{X_1,X_2,\ldots,X D \}\), randomly generate \(D\) berth allocation schemes, and each scheme is represented by three-layer real number encoding:

[0031] [Berth number, ship loader number, operation time period]

[0032] Each individual encoding represents a set of specific berth allocation schemes;

[0033] Introduce a hybrid initialization strategy: Each solution vector \(X D represents the berth and gantry crane allocation scheme for each ship, and it is required that at least 10% of the solutions in the initial population \(P_0\) be generated by heuristic rules to improve the initial quality of the population;

[0034] S62. Detect the solutions randomly generated or generated by heuristic rules in the initial population P0:

[0035] If there are solutions that do not meet the constraint conditions of step S2 and the gantry crane scheduling rules of step S5, repair or remove the invalid solutions to obtain the detected population P;

[0036] S63. Iteration and optimization: Optimize the CSA algorithm using the multi-objective fitness function in step S4, and perform T iterations on the detected population P to output the optimal solution X BEST ; Specifically:

[0037] Judge the current iteration status to determine whether to continue the iteration;

[0038] If the current iteration number t is equal to T max , then jump out of the loop and enter the result output step;

[0039] If the current iteration number t is less than T max , then continue to execute the update step;

[0040] If the global optimal solution remains unchanged in consecutive T iterations (i.e., the fitness value is not further optimized), it is considered that the algorithm has converged and the loop is jumped out;

[0041] If the global optimal solution continues to be optimized, then continue to execute the next step; When any of the following termination conditions for iteration is met (reaching the maximum iteration number T max ; The global optimal solution has no significant change in consecutive K iterations; The fitness values of the population individuals tend to be the same.), the algorithm terminates and outputs the global optimal berth allocation plan. That is, output the final berth and gantry crane allocation plan, including the berth number, gantry crane number and their corresponding operation time periods of each ship.

[0042] If the fitness values among the population individuals tend to be the same (i.e., the diversity decreases significantly), the algorithm may fall into a local optimum;

[0043] At this time, randomly replace some individuals through the nest abandonment strategy (optimized cas algorithm) to increase the population diversity and avoid falling into a local optimum.

[0044] As a possible implementation solution, in step S6, the CSA algorithm includes a Lévy flight mechanism and a discovery mechanism. Among them, the Lévy flight mechanism is optimized to increase random perturbation and change ability. The specific optimization process of the Lévy flight mechanism is:

[0045] X new = X current +α′·Lévy(λ)+e·U(0,1)

[0046]

[0047] wherein, X new is the new solution after iteration; X current is the solution before iteration; α′ is the step size scaling factor, which is dynamically adjusted according to the number of iterations; Lévy(λ) is the flight mechanism function before optimization; e and β are perturbation factors; U(0,1) is a random variable uniformly distributed between [0,1]; α0 is the step size scaling factor before optimization; α′ is the step size scaling factor after optimization; t is the current number of iterations; T is the number of iterations set in step S63; p is the exponential decay parameter, p ∈ [0.5, 5]; N(0,1) is a random variable with a normal distribution between [0,1].

[0048] In the case of adopting the above technical solution, the original CSA algorithm is a fixed value. Now, exponential fractional linear decay and Gaussian perturbation are introduced because the original CSA algorithm is prone to weak local search ability, and the local search ability is improved to discover the global optimal solution.

[0049] As a possible implementation solution, in step S6, the CSA algorithm includes a Lévy flight mechanism and a discovery mechanism. The unoptimized discovery mechanism is a fixed value, which is improved to a sin wave function here. The purpose is to utilize the special effects of the periodic function to dynamically change the cuckoo search ability. Specifically, the optimized discovery mechanism is as follows:

[0050] The discovery mechanism is reflected by the following non-linear dynamic optimization formula:

[0051] P discover = P1·(1 + sin(π·t / T))

[0052] wherein, P discover is the discovery mechanism after optimizing the CSA algorithm; P1 is the discovery mechanism before optimization, set to 0.2 - 0.5; t is the current number of iterations; T is the number of iterations set in step S63.

[0053] In the case of adopting the above technical solution, according to the discovery probability P discover , the solution with a relatively poor fitness value is randomly replaced to generate a new solution; if the population diversity decreases, the population quality is enhanced by randomly replacing some solutions. At the end of each iteration, the individual with the lowest fitness value in the current population is set as the global optimal solution of this round, and the global optimal record is updated.

[0054] As a possible implementation solution, in step S4, the total fitness of the multi-objective fitness function comprehensively considers five components. In the prior art, the total fitness only considers a single objective, and at the same time, non-linear changes are adopted inside to replace the original fixed or linear changes to increase the performance of the algorithm in terms of search. And the multi-objective fitness function of this application is

[0055]

[0056] Among them, the lower the F value, the better the solution; the higher the value, it indicates that there is greater room for improvement in certain aspects of the solution (such as waiting time, resource utilization rate, conflicts, etc.).

[0057] In the formula, w i normalized is the dynamic weight, w i normalized ∈[0,1], and T wait is the waiting time of the ship, a represents the penalty coefficient. When T wait the waiting time is small, its impact is weak; when the waiting time is too long, the penalty is amplified; Y berth is the berth utilization rate, ∈ represents a defined constant, close to 0, used to control the amplitude of the penalty. When the non-utilization rate is very small (1 - Y berth ≤∈), the penalty value is limited to the fixed value ∈ = 1 to avoid the penalty value increasing infinitely; Y crane is the gantry crane utilization rate; P conflict is the conflict penalty term, R robustness is the robustness, represents reducing the impact on fitness when the robustness is poor, maintaining the penalty for high robustness, and is used to balance the impact of each sub-goal on fitness.

[0058] As a possible implementation solution, the waiting time T of the ship wait is

[0059]

[0060] In the formula, T arrival,i is the arrival time of ship i; T service,i is the operation time of ship i;

[0061] The lower the value of the ship waiting time T wait , it indicates that the waiting time of the ship in the port is shorter and the scheduling solution is more efficient. The higher the value of T wait , it indicates poor scheduling and low resource allocation efficiency. The later the departure time departure,i , it may mean low operation efficiency or long waiting time.

[0062] The berth utilization rate Y berth is

[0063]

[0064] In the formula, T total is the total available time of the berth, T berth,i is the operation time of ship i, is the actual berth usage time, and the berth utilization rate is U berth which is used to measure the usage efficiency of berth resources, with a range of [0, 1];

[0065] If 1 - U berth ≤ ∈ (e.g., ∈ = 0.01), then set to represent the penalty for the low under - utilization (i.e., high utilization of 95%) scenario, add the calculation penalty for the high - load scenario, and prioritize optimizing the resource allocation plan.

[0066] The gantry crane utilization rate U crane is

[0067]

[0068] In the formula, T crane,k is the working time of gantry crane k, and T total is the total available time of the gantry crane; when the value of U crane is close to 1, it indicates that the gantry crane resources are efficiently utilized, and when the value of U crane is close to 0, it indicates that the gantry crane is idle and the efficiency is low;

[0069] The conflict penalty term P conflict is added as a comprehensive conflict for the conflict item consideration factor to increase the rationality of allocation. Specifically,

[0070]

[0071] In the formula, Q ij ·Y ik represents that ship S i is assigned to berth B j and at the same time is assigned to gantry crane C k , Conflict ijk represents whether this allocation generates a conflict, which is 0 or 1; the meaning of the conflict penalty term P conflict formula is: accumulate the number of conflicts that occur in all allocation combinations of ships, berths, and gantry cranes. The higher the value of P conflict , the more conflicts there are, and the worse the scheduling plan. When there is no conflict, P conflict = 0, which is the ideal state.

[0072] The present invention adds robustness analysis to judge the adaptability level of the plan. The robustness R robustness is

[0073]

[0074] In the formula, T actual,i is the actual arrival time of ship i, T planned,i : the planned arrival time of ship i, and N is the total number of ships. The robustness Rrobustness The lower the value, the stronger the adaptability of the solution to the fluctuations in the actual arrival time, and the robustness R robustness The higher the value, the more sensitive the solution is to random changes and the poorer the robustness.

[0075] The present invention adds a dynamic weight to dynamically change the proportion of each part of the objective function during iteration. The purpose is to increase the priority of a certain objective in the initial stage and decrease the priority of a certain objective in the later stage. The dynamic weight w i normalized is

[0076]

[0077] In the formula, t is the current iteration number, γ is the rate of decrease of the weight over time, and b r where r is 1, 2, 3, 4 or 5, representing the ship waiting time, berth utilization rate, gantry crane utilization rate, conflict penalty, and robustness respectively, and b r is a constant.

[0078] As a possible implementation solution, in step S5, the process of establishing the gantry crane scheduling rule includes the following steps:

[0079] Define the ship set S = {S1, S2,..., S N} and the gantry crane set C = {C1, C2,..., C K}, and the present invention sets to calculate the ship priority. Considering the waiting time, weight, and operation complexity, comprehensively calculate the priority to avoid often single waiting time or cargo type. Calculate the task priority Priority(i) of the ship according to the following formula:

[0080]

[0081] In the formula: T wait,i is the waiting time of ship i, W i is the total cargo weight of ship i, and Di is the draft of the ship, indicating the operation complexity;

[0082] Sort the ships according to the priority Priority(i), and preferentially allocate gantry crane resources to high-priority ships;

[0083] The maximum capacity of each gantry crane k is C max,k , count the total task volume N crane,k currently allocated to the gantry crane, and determine whether the following restrictions are met:

[0084]

[0085] If the gantry capacity is exceeded, the following dynamic adjustment mechanism is triggered: reassign the excess tasks to other idle gantries; if there are no idle gantries, prioritize adjusting the operation plans of ships with low priorities;

[0086] Combined with the gantry utilization rate U crane,k , the present invention defines a dynamic adjustment strategy, continuously makes judgments during the algorithm, and increases the rationality of algorithm allocation. The specific dynamic task allocation strategy for the gantry is as follows:

[0087]

[0088] In the formula, T assign,k is the remaining task time that the gantry k can allocate, U threshold is the lowest threshold of the gantry utilization rate (such as 50%), T max,k is the maximum time that the gantry k can allocate, and T used,k is the time that the gantry k has already used.

[0089] 3. Beneficial effects

[0090] Compared with the prior art, the beneficial effects of the present invention are as follows:

[0091] (1) By constructing a berth-gantry collaborative scheduling model, the present invention closely combines berth allocation and gantry scheduling. Based on factors such as ship requirements, gantry capacity, and operation time, dynamic collaborative arrangements are made to avoid operation delays caused by improper equipment allocation;

[0092] (2) The present invention adopts an improved CSA algorithm, incorporates a hybrid initialization strategy, improves the quality of the initial population, and makes the starting point of the algorithm better. At the same time, various constraint conditions, such as weather and special ship requirements, are fully considered, which can improve the comprehensive management ability of the port to cope with complex environments. Description of the drawings

[0093] The drawings are not intended to be drawn to scale. In the drawings, each identical or approximately identical component shown in each figure may be represented by the same reference numeral. For clarity, not every component is labeled in each figure. Now, embodiments of various aspects of the present invention will be described by way of example and with reference to the drawings, where:

[0094] Figure 1 is the flow framework diagram of the solution process of the berth-gantry collaborative scheduling model of the present invention;

[0095] Figure 2 is the specific flow schematic diagram of the solution process of the berth-gantry collaborative scheduling model in the embodiment of the present invention;

[0096] Figure 3 is the three-layer real number coding schematic diagram of each scheme in the initial population in the embodiment of the present invention;

[0097] Figure 4 This is the trend chart of the fitness function F in the embodiments of the present invention. Specific embodiments

[0098] To make the objectives, technical solutions, and advantages of the present technical solution clearer, the present technical solution will be further described in detail below in conjunction with specific embodiments. It should be understood that these descriptions are exemplary and are not intended to limit the scope of the present technical solution.

[0099] In a method for intelligent scheduling of terminal port resources according to the present invention, the operation mode of the portal crane includes assigning a portal crane for loading and unloading tasks after a ship arrives at the port. Through a reasonable berth-portal crane collaborative scheduling model, the berthing time of the ship can be reduced, and the turnover efficiency of the port can be improved. In this process, berth allocation and portal crane scheduling restrict each other: the allocation of berths determines the operation scope of the portal crane, and the operation efficiency of the portal crane in turn affects the turnover speed of the berth, achieving efficient allocation and optimization of resources. The specific method includes:

[0100] S1. Collect a data set, and collect basic berth data, basic weather information, portal crane information, and basic information of incoming ships.

[0101] S2. Set constraint conditions, including ship length constraints, portal crane allocation constraints, berth length constraints, ship operation constraints, weather constraints, and berth cargo type constraints.

[0102] S3. Establish a data set and allocation matrices, including a ship set, a berth set, a portal crane set, a berth allocation matrix, and a portal crane allocation matrix; the data set and allocation matrices include:

[0103] Ship set: S = {S1, S2, …, S N}, where S i represents the i-th ship, and i is a positive integer less than or equal to N;

[0104] Berth set: B = {B1, B2, …, B M}, where B j represents the j-th berth, and j is a positive integer less than or equal to M;

[0105] Portal crane set: C = {C1, C2, …, C K}, where C k represents the k-th portal crane, and k is a positive integer less than or equal to K;

[0106] Berth allocation matrix:

[0107] Portal crane allocation matrix:

[0108] S4. Establish a multi-objective fitness function:

[0109]

[0110] In the formula, w i normalized is the dynamic weight, w i normalized ∈ [0, 1], and T wait is the waiting time of the ship, a represents the penalty coefficient; U berth is the berth utilization rate, ∈ represents a defined constant, close to 0, used to control the amplitude of the penalty; U crane is the gantry crane utilization rate; P conflict is the conflict penalty term, R robustness is the robustness.

[0111] The above multi-objective fitness function consists of the ship waiting time, berth utilization rate, gantry crane utilization rate, conflict penalty, robustness, and dynamic weight. The ship waiting time T wait is

[0112]

[0113] In the formula, T arrival,i is the arrival time of ship i; T service,i is the operation time of ship i;

[0114] The berth utilization rate U berth is

[0115]

[0116] In the formula, T total is the total available time of the berth, T berth,i is the operation time of ship i, is the actual use time of the berth. The berth utilization rate U berth is used to measure the utilization efficiency of berth resources, range [0, 1];

[0117] The gantry crane utilization rate U crane is

[0118]

[0119] In the formula, T crane,k is the operation time of gantry crane k, T total is the total available time of the gantry crane; U crane When the value is close to 1, it means that the gantry crane resources are efficiently utilized. When U crane is close to 0, it means that the gantry crane is idle and the efficiency is low;

[0120] The conflict penalty term P conflict is

[0121]

[0122] In the formula, Q ij ·Y ik represents that the ship S i is assigned to the berth B j and at the same time is assigned to the gantry crane C k , Conflict ijk represents whether this assignment generates a conflict, and is 0 or 1;

[0123] The robustness R robustness is

[0124]

[0125] In the formula, T actual,i is the actual arrival time of ship i, T planned,i : the planned arrival time of ship i, and N is the total number of ships;

[0126] The dynamic weight w i normalized is

[0127]

[0128] In the formula, t is the current iteration number, γ is the rate of controlling the weight reduction over time, b r where r is 1, 2, 3, 4, or 5, representing the ship waiting time, berth utilization rate, gantry crane utilization rate, conflict penalty, and robustness respectively, b r is a constant.

[0129] S5. Establish the gantry crane scheduling rule:

[0130] Define the ship set S = {S1, S2,..., S N} and the gantry crane set C = {C1, C2,..., C K}, and calculate the task priority Priority(i) of the ship according to the following formula:

[0131]

[0132] In the formula: T wait,i is the waiting time of ship i, W i is the total cargo weight of ship i, and Di is the draft of the ship;

[0133] Sort the ships according to the priority Priority(i), and give priority to allocating gantry crane resources to high-priority ships;

[0134] The maximum capacity of each gantry crane k is C max,k , count the total task volume N currently assigned to the gantry crane crane,k , and determine whether the following restrictions are met:

[0135]

[0136] If it exceeds the capacity of the gantry crane, trigger the following dynamic adjustment mechanism: reassign the excess tasks to other idle gantry cranes; if there are no idle gantry cranes, give priority to adjusting the operation plan of ships with low priority;

[0137] Combined with the gantry crane utilization rate U crane,k , define the dynamic task allocation strategy of the gantry crane:

[0138]

[0139] In the formula, T assign,k is the remaining task time that gantry crane k can allocate, U threshold is the lowest threshold of the gantry crane utilization rate, T max,k is the maximum time that gantry crane k can allocate, T used is the time that gantry crane k has already used.

[0140] S6. Establish a berth-gantry crane collaborative scheduling model, and the berth-gantry crane collaborative scheduling model meets the following assumptions:

[0141] (1) Each gantry crane continuously executes the material handling task of the same ship;

[0142] (2) Ship requirements: size, weight, quantity, arrival time, operation time;

[0143] (3) After the ship loading operation is completed, it immediately applies to leave the port.

[0144] The allocation method is obtained after solving the berth-gantry crane collaborative scheduling model, and the solving process is as follows:

[0145] S61. Population initialization: Encode the berth and gantry crane scheduling scheme as a solution vector X = {x1, x2,..., x d}; each component in the solution vector X includes a berth allocation matrix and a gantry crane allocation matrix;

[0146] Randomly generate an initial population P0 = {X1, X2,..., X D}; introduce a hybrid initialization strategy: each solution vector X D represents the berth and gantry crane allocation scheme of each ship, and it is required that at least 10% of the solutions in the initial population P0 be generated through heuristic rules to improve the initial quality of the population;

[0147] S62. Detect the solutions randomly generated or generated by heuristic rules in the initial population P0:

[0148] If there are solutions that do not meet the constraint conditions of step S2 and the gantry crane scheduling rules of step S5, repair or remove the invalid solutions to obtain the detected population P;

[0149] S63. Iteration and optimization: Optimize the CSA algorithm using the multi-objective fitness function in step S4, perform T iterations on the detected population P, and output the optimal solution X BEST .

[0150] In step S63, the CSA algorithm includes a Lévy flight mechanism and a discovery mechanism, where:

[0151] The optimization process of the Lévy flight mechanism is:

[0152] X new = X current + α′·Lévy(λ)+ e·U(0,1)

[0153]

[0154] In the formula, X new is the new solution after iteration; X current is the solution before iteration; α′ is the step size scaling factor, dynamically adjusted according to the number of iterations; Lévy(λ) is the flight mechanism function before optimization; e, β are perturbation factors; U(0,1) is a random variable uniformly distributed between [0,1]; α0 is the step size scaling factor before optimization; α′ is the step size scaling factor after optimization; t is the current iteration number; T is the iteration number set in step S63; p is the exponential decay parameter, p ∈ [0.5, 5]; N(0,1) is a random variable normally distributed between [0,1].

[0155] The optimized discovery mechanism is:

[0156] P discoveer = P1·(1 + sim(π·t / T))

[0157] In the formula, P discover is the discovery mechanism after optimizing the CSA algorithm; P1 is the discovery mechanism before optimization, set to 0.2 - 0.5; t is the current iteration number; T is the iteration number set in step S63.

[0158] Embodiment

[0159] In this embodiment, the goal of the berth - gantry crane collaborative scheduling model is to optimize multiple goals including minimizing ship waiting time, berth utilization rate, gantry crane utilization rate, conflict penalty value, and robustness. The constraint conditions are as follows:

[0160] Berth length constraint: Each ship can only berth at one berth:

[0161]

[0162] Ship length constraint: The ship length cannot exceed the berth length:

[0163]

[0164] The first ship operation constraint: The draft of the ship cannot exceed the maximum water depth of the berth:

[0165]

[0166] The second ship operation constraint: The time of two ships at the same berth cannot overlap:

[0167]

[0168] Gantry crane allocation constraint: Each ship is allocated at most one gantry crane:

[0169]

[0170] Weather constraint: Work in normal weather and do not work in extreme weather, including severe weather such as heavy rain and heavy snow;

[0171] Berth cargo type constraint: The information of the cargo received by the berth matches the cargo on the ship.

[0172] Based on the above constraints and the multi-objective fitness function F, a berth-gantry crane collaborative scheduling model is established. According to Figure 1 the solution process of the berth-gantry crane collaborative scheduling model shown, using the above data set, the berth-gantry crane collaborative scheduling model is solved by an algorithm to obtain the berth allocation plan for each berth, that is, the optimal solution X BEST , specifically including the following steps:

[0173] The ship coding set is:

[0174] S = {S1, S2, …, S7}

[0175] Among them, each small set in the S set includes the following data:

[0176] The ship length set is:

[0177] Length SHIP = {40, 50, 60, 70, 80, 90, 100}

[0178] The ship priority set is:

[0179] Order ship={200, 300, 400, 500, 600, 700, 1100}

[0180] The ship load set is

[0181] Weight berth ={200, 400, 600, 800, 900, 1000, 1200}

[0182] The berth set: B = {B1, B2, …, B5}, including the following:

[0183] The berth load set is

[0184] Cargo berth ={3000, 4000, 5000, 7000, 9000}

[0185] The portal crane set is

[0186] C = {C1, C2, …, C7}

[0187] Set the CSA algorithm parameters, including: the population size N = 200, the number of iterations T max = 900, the rate P = 0.2. For the CSA algorithm, optimize and solve the berth-portal crane collaborative scheduling model as Figure 2 shown. The ships to be scheduled are shown in Table 1 below:

[0188] Table 1 Information Table of Ships to be Scheduled

[0189] Vessel Number Captain Deadweight Cargo Type Loading and Unloading Duration s001 20 400 Bulk Cargo 1 s002 40 1000 Steel Bar 2 s003 60 2000 Steel Bar 3 s004 80 3000 Mineral Powder 4 s005 80 3000 Mineral Powder 4 s006 100 5000 Steel Bar 8 s007 100 5000 Steel Bar 8

[0190] Generate the initial population: Generate D = 200 allocation schemes, among which 20 are generated by heuristic rules. Each scheme is represented by three-layer real number coding, and each individual coding represents a set of specific berth allocation schemes, such as Figure 3 .

[0191] Check whether the generated population meets the above constraints (such as berth length, ship draft, etc.)

[0192] Repair or remove the non-conforming schemes. After removal, new schemes need to be supplemented to obtain the detected population P, including 200 allocation schemes.

[0193] At the same time, conduct the portal crane scheduling rule judgment, calculate the ship priority of the current solution, and dynamically adjust the portal crane allocation according to the rules

[0194]

[0195] Such as:

[0196] Gantry crane 1: Current load: 140 tons (Vessel A: 80 tons, Vessel B: 60 tons).

[0197] Adjustment strategy: Reallocate Vessel B (60 tons) to gantry crane 3 with a lower load.

[0198] After adjustment: Gantry crane 1: Vessel A (80 tons) → Total load 80 tons.

[0199] Gantry crane 3: Vessel E (50 tons), Vessel B (60 tons) → Total load 110 tons (overloaded).

[0200] After multiple adjustments, the overloading situation is completely eliminated.

[0201] Calculate the multi-objective fitness function F value for each allocation plan. The fitness function is:

[0202]

[0203] If calculating each component in sequence and then summing them up:

[0204] Sort the population according to the fitness F value. Based on the different F values obtained after each iteration, record the historical fitness set fintness = {250, 250, 249, 249, 250, 248.......} and the current optimal solution X best . Compare according to the formula to check if there is X new <X best , if there is, update the new fitness to F best

[0205] Generate a new solution using the optimized Lévy flight formula: X new = X current + α′·Lévy(λ) + e·U(0, 1)

[0206] For example, if the original solution [1, 2, 9 - 10] flies to generate a new solution, it can be [3, 4, 8 - 9]

[0207] Check if the new solution meets the constraint conditions. If not, repair or discard it.

[0208] Calculate the discovery mechanism probability at the current iteration, such as P discover = 0.2, and randomly replace the solution with a poorer fitness value;

[0209] Judge the current iteration status to determine whether to continue the iteration;

[0210] If the current iteration number t is equal to T max = 900, then jump out of the loop and enter the result output step;

[0211] If the current iteration number t is less than T max = 900, then continue to execute the update step;

[0212] If the global optimal solution remains unchanged in K consecutive iterations (i.e., the fitness value is not further optimized), it is considered that the algorithm has converged and the loop is exited;

[0213] If the global optimal solution continues to be optimized, then continue to execute the next step.

[0214] If the fitness values among the population individuals tend to be the same (i.e., the diversity significantly decreases), the algorithm may fall into a local optimum;

[0215] At the end of each iteration, set the individual with the lowest fitness value in the current population as the global optimal solution of this iteration and update the global optimal record.

[0216] When any iteration termination condition is met (reaching T max , the global optimal solution is stable or the population diversity is too low), the algorithm terminates, and the global optimal berth allocation scheme is output as shown in Table 2, and the iteration target fitness function F curve is as Figure 4 shown. It can be seen from the indicators of the optimized CSA algorithm that when setting 900 iterations, the fitness curve will decline every time after a certain number of iterations, indicating that a better solution X best is found.

[0217] Before optimization, the CSA algorithm usually converges within 200 - 300 times, indicating that the berth - gantry crane collaborative scheduling model of the present invention can obtain a long - term convergence opportunity and obtain a higher - quality global solution.

[0218] Table 2 Optimal Berth Allocation Scheme Table

[0219] Vessel Number Allocated Berth Allocated Gantry Crane Operating Time Waiting Time s001 2 M1 9:00-10:00 0 s002 3 M2 9:00-11:00 0 s003 5 M5 9:00-12:00 0 s004 1 M7 9:30-13:30 0 s005 4 M3 9:30-13:30 0 s006 3 M4 11:00-19:00 1.5 s007 2 M6 11:00-19:00 1

[0220] The above is only an embodiment of this specification and is not used to limit this specification. For those skilled in the art, various changes and modifications can be made to this specification. Any modification, equivalent replacement, improvement, etc. made within the spirit and principle of this specification shall be included within the scope of the claims of this specification.

Claims

1. An intelligent scheduling method for terminal port resources, characterized in that: Including: S1. Collect a data set, and collect basic berth data, basic weather information, gantry crane information, and basic information of incoming ships. S2. Set constraint conditions, including ship length constraint, gantry crane allocation constraint, berth length constraint, ship operation constraint, weather constraint, and berth cargo type constraint. S3. Establish a data set and allocation matrix, including a ship set, a berth set, a gantry crane set, a berth allocation matrix, and a gantry crane allocation matrix. S4. Establish a multi-objective fitness function, which consists of ship waiting time, berth utilization rate, gantry crane utilization rate, conflict penalty, robustness, and dynamic weight. S5. Establish gantry crane scheduling rules. S6. Establish a berth-gantry crane collaborative scheduling model. After solving the berth-gantry crane collaborative scheduling model, an allocation method is obtained. The solving process is as follows: Input the data set and allocation matrix in step S3 into the berth-gantry crane collaborative scheduling model. After ensuring that the solution in the berth-gantry crane collaborative scheduling model meets the constraint conditions in step S2 and the gantry crane scheduling rules in step S5, optimize the CSA algorithm using the multi-objective fitness function in step S4, and then solve the berth-gantry crane collaborative scheduling model to obtain a scheduling plan.

2. The intelligent scheduling method for terminal port resources according to claim 1, wherein: In step S3, the data set and allocation matrix include: Set of ships: S = {S1, S2, …, S N}, where S i represents the i-th ship, and i is a positive integer less than or equal to N; Berth set: B = {B1, B2, …, B M}, where B j represents the j-th berth, and j is a positive integer less than or equal to M; Door machine set: C = {C1, C2, …, C K}, where C k represents the k-th door machine, and k is a positive integer less than or equal to K; Berth allocation matrix: Door machine distribution matrix:

3. The intelligent scheduling method for dock and port resources according to claim 1, characterized in that: In step S6, the berth-gantry crane collaborative scheduling model meets the following assumption conditions: (1) Each gantry crane continuously executes the material handling task of the same ship. (2) Ship requirements: size, weight, quantity, arrival time, operation time. (3) After the ship loading operation is completed, it immediately applies to leave the port.

4. The intelligent scheduling method for dock and port resources according to claim 3, wherein: In step S6, the solving process includes: S61. Population initialization: Encode the berth and gantry crane scheduling scheme into a solution vector X = {x1, x2, …, x d}, and each component in the solution vector X includes a berth allocation matrix and a gantry crane allocation matrix; Randomly generate the initial population P0 = {X1, X2, …, X D}, and introduce a hybrid initialization strategy: each solution vector X D represents the berth and gantry crane allocation plan for each ship. It is required that at least 10% of the solutions in the initial population P0 be generated by heuristic rules to improve the initial quality of the population; S62. Detect the solutions randomly generated or generated by heuristic rules in the initial population P0: If there are solutions that do not meet the constraint conditions in step S2 and the gantry crane scheduling rules in step S5, repair or remove the invalid solutions to obtain the detected population P. S63. Iteration and Optimization: Optimize the CSA algorithm using the multi-objective fitness function in step S4, perform T iterations on the detected population P, and output the optimal solution X BEST .

5. The intelligent scheduling method for dock and port resources according to claim 4, characterized in that: In step S6, the CSA algorithm includes a Lévy flight mechanism and a discovery mechanism. Among them, the optimization process of the Lévy flight mechanism is: X new = X current + α′·Lévy(λ) + e·U(0, 1) where X new is the new solution after iteration; X current is the solution before iteration; α′ is the step size scaling factor, which is dynamically adjusted according to the number of iterations; Lévy(λ) is the flight mechanism function before optimization; e and β are perturbation factors; U(0,1) is a random variable uniformly distributed between [0,1]; α0 is the step size scaling factor before optimization; α′ is the step size scaling factor after optimization; t is the current number of iterations; T is the number of iterations set in step S63; p is the exponential decay parameter, p ∈ [0.5, 5]; N(0,1) is a random variable normally distributed between [0,1].

6. The intelligent scheduling method for terminal port resources according to claim 4, characterized in that: In step S6, the CSA algorithm includes a Lévy flight mechanism and a discovery mechanism. The optimized discovery mechanism is: P discover = P1·(1 + sin(π·t / T)) Wherein, P discover is the discovery mechanism optimized by the CSA algorithm; P1 is the discovery mechanism before optimization, set to 0.2 to 0.5; t is the current iteration number; T is the iteration number set in step S63.

7. A method for intelligent scheduling of terminal port resources according to claim 2, characterized in that: In step S4, the multi-objective fitness function is where w i normalized is the dynamic weight, w i normalized ∈ [0, 1], and T wait is the waiting time of the ship, a represents the penalty coefficient; U berth is the berth utilization rate, ∈ represents a defined constant, close to 0, used to control the amplitude of the penalty; U crane is the gantry crane utilization rate; P conflict is the conflict penalty term, R robustness is the robustness.

8. According to the method for intelligent scheduling of terminal port resources according to claim 7, characterized in that: The waiting time T of the ship wait is where T arrival,i is the arrival time of ship i; T service,i is the operation time of vessel i; The described berth utilization rate U berth is Where, T total is the total available time of the berth, and T berth,i is the operation time of ship i, is the actual use time of the berth. The berth utilization rate U berth is used to measure the utilization efficiency of berth resources, with a range of [0, 1]; The utilization rate U of the door machine crane is where T crane,k is the working time of gantry crane k, and T total is the total available time of the gantry crane; U crane When the value is close to 1, it indicates that the door machine resources are efficiently utilized, U crane When the value is close to 0, it indicates that the door machine is idle and inefficient; The conflict penalty term P conflict is Wherein, Q ij ·Y ik represents that the ship S i is assigned to the berth B j and is simultaneously assigned to the gantry crane C k , Conflict ijk represents whether such an assignment results in a conflict, being 0 or 1; The robustness R robustness is where, T actual,i is the actual arrival time of ship i, and T planned,i : is the planned arrival time of ship i, and N is the total number of ships; The dynamic weight w i normalized is Where \(t\) is the current iteration number, \(\gamma\) is the rate controlling the decrease of the weight over time, and \(b\) r where \(r = 1, 2, 3, 4\) or \(5\), representing the ship waiting time, berth utilization rate, quay crane utilization rate, conflict penalty, and robustness respectively, and \(b\) r is a constant.

9. The intelligent scheduling method for terminal port resources according to claim 1, wherein: In step S5, the process of establishing gantry crane scheduling rules includes: Define the ship set S = {S1, S2, …, S N} and the gantry crane set C = {C1, C2, …, C K}, and calculate the task priority Priority(i) of the ship according to the following formula: Where: T wait,i is the waiting time of ship i, W i is the total cargo weight of ship i, and Di is the draft of the ship; Sort the ships according to the priority Priority(i), and preferentially allocate gantry crane resources to ships with high priority. The maximum capacity of each door machine k is C max,k , count the total task volume N currently assigned to the door machine crane,k , and determine whether the following restrictions are met: If the gantry crane capacity is exceeded, trigger the following dynamic adjustment mechanism: reallocate the excess tasks to other idle gantry cranes; if there are no idle gantry cranes, preferentially adjust the operation plans of ships with low priority. Combined with the utilization rate U of the gantry crane crane,k , define the dynamic task allocation strategy of the gantry crane: Where, T assign,k is the remaining task time allocable to gantry crane k, U threshold is the minimum threshold of the utilization rate of the gantry crane, T max,k is the maximum time allocable to gantry crane k, T used is the time already used by gantry crane k.

Images