A harbor infrastructure emergency resource hierarchical scheduling optimization method and system

Abstract

The application discloses a kind of harbor infrastructure emergency resource hierarchical scheduling optimization method and system, belong to port safety management and intelligent decision-making technical field.The method includes: obtaining multi-source data constructs feature vector;Risk coupling degree coefficient with space attenuation factor is calculated, and the effect of the superposition of multiple risks is quantified;A multi-objective mixed integer programming model is constructed with the shortest response time and the lowest disposal cost as the goal, and the scheduling inertia penalty and resource coordination penalty are introduced;Genetic algorithm is used for solving, and the coupling coefficient is embedded into the fitness and the crossover probability is self-adaptive;The candidate scheme is screened by Pareto non-dominated verification;The scheduling is dynamically adjusted in combination with real-time risk update.The application overcomes the defect of isolated risk assessment, and improves the dynamic adaptability and resource coordination of the scheme.

Description

Technical Field

[0001] This invention relates to the field of port safety management and intelligent decision-making technology, and in particular to a hierarchical scheduling optimization method and system for emergency resources of seaport infrastructure. Background Technology

[0002] As a core node in the seaport logistics system, the safe operation of container terminals directly impacts regional economic stability and the security of international trade supply chains. Container terminals are equipped with numerous large-scale loading and unloading equipment, including quay cranes, rubber-tired gantry cranes, and rail-mounted gantry cranes. Large container ships moored at the berths are connected to the terminal structure via mooring cables. Under extreme weather conditions such as typhoons and storm surges, terminals face multiple risks: strong winds and high waves may cause violent collisions between moored vessels and berths, resulting in structural damage to the terminal and damage to ships; quay cranes may collapse or derail under strong winds; and rubber-tired gantry cranes are prone to slippage or overturning in strong winds. Following such emergencies, it is necessary to rapidly deploy various emergency resources, such as spare parts, maintenance personnel, and emergency supplies, to the incident site. Therefore, how to quickly and economically deploy emergency resources from various reserve points to the point of need in scenarios involving multiple overlapping risks caused by extreme weather is one of the core issues in the field of container terminal emergency management.

[0003] Currently, emergency resource scheduling schemes at container terminals mainly rely on emergency plans combined with human experience. Scheduling decisions typically prioritize minimizing response time as the sole optimization objective, independently allocating resources from each reserve point to each demand point. Some existing technologies introduce linear programming or integer programming models to solve for the optimal scheduling quantity given the total resource amount and transportation routes. Other research employs heuristic algorithms such as genetic algorithms and particle swarm optimization to optimize scheduling schemes. In recent years, some scholars have attempted to consider the balance between multiple optimization objectives, constructing multi-objective optimization models and using Pareto analysis to filter non-dominated solution sets. Furthermore, a few existing methods incorporate real-time monitoring data, adjusting the scheduling scheme by periodically updating input parameters.

[0004] The existing technologies still have the following shortcomings. Existing methods generally treat each risk source as an independent event, for example, considering the risk of ship-berth collision and the risk of large machinery collapse separately. This ignores the coupling and amplifying effects of multiple risk sources existing simultaneously in extreme weather conditions within a container terminal area—strong winds increase both the probability of ship collisions with berths and the risk of crane collapse. These two risks are highly correlated in time and space. Existing methods, lacking coupled assessment, tend to underestimate the resource demand in high-risk areas, leading to insufficient spare parts and repair personnel. This invention quantifies the superposition effect of multiple risks by constructing a risk coupling coefficient with a spatial attenuation factor and embedding it into the fitness weights and crossover probabilities of a genetic algorithm. Simultaneously, it introduces scheduling inertia penalties and resource coordination penalties into the multi-objective optimization model, solving the problems of isolated risk assessment, frequent fluctuations in scheduling schemes, and poor resource supply resilience in existing methods. Summary of the Invention

[0005] To address the issues of lack of risk coupling assessment, poor dynamic adaptability of solutions, and insufficient resource coordination in emergency resource scheduling for seaports, this invention provides a hierarchical scheduling optimization method and system for emergency resources of seaport infrastructure.

[0006] The technical solution adopted by this invention to solve its technical problem is as follows: S1: Obtain real-time monitoring data and historical operation data of container terminal infrastructure, and construct a multi-source feature vector set based on the real-time monitoring data and historical operation data; set the conditional probability distribution of each risk event, calculate the occurrence probability of the risk event under the condition of the multi-source feature vector set, identify the risk source type according to the occurrence probability of the risk event, and then output the occurrence probability value of each risk source according to the occurrence probability of the risk event and the risk source type. The risk source type includes the risk of collision between ships and berths under extreme weather conditions and the risk of large machinery failure. The occurrence probability value of each risk source is used as the basic input for subsequent risk coupling analysis; S2: Calculate the risk coupling coefficient between each risk source according to the occurrence probability value of each risk source in S1. The risk coupling coefficient is used to characterize the risk superposition effect caused by spatial proximity and joint occurrence between different risk sources, and serves as a dynamic adjustment factor for the fitness function and cross probability of the subsequent optimization model; S3: Using the response time as the most With the shortest response time and lowest disposal cost as optimization objectives, and constrained by emergency resource capacity, response time, and resource scheduling continuity, a multi-objective mixed-integer programming model is constructed by introducing a resource scheduling inertia coefficient and a resource synergy penalty term. The objective function of the multi-objective mixed-integer programming model is based on the shortest response time and lowest disposal cost. S4: The multi-objective mixed-integer programming model constructed in S3 is solved using a genetic algorithm to generate an initial feasible scheduling scheme. The genetic algorithm introduces the risk coupling coefficient mentioned in S2 as a dynamic weight factor in the fitness function, and the crossover operation of the genetic algorithm adopts a crossover probability based on the risk coupling coefficient. S5: The initial feasible scheduling scheme generated in S4 is subjected to multi-objective Pareto non-dominated verification, and then the non-dominated solution set on the Pareto front is selected as the candidate scheduling scheme set. S6: Real-time operation data of the container terminal is continuously acquired. Based on the candidate scheduling scheme set mentioned in S5 and combined with the risk probability output updated in real time in S1, the emergency resource scheduling scheme is dynamically adjusted.

[0007] Preferably, the specific method for calculating the probability of occurrence of each risk event in S1 is as follows: A conditional probability distribution for each risk event is established, and the probability of occurrence of each risk event under a given multi-source feature vector is calculated according to the following formula: ; in, For multi-source feature vectors, For the first One risk event For the first One risk event In order to observe Under the conditions The posterior probability of occurrence is used as the probability value of the occurrence of this risk source. Known Observed when it occurred The likelihood function, for The prior probability is the sum of the marginal probabilities of all possible risk events in the denominator; the multi-source feature vector includes meteorological data, hydrological data, operational status data of large terminal equipment, berth vessel berthing status data, and container yard cargo information data.

[0008] Preferably, the formula for calculating the risk coupling coefficient in S2 is as follows: ; in, For the first The first risk source and the first Risk coupling coefficient between risk sources and The first The first risk source and the first The probability of occurrence of each risk source is obtained from the output of S1. The joint probability of two risk sources occurring simultaneously. For the first The first risk source and the first The spatial distance between the risk sources within the container terminal area This is the spatial attenuation coefficient, which takes a positive value.

[0009] Preferably, the objective function of the multi-objective mixed-integer programming model in S3 is: minimizing the response time function. Minimize the disposal cost function ,in, Total response time Total disposal cost, For the collection of emergency resource reserve points, For collection of emergency demand points, This refers to a set of emergency resource types, including spare parts and components, human resources, and emergency supplies. To the reserve point To the demand point The quantity of emergency resources to be allocated is a decision variable. To the reserve point To demand point The delivery time The congestion coefficient for internal roads within the dock is a factor that is greater than or equal to 1. To the reserve point To demand point The unit resource transportation cost For the first Inventory levels of similar resources For the first The unit inventory holding cost of this type of resource For demand points Resource shortage penalty coefficient, For demand points Unmet resource demand.

[0010] Preferably, the formula for calculating the resource scheduling inertia coefficient in S3 is as follows: ; in, For the collection of emergency resource reserve points, For collection of emergency demand points, This represents the resource scheduling inertia coefficient. For the current moment from the reserve point To the demand point The amount of resources scheduled. For the current time index, For the previous moment from the reserve point To the demand point The amount of resources scheduled. To the reserve point To demand point The scheduling path stability weight coefficient is positively correlated with the historical success rate of the path; the resource coordination penalty term is applied to the objective function by adding an inertia penalty term to the objective function. ,in Let be the inertia penalty coefficient, when The inertia penalty term takes effect when the preset inertia threshold is exceeded.

[0011] Preferably, the calculation formula for the resource synergy penalty term in S3 is as follows: ; in, For collection of emergency demand points, This is a penalty item for resource synergy. To meet demand The first scheduling The number of reserve points from which these resources originate. To meet demand The total number of source reserve points for all resources being scheduled. The preset threshold for the repetition rate of resource sources. For the first The duplicate penalty coefficient for resource class. It is a set of resource types.

[0012] Preferably, the genetic algorithm in S4 is encoded as follows: the fitness function incorporates the risk coupling coefficient calculated in S2. The fitness function of the genetic algorithm is calculated using the following formula, which serves as a dynamic weighting factor: ; in, Here is the fitness function of the genetic algorithm. Total response time Total disposal cost, This is a comprehensive index of risk coupling. The weighting factor for response time. As a weighting factor for disposal costs, These are the weighting coefficients of the risk coupling comprehensive index, and all three are positive numbers; the risk coupling comprehensive index Risk coupling coefficient calculated based on S2 To build, ,in For the first The first risk source and the first The product of the resource requirements of each risk source's demand point; the crossover probability is calculated as follows: for two parent individuals participating in the crossover operation, calculate the average risk coupling coefficient corresponding to the emergency demand points they cover. Crossover probability ;in, For crossover probability, Based on the basic crossover probability, This is the average risk coupling coefficient corresponding to the emergency demand points covered by the two parent individuals. This is the adjustment range for the crossover probability.

[0013] Preferably, the method for dynamically adjusting the emergency resource scheduling scheme in S6 is as follows: when the total deviation between the risk probability output updated in real time in S1 and the risk probability on which the current scheduling scheme is based exceeds a preset deviation threshold, a rescheduling process is triggered; the total deviation value is the sum of the absolute values ​​of the probability deviations of each risk source; during rescheduling, the currently executed scheduling scheme is used as an individual in the initial population of the genetic algorithm, and several mutated individuals are replicated to form the initial population, and the adjusted scheduling scheme is regenerated in combination with the updated risk probability; if the ranking of the current scheme on the Pareto front is higher than the preset ranking threshold, rescheduling is not triggered.

[0014] On the other hand, a hierarchical scheduling optimization system for emergency resources of port infrastructure includes: a data acquisition module for acquiring real-time monitoring data and historical operation data of container terminal infrastructure; a risk identification and quantification module for constructing a terminal operation risk probability model and outputting the occurrence probability value of each risk source; a risk coupling analysis module for calculating the risk coupling coefficient between each risk source based on the occurrence probability value of each risk source; an optimization decision module for constructing a multi-objective mixed integer programming model with the shortest response time and lowest disposal cost as optimization objectives, introducing a resource scheduling inertia coefficient and a resource synergy penalty term; a scheme generation and verification module for solving the model using a genetic algorithm and performing multi-objective Pareto non-dominated verification to select a set of candidate scheduling schemes; and a real-time monitoring and feedback adjustment module for dynamically adjusting the emergency resource scheduling scheme based on the set of candidate scheduling schemes and the real-time updated risk probability output.

[0015] The beneficial effects of this invention are as follows: By introducing a risk coupling coefficient, the spatial correlation and joint occurrence probability of multiple risk sources, such as the risk of ship-berth collisions and large machinery failures under extreme weather conditions, are incorporated into emergency demand assessment. This overcomes the shortcomings of traditional methods that consider each risk source in isolation, making the assessment of spare parts and repair personnel needs in high-risk overlapping areas more realistic. By adding a resource scheduling inertia coefficient to the optimization model, penalties are imposed on the deviation of scheduling schemes at adjacent times, avoiding on-site repair command chaos and transportation resource waste caused by frequent and large-scale changes in schemes. The resource synergy penalty term constrains the distribution of source reserve points for various resources such as spare parts, human resources, and emergency supplies received by the same demand point, preventing excessive concentration of resources in a few supply nodes and improving the resilience and supply reliability of the scheduling scheme. Attached Figure Description

[0016] To more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are merely exemplary, and those skilled in the art can derive other embodiments based on the provided drawings without creative effort.

[0017] Figure 1 This is a schematic diagram of the optimal embodiment of the hierarchical scheduling optimization method for emergency resources of port infrastructure according to the present invention; Figure 2 A schematic diagram of the optimal embodiment of the hierarchical scheduling and optimization system for emergency resources of port infrastructure according to the present invention. Detailed Implementation

[0018] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0019] This embodiment provides a hierarchical scheduling optimization method and system for emergency resources of seaport infrastructure. Its core concept lies in achieving efficient, stable, and economical scheduling of emergency resources at container terminals under extreme weather conditions through hierarchical collaboration across four levels: risk coupling quantification, multi-objective optimization modeling, genetic algorithm solution, and real-time feedback adjustment. The specific implementation of this invention will be described in detail below from two aspects: method steps and system composition.

[0020] Example 1 like Figure 1 As shown, the present invention provides a hierarchical scheduling optimization method for emergency resources of seaport infrastructure, comprising: S1: Acquire real-time monitoring data and historical operation data of container terminal infrastructure, and construct a multi-source feature vector set based on the real-time monitoring data and historical operation data; set the conditional probability distribution of each risk event, calculate the occurrence probability of the risk event under the condition of the multi-source feature vector set, identify the risk source type according to the occurrence probability of the risk event, and then output the occurrence probability value of each risk source according to the occurrence probability of the risk event and the risk source type. The risk source type includes the risk of collision between ship and berth under extreme weather conditions and the risk of large machinery failure. The occurrence probability value of each risk source serves as the basic input for subsequent risk coupling analysis.

[0021] In S1, the multi-source feature vector This includes meteorological data, hydrological data, operational status data of large terminal equipment, berth vessel berthing status data, and container yard cargo information data. Meteorological data includes wind speed, wind direction, typhoon path, and intensity level. Hydrological data includes tide level, wave height, and current velocity. Operational status data of large terminal equipment includes the fault status of quayside container cranes, the operational status of rubber-tired gantry cranes, and pressure monitoring values ​​of oil pipelines. Berth vessel berthing status data includes vessel draft, mooring line tension, and relative displacement between the vessel and the berth. Container yard cargo information data includes the stacking location and number of stacking layers of dangerous goods containers. Historical operational data is stored in the system's historical database and can be accessed through a database interface.

[0022] In S1, by defining the conditional probability distribution of each risk event, the probability of occurrence of each risk event is calculated according to the following formula, given the multi-source feature vectors: ; in, It is a multi-source feature vector, which includes the aforementioned meteorological data, hydrological data, operational status data of large terminal equipment, berth vessel berthing status data, and container yard cargo information data. For the first One risk event This is the index number of the risk event, and its value is a positive integer. For the first One risk event This is the index number of the risk event, and its value is a positive integer. In order to observe multi-source feature vectors Under the condition of the first One risk event The posterior probability of occurrence is output as the probability value of occurrence of the risk source. For the known number One risk event Observed multi-source feature vectors during occurrence The likelihood function is estimated based on the distribution of multi-source feature vectors when the risk event occurs in historical operating data. For the first One risk event The prior probability is estimated statistically based on the historical frequency of the risk event in historical operational data. The denominator is... Let be the sum of the marginal probabilities of all possible risk events, where Iterate through all risk event types. Risk sources include the risk of ship-berth collision under extreme weather conditions and the risk of large machinery failure. The risk of large machinery failure includes the risk of quayside container crane collapse and the risk of rubber-tired gantry crane slippage or overturning.

[0023] S2: Based on the occurrence probability values ​​of each risk source described in S1, calculate the risk coupling coefficient between each risk source. The risk coupling coefficient is used to characterize the risk superposition effect caused by spatial proximity and joint occurrence between different risk sources, and serves as a dynamic adjustment factor for the fitness function and crossover probability of the subsequent optimization model.

[0024] The formula for calculating the risk coupling coefficient in S2 is: ; in, For the first The first risk source and the first Risk coupling coefficient between risk sources and All are index numbers of the risk sources, and their values ​​are positive integers. Not equal to . For the first The probability of occurrence of each risk source is obtained from the output of S1. For the first The probability of occurrence of each risk source is obtained from the output of S1. For the first The first risk source and the first The joint probability of two risk sources occurring simultaneously is estimated based on the statistical frequency of the simultaneous occurrence of two risk sources in historical operating data, or by multiplying the product of the individual occurrence probabilities of the two risk sources by the correlation coefficient. For the first The first risk source and the first The spatial distance between two risk sources within the container terminal area is calculated using Euclidean distance or Manhattan distance based on their geographical coordinates within the terminal area. This is the spatial attenuation coefficient, which takes a positive value. The value of is set based on the spatial scale of the dock area and the attenuation characteristics of the risk impact range. Preferably, The value ranges from 0.01 to 0.1. Spatial attenuation coefficient. This is used to characterize the rate at which the risk coupling effect between two risk sources decays with increasing spatial distance. The larger the value, the more significant the attenuation effect of spatial distance on the coupling effect. For the natural constant An exponential decay factor with a base of is used to spatially weight the attenuation of the risk coupling coefficient. Risk coupling coefficient The larger the value, the more significant the number of . The first risk source and the first The more significant the risk superposition effect between two risk sources, the greater the spatial distance between them. As it approaches zero, the exponential decay factor When the spatial attenuation value approaches 1, the spatial decay effect is negligible; when the spatial distance between two risk sources is close to 1... As it approaches infinity, the exponential decay factor Approaching 0, the spatial decay effect causes the risk coupling coefficient to approach zero.

[0025] S3: With the shortest response time and lowest disposal cost as optimization objectives, and with constraints such as emergency resource capacity, response time limit, and resource scheduling continuity as constraints, and with the introduction of resource scheduling inertia coefficient and resource coordination penalty term, a multi-objective mixed integer programming model is constructed. The shortest response time and lowest disposal cost are used as the objective function of the multi-objective mixed integer programming model.

[0026] The objective function of the multi-objective mixed integer programming model in S3 is: Minimize response time function: ; Minimize the disposal cost function: ; in, Total response time Total disposal cost. For the collection of emergency resource reserve points, For the index number of the reserve point, Belongs to set . For collection of emergency demand points, For the index number of the demand point, Belongs to set . A collection of emergency resource types. For the index number of the resource type, Belongs to set The emergency resources include spare parts, human resources, and emergency supplies. Spare parts include replacement parts for quay container cranes, tire-mounted gantry cranes, and mooring ropes. Human resources include maintenance personnel, repair personnel, and operators. Emergency supplies include flood control sandbags, emergency lighting equipment, and drainage equipment.

[0027] To the reserve point To the demand point The amount of emergency resources allocated, and As decision variables, The value is a non-negative integer. To the reserve point To demand point The transportation time depends on the storage point. With demand points The path length and average transport speed between them are calculated. The congestion coefficient of the internal roads of the dock. The value of is greater than or equal to 1. The settings are based on real-time traffic flow, road construction status, and traffic conditions within the wharf's internal roads. When the path is clear... The value is 1, when the path is congested. The value is greater than 1. To the reserve point To demand point The unit resource transportation cost includes fuel cost, vehicle depreciation cost, and labor cost. For the first The inventory level of this type of resource, which is the first in the reserve point. The available inventory quantity of this type of resource. For the first The unit inventory holding cost of this type of resource includes warehousing costs, maintenance costs, and capital occupation costs. For demand points Resource shortage penalty coefficient, Based on the demand points output in S1 The risk level is set accordingly; the higher the risk level, the greater the penalty for resource shortage. The larger the value, the better. For demand points Unmet resource demand equal to demand point Total resource demand minus the amount transferred from various reserve points to demand points The total amount of resources scheduled, i.e. in For demand points The total demand for resources, when hour, The value is 0.

[0028] The constraints of the multi-objective mixed-integer programming model include emergency resource capacity constraints, response time constraints, and resource scheduling continuity constraints. The emergency resource capacity constraint is: for each reserve point... and each resource type From the reserve point The retrieved number The total amount of this type of resource shall not exceed the first [number] of this reserve point. The inventory of such resources, i.e. ,in To the reserve point To the demand point The first scheduling Number of resource types Reserve point The Middle Inventory levels of the resource type. Response time constraints are: for each demand point... From various reserve points to demand points The response time of the scheduled resources shall not exceed the maximum allowable response time for that demand point, i.e. ,in For demand points Maximum allowable response time. Resource scheduling continuity constraint: the change in the scheduling scheme between adjacent time points does not exceed a preset maximum change threshold to ensure the continuity of the scheduling scheme.

[0029] The formula for calculating the resource scheduling inertia coefficient in S3 is as follows: ; in, For the collection of emergency resource reserve points, For collection of emergency demand points, This represents the resource scheduling inertia coefficient. For the current moment from the reserve point To the demand point The amount of resources scheduled. This is the index number for the current time. The value is a positive integer. For the previous moment from the reserve point To the demand point The amount of resources scheduled. This is the index number from the previous time step. It represents the absolute difference between the number of resources scheduled at the current time and the previous time. To the reserve point To demand point The scheduling path stability weight coefficient, the stability weight coefficient It is positively correlated with the historical success rate of the path. The higher the historical success rate of the path, the better. The larger the value, the lower the historical success rate of the path. The smaller the value, the better. The value range is from 0 to 1. Resource scheduling inertia coefficient. The larger the value, the greater the change in the current scheduling scheme compared to the previous scheduling scheme, and the worse the inertia of the scheduling scheme.

[0030] The resource synergy penalty term is applied to the objective function by adding an inertia penalty term to the objective function. ,in The inertia penalty coefficient, The value is a positive number. Inertia penalty coefficient. The value is set according to the degree of continuity required for the scheduling scheme. The larger the value, the heavier the penalty for significant changes in the scheduling scheme. When the resource scheduling inertia coefficient... When the preset inertia threshold is exceeded, the inertia penalty term... Effective, meaning this term is added to the objective function as a penalty. When the resource scheduling inertia coefficient... If the preset inertia threshold is not exceeded, the inertia penalty item will not take effect, that is, no inertia penalty will be applied. The inertia threshold is determined based on the maximum allowable variation range of the actual dock scheduling.

[0031] The formula for calculating the resource synergy penalty in S3 is as follows: ; in, For collection of emergency demand points, This is a penalty item for resource synergy. To meet demand The first scheduling The number of reserve points from which the resource originates. For example, if it is supplied to the demand point... The spare parts and components resources to be dispatched come from reserve point 1, reserve point 2, and reserve point 3, respectively. Therefore, for these spare parts and components resources... It equals 3. To meet demand The total number of source reserve points for all resources to be scheduled. For example, if resources are allocated to the demand point... The spare parts and components resources are sourced from 3 reserve points, human resources from 2 reserve points, and emergency supplies from 4 reserve points. It equals the sum of 3, 2, and 4, which is 9. To meet demand The first scheduling The proportion of the number of source reserve points for this type of resource to the total number of all resource source reserve points. The preset threshold for the repetition rate of resource sources. The value range is from 0 to 1. The value is set according to the port's requirements for the diversity of resource supply. The larger the value, the lower the requirement for diversity of resource sources; The smaller the value, the higher the requirement for diversity of resource sources. To take 0 and The maximum value between, when When the difference is greater than 0, take the difference; when... If it is less than or equal to 0, take 0. For the first The duplicate penalty coefficient for resource class. Values ​​are positive. The duplicate penalty coefficient for different types of resources. The values ​​can differ; for critical resources, the duplicate penalty coefficient is larger, while for non-critical resources, the duplicate penalty coefficient is smaller. For a set of resource types, For the index number of the resource type, Belongs to set Resource synergy penalty item It is added as a penalty term to the total disposal cost function within the objective function. In the middle, that is, the total disposal cost function Updated to Resource synergy penalty item The higher the value, the higher the concentration of resource sources, the worse the resilience of resource supply, and the heavier the punishment.

[0032] S4: The genetic algorithm is used to solve the multi-objective mixed integer programming model constructed in S3 to generate an initial feasible scheduling scheme. The genetic algorithm introduces the risk coupling coefficient mentioned in S2 as a dynamic weight factor in the fitness function. The crossover operation of the genetic algorithm adopts the crossover probability based on the risk coupling coefficient.

[0033] In S4, the specific process of the genetic algorithm is as follows. First, initialize the genetic algorithm population. Each individual in the population represents a scheduling scheme, that is, a set of decision variables. The population size is set according to the problem size, preferably between 100 and 500. Individuals are encoded using real numbers or binary codes, and the decision variables... The value of is a non-negative integer and satisfies all the constraints in S3.

[0034] The fitness function in a genetic algorithm is used to evaluate the quality of each individual. The fitness function incorporates a risk coupling coefficient calculated using S2. As a dynamic weighting factor. Specifically, the fitness function The calculation method is as follows: ; in, Here is the fitness function of the genetic algorithm. Total response time Total disposal cost, This is a comprehensive index for risk coupling. The weighting factor for response time. As a weighting factor for disposal costs, These are the weighting coefficients of the risk coupling composite index; all three are positive. Risk Coupling Composite Index Risk coupling coefficient calculated based on S2 To build, ,in For the first The first risk source and the first The product of resource demands at each demand point where a risk source is located is used to characterize the degree of impact of risk coupling effects on resource demand. Fitness function The higher the value, the better the individual's overall performance.

[0035] The crossover operation in the genetic algorithm uses a crossover probability based on the risk coupling coefficient. The calculation method is as follows: ; in, For crossover probability, Based on the basic crossover probability, This is the average risk coupling coefficient corresponding to the emergency demand points covered by the two parent individuals. Adjustment magnitude for crossover probability. When the average risk coupling coefficient... When the crossover probability is large, Increase the average risk coupling coefficient to enhance the algorithm's search capability in high-risk coupling regions; When the crossover probability is small, The mutation rate is reduced to maintain the convergence of the algorithm. The mutation operation in the genetic algorithm can use uniform or non-uniform mutation, with the mutation probability... Set to a fixed value, preferably, the mutation probability. The value range is from 0.01 to 0.1.

[0036] The termination condition of a genetic algorithm is reaching a preset maximum number of iterations or the fitness function value no longer significantly improving within a certain number of consecutive generations. The maximum number of iterations is set according to the problem size; preferably, it is set to 100 to 500 generations.

[0037] S5: Perform multi-objective Pareto non-dominated verification on the initial feasible scheduling scheme generated in S4, and then select the non-dominated solution set on the Pareto front as the candidate scheduling scheme set.

[0038] In S5, the specific method for Pareto non-domination verification is as follows. For any two scheduling scheme solutions... and If the solution Non-inferior solution on all objective functions And it is better than the solution on at least one objective function. Then it is called a solution. Dominant Solution Specifically, for the minimization problem, if and ,in, For scheduling scheme Total response time For scheduling scheme Total response time For scheduling scheme Total disposal cost For scheduling scheme Given the total disposal cost, and at least one objective function such that the strict inequality holds, then the solution is... Dominant Solution A solution is called a non-dominated solution, or Pareto optimal solution, if it is not dominated by any other solution in the population. The set of all non-dominated solutions is called the Pareto front. The set of non-dominated solutions on the Pareto front is output as a set of candidate scheduling schemes. Each scheme in the candidate scheduling scheme set achieves a different trade-off between response time and disposal cost, allowing decision-makers to choose according to actual needs.

[0039] S6: Continuously acquire real-time operational data of the container terminal, and dynamically adjust the emergency resource scheduling scheme based on the candidate scheduling scheme set described in S5 and the risk probability output updated in real time in S1.

[0040] The method for dynamically adjusting the emergency resource scheduling plan in S6 is as follows: when the total deviation between the real-time updated risk probability output in S1 and the risk probability on which the current scheduling plan is based exceeds a preset deviation threshold, a rescheduling process is triggered. The total deviation is the sum of the absolute values ​​of the probability deviations of each risk source. Specifically, let the probability of occurrence of each risk source on which the current scheduling plan is based be... The probability of occurrence of each risk source is updated in real time. Then the total deviation value The calculation formula is: ; in, Iterate through all risk source types. The preset deviation threshold is set based on the terminal's sensitivity to risk changes; preferably, the preset deviation threshold ranges from 0.1 to 0.5. When the total deviation value... When the preset deviation threshold is exceeded, a rescheduling process is triggered.

[0041] During rescheduling, the currently executed scheduling scheme is used as an individual in the initial population of the genetic algorithm, and several mutated individuals are generated by copying and mutation of that individual to form the new initial population. The adjusted scheduling scheme is then regenerated based on the updated risk probabilities. Specifically, the initial population contains an individual from the current scheduling scheme, and multiple mutated individuals generated by copying and mutation of that individual. The mutation operation involves modifying the decision variables... Random additions, reductions, or swaps are performed. The number of mutated individuals is set based on the population size, preferably between 10% and 30% of the population size. If the current scheme ranks higher than a preset ranking threshold on the Pareto front, rescheduling is not triggered. The ranking on the Pareto front is determined based on non-dominated sorting; a higher ranking indicates a better scheme. The preset ranking threshold is set based on the stability requirements of the scheduling scheme. When the current scheme is already in a relatively good position on the Pareto front, rescheduling is not triggered even if the risk probability changes, to avoid frequent changes in the scheduling scheme. After rescheduling is completed, the updated scheduling scheme is output and executed.

[0042] Example 2 like Figure 2 As shown, the present invention also provides a hierarchical scheduling optimization system for emergency resources of seaport infrastructure, which is equipped with the above-mentioned hierarchical scheduling optimization method for emergency resources of seaport infrastructure to realize the automated operation of the method.

[0043] The system includes a data acquisition module, a risk identification and quantification module, a risk coupling analysis module, an optimization decision-making module, a scheme generation and verification module, and a real-time monitoring and feedback adjustment module.

[0044] The data acquisition module is used to obtain real-time monitoring data and historical operational data of the container terminal infrastructure. The module includes sub-modules for meteorological data acquisition, hydrological data acquisition, large terminal equipment operation status acquisition, berth vessel berthing status acquisition, and container yard cargo information acquisition. The meteorological data acquisition sub-module acquires real-time meteorological data, including wind speed, wind direction, typhoon path, and intensity level, through meteorological sensors, weather radar, and meteorological satellite data interfaces. The hydrological data acquisition sub-module acquires real-time hydrological data, including tide height, wave height, and current velocity, through tide gauges, wave buoys, and hydrological monitoring stations. The large terminal equipment operation status acquisition sub-module acquires operational status data of large terminal equipment through equipment sensors and the SCADA system, including the fault status of quay container cranes, the operational status of rubber-tired gantry cranes, and pressure monitoring values ​​of oil pipelines. The berth vessel berthing status acquisition sub-module acquires berth vessel berthing status data, including vessel draft, mooring line tension, and relative displacement between the vessel and the berth, through the Automatic Identification System (AIS), mooring line tension sensors, and vessel draft monitoring system. The container yard cargo information collection submodule obtains the storage location and stacking layer of dangerous goods containers through the container management system. Historical operational data is stored in the system's historical database, and the data collection module reads historical data through the database interface.

[0045] The risk identification and quantification module communicates with the data acquisition module to construct a probability model of operational risks at the container terminal, identify the types and levels of risk sources within the terminal area, and output the probability value of each risk source. The specific workflow of the risk identification and quantification module is consistent with the method described in S1. The risk coupling analysis module communicates with the risk identification and quantification module to calculate the risk coupling coefficient between risk sources based on their probability values. The formula for calculating the risk coupling coefficient is: ; in, For the first The first risk source and the first Risk coupling coefficient between risk sources and The first The first risk source and the first The probability of occurrence of each risk source. The joint probability of two risk sources occurring simultaneously. For the first The first risk source and the first The spatial distance between the risk sources within the container terminal area This is the spatial attenuation coefficient.

[0046] The decision-making module and the risk coupling analysis module are connected to optimize the communication. The optimization objectives are to minimize response time and disposal cost. The constraints are emergency resource capacity constraints, response time constraints, and resource scheduling continuity constraints. Resource scheduling inertia coefficient and resource synergy penalty term are introduced to construct a multi-objective mixed integer programming model.

[0047] The scheme generation and verification module is communicatively connected to the optimization decision module. It is used to solve the multi-objective mixed-integer programming model using a genetic algorithm, generate initial feasible scheduling schemes, and perform multi-objective Pareto non-dominated verification on these schemes. The non-dominated solution set on the Pareto front is then selected as the candidate scheduling scheme set. A risk coupling coefficient is introduced as a dynamic weighting factor into the fitness function of the genetic algorithm, and the crossover operation of the genetic algorithm uses a crossover probability based on the risk coupling coefficient.

[0048] The real-time monitoring and feedback adjustment module is used to dynamically adjust the emergency resource scheduling scheme based on the candidate scheduling scheme set and the real-time updated risk probability output. The real-time monitoring and feedback adjustment module acquires real-time data through the data acquisition module to dynamically update the risk probability.

[0049] This document uses specific examples to illustrate the principles and implementation methods of this application. The descriptions of the above embodiments are only for the purpose of helping to understand the methods and core ideas of this application. The above descriptions are only preferred embodiments of this application. It should be noted that due to the limitations of written expression, while there are objectively infinite specific structures, those skilled in the art can make several improvements, modifications, or changes without departing from the principles of this application, and can also combine the above technical features in an appropriate manner. These improvements, modifications, changes, or combinations, or the direct application of the inventive concept and technical solution to other situations without modification, should all be considered within the scope of protection of this application.

Claims

1. A hierarchical scheduling optimization method for emergency resources of seaport infrastructure, characterized in that, include: S1: Obtain real-time monitoring data and historical operation data of container terminal infrastructure, and construct a multi-source feature vector set based on the real-time monitoring data and historical operation data; The conditional probability distribution of each risk event is set. Under the condition of the multi-source feature vector set, the probability of occurrence of the risk event is calculated. The risk source type is identified according to the probability of occurrence of the risk event. Then, the probability value of occurrence of each risk source is output according to the probability of occurrence of the risk event and the risk source type. The risk source type includes the risk of collision between ship and berth under extreme weather and the risk of failure of large machinery. The probability value of occurrence of each risk source is used as the basic input for subsequent risk coupling analysis. S2: Based on the occurrence probability values ​​of each risk source mentioned in S1, calculate the risk coupling coefficient between each risk source. The risk coupling coefficient is used to characterize the risk superposition effect caused by spatial proximity and joint occurrence between different risk sources, and serves as a dynamic adjustment factor for the fitness function and crossover probability of the subsequent optimization model. S3: With the shortest response time and lowest disposal cost as optimization objectives, and with constraints such as emergency resource capacity, response time limit, and resource scheduling continuity as constraints, and with the introduction of resource scheduling inertia coefficient and resource coordination penalty term, a multi-objective mixed integer programming model is constructed. The shortest response time and lowest disposal cost are used as the objective function of the multi-objective mixed integer programming model. S4: The genetic algorithm is used to solve the multi-objective mixed integer programming model constructed in S3 to generate an initial feasible scheduling scheme. The genetic algorithm introduces the risk coupling coefficient mentioned in S2 as a dynamic weight factor in the fitness function. The crossover operation of the genetic algorithm adopts the crossover probability based on the risk coupling coefficient. S5: Perform multi-objective Pareto non-dominated verification on the initial feasible scheduling scheme generated in S4, and then select the non-dominated solution set on the Pareto front as the candidate scheduling scheme set. S6: Continuously acquire real-time operational data of the container terminal, and dynamically adjust the emergency resource scheduling scheme based on the candidate scheduling scheme set described in S5 and the risk probability output updated in real time in S1.

2. The hierarchical scheduling optimization method for emergency resources of seaport infrastructure according to claim 1, characterized in that, The specific method for calculating the probability of occurrence of each risk event in S1 is as follows: Define the conditional probability distribution for each risk event, and calculate the probability of occurrence of each risk event given a multi-source feature vector using the following formula: ； in, For multi-source feature vectors, For the first One risk event For the first One risk event In order to observe Under the conditions The posterior probability of occurrence is used as the probability value of the occurrence of this risk source. Known Observed when it occurred The likelihood function, for The prior probability is the sum of the marginal probabilities of all possible risky events in the denominator. The multi-source feature vector includes meteorological data, hydrological data, operational status data of large terminal equipment, berthing status data of ships at berths, and cargo information data of container yards.

3. The hierarchical scheduling optimization method for emergency resources of seaport infrastructure according to claim 1, characterized in that, The formula for calculating the risk coupling coefficient in S2 is as follows: ； in, For the first The first risk source and the first Risk coupling coefficient between risk sources and The first The first risk source and the first The probability of occurrence of each risk source is obtained from the output of S1. The joint probability of two risk sources occurring simultaneously. For the first The first risk source and the first The spatial distance between the risk sources within the container terminal area This is the spatial attenuation coefficient, which takes a positive value.

4. The hierarchical scheduling optimization method for emergency resources of seaport infrastructure according to claim 1, characterized in that, The objective function of the multi-objective mixed integer programming model in S3 is: Minimize response time function , Minimize disposal cost function , in, Total response time Total disposal cost, As a collection of emergency resource reserve points, For collection of emergency demand points, This refers to a set of emergency resource types, including spare parts and components, human resources, and emergency supplies. To the reserve point To the demand point The quantity of emergency resources to be allocated is a decision variable. To the reserve point To demand point The delivery time The congestion coefficient for internal roads within the dock is a factor that is greater than or equal to 1. To the reserve point To demand point The unit resource transportation cost For the first Inventory levels of similar resources For the first The unit inventory holding cost of this type of resource For demand points Resource shortage penalty coefficient, For demand points Unmet resource demand.

5. The hierarchical scheduling optimization method for emergency resources of seaport infrastructure according to claim 1, characterized in that, The formula for calculating the resource scheduling inertia coefficient in S3 is as follows: ； in, As a collection of emergency resource reserve points, For collection of emergency demand points, This represents the resource scheduling inertia coefficient. For the current moment from the reserve point To the demand point The amount of resources scheduled. For the current time index, For the previous moment from the reserve point To the demand point The amount of resources scheduled. To the reserve point To demand point The scheduling path stability weight coefficient, which is positively correlated with the historical success rate of the path; The resource synergy penalty term is applied to the objective function by adding an inertia penalty term to the objective function. ,in Let be the inertia penalty coefficient, when The inertia penalty term takes effect when the preset inertia threshold is exceeded.

6. The hierarchical scheduling optimization method for emergency resources of seaport infrastructure according to claim 1, characterized in that, The formula for calculating the resource synergy penalty term in S3 is as follows: ； in, For collection of emergency demand points, This is a penalty item for resource synergy. To meet demand The first scheduling The number of reserve points from which these resources originate. To meet demand The total number of source reserve points for all resources being scheduled. The preset threshold for the repetition rate of resource sources. For the first The duplicate penalty coefficient for resource class. It is a set of resource types.

7. The hierarchical scheduling optimization method for emergency resources of seaport infrastructure according to claim 1, characterized in that, The encoding method of the genetic algorithm in S4 is as follows: The fitness function incorporates the risk coupling coefficient calculated using S2. The fitness function of the genetic algorithm is calculated using the following formula, which serves as a dynamic weighting factor: ； in, Here is the fitness function of the genetic algorithm. Total response time Total disposal cost, This is a comprehensive index of risk coupling. The weighting factor for response time. As a weighting factor for disposal costs, These are the weighting coefficients of the risk coupling comprehensive index, and all three are positive numbers; The risk coupling comprehensive index Risk coupling coefficient calculated based on S2 To build, ,in For the first The first risk source and the first The product of the resource demand at each demand point where a risk source is located; The crossover probability is calculated as follows: for the two parent individuals participating in the crossover operation, the average risk coupling coefficient corresponding to the emergency demand points they cover is calculated. Crossover probability ; in, For crossover probability, Based on the basic crossover probability, This is the average risk coupling coefficient corresponding to the emergency demand points covered by the two parent individuals. This is the adjustment range for the crossover probability.

8. The hierarchical scheduling optimization method for emergency resources of seaport infrastructure according to claim 1, characterized in that, The method for dynamically adjusting the emergency resource scheduling scheme in S6 is as follows: When the total deviation between the risk probability output updated in real time in S1 and the risk probability on which the current scheduling scheme is based exceeds the preset deviation threshold, the rescheduling process is triggered. The total deviation value is the sum of the absolute values ​​of the probability deviations of each risk source; During rescheduling, the currently executed scheduling scheme is used as an individual in the initial population of the genetic algorithm, and several mutated individuals are generated by replication to form the initial population. The adjusted scheduling scheme is then regenerated by combining the updated risk probability. If the current scheme ranks higher than the preset ranking threshold on the Pareto front, rescheduling will not be triggered.

9. A hierarchical scheduling and optimization system for emergency resources of seaport infrastructure, characterized in that, The method for hierarchical scheduling and optimization of emergency resources for seaport infrastructure as described in any one of claims 1-8 includes: The data acquisition module is used to acquire real-time monitoring data and historical operational data of container terminal infrastructure; The risk identification and quantification module is used to build a probability model of terminal operation risks and output the probability value of occurrence of each risk source. The risk coupling analysis module is used to calculate the risk coupling coefficient between risk sources based on the probability of occurrence of each risk source. The optimization decision module is used to construct a multi-objective mixed integer programming model with the optimization objectives of minimizing response time and minimizing disposal cost, by introducing a resource scheduling inertia coefficient and a resource synergy penalty term. The scheme generation and verification module is used to solve the model using a genetic algorithm, perform multi-objective Pareto non-dominated verification, and select a set of candidate scheduling schemes. The real-time monitoring and feedback adjustment module is used to dynamically adjust the emergency resource scheduling scheme based on the candidate scheduling scheme set and the real-time updated risk probability output.

Images