A multi-section cross-operation construction ship scheduling management method

Abstract

The application relates to the technical field of construction ship scheduling management, and discloses a construction ship scheduling management method under multi-section cross operation, which comprises the following steps: S1, acquiring real-time data of each section to construct a multi-dimensional feature vector containing time, resources, operation and risk; S2, establishing a dynamic coupling network model representing time-varying coupling relationship based on the vector; S3, adaptively adjusting the task scheduling priority according to the global state change of the network; S4, using a leader-follower game optimization method combining the network model and the priority to solve the scheduling strategy in a rolling window; and S5, outputting an execution strategy and performing cyclic scheduling through a sliding time window until the task is completed. Through the dynamic coupling network model and the game optimization method combined with the dynamic priority, the application can resolve conflicts among multiple parties and improve the global coordination and dynamic adaptability of ship scheduling under a complex cross-operation environment.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of construction ship scheduling management, in particular to a construction ship scheduling management method under multi-section cross operation. BACKGROUND

[0002] In large water conservancy and hydropower, cross-sea channel, port channel and other projects, the construction ship scheduling management method under multi-section, high-density and cross operation scenarios, etc. Large projects, multiple construction units or sections simultaneously carry out work in limited operation water area, with numerous ships, various types, strong task correlation, making ship scheduling a very complex system engineering problem, whose efficiency and safety are directly related to the overall progress, cost and risk control of the project.

[0003] At present, for the scheduling management of construction ships, the industry has developed various technical methods. In the project planning stage, a static planning method based on the critical path method (CPM) and other network planning techniques is usually used to make macro, long-period planning arrangements for construction processes and ship use. In the actual execution stage, to deal with dynamic changes on site, some methods introduce optimization scheduling models based on heuristic algorithms (such as genetic algorithm, ant colony algorithm, etc.). These models abstract the scheduling problem as a mathematical optimization problem, and solve the allocation and path of the ship at a specific time point, in order to achieve the optimization of one or more objectives (such as the shortest total duration, the lowest total cost). At the same time, at a more microscopic real-time scheduling level, a dispatch strategy based on preset rules, such as first-come-first-served, shortest operation time priority, etc., is often used to handle immediate ship scheduling requests.

[0004] Although the prior art improves the planning and automation level of construction ship scheduling to some extent, there are still some deficiencies in the face of multi-section cross-operation, which is a highly dynamic and strongly coupled complex scene. First, the existing methods have limitations in recognizing the internal coupling relationship of the construction system. These methods often build models based on simplified, static process logic or resource constraints, which makes it difficult to capture and represent the dynamic coupling relationship between sections in the real world, which is intertwined by multiple factors such as time, resources, operations, risks, etc. Second, the scheduling priority decision mechanism in the prior art lacks dynamic response capability to the global state of the system. Whether it is a fixed priority set at the beginning of the project or a simple rule based on local task attributes (such as urgency, waiting time), their common defect is that the priority is determined statically or only depends on local information. Third, the existing optimization scheduling model has a model mismatch problem when dealing with the relationship between multiple interest subjects. Traditional centralized optimization methods try to calculate a global optimal solution for the entire system, but this often ignores the inherent driving force of each construction section as an independent economic entity to pursue its own interests, resulting in the "optimal" plan being difficult to be fully accepted and effectively executed by all parties in reality. SUMMARY

[0005] In view of the deficiencies of the prior art, the present application provides a construction ship scheduling management method under multi-section cross-operation, which solves the problem that the existing construction ship scheduling method lacks a deep understanding of the dynamic coupling relationship between multiple sections and an effective coordination mechanism for multi-party game conflicts, resulting in insufficient globality, adaptability and collaboration of scheduling decisions in a complex cross-operation environment.

[0006] To achieve the above purpose, the present application is implemented by the following technical scheme: a construction ship scheduling management method under multi-section cross-operation, the method comprising the following steps:

[0007] S1, obtaining real-time operation data of each construction section, and constructing a multi-dimensional feature vector containing time, resource, operation and risk information for each section;

[0008] S2, establishing a dynamic coupling network model that can represent the time-varying coupling relationship between sections based on the constructed multi-dimensional feature vector;

[0009] S3, adaptively adjusting the scheduling priority of each operation task according to the global state change of the dynamic coupling network model;

[0010] S4, obtaining a forward-rolling limited time window, and using a leader-follower game optimization method within the limited time window to obtain a ship scheduling strategy in the current scheduling period, wherein the dynamic coupling network model and the adjusted scheduling priority are combined in the leader-follower game optimization method.

[0011] S5, output and execute the ship scheduling strategy, synchronously slide the time window forward to schedule the next cycle, and cyclically execute until all construction tasks are completed.

[0012] The application provides a construction ship scheduling management method under multi-bid section cross operation.

[0013] 1. The application can realize accurate description of complex and time-varying correlation between construction units in a multi-bid section cross operation environment by constructing a multi-dimensional feature vector containing time, resource, operation and risk information and establishing a dynamic coupling network model on this basis. Compared with traditional methods that rely on static or empirical models, the application can more deeply understand the overall situation of the system, dynamically identify potential coupling risks caused by multiple factors interweaving, which are difficult to be found by traditional methods, and thus significantly improve the overall situation awareness ability and the accuracy of risk prediction of the complex construction environment.

[0014] 2. The application proposes a mechanism for dynamically adjusting the operation task scheduling priority according to the global state change of the dynamic coupling network model. This mechanism enables the scheduling focus to shift from following fixed and preset rules to actively and adaptively focusing on the key link that has the greatest impact on the overall system stability. When a bid section becomes the main disturbance source of the system, the priority of its related tasks will be automatically improved, guiding the scheduling resources to solve the root problems that are most likely to cause chain delays or conflicts, so that the entire scheduling management process is more forward-looking and targeted in risk response.

[0015] 3. The application adopts a leader-follower game optimization method and integrates the dynamic coupling network model and the adjusted scheduling priority, realizing the organic unification of global planning and local autonomous decision-making. By introducing the utility function, the method enables each bid section to consciously consider the impact of its decision on the stability of the global network while pursuing its own benefits. This can effectively alleviate conflicts between bid sections caused by resource competition or process connection, and promote all parties to form a synergistic effect rather than a zero-sum game. The final ship scheduling strategy not only has optimization in mathematics. BRIEF DESCRIPTION OF DRAWINGS

[0016] Figure 1 The figure is a flowchart of the method of the application. DETAILED DESCRIPTION

[0017] With reference to the drawings of the present application in the specification, the technical solutions in the embodiments of the present application will be clearly and completely described. Obviously, the described embodiments are only a part of the embodiments of the present application, rather than all the embodiments. Based on the embodiments in the present application, all other embodiments obtained by those of ordinary skill in the art without creative work fall within the protection scope of the present application.

[0018] Please refer to the drawings of the present application in the specification, the technical solutions in the embodiments of the present application will be clearly and completely described. Obviously, the described embodiments are only a part of the embodiments of the present application, rather than all the embodiments. Based on the embodiments in the present application, all other embodiments obtained by those of ordinary skill in the art without creative work fall within the protection scope of the present application. Figure 1 The embodiment of the present application provides a construction ship scheduling management method under multi-section cross operation, and the method comprises the following steps:

[0019] S1, real-time operation data of each construction section is obtained, and a multi-dimensional feature vector containing time, resource, operation and risk information is constructed for each section;

[0020] The core purpose of step S1 is to convert the original information of multiple sources, heterogeneous and dynamic changes in the construction site into a structured and standardized mathematical expression, i.e. a multi-dimensional feature vector, so as to provide accurate and reliable input for the construction of dynamic coupling network and game optimization decision in the subsequent steps.

[0021] In a specific embodiment, first, the operation data of each construction section is obtained in real time or quasi-real time through a set of data acquisition system. Preferably, the data acquisition system can integrate ship automatic identification system (AIS), global positioning system (GPS), various Internet of Things (IoT) sensors deployed in the construction area and ships, electronic construction log system, resource scheduling management platform and third-party weather forecast service, etc. to ensure the comprehensiveness and real-time of the obtained data.

[0022] After obtaining the original operation data, a multi-dimensional feature vector of each independent construction section k at time is constructed. The construction of the vector is the key to comprehensively and quantitatively describe the state of the section. The structure of the vector is defined as follows:

[0023] ;

[0024] In the formula, is the multi-dimensional feature vector of the section at time ; is a time coupling factor; is a resource competition degree; is an operation fluctuation coefficient; is an emergency correlation degree.

[0025] The calculation method and technical connotation of each component in the vector are described in detail as follows:

[0026] Time coupling factor :

[0027] This factor aims to quantify the timing impact of a bid section's progress change on other bid sections that have process flow dependencies. In multi-bid section cross-operations, bid sections are not isolated, and the advance or delay of a bid section will spread to its adjacent subsequent bid sections like ripples, thereby affecting the entire project duration. Therefore, accurately depicting this time coupling relationship is a prerequisite for collaborative scheduling, so as to obtain a value that can reflect its average time impact strength.

[0028] Resource competition degree :

[0029] This factor is used to represent the intensity of competition for common or scarce resources by a bid section within a specific time window. In large water projects, such as channels, anchorages, large dredgers, heavy lifting ships, etc., are often key resources that multiple bid sections need to share, and the competition for these resources is the main reason for operation conflicts and efficiency bottlenecks.

[0030] In a specific implementation, the calculation method of this factor is as follows: first, a global resource demand pool is counted and updated, which contains all the resource usage requests submitted by the current bid sections. Then, for the bid section , identify how many of the resources it requests are also being requested by other bid sections (i.e., the intersection of resource demand). Finally, calculate the ratio of this intersection to the total number of different resources in the global resource demand pool, and normalize it to obtain a value that objectively reflects the resource competition pressure.

[0031] Operation fluctuation coefficient :

[0032] This factor aims to measure the internal stability of bid section k's operation state. A bid section with a stable operation state and predictable efficiency is reliable for the scheduling system; on the contrary, a bid section with fluctuating operation efficiency and unstable state is a source of uncertainty in the system and needs to be focused on.

[0033] In a specific implementation, this factor is obtained by calculating the variance of the key operation efficiency indicators (KPIs) of bid section in a pre-set time window in the past (e.g., the past 8 hours). The key operation efficiency indicators here can be the dredging volume per unit time, the rock volume, the pipe section installation speed, etc., which can directly reflect the physical quantities of construction progress. The larger the calculated variance value, the worse the operation stability of the bid section, and the higher the risk it may introduce in subsequent scheduling.

[0034] Emergency correlation degree :

[0035] This factor is used to prospectively assess the degree of correlation of external emergencies or potential risks to the bid section . Traditional scheduling methods tend to be passive responses after the event occurs, and the present application aims to change risk management from post-response to pre-judgment by introducing this factor.

[0036] In a specific implementation, first, a risk event library containing a variety of typical emergencies (such as gale weather warning, temporary closure of the channel, key equipment failure warning, etc.) needs to be established. Then, according to the influence range and severity level of each event, an influence weight is defined for each event. Finally, by monitoring external information in real time, once it is found that a certain event may affect the bid section , the corresponding weight of the event is added to the emergency correlation degree of the bid section .

[0037] This factor is the weighted sum of the influence of multiple potential risk events on the bid section , and the higher the value, the greater the external environmental risk currently faced by the bid section .

[0038] Through the above steps, the originally complex on-site information is systematically and quantitatively converted into a multi-dimensional feature vector in a unified format.

[0039] S2, based on the constructed multi-dimensional feature vector, a dynamic coupling network model is established, which can represent the time-varying coupling relationship between the bid sections.

[0040] The core purpose of step S2 is to sublimate these discrete vectors describing the state of a single bid section into a dynamic coupling network model that can macroscopically and comprehensively represent the time-varying coupling relationship between bid sections.

[0041] In a specific embodiment, the dynamic coupling network model is mathematically embodied as a dynamic adjacency matrix evolving over time. This matrix is not a static, pre-defined structure, but can be continuously reconstructed according to the real-time feature vectors input in step S1, thereby capturing the rapidly changing correlation characteristics in the construction environment. The establishment process of this model includes two core links: dynamic neighborhood identification and coupling weight calculation.

[0042] First, dynamic neighborhood identification is performed. In a large and complex construction system, treating all sections as pairs of connected components would not only introduce a large amount of redundant calculations, but could also affect critical decisions due to weakly correlated noise interference. Therefore, this invention preferably introduces a screening mechanism to identify, before calculating the specific connection strength, which sections' interactions have the most significant impact on the stability of the entire system at the current moment.

[0043] This identification process introduces global network information entropy. This is achieved through the concept of information entropy. As a physical quantity measuring the complexity and uncertainty of a system, its changes can sensitively reflect the evolution of the system's state. This step is achieved by calculating... Regarding any two sections (e.g., section) and section (multidimensional feature vectors) and The second-order mixed partial derivatives are used to measure the contribution of the interaction between these two segments to the overall uncertainty of the system. Accordingly, each segment... dynamic neighborhood set Determined by the following formula:

[0044] ;

[0045] In the formula, For section At any moment The dynamic neighborhood set; The global network information entropy of the dynamically coupled network model; For section The multidimensional feature vector; For section The multidimensional feature vector; For a moment The dynamic threshold is used to filter out segments with strong coupling relationships; It is a second-order mixed partial derivative operator.

[0046] Preferably, this threshold can be adaptively adjusted based on the system's entropy change history over a period of time. When the system tends to be stable, the threshold can be appropriately increased to focus on the most critical coupling relationships; when the system state fluctuates drastically, the threshold can be appropriately decreased to capture more potential correlations. This makes the construction of the network topology itself environmentally adaptable, and since this is existing technology, it will not be elaborated further in this paper.

[0047] After determining the dynamic neighborhood set, the coupling strength of the segment pairs within the neighborhood is quantified, i.e., the dynamic adjacency matrix is ​​filled. elements The calculation of this element follows a basic principle: the more similar the states of two segments, the greater the potential for mutual influence. Here, "state" refers to the multidimensional feature vector constructed in step S1.

[0048] In a specific implementation, element It is given by the following formula:

[0049] ;

[0050] In the formula, For the dynamic adjacency matrix at time... Element; For section At any moment The multidimensional feature vector; For section At any moment The multidimensional feature vector; For section At any moment The multidimensional feature vector; It is a natural exponential function; For section At any moment A dynamic neighborhood set.

[0051] The formula is constructed using the following logic:

[0052] The core of the molecule is the calculation of the standard segment. and section Feature vector and Euclidean distance between This distance intuitively reflects the degree of difference between the two in the four-dimensional feature space.

[0053] This distance is transformed into a similarity score using a negative exponential function; the smaller the distance, the higher the score. Parameters As a preset distance scaling factor, it is used to adjust the sensitivity of the influence of distance on the final weight.

[0054] The denominator is a normalization term, which will divide the segment. For all other segments in its neighborhood The similarity scores are summed. The significance of this normalization operation is to make the values ​​from the standard segment... The sum of the weights of all starting edges is a constant, which makes It can be understood as a section its neighboring sections The distribution of relative influence.

[0055] S3. Adaptively adjust the scheduling priority of each task based on the global state changes of the dynamic coupled network model.

[0056] Step S3 is a crucial transitional step after obtaining the dynamically coupled network model constructed in step S2, but before making game-theoretic optimization decisions. It abandons the static and fixed priority rules of traditional scheduling methods and instead establishes an adaptive priority adjustment mechanism that can proactively perceive and respond to changes in the global state of the system. This aims to dynamically direct the attention of the scheduling system towards the operational tasks that have the greatest impact on the stability and coordination of the entire construction system.

[0057] In one specific embodiment, this adaptive adjustment process is not based on isolated, local task information, but rather directly on the network information entropy that characterizes the global complexity of the system. The process is hooked up. An integral-decrease hybrid dynamics model is used for each job to be scheduled. Scheduling priority at time t The model will be continuously updated. Its specific mathematical expression is as follows:

[0058] ;

[0059] In the formula, For homework assignments At any moment Scheduling priority; The initial priority of the task; This is the time decay coefficient; It is a non-linear activation function; The global network information entropy of the dynamically coupled network model; For the task The corresponding multidimensional feature vector of the segment; is the base of the natural logarithm; The current moment; This is the starting time of integration; It is the integral variable.

[0060] The ingenuity of this formula lies in the fact that it consists of two complementary parts with clear physical meanings:

[0061] The first part is the priority decay term, i.e. .

[0062] Its technical essence lies in introducing a "time forgetting" mechanism into the priority system. In the formula, Represents the task / assessment The baseline priority assigned during the initial planning, and It is a time-varying phenomenon. an exponential decay factor with a preset time decay coefficient Control.

[0063] The purpose of this item is to ensure that the priority of any task will not be permanently high due to its initial value, and if a task has not produced significant dynamic effects on the system for a long time, its priority should be controlled to fall back. This is a necessary design to ensure the fairness of the scheduling system and prevent the phenomenon of "task starvation", that is, to avoid some low initial priority tasks being delayed indefinitely.

[0064] The second part is the entropy feedback integral term, that is .

[0065] This item is the core of the priority adaptive adjustment of the present application, which establishes a direct feedback path from the "global system state" to the "local task priority". The specific role of each component part is described as follows:

[0066] Core driving force : This item is the first-order partial derivative of the global network information entropy of the task with respect to the multi-dimensional feature vector of the bid section. It directly quantifies the marginal contribution rate of the state of the bid section to the overall system uncertainty (i.e. entropy) in a physical sense. If the partial derivative value is large, it means that the bid section is a "sensitive point" or "unstable source" in the current system, and any change in it may cause a severe fluctuation in the overall system.

[0067] Nonlinear regulator : In order to prevent the partial derivative value from being extreme and causing impact on the priority system, a nonlinear activation function (such as Sigmoid or Tanh function) is preferably introduced to process the partial derivative value. The function can smoothly map any range of input to a bounded and normalized interval, playing a role in stabilizing the adjustment and highlighting the importance.

[0068] Historical impact accumulator : The definite integral operator here embodies the "memory effect". The scheduling system not only focuses on the influence of the task at the current instant, but also accumulates its continuous influence in the past period of time. If a task continuously becomes a source of system instability, its entropy contribution will be continuously accumulated through integration, so that its priority is continuously and significantly improved until the problem associated with it is solved.

[0069] In summary, step S3 realizes an intelligent priority management by dynamically coupling the task priority with the system entropy change.

[0070] S4, acquire a limited time window of forward rolling, and adopt a leader-follower game optimization method in the limited time window, wherein a dynamic coupling network model and an adjusted scheduling priority are combined to obtain a ship scheduling strategy in a current scheduling period;

[0071] Step S4 adopts a more flexible and adaptive leader-follower game optimization method, that is, a Stackelberg game model. This model can accurately simulate the hierarchical decision structure in the multi-section cross-operation scene: there is a "leader" (that is, a global scheduling system) that needs to be overall coordinated from a global perspective, and there are multiple "followers" (that is, each construction section) that have certain autonomous decision-making power and pursue their own maximum benefits.

[0072] In this game framework, the optimization goal of the leader is macroscopic. It does not directly intervene in the micro-operation of each ship, but is committed to minimizing two global costs: one is the adjustment cost of the scheduling strategy, which aims to maintain the continuity and stability of the scheduling plan and avoid chaos caused by drastic changes in the strategy; the other is the network coupling risk of the entire system, which is quantified through the dynamic coupling network model constructed in step S2. The leader tends to adopt macro strategies that can reduce the weight of the strongly coupled edges in the network and promote the decoupling of the system.

[0073] Correspondingly, each follower, that is, the section , makes decisions under the macro strategy framework given by the leader to maximize its own utility function . The specific mathematical expression is:

[0074] ;

[0075] In the formula, is the utility function of the follower, which represents the section ; is the utility weight coefficient of the section that can be adaptively adjusted; , , is the utility weight coefficient of the section that is preset to balance different objectives; is the time cost of the section ; is the conflict index of the section ; is the resource utilization rate of the section ; is the global network information entropy of the dynamic coupling network model; and is the multi-dimensional feature vector of the bid ; is the natural logarithm function ; is the first-order time derivative operator.

[0076] is the core cost-risk trade-off term. It combines the time cost with the conflict index in a non-linear power function form. The negative impact on the total utility can be amplified exponentially as the conflict risk increases, thus strongly discouraging the bid from choosing high conflict risk behaviors.

[0077] is the resource utilization benefit term. The logarithm function form is adopted to reflect the law of “diminishing marginal utility” of resource utilization. That is, when the resource utilization rate is low, improving its utilization rate will bring significant utility growth; while when the utilization rate is already high, the utility growth brought by further improvement will slow down.

[0078] is a design that embodies the global perspective. It directly incorporates the rate of change of the bid 's impact on the overall system uncertainty into its own utility function. Thus, when a bid's behavior will lead to an accelerated increase in the system's entropy, its own utility will be impaired. It can internalize the consideration of global stability into the self-conscious behavior of each follower, guiding it to spontaneously choose those schemes that are conducive to the system's tendency towards stability and order when making decisions.

[0079] This step also seamlessly integrates the dynamically adjusted scheduling priority in step S3 into the game process. The integration method is to dynamically adjust the core weight coefficient in the above utility function, and the adjustment process is described by the following formula:

[0080] ;

[0081] wherein, is the adaptive adjustment value of the utility weight coefficient at time ; is the initial value of the weight coefficient; is the hyperbolic tangent function; is the preset adjustment sensitivity parameter; is the variance calculation function; is the numerical set of the task priority related to the bid in the past time window ; is the preset time window length for calculating the priority variance.

[0082] The formula links the "influence" or "decision-making power" of a bid in the game with the "instability" of its own state. In the formula, the instability of the bid in the recent period is measured by calculating the variance of the priority of the task related to the bid i in the past time window If the variance is large, it means that the bid is a key disturbance source or bottleneck in the current system, and at this time, the weight of the bid is significantly increased by the function , so that the component of its utility function in the whole game system is increased, and its demand will be given priority. Thus, the focus of decision-making of the scheduling system can be adaptively aligned with the most critical and most unstable link.

[0083] Finally, since the game between the leader and the follower does not go on indefinitely, the present application sets a double convergence criterion as the termination condition of the game iteration. The criterion requires that not only the gradient norm of the global total utility function of the game needs to be less than a first preset threshold for judging the convergence of the optimization process to ensure the mathematical optimality of the solution, but also the relative change rate of the global network information entropy of the dynamic coupling network model must be less than a second preset threshold for judging the stability of the network model to ensure that the decision is made based on a relatively stable system state. Only when both conditions are met, the iteration process will be terminated, and the equilibrium solution obtained at this moment will be output as the final ship scheduling strategy in the current scheduling period.

[0084] S5, output and execute the ship scheduling strategy, and synchronously slide the time window forward to perform scheduling in the next period, and perform the cycle until all construction tasks are completed;

[0085] This step is responsible for converting the optimized scheduling strategy generated in the previous step into actual work on site and realizing the cyclic iteration of the entire scheduling process.

[0086] First, this step outputs the ship scheduling strategy obtained in step S4 as a set of work instructions, which is then issued to each construction unit for execution.

[0087] While the strategy in the current period is being executed, the system synchronously slides the limited time window of scheduling optimization forward. Then, the entire process returns to step S1, and a new round of scheduling optimization process is started based on the latest real-time data on site.

[0088] At this time, the cycle will continue until all scheduled construction tasks are completed, thereby realizing dynamic and closed-loop control of the entire construction process.

[0089] In a preferred embodiment, to ensure the stability of the system in long-term operation, this step can further include a high-order coupling control mechanism. This mechanism unifies the evolution of both by solving a set of coupled differential equations describing the mutual evolution of the scheduling strategy matrix and the dynamic adjacency matrix ;

[0090] ;

[0091] wherein, is the derivative operator with respect to time t; is the scheduling strategy matrix; is the dynamic adjacency matrix; is a preset coupling function describing the evolution law of the scheduling strategy matrix ; is a preset coupling function describing the evolution law of the dynamic adjacency matrix . This gives the entire scheduling management method stronger long-term operation robustness by dynamically binding the decision-making behavior (represented by ) and the system model cognition (represented by ).

[0092] While the embodiments of the application have been illustrated and described, it will be understood by those skilled in the art that various changes, modifications, substitutions and alterations can be made therein without departing from the spirit and scope of the application, which is defined by the appended claims and their equivalents.

Claims

1. A method for scheduling and managing construction vessels under multi-section overlapping operations, characterized in that, The method includes the following steps: S1. Obtain real-time operation data for each construction section and construct a multi-dimensional feature vector for each section that includes time, resources, operation and risk information; S2. Based on the constructed multidimensional feature vectors, establish a dynamic coupling network model that can characterize the time-varying coupling relationship between each segment; In step S2, the specific process for establishing the dynamically coupled network model is as follows: First, the dynamically coupled network model is transformed into a dynamic adjacency matrix. express; Then, the dynamic adjacency matrix is ​​calculated according to the following formula. elements This element represents a segment. benchmark section At any moment Influence weights: ； In the formula, For the dynamic adjacency matrix at time... Element; For section At any moment The multidimensional feature vector; For section At any moment The multidimensional feature vector; For section At any moment The multidimensional feature vector; It is a natural exponential function; For section At any moment The dynamic neighborhood set; The process of establishing the dynamic coupling network model also includes: Before calculating the dynamic adjacency matrix, the dynamic neighborhood set is determined. The specific process is as follows: First, calculate the global network information entropy of the dynamically coupled network model at time t. ; Then, the dynamic neighborhood set is determined by the following formula. : ； In the formula, For section At any moment The dynamic neighborhood set; The global network information entropy of the dynamically coupled network model; For section The multidimensional feature vector; For section The multidimensional feature vector; For a moment The dynamic threshold is used to filter out segments with strong coupling relationships; It is a second-order mixed partial derivative operator; S3. Adaptively adjust the scheduling priority of each task based on the global state changes of the dynamic coupled network model. S4. Obtain the finite time window for forward rolling, and within the finite time window, adopt the leader-follower game optimization method. In the leader-follower game optimization method, combine the dynamic coupling network model and the adjusted scheduling priority to obtain the ship scheduling strategy within the current scheduling cycle. S5. Output and execute the ship scheduling strategy, and simultaneously slide the time window forward to schedule the next cycle, repeating the process until all construction tasks are completed.

2. The method for scheduling and managing construction vessels under multi-section cross-operation as described in claim 1, characterized in that, In step S1, a multidimensional feature vector is constructed for each segment k. The specific process is as follows: First, define the multidimensional feature vector. The structure is as follows: ； In the formula, For section At any moment The multidimensional feature vector; This is the time coupling factor; For resource competitiveness; This is the operational fluctuation coefficient; For emergency relevance; Then, based on the acquired real-time job data, each component in the vector is calculated: Time coupling factor It is obtained by averaging the rate of change of the time impact of the related sections; resource competition By calculating the section The ratio of the intersection of the required resources with those of other sections to the total resource set is obtained; Operational fluctuation coefficient By calculating the section The variance of the work efficiency index within the past time window was obtained; Emergency Relevance By analyzing the affected sections The weighted summation of various emergencies is obtained.

3. The method for scheduling and managing construction vessels under multi-section cross-operation as described in claim 1, characterized in that, In step S3, the step of adaptively adjusting the scheduling priority of each job task includes: Update each task using the following formula: At any moment scheduling priority : ； In the formula, For homework assignments At any moment Scheduling priority; The initial priority of the task; This is the time decay coefficient; It is a non-linear activation function; The global network information entropy of the dynamically coupled network model; For the job task The corresponding multidimensional feature vector of the segment; is the base of the natural logarithm; The current moment; This is the starting time of integration; It is the integral variable.

4. The method for scheduling and managing construction vessels under multi-section cross-operation as described in claim 3, characterized in that, The steps of using the leader-follower game optimization method include: First, the global scheduler, acting as the leader, solves its optimization problem with the goal of minimizing the stability of scheduling policy adjustments and network coupling risks. Then, each segment, acting as a follower, maximizes its own utility function within the strategy framework given by the leader. Decisions are made for the objective until the entire game system reaches Stackelberg equilibrium, thereby obtaining the ship scheduling strategy.

5. The method for scheduling and managing construction vessels under multi-section cross-operation as described in claim 4, characterized in that, The follower's utility function The function, determined by the following formula, integrates time cost, conflict risk, resource utilization, and contribution to the global network state: ； In the formula, For follower, indicating segment Utility function; For section Adaptively adjustable utility weighting coefficients; , , For section Pre-defined utility weighting coefficients used to balance different objectives; For section Time cost; For section Conflict index; For section Resource utilization rate; The global network information entropy of the dynamically coupled network model; For section The multidimensional feature vector; It is the natural logarithm function; It is the first-order partial derivative operator with respect to time.

6. The method for scheduling and managing construction vessels under multi-section cross-operation as described in claim 5, characterized in that, In step S4, the steps of using the adjusted scheduling priority include: The task scheduling priorities obtained in step S3 are used to... Recent changes in the value are used to adaptively adjust the weighting coefficients of the follower utility function in step S4. Its adjustment formula is: ； In the formula, Utility weighting coefficient At any moment The adaptive adjustment value; The initial value of the weighting coefficient; It is the hyperbolic tangent function; These are preset sensitivity adjustment parameters; This is the variance calculation function; To be consistent with the section The relevant task priorities in the past time window The set of values ​​within; This is the preset time window length used to calculate the priority variance.

7. The method for scheduling and managing construction vessels under multi-section cross-operation as described in claim 6, characterized in that, Step S4, which involves obtaining the ship scheduling strategy within the current scheduling period, also includes a termination condition for determining whether the game optimization is complete. This termination condition is a dual convergence criterion.

8. The method for scheduling and managing construction vessels under multi-section cross-operation as described in claim 7, characterized in that, The method also includes a coupling control step to ensure that the formulation of the scheduling strategy is coordinated with the evolution of the network model. The specific process is as follows: The scheduling strategy matrix is ​​unified by solving the following system of differential equations. With dynamic adjacency matrix The evolutionary process: ； In the formula, For the derivative operator with respect to time t; The scheduling strategy matrix; It is a dynamic adjacency matrix; A preset matrix used to describe scheduling strategies. The coupling function of evolutionary laws; A pre-defined matrix used to describe dynamic adjacency matrices. The coupling function of evolutionary laws.

Images