Intelligent stowage method in container wharf ship cabin

Abstract

The invention provides an intelligent stowage method in a container wharf ship cabin. The intelligent stowage method comprises the following steps that S1, information of a container to be assembled and pre-stowage information of a target ship are obtained; s2, based on the information of the container to be assembled and the pre-assembly information of the target ship, establishing a multi-target optimization function including a dynamic weight adjustment mechanism; s3, carrying out iterative solution through an improved Monte Carlo tree search algorithm; and S4, when a preset number of iterations is reached, outputting an optimal stowage scheme. Through the Monte Carlo search algorithm for dynamic weight adjustment, the defect of parameter solidification in multi-objective optimization of a traditional method is effectively overcome, the convergence speed of the algorithm is increased, and meanwhile, a local optimal trap is avoided. According to the multi-target cooperation mechanism, core indexes such as box turnover operation, box area conflicts and field bridge movement are brought into dynamic optimization, the effects of reducing the box turnover amount, reducing the box area conflicts and reducing invalid field bridge movement are achieved, and the operation cost can be obviously reduced.

Description

Technical Field

[0001] The present application relates to the technical field of container terminal stowage, and in particular to an intelligent stowage method within a ship cabin at a container terminal. Background Art

[0002] With the advancement of smart port construction, the traditional loading model that relies on the experience of loaders can no longer meet the efficient operation requirements of modern container terminals. Although existing intelligent loading technology can simulate human decision-making, it has significant shortcomings when handling large-scale loading tasks: Explosion of solution space: The complexity of container stowage combinations increases exponentially with the number of bays, making it difficult for the algorithm to converge (the solution time exceeds 2 hours when the number of bays on a conventional ship is >200); Poor constraint coupling: Existing methods use static thresholds for key factors such as ship stability constraints (such as heavy-pressure-light limits) and equipment coordination constraints (crane-quay crane linkage), and are unable to dynamically adapt to changes in the operating environment. Single evaluation dimension: Over-focusing on the single indicator of container turnover rate, ignoring key factors affecting operational safety, such as container area conflicts and three-dimensional moment balance; Parameter fixation defect: Key parameters such as the balance coefficient c in Monte Carlo search use fixed values, which can easily lead to local optimal traps. Summary of the Invention

[0003] The present invention aims to provide an intelligent stowage method for a container terminal ship cabin to solve the problems raised in the background technology, comprising the following steps: S1. Obtain the container information to be assembled and the pre-allocation information of the target ship, the container information to be assembled includes the weight of the container, the box area to be assembled, and the container group information; the pre-allocation information of the target ship includes the ship bay, the container group requirements of the ship bay, the pre-allocated ship bay within the ship, and the upper limit of the weight difference between the heavy box and the light box in a single row of the ship; S2. Based on the container information to be assembled and the pre-allocation information of the target ship, a multi-objective optimization function including a dynamic weight adjustment mechanism is established, and the multi-objective optimization function is expressed as:

[0004] Among them, M is the total number of optimization objectives, is the dynamic weight of the k-th objective function at the iteration number iter, which is calculated by the following formula:

[0005] Among them, iter is the current iteration number of Monte Carlo tree search, and are the maximum and minimum observed values of target k in the iter generation respectively; S3. Iterative solution is performed using an improved Monte Carlo tree search algorithm, wherein the improvement includes injecting dynamic weight coefficients into the node evaluation stage: Among them, Q is the cumulative reward value of the node, N is the number of node visits, t is the total number of simulations, and c is the balance coefficient. is the dynamic weight sum; S4. When the preset number of iterations is reached, the optimal loading solution is output, which includes a dynamic weight iteration curve, a three-dimensional moment distribution thermal map, and a ship resonance spectrum analysis report.

[0006] Furthermore, the objective function includes: a target for the number of container turnovers, a target for the energy consumption of the field bridge, and a target for the conflict between container areas. Specifically, Target number of box turnovers: ; in Indicates the difficulty coefficient of box rummaging; The number of containers to be turned over; Energy consumption targets for field cranes:

[0007] in represents the energy consumption coefficient per unit distance, Indicates the number of moving bays of field bridge b; Box area conflict target:

[0008] in is the conflict type weight, The number of times the corresponding type of conflict occurs.

[0009] Furthermore, the dynamic weight satisfies the following adjustment rules: when When the weight enhancement is triggered, Increase by 20%.

[0010] Furthermore, the improvements to the Monte Carlo tree search algorithm include: Dynamic weights are updated every 10 generations; When the number of node visits N ≥ 50, the balance coefficient c is dynamically adjusted according to the current device status.

[0011] Furthermore, the method also includes the following constraints: Ship resonance frequency constraints:

[0012] in, is the box vibration weight coefficient, is the indicator variable for 40-foot container, is the propagation natural frequency, is the maximum allowable vibration amplitude; Three-dimensional moment equilibrium constraint:

[0013] in, is the container mass parameter, (x, y, z) is the spatial reference coordinate, is the dynamic center of gravity of the ship, is the torque imbalance threshold.

[0014] The present invention also provides an intelligent loading device in a container terminal ship cabin, comprising: The data interface module is connected to the TOS system to obtain real-time status data of the quay crane / yard crane equipment to adjust the objective function coefficient; Dynamic weight calculation module, used to calculate the intergenerational normalized weights of multi-objective functions in real time; The Monte Carlo tree search node extension module dynamically generates, screens, and optimizes stowage paths to generate stowage plans within the ship's hold under multiple constraints. The three-dimensional moment check module is used to trigger the three-dimensional space moment balance check when the height of the container is ≥ 8 floors; The resonance suppression module is used to perform coupled calculations of the ship's natural frequency and tank position distribution.

[0015] Furthermore, the dynamic weight calculation module is linked with the Monte Carlo tree search node expansion module to synchronously update the weight parameters every 10 generations of iterations.

[0016] The present invention also provides an electronic device, comprising a memory and a processor, wherein the memory stores a computer program, and the processor implements the method according to any one of claims 1 to 5 when executing the computer program.

[0017] The present invention further provides a computer-readable storage medium storing a computer program, wherein when the computer program is executed by a processor, the computer program generates a loading plan including the following data: Dynamic weight iteration curve: contains three sets of data: weight value, intergenerational extreme value, and objective function value; 3D torque distribution heat map: color gradient indicates the degree of torque imbalance, threshold Dynamic adjustment according to draft depth; Ship vibration spectrum report: including natural frequencies and the actual vibration amplitude .

[0018] Beneficial effects of the present invention: This invention utilizes a Monte Carlo search algorithm with dynamic weight adjustment to effectively overcome the parameter rigidity inherent in traditional multi-objective optimization methods, accelerating algorithm convergence while avoiding local optimality traps. This multi-objective collaborative mechanism incorporates key metrics such as container turnover, container-area conflicts, and crane movements into dynamic optimization, reducing container turnover, container-area conflicts, and ineffective crane movements, significantly lowering operational costs.

[0019] This invention incorporates constraints that consider vessel stability and operational safety. By combining a dynamic resonant frequency suppression model with the coupled calculation of the ship's natural frequency and container space distribution, vibration spectrum analysis is automatically triggered when the number of stacked layers exceeds the specified value, reducing the ship's heel angle. A three-dimensional moment balance constraint significantly improves the accuracy of moment imbalance warnings by adjusting the threshold via a draft gradient, allowing for early detection of unsafe loading scenarios. Furthermore, a dual-loop parameter adjustment system enables real-time linkage between key coefficients (such as the balance coefficient c) and equipment status, improving algorithm iteration efficiency and significantly enhancing the flexibility and adaptability of the optimization process. DETAILED DESCRIPTION

[0020] The technical solutions in the embodiments of the present invention are clearly and completely described. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of them. All other embodiments derived by ordinary technicians in this field based on the embodiments of the present invention without making any creative efforts are within the scope of protection of the present invention. Example 1:

[0021] This embodiment provides a method for intelligent loading in a container terminal ship cabin, and the specific steps include: S1. Obtain the container information to be assembled and the pre-configured information of the target vessel; the above information is obtained from the TOS (Terminal Operating System) in real time through a standardized interface, and the communication protocol ensures dynamic data updates. The container information to be assembled includes: The container weight, i.e. the gross weight of the container, is obtained in real time through the container weighing data of the TOS system; Container area information to be matched: the container area number and container position number of the container yard; Container group information: group number by destination, size, and type.

[0022] The pre-assigned information of the target ship includes: Ship bay and container slot information: including the following fields: bay distribution, container slot load limit (e.g., maximum stack height 8 layers), container slot unblocking status (initial list of available loading locations); Container group requirements for ship container slots: The pre-allocated ship plan specifies the allowed container group number for each ship container slot; Pre-allocated container spaces within the vessel: The location, weight, and grouping information of loaded containers can be used to calculate the vessel's current center of gravity and remaining loading capacity; Upper limit of the weight difference between heavy containers and light containers in a single row of a ship: The weight difference threshold of the upper and lower containers in the same row (e.g. ≤ 5 tons) is used as one of the constraints of the overall algorithm (e.g. if the weight of the upper container exceeds that of the lower container and the difference is greater than the threshold, a stowage alarm is triggered).

[0023] S2. Based on the container information to be assembled and the pre-configured information of the target ship, a multi-objective optimization function including a dynamic weight adjustment mechanism is established. The multi-objective optimization function given in this embodiment is expressed as:

[0024] Among them, M is the total number of optimization objectives, is the corresponding objective function, is the dynamic weight of the k-th objective function at the iteration number iter.

[0025] This embodiment gives some objective functions For example, in addition to the examples given in this embodiment, other objective functions in the prior art can also be added to the overall objective function.

[0026] (1) Target number of box turnovers:

[0027] in The difficulty coefficient of container retrieval is used to characterize the complexity of container retrieval operations for container group i in the yard. It is positively correlated with the container stacking depth and equipment accessibility. Its value range is 1.0 (easy to retrieve from shallow layers) to 2.5 (difficult to retrieve from deep layers). A dynamic adjustment rule is introduced: when the number of stacking layers above container group i is ≥ 3, Increase by 0.3, if the container group i is located at the intersection of cross-regions (such as the intersection of Q01 and Q02), Upgraded to 2.0.

[0028] is the number of containers to be turned over, representing the number of containers in group i that need to be turned over due to being covered by other containers in the stacking yard. When there are 3 or more layers of stacking above group i, each container that is pressed is counted as one count. When stacking across bays, each cross-area container press count is counted as an additional 0.5 count.

[0029] (2) Energy consumption target of field crane:

[0030] in Represents the energy consumption coefficient per unit distance, which is used to reflect the energy consumption characteristics of the bridge b in different operating states, including no-load and full-load modes. Its value is dynamically adjusted by the TOS system in real time to obtain the bridge load state and is normalized based on the rated power of the bridge. This embodiment provides a specific value method: 0.8: The yard crane moves empty (no container lifting); 1.0: Standard load (single 20-foot container); 1.2: Overload operation (double 40-foot container).

[0031] Of course, it can also be modified according to the actual situation according to the different port operation conditions and container standards. The specific value of .

[0032] It indicates the number of moving bays of the yard crane b, that is, the number of times the yard crane moves across bays during the loading process. The actual movement trajectory of the yard crane is obtained in real time through the TOS system for verification.

[0033] (3) Box area conflict target:

[0034] in is the conflict type weight. In this embodiment, different conflict weights are divided according to the container type. Specifically: Dangerous goods conflicts are triggered when dangerous goods boxes and ordinary boxes are mixed in the same column. The value is 3.0; When the refrigerated container area and the ordinary container area are operated alternately, a refrigerated container conflict is triggered. The value is 2.0; When extra-wide / extra-high containers and standard containers are placed in the same bay, an oversized container conflict is triggered. The value is 2.5; When the above-mentioned multiple types of conflicts coexist, the maximum weight is taken, and the container type is identified through the quay crane OCR system to trigger the conflict counting.

[0035] The number of conflicts of corresponding types is counted as follows: Dangerous Goods conflicts are counted as one per column of mixed dangerous goods containers; Refrigerated Container conflicts are counted as one per alternating refrigerated container area; and Out-of-Gauge Container conflicts are counted as two per bay of mixed Out-of-Gauge Containers. These counts are automatically calculated by scanning the container type labels on the quay crane's OCR system.

[0036] In addition to the above description, there is also a dynamic correlation between parameters between the objective functions of this embodiment: (1) When >2.0 automatically triggers the field bridge Coefficient compensation mechanism ( =min(1.2, ×1.1)).

[0037] When the difficulty of turning the box When it exceeds 2.0 (usually caused by stacking across bays or stacking layers ≥ 5), it indicates that the current loading plan has a highly complex container turning operation. High difficulty container turning will inevitably lead to extended yard crane operation time. By improving The coefficient is used to strengthen the weight of the energy consumption of the yard bridge movement in the objective function, and to suppress the algorithm from generating loading plans that require frequent cross-zone movement.

[0038] When the difficulty coefficient of container turnover is too high, the yard crane may be overloaded during actual yard operations. Compensation limits the number of field bridge movements through mathematical constraints to avoid equipment overload.

[0039] (2) Dangerous goods conflicts ( =3.0) occurs, the difficulty coefficient of the box Forced to upgrade to 2.0 or above.

[0040] Based on the principle of safety first, dangerous goods containers must be strictly isolated and upgraded. The coefficient can make the algorithm give priority to allocating low-risk container locations to dangerous goods, reducing the probability of misoperation, and additional safety checks are required when dangerous goods are turned over. The coefficient amplifies the cost of turning over the container, guiding the algorithm to generate a loading plan that does not require a second overturn; In summary, by introducing the dangerous goods conflict adjustment coefficient mechanism, the The coefficient enables the algorithm to block high-risk loading paths in advance at the objective function level to avoid chain reactions that may be caused by conflicts between dangerous goods.

[0041] (3) Number of times the box is turned over >5 triggers the number of bridge moves Compensation mechanism = ceil (Actual number of moves × 1.2)).

[0042] High turnover rate ( >5) usually means that there are multiple layers (≥3 layers) or cross-bay stacking of containers in the yard. In this case, the yard crane needs to move frequently to perform the overturning operation, which means that the solution is energy-intensive and the complexity of the overturning of a single container group exceeds the normal threshold. Compensation increases the actual number of moves by 20%, emphasizing the cost of crane movement in the objective function to prevent the algorithm from generating high-risk scenarios requiring frequent cross-zone operations. The formula ceil(⋅) ensures that the number of moves is rounded to an integer, compensating for the additional energy consumption caused by container transfers (such as crane start-up and shutdown losses), thus meeting the core goal of minimizing crane movement.

[0043] In addition to the above-mentioned field bridge movement angle, from the overall consideration of the algorithm, when the number of box turning When the value is >5, the Monte Carlo tree search is prone to fall into local optimality (only optimizing the box flipping and ignoring the movement cost). Compensation, increase the weight of the mobile cost in the UCB formula (later The proportion of users increases), guiding the algorithm to explore the equilibrium solution of low movement and low box turnover.

[0044] Multi-objective function is the dynamic weight of the k-th objective function at the iteration number iter, which is calculated by the following formula:

[0045] Among them, iter is the current iteration number of Monte Carlo tree search, and are the maximum and minimum observed values of target k in the iter generation, respectively.

[0046] Dynamic weights must meet the following adjustment rules: when When the weight enhancement is triggered, Increase by 20%.

[0047] From the perspective of ship stowage, there is a natural conflict between container turning operation (F1) and yard crane movement (F2). When the F1 range exceeds the threshold, it indicates that the cost of container turning has increased sharply (for example, the number of cross-bay storage layers is ≥5 layers). At this time, the container turning cost is strengthened. High-risk container transfer scenarios can be prioritized. When the range of the crane movement target (F2) exceeds the standard, it usually corresponds to the crane's cross-zone operation frequency exceeding the safety threshold (in this embodiment, the cross-zone movement frequency for a single crane per day is set to ≤15 times). Weight reinforcement uses mathematical constraints to limit the number of movements to avoid equipment overload.

[0048] From the algorithm level, when the range of a single target exceeds 60% of the total range, it indicates that the impact of this target on the solution space in the current iteration is significantly higher than that of other targets. Weight enhancement aims to balance the optimization weights among multiple targets and avoid excessive dominance of the search direction by a single target. In addition, in the Monte Carlo tree search, if the range of a target continues to expand, it may cause the algorithm to fall into a local optimum. By improving Weight, forcibly increase the exploration cost of the target and guide the algorithm to jump out of the current search path.

[0049] S3. Iterate the solution using an improved Monte Carlo tree search algorithm. The improvement includes injecting dynamic weight coefficients into the node evaluation phase. The dynamic weights are updated every 10 generations. Specifically, in: Q is the node's cumulative reward value, which represents the weighted objective function improvement obtained by the node in the historical simulation. According to the embodiment data, the typical value range is 10 3 -10 5 ; N is the number of node visits, which is the number of times the node is selected during the Monte Carlo search process. When N ≥ 50, the weight correction mechanism is triggered; t is the total number of simulations, linked to iter and 100≤t≤2000; c is the balance coefficient, ranging from 1.5≤c≤2.5; is the dynamic weight sum.

[0050] S4. When the preset number of iterations is reached, the optimal loading solution is output, which includes a dynamic weight iteration curve, a three-dimensional moment distribution thermal map, and a ship resonance spectrum analysis report. Example 2:

[0051] This embodiment is based on the intelligent stowage method for a container terminal ship cabin provided in Example 1, and further adds physical constraints to the method, including ship resonance frequency constraints and three-dimensional moment balance constraints. Specifically: (1) Ship resonance frequency constraints When the container layer height is greater than 5 layers, the stacking mass distribution changes the ship's natural frequency , which easily induces resonance. When the vibration amplitude Γ actual ≥0.8Γ max When , a red warning area is marked in the heat map of the output resonance frequency, prompting manual intervention.

[0052] The specific constraint formula is:

[0053] in: is the box vibration weight coefficient, and the calculation formula is:

[0054] Indicates the position b Gross mass of inner container, Indicates the total mass of containers on board. represents the stockpile height of the bay b, Indicates the maximum storage height allowed for a ship. In this embodiment, the maximum height given is 12 meters.

[0055] and It is expressed as a mass proportion term, reflecting the dominance of the mass at this position on the overall vibration; It is expressed as a height ratio term, quantifying the effect of high stockpiling on vibration amplification.

[0056] It is the indicator variable of 40-foot container, which is used to reflect the dominant vibration of 40-foot container. The length of 40-foot container is twice that of 20-foot container. Its stacking is more likely to induce lateral vibration mode. Therefore, when there is a 4-foot container at position b, Take 1, and the position b is a 20-foot container. Take 0.5; is the ship's natural frequency, and the change in the ship's mass distribution after loading directly affects , needs to be updated in real time to match the actual working conditions. The calculation formula is: is the ship stiffness coefficient (unit: N / m), which is related to the hull material and structure; is the effective mass (unit: kg), and the calculation formula is:

[0057] is the empty ship mass, and the coefficient of 0.7 reflects the inertial coupling effect between the loaded containers and the hull vibration.

[0058] is the maximum allowable vibration amplitude, which can be calculated by the following formula: .

[0059] This is the maximum number of stacking layers in the current stowage solution. High-level stacking (≥4 layers) leads to stress concentration on the hull, and the threshold needs to be relaxed to balance safety and solution feasibility. The gradient design of increasing by 0.02 every 3 layers is based on experimental data to ensure flexible adjustment under different storage conditions.

[0060] From the algorithmic point of view, this embodiment sets resonance constraints, mass distribution ( ) and box type ( ) quantifies the vibrational energy, Dynamically match ship status, Threshold level control forms a multi-dimensional security verification system. In Monte Carlo tree search, resonance constraints can directly prune illegal paths and reduce invalid calculations; dynamic weight A penalty factor is imposed on the illegal path to reduce its selection probability.

[0061] In practical operation, resonance constraints are set to avoid equipment loss and operation interruption caused by resonance. Threshold classification is used to increase the number of feasible solutions and improve the diversity of solutions.

[0062] (2) Three-dimensional moment balance constraint High-center-of-gravity storage (e.g., dangerous goods concentrated on the upper layer) will significantly increase the moment. If the lateral / longitudinal moment imbalance exceeds the threshold, it will cause the risk of capsizing the ship. Therefore, this embodiment provides a three-dimensional moment balance constraint:

[0063] in: is the mass of the i-th container, (x, y, z) is the spatial reference coordinate; The dynamic center of gravity of the ship is updated every time 10 new container stowage spaces are added. is the moment imbalance threshold, with a base value of 0.25, and is adjusted according to the current draft d:

[0064] 5. The intelligent stowage method for a container terminal ship cabin according to claim 1, characterized in that the method further includes the following constraints: Ship resonance frequency constraints:

[0065] in, is the box vibration weight coefficient, is the indicator variable for 40-foot container, is the propagation natural frequency, is the maximum allowable vibration amplitude; Three-dimensional moment equilibrium constraint:

[0066] in, is the container mass parameter, (x, y, z) is the spatial reference coordinate, is the dynamic center of gravity of the ship, is the torque imbalance threshold.

[0067] The ship's resonant frequency constraint causes moment imbalance, which triggers box-turning compensation operations. The constraint condition reduces such inefficient operations at the source.

[0068] In the output three-dimensional torque heat map, the imbalance area is marked with a color gradient: Green (<0.8 ): safe zone; Yellow (0.8 ∼ ): Warning area; Red (> ): Violation zone (automatic interception).

[0069] The ship resonance frequency constraint and three-dimensional torque balance constraint given in this embodiment are linked with the container flipping operation and yard bridge movement targets. A penalty factor is applied to the torque-exceeding path in the Monte Carlo tree search, reducing the probability of selecting its UCB value. Overall, this guides the algorithm to generate a low-torque, low-tumble, and low-movement equilibrium solution, effectively improving the quality of the final solution. Example 3:

[0070] This embodiment provides an intelligent loading device for a container terminal ship cabin, comprising: The data interface module is connected to the TOS system to obtain real-time status data of the quay crane / yard crane equipment to adjust the objective function coefficient; Dynamic weight calculation module, used to calculate the intergenerational normalized weights of multi-objective functions in real time; The Monte Carlo tree search node extension module dynamically generates, screens, and optimizes stowage paths to generate stowage plans within the ship's hold under multiple constraints. The three-dimensional moment check module is used to trigger the three-dimensional space moment balance check when the height of the container is ≥ 8 floors; The resonance suppression module is used to perform coupled calculations of the ship's natural frequency and tank position distribution.

[0071] Among them, the dynamic weight calculation module is linked with the Monte Carlo tree search node expansion module to synchronously update the weight parameters every 10 generations of iteration. Example 4:

[0072] This embodiment provides an electronic device, including a memory and a processor. The memory stores a computer program, and when the processor executes the computer program, any of the methods described in Embodiments 1 and 2 is implemented. Example 5:

[0073] This embodiment provides a computer-readable storage medium storing a computer program. When the computer program is executed by a processor, it generates a loading plan including the following data: Dynamic weight iteration curve: contains three sets of data: weight value, intergenerational extreme value, and objective function value; 3D torque distribution heat map: color gradient indicates the degree of torque imbalance, threshold Dynamic adjustment according to draft depth; Ship vibration spectrum report: including natural frequencies and the actual vibration amplitude .

[0074] The various embodiments in this specification are described in a progressive manner, and each embodiment focuses on the differences from other embodiments. The same or similar parts between the various embodiments can be referenced to each other.

[0075] The preferred embodiments of the present invention are described in detail above, but the present invention is not limited to the specific details in the above embodiments. Within the technical concept of the present invention, various equivalent transformations (such as quantity, shape, position, etc.) can be made to the technical solution of the present invention, and these equivalent transformations are all protected by the present invention.

Claims

1. An intelligent stowage method for a container terminal ship cabin, characterized in that: The steps include: S1. Obtain the container information to be assembled and the pre-allocation information of the target ship, the container information to be assembled includes the weight of the container, the box area to be assembled, and the container group information; the pre-allocation information of the target ship includes the ship bay, the container group requirements of the ship bay, the pre-allocated ship bay within the ship, and the upper limit of the weight difference between the heavy box and the light box in a single row of the ship; S2. Based on the container information to be assembled and the pre-allocation information of the target ship, a multi-objective optimization function including a dynamic weight adjustment mechanism is established, and the multi-objective optimization function is expressed as: ； Among them, M is the total number of optimization objectives, is the dynamic weight of the k-th objective function at the iteration number iter, which is calculated by the following formula: ；； Among them, iter is the current iteration number of Monte Carlo tree search, and are the maximum and minimum observed values of target k in the iter generation respectively; S3. Iterative solution is performed using an improved Monte Carlo tree search algorithm, wherein the improvement includes injecting dynamic weight coefficients into the node evaluation stage: ； Among them, Q is the cumulative reward value of the node, N is the number of node visits, t is the total number of simulations, and c is the balance coefficient. is the dynamic weight sum; S4. When the preset number of iterations is reached, the optimal loading solution is output, which includes a dynamic weight iteration curve, a three-dimensional moment distribution thermal map, and a ship resonance spectrum analysis report.

2. The intelligent stowage method for a container terminal ship cabin according to claim 1, characterized in that: The objective function includes: the number of container turnover targets, the field bridge energy consumption target and the container area conflict target. Specifically, Target number of box turnovers: ； in Indicates the difficulty coefficient of box rummaging; The number of containers to be turned over; Energy consumption targets for field cranes: ； in represents the energy consumption coefficient per unit distance, Indicates the number of moving bays of field bridge b; Box area conflict target: ； in is the conflict type weight, The number of times the corresponding type of conflict occurs.

3. The intelligent stowage method in a container terminal ship cabin according to claim 1, characterized in that: The dynamic weights satisfy the following adjustment rules: when When the weight enhancement is triggered, Increase by 20%.

4. The intelligent stowage method for a container terminal ship cabin according to claim 1, characterized in that: Improvements to the Monte Carlo tree search algorithm include: Dynamic weights are updated every 10 generations; When the number of node visits N ≥ 50, the balance coefficient c is dynamically adjusted according to the current device status.

5. The intelligent stowage method in a container terminal ship cabin according to claim 1, characterized in that: The method also includes the following constraints: Ship resonance frequency constraints: ； in, is the box vibration weight coefficient, is the indicator variable for 40-foot container, is the propagation natural frequency, is the maximum allowable vibration amplitude; Three-dimensional moment equilibrium constraint: ； in, is the container mass parameter, (x, y, z) is the spatial reference coordinate, is the dynamic center of gravity of the ship, is the torque imbalance threshold.

6. An intelligent loading device for a container terminal ship cabin, characterized in that: include: The data interface module is connected to the TOS system to obtain real-time status data of the quay crane / yard crane equipment to adjust the objective function coefficient; Dynamic weight calculation module, used to calculate the intergenerational normalized weights of multi-objective functions in real time; The Monte Carlo tree search node extension module dynamically generates, screens, and optimizes stowage paths to generate stowage plans within the ship's hold under multiple constraints. The three-dimensional moment check module is used to trigger the three-dimensional space moment balance check when the height of the container is ≥ 8 floors; The resonance suppression module is used to perform coupled calculations of the ship's natural frequency and tank position distribution.

7. The loading device according to claim 6, characterized in that: The dynamic weight calculation module is linked with the Monte Carlo tree search node expansion module to synchronously update the weight parameters every 10 generations of iterations.

8. An electronic device comprising a memory and a processor, wherein the memory stores a computer program, wherein: When the processor executes the computer program, the method according to any one of claims 1 to 5 is implemented.

9. A computer-readable storage medium storing a computer program, characterized in that: When the computer program is executed by a processor, it generates a loading plan including the following data: Dynamic weight iteration curve: contains three sets of data: weight value, intergenerational extreme value, and objective function value; 3D torque distribution heat map: color gradient indicates the degree of torque imbalance, threshold Dynamic adjustment according to draft depth; Ship vibration spectrum report: including natural frequencies and the actual vibration amplitude .

Images