Palletizing method for air freight, storage medium, and program product

Abstract

This application provides a palletizing method, storage medium, and program product for air freight. By acquiring first information input by the user in natural language, the method directly identifies the user's intent to determine whether it is for palletizing. Upon successful identification, it automatically extracts key second information such as the target aircraft type, target pallet type, business priority, business attributes, and cargo dimensions and weight. Based on the extracted second information, it automatically determines the corresponding first palletizing scheme. Subsequently, based on the palletizing scheme, it uniformly generates and directly outputs a first task identifier, a first natural language summary, a first three-dimensional visualization model, and a first cargo loading list. This eliminates a large number of tedious operations such as manual information processing, manual scheme formulation, manual list organization, and manual result summarization, thereby effectively shortening the overall palletizing processing cycle and significantly improving palletizing efficiency.

Description

Technical Field

[0001] This application relates to the field of intelligent air freight technology, and in particular to a palletizing method, storage medium and program product for air freight. Background Technology

[0002] With the continuous growth of air cargo volume, palletizing is a key step before cargo loading, and its operational quality directly affects the utilization rate of flight cabins and flight safety.

[0003] Currently, the existing technology for air cargo palletizing mainly adopts a manual palletizing mode: operators need to manually collect basic data such as target aircraft type, pallet and container information, and the size and weight of each cargo, and combine business priorities and loading rules to determine the cargo placement order, pallet and container allocation and loading layout through experience.

[0004] The technical problem with the aforementioned existing technology is that manual pattern making is extremely inefficient: manually processing a single batch of pattern making tasks requires a lot of time for information verification, rule sorting and scheme calculation. When faced with complex scenarios with multiple goods and multiple constraints, the pattern making cycle is long and prone to oversights. Summary of the Invention

[0005] The air freight palletizing method, storage medium, and program product provided in this application are intended to improve palletizing efficiency.

[0006] In a first aspect, embodiments of this application provide a palletizing method for air freight, including:

[0007] Obtain the first information, which is the information input by the user using natural language;

[0008] The intent of the first information is identified. When the intent identified from the first information is to measure the board box, the second information is extracted from the first information. The second information includes at least one of the following: target machine type, target board type, business priority, business attributes, and cargo information of each cargo. The cargo information includes at least one of the following: length, height, width, and weight.

[0009] Based on the second piece of information, the first board-making scheme is determined;

[0010] Based on the first template-making scheme, the first task identifier, the first natural language summary, the first three-dimensional visualization model, and the first cargo loading list are generated and output.

[0011] Secondly, embodiments of this application provide a palletizing device for air freight, comprising:

[0012] The acquisition module is used to acquire the first information, which is the information input by the user using natural language.

[0013] The processing module is used to perform intent recognition on the first information. When the intent recognized from the first information is to measure the board box, the second information is extracted from the first information. The second information includes at least one of the following: target machine model, target board type, business priority, business attributes, and cargo information of each cargo. The cargo information includes at least one of the following: length, height, width, and weight.

[0014] The processing module is also used to determine the first board-making scheme based on the second information;

[0015] The output module is used to generate and output a first task identifier, a first natural language summary, a first three-dimensional visualization model, and a first cargo loading list based on the first plate-making scheme.

[0016] Thirdly, embodiments of this application provide an electronic device, including: a memory and a processor;

[0017] The memory stores the instructions that the computer executes;

[0018] The processor executes computer execution instructions stored in memory, causing the processor to perform the first aspect and / or various possible implementations of the first aspect as described above.

[0019] Fourthly, embodiments of this application provide a computer-readable storage medium storing computer-executable instructions, which, when executed by a processor, are used to implement the first aspect and / or various possible implementations of the first aspect.

[0020] Fifthly, embodiments of this application provide a computer program product, including a computer program that, when executed by a processor, implements the first aspect and / or various possible implementations of the first aspect.

[0021] The air freight palletizing method, storage medium, and program product provided in this application embodiment obtain first information input by the user in natural language, directly identify the intent to determine whether it is for palletizing, and automatically extract key second information such as target aircraft type, target pallet type, business priority, business attributes, and cargo size and weight after successful identification; then, based on the extracted second information, the corresponding first palletizing scheme is automatically determined, replacing the process of manual experience-based trial calculation and repeated adjustments; subsequently, based on the palletizing scheme, a first task identifier, a first natural language summary, a first three-dimensional visualization model, and a first cargo loading list are uniformly generated and directly output, realizing the fully automated processing from natural language instruction parsing, key information extraction, palletizing scheme generation to multimodal result output, saving a lot of tedious operations such as manual information processing, manual scheme formulation, manual list organization, and manual result summarization, thereby effectively shortening the overall palletizing processing cycle and greatly improving palletizing efficiency. Attached Figure Description

[0022] The accompanying drawings, which are incorporated in and form part of this specification, illustrate embodiments consistent with this application and, together with the description, serve to explain the principles of this application.

[0023] Figure 1 A flowchart illustrating the palletizing method for air freight provided in this application embodiment. Figure 1 ;

[0024] Figure 2 A flowchart illustrating the palletizing method for air freight provided in this application embodiment. Figure 2 ;

[0025] Figure 3 A flowchart illustrating the palletizing method for air freight provided in this application embodiment. Figure 3 ;

[0026] Figure 4 A flowchart illustrating the palletizing method for air freight provided in this application embodiment. Figure 4 ;

[0027] Figure 5 A schematic diagram of the test board box assembly provided in the embodiments of this application;

[0028] Figure 6 A schematic diagram of the three-dimensional visualization model provided in the embodiments of this application;

[0029] Figure 7 A flowchart illustrating the loading planning provided in this application embodiment. Figure 1 ;

[0030] Figure 8 A flowchart illustrating the loading planning provided in this application embodiment. Figure 2 ;

[0031] Figure 9 A schematic diagram showing the orientation of the goods provided in the embodiments of this application;

[0032] Figure 10 This is a schematic diagram of the discretization of the board box provided in the embodiments of this application;

[0033] Figure 11 A simplified schematic diagram of the goods provided in the embodiments of this application;

[0034] Figure 12 A schematic diagram of poles and remaining space provided for embodiments of this application;

[0035] Figure 13 This is a schematic diagram of the task query provided in this embodiment;

[0036] Figure 14This is a schematic diagram of cargo plane tracking provided in an embodiment of this application;

[0037] Figure 15 A schematic diagram illustrating the query of knowledge related to plate making provided in an embodiment of this application;

[0038] Figure 16 A schematic diagram of the structure of the palletizing device for air freight provided in the embodiments of this application;

[0039] Figure 17 This is a schematic diagram of the structure of an electronic device provided in an embodiment of this application.

[0040] The accompanying drawings illustrate specific embodiments of this application, which will be described in more detail below. These drawings and descriptions are not intended to limit the scope of the concept in any way, but rather to illustrate the concept of this application to those skilled in the art through reference to particular embodiments. Detailed Implementation

[0041] Exemplary embodiments will now be described in detail, examples of which are illustrated in the accompanying drawings. When the following description relates to the drawings, unless otherwise indicated, the same numbers in different drawings denote the same or similar elements. The embodiments described in the following exemplary embodiments do not represent all embodiments consistent with this application. Rather, they are merely examples of apparatuses and methods consistent with some aspects of this application as detailed in the appended claims.

[0042] To address the aforementioned technical challenges, the following technical concept is proposed: Abandoning the outdated model of traditional air cargo palletizing that relies on manual information collection, intent analysis, and scheme formulation, this approach uses natural language interaction as the user entry point. First, it acquires the user's input in natural language, then performs targeted intent recognition for the palletizing scenario, accurately determining whether the information's purpose is palletizing calculation. After confirming the calculation intent, without manual analysis, it automatically extracts core secondary information from the primary information, including the target aircraft type, target pallet size, business priority, business attributes, and cargo dimensions (length, width, height, and weight), mitigating the risk of errors in manual parameters through standardized extraction results. Then, based on this structured secondary information, it intelligently matches and determines a suitable primary palletizing scheme, replacing manual trial calculations. Finally, based on the generated palletizing scheme, it simultaneously produces a uniquely identifiable primary task identifier, a readily viewable primary natural language summary, an intuitive 3D visualization model of the layout, and a detailed primary cargo loading list. Integrating these multiple outputs achieves a closed-loop intelligent processing from user natural language command input to the delivery of all palletizing elements, simplifying manual operations and streamlining the palletizing process.

[0043] The technical solution of this application and how the technical solution of this application solves the above-mentioned technical problems are described in detail below with specific embodiments. These specific embodiments can be combined with each other, and the same or similar concepts or processes may not be described again in some embodiments. The embodiments of this application will now be described with reference to the accompanying drawings.

[0044] Figure 1 A flowchart illustrating the palletizing method for air freight provided in this application embodiment. Figure 1 The methods described above in this application's embodiments can be applied to any electronic device. For example... Figure 1 As shown, the method includes:

[0045] S101. Obtain the first information, which is the information input by the user using natural language.

[0046] The first piece of information is the prototype request submitted by the user in the form of natural language or a structured template. It can be either spoken dialogue text or structured data files such as Excel. It is the original input source for the entire prototype process and carries the user's core demands and constraints.

[0047] Specifically, the system proactively receives natural language text input by the user or uploaded structured files through a dialogue interface or file upload interface. It comprehensively collects the user's prototyping requirements, including but not limited to cargo information, machine model and prototyping requirements, and business constraints. Simultaneously, it performs preliminary format and completeness checks on the input to ensure smooth subsequent semantic parsing and intent recognition. For example, users can directly input descriptive text or upload a pre-filled Excel spreadsheet of cargo information; the system will uniformly receive and convert both into processable text or structured data formats.

[0048] For example, such as Figure 5 As shown, Figure 5 The schematic diagram of the calculation board box palletizing provided in the embodiment of this application includes: the user inputs "29 pieces of goods with dimensions of 135&95&63, weight of 120kg, B752PAG board, how to palletize?", this text is the first information; if the user uploads an Excel template containing the dimensions of the goods, weight, and model board type in another way, this file will also be completely obtained by the system as the first information.

[0049] S102. Perform intent recognition on the first information. When the intent recognized from the first information is to measure the board box, extract the second information from the first information. The second information includes at least one of the following: target machine type, target board type, business priority, business attributes, and cargo information of each cargo. The cargo information includes at least one of the following: length, height, width, and weight.

[0050] The second information is a set of core parameters that are decomposed, extracted and standardized from the first information. It includes the necessary constraints for pattern making, such as the target model, target board type, cargo size and weight, special attributes, and business priority. It serves as a bridge connecting user needs and algorithm calculations.

[0051] Specifically, the system's built-in large language model first performs sentence-by-sentence semantic decomposition and contextual understanding on the initial information to identify the core intent—for example, determining whether the user is asking about the palletizing solution, querying historical tasks, or consulting domain knowledge. Once the intent is confirmed to be "calculating palletizing boxes," key information such as the target model, pallet type, cargo dimensions (length, width, height), weight, whether it is flippable, and whether it needs to be placed at the bottom layer is automatically extracted from unstructured natural language or structured files through entity extraction, rule matching, and other methods. At the same time, missing default parameters are filled in (such as the default that the cargo can be stacked normally and the business priority is normal), transforming the scattered user input into a standardized parameter set that the machine can directly read, providing clear constraint inputs for subsequent algorithm calculations.

[0052] For example, such as Figure 5 As shown, after recognizing the user's input of natural language as the requirement for prototyping and measurement, the system extracts the second information: the target model is B752, the target board type is PAG, the cargo size is 135×95×63cm, the single piece weight is 120kg, and the cargo quantity is 29 pieces. At the same time, it adds the special attribute of "flippable / cannot be placed at the bottom". Just like the cargo information table and board model information shown in the figure, the original colloquial description is transformed into a structured parameter list.

[0053] S103. Based on the second information, determine the corresponding first plate-making scheme.

[0054] Specifically, the system transmits the standardized second information to the intelligent palletizing algorithm engine in the background. The engine first loads the physical constraints of the target model and pallet type (such as the maximum volume, weight limit, and size boundary of a certain pallet box), and then optimizes the loading volume and weight as the core objective, while taking into account the special requirements of the goods (such as non-rotatable, need to be placed at the bottom), load balance, stacking stability and other constraints. Through feasible combination optimization strategies such as greedy algorithms, simulated annealing or genetic algorithms, it iteratively solves the feasible loading layout step by step.

[0055] S104. Based on the first plate-making scheme, generate and output the first task identifier, the first natural language summary, the first three-dimensional visualization model, and the first cargo loading list.

[0056] Among them, the first task identifier is a globally unique ID generated for this stencil calculation, used for persistent storage and traceability of the solution; the first natural language summary is a summary of the key conclusions of the stencil results, which helps users quickly understand the core of the solution; the first 3D visualization model is a loading layout scene rendered by a 3D graphics engine, which intuitively shows the specific position of the goods on the pallet; the first cargo loading list is a detailed table that records the size, weight, placement position and pallet information of each item. The four together constitute a complete stencil result set.

[0057] Specifically, based on the generated palletizing scheme, the system first creates a unique task ID (such as a timestamp + random sequence) and associates and binds all input and output data of this calculation for subsequent querying and tracing. Then, the large language model performs in-depth analysis and summarization on the original results output by the algorithm (such as space utilization, number of pallets and boxes used, cargo location sequence, and remaining cargo status) to generate easy-to-understand natural language summaries and extract core conclusions such as "the number of cargo that can be loaded, the utilization rate of pallets and boxes, and the status of remaining cargo". At the same time, the system calls the 3D graphics engine to render an interactive 3D loading model based on the location coordinates and size data of the cargo, supporting users to rotate, zoom, and cross-section view. Finally, the system compiles a loading list containing cargo number, size, weight, placement position, and corresponding pallet and box information, and generates and associates the four types of results simultaneously under the same task ID. The system integrates the four types of results and pushes them to users through the front-end interactive interface. Users can view the unique task ID, read a concise result summary, and click links to view an interactive 3D loading model. They can also download or view a detailed cargo loading list and remaining cargo details online. If connecting to external systems, the system can also output structured data such as task ID, summary, 3D model link, and list data through API interfaces, enabling cross-system transfer and reuse of solutions and completing a closed loop from user requirement input to prototype delivery.

[0058] For example, such as Figure 5 As shown, for this calculation, the system generated task ID "20251226090252806771", which can be summarized as "This calculation is for a B752 aircraft PAG pallet, containing 29 pieces of cargo measuring 135×95×63cm / 120kg. 10 pieces can be loaded, resulting in a pallet / container volume utilization rate of approximately 66.8%. The remaining 19 pieces will be processed later." Figure 6 As shown, Figure 6This is a schematic diagram of the three-dimensional visualization model provided in the embodiments of this application. The three-dimensional visualization model will intuitively display the stacking position and orientation of 10 goods on the PAG board, as well as the loading volume ratio, loading weight ratio and model identifier. Users can click on any goods to view its attributes. The loading list records in detail the size, weight, placement coordinates and corresponding pallet information of each goods, just like the calculation result table, remaining goods details and 3D map link shown in the figure.

[0059] The air freight palletizing method provided in this application obtains the first information input by the user in natural language, directly identifies its intent to determine whether it is for palletizing, and automatically extracts key second information such as target aircraft type, target pallet type, business priority, business attributes, and cargo size and weight. Based on the extracted second information, the corresponding first palletizing scheme is automatically determined, replacing the process of manual trial calculation and repeated adjustments. Subsequently, based on the palletizing scheme, a first task identifier, a first natural language summary, a first three-dimensional visualization model, and a first cargo loading list are uniformly generated and directly output. This realizes the fully automated processing from natural language instruction parsing, key information extraction, palletizing scheme generation to multimodal result output, eliminating a large number of tedious operations such as manual information processing, manual scheme formulation, manual list organization, and manual result summarization, thereby effectively shortening the overall palletizing processing cycle and greatly improving palletizing efficiency.

[0060] In one possible implementation, Figure 7 A flowchart illustrating the loading planning provided in this application embodiment. Figure 1 ,like Figure 7 As shown, step S103 can also be:

[0061] Based on the second information, determine the board-making scheme for the i-th time, where i=0;

[0062] For the i-th board-making scheme, the following steps are performed:

[0063] The loading plan is performed for the i-th board-making scheme, and a score is obtained for the i-th scheme.

[0064] When i equals 0, the score of the i-th solution is determined as the historical score;

[0065] If the score of the solution in the i-th iteration is greater than the historical score, then the optimized solution in the i-th iteration is determined as the first solution.

[0066] Based on the i-th board-making optimization scheme and preset rules, generate board-making schemes for the neighborhood;

[0067] The board-making scheme of the neighborhood is determined as the board-making scheme of the (i+1)th time.

[0068] i is an integer, ranging from 0 to N, where N is a positive integer.

[0069] Specifically, firstly, the system generates an initial palletizing scheme for the 0th iteration based on the second information, serving as the starting point for the entire optimization process. This initial scheme only needs to satisfy basic loading constraints and does not require to be optimal. Then, it enters an iterative loop: for the current ith palletizing scheme, the system adjusts the loading plan, such as fine-tuning the order of goods placement, pallet / box allocation, or orientation, to obtain the ith optimized palletizing scheme. Next, this optimized scheme is scored, with scoring dimensions typically including pallet / box space utilization, weight distribution balance, and compliance. In the first iteration (i=0), the score of the 0th scheme is directly set as the historical score, serving as the benchmark for subsequent comparisons. If the score of the ith scheme is higher than the current historical score, this optimized palletizing scheme is updated as the first palletizing scheme, and the historical score is also updated. Afterward, based on the current optimal palletizing scheme and preset rules, a neighboring palletizing scheme is generated, which involves making minor adjustments to the current optimal layout, serving as the (i+1)th palletizing scheme for the next iteration. This loop will continue to execute until the number of iterations i reaches the preset maximum value N, and the final output of the first board-making scheme is the approximate optimal result obtained after multiple rounds of iteration.

[0070] For special cases: If the initial solution in step 0 is already optimal, after the first scoring, since the scores of neighboring solutions generated in subsequent iterations will not exceed the initial score, the optimized solution in step 0 will be directly determined as the first solution, and subsequent iterations will not change this result; if the score of a solution in a certain iteration is equal to the historical score, the first solution will not be updated, and neighboring solutions will continue to be generated for the next round; if the scores of all neighboring solutions cannot exceed the historical scores, the iteration will continue until the maximum number of iterations N, and finally the current optimal first solution will be retained; if a neighboring solution generated in a certain step during the iteration violates the loading constraint, the system will automatically discard the solution and regenerate it, ensuring that the solution in each iteration is feasible.

[0071] The neighborhood slab-breaking scheme is a new candidate scheme generated by small-scale, local perturbations based on the current optimal slab-breaking scheme. The core idea is to explore better loading possibilities without destroying the rationality of the original layout.

[0072] It is possible that the system will generate a neighborhood scheme by performing the following typical operations based on the current i-th stencil optimization scheme and preset rules:

[0073] Randomly select 2-3 items from the already placed goods and swap their positions or orientations on the pallet. For example, swap the positions of a piece of goods in the corner of the pallet with a piece of goods in the middle area, or adjust the goods that were originally laid flat to be placed vertically, thus changing the layout structure within a small range.

[0074] If the current solution uses multiple pallets / containers, randomly select a piece of goods, move it from the current pallet / container to another pallet / container, and readjust the local layout of the target pallet / container to ensure that the weight and volume constraints are still met after the move, and try to optimize the overall pallet / container utilization rate.

[0075] Reorder segments of the current cargo loading sequence, such as shuffling the order of 5 consecutive items and replanning them, to explore more compact stacking methods without changing the overall pallet and container allocation.

[0076] Select a local area on the pallet (such as a stack of goods), disassemble it, and re-plan the loading. Try to fill the area in a more efficient way to improve the utilization of local space.

[0077] These operations are all small-scale perturbations that do not fundamentally modify the entire prototyping scheme. They retain the reasonable parts of the current optimal scheme while exploring potential better layouts. After generation, the system verifies whether the new scheme meets constraints such as load safety, size, and weight. If it does, it is used as the (i+1)th prototyping scheme to enter the next iteration; if it does not, it is discarded and regenerated until a feasible neighborhood scheme is obtained.

[0078] Starting from the initial palletizing scheme, the process of "loading planning → scheme scoring → updating the optimal solution → generating neighboring schemes" is executed iteratively to gradually approach a better loading layout, and finally outputs the first palletizing scheme that is close to the optimal solution. This not only ensures the feasibility of the solution, but also continuously improves the utilization rate of pallet space and the rationality of loading through multiple iterations. It avoids the limitations of manual palletizing or single-time planning, making the palletizing scheme more in line with the efficiency and safety requirements of air freight. At the same time, the computational cost is controlled by setting the number of iterations N, ensuring the efficiency and practicality of the solution generation.

[0079] In one possible implementation, the i-th palletizing scheme includes a cargo sequence and a pallet-box sequence. Figure 8 A flowchart illustrating the loading planning provided in this application embodiment. Figure 2 ,like Figure 8 As shown, loading planning is performed on the i-th board-making scheme to obtain the optimized board-making scheme for the i-th time, including:

[0080] The first board in the board sequence is identified as the current board, and the pole set of the current board is initialized, which includes multiple poles.

[0081] For each item in the item sequence, perform the following steps:

[0082] If the goods cannot be placed at any pole in the pole set with any orientation in the orientation set, then the next pallet is selected in ascending order from the pallet sequence as the current pallet, and the pole set of the current pallet is initialized. The orientation set includes flat, vertical, side-vertical, horizontal, side-horizontal, and side-placed.

[0083] If the cargo can be placed at a pole in the pole set with one of the orientations in the orientation set, then update the current pole set of the pallet, calculate the remaining space score of each pole of the current pallet, and place the cargo at the pole with the highest remaining space score in the current pallet with one of the orientations in the orientation set.

[0084] The process continues until all goods in the cargo sequence have been placed, resulting in the optimized palletizing scheme for the i-th iteration.

[0085] Specifically, firstly, the first pallet in the pallet sequence is taken as the current pallet, and its set of poles is initialized. Here, the poles are the projection points of the vertices of the placed goods onto the plane, representing the candidate starting positions in the pallet where new goods can be placed. Each pole corresponds to a segment of remaining space (the available volume from that point to the boundary of the pallet), which is used to determine whether the goods can be placed and to evaluate the space utilization rate.

[0086] Next, iterate through each item in the item sequence, trying to find the item in one of the six possible orientations (e.g., ...). Figure 9 As shown, Figure 9 The orientation diagram of the goods provided in this application embodiment includes: flat, vertical, side-vertical, horizontal, side-horizontal, and side-placed) will be placed to any pole of the current pallet. If the goods cannot be placed into the remaining space of any pole with any orientation set, the process switches to the next pallet in the forward order of the pallet sequence and initializes its pole set, and continues to try to place the current goods.

[0087] If the cargo can be placed into a certain pole in a certain orientation of the orientation set, then update the current pole set of the pallet (add new poles generated by cargo vertices and remove occupied old poles), calculate the remaining space score of each pole, select the pole with the highest score to place the cargo in a certain orientation, and complete this loading.

[0088] Among them, the pole ep is calculated The scoring formula is as follows:

[0089] +

[0090] Where L, W, and D represent points ep respectively. Along The projection points of the positive coordinate axes onto the inner wall of the container , , Coordinate values Let be the three-dimensional coordinates of the pole ep. The coordinates of the bottom left rear vertex of the cargo. The coordinates are the upper right front vertex of the cargo.

[0091] The process continues until all goods are placed, resulting in the optimized palletizing solution for the i-th time. During this process, the discretized height matrix of the pallet and box is used to determine whether the goods are inside the container, and the coordinates of the diagonal vertices of the cuboid are used to accurately record the position of the goods, ensuring compliance with air cargo safety and space utilization regulations.

[0092] Possibly, Figure 10 This is a schematic diagram of the discretization of the board box provided in the embodiments of this application, as shown below. Figure 10 As shown, it includes:

[0093] The left side presents the true 3D outline of the pallet box, showcasing its complex shape, which is not a standard cuboid. The right side is a visualization of the pallet box discretized into an L×W two-dimensional height matrix, with dense point clouds marking the height values ​​at various coordinate positions on the inner wall. For a pallet box with an outer dimension of L×W×D, the height information is stored in a two-dimensional matrix H after discretization, where H[i,j] represents the height value of the inner wall along the vertical direction at coordinate (i,j). To determine whether the cargo is inside the container, it is only necessary to compare the cargo bottom height z with the cargo height h. c The sum of z+h c And the size of the corresponding position H[i,j], if z+h c If the value is less than or equal to H[i,j], then the cargo c is inside the container. This method provides a quantifiable data basis for spatial calculation of irregular platters.

[0094] Figure 11 A simplified schematic diagram of the goods provided in the embodiments of this application, as shown below. Figure 11 As shown, it includes:

[0095] Taking a standard cuboid as an example, the position and size of the cuboid in three-dimensional space can be completely determined using only two coordinate points: the lower left rear vertex (x1, y1, z1) and the upper right front vertex (x2, y2, z2). This can be mathematically represented as a two-dimensional array [(x1, y1, z1), (x2, y2, z2)]. This simplified representation replaces the traditional method of recording eight vertices, significantly reducing data storage and computational complexity. Simultaneously, it accurately matches the height field after discretization of the pallet / box, providing an efficient data structure for subsequent cargo placement and collision detection.

[0096] Figure 12 A schematic diagram of poles and remaining space provided for embodiments of this application, as shown below. Figure 12 As shown, it includes:

[0097] The left side uses red markers to show the projections of the vertices of the placed cargo c (represented as [(x1,y1,z1),(x2,y2,z2)]) onto the plane. These projection points are called "vertices," representing candidate starting positions within the pallet box where new cargo can be placed. The right side uses purple areas to visualize the remaining space corresponding to a specific pole ep(x0,y0,z0) in the pole set EP, i.e., the available volume (L-x0,W-y0,D-z0) from that pole to the boundary of the pallet box. It also clearly shows the necessary constraints for cargo to be placed at that pole.

[0098] x2-x1≤L-x0;

[0099] y2-y1≤W-y0;

[0100] z2-z1≤D-z0;

[0101] This diagram visually illustrates the correspondence between the poles and the remaining space, as described above. Figure 8 The placement decisions in the loading plan provide a visual basis.

[0102] By employing a heuristic packing logic based on a pole placement strategy and multi-orientation trials, goods are planned one by one into pallets and boxes. Priority is given to placing goods in poles with the highest remaining space scores. Six placement directions are supported, which maximizes the utilization of pallet and box space while ensuring the stability and compliance of goods stacking. Through the dynamic updating of pole sets and pallet and box switching mechanism, complex loading scenarios involving multiple goods and multiple pallets and boxes are handled efficiently. This provides a precise and feasible loading planning solution for iterative optimization, further improving the space utilization efficiency and operational standardization of the palletizing solution.

[0103] In one possible implementation, the score of the i-th scheme is determined based on the pallet box volume, cargo volume, cargo spatial coordinates, pallet box usage cost, first preset weight, second preset weight, and third preset weight in the i-th pallet optimization scheme.

[0104] Specifically, as shown in the following formula:

[0105]

[0106] in, Score the solution for the i-th iteration; Let p be the total volume of goods in the container. This refers to the volume of the tray / box. Let c be the spatial coordinates of the cargo; p Cost of using board box p; , , The preset weights are, in order, the first preset weight, the second preset weight, and the third preset weight.

[0107] The cost of using the panel box is predetermined.

[0108] Figure 2 A flowchart illustrating the palletizing method for air freight provided in this application embodiment. Figure 2 ,like Figure 2 As shown, it includes:

[0109] S201. When the intent identified from the first information is a task query, extract the third information from the first information. The third information includes the second task identifier.

[0110] Specifically, the system's built-in large language model performs semantic decomposition and contextual understanding on the first piece of information. After identifying the user's core intent as "task query", it accurately extracts the task identifier from the text through entity extraction technology, transforming the unstructured natural language query into a structured index that can be used for data retrieval, ensuring that the corresponding historical template solution can be accurately located.

[0111] S202. Based on the second task identifier, determine and output the corresponding second natural language summary, second three-dimensional visualization model, and second cargo loading list.

[0112] Specifically, based on the extracted second task identifier, the system calls the persistently stored database through the internal API to associate and retrieve the complete structured data package of the task, including the original input parameters, constraint rules, loading layout sequence, and other data. Then, based on this data, a concise natural language summary is generated by the large language model, and at the same time, the 3D graphics engine is called to reconstruct an interactive 3D loading model in real time according to the layout sequence. Finally, a loading list containing cargo size, weight, placement position, and pallet information is generated, completing the construction of multimodal results of historical solutions.

[0113] The system integrates the generated second task identifier, second natural language summary, second 3D visualization model, and second cargo loading list, and pushes them to users in a multimodal format with rich graphics and text. Users can view the task identifier, read the result summary, click links to view interactive 3D models, browse or download detailed loading lists, and complete an immersive review and assessment of historical plans.

[0114] For example, such as Figure 13 As shown in the figure, the system will present the task ID "249", data summary, data table containing package quantity and volume information, 3D map link, etc. Users can click on the 3D link to view the specific layout of the goods on the PAG board, and can also obtain detailed information of each item from the table, so as to achieve a complete review of the historical palletizing scheme.

[0115] The air freight palletizing method provided in this application obtains task query information input by the user in natural language, identifies the intent and extracts the task identifier, quickly associates and retrieves the corresponding historical palletizing results, generates and outputs natural language summaries, 3D visualization models and cargo loading lists, realizing closed-loop traceability of "task identifier → data → scenario", allowing users to efficiently review historical palletizing plans, avoiding the tediousness of manually searching documents and repeating calculations, greatly improving the reusability and management efficiency of palletizing plans, and further ensuring the accuracy and intuitiveness of plan review by reproducing the loading scenario through 3D visualization.

[0116] Figure 3 A flowchart illustrating the palletizing method for air freight provided in this application embodiment. Figure 3 ,like Figure 3 As shown, it includes:

[0117] S301. When the intent identified from the first information is a cargo plane inquiry, the third task identifier and the fourth information are extracted from the first information. The fourth information includes the target date, origin and destination.

[0118] Specifically, the system's built-in large language model performs semantic decomposition on the first piece of information, identifies the core intent of "cargo aircraft query," and then accurately extracts the third task identifier and the fourth piece of information through entity extraction technology, transforming the unstructured natural language query into structured parameters that can be used for flight retrieval and pallet matching.

[0119] For example, after the system recognizes that the user's input "@cargo plane query which cargo planes are available from point A to point B tonight, and needs to adapt to the palletizing scheme with task ID 100" is a cargo plane query intention, it extracts the third task identifier "100" and the fourth information: target date "XXXX-XX-XX", origin "point A", and destination "point B".

[0120] S302. Based on the fourth piece of information, determine the first flight information of multiple cargo planes.

[0121] Specifically, based on the extracted target date, origin, and destination, the system pulls real-time data of eligible all-cargo aircraft flights from the entire network through a deep interface with the aviation production system, and organizes it into a first flight information list containing core flight information, in preparation for subsequent pattern making scheme adaptation.

[0122] For example, based on the query conditions of "XXXX-XX-XX, location A, location B", the system retrieves three all-cargo flights from the aviation production system, including flight numbers A100 (aircraft type B200), A101 (aircraft type B201), and A102 (aircraft type B202), forming the first set of flight information.

[0123] S303. Based on the third task identifier, determine the corresponding third board-making scheme.

[0124] Specifically, based on the third task identifier, the system calls the database via internal API to retrieve the complete palletizing scheme data corresponding to the task, including key information such as pallet type, pallet size, total weight, and total volume, providing input for flight compatibility verification.

[0125] S304. Based on the third stencil scheme, the information of the first flight is filtered to obtain the information of the second flight and then output.

[0126] Specifically, the system uses a built-in aircraft loading knowledge graph and rule engine to automatically match the pallet and container size, weight, and type of the third palletizing scheme with the aircraft type, cargo hold configuration, weight limit, height limit, and available container type of each flight in the first flight information. It verifies whether the pallet and container can be successfully loaded into the cargo hold without exceeding the load limit, filters out flights that do not meet the compatibility requirements, and finally obtains the second flight information that meets the loading requirements.

[0127] The system integrates the second flight information and presents it to the user in tabular form, displaying key information such as flight number, aircraft type, and scheduled departure and arrival times. It also provides a link to download the full data, making it convenient for users to obtain detailed information and perform subsequent operations.

[0128] For example, Figure 14 This is a schematic diagram of cargo plane tracking provided in an embodiment of this application, such as... Figure 14 As shown, the process includes: After a user inputs "What all-cargo flights are available from Shenzhen to Beijing tonight?" in natural language, the system first identifies the "cargo flight query" intent. Then, based on the extracted target date, origin, and destination, it retrieves all-cargo flight data that meets the criteria from the aviation production system. Finally, it presents key information in tabular form, including resource type, origin / destination city and airport, flight status, flight number, aircraft type, and planned take-off and landing times. At the same time, it provides a link to download the full data, making it easy for users to obtain complete flight information and providing data support for subsequent compatibility verification between the template scheme and the flights.

[0129] The air cargo palletizing method provided in this application obtains flight query information input by the user in natural language, identifies the intent, extracts the task identifier and basic flight conditions, first retrieves cargo flights that meet the conditions, then performs compatibility verification between the aircraft type and pallet based on the associated palletizing scheme, and finally outputs recommended flights that can be safely loaded according to the palletizing scheme. This achieves a leap from passively querying flights to actively and intelligently recommending compliant loading flights, which not only solves the inefficiency problem of manually matching aircraft type and pallet constraints, but also avoids loading risks caused by improper adaptation, significantly improving the accuracy and operational efficiency of air cargo palletizing.

[0130] Figure 4A flowchart illustrating the palletizing method for air freight provided in this application embodiment. Figure 4 ,like Figure 4 As shown, it includes:

[0131] S401. When the intent identified from the first information is a query for knowledge related to board making, determine the graph query statement based on the first information.

[0132] Specifically, the system's built-in large language model performs semantic decomposition and contextual understanding on the first piece of information. After identifying the user's core intent as "querying knowledge related to board making," it uses natural language understanding technology to transform the colloquial or instructional query requirements into query statements supported by the knowledge graph. This clarifies the entity type (such as "aircraft type" and "load rules"), attributes (such as "maximum load" and "remaining load requirements"), and relationships (such as "aircraft type-load rules-requirements") to be retrieved, providing precise instructions for subsequent graph retrieval.

[0133] For example, Figure 15 This is a schematic diagram illustrating the query of knowledge related to plate making provided in an embodiment of this application, such as... Figure 15 As shown, the user entered "@Knowledge Base Aviation Loading Rules and Operation Standards", and this text is the first piece of information.

[0134] S402. Based on the graph query statement, query the corresponding entities, attributes and relationships from the preset graph, integrate the entities, attributes and relationships, obtain the fifth information and output it.

[0135] Specifically, based on the generated graph query statement, the system traverses the preset knowledge graph of the board-making field, accurately retrieves the corresponding entities (such as "B737-300F model" and "maximum load"), attributes (such as "remaining load reserved 300KG" and "maximum load calculation formula"), and relationships (such as "model - remaining load requirement - must reserve 300KG"). It integrates the scattered graph node data into complete knowledge fragments, and at the same time associates relevant rule clauses, remarks and other supplementary information to form comprehensive fifth information.

[0136] The system transforms the structured knowledge fragments in the fifth piece of information into clear and easy-to-understand natural language answers through a large language model. At the same time, it retains key formulas, rule entries, and notes, presenting them to users in a graphic and textual format to facilitate users' quick access to professional knowledge and assist in decision-making.

[0137] The air cargo palletizing method provided in this application obtains knowledge query information input by the user in natural language, identifies the intent, and converts it into a graph query statement. It accurately retrieves relevant entities, attributes, and relationships from a preset palletizing knowledge graph, integrates them into professional knowledge, and outputs them in natural language. This structured and semantically represented professional knowledge scattered in manuals and documents, achieving an upgrade from passive "document review" to proactive "intelligent question answering." It allows users to quickly obtain professional knowledge such as air loading rules and operating standards, providing efficient knowledge support for palletizing decisions and reducing the cost and error of manual review.

[0138] Figure 16 This is a schematic diagram of the structure of the palletizing device for air freight provided in the embodiments of this application, as shown below. Figure 16 As shown, the air cargo palletizing device 160 provided in this embodiment includes an acquisition module 1601, a processing module 1602, and an output module 1603.

[0139] The acquisition module 1601 is used to acquire first information, which is information input by the user using natural language.

[0140] The processing module 1602 is used to perform intent recognition on the first information. When the intent recognized from the first information is to measure the board box, the second information is extracted from the first information. The second information includes at least one of the following: target machine model, target board type, business priority, business attributes, and cargo information of each cargo. The cargo information includes at least one of the following: length, height, width, and weight.

[0141] The processing module 1602 is also used to determine the first board-making scheme based on the second information;

[0142] The output module 1603 is used to generate and output a first task identifier, a first natural language summary, a first three-dimensional visualization model, and a first cargo loading list based on the first plate-making scheme.

[0143] In one possible implementation, the processing module 1602 is further configured to:

[0144] When the intent identified from the first information is a task query, the third information is extracted from the first information, and the third information includes the second task identifier;

[0145] Based on the second task identifier, determine and output the corresponding second natural language summary, second three-dimensional visualization model, and second cargo loading list.

[0146] In one possible implementation, the processing module 1602 is further configured to:

[0147] When the intent identified from the first information is a cargo plane inquiry, the third task identifier and the fourth information are extracted from the first information. The fourth information includes the target date, origin and destination.

[0148] Based on the fourth piece of information, the first flight information of multiple cargo planes was determined;

[0149] Based on the third task identifier, the corresponding third plate-making scheme is determined;

[0150] Based on the third stencil scheme, the information of the first flight is filtered to obtain the information of the second flight and then output.

[0151] In one possible implementation, the processing module 1602 is further configured to:

[0152] When the intent identified from the first information is a query for knowledge related to board making, the query statement for the graph is determined based on the first information;

[0153] Based on the graph query statement, the corresponding entities, attributes and relationships are queried from the preset graph, and the entities, attributes and relationships are integrated to obtain the fifth information and output it.

[0154] In one possible implementation, the processing module 1602 is specifically used for:

[0155] Based on the second information, determine the board-making scheme for the i-th time, where i=0;

[0156] For the i-th board-making scheme, the following steps are performed:

[0157] The loading plan is performed for the i-th board-making scheme, and a score is obtained for the i-th scheme.

[0158] When i equals 0, the score of the i-th solution is determined as the historical score;

[0159] If the score of the solution in the i-th iteration is greater than the historical score, then the optimized solution in the i-th iteration is determined as the first solution.

[0160] Based on the i-th board-making optimization scheme and preset rules, generate board-making schemes for the neighborhood;

[0161] The board-making scheme of the neighborhood is determined as the board-making scheme of the (i+1)th time.

[0162] i is an integer, ranging from 0 to N, where N is a positive integer.

[0163] In one possible implementation, the processing module 1602 is specifically used for:

[0164] The first board in the board sequence is identified as the current board, and the pole set of the current board is initialized, which includes multiple poles.

[0165] For each item in the item sequence, perform the following steps:

[0166] If the goods cannot be placed at any pole in the pole set with any orientation in the orientation set, then the next pallet is selected in ascending order from the pallet sequence as the current pallet, and the pole set of the current pallet is initialized. The orientation set includes flat, vertical, side-vertical, horizontal, side-horizontal, and side-placed.

[0167] If the cargo can be placed at a pole in the pole set with one of the orientations in the orientation set, then update the current pole set of the pallet, calculate the remaining space score of each pole of the current pallet, and place the cargo at the pole with the highest remaining space score in the current pallet with one of the orientations in the orientation set.

[0168] The process continues until all goods in the cargo sequence have been placed, resulting in the optimized palletizing scheme for the i-th iteration.

[0169] In one possible implementation, the processing module 1602 is specifically used for:

[0170] The score of the i-th scheme is determined based on the pallet volume, cargo volume, cargo spatial coordinates, pallet usage cost, first preset weight, second preset weight, and third preset weight in the i-th pallet optimization scheme.

[0171] The air cargo palletizing device provided in this embodiment can execute the method provided in the above method embodiment. Its implementation principle and technical effect are similar, and will not be described in detail here.

[0172] Figure 17 This is a schematic diagram of the structure of an electronic device provided in an embodiment of this application. Figure 17 As shown, the electronic device 170 provided in this embodiment includes at least one processor 1701 and a memory 1702. Optionally, the device 170 further includes a communication component 1703. The processor 1701, memory 1702, and communication component 1703 are connected via a bus.

[0173] In a specific implementation, at least one processor 1701 executes computer execution instructions stored in memory 1702, causing at least one processor 1701 to perform the above-described method.

[0174] The specific implementation process of processor 1701 can be found in the above method embodiments, and its implementation principle and technical effect are similar. It will not be repeated here.

[0175] In the above embodiments, it should be understood that the processor can be a Central Processing Unit (CPU), or other general-purpose processors, digital signal processors (DSPs), application-specific integrated circuits (ASICs), etc. The general-purpose processor can be a microprocessor or any conventional processor. The steps of the method disclosed in this invention can be directly implemented by a hardware processor, or implemented by a combination of hardware and software modules within the processor.

[0176] The memory may include high-speed memory (Random Access Memory, RAM) and may also include non-volatile memory (NVM), such as at least one disk storage device.

[0177] The bus can be an Industry Standard Architecture (ISA) bus, a Peripheral Component Interconnect (PCI) bus, or an Extended Industry Standard Architecture (EISA) bus, etc. Buses can be categorized as address buses, data buses, control buses, etc. For ease of illustration, the buses shown in the accompanying drawings are not limited to a single bus or a single type of bus.

[0178] This application also provides a computer program product, including a computer program that, when executed by a processor, implements the above-described method.

[0179] This application also provides a computer-readable storage medium storing computer-executable instructions, which, when executed by a processor, implement the above-described method.

[0180] The aforementioned readable storage medium can be implemented by any type of volatile or non-volatile storage device or a combination thereof, such as static random access memory (SRAM), electrically erasable programmable read-only memory (EEPROM), erasable programmable read-only memory (EPROM), programmable read-only memory (PROM), read-only memory (ROM), magnetic storage, flash memory, magnetic disk, or optical disk. The readable storage medium can be any available medium accessible to a general-purpose or special-purpose computer.

[0181] An exemplary readable storage medium is coupled to a processor, enabling the processor to read information from and write information to the readable storage medium. Of course, the readable storage medium can also be a component of the processor. The processor and the readable storage medium can reside in application-specific integrated circuits (ASICs). Alternatively, the processor and the readable storage medium can exist as discrete components in the device.

[0182] The division of units is merely a logical functional division; in actual implementation, there may be other division methods. For example, multiple units or components may be combined or integrated into another system, or some features may be ignored or not executed. Furthermore, the coupling or direct coupling or communication connection shown or discussed may be indirect coupling or communication connection through some interfaces, devices, or units, and may be electrical, mechanical, or other forms.

[0183] The units described as separate components may or may not be physically separate. The components shown as units may or may not be physical units; that is, they may be located in one place or distributed across multiple network units. Some or all of the units can be selected to achieve the purpose of this embodiment according to actual needs.

[0184] In addition, the functional units in the various embodiments of the present invention can be integrated into one processing unit, or each unit can exist physically separately, or two or more units can be integrated into one unit.

[0185] If a function is implemented as a software functional unit and sold or used as an independent product, it can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of this invention, or the part that contributes to the prior art, or a part of the technical solution, can be embodied in the form of a software product. This computer software product is stored in a storage medium and includes several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) to execute all or part of the steps of the methods of the various embodiments of this invention. The aforementioned storage medium includes various media capable of storing program code, such as USB flash drives, portable hard drives, read-only memory (ROM), random access memory (RAM), magnetic disks, or optical disks.

[0186] Those skilled in the art will understand that all or part of the steps of the above-described method embodiments can be implemented by hardware related to program instructions. The aforementioned program can be stored in a computer-readable storage medium. When executed, the program performs the steps of the above-described method embodiments; and the aforementioned storage medium includes various media capable of storing program code, such as ROM, RAM, magnetic disks, or optical disks.

[0187] Finally, it should be noted that other embodiments of the invention will readily occur to those skilled in the art upon consideration of the specification and practice of the invention disclosed herein. This invention is intended to cover any variations, uses, or adaptations of the invention that follow the general principles of the invention and include common knowledge or customary techniques in the art not disclosed herein, and is not limited to the precise structures described above and shown in the accompanying drawings, and various modifications and changes can be made without departing from its scope. The scope of the invention is limited only by the appended claims.

Claims

1. A method for palletizing air freight, characterized in that, include: Obtain first information, which is information input by the user using natural language; The first information is used to identify intent. When the intent identified from the first information is to measure the board box, the second information is extracted from the first information. The second information includes at least one of the following: target machine model, target board type, business priority, business attributes, and cargo information of each cargo. The cargo information includes at least one of the following: length, height, width, and weight. Based on the second information, the first plate-making scheme is determined; Based on the first plate-making scheme, a first task identifier, a first natural language summary, a first three-dimensional visualization model, and a first cargo loading list are generated and output.

2. The method according to claim 1, characterized in that, Also includes: When the intent identified from the first information is a task query, third information is extracted from the first information, and the third information includes a second task identifier; Based on the second task identifier, determine and output the corresponding second natural language summary, second three-dimensional visualization model, and second cargo loading list.

3. The method according to claim 1, characterized in that, Also includes: When the intent identified from the first information is a cargo plane inquiry, a third task identifier and a fourth piece of information are extracted from the first information. The fourth piece of information includes the target date, origin, and destination. Based on the aforementioned fourth information, the first flight information of multiple cargo aircraft is determined; Based on the third task identifier, the corresponding third plate-making scheme is determined; Based on the third stencil scheme, the first flight information is filtered to obtain the second flight information and output it.

4. The method according to claim 1, characterized in that, Also includes: When the intent identified from the first information is a query for knowledge related to board making, a graph query statement is determined based on the first information; Based on the graph query statement, the corresponding entities, attributes and relationships are queried from the preset graph, and the entities, attributes and relationships are integrated to obtain the fifth information and output it.

5. The method according to claim 1, characterized in that, The step of determining the corresponding first plate-making scheme based on the second information includes: Based on the second information, determine the board-making scheme for the i-th time, where i=0; For the i-th board-making scheme, the following steps are performed: The loading plan is performed for the i-th board-making scheme, and a score is obtained for the i-th scheme. When i equals 0, the score of the i-th solution is determined as the historical score; If the score of the i-th solution is greater than the historical score, then the i-th optimized solution is determined as the first solution. Based on the i-th board-making optimization scheme and preset rules, generate board-making schemes for the neighborhood; The board-making scheme of the neighborhood is determined as the board-making scheme for the (i+1)th time. The value of i is an integer, ranging from 0 to N, where N is a positive integer.

6. The method according to claim 5, characterized in that, The i-th palletizing scheme includes a cargo sequence and a pallet box sequence. The loading planning for the i-th palletizing scheme to obtain the i-th optimized palletizing scheme includes: The first board in the board sequence is determined as the current board, and the pole set of the current board is initialized, the pole set including multiple poles; For each item in the sequence of goods, perform the following steps: If the goods cannot be placed at any pole in the pole set with any orientation in the orientation set, then the next pallet is selected in ascending order from the pallet sequence as the current pallet, and the pole set of the current pallet is initialized. The orientation set includes flat, vertical, side-vertical, horizontal, side-horizontal, and side-placed. If the cargo can be placed in the pole of the pole set with one of the orientations in the orientation set, then update the pole set of the current pallet, calculate the remaining space score of each pole of the current pallet, and place the cargo in the pole of the current pallet with the highest remaining space score with one of the orientations in the orientation set. The process continues until all goods in the cargo sequence have been placed, resulting in the i-th optimized palletizing scheme.

7. The method according to claim 5, characterized in that, The process of scoring the optimization scheme for the i-th iteration to obtain the scheme score for the i-th iteration includes: The score of the i-th scheme is determined based on the pallet volume, cargo volume, cargo spatial coordinates, pallet usage cost, first preset weight, second preset weight, and third preset weight in the i-th pallet optimization scheme.

8. A palletizing device for air freight, characterized in that, include: The acquisition module is used to acquire first information, which is information input by the user using natural language. The processing module is used to perform intent recognition on the first information. When the intent recognized from the first information is to test the board box, the second information is extracted from the first information. The second information includes at least one of the following: target model, target board type, business priority, business attributes, and cargo information of each cargo. The cargo information includes at least one of the following: length, height, width, and weight. The processing module is further configured to determine a first board-making scheme based on the second information; The output module is used to generate and output a first task identifier, a first natural language summary, a first three-dimensional visualization model, and a first cargo loading list based on the first plate-making scheme.

9. An electronic device, characterized in that, include: Memory, processor; The memory stores computer-executed instructions; The processor executes computer execution instructions stored in the memory, causing the processor to perform the method as described in any one of claims 1-7.

10. A computer-readable storage medium, characterized in that, The computer-readable storage medium stores computer-executable instructions, which, when executed by a processor, are used to implement the method as described in any one of claims 1-7.

11. A computer program product, characterized in that, Includes a computer program that, when executed by a processor, implements the method described in any one of claims 1-7.

Images