Ship pipeline design drawing stage management method and system

Abstract

The invention relates to the technical field of ship manufacturing intelligent management, in particular to a ship pipeline design drawing stage management method and a ship pipeline design drawing stage management system. Production schedule and process specification data are acquired and processed in a structured mode, a standardized input set is constructed, conflict resolution and progress and cost optimization are carried out by utilizing a multi-objective optimization engine; a four-dimensional model, semantic verification and dynamic scheduling are combined to generate a preliminary pipeline scheme, construction drawings and animations are output after three-dimensional simulation verification, feedback is collected in real time in construction, secondary scheduling and optimization are triggered, continuous iteration and accurate execution of design are achieved, and through intelligent decision making of a multi-target optimization engine, the design drawing efficiency can be improved by 40% or above. Meanwhile, the design error rate is reduced by 90%, and later reworking is greatly reduced.

Description

Technical Field

[0001] The present invention relates to the technical field of intelligent management of shipbuilding, and specifically to a method and system for managing the design and drawing stage of ship piping, which is particularly suitable for the refined design, intelligent decision-making and collaborative management of piping systems during the construction of large ships. Background Art

[0002] The ship's piping system is the vessel's blood vessel network, and its design quality directly affects the ship's construction efficiency and operational safety. Traditional ship piping design and drawing management mainly relies on manual coordination by designers' experience, which has the following technical problems: First, there is a serious disconnect between design and construction. Design departments often lack a full understanding of the actual construction site when developing drawings and plans. This leads to frequent spatial conflicts and process inconsistencies in the design plans during actual construction, resulting in a high rework rate.

[0003] Secondly, multi-disciplinary collaboration is inefficient. Ship piping involves multiple disciplines, including structure, machinery, and electrical engineering. There is a lack of effective information sharing and decision-making coordination mechanisms among these disciplines, leading to frequent duplication of designs and resource conflicts.

[0004] Thirdly, change management lags in response. When problems are discovered on the construction site and design changes are required, traditional management methods are slow to respond, unable to quickly assess the impact of the changes and adjust subsequent plans, seriously affecting project progress.

[0005] Finally, material management was crude and inefficient. The pallet division and distribution of piping materials lacked scientific planning, leading to frequent material backlogs or shortages, increasing storage costs and construction wait times.

[0006] With the development trend of ships becoming larger and more complex, the limitations of traditional management methods are becoming increasingly apparent. There is an urgent need to introduce advanced technologies such as artificial intelligence to realize intelligent management in the pipeline design and drawing stage. Summary of the Invention

[0007] The purpose of the present invention is to overcome the shortcomings of the existing technology and provide a method and system for managing the design and drawing stage of ship piping. By building an intelligent decision-making system based on artificial intelligence, deep collaboration and dynamic optimization of multiple links such as design, production, and materials can be achieved.

[0008] This invention utilizes four-dimensional spatiotemporal collision detection technology, which not only considers static spatial interference but also incorporates the temporal dimension into collision prediction, enabling early identification of dynamic conflicts during construction. Through semantically informed process specification detection, abstract textual specifications are automatically converted into computable constraint parameters, significantly improving the accuracy and efficiency of specification checks. Cognitive digital twin technology is introduced to achieve in-depth perception and understanding of construction site conditions, supporting intelligent scheduling decisions based on causal reasoning.

[0009] The present invention discloses a method for managing the drawing production stage of ship piping design, comprising: Obtain production schedule data and specification parameters from the process specification database, structure the acquired data, and establish a data timestamp mechanism to form a standardized input data set; Based on the generated input data set, a multi-objective optimization engine performs a comprehensive trade-off analysis to minimize construction conflicts, schedule deviations, and material costs. Simultaneously, it performs collision detection based on a four-dimensional spatiotemporal model, process specification compliance verification based on semantic understanding, dynamic scheduling optimization considering real-time deviations, and pallet allocation planning based on on-site constraints. This generates a preliminary piping design plan that includes construction phase division, drawing time schedule, and bill of materials. Based on the generated preliminary piping design, conduct 3D simulation verification and feasibility assessment to identify potential design conflicts and construction risks, and form a verified optimized solution. Based on the verified optimization plan, generate 2D drawings with construction stage markings and construction process animation videos, and output executable construction guidance documents; During the construction execution process, on-site feedback information and design change requirements are collected. When the impact of the change exceeds the preset threshold, secondary scheduling is automatically triggered, the drawing plan is updated, and the above analysis and optimization process is re-executed to achieve iterative optimization of the design solution.

[0010] Preferably, the multi-objective optimization engine performs comprehensive trade-off analysis including: The construction space is divided into three-dimensional grid units and the occupancy time window of each grid unit is marked. By identifying the situation where multiple pipelines occupy the same space at the same time, the spatial conflict severity index is calculated. At the same time, the equipment sharing conflicts and process parallel feasibility between different pipeline construction processes are analyzed to quantify the impact of process conflicts on the overall progress. Construct a construction network diagram and calculate the earliest start time and latest completion time for each activity. Identify the key activity sequence with zero total float time to form the critical path. Analyze the construction uncertainty of various pipelines based on historical data and set time buffers for high-risk activities. Analyze the demand distribution of pipes of the same specifications and calculate the unit cost under different batches. Comprehensively evaluate the capital cost of advance procurement and on-site storage space limitations to determine the optimal material procurement batch and timing. The weight coefficients of the three optimization objectives of construction conflict, schedule deviation and material cost are dynamically adjusted according to the project stage. Through iterative optimization, multiple candidate plans are generated and the comprehensive score of each plan is calculated. The plan with the best overall performance is selected as the output result.

[0011] Preferably, the collision detection based on the four-dimensional space-time model specifically includes: Construct a four-dimensional space-time occupancy tensor containing three-dimensional spatial coordinates and time dimensions, where the first three dimensions represent the position coordinates of the pipeline in space, and the fourth dimension represents the construction time series; For each pipeline section, calculate its temporal and spatial occupancy trajectory throughout the construction period, including the hoisting path, the space occupied during the installation process, and the space required for welding operations; Tensor operations are used to identify overlapping areas of different pipelines in four-dimensional space-time. A probability distribution model is introduced to account for the uncertainty of the construction process. Monte Carlo simulation is used to generate multiple possible construction scenarios and calculate the probability distribution of collisions. When the collision probability exceeds a preset threshold, a collision resolution strategy is generated based on deep reinforcement learning, including a spatial avoidance plan that adjusts the pipeline path and a temporal avoidance plan that adjusts the construction sequence. The impact of different resolution strategies on the overall construction progress and cost is evaluated, and the optimal solution is selected.

[0012] Preferably, the process specification compliance verification based on semantic understanding specifically includes: A pre-trained language model is used to perform in-depth semantic analysis on the process specification text to extract specific control elements from the process specification text, including: the pipeline components to be constrained (constraint objects), the technical requirements to be followed (constraint conditions), the specific numerical range (constraint values), and the ship areas or pipeline types to which the specifications apply (scope of application). This generates a high-dimensional semantic vector representation of the specifications. Construct a specification semantic space and calculate the semantic similarity between different specifications, automatically identify the association, conflict and complementary relationship between specifications, and establish a mapping relationship between specification semantics and design parameters; Automatically convert abstract regulatory requirements into specific ship piping design parameters, including quantifiable process indicators such as the minimum bend radius required at pipe bends, the maximum allowable flow velocity variation percentage when the fluid flows through the pipe, the minimum and maximum spacing between adjacent pipe supports, and the groove angle requirements required at welds. Construct a specification dependency graph and identify potential specification conflict patterns through graph neural networks. Specs are divided into hard constraints and soft constraints. A multi-objective optimization algorithm is used to find a design solution that maximizes overall constraint satisfaction. For conflicts that cannot be completely resolved, detailed conflict reports and resolution suggestions are generated.

[0013] Preferably, the dynamic scheduling optimization considering real-time deviation specifically includes: Obtain real-time information on the actual drawing completion time of each pipeline, the time and content of design changes, and on-site feedback on problems and delays. Calculate the time deviation between the actual and planned progress and analyze the development trend of the deviation. Build a cognitive digital twin system that integrates multiple sources of on-site data, including environmental and equipment parameter data collected by IoT sensors installed at the construction site, construction footage captured by the on-site video surveillance system, quality and progress feedback submitted by construction workers via mobile devices, and formal progress report data from the project management system. This system then fuses heterogeneous data using a deep learning model to form a unified cognitive state representation. An attention mechanism is introduced to identify key state changes, and a causal reasoning network is used to analyze the causal relationship in the construction process and predict the causal effects of different scheduling decisions. Based on the current schedule deviation, resource constraints and downstream process dependencies, a priority scoring model is established and the drawing priority of each piping system is dynamically adjusted, and design resources are reallocated to ensure that tasks on the critical path are completed first.

[0014] Preferably, the pallet partitioning planning based on site constraints specifically includes: Analyze the spatial layout constraints of the construction site, including the dimensions of available storage space, width and height restrictions of transport aisles, load ratings of lifting equipment, and loading capacities of transport vehicles; Build a digital twin model of material flow, simulate the entire process of material flow from the warehouse to the construction site through discrete event simulation technology, and identify potential bottlenecks and conflicts under different pallet partitioning schemes; Pipelines with the same specifications and in similar construction areas are grouped according to similarity, ensuring that the weight of each pallet does not exceed the equipment load limit and the size meets transportation constraints. A three-dimensional packing algorithm is used to optimize pallet space utilization. Monte Carlo simulations are used to evaluate the resilience of pallet solutions under abnormal circumstances, including transport equipment failures and temporary unavailability of storage areas. An efficient and resilient pallet partitioning scheme is selected, and a unique identification code is generated for each pallet, containing area, batch, and content information.

[0015] Preferably, the collecting of on-site feedback information and design change requirements and automatically triggering secondary scheduling specifically includes: Identify and categorize sources of changes, including proactive changes due to design optimization, reactive changes due to field feedback, and mandatory changes due to specification updates. Based on the scope of impact, these changes are categorized as Level 1 changes affecting multiple systems, Level 2 changes within a single system, and Level 3 changes that do not affect interfaces. Analyze the direct impact of the change, including changes to upstream and downstream connections of the pipeline, the need for simultaneous revision of related drawings, and the impact on completed construction. Simultaneously assess the indirect impact, including the chain reaction changes in space occupancy, the impact on other disciplines, and the impact on the overall project schedule. Setting a change impact threshold, and automatically triggering a rescheduling process when the cumulative impact exceeds the threshold, updating the status of the affected tasks and re-running the multi-objective optimization engine to generate a revised output plan; Generate a new version number for each change and save complete data before and after the change, support version comparison and rollback, determine the corresponding approval process based on the change level, and record approval opinions and decision-making basis to ensure the traceability of changes.

[0016] Preferably, generating a two-dimensional drawing with construction stage identification and an animation video of the construction process specifically includes: Extract key information for on-site construction guidance from the 3D design model of the ship's pipelines, including the specific location coordinates of the pipelines to be fabricated in sections, the type and parameters of the welds used at each connection (such as butt welds and fillet welds), the precise installation location coordinates and types of the pipeline supports, and other key technical information necessary for construction personnel. Based on the optimization results, a construction sequence number is generated and the dependencies between the processes are marked. Automatically add segment numbers, positioning dimensions, welding symbols, quality requirements, material specifications, and quantity information to 2D drawings. Avoid overlapping and occlusion of annotations through layout optimization algorithms, and use layers to manage different types of annotation information. Build a 3D scene including a ship structure model, pipeline model, and construction equipment model. Set the animation rhythm and determine the key frame positions according to the actual construction time to simulate the complete process of pipeline lifting, welding operations, and inspection and testing. Use color changes in animations to indicate construction status, add arrows to indicate installation direction, set translucent effects to display internal structures, overlay current step instructions, progress timelines, and safety warning signs, and generate multi-format output files that support pause control, chapter jumps, and on-site annotation functions.

[0017] As an option, it also includes: Adopting a distributed collaboration framework based on blockchain, with the design, production, and materials departments as blockchain network nodes, verifying and synchronizing data through a consensus mechanism, and using smart contracts to automatically trigger cross-departmental collaboration processes; Build a distributed intelligent optimization mechanism based on federated learning to achieve multi-party collaborative training without sharing original data, and protect data privacy through differential privacy and secure multi-party computing technologies; Establish a spiral mechanism among data, knowledge, and intelligence. Learn design patterns from historical project data to form a domain knowledge base. This knowledge guides intelligent decision-making in current projects. Decision-making results generate new data, which is then updated through continuous learning. A hierarchical optimization architecture is set up to balance local optimization and global coordination. The bottom-level modules perform real-time local optimization, the middle-level modules perform intra-domain coordination, and the top-level modules perform periodic global strategic optimization. The optimized resource allocation of each module is controlled through the optimized budget mechanism.

[0018] A system for managing the drawing production stage of ship piping design, comprising: The data acquisition and preprocessing module is used to obtain production schedule data and specification parameters from the process specification database, structure the acquired data, and establish a data timestamp mechanism to form a standardized input data set; An intelligent analysis and decision-making module, including a multi-objective optimization engine, a collision detection unit, a process specification detection unit, a dynamic scheduling unit, and a pallet division unit, is used to perform a comprehensive trade-off analysis based on the input data set to minimize construction conflicts, minimize schedule deviations, and optimize material costs. It performs collision detection based on a four-dimensional spatiotemporal model, process specification compliance verification based on semantic understanding, dynamic scheduling optimization considering real-time deviations, and pallet division planning based on on-site constraints, generating a preliminary piping design plan that includes construction phase division, drawing time plan, and material list; A scheme generation and verification module is used to perform three-dimensional simulation verification and feasibility assessment based on the preliminary piping design scheme, identify potential design conflicts and construction risks, and form a verified optimized scheme; A construction guidance output module is used to generate two-dimensional drawings with construction stage identification and construction process animation videos based on the verified optimization plan, and output executable construction guidance documents; The execution feedback and iterative optimization module is used to collect on-site feedback information and design change requirements during the construction execution process. When the impact of the change exceeds the preset threshold, it automatically triggers secondary scheduling, updates the drawing plan, and calls the intelligent analysis and decision-making module to re-execute the comprehensive trade-off analysis of minimizing construction conflicts, minimizing schedule deviations, and optimizing material costs, collision detection based on a four-dimensional space-time model, process specification compliance verification based on semantic understanding, dynamic scheduling optimization considering real-time deviations, and analysis and optimization process of pallet division planning based on on-site constraints, so as to achieve iterative optimization of the design scheme.

[0019] Compared with the prior art, the present invention has the following beneficial effects: First, significantly improve design efficiency. Through the intelligent decision-making of the multi-objective optimization engine, design drawing efficiency can be increased by more than 40%, while the design error rate can be reduced by 90%, significantly reducing later rework.

[0020] Second, it enables deep collaborative management. The distributed collaborative framework based on blockchain ensures data consistency and decision-making synchronization across departments, eliminates information silos, and improves overall operational efficiency by 15% to 20%.

[0021] Third, it enhances dynamic response capabilities. The system can detect progress deviations in real time and automatically trigger scheduling optimization, shortening change response time from traditional days to hours, ensuring project completion on schedule.

[0022] Fourth, optimize resource allocation. Through intelligent pallet division and material flow simulation, material utilization rate increased by 20%, storage costs decreased by 15%, and construction waiting time was effectively reduced.

[0023] Fifth, it improves the scientific nature of decision-making. A decision-making mechanism based on knowledge graphs and causal reasoning makes management decisions more scientific and reasonable, shortening the overall project duration by 20% to 30%. BRIEF DESCRIPTION OF THE DRAWINGS

[0024] Figure 1 This is a diagram of the overall architecture of the ship piping design and drawing stage management system provided by an embodiment of the present invention.

[0025] Figure 2 This is a workflow diagram of the multi-objective optimization engine in an embodiment of the present invention.

[0026] Figure 3 Schematic diagram of the principle of four-dimensional space-time collision detection in an embodiment of the present invention.

[0027] Figure 4 This is a process flow chart of change management in an embodiment of the present invention.

[0028] Figure 5 2 is a structural diagram of a distributed collaborative framework in an embodiment of the present invention. DETAILED DESCRIPTION

[0029] In order to make the objectives, technical solutions and advantages of the present invention more clear, specific embodiments of the present invention will be further described in detail below with reference to the accompanying drawings.

[0030] See also Figure 1 The present invention provides a method for managing the design and drawing stage of ship piping. The method realizes the optimization of the entire process from data collection to construction execution by building an intelligent management system.

[0031] First, the system acquires production schedule data and specification parameters from a process specification database. In one embodiment of the present invention, production schedule data is acquired through a real-time interface with the shipyard's ERP system, including information such as the current progress percentage of each construction area, staffing status for each type of work, and equipment resource utilization. For example, the system automatically synchronizes data hourly, capturing specific information such as the pipeline installation progress in Area A is 75%, there are 8 welders, and the crane utilization rate is 60%. Process specification data is extracted from a pre-built structured database, containing technical standards, material requirements, installation specifications, and more for various pipeline types.

[0032] Preferably, during the data preprocessing stage, the system cleans and standardizes the acquired raw data. Specifically, the system establishes a unified data model to map data from different sources into a standard format. For example, abstract specification requirements are automatically converted into specific ship pipeline design parameters, including: the minimum bending radius value required to be met at the pipeline bend (such as the minimum bending radius ≥ 3D, where D is the pipe diameter), the maximum flow rate change percentage allowed when the fluid flows in the pipeline, the minimum and maximum spacing ranges between adjacent pipeline supports, the groove angle numerical requirements required at the weld, and other quantifiable process indicators. These quantitative indicators directly affect design decisions. For example, when designing a pipeline bend, the system will automatically ensure that the bending radius is greater than or equal to the minimum allowable value. At the same time, the system establishes a timestamp mechanism for each data item to record the acquisition time and validity period of the data to ensure that decisions are based on the latest information.

[0033] Based on the standardized input data set, the system performs a comprehensive analysis using a multi-objective optimization engine. This is one of the core innovations of this invention, as the engine simultaneously considers the three objectives of minimizing construction conflicts, minimizing schedule deviations, and optimizing material costs. To minimize construction conflicts, the system divides the construction space into three-dimensional grid cells, each of which is set to 0.5m×0.5m×0.5m. This accuracy accurately reflects pipeline occupancy without incurring excessive computational burden. The system marks the occupancy time window for each grid, for example, grid (10,20,15) is occupied by the DN100 pipeline from days 5 to 7.

[0034] In one embodiment of the present invention, the spatial conflict severity index is calculated using the following formula: , in, is the spatial conflict severity index; is the total number of pipelines; For pipeline and pipelines The weight coefficient between them ranges from [0,1] and is determined according to the importance of the pipeline; is the spatial overlap, which indicates the spatial overlap volume ratio of the two pipelines; The time overlap ratio represents the overlap ratio of the construction time of the two pipelines. Minimizing schedule deviation is based on the critical path method (CPM). The system first constructs a complete construction network diagram and calculates the earliest start time (ES), latest start time (LS), earliest finish time (EF), and latest finish time (LF) for each activity. The total float time TF is calculated as follows: , When TF = 0, the activity is on the critical path. The system also incorporates uncertainty analysis based on historical data to set time buffers for high-risk activities. For example, based on historical data analysis, the standard deviation of completion time for welding operations on large-diameter pipelines (DN ≥ 200) is large. Therefore, the system automatically adds a 15% time buffer for these activities.

[0035] Material cost optimization is achieved through an improved version of the economic batch size model (EOQ). Taking into account the particularity of shipbuilding, the cost function adopted by the system is: , in, is the total cost; The purchase cost includes the unit price of materials and volume discounts; The storage cost is calculated on a daily basis, taking into account capital occupation and site costs; =Stock-out costs, i.e., the construction delay losses caused by insufficient materials. In one embodiment of the present invention, the system dynamically adjusts the weights of the three objectives based on the project stage. In the early stages of a project (0% to 30% progress), the progress weight is set to 0.5, the conflict weight to 0.3, and the cost weight to 0.2, with the focus on ensuring a smooth project launch. In the mid-stage of a project (30% to 70% progress), the three weights are balanced at 0.33. In the late stages of a project (70% to 100% progress), the conflict weight is increased to 0.5 to ensure construction quality during the final stages.

[0036] The collision detection of the present invention adopts an innovative four-dimensional space-time model. Figure 3 , the model constructs the space-time occupancy tensor (STOT), which is a fourth-order tensor ,in 、 、 is the spatial dimension, is the time dimension. In the specific implementation, each element of the tensor Indicates spatial location At the moment occupancy probability. Preferably, the system does not use a binary representation of occupancy, but instead uses a continuous probability value in the range [0, 1]. This design takes into account the uncertainty of the construction process. For example, pipeline hoisting may have time deviations due to factors such as weather and equipment.

[0037] For each pipeline segment, the system calculates its complete temporal and spatial occupancy trajectory. For example, for a DN150, 6-meter-long straight pipe, the temporal and spatial occupancy calculation for the lifting process includes the spatial occupancy at the lifting point (taking into account the size of the lifting equipment), the swept space along the lifting path (accounting for pipeline sway), and the final occupancy at the installation location (accounting for welding space). The system uses Monte Carlo simulation to generate 1,000 possible construction scenarios, each taking into account different random factors, such as lifting speed deviation (±10%) and positioning error (±50mm).

[0038] The collision probability is calculated using the following formula: , in, For pipeline and pipelines The collision probability of and are the space-time occupancy tensors of the two pipelines respectively; the integration range For the entire construction cycle. When the collision probability exceeds a preset threshold (set to 0.15 in this embodiment, this value is determined based on the shipyard's safety management requirements), the system triggers the collision resolution mechanism. The resolution strategy is generated using a deep reinforcement learning method, specifically the Deep Deterministic Policy Gradient (DDPG) algorithm. State space Contains 26 features such as current pipeline location, construction progress, available resources, etc.; action space It includes eight action dimensions, including path adjustment (continuous value, range ±2 meters), delay time (discrete value, 0-7 days), etc. The reward function is designed as: , in, For immediate rewards; Number of days of progress delay; is the additional cost (10,000 yuan); is the number of other pipelines affected; 、 、 are weight coefficients, and these parameters are determined through regression analysis of historical project data.

[0039] The process specification detection module of this invention uses deep semantic understanding technology, which is a key innovation that distinguishes it from traditional rule-matching methods. The system first uses a pre-trained BERT model (fine-tuned on ship engineering corpus) to encode the specification text.

[0040] A pre-trained language model is used to perform deep semantic analysis on the process specification text, extracting specific control elements from the text. These elements include the constrained piping components (constraint objects), the technical requirements to be followed (constraint conditions), the specific numerical range (constraint values), and the vessel areas or pipeline types to which the specification applies (scope of application). This generates a high-dimensional semantic vector representation of the specification. For example, the system uses an attention mechanism to identify the key semantic elements in the specification text, "A reinforcement structure shall be installed at pipeline penetrations, and the thickness of the reinforcement plate shall not be less than 0.7 times the bulkhead thickness." The constraint object is "pipeline penetrations," the constraint type is "structural requirements," and the constraint parameter is "reinforcement plate thickness ≥ 0.7 × bulkhead thickness."

[0041] The specification semantic space is constructed using knowledge graph technology. Nodes in the graph represent specification clauses, and edges represent relationships between specifications. Relationship types include dependencies (e.g., welding specifications depend on material specifications), mutual exclusions (e.g., specifications for different pressure levels are mutually exclusive), and inheritance (e.g., special specifications inherit general specifications). The system uses a graph neural network (GNN) for specification conflict detection. The network employs three layers of graph convolutional layers, with the message passing formula for each layer being: , in, Represents a node in the graph In the Layer ( Iterations) feature representation vector in graph neural network, For nodes In the The feature representation of the layer, is a node Neighbor ; For the The learnable weight matrix of the layer; For nodes AGG is the aggregation function, and this embodiment adopts mean aggregation; As the activation function, ReLU is used. Represents a collection node All neighbors of The characteristics of the layer are called messages or neighbor information.

[0042] When a specification conflict is detected, the system models it as a constraint satisfaction problem (CSP). Preferably, the system distinguishes between hard and soft constraints. Hard constraints must be satisfied, such as safety-related specifications; soft constraints must be satisfied as much as possible, such as optimization suggestion specifications. The constraint satisfaction calculation formula is: , in, For overall satisfaction; is a set of hard constraints; is a set of soft constraints; is the indicator function, hard constraint 1 if satisfied, 0 otherwise; Soft constraint The satisfaction function of , the value range is [0,1].

[0043] Dynamic scheduling optimization is another core innovation of this invention. The system builds a cognitive digital twin system to achieve in-depth perception and understanding of the construction site.

[0044] At the data collection level, a cognitive digital twin system is built to integrate field data from multiple sources, including: environmental data (temperature, humidity, wind speed, etc.) and equipment status (crane load, welding machine current, etc.) collected every 5 seconds by IoT sensors installed at the construction site, 25 frames per second of construction image data provided by the on-site video surveillance system (using computer vision algorithms to identify personnel locations, material stacking, work progress, etc.), quality and progress feedback information submitted by construction workers through mobile terminals (such as good welding quality, slight deformation of materials, etc.), and formal progress report data in the project management system. Various different types of data are integrated through deep learning models to form a unified cognitive state representation.

[0045] Multimodal data fusion utilizes a Transformer-based architecture. Data from different modalities first passes through their respective encoders: Numerical sensor data is encoded into 128-dimensional vectors using a fully connected network; video frames are processed through ResNet to extract 2048-dimensional features, which are then mapped to 128 dimensions via a linear transformation; and text feedback is encoded into 768 dimensions using BERT and then also mapped to 128 dimensions. These uniformly dimensional feature vectors are then fused using a multi-head attention mechanism: , in, are query, key, and value matrices respectively; is the feature dimension. Through the parallel computation of 8 attention heads, the system can capture the complex correlations between different modalities.

[0046] In one embodiment of the present invention, the system constructs a causal graph of the construction process using a causal discovery algorithm. Using the Peter-Clark (PC) algorithm, the system learns from historical data the causal chain of "insufficient welder numbers," "welding schedule delays," and "waiting for subsequent processes." Based on this causal relationship, when the system detects that the number of welders is 80% below the planned value, it can predict a schedule delay three days later and take proactive action. Scheduling decisions are made based on multiple factors. The priority scoring model is: , in, For the task Priority score; is the criticality (whether it is on the critical path, the value is 0 or 1); Urgency (countdown of days to deadline); is the dependency (normalized value of the number of dependent tasks); is the resource readiness (the proportion of available resources required); weight coefficient 、 、 、 Determined based on expert experience and historical data analysis.

[0047] The pallet partitioning planning module fully considers the actual constraints on site. The system first conducts a detailed constraint analysis, including space constraints and equipment constraints.

[0048] To address spatial constraints, the system uses laser scanning to capture 3D point cloud data on-site, achieving an accuracy of ±10mm. Based on this point cloud data, the system automatically identifies available storage areas and calculates their length, width, and height. For example, the temporary storage yard in Area A measures 20m × 15m × 6m, but considering safety aisle requirements, the actual available space is 18m × 13m × 5.5m. Analysis of transport aisles is equally important. The system identifies a main aisle width of 4.5m, a branch aisle width of 3m, and a minimum turning radius of 6m. These parameters directly influence the design of pallet dimensions.

[0049] Equipment capacity constraints are obtained through querying equipment records and real-time monitoring. For example, a 25-ton bridge crane has a rated load of 25 tons, but considering a safety factor of 0.8, the actual load limit is 20 tons. A 3-ton forklift has a fork length of 1.2m and a maximum lifting height of 4m. This means that the maximum dimensions of a pallet cannot exceed 1.2m × 1.2m × 3.5m (with a 0.5m safety margin).

[0050] The material flow digital twin model is implemented using discrete event simulation technology. Entities in the model include pallets (attributes include size, weight, and contents list), transport equipment (attributes include type, capacity, speed, and current location), storage locations (attributes include capacity and occupancy status), and construction sites (attributes include required materials and time of demand). The simulation process follows the following logic: materials depart from the warehouse → are loaded onto transport equipment → move along the planned route → arrive at the designated storage location → await construction demand → are delivered to the construction site.

[0051] In one embodiment of the present invention, the system evaluates the resilience of the solution through 1,000 Monte Carlo simulations. Each simulation randomly injects failure events, such as a forklift failure with a 5% probability and an average repair time of 2 hours, or a temporary aisle occupancy with a 10% probability and an average occupancy time of 30 minutes. The resilience index is defined as: , in, is an indicator of resilience; The completion time under normal circumstances; is the mean time to completion under failure conditions; Additional costs to deal with failures; =Total cost. Solutions with a toughness index greater than 0.85 are considered acceptable. Pallet loading optimization utilizes an improved three-dimensional packing algorithm. The algorithm first prioritizes pipelines, taking into account construction sequence (earlier use leads to higher priority), vulnerability (fragile parts have higher priority and require better protection), and weight (heavy items are placed at the bottom, light items at the top). Loading utilizes a "best fit descending" strategy, placing pipelines in locations that maximize space utilization.

[0052] The change management mechanism is key to ensuring system flexibility. The system classifies changes into three levels based on their impact, with each level having different processing procedures and approval requirements.

[0053] Level 1 changes are major modifications that affect multiple systems. For example, adjusting the main pipeline path may affect the connections of multiple branch pipelines. These changes require a comprehensive impact analysis process. The system first constructs a change propagation graph, where nodes represent affected objects (pipelines, drawings, construction tasks, etc.) and edges represent impact relationships. Using a breadth-first search of the graph, the system identifies all directly and indirectly affected objects.

[0054] The degree of impact is quantified using the following formula: , in, is the total impact level; is the number of affected objects; For the object Importance weight of is the direct impact degree, the value range is is the degree of indirect impact; is the attenuation coefficient of the indirect effect.

[0055] In one embodiment of the present invention, the change impact threshold is set at 0.3. This threshold is determined based on historical project data analysis: when the impact is less than 0.3, the chain reaction caused by the change can usually be resolved through local adjustments; when it exceeds 0.3, systematic re-planning is required. When the cumulative impact exceeds the threshold, the system automatically triggers the secondary scheduling process.

[0056] Version management utilizes a Git-like mechanism. Each change generates a new version, with a version number format of major.minor.revision. Major versions are updated at project milestones, minor versions at level one changes, and revisions at level two and three changes. The system maintains a complete snapshot of each version, supporting version comparison and historical tracking. The system also records the context of each change, including the reason for the change, the decision-making basis, and the individuals involved, to provide a complete change tracking chain.

[0057] The construction guidance output module converts optimization results into on-site executable guidance documents. The intelligent annotation process of 2D drawings fully considers the usage habits of construction workers.

[0058] The system extracts key construction information from the 3D model using a combination of rules-based and machine learning. For standardized information, such as the location of pipeline segments, the system directly identifies them through geometric analysis. For complex information, such as the optimal welding sequence, the system generates recommendations using a sequence model trained using historical construction data.

[0059] Key information needed to guide on-site construction is extracted from the 3D design model of the vessel's piping. This includes the specific location coordinates of the pipeline segments to be fabricated, the type and parameters of welds used at each connection (such as butt welds and fillet welds), and the precise installation coordinates and types of pipe supports. Based on the optimization results, a construction sequence is generated, and the dependencies between the various steps are annotated. This extracted information is directly translated into specific annotations in the construction guidance documents, ensuring that the design intent is accurately conveyed to the construction personnel.

[0060] Annotation layout optimization is an important technical detail. The system divides the drawing into a grid, and each annotation occupies a certain grid cell. The layout optimization problem is modeled as: , in, is the marked quantity; For annotation and The center distance; is an overlap indicator variable, which is 1 when there is overlap and 0 otherwise. By solving with a genetic algorithm, the system can find a label layout that is both clear and compact.

[0061] Construction animations are generated using procedural animation technology. The system first constructs a sequence of keyframes, each corresponding to a key construction node. Interpolation algorithms automatically generate transitions between frames. For example, the pipeline hoisting animation follows the following process: initial position (ground) → hoisting (Bezier curve path) → aerial movement (accounting for swing) → positioning (precise alignment) → final securing. The animation's timescale is compressed proportionally to the actual construction time, typically using a 1:60 ratio (one minute of animation corresponds to one hour of construction).

[0062] The distributed collaboration mechanism of this invention uses a variety of advanced technologies to ensure the efficient operation of the system. The blockchain-based collaborative framework solves many problems of traditional centralized systems.

[0063] In one embodiment of the present invention, the blockchain utilizes a consortium chain architecture, with participating nodes including the design department, production department, materials department, and quality department. Each department runs a full node, maintaining a complete copy of the ledger. The consensus mechanism utilizes the Practical Byzantine Fault Tolerance (PBFT) algorithm, ensuring system operation even in the event of partial node failure.

[0064] Smart contracts are designed with the business logic of shipbuilding in mind. For example, the trigger conditions for completing a piping design contract include: passing a 3D model integrity check, no serious conflicts detected during collision detection, a compliance inspection pass rate exceeding 95%, and completeness of all relevant drawings. When these conditions are met, the contract automatically executes, notifying relevant departments and updating the task status.

[0065] The federated learning mechanism enables different shipyards to share AI model improvements without disclosing commercial secrets. During model training, each shipyard uses local data to calculate the gradient: , in, For shipyards gradient; are model parameters; For shipyards After the central server collects the gradients of each shipyard, it uses the federated average algorithm to update the global model: , in, is the learning rate; is the number of participating shipyards. To protect privacy, each shipyard adds noise that satisfies differential privacy before uploading the gradient: , in, The mean is , the variance is Gaussian noise; The value of needs to balance privacy protection and model accuracy. In this embodiment, it is set to .

[0066] The system's continuous optimization mechanism ensures long-term performance improvement. The spiral growth of data, knowledge, and intelligence is achieved through multiple mechanisms.

[0067] Knowledge extraction utilizes pattern mining technology. Systematic analysis of historical project data identifies successful design patterns. For example, association rule mining revealed that when the pipe density exceeds 3 pipes / m², the success rate of a layered layout is 92%. These patterns are formalized into knowledge rules and stored in a knowledge base.

[0068] Knowledge is applied through a reasoning engine. When faced with a new design task, the system first searches for similar cases and then applies relevant knowledge rules to generate an initial solution. Knowledge is updated using an incremental learning approach, allowing experience from new projects to be promptly incorporated into the knowledge base.

[0069] Our layered optimization architecture effectively balances responsiveness and global optimization. Local optimization at the bottom layer is performed every minute to address urgent adjustments; domain coordination at the middle layer is performed hourly to ensure consistency within disciplines; and global optimization at the top layer is performed daily to strategically adjust resource allocation.

[0070] An optimization budget mechanism prevents over-optimization. Each module is allocated a fixed amount of computing resources per cycle. For example, the collision detection module can use 100 core-minutes of computing resources per hour. Modules are required to allocate resources based on task importance, performing detailed inspections on critical pipelines and rapid inspections on less important pipelines.

[0071] See also Figure 1 and Figure 5 The present invention also provides a system for managing the ship piping design drawing stage. The system includes five main modules, and each module works together to achieve the overall function.

[0072] Data Acquisition and Preprocessing Module 1 is responsible for data acquisition and standardization. This module includes multiple sub-units: the ERP interface unit connects to the shipyard management system, using a RESTful API and supporting data exchange in JSON format; the specification database management unit maintains structured process specification knowledge, supporting version management and incremental updates; the data cleaning unit identifies and processes abnormal data and fills in missing values; and the timestamp management unit adds a time stamp to each data item, supporting temporal queries.

[0073] The Intelligent Analysis and Decision-Making Module 2 is the core of the system and consists of five key units. The multi-objective optimization engine utilizes a parallel computing architecture, capable of simultaneously processing multiple optimization tasks. The collision detection unit uses GPU-accelerated tensor operations to increase detection speed by over 10 times. The process specification detection unit integrates natural language processing capabilities, capable of understanding and translating textual specifications. The dynamic scheduling unit achieves millisecond-level response speeds, ensuring timely handling of emergencies. The pallet partitioning unit uses simulation technology to predict material flow and optimize storage and transportation solutions.

[0074] Solution Generation and Verification Module 3 converts analysis results into executable solutions. Its simulation engine supports multiple simulation modes: deterministic simulation for verifying basic feasibility, stochastic simulation for risk assessment, and optimization simulation for parameter tuning. Verification rules can be flexibly configured to meet the needs of different projects.

[0075] The Construction Instruction Output Module 4 is responsible for generating on-site guidance documents. The drawing generation unit supports multiple CAD formats, ensuring compatibility with existing design software. The animation rendering unit utilizes real-time rendering technology to quickly generate high-quality construction animations. Output formats include PDF documents, MP4 videos, and interactive 3D models to meet diverse usage scenarios.

[0076] The Execution Feedback and Iterative Optimization Module 5 achieves closed-loop management. The feedback collection unit supports multiple input methods: mobile app input, voice recognition, image recognition, and more. The Change Impact Analysis Unit can complete impact assessments within 30 seconds, supporting rapid decision-making. The version control unit utilizes distributed storage to ensure data security and traceability.

[0077] Modules communicate loosely via message queues, employing a publish-subscribe model to ensure system scalability and reliability. For example, when the data acquisition module acquires new data, it publishes a data update event. The intelligent analysis module subscribes to this event and triggers the corresponding analysis process. This architectural design enables the system to flexibly respond to changing requirements, facilitating functional expansion and performance optimization.

[0078] It can be seen from the above embodiments that the ship piping design and drawing stage management method and system provided by the present invention, by introducing advanced technologies such as artificial intelligence, digital twins, and blockchain, realizes deep collaboration and intelligent optimization of design, production, and materials, significantly improves the efficiency and quality of ship construction, and has important engineering application value.

[0079] The above-described embodiments merely represent specific implementations of the present invention. While the descriptions are relatively specific and detailed, they should not be construed as limiting the scope of the present invention. It should be noted that a person skilled in the art would be able to make numerous variations and improvements without departing from the spirit of the present invention, and all such variations and improvements fall within the scope of protection of the present invention.

Claims

1. A method for managing the drawing stage of ship piping design, characterized in that: include: Obtain production schedule data and specification parameters from the process specification database, structure the acquired data, and establish a data timestamp mechanism to form a standardized input data set; Based on the generated input data set, a multi-objective optimization engine performs a comprehensive trade-off analysis to minimize construction conflicts, schedule deviations, and material costs. Simultaneously, it performs collision detection based on a four-dimensional spatiotemporal model, process specification compliance verification based on semantic understanding, dynamic scheduling optimization considering real-time deviations, and pallet allocation planning based on on-site constraints. This generates a preliminary piping design plan that includes construction phase division, drawing time schedule, and bill of materials. Based on the generated preliminary piping design, conduct 3D simulation verification and feasibility assessment to identify potential design conflicts and construction risks, and form a verified optimized solution. Based on the verified optimization plan, generate 2D drawings with construction stage markings and construction process animation videos, and output executable construction guidance documents; During the construction execution process, on-site feedback information and design change requirements are collected. When the impact of the change exceeds the preset threshold, secondary scheduling is automatically triggered, the drawing plan is updated, and the comprehensive trade-off analysis based on the multi-objective optimization engine, collision detection based on the four-dimensional space-time model, process specification compliance verification based on semantic understanding, dynamic scheduling optimization considering real-time deviations, and analysis and optimization process of pallet division planning based on on-site constraints are re-executed to achieve iterative optimization of the design scheme.

2. The method for managing the ship piping design drawing stage according to claim 1 is characterized in that: The comprehensive trade-off analysis performed by the multi-objective optimization engine specifically includes: The construction space is divided into three-dimensional grid units and the occupancy time window of each grid unit is marked. By identifying the situation where multiple pipelines occupy the same space at the same time, the spatial conflict severity index is calculated. At the same time, the equipment sharing conflicts and process parallel feasibility between different pipeline construction processes are analyzed to quantify the impact of process conflicts on the overall progress. Construct a construction network diagram and calculate the earliest start time and latest completion time for each activity. Identify the key activity sequence with zero total float time to form the critical path. Analyze the construction uncertainty of various pipelines based on historical data and set time buffers for high-risk activities. Analyze the demand distribution of pipes of the same specifications and calculate the unit cost under different batches. Comprehensively evaluate the capital cost of advance procurement and on-site storage space limitations to determine the optimal material procurement batch and timing. The weight coefficients of the three optimization objectives of construction conflict, schedule deviation and material cost are dynamically adjusted according to the project stage. Through iterative optimization, multiple candidate plans are generated and the comprehensive score of each plan is calculated. The plan with the best overall performance is selected as the output result.

3. The method for managing the ship piping design drawing stage according to claim 1 is characterized in that: The collision detection based on the four-dimensional space-time model specifically includes: Construct a four-dimensional space-time occupancy tensor containing three-dimensional spatial coordinates and time dimensions, where the first three dimensions represent the position coordinates of the pipeline in space, and the fourth dimension represents the construction time series; For each pipeline section, calculate its temporal and spatial occupancy trajectory throughout the construction period, including the hoisting path, the space occupied during the installation process, and the space required for welding operations; Tensor operations are used to identify overlapping areas of different pipelines in four-dimensional space-time. A probability distribution model is introduced to account for the uncertainty of the construction process. Monte Carlo simulation is used to generate multiple possible construction scenarios and calculate the probability distribution of collisions. When the collision probability exceeds a preset threshold, a collision resolution strategy is generated based on deep reinforcement learning, including a spatial avoidance plan that adjusts the pipeline path and a temporal avoidance plan that adjusts the construction sequence. The impact of different resolution strategies on the overall construction progress and cost is evaluated, and the optimal solution is selected.

4. The method for managing the ship piping design drawing stage according to claim 1 is characterized in that: The process specification compliance verification based on semantic understanding specifically includes: A pre-trained language model is used to perform deep semantic analysis on the process specification text to extract specific control elements in the process specification text, including: constrained pipeline components, technical requirements that should be followed, specific limited numerical ranges, and the core semantic elements of the ship area or pipeline type to which the specification applies, generating a high-dimensional semantic vector representation of the specification; Construct a specification semantic space and calculate the semantic similarity between different specifications, automatically identify the association, conflict and complementary relationship between specifications, and establish a mapping relationship between specification semantics and design parameters; Automatically convert abstract regulatory requirements into specific ship piping design parameters, including: minimum bend radius values ​​required at pipe bends, maximum allowable velocity variation percentage when the fluid flows through the pipe, minimum and maximum spacing ranges between adjacent pipe supports, and groove angle values ​​required at welds, all of which are quantifiable process indicators; Construct a specification dependency graph and identify potential specification conflict patterns through graph neural networks. Specs are divided into hard constraints and soft constraints. A multi-objective optimization algorithm is used to find a design solution that maximizes overall constraint satisfaction. For conflicts that cannot be completely resolved, detailed conflict reports and resolution suggestions are generated.

5. The method for managing the ship piping design drawing stage according to claim 1 is characterized in that: The dynamic scheduling optimization considering real-time deviation specifically includes: Obtain real-time information on the actual drawing completion time of each pipeline, the time and content of design changes, and on-site feedback on problems and delays. Calculate the time deviation between the actual and planned progress and analyze the development trend of the deviation. Build a cognitive digital twin system that integrates field data from multiple sources, including: environmental and equipment parameter data collected by IoT sensors installed on the construction site, construction footage captured by the on-site video surveillance system, quality and progress feedback submitted by construction workers via mobile terminals, and formal progress report data from the project management system. This system then fuses heterogeneous data using a deep learning model to form a unified cognitive state representation. An attention mechanism is introduced to identify key state changes, and a causal reasoning network is used to analyze the causal relationship in the construction process and predict the causal effects of different scheduling decisions. Based on the current schedule deviation, resource constraints and downstream process dependencies, a priority scoring model is established and the drawing priority of each piping system is dynamically adjusted, and design resources are reallocated to ensure that tasks on the critical path are completed first.

6. The method for managing the ship piping design drawing stage according to claim 1 is characterized in that: The pallet partitioning planning based on on-site constraints specifically includes: Analyze the spatial layout constraints of the construction site, including the dimensions of available storage space, width and height restrictions of transport aisles, load ratings of lifting equipment, and loading capacities of transport vehicles; Build a digital twin model of material flow, simulate the entire process of material flow from the warehouse to the construction site through discrete event simulation technology, and identify potential bottlenecks and conflicts under different pallet partitioning schemes; Pipelines with the same specifications and in similar construction areas are grouped according to similarity, ensuring that the weight of each pallet does not exceed the equipment load limit and the size meets transportation constraints. A three-dimensional packing algorithm is used to optimize pallet space utilization. Monte Carlo simulations are used to evaluate the resilience of pallet solutions under abnormal circumstances, including transport equipment failures and temporary unavailability of storage areas. An efficient and resilient pallet partitioning scheme is selected, and a unique identification code is generated for each pallet, containing area, batch, and content information.

7. The method for managing the ship piping design drawing stage according to claim 1 is characterized in that: The collection of on-site feedback information and design change requirements and automatic triggering of secondary scheduling specifically includes: Identify and categorize sources of changes, including proactive changes due to design optimization, reactive changes due to field feedback, and mandatory changes due to specification updates. Based on the scope of impact, these changes are categorized as Level 1 changes affecting multiple systems, Level 2 changes within a single system, and Level 3 changes that do not affect interfaces. Analyze the direct impact of the change, including changes to upstream and downstream connections of the pipeline, the need for simultaneous revision of related drawings, and the impact on completed construction. Simultaneously assess the indirect impact, including the chain reaction changes in space occupancy, the impact on other disciplines, and the impact on the overall project schedule. Setting a change impact threshold, and automatically triggering a rescheduling process when the cumulative impact exceeds the threshold, updating the status of the affected tasks and re-running the multi-objective optimization engine to generate a revised output plan; Generate a new version number for each change and save complete data before and after the change, support version comparison and rollback, determine the corresponding approval process based on the change level, and record approval opinions and decision-making basis to ensure the traceability of changes.

8. The method for managing the ship piping design drawing stage according to claim 1 is characterized in that: The generating of the two-dimensional drawings with construction stage identification and the construction process animation video specifically includes: Extract key information for on-site construction guidance from the 3D design model of the ship's pipelines, including the specific location coordinates of the pipelines to be fabricated in sections, the type and parameters of the welding points used at each connection, and the precise installation location coordinates and types of the pipeline brackets—key technical information necessary for construction personnel. Based on the optimization results, a construction sequence number is generated and the dependencies between the processes are marked. Automatically add segment numbers, positioning dimensions, welding symbols, quality requirements, material specifications, and quantity information to 2D drawings. Avoid overlapping and occlusion of annotations through layout optimization algorithms, and use layers to manage different types of annotation information. Build a 3D scene including a ship structure model, pipeline model, and construction equipment model. Set the animation rhythm and determine the key frame positions according to the actual construction time to simulate the complete process of pipeline lifting, welding operations, and inspection and testing. Use color changes in animations to indicate construction status, add arrows to indicate installation direction, set translucent effects to display internal structures, overlay current step instructions, progress timelines, and safety warning signs, and generate multi-format output files that support pause control, chapter jumps, and on-site annotation functions.

9. The method for managing the ship piping design drawing stage according to any one of claims 1 to 8, characterized in that: Also includes: Adopting a distributed collaboration framework based on blockchain, with the design, production, and materials departments as blockchain network nodes, verifying and synchronizing data through a consensus mechanism, and using smart contracts to automatically trigger cross-departmental collaboration processes; Build a distributed intelligent optimization mechanism based on federated learning to achieve multi-party collaborative training without sharing original data, and protect data privacy through differential privacy and secure multi-party computing technologies; Establish a spiral mechanism among data, knowledge, and intelligence. Learn design patterns from historical project data to form a domain knowledge base. This knowledge guides intelligent decision-making in current projects. Decision-making results generate new data, which is then updated through continuous learning. A hierarchical optimization architecture is set up to balance local optimization and global coordination. The bottom-level modules perform real-time local optimization, the middle-level modules perform intra-domain coordination, and the top-level modules perform periodic global strategic optimization. The optimized resource allocation of each module is controlled through the optimized budget mechanism.

10. A system for managing the drawing stage of ship piping design, characterized in that: include: The data acquisition and preprocessing module is used to obtain production schedule data and specification parameters from the process specification database, structure the acquired data, and establish a data timestamp mechanism to form a standardized input data set; An intelligent analysis and decision-making module, including a multi-objective optimization engine, a collision detection unit, a process specification detection unit, a dynamic scheduling unit, and a pallet division unit, is used to perform a comprehensive trade-off analysis based on the input data set to minimize construction conflicts, minimize schedule deviations, and optimize material costs. It performs collision detection based on a four-dimensional spatiotemporal model, process specification compliance verification based on semantic understanding, dynamic scheduling optimization considering real-time deviations, and pallet division planning based on on-site constraints, generating a preliminary piping design plan that includes construction phase division, drawing time plan, and material list; A scheme generation and verification module is used to perform three-dimensional simulation verification and feasibility assessment based on the preliminary piping design scheme, identify potential design conflicts and construction risks, and form a verified optimized scheme; A construction guidance output module is used to generate two-dimensional drawings with construction stage identification and construction process animation videos based on the verified optimization plan, and output executable construction guidance documents; The execution feedback and iterative optimization module is used to collect on-site feedback information and design change requirements during the construction execution process. When the impact of the change exceeds the preset threshold, it automatically triggers secondary scheduling, updates the drawing plan, and calls the intelligent analysis and decision-making module to re-execute the comprehensive trade-off analysis of minimizing construction conflicts, minimizing schedule deviations, and optimizing material costs, collision detection based on a four-dimensional space-time model, process specification compliance verification based on semantic understanding, dynamic scheduling optimization considering real-time deviations, and analysis and optimization process of pallet division planning based on on-site constraints, so as to achieve iterative optimization of the design scheme.

Images