Tower crane automatic type selection and arrangement optimization modeling method based on generative adversarial network
By automatically generating tower crane selection and layout solutions through the generation of adversarial network (GAN) models, the problems of insufficient scientificity and time-consuming data processing in traditional methods are solved, and efficient and stable tower crane optimization is achieved, adapting to complex construction environments, and improving construction efficiency and safety are improved.
Patent Information
- Application Number
- CN202510396464.7
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2025-03-31
- Publication Date
- 2025-07-18
AI Technical Summary
Traditional tower crane selection and layout methods rely on manual experience, lack scientificity and objectivity, and are difficult to deal with complex construction scenarios, the optimization results are stable and efficient, and data processing is time-consuming and susceptible to human errors.
Using a generative adversarial network (GAN) model, through adversarial training generators and discriminators, we learn construction site characteristics, and automatically generate tower crane selection and layout solutions that meet the multi-objective optimization needs. Combining construction goals and restrictions, multiple candidate solutions are generated and simulated and evaluated.
It significantly improves the intelligence and automation of tower crane selection and layout, improves construction management efficiency, reduces project costs, and can flexibly adapt to dynamic construction changes. The generated solutions are highly accurate and stable in complex scenarios.
Smart Images

Figure CN120337528A_ABST
Abstract
Description
Technical Field
[0001] The present invention relates to a method for automatic selection and layout optimization modeling of tower cranes based on generative adversarial networks, belonging to the field of engineering construction management. Background Technique
[0002] The role of the construction industry as a pillar industry of the national economy is constantly strengthening [1] , with the rapid development of the construction industry, the construction demand for high-rise and super-high-rise buildings is continuously increasing. Tower cranes (hereinafter referred to as tower cranes) are important mechanical equipment in modern construction. They can carry out material transportation tasks in the vertical and horizontal directions within a large coverage area. Therefore, they play an indispensable role in improving the transportation efficiency of the construction site and ensuring the project progress [2] . However, due to the limitation of site space and the need for parallel construction in multiple construction sections, some hoisting areas between adjacent tower cranes will overlap when arranging tower crane groups. There is a risk of collision between the boom and the boom or the boom and the lifting rope in the overlapping area [3] . In large-scale connected construction projects, due to the wide operation range and frequent cross-operation of tower crane groups, the collision risk increases significantly, which not only has a negative impact on construction efficiency, but may also cause serious safety hazards and even lead to major engineering accidents [4] . Therefore, how to reasonably select and optimize the layout of tower cranes has become one of the key problems that need to be solved urgently in construction management.
[0003] Traditional methods for tower crane selection and layout mainly rely on manual experience to determine the position of tower cranes. This experience-driven method not only lacks scientificity and objectivity, but also the decision-making quality often drops significantly when dealing with diverse building layout requirements or emerging hoisting technologies (such as modular construction) [5] . In addition, it is difficult for such methods to comprehensively consider the characteristics of the construction site, the dynamic changes of the construction progress, and the multi-objective optimization requirements such as economy, safety, and efficiency, thus restricting their application scope and optimization effect.
[0004] In view of the deficiencies of traditional methods, relevant researchers have introduced numerical calculation methods, including meta-heuristic algorithms (such as Genetic Algorithm (GA) [6] , Firefly Algorithm (FA) [2] ) and hybrid meta-heuristic algorithms (such as Artificial Neural Network (ANN)-GA method [7] and Probability Bees Algorithm (PBA) [8])。These methods can generate near-optimal solutions within limited computational time, but there are still limitations in many aspects. On the one hand, it is difficult to guarantee the accuracy and stability of the optimization results in complex construction scenarios, and the success rate fluctuates greatly. On the other hand, these methods usually require explicit manual data input, and even experienced operators need to spend several hours sorting and inputting relevant information. This cumbersome operation process not only significantly increases the time cost but also may affect the optimization results due to human errors. Therefore, traditional numerical calculation methods are difficult to achieve efficient tower crane selection and layout optimization in dynamic and complex construction sites, and there is an urgent need for more intelligent and automated innovative solutions.
[0005] In recent years, with the rapid development of artificial intelligence technology, deep learning methods (such as Generative Adversarial Network, GAN [9] ) have shown significant potential in solving complex optimization problems. Through the adversarial training of the generator and discriminator, GAN can capture the characteristics of complex data distributions and generate high-quality candidate solutions. This method has been successfully applied in many fields, such as automated architectural and home design
[10] , indoor layout optimization
[11] and shear wall design
[12] , demonstrating high planning efficiency and effectiveness in problem-solving. With the powerful generation ability of GAN, it is expected to break through the limitations of traditional methods and provide a more scientific, efficient, and automated solution for tower crane selection and layout optimization.
[0006] In summary, the present invention proposes a method for automatic tower crane selection and layout optimization modeling based on GAN, aiming to solve the problems of insufficient scientificity, time-consuming manual data processing, and low stability of optimization results in traditional methods. By introducing the GAN model, this method makes full use of the adversarial training mechanism between the generator and discriminator, can learn the complex characteristics of the construction site, and generate tower crane selection and layout plans that meet the multi-objective optimization requirements (such as economy, safety, and efficiency). Compared with traditional manual experience or numerical optimization methods, this method not only significantly improves the intelligence and automation of plan generation but also shows stronger adaptability and real-time optimization ability in dynamic construction environments. Summary of the Invention
[0007] The present invention provides a method for automatic tower crane selection and layout optimization modeling based on a generative adversarial network, aiming to provide an automated and intelligent solution for tower crane selection and layout, thereby improving construction management efficiency, reducing project costs, and being able to flexibly adapt to dynamic changes during the construction process.
[0008] The present invention solves the above technical problems through the following technical solutions:
[0009] The present invention provides a method for automatic selection and layout optimization modeling of tower cranes based on a generative adversarial network. The working steps of this method include:
[0010] Step 1: Acquisition and preprocessing of construction site information. (1) Obtain the general design drawings of the proposed building complex at the construction site, and obtain the original construction data such as the CAD floor plans of the proposed buildings, the site conditions of the construction site, construction resources, and construction progress; (2) Preprocess the original construction data to extract the information related to the tower crane layout, including but not limited to: the construction positions and boundaries of the proposed buildings, the positions of temporary facilities and obstacles in the site, the potential layout positions of tower cranes, the positions of building material yards, and the distribution of other construction resources; (3) Generate the image to be designed based on the extracted information, and classify and label different types of information in the image, and distinguish them with a unified color coding, so as to be used as the input of the subsequent generative adversarial network (GAN) model.
[0011] Step 2: Requirement definition and constraint determination. Determine the core requirements for the selection and layout optimization of tower cranes, including: (1) Determine the potential models of tower cranes that meet the construction requirements; (2) Determine the number of tower cranes that meet the building construction coverage requirements; (3) Optimize the spatial layout of tower cranes to maximize the coverage range and lifting efficiency. Determine the limiting conditions for the selection and layout optimization of tower cranes, including but not limited to: economic constraints (such as cost budget), technical constraints (such as construction process requirements), and safety constraints (such as operation safety and interference avoidance).
[0012] Step 3: Construction of the generative adversarial network model. (1) Construct a generator model for generating the selection and layout plan of tower cranes that meets the construction requirements according to the input construction site characteristics; (2) Construct a discriminator model for discriminating the similarity between the generated plan and the historical real plan; (3) Define a loss function to improve the accuracy and applicability of the generated plan by synchronously optimizing the performance of the generator and the discriminator.
[0013] Step 4: Training of the generative adversarial network model. (1) Initialize the generator and the discriminator using a pre-trained model, and normalize the input image to be designed; (2) During the model training process, the generator generates the selection and layout plan of tower cranes, the discriminator evaluates the rationality of the generated plan, and improves the plan quality by optimizing the model parameters; (3) By learning the layout planning mode in actual cases, ensure that the tower crane layout plan generated by the model can meet the requirements of the actual construction scenario.
[0014] Step 5: Selection and optimization of the optimal solution. (1) Based on the output of the GAN model, generate multiple candidate solutions, covering different tower crane types, quantities, and layout methods; (2) Use construction objectives (economy, efficiency, safety, etc.) and on-site constraints to evaluate the candidate solutions through construction simulation tools, and verify their feasibility in terms of hoisting paths, construction cycles, and resource utilization.
[0015] The flow chart of the optimization modeling method for automatic selection and layout of tower cranes based on GAN is as Figure 1 shown.
[0016] As a further solution of the invention, in step 1, the invention extracts and processes the image data in the original construction materials to generate a structured design drawing, so as to provide an effective input for the subsequent generative adversarial network (GAN) model. The extracted image content includes information such as the construction and building boundary positions, identified obstacles (such as existing buildings and planned temporary facilities), potential tower crane layout positions, building floor plans, and material yard positions. The image extraction process for tower crane selection and layout planning is as Figure 2 shown.
[0017] The invention unifies the legends and color markings of the to-be-designed images, which are consistent with the existing classification standards in the dataset, ensuring that the GAN model can accurately identify and apply the learned planning patterns to predict the feasible layout positions of tower cranes. In the image generation process, the to-be-designed drawings of the invention are generated based on image processing technology without explicit preprocessing of data, and can be implemented in common design software (such as AutoCAD). AutoCAD, as the most widely used CAD software globally, occupies approximately 40% of the market share
[13] , and its simplicity and popularity make the implementation of this method more convenient. The invention specially designs an image generation process, which is compatible with AutoCAD and similar design tools (such as TCLP software), to ensure that decision-makers can efficiently utilize existing tools for construction site planning and design.
[0018] Through the above process, the invention generates the to-be-designed images as the input of the GAN model, laying a data foundation for the subsequent tower crane selection and layout optimization, and at the same time improving the operability and applicability of the system.
[0019] As a further aspect of the invention, when performing step two, determine the core requirements for tower crane selection and layout optimization, including the following three aspects: (1) Selection of potential tower crane types: According to the conditions of the construction site and the requirements of the construction tasks, determine the tower crane types that meet the construction requirements, such as fixed tower cranes, self-erecting tower cranes, etc.; (2) Determination of the number of tower cranes: By analyzing the requirements for the coverage of building construction, reasonably plan the number of tower cranes to ensure the coverage of all construction areas and the completion of lifting tasks; (3) Optimization of the spatial layout: On the premise of meeting the requirements of construction tasks, optimize the spatial layout of the tower cranes to maximize the coverage range and lifting efficiency of the tower cranes, reduce the cross-operation area, and thus improve the overall construction efficiency.
[0020] Secondly, clarify the limiting conditions for tower crane selection and layout optimization, including but not limited to the following three aspects: (1) Economic constraints: Within the scope of the cost budget, achieve tower crane selection and layout optimization, and minimize the investment and operating costs of construction machinery and equipment; (2) Technical constraints: Comprehensively consider the requirements of construction technology, such as the compatibility of construction machinery and equipment in different construction stages and the applicability of tower cranes to specific building structures, to ensure the feasibility and efficiency of construction; (3) Safety constraints: Pay attention to operation safety and avoid space interference, especially when multiple tower cranes are operating in parallel. Ensure the safety distance between the boom and the boom or between the boom and the lifting rope, reduce the collision risk, and ensure construction safety.
[0021] As a further aspect of the invention, when performing step three, for the GAN-based automatic tower crane selection and layout optimization model, its key steps include the design of the generator (Generator) and the discriminator (Discriminator) and the definition of the loss function. The generator and the discriminator are optimized with each other through adversarial training to jointly complete the automation and intelligence of tower crane selection and layout. The generator aims to generate a tower crane selection and layout plan that meets the construction requirements based on the input construction site characteristics. Specifically, the generator receives the conditional image (x) and the noise vector (z) as inputs and outputs the predicted tower crane layout plan G(z|x). The conditional image x includes information such as the construction site layout, potential tower crane locations, building boundaries, and material stacking locations, while the noise z introduces diversity to the generated plan. The goal of the generator is to generate a tower crane selection and layout plan that is highly similar to the real plan, so as to "confuse" the discriminator to the greatest extent and make it difficult to accurately distinguish the difference between the generated plan and the historical real plan.
[0022] The discriminator is used to evaluate the rationality of the generation scheme, and its goal is to accurately judge whether the input data is real data. The discriminator receives real data pairs (x, y) and generated data pairs (x, G(z|x)), and outputs a probability value D(x, y), indicating the possibility that the input data is real data. By maximizing the recognition ability of real data and minimizing the misjudgment of generated data, the discriminator ensures that the model can generate higher-quality schemes.
[0023] To optimize the performance of the generator and the discriminator, Equation (1) defines the following loss function:
[0024]
[0025] The design of this loss function aims to maximize the recognition probability D(x, y) of real data and minimize the probability D(x, G(z|x)) that generated data is misidentified as real data.
[0026] As a further solution of the invention, in the initialization and training process of the generative adversarial network model in Step Four, the setting of hyperparameters is a key link, which is directly related to the convergence efficiency, generation effect and stability of the model. Aiming at the problem of automatic tower crane type selection and layout optimization, the present invention combines the actual requirements and the characteristics of the GAN model, and scientifically and reasonably sets and optimizes the main hyperparameters, including but not limited to the following aspects:
[0027] (1) Learning Rate: Appropriate learning rates are set for the generator and the discriminator respectively to ensure the balance of their training, avoid the problem of performance imbalance of the generator or the discriminator, and improve the stability and training efficiency of the model.
[0028] (2) Batch Size: During the training process, the number of samples in each iteration (such as 64) is set, taking into account both the training speed and the stability of the model performance. A reasonable choice of batch size can accelerate the update of model parameters and avoid overfitting caused by data fluctuations. (3) Latent Vector Dimension: The dimension of the noise input vector (such as 100) is set to increase the diversity of the generated results of the generator, so that it can more flexibly meet the requirements of tower crane type selection and layout in different construction scenarios.
[0029] (4) Loss Function Weights: In the loss function, the weights of the adversarial loss and the constraint loss are dynamically adjusted (for example, set to 1 and 0.5 respectively) to balance the accuracy of the generated results of the generator and the practical feasibility of the layout scheme, and ensure that the model converges quickly during training while improving the quality of the output scheme.
[0030] (5) Input image normalization: Normalize the input image to be designed, map the pixel values uniformly between 0 and 1, reduce the interference of data amplitude on the training process, and thus improve the convergence speed and generation stability of the model.
[0031] As a further solution of the invention, in step five, the invention realizes the intelligent decision-making of tower crane selection and layout through the results output by the generative adversarial network model, and can dynamically adjust according to the changes of actual construction conditions. The specific steps are as follows:
[0032] (1) Generation of candidate solutions: Based on the output of the GAN model, generate multiple candidate solutions, each solution covering different tower crane types, quantities and layout methods. These candidate solutions are automatically generated according to the construction site characteristics and the planning mode learned by the GAN model to ensure that they meet the basic requirements in terms of coverage, construction efficiency and safety.
[0033] (2) Scheme evaluation and verification: Comprehensively evaluate the performance of the candidate solutions under multiple objectives such as economy (such as total cost and resource consumption), efficiency (such as hoisting path optimization and construction period shortening) and safety (such as collision risk and operation stability). Check whether the candidate solutions meet the on-site restriction conditions, including constraints such as site boundaries, obstacle distribution, tower crane spacing, hoisting path and operation range overlap. Finally, simulate the path during the hoisting process to verify the rationality of the path planning and resource utilization rate to ensure that the solution can be implemented without obstacles in actual construction. Description of the drawings
[0034] Figure 1 Flowchart of the automatic tower crane selection and layout optimization modeling method based on GAN.
[0035] Figure 2 Image extraction process of tower crane selection and layout planning.
[0036] Figure 3 Comparison of layout results of different tower crane layout planning methods. Detailed implementation manners
[0037] The invention provides an automatic tower crane selection and layout optimization modeling method based on a generative adversarial network. The following combines the drawings and engineering examples to detail the specific implementation of the invention to verify the applicability and optimization effect of the model.
[0038] Taking a certain engineering project as an example, optimize the design of the tower crane selection and layout scheme, and evaluate the advantages of the optimization model based on the generative adversarial network proposed by the invention in terms of transportation efficiency and calculation time by comparing with the exhaustive search algorithm, genetic algorithm (GA) and the layout scheme of the actual engineering project.
[0039] This project is a high-rise building project, involving complex construction site conditions, including multiple obstacles, limited tower crane layout space, and high material transportation requirements. The specific project overview is as Figure 3 shown, including key information such as construction boundaries, building floor plans, material yard locations, potential tower crane locations, etc.
[0040] The tower crane selection and layout are optimized and calculated respectively through the exhaustive traversal algorithm, GA algorithm, and GAN. The results are shown in Table 1. In terms of transportation efficiency, the tower crane layout plan generated by GAN is significantly better than the layout plan in the actual project, shortening the transportation time from 7847.85 seconds to 5264.56 seconds, a reduction of 32.92%. Compared with the plan generated by the GA algorithm, GAN further shortens the total transportation time by 8.58%, fully demonstrating its effectiveness and accuracy in optimizing the plan quality. In terms of calculation efficiency, the calculation times of the exhaustive traversal algorithm, GA algorithm, and GAN are 31.45 seconds, 9.29 seconds, and 1.83 seconds respectively, among which GAN takes the shortest time. This result indicates that GAN can not only generate high-quality tower crane layout plans but also has extremely high calculation efficiency, showing significant advantages in dealing with dynamic construction scenarios and real-time optimization requirements.
[0041] Table 1 Comparison of different tower crane transportation path planning methods and the actual project layout in terms of transportation efficiency and calculation time
[0042]
[0043] References:
[0044] [1] Liu Meng, Huang Chun, Wang Jingjing, Wang Wenqi, et al. Optimization of Tower Crane Selection and Layout Based on Mixed Integer Linear Programming [J]. Journal of Civil Engineering and Management, 2020, 37(02): 142-150.
[0045] [2] Wang J, Zhang X, Shou W, et al. A BIM-based Approach for Automated Tower Crane Layout Planning [J]. Automation in Construction, 2015, 59: 168-178.
[0046] [3] Zhang C, Hammad A. Improving lifting motion planning and re-planning of cranes with consideration for safety and efficiency[J]. Advanced Engineering Informatics, 2012, 26(2): 396-410.
[0047] [4] Fang, Y., & Cho, Y. K. (2017). Effectiveness analysis from a cognitive perspective for a real-time safety assistance system for mobile crane lifting operations. Journal of Construction Engineering and Management, 143(4), 05016025.
[0048] [5] Lin J, Fu Y, Li R, et al. An algorithm for optimizing the location and type selection of attached tower cranes based on value engineering[C] / / International Conference on Construction and Real Estate Management 2020. Reston, VA: American Society of Civil Engineers, 2020: 106-117.
[0049] [6] Abdelmegid M A, Shawki K M, Abdel-Khalek H. GA optimization model for solving tower crane location problem in construction sites[J]. Alexandria Engineering Journal, 2015, 54(3): 519-526.
[0050] [7]Tam C M, Tong T K L. GA-ANN model for optimizing the locations of tower crane and supply points for high-rise public housing construction[J]. Construction Management and Economics, 2003, 21(3): 257-266.
[0051] [8]Lien L C, Cheng M Y. Particle bee algorithm for tower crane layout with material quantity supply and demand optimization[J]. Automation in Construction, 2014, 45: 25-32.
[0052] [9]Creswell A, White T, Dumoulin V, et al. Generative adversarial networks: An overview[J]. IEEE signal processing magazine, 2018, 35(1): 53-65.
[0053]
[10] Huang W, Zheng H. Architectural drawings recognition and generation through machine learning[C]. Proceedings of the 38th annual conference of the association for computer aided design in architecture, Mexico City, Mexico. 2018: 18-20.
[0054]
[11] Zhao B, Yin W, Meng L, et al. Layout2image: Image generation from layout[J]. International Journal of Computer Vision, 2020, 128(10): 2418-2435.
[0055]
[12] Liao W, Lu X, Huang Y, et al. Automated structural design of shearwall residential buildings using generative adversarial networks[J]. Automation in Construction, 2021, 132: 103931.
[0056]
[13] Ma K, Mao Z, He D, et al. Design a network architectural teaching system by auto CAD[J]. Computer-Aided Design and Applications, 2020, 17(S2): 1-10.
Claims
1. An automatic selection and layout optimization modeling method for tower cranes based on generative adversarial networks, characterized in that, The steps of the automatic tower crane type selection and layout optimization modeling method include: Step 1: Acquisition and preprocessing of construction site information; (1) Obtain the general design drawings of the buildings to be constructed at the construction site, and obtain the CAD floor plans of the buildings to be constructed, the construction site conditions, construction resources, and the original construction data of the construction progress; (2) Preprocess the original construction data, and extract the information related to the tower crane layout, including: the construction positions and boundaries of the buildings to be constructed, the positions of temporary facilities and obstacles within the site, the potential layout positions of the tower cranes, the positions of building material yards, and the distribution of other construction resources; (3) Generate the image to be designed based on the extracted information related to the tower crane layout, classify and label different types of information in the image to be designed, and distinguish them with a unified color coding, so as to be used as the input of the subsequent generative adversarial network GAN model; Step 2: Requirement definition and constraint determination; Determine the core requirements for tower crane type selection and layout optimization, including: (1) Determine the potential tower crane models that meet the construction requirements; (2) Determine the number of tower cranes that meet the building construction coverage requirements; (3) Optimize the spatial layout of the tower cranes to maximize the coverage range and lifting efficiency; Determine the limiting conditions for tower crane type selection and layout optimization, including: economic constraints, technical constraints, and safety constraints; Step 3: Construct a generative adversarial network GAN model; (1) Construct a generator model for generating tower crane type selection and layout plans that meet the construction requirements according to the input construction site characteristics; (2) Construct a discriminator model for discriminating the similarity between the generated plan and the historical real plan; (3) Define a loss function, and improve the accuracy and applicability of the generated plan by synchronously optimizing the performance of the generator and the discriminator; Step 4: Training of the generative adversarial network GAN model. (1) Initialize the generator and the discriminator with a pre-trained model, and normalize the input image to be designed; (2) During the training process of the generative adversarial network GAN model, the generator generates tower crane type selection and layout plans, the discriminator evaluates the rationality of the generated plans, and improves the plan quality by optimizing the parameters of the generative adversarial network GAN model; (3) By learning the layout planning mode in actual cases, ensure that the tower crane layout plans generated by the model can meet the requirements of the actual construction scenario; Step 5: Selection and optimization of the optimal plan; (1) Based on the output of the generative adversarial network GAN model, generate multiple candidate plans, covering different tower crane types, quantities, and layout methods; (2) Use the construction objectives and on-site limiting conditions to evaluate the candidate plans through a construction simulation tool, and verify their feasibility in terms of lifting paths, construction periods, and resource utilization.
2. The tower crane automatic selection and layout optimization modeling method based on the generative adversarial network according to claim 1, characterized in that, In step one, the construction site conditions include the site size, topographic features, and the distribution of foundation bearing capacity; the construction resources include the types and performance parameters of construction machinery and equipment, the types of building construction materials, and the locations of material yards; the construction schedule includes the processes on the critical path of construction and their time arrangements. At the same time, obtain the resource requirement plan for each stage, including the number of equipment, manpower allocation, and material supply, and clarify the total construction period and the time constraint conditions for each stage.
3. The tower crane automatic selection and layout optimization modeling method based on the generative adversarial network according to claim 2, characterized in that, Among the construction site conditions, the topographic features include the elevation changes, flatness, and the distribution of obstacles, which are used to evaluate whether the construction site meets the stability requirements for the installation of tower crane foundations.
4. The automatic type selection and layout optimization modeling method for tower cranes based on a generative adversarial network according to claim 1, wherein In step one, use AutoCAD software to process and analyze the relevant data of the construction site, and extract the key information affecting the construction site conditions and the tower crane layout from the design drawings; the extracted content includes the boundaries of the construction and the planned buildings, the identified obstacles, the potential locations for tower crane layout, and the locations of building material yards, generating a structured design drawing containing various types of key information to provide a data basis for subsequent optimization.
5. A tower crane automatic selection and layout optimization modeling method based on a generative adversarial network according to claim 1, characterized in that, In step one, the structured design drawing classifies and labels different types of information, including the boundaries of the planned buildings, the locations of material yards, the potential locations for tower crane layout, and temporary facilities; Use a unified color coding to distinguish different types of information, including: the potential locations for tower crane layout are marked in yellow, the building material supply areas are marked in blue, and other construction information is marked in black; The generated image to be designed is used as the input for the subsequent generative adversarial network (GAN) model, providing intuitive and standardized data support for the GAN model to accurately learn the construction site characteristics and generate a reasonable tower crane layout plan.
6. The automatic selection and layout optimization modeling method of tower crane based on generative adversarial network according to claim 1, characterized in that In step two, determine the core requirements for tower crane selection and layout optimization by analyzing the construction requirements. First, according to the specific requirements and technical parameters of building construction, screen the potential types of tower cranes that meet the requirements for lifting capacity, boom length, and stability; based on the construction site area and the distribution characteristics of the buildings, evaluate the number of tower cranes required to ensure full coverage of the construction site and construction efficiency; finally, optimize the spatial layout of the tower cranes to maximize their coverage range, while ensuring the shortest lifting path and the rational use of resources to improve the lifting efficiency.
7. A tower crane automatic selection and layout optimization modeling method based on a generative adversarial network according to claim 1, characterized in that In step two, the economic constraints specifically include: (1) Evaluate the impact of the tower crane selection and layout plan on the project budget, considering the equipment purchase, rental, and transportation costs comprehensively; (2) Combine the construction schedule arrangement and analyze the operation and maintenance costs during the tower crane usage period to ensure that the total cost is controlled within the budget range; The technical constraints specifically include: (1) Determine the degree of support of the tower crane type and layout location for the key processes according to the construction process requirements; (2) Consider the lifting capacity, lifting height, and boom swing range of the tower crane to ensure compliance with the construction requirements of the building; The specific safety constraints include: (1) the impact of the tower crane layout plan on operation safety, including anti-collision measures between tower cranes and the safety distances between tower cranes and facilities and buildings within the site; (2) the impact of the tower crane hoisting path on construction site personnel, equipment, and material yards to avoid potential safety hazards.
8. A tower crane automatic selection and layout optimization modeling method based on a generative adversarial network according to claim 1, characterized in that, When performing Step 3, for the construction of the generator model, first, a convolutional neural network (CNN) is used to extract the input construction site features to generate a tower crane selection and layout plan that meets the construction requirements; according to the construction feature embedding module, the generated plan is optimized for spatial layout by combining site information and task requirements; finally, a multi-layer neural network structure is designed to ensure that the generated tower crane layout plan meets the actual requirements in terms of applicability and diversity. When performing Step 3, for the construction of the discriminator model, first, a multi-layer perceptron (MLP) is used to analyze the similarity between the input plan and the historical real plan to evaluate the rationality of the generated plan. Then, an efficient discrimination module is designed to provide real-time feedback on the performance of the generator and improve the overall training efficiency of the generative adversarial network.
9. A tower crane automatic selection and layout optimization modeling method based on a generative adversarial network according to claim 1, characterized in that, In Step 4, the generator and the discriminator are initialized using a pre-trained model, and transfer learning is performed based on the data in actual construction cases to improve the initial performance and convergence speed of the pre-trained model; the generator repeatedly generates tower crane selection and layout plans, and the discriminator is used to optimize the rationality of the generated plans round by round, so that the generated hoisting plan meets the actual construction requirements.
10. A tower crane automatic selection and layout optimization modeling method based on a generative adversarial network according to claim 1, characterized in that In Step 5, the specific methods for optimal plan selection and dynamic optimization include: candidate plan generation and preliminary screening, multi-objective evaluation and optimization. In Step 5, for candidate plan generation and preliminary screening, first, multiple tower crane selection and layout plans output by the generative adversarial network (GAN) model are used, covering different tower crane types, quantities, and layout methods; then, the characteristics and limiting conditions of the construction site are used to screen out candidate plans that meet the basic construction requirements to ensure that the candidate plans meet the tower crane coverage range, hoisting efficiency, and reasonable resource allocation. In Step 5, for multi-objective evaluation and optimization, first, a construction simulation tool is used to perform multi-dimensional evaluation on the candidate plans to verify the performance of the plans in terms of economy, efficiency, and safety, and the hoisting plan is selected in combination with the simulation evaluation results.
Citation Information
Cited By
Engineering cost data calculation management method and system
CN120543220A
Bridge model intelligent construction method and system based on BIM
CN120671480A