Research on Construction Management Method of Building Floor Components Based on Improved Particle Swarm Optimization Algorithm

Dong Yan^*

Infrastructure Office，Bengbu University, Bengbu 233000, Anhui, China

 ^*Email：yandong930504@163.com

Abstract: In the context of complex structure, high process coupling, and strict spatial constraints in the construction of super high-rise building floor systems, traditional management methods are difficult to adapt to dynamic uncertainty. This paper proposes an improved PSO algorithm based on a dual core driving mechanism to address the problems of dimensionality disaster, premature convergence, and discrete continuous mapping distortion in the construction management of building floor components using particle swarm optimization algorithm. This algorithm achieves precise adaptation to construction scenarios by designing a nonlinear inertia weight decay function based on process urgency and a neighborhood topology dynamic reconstruction strategy guided by resource conflict intensity, combined with a discrete continuous transformation operator. At the same time, a management model covering component level construction information modeling, multi-level construction conflict detection, and algorithm and model coupling interfaces will be constructed to form a complete intelligent management system. Taking the relevant construction data of Shanghai center Building as a sample for experimental verification, the results show that this method performs better in terms of construction period deviation rate and other indicators than the classic PSO, genetic algorithm, critical path method and BIM 5D platform, effectively improving the efficiency, quality and risk resistance of construction management of building floor components, and providing theoretical support and practical path for construction management optimization under the background of intelligent building.

Keywords: improved particle swarm algorithm; Building floor components; Construction management; Dual core driving algorithm

I. Introduction

In the macro context of the deepening of urbanization, as the core carrier of urban space intensive utilization, the construction of complex floor systems in super high-rise buildings is facing systematic difficulties in dynamic coordination mechanisms. This type of floor system usually presents significant characteristics such as diverse structural topology, high coupling degree of construction processes, and strict constraints on working space [2]. In key aspects such as controlling the initial setting time of concrete pouring, ensuring the accuracy of steel bar binding nodes, and optimizing the load transmission path of formwork support, it is highly susceptible to chain reactions caused by spatiotemporal conflicts in process overlap and Pareto improvement failure in resource allocation, resulting in marginal decline in construction efficiency and cumulative effects of quality risks [3]. This reality poses a deep challenge to traditional construction management paradigms.

Traditional schedule management methods often rely on static discrete models such as Critical Path Method (CPM) [4] and Program Evaluation and Review Technique (PERT) [5]. Their parameter matrices have significant solidification characteristics and are difficult to dynamically adapt to uncertain variables such as the bullwhip effect of material supply and random disturbances of meteorological factors during the construction process. In addition, there is a significant time delay in the early warning mechanism for resource conflicts, which often leads to a closed-loop dilemma of problem manifestation and passive adjustment, making it difficult to achieve forward-looking control and systematic optimization of the construction process [6]. At the same time, under the background of industrial transformation with the deep implementation of intelligent construction strategy, the construction industry is undergoing a transition from traditional construction mode to digital construction paradigm [7]. The digital twin and intelligent decision-making of the construction process have become the core demands for improving engineering management efficiency. This requires the construction of an intelligent management method system that can accurately depict the dynamic evolution law of construction and efficiently achieve Pareto optimal resource allocation, in order to adapt to the inherent requirements of high-quality development of construction projects in the new era [8].

From the perspective of the evolution of research at home and abroad, the research paradigm of construction optimization algorithms has gone through an iterative process from traditional mathematical programming methods to intelligent biomimetic algorithms [9]. Early research often used deterministic mathematical methods such as linear programming and integer programming to solve construction optimization problems, such as the widespread application of CPM in schedule planning. However, these methods have inherent drawbacks such as low solution space search efficiency and limited adaptability to constraint conditions when dealing with complex construction systems with multi-objective coupling and multiple constraints interweaving [10]. In recent years, metaheuristic algorithms represented by genetic algorithms and particle swarm optimization (PSO) have been systematically applied and extensively researched in the field of construction optimization due to their powerful global optimization capabilities [11]. Among them, PSO has shown good application prospects in engineering scheduling problems due to its concise principle, fast convergence speed, and strong robustness. It has achieved a series of breakthrough results in research directions such as resource balance optimization, schedule and cost coordination, and construction site layout [12]. However, as shown in Figure 1, existing research has gradually exposed its application boundaries and technical bottlenecks in the field of engineering scheduling. When dealing with high-dimensional and multi peak complex construction optimization problems, PSO is prone to premature convergence due to population diversity attenuation, leading to optimization results falling into local optimal solutions [13]; Meanwhile, the inertia weights, learning factors, and other parameters of the algorithm have high sensitivity, and their performance is significantly affected by parameter combination strategies. The adaptive ability in different construction scenarios needs further improvement [14].

Figure 1 Schematic diagram of existing research limitations

Through in-depth analysis, it can be concluded that there is a significant methodological gap in the modeling of dynamic response mechanisms at the building component level in existing research. The current research on the application of PSO in construction management mainly focuses on multi-objective optimization of project progress, cost, and quality at the overall level. There is insufficient attention paid to the dynamic coupling relationship during the construction process of key components such as floor slabs, and micro parameters such as process logic constraints, spatial resource constraints, and resource consumption characteristics of component construction have not been fully incorporated, resulting in certain mapping deviations between algorithm models and actual construction scenarios [15]. In addition, there are many areas in the existing research that urgently need improvement, such as weight assignment methods for multi-objective optimization and quantitative processing mechanisms for constraint conditions, which cannot meet the practical needs of refined and intelligent construction management of building floor components. Overall, existing research still has significant shortcomings in terms of scene adaptability of construction optimization algorithms, modeling accuracy of component level dynamic response mechanisms, and scientific construction of multi-objective optimization models. It is urgent to develop an improved particle swarm algorithm that can accurately adapt to the construction characteristics of building floor components, in order to break through the theoretical limitations and technical bottlenecks of existing research and provide new theoretical support and methodological guidance for improving the intelligent level of construction management of building floor components.

Based on this, this article focuses on the intelligent optimization of construction management of building floor components, and deeply studies the solution based on improved particle swarm algorithm. The study first analyzed the characteristics of structural topology diversity, high process coupling, and strict spatial constraints in the construction of super high-rise building floor systems, as well as the limitations of traditional management methods in dynamically adapting uncertain variables and resource conflict warning. It pointed out the bottlenecks of PSO in dealing with this scenario, such as dimensionality disaster, premature convergence, and discrete continuous mapping distortion. On this basis, a problem model based on multi constraint analysis was constructed, and a dual core driven algorithm improvement mechanism was proposed, including a nonlinear inertia weight decay function based on process urgency and a neighborhood topology dynamic reconstruction strategy guided by resource conflict intensity. A discrete continuous transformation operator was designed to solve the mapping problem. At the same time, a complete management model covering component level construction information modeling specifications, multi-level construction conflict detection mechanisms, and improved algorithm and construction model coupling interfaces has been established. By taking the construction data of the core tube cantilever floor complex of Shanghai center Building as the sample, the proposed method is compared with the classic PSO, genetic algorithm, critical path method, BIM 5D platform, etc., and its superiority in multiple indicators such as construction period deviation rate, process waiting time, and tower crane idle rate is verified. This article breaks through the theoretical limitations of traditional particle swarm optimization algorithm in the construction management of building floor components, constructs an intelligent management method system that adapts to complex construction scenarios, and enriches the research on algorithm applications and management models in the field of intelligent construction; At the practical level, it provides precise and efficient management solutions for the construction of building floor components, effectively improving the scientific and intelligent level of construction progress control, resource allocation, and conflict resolution, reducing quality risks and cost consumption, and has important guiding value for promoting the transformation of the construction industry from traditional construction models to digital paradigms.

II. Theoretical framework for improving particle swarm optimization algorithm

Modeling of Construction Management Issues for Floor Components

As shown in Figure 2, the problem modeling of floor component construction management needs to be based on multi constraint analysis, covering core elements such as process logic, tower crane coordination, and pouring continuity. The constraints of process logic are reflected in the sequential dependence of each construction process, such as the completion of steel bar binding before template support can be carried out [16]; The collaborative constraint of tower cranes involves the spatial division and scheduling priority of the operating range of multiple tower cranes, and it is necessary to avoid conflicts caused by overlapping operating radii [17]; The continuity constraint of pouring requires the time connection and strength maintenance of the concrete pouring process to prevent cold joints from affecting the structural performance. On this basis, a mathematical description of four-dimensional spatiotemporal conflict detection is constructed, quantifying the parameters of time, space, process, and resource dimensions into multidimensional vectors, and identifying potential conflicts through operations between vectors. At the same time, with the goal of minimizing resource consumption intensity and minimizing project duration, a Pareto optimal objective function is constructed for both. By introducing weight coefficients to balance the priority of the dual objectives, a complete problem model framework is formed.

2. Engineering adaptation bottleneck of classic PSO algorithm

Although PSO algorithm has shown certain advantages in many optimization fields, there are significant engineering adaptation bottlenecks in the specific engineering scenario of floor component construction management, which seriously restrict its optimization effect and application scope.

In the scenario of large-scale floor slab scheduling, the dimensional disaster problem exhibits obvious emergent characteristics and becomes the primary obstacle affecting algorithm efficiency [18]. With the increase in the number of floor components and the complexity of each construction process, the variable dimensions required for algorithm optimization have sharply increased. The construction of each floor component involves multiple processes, each of which is associated with multiple parameters such as time, resources, and space. This exponentially expands the search space for particles. In this high-dimensional space, the search efficiency of particles decreases significantly, and a large number of search steps are wasted in invalid areas, making it difficult for algorithms to find satisfactory optimization solutions in a reasonable time, and even leading to the phenomenon of search process stagnation. For example, when a floor slab group contains dozens or even hundreds of components, the dimensions of optimization variables can reach hundreds or even thousands of dimensions. The particles of classical PSO algorithm often exhibit blindness in such a large space, making it difficult to focus on valuable search areas.

The mechanism of premature convergence of particles and the formation of local optimal traps is another important bottleneck faced by classical PSO algorithms, which is rooted in the premature loss of population diversity [19]. In the process of algorithm iteration, particles adjust their motion direction by learning from their own historical optimal position and the population’s historical optimal position. This learning mechanism can accelerate the convergence speed of the algorithm in the early stage, but as the number of iterations increases, when a local optimal solution is recognized and followed by the majority of particles, and the solution is not the global optimal solution, particles will excessively cluster in that local area, leading to a rapid decrease in population diversity. At this point, the algorithm loses the ability to explore other potential better solutions, thus falling into the trap of local optima. In the construction management of floor components, due to the complexity and multiple constraints of the construction scene, there are numerous local optimal solutions, and there may be a significant gap between these local optimal solutions and the global optimal solution. This makes it easy for the classical PSO algorithm to converge prematurely during the iteration process, making it difficult to find the true global optimal construction plan.

In addition, the mapping distortion problem between discrete process space and continuous algorithm space further exacerbates the engineering adaptation challenge of classical PSO algorithm [20]. The construction process of floor components has significant discreteness characteristics, with each process being an independent and inseparable construction link, and the transition between processes being skip like rather than continuous. However, the classic PSO algorithm is designed based on continuous space, where the position and velocity of particles are continuous variables, and the motion mode of particles is smooth movement in continuous space. There is a fundamental difference between the discrete actual construction process and the continuous algorithm space, which leads to distortion when mapping the construction process to the algorithm space. For example, the start and end times of a process must be discrete integer moments, while the particle positions generated by the algorithm may be continuous real numbers, which requires complex rounding or conversion processing. This processing often loses a lot of effective information, reduces the accuracy of the algorithm in representing the actual construction process, and makes it difficult to accurately reflect the dynamic changes and constraint relationships of the construction process.

Figure 2 Schematic diagram of theoretical architecture

3. Dual core driving mechanism for algorithm improvement

Aiming at the bottleneck of the classic PSO algorithm, the improved algorithm adopts a dual core driving mechanism to achieve performance improvement. Firstly, design a nonlinear inertia weight decay function based on the urgency of the process [21], dynamically adjust the inertia weight according to the time sensitivity of the process, and enhance the algorithm’s optimization ability in different construction stages. Processes with high urgency correspond to larger weights to accelerate convergence speed.

Specifically, the process time margin entropy ​​TE_i in the construction network plan is introduced as a moderating variable:

（1）

The time margin entropy is constructed by the total floating time TF_i and the criticality coefficient α:

（2）

Where R is the resource conflict intensity function and λ is the decay rate control factor.

Secondly, a neighborhood topology dynamic reconstruction strategy guided by resource conflict intensity is proposed, which adjusts the information exchange network between particles based on real-time resource conflict data, and enables particles to form a closer collaborative search pattern in the conflict area. At the same time, a conversion operator between process discretization encoding and continuous particle position is constructed to achieve precise docking between discrete construction processes and continuous algorithm space through encoding mapping, solving the problem of mapping distortion.

Establish a construction conflict intensity matrix SCM∈R^n×n for dynamic bottlenecks:

（3）

Among them, δ is the Boolean conflict function, and β is the weight coefficient of the conflict type. In the initial stage, the von Neumann topology is used to ensure global exploration capability. When ||SCM||_F>ξ conflict detection is triggered, high conflict process nodes (diag (SCM)>η) are aggregated into subgroups, and switching to a star topology within the subgroups achieves local fine optimization. The optimal solution transfer between subgroups is achieved through pheromone diffusion mechanism.

Design a dual layer position mapping mechanism for engineering constraint coding, and express the continuous spatial particle position vector as follows:

（4）

The expression for the priority sequence of discrete process space processes is as follows:

（5）

Conversion operator design:

（6）

The resource correction matrix M_resource is generated by the tower crane service radius constraint:

（7）

Where Δ is the feasible solution bias compensation amount.

III. Construction management model for building floor components

1. Component level Construction Information Modeling Specification

The core objective of the Component Level Construction Information Modeling Specification is to ensure the accuracy, completeness, and reusability of construction information throughout its entire lifecycle through unified standards and methods. The extraction and semantic enhancement of BIM model component attributes are the primary steps in implementing this specification, with a focus on efficiently extracting basic attributes such as geometric parameters, material properties, and mechanical properties of components from BIM models. At the same time, these attributes are expanded and deepened at the semantic level by combining professional knowledge in the field of architecture [22].

As shown in Figure 3, based on the semantic enhancement of the BIM model, the construction process library’s process level knowledge graph is constructed as a key link between information and knowledge. This work aims to systematically sort and integrate construction process knowledge scattered in construction manuals, technical specifications, and expert experience, with construction processes as the core nodes, to construct a knowledge graph covering multi-dimensional information such as logical dependencies between processes, resource requirement lists, quality control point parameters, safety operating procedures, etc. Through the structured expression of knowledge graphs, efficient correlation queries and intelligent reasoning of construction process knowledge are achieved, providing strong knowledge driven support for the intelligent generation, optimization and adjustment of construction plans, as well as on-site technical disclosure.

Figure 3 Schematic diagram of knowledge graph architecture

The feature vectorization processing of historical construction data is an important means to achieve data value mining in the component level information modeling specification. By applying data mining algorithms and feature engineering techniques, the massive historical construction data shown in Table 1 is cleaned, transformed, and refined into feature vectors that can characterize key elements such as construction efficiency indicators, cost consumption patterns, and quality risk probabilities. These feature vectors can not only serve as input parameters for machine learning models for predictive analysis of construction processes, but also provide data-driven quantitative basis for optimization decisions of construction plans, thus achieving the transformation of construction management from experience driven to data-driven.

Table 1 Feature Engineering Framework

​​Deep feature extraction uses TCN temporal convolutional network to capture construction disturbance patterns, expressed as follows:

（8）

Among them, Dₕᵢₛₜ is the historical data tensor, and * represents the convolution operation.

2. Multi level construction conflict detection mechanism

The multi-level construction conflict detection mechanism is a key technical system to ensure the orderly progress of the construction process, reduce rework and delays. Its core function is to comprehensively identify various problems that may occur during the construction process, such as time conflicts, spatial interference, and process contradictions. The application of spatiotemporal cube theory in process collision detection provides innovative ideas for solving spatiotemporal conflicts between processes. This method integrates the time interval and spatial range of each construction process into a four-dimensional spatiotemporal cube, and uses spatial intersection calculation and time overlap analysis between cubes to accurately detect whether there is temporal and spatial overlap collision between different processes during the construction process. Construct a discrete spatiotemporal coordinate system (x, y, z, t) and abstract the process as a hypercube:

（9）

This visual detection method can intuitively present the unreasonable aspects in the process arrangement, providing clear basis for optimizing and adjusting the construction schedule plan.

The collision simulation technology of the motion envelope in the interference area of tower cranes has become an important component of conflict detection for high-risk tower crane operations in large-scale construction sites.

The D-H parameter table for establishing the joint coordinate system of the tower crane through kinematic modeling is shown in Table 2.

Table 2 D-H parameter table

Envelope generation based on Monte Carlo method to solve the reachable space of the hook:

（10）

​​Establish a safe distance field between tower cranes through dynamic collision avoidance:

（11）

This technology accurately simulates the motion trajectory and coverage range of tower cranes at different operating radii and lifting heights by establishing a three-dimensional motion envelope model of the tower crane, and then conducts simulation analysis of potential collision risks between multiple tower cranes, as well as between tower cranes and buildings and construction machinery. Through this rehearsal, it is possible to plan the operating path and avoidance strategy of the tower crane in advance, greatly reducing the probability of on-site collision accidents.

The risk warning model for interruption of concrete pouring continuity plays an irreplaceable role in ensuring the quality of key processes.

The interruption probability model adopts Weibull survival analysis [23]:

（12）

This model constructs a multi factor coupled risk assessment model by real-time collection of key parameters that affect the continuity of pouring, such as concrete supply speed, pouring equipment operation status, on-site weather changes, and personnel configuration. When a parameter exceeds the normal range and may cause pouring interruption, the model can automatically issue a warning signal and provide corresponding emergency response suggestions, effectively avoiding quality problems such as cold joints and insufficient strength caused by pouring interruption, and ensuring the construction quality of concrete structures.

3. Improve the coupling interface between algorithms and construction models

Improving the coupling interface between algorithms and construction models is the core technology bridge for achieving intelligent optimization of the construction process. Its core goal is to establish an efficient and accurate communication mechanism and data exchange channel between algorithms and models, so that the optimization capabilities of algorithms can seamlessly connect with the actual construction needs. The compilation rule from particle position vectors to process priority sequences is the fundamental link of this interface. It defines a rigorous set of mapping rules to transform the position vectors of particles in particle swarm optimization algorithms into process priority sequences with practical construction significance. This transformation not only realizes the mathematical mapping between algorithm variables and construction problems, but also ensures that the optimization results of the algorithm can directly guide the formulation of construction plans.

Considering the inevitable uncertainty in the construction process, the particle swarm reinitialization strategy driven by construction disturbance events has become a key mechanism to ensure the adaptability of the algorithm. When there are sudden equipment failures, temporary design changes, material supply delays, and other disturbances on the construction site, this strategy can automatically trigger the initialization process of the particle swarm algorithm, and regenerate the initial particle swarm based on new construction constraints and objective functions. This dynamic adjustment mechanism enables the algorithm to quickly adapt to changes in the construction environment, ensuring that the optimization results are always consistent with actual construction needs.

In order to achieve optimal allocation of resources, the adaptive value calculation mechanism for real-time feedback of resource load balancing degree is incorporated into the design of coupling interfaces. This mechanism takes resource load balancing as an important evaluation indicator for algorithm fitness value calculation, and dynamically adjusts the calculation weight of fitness value by collecting real-time load status data of various resources during the construction process. When a certain type of resource is overloaded or idle, the fitness value will be adjusted accordingly, guiding the algorithm to pay more attention to the balanced utilization of resources in the optimization process. In this way, algorithm optimization results can not only meet basic goals such as minimizing construction period, but also achieve reasonable allocation of resources, thereby improving construction efficiency while reducing construction costs.

IV. Experimental design

1. Dataset

This study uses the construction data of the core tube cantilever floor complex of Shanghai center Building to build a basic data set. The raw data covers 18 standard floor units (including 36 irregular components), with a total of 8460 construction task instances. The data dimensions include: ①Component level geometric attributes derived from IFC4.0 files exported from Revit models; ②The historical construction records include 48 types of time-series data such as tower crane operation trajectories and concrete tanker dispatch logs; ③The on-site monitoring system collects physical field information such as temperature, humidity, vibration frequency, etc. In the data preprocessing stage, sliding window normalization and DBSCAN density clustering are used to eliminate outliers, ultimately generating a structured dataset with spatiotemporal labels. The training and testing sets are divided in a 7:3 ratio, and the disturbance event data during Typhoon Lekima in 2020 is retained as a special validation set.

2. Comparison plan

The specific comparison schemes selected in this article are shown in Table 3. This study adopts dynamic inertia weight mechanism, topology reconstruction, and discrete continuous mapping in the core mechanism of the algorithm. The value of ω in the parameter configuration has a range and includes ξ and Δ. The constraints are processed through real-time resource correction matrix. In the face of disturbances, hierarchical reinitialization and ω mutation response strategies are adopted. TCN feature extraction and sliding window normalization are used for data processing, making it more dynamically adaptable and targeted overall. Classic PSO relies on fixed inertia weights and static ring topology, with fixed parameters and simple constraint handling. When disturbed, the entire population is randomly reset, and the data is standardized using Z-score, which is relatively traditional. GA focuses on tournament selection, OX crossover, and exchange mutation, with parameters including crossover probability, mutation probability, and population size. Constraints are handled using penalty functions, and elites are retained during perturbations. Data is discretized using equal width bins, with a focus on probability based evolutionary search [24]. CPM is based on the critical path method and resource balancing heuristic rules, with parameters involving floating time thresholds and tower crane spacing. Constraints are adjusted through Gantt charts and negotiated on-site, and emergency meetings are used to respond to disturbances. The data relies on manual experience screening, which heavily relies on manpower and experience [25]. The BIM 5D platform utilizes rule engine conflict detection and 4D simulation deduction, with parameters focusing on collision accuracy and warning delay. It directly analyzes IFC processing constraints, pauses and manually recalculates when disturbed, and aggregates data through SQL queries, emphasizing the application of digital platforms [26].

Table 3 Comparison Scheme

3. Experimental environment

The experimental environment of this article is shown in Table 4.

Table 4 Experimental Environment

4. Evaluation indicators

The key evaluation indicators selected in this article are shown in Table 5.

Table 5 Evaluation Indicators

V. Results

1. Performance comparison

The proposed solution in this article demonstrates significant advantages in multiple key dimensions. From the perspective of δ_T, the deviation of this study is -4.2, which is significantly better than classical PSO, genetic algorithm, critical path method, and BIM 5D platform. This smaller negative deviation indicates that the actual construction period is closer to the planned schedule, reflecting a more accurate schedule planning ability. In terms of ∑t_wait, the value recorded in this study is 38.6, far lower than classical PSO and others. The reduction of cumulative downtime means that this study can effectively alleviate resource conflicts during process handover, achieve smoother process transitions, and improve overall efficiency. The situation of η idle is similar, with an idle rate of 13.8 in this study, which is much lower than other methods. This indicates that the use of tower cranes in this study is more efficient, minimizing ineffective time and fully utilizing their role in the construction process. Hres is used to measure the balance of resource usage distribution, and the value in this study is 1.24, which is the lowest among all methods. A lower entropy value means a more balanced distribution of resource usage duration, indicating outstanding ability in stable and efficient resource allocation.

τ_resp is a key indicator for measuring adaptability, and this study has shown outstanding performance. This rapid response to emergencies reflects the robustness and flexibility of this study in the dynamic construction environment. Zeta is 0.982, which is the highest among all comparison methods. This high conflict resolution indicates that this study has a significant effect in avoiding potential construction conflicts and reducing the risk of delays and interruptions. The value of k is 0.037, which is the lowest among all values. A lower index reflects better concrete quality, smaller strength deviation, and fewer cracks.

Finally, the value of ν in this study is 1.63, indicating that it can achieve faster and more effective resumption of work after a typhoon, demonstrating its ability to withstand risks under adverse weather conditions.

The sustained excellent performance of this study on all indicators indicates that its comprehensive approach, which integrates dynamic mechanisms, efficient resource allocation, and rapid adaptability, is more effective in addressing the complex challenges of construction management than traditional and other alternative methods. Lower deviation, less waiting time, higher resource utilization, faster disturbance response, better conflict resolution, better quality control, and stronger recovery capability all highlight the transformative potential of this study in optimizing construction processes.

Table 6 Comprehensive Comparison Results of Construction Management Optimization Plans

2. Ablation experiment

The results of the ablation experiment are shown in Table 7. The experimental results clearly reveal the incremental contribution of each mechanism to the overall performance, demonstrating how specific components enhance the effectiveness of the algorithm. There is significant room for improvement in performance indicators based on the benchmark PSO. After adding the dynamic inertia weight mechanism, all indicators showed measurable improvement. The deviation rate of the construction period has decreased to -8.3, reflecting a higher degree of conformity between the actual and planned construction periods; The conflict resolution degree has increased to 0.814, indicating a better conflict avoidance effect; The response time to disturbances has been shortened to 12.7, reflecting a faster adaptation speed to disturbances; The decrease in resource fluctuation entropy to 2.05 indicates a more balanced allocation of resources; The idle rate of tower cranes has dropped to 23.6, indicating an improvement in the utilization rate of tower cranes. After introducing the topology reconstruction strategy, the performance of all indicators was further improved. The more significant improvement here indicates that reconstructing the communication topology between particles can achieve better information sharing and collaboration, making the algorithm more effective in escaping local optima. This brings about more robust schedule planning decisions, reduces conflicts, shortens response times, and balances resource utilization. After adding the discrete continuous transformation operator, it reduces the inefficiency caused by the mismatch between algorithm results and real-world constraints by converting abstract particle positions into executable sequences.

Table 7 Ablation Experiment

VI. Conclusion

This study addresses the complexity and challenges in the construction management of floor components in super high-rise buildings. A PSO based intelligent management method system is constructed, and through theoretical modeling, mechanism innovation, and empirical testing, the following conclusions are drawn: (1) Classic PSO has bottlenecks such as dimensional disaster, premature convergence, and discrete continuous mapping distortion in high-dimensional construction scenarios, making it difficult to adapt to the multi constraint coupling characteristics of floor component construction. However, by introducing a dual core driving mechanism and discrete continuous transformation operator, the above limitations can be effectively overcome, and the algorithm’s global optimization ability and scene adaptability can be improved. (2) The constructed component level construction information modeling specification, multi-level conflict detection mechanism, and algorithm model coupling interface form a complete management chain covering information perception, conflict recognition, and intelligent optimization, achieving dynamic control of the construction process. (3) The experimental verification with the construction data of Shanghai center Building as the sample shows that, compared with the classic PSO, genetic algorithm, critical path method and BIM 5D platform, the method proposed in this study performs optimally in the core indicators such as construction period deviation rate, process waiting time, tower crane idle rate, resource fluctuation entropy, disturbance response time, and fully verifies its remarkable effectiveness in improving construction efficiency, reducing quality risk, and enhancing anti-interference capability.

Based on the research results, the following suggestions are proposed for the construction management practice of building floor components: (1) Promote the engineering implementation of improved algorithms. In complex floor construction such as super high-rise and large-span, the improved PSO algorithm proposed in this study can be prioritized, combined with BIM platform to build an intelligent scheduling system, focusing on optimizing key links such as tower crane coordination and concrete pouring continuity, reducing resource waste and project delay risks. (2) Strengthening the standardized collection and deep mining of construction data, it is recommended to establish a database covering dimensions such as component geometric attributes, process parameters, resource consumption, and environmental monitoring. Hidden features can be extracted through techniques such as temporal convolutional networks to provide high-quality data support for algorithm optimization. (3) Improve conflict warning and emergency response mechanisms, based on a multi-level conflict detection model, conduct spatiotemporal conflict rehearsals before construction, monitor risks such as tower crane interference and pouring interruption in real time during construction, and develop emergency plans in combination with disturbance response strategies to enhance the resilience of the construction system.

Reference

• Ptuhina, I., Spiridonova, T., Musorina, T., Kanyukova, S., & Rezvaia, A. (2016). Efficiency of urban development under high-rise construction of districts. In MATEC Web of Conferences (Vol. 53, p. 01049). EDP Sciences.
• Aldwaik, M., & Adeli, H. (2014). Advances in optimization of highrise building structures. Structural and Multidisciplinary Optimization, 50(6), 899-919.
• Ruan, R., Lai, M., Jiang, C., Wang, J., & Lin, Y. (2023). Integral lifting of steel structure corridor between two super high-rise buildings under wind load. Buildings, 13(10), 2441.
• Ariza, F., Suroso, A., & Amin, M. (2025). Analysis of the Critical Chain Project Management (CCPM) method, BIM 3D, and Last Planner System on Contractor Performance and High Rise Building Project Duration. Eduvest-Journal of Universal Studies, 5(3), 2716-2730.
• Husin, A. E., Prawina, R. S., Priyawan, P., Pangestu, R., Kussumardianadewi, B. D., Sinaga, L., & Kristiyanto, K. (2023). Optimizing time performance in implementing green retrofitting on high-rise residential by using system dynamics and M-Pert. Civil Engineering Journal, 9(12), 3060-3074.
• Chen, C. X., Liang, J., Yang, S., & Zhu, J. (2024). The bullwhip effect, demand uncertainty, and cost structure. Contemporary Accounting Research, 41(1), 195-225.
• Volkov, A. (2018). Digital models in construction: Classification paradigm. In MATEC Web of Conferences (Vol. 251, p. 03066). EDP Sciences.
• Hyun Park, S., Seon Shin, W., Hyun Park, Y., & Lee, Y. (2017). Building a new culture for quality management in the era of the Fourth Industrial Revolution. Total Quality Management & Business Excellence, 28(9-10), 934-945.
• Li, H., Liao, B., Li, J., & Li, S. (2024). A Survey on Biomimetic and Intelligent Algorithms with Applications. Biomimetics, 9(8), 453.
• Aboelmagd, Y. M. (2018). Linear programming applications in construction sites. Alexandria Engineering Journal, 57(4), 4177-4187.
• Hu, W., Zhang, Y., Liu, L., Zhang, P., Qin, J., & Nie, B. (2024). Study on Multi-Objective optimization of construction project based on improved genetic algorithm and particle swarm optimization. Processes, 12(8), 1737.
• Pereira, C. M., Lapa, C. M., Mol, A. C., & Da Luz, A. F. (2010). A particle swarm optimization (PSO) approach for non-periodic preventive maintenance scheduling programming. Progress in Nuclear Energy, 52(8), 710-714.
• Jia, D., Zheng, G., Qu, B., & Khan, M. K. (2011). A hybrid particle swarm optimization algorithm for high-dimensional problems. Computers & Industrial Engineering, 61(4), 1117-1122.
• Liu, P., Xie, M., Bian, J., Li, H., & Song, L. (2020). A hybrid PSO–SVM model based on safety risk prediction for the design process in metro station construction. International journal of environmental research and public health, 17(5), 1714.
• Chau, K. W. (2007). Application of a PSO-based neural network in analysis of outcomes of construction claims. Automation in construction, 16(5), 642-646.
• Tommelein, I. D., & Weissenberger, M. (1999, July). More just-in-time: Location of buffers in structural steel supply and construction processes. In Proceedings IGLC (Vol. 7, p. 109).
• Yin, J., Li, J., Yang, A., & Cai, S. (2024). Optimization of service scheduling problem for overlapping tower cranes with cooperative coevolutionary genetic algorithm. Engineering, Construction and Architectural Management, 31(3), 1348-1369.
• Zou, Z., Zeng, H., Zheng, X., & Chen, J. (2025). Research on Multi-Center Path Optimization for Emergency Events Based on an Improved Particle Swarm Optimization Algorithm. Mathematics, 13(4), 654.
• Rahman, C. M., & Rashid, T. A. (2019). Dragonfly algorithm and its applications in applied science survey. Computational Intelligence and Neuroscience, 2019(1), 9293617.
• Zhang, Y., Wang, S., & Ji, G. (2015). A comprehensive survey on particle swarm optimization algorithm and its applications. Mathematical problems in engineering, 2015(1), 931256.
• Xu, B., Deng, B. C., Li, J., & He, J. (2019). Structural nonlinearity and mass identification with a nonparametric model using limited acceleration measurements. Advances in Structural Engineering, 22(4), 1018-1031.
• Hu, Z. Z., Leng, S., Lin, J. R., Li, S. W., & Xiao, Y. Q. (2022). Knowledge extraction and discovery based on BIM: a critical review and future directions. Archives of Computational Methods in Engineering, 29(1), 335-356.
• Erik Bradley, P., & Kohler, N. (2007). Methodology for the survival analysis of urban building stocks. Building Research & Information, 35(5), 529-542.
• Li, T., Shao, G., Zuo, W., & Huang, S. (2017, February). Genetic algorithm for building optimization: State-of-the-art survey. In Proceedings of the 9th international conference on machine learning and computing (pp. 205-210).
• Olivieri, H., Seppänen, O., & Denis Granja, A. (2018). Improving workflow and resource usage in construction schedules through location-based management system (LBMS). Construction management and economics, 36(2), 109-124.
• Cheng, J. C., Tan, Y., Song, Y., Liu, X., & Wang, X. (2017). A semi-automated approach to generate 4D/5D BIM models for evaluating different offshore oil and gas platform decommissioning options. Visualization in Engineering, 5(1), 12.