A BIM-based smart construction site data processing method

Abstract

The present application discloses a method for processing data of a smart construction site based on BIM, which involves BIM, including: using a BIM model to obtain basic data of the construction site and determine the construction resources to be optimized; comparing the construction site data collected in real time with the basic data obtained in the BIM model and calculating key deviation values; establishing a construction resource optimization model based on the key deviation values; solving the construction resource optimization model using an improved genetic algorithm to obtain an optimal construction resource allocation plan; the improved genetic algorithm includes: generating an initial solution set based on the BIM model; dividing the population into a first zone and a second zone, using a high mutation rate for global search in the first zone, and using an elite retention strategy to accelerate local optimization in the second zone; setting a co-evolution mechanism between the first zone and the second zone; in view of the low resource utilization rate of smart construction sites, the present application integrates data through a BIM model, considers process logic and storage space constraints, and uses an improved dual-zone co-evolution genetic algorithm to improve optimization efficiency.

Description

Technical Field

[0001] This application relates to the BIM field, and in particular to a BIM-based smart construction site data processing method. Background Art

[0002] Hospital projects involve the simultaneous scheduling of common construction resources (such as concrete and steel) and specialized medical resources (such as medical equipment and cleanroom systems). These different types of resources have different optimization objectives and constraints. While common construction resources primarily consider cost and schedule, medical equipment, especially large, sophisticated equipment like MRI machines and CT scanners, places extremely high demands on the installation environment, scheduling, and logistics.

[0003] On the one hand, traditional methods often optimize general construction resources separately from medical equipment resources, resulting in poor overall coordination, frequent process conflicts, and wasted resources. For example, after the main building is completed, it is discovered that the reserved holes do not match large medical equipment, or when precision equipment arrives, the construction environment is found to be substandard. Existing methods are ineffective when dealing with complex constraints such as limited material storage space and the overlapping operations of multiple processes. In particular, they lack effective support for the allocation of construction resources in special areas such as hospital clean rooms and radiation protection zones.

[0004] On the other hand, genetic algorithms, as a heuristic optimization method, have been widely used in construction resource optimization. However, traditional genetic algorithms still have significant shortcomings when dealing with complex resource scheduling problems such as hospital projects: they often use a single population evolution strategy, which cannot simultaneously balance global search and local optimization. When dealing with complex optimization problems with multiple objectives and constraints, they are prone to being trapped in local optimal solutions or converge too slowly. In most applications, genetic algorithms use a random method to generate initial solutions, which does not fully utilize the engineering information in the BIM model. This results in low initial population quality and limited optimization efficiency.

[0005] Therefore, an effective smart construction site data processing method is proposed to improve the construction efficiency and resource utilization of hospital projects. Summary of the Invention

[0006] In response to the low resource utilization rate of smart construction sites, this application provides a BIM-based smart construction site data processing method, which integrates data through the BIM model, considers process logic and storage space constraints, and adopts an improved dual-zone co-evolutionary genetic algorithm to improve optimization efficiency.

[0007] This application provides a BIM-based smart construction site data processing method, comprising: utilizing a BIM model to obtain basic construction site data and determine the construction resources to be optimized; the basic data includes design parameters, design duration, and process dependencies; the construction resources to be optimized include material supply; wherein, design parameters refer to architectural and engineering technical indicators extracted from the BIM model and serve as the basic input data for construction resource optimization. These parameters may include: geometric parameters: spatial information such as the building's size, shape, area, and volume; material parameters: information such as the material type, specifications, and performance requirements of each component; equipment parameters: technical parameters of hospital-specific medical equipment, such as size, weight, installation requirements, and environmental conditions; process dependencies: the order and logical connections between various processes during the construction process, which are fundamental constraints that must be followed in construction resource optimization. For example, these include the competition between multiple processes for the same resource, such as the simultaneous need for tower cranes in multiple areas; the order in which multiple processes are implemented in the same space, particularly the spatial coordination of dense pipeline installation in a hospital; and the dependencies on special process flows such as clean areas and radiation protection zones that are unique to hospitals.

[0008] The construction site data collected in real time is compared with the basic data obtained from the BIM model to calculate the key deviation values, which include construction progress deviation and resource utilization deviation. Based on the key deviation values, a construction resource optimization model is established. The construction resource optimization model aims to minimize the deviation value and sets construction constraints based on the basic data. The improved genetic algorithm is used to solve the construction resource optimization model to obtain the optimal construction resource allocation plan.

[0009] The improved genetic algorithm includes: generating an initial solution set based on the BIM model, which satisfies key construction constraints, including process logic constraints; dividing the population into a first zone and a second zone, using a high mutation rate for global search in the first zone and an elite retention strategy for accelerated local optimization in the second zone; specifically, the initial population can be divided into two parts based on the distribution of fitness values: individuals with lower fitness but higher diversity are assigned to the first zone for global search, while individuals with higher fitness are assigned to the second zone for local optimization. Alternatively, the population can be divided proportionally, with the first zone typically comprising 40% to 60% of the population and the second zone the remainder. A co-evolutionary mechanism is then established between the first and second zones.

[0010] Furthermore, the real-time collected construction site data is compared with the basic data obtained from the BIM model to calculate key deviation values, including: using the Internet of Things settings deployed on the construction site to collect real-time construction site data, the real-time construction site data includes construction progress data and resource consumption data; structuring the real-time construction site data according to the component coding system of the BIM model to associate the spatial position relationship between the real-time construction site data and the basic data of the BIM model; calculating the construction progress deviation value by comparing the actual construction progress with the BIM model design period; and calculating the resource utilization deviation value by comparing material consumption with the BIM model design parameters.

[0011] Furthermore, based on the key deviation values, a construction resource optimization model is established, including: constructing a multi-objective optimization function based on the construction progress deviation value and the resource utilization deviation value; extracting construction constraints based on the basic data of the BIM model, and the construction constraints include: material storage space constraints; among them, material storage space constraints refer to the material storage capacity constraints of the construction site that must be considered in the process of smart construction site resource optimization. Its mathematical expression is: This constraint ensures that the total space occupied by all materials at any time point t does not exceed the total storage space available on the construction site at that time point.

[0012] Furthermore, the multi-objective optimization function is: Among them, PD i Represents the construction progress deviation value of the i-th process, RD j represents the resource utilization deviation value of the jth type of material, w1 and w2 are weight coefficients, and w1+w2=1; the material storage space constraint is: in, represents the storage capacity of the kth type of material at time t, S k Indicates the storage space occupied by the k-th type of material, C t Represents the total amount of storage space available on the construction site at time t.

[0013] Specifically, by taking the progress deviation (PD i ) and resource utilization deviation (RD j ) into the same function, realizing a unified optimization framework for general construction resources and special medical resources, solving the process conflict problem caused by separate optimization in traditional methods. In addition, by introducing the time dimension t, the static space constraint is transformed into a dynamic constraint, enabling the system to handle the characteristics of the site storage capacity changing with the construction stage, avoiding the resource scheduling conflict caused by ignoring the time sequence change in the traditional model. Finally, through S k The parameters distinguish the unit space occupied by different materials, accurately expressing the huge difference in space requirements between ordinary building materials and large medical equipment (such as MRI machines and CT scanners), laying the foundation for subsequent adjustments to the equipment sensitivity level (SE).

[0014] Furthermore, an improved genetic algorithm is used to solve the construction resource optimization model to obtain the optimal construction resource allocation plan, including: encoding the construction resource allocation plan into a chromosome, which contains the material supply time and quantity; constructing a fitness function based on a multi-objective optimization function; generating an initial solution set based on the BIM model as the initial population; dividing the population into a first zone and a second zone: in the first zone, a mutation operation with a mutation rate greater than a threshold ε1 is used for global search, and the mutation operation is to randomly adjust the material supply time and quantity to explore a larger solution space; in the second zone, a mutation rate less than a threshold ε2 and an elite retention strategy are used for local optimization, and P% individuals with the highest fitness in each generation are selected to directly enter the next generation; a co-evolution mechanism is set between the first zone and the second zone, and the co-evolution mechanism balances the exploration and development of the algorithm by periodically exchanging interval individuals and adjusting the fitness of migrated individuals; when the iteration termination condition is met, the optimal individual is output as the optimal construction resource allocation plan.

[0015] Furthermore, the fitness function is: Among them, SE is the equipment sensitivity level. The SE value of large precision equipment such as MRI machines and CT scanners is greater than 1, and the SE value of ordinary medical equipment is equal to 1. α is the weight coefficient; F is the value of the multi-objective optimization function.

[0016] Furthermore, an initial solution set is generated based on the BIM model, including: extracting process dependencies and material demand information from the BIM model, and constructing a process network model in the form of a directed acyclic graph; calculating the type and quantity of materials required for each process based on the construction resources to be optimized, and forming a material demand matrix M; generating an initial material supply plan based on the process network model and the material demand matrix M through heuristic rules, and the initial supply plan includes the supply time point of each batch; by moving the supply time point forward and backward, the initial supply plan is revised to meet the process dependencies; and checking whether the revised supply plan meets the material storage space constraint. If not, the supply plan is revised again until all time points meet the storage space constraint; the revised supply plan is encoded as a chromosome as the initial population.

[0017] In particular, on the one hand, traditional genetic algorithms generally use a completely random method to generate the initial population, without considering the constraints and professional knowledge of the problem domain, resulting in random initialization and low population quality. This application extracts process dependencies and material demand information from the BIM model to construct a process network model in the form of a directed acyclic graph (DAG). The individuals in the initial population have met the key construction constraints, avoiding the generation and evaluation of a large number of invalid solutions in traditional algorithms, and reducing the search space from the theoretical possible set to the actual feasible set for engineering. Through the topological sorting characteristics of the directed acyclic graph, it is ensured that the initial supply plan is reasonably generated under the premise of satisfying the process logic, avoiding a large number of logical contradictions that may be introduced by random search.

[0018] On the other hand, traditional algorithms generally use a single population evolution strategy and a simple penalty function mechanism when dealing with multi-objective and multi-constraint problems. Therefore, they are prone to falling into local optimal solutions, resulting in slow convergence and low optimization efficiency in complex scenarios such as hospital engineering. This application not only improves the quality of the initial solution through structured diversity design and progressive constraint correction methods, but also provides high-quality genetic material for subsequent dual-region co-evolution. At the same time, this method builds a bridge between engineering domain knowledge and computational optimization by converting the engineering semantics in the BIM model into the chromosome encoding of the genetic algorithm, enabling the algorithm to directly use the professional knowledge of hospital engineering to guide the optimization process, fundamentally improving the blindness and inefficiency of traditional genetic algorithms when dealing with complex engineering optimization problems.

[0019] Furthermore, in the first zone, a mutation operation with a mutation rate greater than the threshold ε1 is used for global search. The mutation operation is to randomly adjust the material supply time and quantity to explore a larger solution space, including: setting the mutation rate threshold ε1 = 0.15, and randomly generating the actual mutation rate v in each mutation operation, so that v∈[0.15, 0.3]; for the individuals in the first zone in the initial population, randomly select the material supply time gene bit on the chromosome for mutation, with a mutation probability of v; the material supply time represented by the gene bit is randomly offset within the range allowed by the obtained process dependency, with an offset D∈[-5, 5] time units; the process logic constraint test is performed on the mutated individuals. If the process logic constraint is violated, the earliest feasible time point after the completion of the previous process is found in the process network model as the valid time point, and the supply time is adjusted to the nearest valid time point; the material storage space constraint is performed on the mutated individuals If there is a violation, the material supply at the violation time point is reduced and the supply at the adjacent time points is increased accordingly; using the fitness function Evaluate the fitness value of the individual after mutation, retain the individuals with improved fitness after mutation, and retain the individuals with reduced fitness with a probability of P1% to maintain population diversity.

[0020] In particular, traditional genetic algorithms face an "exploration-exploitation dilemma," meaning the algorithm either focuses on exploring new solution spaces (potentially missing local optimization opportunities) or on exploiting known solutions (prone to falling into local optima). This solution, through a regional differentiation strategy, achieves parallel advancement of exploration and exploitation at the algorithmic level, significantly improving the algorithm's global optimization capabilities. Furthermore, the random offset within the range of [-5, 5] is not blind, but is executed within the allowable range of process dependencies. Constraints are checked and repaired through the process network model, enabling efficient exploration while ensuring solution feasibility and avoiding a large number of ineffective searches. Finally, different correction strategies are adopted for individuals that violate process logic constraints and material storage space constraints, reflecting the refined handling of different types of constraints. This is particularly important in the complex scenario of hospital projects where ordinary building resources coexist with specialized medical equipment.

[0021] Furthermore, in the second zone, a mutation rate less than a threshold ε2 and an elite retention strategy are used for local optimization, and P% individuals with the highest fitness in each generation are selected to directly enter the next generation, including: setting a mutation rate threshold ε2 = 0.05, and randomly generating an actual mutation rate v' in each mutation operation, so that v'∈[0.01, 0.05]; sorting the individuals in the second zone of the initial population from high to low according to their fitness values, and selecting P2% individuals with the highest fitness as elite individuals to directly enter the next generation; for the remaining individuals, the material supply time gene bit on the chromosome is selected for mutation, with a mutation probability of v'; the material supply time represented by the selected gene bit is offset within the range allowed by the process dependency, with an offset D'∈[-2, 2] time units; the process logic constraint check is performed on the mutated individuals, and if there is a violation, the earliest feasible time point after the completion of the previous process is found in the process network model as the valid time point, and the supply time is adjusted to the nearest valid time point; the material storage space constraint is performed on the mutated individuals Check, and if there is a violation, adjust the supply at multiple time points; use the fitness function Evaluate the fitness value of the individual after mutation, retain only the individuals with improved fitness after mutation, and eliminate the individuals with reduced fitness to enhance the efficiency of local optimization.

[0022] In particular, setting the mutation rate threshold ε2 = 0.05 and randomly generating v′∈[0.01, 0.05] yields a significantly lower actual mutation rate than the first region [0.15, 0.3]. This refined mutation control enables the algorithm to deeply explore promising regions of the solution space, avoiding the dilemma inherent in traditional genetic algorithms, which suffer from a single mutation rate: either excessive perturbation destroys high-quality solutions, or insufficient mutation hinders improvement. Furthermore, a mechanism that selects the P2% individuals with the highest fitness to advance directly into the next generation ensures that the algorithm does not lose discovered high-quality solutions, addressing the issue of historically optimal solutions being easily lost in traditional genetic algorithms and providing a stable optimization path for complex constrained problems.

[0023] The first zone employs a high mutation rate (v∈[0.15, 0.3]) and a tolerant selection strategy, focusing on global exploration. The second zone employs a low mutation rate (v′∈[0.01, 0.05]) and a strict selection strategy, focusing on local development. Both zones periodically exchange high-quality and diverse individuals to achieve information sharing and improve optimization efficiency. This dual-zone co-evolutionary strategy addresses the core contradiction of traditional single-population genetic algorithms—the inability to simultaneously balance search space breadth and convergence speed. It is particularly well-suited for complex optimization scenarios with heterogeneous resource allocation requirements, such as hospital engineering. By ensuring that the algorithm maintains sufficient exploration capabilities while further developing discovered high-quality solutions, it ultimately obtains high-quality optimization solutions that are implementable in real-world engineering applications.

[0024] Furthermore, a co-evolution mechanism is set up between the first zone and the second zone. The co-evolution mechanism balances the exploration and development of the algorithm by periodically exchanging interval individuals and adjusting the fitness of migrated individuals, including: setting an interval exchange period T, and exchanging intervals every T generations of iterations, where T is P3% of the number of iterations; calculating the gradient diversity index MD of individuals in the first zone, and selecting the top P4% individuals with the highest MD value as diversity representatives; selecting the top P5% individuals with the highest fitness from the elite individuals in the second zone as high-quality solution representatives; migrating the diversity representatives selected from the first zone to the second zone, and migrating the high-quality solution representative individuals selected from the second zone to the first zone; for individuals migrating from the first zone to the second zone, the random perturbation range of the material supply time of the corresponding individuals is reduced to [-1, 1] to adjust their adaptability; for individuals migrating from the second zone to the first zone, the mutation probability of the corresponding individuals is increased to v+0.1, and the random perturbation range of the material supply time is expanded to [-7, 7] to enhance the exploration ability; after migration, the populations in the two zones are re-evaluated, and the fitness ranking is updated to prepare for the next generation of iteration.

[0025] In particular, this application simulates the phenomenon of gene exchange between populations in different ecological niches during biological evolution, upgrading a single-population genetic algorithm to a multi-population coevolutionary system. This biomimetic approach enables the algorithm to simultaneously maintain both "exploration" and "exploitation" evolutionary pressures, resolving the convergence dilemma caused by a single evolutionary strategy in traditional genetic algorithms. Furthermore, for individuals migrating from the first zone to the second, the perturbation range is reduced to [-1, 1], promoting their transformation from "explorers" to "optimizers." For individuals migrating from the second zone to the first, the mutation probability is increased and the perturbation range is expanded to [-7, 7], stimulating their exploration potential.

[0026] Compared with the existing technology, the advantages of this application are:

[0027] Hospital projects involve the simultaneous scheduling of common construction resources (concrete, rebar) and specialized resources (medical equipment, cleanroom systems). Different resource types have distinct optimization objectives and constraints, and traditional approaches often optimize them separately, leading to overall disharmony. Furthermore, traditional genetic algorithms often employ a single population evolution strategy, failing to balance global search with local optimization. They are prone to falling into local optimal solutions or experiencing slow convergence, and lack consideration for the sensitivity of specialized equipment, resulting in irrational resource allocation and poor optimization results.

[0028] On the one hand, this application integrates multi-source data through the BIM model and constructs a multi-objective optimization function that takes into account material storage space constraints, thereby achieving unified and coordinated optimization of general building resources and special medical resources. On the other hand, through an improved dual-zone co-evolutionary genetic algorithm, the population is divided into a global search zone with a high mutation rate and a local optimization zone for elite retention, and an interval exchange mechanism is set up to achieve a balance between search space and optimization efficiency. At the same time, the introduction of equipment sensitivity parameters to adjust the fitness function enables the algorithm to make more reasonable optimization decisions for special resources such as medical precision equipment, thereby significantly improving overall resource utilization and construction efficiency. BRIEF DESCRIPTION OF THE DRAWINGS

[0029] The present application will be further described in the form of exemplary embodiments, which will be described in detail with reference to the accompanying drawings. These embodiments are not limiting, and in these embodiments, the same numbers represent the same structures, wherein:

[0030] Figure 1 is an exemplary flow chart of a BIM-based smart construction site data processing method according to some embodiments of the present application;

[0031] Figure 2 is an exemplary flow chart of generating an initial population according to some embodiments of the present application;

[0032] Figure 3 is an exemplary flowchart of individual iteration of the first zone according to some embodiments of the present application;

[0033] Figure 4 This is an exemplary flow chart of the second zone individuals according to some embodiments of the present application. DETAILED DESCRIPTION

[0034] The method and system provided in the embodiments of the present application are described in detail below with reference to the accompanying drawings.

[0035] like Figure 1 As shown, the BIM model is used to obtain basic data of the construction site and determine the construction resources to be optimized; the basic data includes design parameters, design construction period and process dependency; the construction resources to be optimized include material supply; the real-time collected construction site data is compared with the basic data obtained from the BIM model to calculate the key deviation value, which includes the construction progress deviation and the resource utilization deviation; according to the key deviation value, a construction resource optimization model is established, the construction resource optimization model aims to minimize the deviation value, and construction constraints are set according to the basic data; the construction resource optimization model is solved using an improved genetic algorithm to obtain the optimal construction resource allocation plan; wherein the improved genetic algorithm includes: generating an initial solution set according to the BIM model, the initial solution set satisfies the key construction constraints, and the key construction constraints include process logic constraints; dividing the population into a first area and a second area, using a high mutation rate for global search in the first area, and using an elite retention strategy to accelerate local optimization in the second area; setting a co-evolution mechanism between the first area and the second area.

[0036] Obtain BIM model files in Revit, ArchiCAD, or Navisworks format from the hospital's engineering design firm. Establish a multi-level coding system based on functional areas, disciplines, and component types, based on the "Building Information Model Classification and Coding Standard." Specific codes are assigned to specialized medical areas such as operating rooms, radiology departments, and nuclear medicine departments. A mapping relationship between medical equipment and BIM model components is established. Geometric parameters are extracted, specifically the actual dimensions and installation space requirements of large medical equipment (such as MRI machines and CT scanners), as well as the location and size parameters of various reserved holes.

[0037] Extract design duration data from BIM model attached documents or project management systems, identify and extract key milestone nodes and time control points, decompose the total duration into sub-durations for different areas, disciplines, and systems, and extract duration constraints for special processes and medical equipment installation.

[0038] Based on the spatial position and physical connection of components, preliminary process dependencies are automatically derived, and engineering experts supplement the logical dependencies that cannot be directly derived from physical relationships. In particular, process dependencies are established between medical equipment and construction projects, and a complete process dependency matrix is generated.

[0039] A multi-level IoT sensing system is deployed on the construction site to collect construction progress and resource consumption data. RFID tags and QR code recognition are used to mark key components and process nodes. Workers scan completion status using handheld terminals, while fixed cameras are deployed in key areas for visual recognition. The system automatically summarizes progress data every four hours, enabling real-time monitoring of key processes such as medical equipment installation.

[0040] Convert the collected heterogeneous data into structured data that can be integrated with the BIM model: Based on the ISO16739 standard, a hierarchical coding structure is constructed, including: project ID - professional code - floor number - component type - component number. For example, "MH001-MEP-03-EQP-0021" represents medical equipment on the third floor.

[0041] The calculation of progress deviation adopts a multi-index comprehensive evaluation method: Calculation of process completion rate (CR): CR i = Number of components actually completed in process i / Number of components that should have completed in process i; Time deviation (TD) calculation: TD i = Actual completion time - planned completion time; Comprehensive calculation formula for schedule deviation (PD): Among them, T max is the maximum acceptable time deviation, w T and w C is the weight coefficient.

[0042] The resource utilization deviation is calculated using a hierarchical classification comparison method: Material Consumption Efficiency (MCE) calculation: MCE j = Actual workload completed / Actual material consumption j ÷ Standard workload / Standard material consumption j. Resource utilization (RU) calculation: RU j = Actual material usage j / design material requirement j×100%. Resource utilization deviation (RD) calculation: RD j =|RU j -100%|×β j , where β j is the material importance coefficient, and the β value of special medical equipment is higher than that of ordinary building materials.

[0043] According to the key deviation values, a construction resource optimization model is established, including: constructing a multi-objective optimization function according to the construction progress deviation value and the resource utilization deviation value; Among them, PD i Represents the construction progress deviation value of the i-th process, RD j represents the resource utilization deviation value of the jth type of material, w1 and w2 are weight coefficients, and satisfy w1+w2=1;

[0044] Based on the basic data of the BIM model, construction constraints are extracted. The construction constraints include: material storage space constraints; the material storage space constraints are: in, represents the storage capacity of the kth type of material at time t, S k Indicates the storage space occupied by the k-th type of material, C t represents the total amount of storage space available at the construction site at time t;

[0045] like Figure 2 As shown, this solution uses hybrid coding technology to transform the construction resource allocation plan into a computable chromosome structure. Each chromosome consists of two parallel sequences: a material supply time series and a corresponding quantity series. The time series uses integer encoding, representing time units relative to the project start; the quantity series uses floating-point encoding, representing the amount of material supplied at each time. In view of the diverse material types found in hospital projects, a hierarchical chromosome structure is adopted, with different material types occupying distinct segments within the chromosome, with separate coding segments specifically designated for precision medical equipment.

[0046] This solution designs an innovative fitness function that takes into account the sensitivity of medical equipment: F represents the value of the multi-objective optimization function, encompassing both schedule and resource deviations. SE represents the equipment sensitivity level, which incorporates the characteristics of medical equipment. α is the adjustment coefficient, controlling the impact of SE on fitness. To address the specificity of hospital projects, an SE parameter is introduced: for large precision equipment such as MRI machines and CT scanners, SE is greater than 1 (typically set to 1.5-2.0); for general medical equipment, SE is equal to 1; for conventional building materials, SE is not explicitly introduced.

[0047] Generate high-quality initial solution sets based on BIM models, data parser based on IFC standards, support IFC2x3 and IFC4 standards, directly extract process sequence relationships from 4D-BIM time information, graph structure: G = (V, E), V = {v1, v2, ..., V n}, represents the process set; E={(v i , v j )}, process v i Must be in process v j Each process node contains key attributes, process ID and name, planned start time and duration, resource type requirement list, process priority and sensitivity tag.

[0048] Extract material parameters (specifications, quantity, performance) from BIM model component attributes, integrate resource consumption quotas in the project management system, and construct a material demand matrix, M = [m ij ] n×k , where m ijDenote the demand of the i-th process for the j-th type of material. n is the total number of processes, and k is the total number of material types. Perform time expansion on the demand matrix, M(t) = [m ij (t)] n×k , which represents the demand of the i-th process for the j-th type of material at time t.

[0049] Set a multi-objective heuristic rule set. Critical path priority rule: Prioritize meeting the material demands of processes on the critical path; balanced supply rule: Avoid spikes in the supply curve and smooth resource allocation; latest supply rule: Supply materials as close as possible to the usage time to reduce storage pressure; batch optimization rule: Determine economically reasonable supply batches according to material characteristics; equipment priority guarantee rule: Prioritize determining the supply plan for precision medical equipment.

[0050] Calculation of process priority: π(v i ) = α1 × CP i + α2 × TO i + α3 × SE i ; CP is the critical path index (0 or 1); TO is the topological order of the process; SE is the equipment sensitivity level; α is the weight coefficient. Generate an initial supply plan. The supply time point t = the start time of the process - the lead coefficient β, where β = f(material type, storage conditions, usage urgency). The batch supply quantity where γ is the material loss coefficient, which is determined according to the material type and project experience.

[0051] Detection of violation of process dependency constraints. For any process v j and its predecessor process v i : If t(supply, j) < t(complete, i), then the process constraint is violated. If the supply time point violates the process dependency: t′(supply, j) = max(t(supply, j), t(complete, i) + safety margin.

[0052] Check for storage space constraints. Ct = baseline capacity ± phased capacity adjustment. Calculate the available storage space at each time point based on the 4D-BIM site layout information. The occupied space at time t = (supply quantity) - ∑(consumption quantity) for all k types of materials up to time t. For those that violate the storage space constraint check, large batches can be split into multiple small batches to disperse the supply pressure.

[0053] Such as Figure 3As shown, the first zone uses a high-intensity mutation strategy. Its purpose is to fully explore the decision space, discover potential high-quality solutions, and prevent the algorithm from falling into local optima. A baseline mutation rate threshold ε1 = 0.15 is set, and a random floating technique is used to generate the actual mutation rate v∈[0.15, 0.3]. When the mutation rate is lower than 0.15, the algorithm is susceptible to being trapped in local optima due to the quality of the initial solution. The upper limit of 0.3 is set to maintain algorithm stability. Exceeding this value will lead to excessive destruction of high-quality genetic features, causing the evolutionary process to degenerate into random search. A dynamic mutation rate value is generated before each mutation to increase population diversity and prevent homogenization of search trajectories.

[0054] For each individual in the first zone, the material supply time locus on the chromosome is selected for mutation with a probability of v. Mutation selection is not completely random; instead, the actual selection probability is adjusted based on the importance of the material to ensure that key medical device loci have a reasonable chance of mutation. By limiting the number of consecutively selected loci, we avoid excessive concentration of mutations in specific chromosome regions and ensure comprehensive coverage of the search space.

[0055] For the selected material supply time gene positions, a random offset of ±5 time units is applied within the allowed range of process dependencies. The offset range of [-5, 5] is the optimal value determined through simulation and analysis of the hospital construction cycle characteristics. It provides sufficient perturbation strength while maintaining solution feasibility.

[0056] The mutated individuals may violate the engineering constraints. The system uses a two-stage repair technology to ensure the feasibility of the solution: by backtracking in the process network model, accurately calculating the earliest feasible time point that satisfies the completion of the previous process, and adjusting the supply time that violates the constraints to this point. At the time point of the material supply, the “peak shaving and valley filling” strategy is adopted to reduce the supply according to the priority of material importance, and intelligently allocate the reduced amount to adjacent time points where no constraints are violated.

[0057] Using fitness function Evaluate the quality of mutated individuals and adopt a probabilistic acceptance mechanism: Individuals with reduced fitness are retained with a probability of P1%. This soft selection mechanism helps the algorithm navigate through suboptimal areas. The P1 value is dynamically adjusted based on population diversity indicators. When population diversity decreases, the acceptance probability is increased; otherwise, it is decreased, achieving an adaptive balance. A P1 value of 20% is preferred.

[0058] like Figure 4As shown, the second zone employs a low-intensity mutation and elite retention strategy, aiming to fine-tune optimization within the space of discovered high-quality solutions and improve the algorithm's convergence efficiency. Individuals in the second zone are sorted by fitness, and the top P2% of individuals are selected to advance directly into the next generation. The P2 value is not fixed but dynamically adjusted based on the algorithm's stage of operation and the concentration of solution quality. It is initially low (approximately 10%) and increases to 20%-30% with each iteration. When selecting elite individuals, not only the absolute fitness value but also the quality differences between individuals are considered to ensure diversity within the elite group.

[0059] For non-elite individuals, a low mutation rate v′∈[0.01, 0.05] is used for fine-grained mutation: a low mutation rate ensures fine-tuning of existing high-quality solutions rather than drastic changes, making it suitable for localized, refined search. The actual mutation rate is adjusted based on the material type, with lower mutation rates for medical precision equipment and relatively higher rates for common building materials.

[0060] Apply a small offset to the selected loci within the range [-2, 2]: Small offsets focus on detailed exploration of the solution space and are suitable for finding the optimal solution within a region of known good solutions. Offset values are generated according to a normal distribution, increasing the probability of small offsets and enhancing the ability to accurately locate the optimal supply time point.

[0061] The mutated solution undergoes detailed constraint verification and repair: This not only checks the completion time of the preceding process but also considers resource preparation time and process requirements to ensure the engineering feasibility of the repaired solution. When a storage constraint is violated, the algorithm not only adjusts a single time point but also analyzes resource allocation across multiple adjacent time points to achieve constraint satisfaction with minimal interference. This dual-zone differentiated evolutionary strategy allows the algorithm to seamlessly integrate global search and local optimization, making it particularly well-suited for resource scheduling optimization problems such as hospital projects, which involve multiple resource types and complex constraints.

[0062] This solution employs a unique dual-zone co-evolution mechanism, breaking through the limitations of traditional genetic algorithms for single-population evolution. This mechanism achieves dual population evolution by establishing a high-mutation global search zone (the first zone) and an elite-retaining local optimization zone (the second zone). Periodic individual migration and adaptive adjustments ensure effective information exchange and complementary characteristics between the two zones.

[0063] This mechanism is based on the "island model" theory in evolutionary computation, but is innovatively adapted to the specific characteristics of hospital engineering resource scheduling, incorporating gradient diversity-driven and regional adaptive transformation techniques. Through this "exploration-exploitation" balancing mechanism, it resolves the contradiction between the vast search space and the high optimization precision required in hospital engineering resource scheduling optimization.

[0064] The interval exchange period T is set to P3% of the total number of iterations (usually 5% to 10%). This dynamic period mechanism is adaptively adjusted based on the problem scale, avoiding the problems of premature convergence or inefficient search that may be caused by a fixed period. A period that is too short will lead to unstable regional characteristics, while a period that is too long will cause excessive regional differentiation. In hospital engineering optimization, a three-level period adjustment strategy is adopted based on project complexity: structural phase: longer period (approximately 10%) to ensure sufficient exploration; decoration phase: medium period (approximately 7%) to balance exploration and utilization; equipment installation phase: shorter period (approximately 5%), focusing on precise optimization.

[0065] The gradient diversity index is the core indicator for measuring the ability of individuals to cover the search space. The calculation formula is: Among them, T ikj represents the supply time of the jth batch of the kth type of material in individual i, is the population mean, w k is the material type weight. The top 4% (usually 5%-8%) individuals with the highest MD values are selected as diversity representatives. While these individuals may not have the highest fitness, they can represent different search directions, providing the algorithm with multiple exploration paths.

[0066] The individuals in the second zone have been selected to be of high quality through elite selection. The top P5% (usually 3% to 5%) of individuals with the highest quality from the second zone are selected as representatives of high-quality solutions. The design of P5% being lower than P4% ensures that only the best solutions are introduced into the first zone, avoiding premature convergence.

[0067] Establish a two-way migration channel to migrate diversity representatives from the first zone to the second zone, and migrate high-quality solution representatives from the second zone to the first zone, forming a dual mechanism of "diversity injection" and "high-quality solution sharing".

[0068] Adaptive adjustments for individuals migrating to the first zone: Reduce the random perturbation range from [-5, 5] to [-1, 1], allowing high-diversity individuals to stabilize after entering the refined optimization zone and conduct local refined search. Adaptive adjustments for individuals migrating to the second zone: Increase the mutation rate to v+0.1 (usually reaching 0.25-0.4) and expand the perturbation range from [-2, 2] to [-7, 7], allowing high-quality solutions in the first zone to serve as the search center, exploring potential better solutions in a wider space.

[0069] After migration is complete, the two populations are fully reassessed, and the fitness rankings are updated. Based on the reassessment results, the selection pressure and mutation parameters of subsequent iterations are dynamically adjusted to accommodate changes in population structure. This co-evolutionary mechanism achieves a dynamic balance between exploration and exploitation in hospital engineering resource optimization through individual migration strategies driven by gradient diversity indicators and post-migration environmental adaptation. It is particularly suitable for co-optimizing common building materials and precision medical equipment.

[0070] The algorithm sets three types of termination conditions: ① reaching the maximum number of iterations (usually 100-200 generations); ② the optimal fitness improvement is less than the threshold ε (usually 0.1%) for n consecutive generations (usually 20 generations); ③ finding a solution that meets the target fitness value (the fitness value is less than the preset target). The iteration is terminated when any of the conditions is met. When the iteration terminates, the Pareto optimal solution set is extracted from the last generation of the population, and a comprehensive scoring mechanism is applied to the individuals in the Pareto optimal solution set to calculate the weighted fitness value. The individual with the highest overall score is selected as the optimal solution. The optimization quality of related processes in special areas of the hospital (such as operating rooms and radiology departments) will be specially considered in the scoring.

[0071] A hierarchical decoding strategy is adopted to decompose the chromosome into three levels: material type layer, time point layer, and supply quantity layer. First, the material type code is identified, then the supply time point of each type of material is analyzed, and finally the specific supply quantity at each time point is determined. The supply time gene bit in the chromosome is decoded into the actual calendar time of the project. Based on the project start date and the work calendar (taking into account holidays, working time restrictions, etc.), the system maps relative time points to absolute dates to ensure that the scheduling plan corresponds to the actual construction schedule. Based on the decoded supply time and quantity, the system intelligently divides each type of material into batches. The economic batch principle is adopted for ordinary building materials, and for special medical equipment, the installation timing requirements are given priority.

[0072] The invention of the present application and its implementation methods are described schematically above. This description is not restrictive. Without departing from the spirit or basic features of the present application, the present application can be implemented in other specific forms. What is shown in the accompanying drawings is only one of the implementation methods of the invention of the present application, and the actual structure is not limited to this. Therefore, if a person of ordinary skill in the art is inspired by it, without departing from the purpose of the invention, a structural method and embodiment similar to the technical solution are designed without creativity, which should all fall within the scope of protection of the present application. In addition, the word "including" does not exclude other elements or steps, and the word "one" before an element does not exclude the inclusion of "multiple" elements. Words such as first and second are used to indicate names and do not indicate any specific order.

Claims

1. A method for processing smart construction site data based on BIM, characterized in that: include: Use BIM models to obtain basic site data and determine construction resources to be optimized; The basic data includes design parameters, design duration and process dependency; The construction resources to be optimized include material supply; wherein the material supply includes equipment supply; Compare the real-time collected construction site data with the basic data obtained from the BIM model to calculate key deviation values, including construction progress deviation and resource utilization deviation; According to the key deviation value, a construction resource optimization model is established, wherein the construction resource optimization model aims to minimize the deviation value and sets construction constraints according to the basic data; The improved genetic algorithm is used to solve the construction resource optimization model and obtain the optimal construction resource allocation plan, including: Encoding the construction resource allocation plan into a chromosome, wherein the chromosome includes the material supply time and quantity; Constructing a fitness function based on a multi-objective optimization function; Generate an initial solution set based on the BIM model as the initial population; Divide the population into the first and second zones: In the first zone, a mutation operation with a mutation rate greater than the threshold ε1 is used for global search; In the second zone, local optimization is performed using a mutation rate less than the threshold ε2 and an elite retention strategy; Setting up a co-evolution mechanism between the first zone and the second zone; When the iteration termination condition is met, the optimal individual is output as the optimal construction resource allocation plan; The fitness function is: Where α is the weight coefficient; SE is the equipment sensitivity level; F is the value of the multi-objective optimization function; the SE value of large precision equipment is greater than that of ordinary medical equipment; Among them, the improved genetic algorithm includes: generating an initial solution set according to the BIM model, wherein the initial solution set satisfies key construction constraints, wherein the key construction constraints include process logic constraints; Set up a co-evolution mechanism between the first zone and the second zone, including: Set the interval exchange period and perform an interval exchange according to the preset ratio of iterations; Select representative individuals from the first zone based on the diversity evaluation index, and select high-quality solution individuals from the second zone based on the fitness value; Perform two-way migration, migrate the representative individuals in the first area to the second area, and migrate the high-quality solution individuals in the second area to the first area; Adaptively adjust the migrating individuals so that the individuals migrating from the first area to the second area can adapt to the fine search environment, and enhance the global exploration ability of the individuals migrating from the second area to the first area; The fitness function is used to evaluate the first and second zones after migration, and the fitness ranking is updated.

2. The BIM-based smart construction site data processing method according to claim 1, characterized in that: Calculate key deviation values, including: Utilizing the IoT deployed at the construction site to collect real-time construction site data, including construction progress data and resource consumption data; Structuring the real-time construction site data according to the component coding system of the BIM model to correlate the spatial position relationship between the real-time construction site data and the basic data of the BIM model; By comparing the actual construction progress with the BIM model design period, the construction progress deviation value is calculated; By comparing material consumption with BIM model design parameters, the resource utilization deviation value is calculated.

3. The BIM-based smart construction site data processing method according to claim 1, characterized in that: Based on the key deviation values, a construction resource optimization model is established, including: According to the construction progress deviation value and resource utilization deviation value, a multi-objective optimization function is constructed; Based on the basic data of the BIM model, construction constraints are extracted, wherein the construction constraints include material storage space constraints.

4. The BIM-based smart construction site data processing method according to claim 3, characterized in that: The multi-objective optimization function is: Among them, PD i Represents the construction progress deviation value of the i-th process, RD j represents the resource utilization deviation value of the jth type of material, w1 and w2 are weight coefficients, and satisfy w1+w2=1; The material storage space constraints are: in, represents the storage capacity of the kth type of material at time t, S k Indicates the storage space occupied by the k-th type of material, C t Represents the total amount of storage space available on the construction site at time t.

5. The BIM-based smart construction site data processing method according to claim 1, characterized in that: Generate an initial solution set based on the BIM model, including: Extract process dependencies and material requirements from the BIM model and construct a process network model in the form of a directed acyclic graph; Based on the construction resources to be optimized, calculate the type and quantity of materials required for each process to form the material demand matrix M; Based on the process network model and the material demand matrix M, an initial material supply plan is generated by heuristic rules, wherein the initial supply plan includes the supply time point of each batch; By moving the supply time points forward or backward, the initial supply plan is modified to meet the process dependencies; Check whether the revised supply plan meets the material storage space constraints If not, the supply plan is revised again until the storage space constraints are met at all time points; The revised supply plan is encoded into chromosomes as the initial population.

6. The BIM-based smart construction site data processing method according to claim 1, characterized in that: In the first zone, a global search is performed using mutation operations with a mutation rate greater than the threshold ε1, including: Set the mutation rate threshold ε1 and generate the actual mutation rate v within the preset range in each mutation operation; Randomly select the material supply time gene position of the individuals in the first zone of the population and mutate it; Randomly offset the supply time of the selected gene position within the allowed range of the obtained process dependency; The process logic constraint check is performed on the mutated individuals. For individuals that violate the constraint, the effective time point is determined through the process network for repair; The material storage space constraint of the mutated individuals is checked. For individuals that violate the constraint, the material supply at the relevant time point is adjusted to balance the individuals. The fitness function is used to evaluate the fitness value of the mutated individuals, retaining the individuals with improved fitness and accepting the individuals with reduced fitness with a preset probability to maintain population diversity.

7. The BIM-based smart construction site data processing method according to claim 1, characterized in that: In the second zone, local optimization is performed using a mutation rate less than a threshold ε2 and an elite retention strategy, including: Set the mutation rate threshold ε2 and generate a mutation rate v′ within the preset range during each mutation operation; Sort the individuals in the second zone of the population by their fitness values, and select the individuals with the highest fitness in a preset proportion as elite individuals to directly enter the next generation; Mutate the gene loci for selecting material supply time for non-elite individuals; Shift the supply time of the selected gene position within the allowed range of process dependencies; The process logic constraint check is performed on the mutated individuals. For individuals that violate the constraint, the effective time point is determined through the process network for repair; The material storage space constraint of the mutated individuals is checked. If the constraint is violated, the material supply at multiple time points is adjusted to achieve balance. The fitness function is used to evaluate the fitness value of the individual after mutation. Only individuals with improved fitness are retained, and individuals with reduced fitness are eliminated to enhance the efficiency of local optimization.

Images