A digital management method and system for the entire engineering construction process

Abstract

This application relates to the field of data processing technology and discloses a method and system for digital management of the entire engineering construction process. The method includes: collecting and standardizing multi-source data throughout the entire engineering construction cycle; constructing a stacked integrated prediction system using the standardized data to obtain predicted values ​​for three-dimensional indicators; allocating resources based on the predicted values ​​and an ergonomic risk index to obtain a configuration plan; quantifying the uncertainty factors of the configuration plan; calculating the optimal solution set through a multi-objective symbiotic algorithm to form a management strategy. This application achieves high-precision prediction of three-dimensional indicators for engineering projects through a stacked integrated algorithm, combining ergonomic risk assessment with resource allocation optimization, and solving the technical shortcomings of traditional methods that neglect human factors engineering.

Description

Technical Field

[0001] This application relates to the field of data processing technology, and in particular to a digital management method and system for the entire process of engineering construction. Background Technology

[0002] With the development of the construction industry, engineering construction management has gradually shifted from traditional paper-based document management to digitalization. Currently, most engineering construction projects utilize some digital management tools, such as CAD design software, BIM modeling technology, and project management systems, enabling digital processing of specific stages such as design, construction, and supervision. The application of these technologies has, to some extent, improved the efficiency of engineering construction, reduced human error, and promoted information sharing and collaboration.

[0003] However, existing digital management methods for engineering construction have significant limitations. First, most digital tools are only designed for a specific stage or link in the construction process, leading to severe information silos and hindering effective data flow and sharing between systems. Second, the lack of a unified data standard and management platform throughout the entire project lifecycle makes it difficult to manage and trace data from planning and design to final acceptance in an integrated manner. Third, existing systems have limited real-time response capabilities for engineering changes, cost control, and progress monitoring, failing to meet the needs for rapid decision-making during construction. Furthermore, traditional digital tools typically focus on static data management, lacking adequate real-time monitoring and early warning mechanisms for dynamic construction processes.

[0004] To address the issue of information fragmentation during engineering construction, particularly in areas such as standardized engineering data collection, real-time multi-party collaboration and interaction, data security and integrity assurance, and intelligent decision support in complex scenarios, there is a lack of systematic technical approaches and effective tools. Furthermore, the absence of an adaptive framework capable of real-time integration of stacked predictive algorithms, ergonomic risk assessment, and multi-objective symbiotic optimization algorithms prevents the automatic adjustment of decision parameters and re-optimization of subsequent construction phases when project deviations occur. Summary of the Invention

[0005] This application provides a digital management method and system for the entire engineering construction process. It achieves high-precision prediction of three-dimensional engineering indicators through a stacked ensemble algorithm, combining ergonomic risk assessment with resource allocation optimization, thus overcoming the technical shortcomings of traditional methods that neglect human factors engineering. Furthermore, it addresses engineering uncertainties through a multi-objective symbiotic biological search algorithm, overcoming the challenge of existing technologies failing to provide optimal decision-making solutions under multi-level uncertainties.

[0006] Firstly, this application provides a digital management method for the entire process of engineering construction. This method includes: collecting multi-source heterogeneous data from the entire lifecycle of engineering construction and performing standardized processing to obtain a standardized engineering dataset; constructing a stacked integrated prediction system using the standardized engineering dataset, integrating the outputs of linear regression, support vector machines, and neural networks through a gradient boosting decision tree to obtain predicted values ​​for three-dimensional indicators of engineering time, cost, and quality; allocating engineering resources based on the predicted values ​​of the three-dimensional indicators, combined with an engineering task ergonomic risk index calculated using the occupational repetitive action method, to obtain a resource allocation scheme; quantifying engineering uncertainty factors based on the resource allocation scheme, calculating a set of schemes under multi-level uncertainty using a multi-objective symbiotic biological search algorithm, to obtain an engineering construction management strategy.

[0007] Secondly, this application provides a digital management system for the entire process of engineering construction, the digital management system for the entire process of engineering construction comprising:

[0008] The processing module is used to collect multi-source heterogeneous data throughout the entire life cycle of engineering construction and perform standardization processing to obtain a standardized engineering dataset.

[0009] The integration module is used to construct a stacked integrated prediction system using the standardized engineering dataset, and integrates the output results of linear regression, support vector machine and neural network through gradient boosting decision tree to obtain the predicted values ​​of the three-dimensional indicators of engineering time-cost-quality.

[0010] The allocation module is used to allocate engineering resources based on the predicted values ​​of the three-dimensional indicators, combined with the engineering task ergonomic risk index calculated by the occupational repetitive action method, and to obtain a resource allocation scheme.

[0011] The quantification module is used to quantify the uncertainties of the project based on the resource allocation scheme, and to calculate the set of schemes under multi-level uncertainties through a multi-objective symbiotic biological search algorithm to obtain the project construction management strategy.

[0012] Thirdly, a digital management device for the entire process of engineering construction is provided, comprising: a memory and at least one processor, wherein the memory stores instructions; the at least one processor invokes the instructions in the memory to cause the digital management device for the entire process of engineering construction to execute the aforementioned digital management method for the entire process of engineering construction.

[0013] Fourthly, a computer-readable storage medium is provided, wherein instructions are stored in the computer-readable storage medium, which, when executed on a computer, cause the computer to perform the aforementioned digital management method for the entire process of engineering construction.

[0014] The technical solution provided in this application solves the technical problems of data fragmentation and inconsistent formats in traditional engineering management by collecting multi-source heterogeneous data throughout the entire lifecycle of engineering construction and performing standardized processing. This enables data from different sources such as BIM models, cost estimation systems, progress monitoring, and quality inspection to be integrated and analyzed on a unified platform, providing a data foundation for subsequent intelligent decision-making. The establishment of standardized engineering datasets achieves spatiotemporal consistency and semantic interoperability of engineering information, significantly reducing information loss and errors during data conversion and integration. Based on this, this invention utilizes a stacked integrated prediction system that integrates three complementary basic models: linear regression, support vector machines, and neural networks. Gradient boosting decision tree algorithms are used as meta-learners to integrate the prediction results of multiple models, effectively addressing the limitations of single prediction models in predicting the three-dimensional indicators of engineering time, cost, and quality. This stacked integrated architecture fully leverages the advantages of linear regression models in capturing linear relationships, the ability of support vector machines in processing nonlinear data, and the strengths of neural networks in extracting complex interactions of high-dimensional features. The combined prediction results of multiple models significantly improve prediction accuracy, providing a reliable basis for engineering management decisions. This invention innovatively introduces the occupational repetitive action method into the engineering resource allocation process. It systematically considers the impact of the ergonomic risk index on the efficiency of engineering execution. By calculating the ergonomic risk index matrix of engineering tasks and combining it with the predicted values ​​of three-dimensional indicators, it optimizes resource allocation, effectively solving the technical defects of traditional resource allocation methods that ignore human factors engineering factors, while improving the health and safety level of workers and work efficiency.

[0015] This invention quantifies and models engineering uncertainties based on resource allocation schemes, applies the α-cut fuzzy set method to handle uncertainties, and uses a multi-objective symbiotic biological search algorithm to find optimal solutions under multi-level uncertainty conditions. This algorithm evolves solutions by simulating symbiotic relationships in nature, and can find the optimal balance point among the three conflicting objectives of schedule, cost, and quality, providing project managers with multi-scheme decision support under different risk preferences. Overall, this invention achieves intelligent management of the entire process from data collection, predictive analysis, resource allocation to uncertainty handling, transforming engineering construction management from traditional experience-based decision-making to data-driven scientific decision-making, significantly improving the accuracy, scientific nature, and adaptability of engineering project management, and is particularly suitable for the full-process digital management of complex large-scale engineering projects. Attached Figure Description

[0016] To more clearly illustrate the technical solutions of the embodiments of the present invention, the drawings used in the description of the embodiments will be briefly introduced below. Obviously, the drawings described below are some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0017] Figure 1 This is a schematic diagram of one embodiment of the digital management method for the entire construction process in this application.

[0018] Figure 2 This is a schematic diagram of one embodiment of the digital management system for the entire construction process in this application.

[0019] Figure 3 This is a schematic block diagram of the structure of the digital management equipment for the entire construction process in this embodiment of the invention. Detailed Implementation

[0020] This application provides a method and system for digital management of the entire engineering construction process. The terms "first," "second," "third," "fourth," etc. (if present) in the specification, claims, and accompanying drawings of this application are used to distinguish similar objects and are not necessarily used to describe a specific order or sequence. It should be understood that such data can be interchanged where appropriate so that the embodiments described herein can be implemented in a sequence other than that illustrated or described herein. Furthermore, the terms "comprising" or "having" and any variations thereof are intended to cover a non-exclusive inclusion; for example, a process, method, system, product, or device that includes a series of steps or units is not necessarily limited to those steps or units explicitly listed, but may include other steps or units not explicitly listed or inherent to such processes, methods, products, or devices.

[0021] For ease of understanding, the specific process of the embodiments of this application is described below. Please refer to [link / reference]. Figure 1 One embodiment of the digital management method for the entire construction process in this application includes:

[0022] Step S101: Collect multi-source heterogeneous data throughout the entire lifecycle of the project construction and perform standardization processing to obtain a standardized project dataset;

[0023] Step S102: Construct a stacked ensemble prediction system using a standardized engineering dataset. Integrate the outputs of linear regression, support vector machine, and neural network through a gradient boosting decision tree to obtain predicted values ​​of the three-dimensional indicators of engineering time, cost, and quality.

[0024] Step S103: Based on the predicted values ​​of the three-dimensional indicators, the engineering task ergonomic risk index calculated by the occupational repetitive action method is combined with the engineering resources to obtain the resource allocation plan.

[0025] Step S104: Quantify the uncertainties of the project based on the resource allocation scheme, and calculate the scheme set under multi-level uncertainty through the multi-objective symbiotic biological search algorithm to obtain the project construction management strategy.

[0026] It is understood that the executing entity of this application can be a digital management system for the entire process of engineering construction, or it can be a terminal or a server; the specific implementation is not limited here. This application's embodiment uses a server as an example for illustration.

[0027] Specifically, a distributed IoT system collects multi-source heterogeneous data, including design parameters, cost, schedule, quality, safety, and human resource data, forming raw multi-source heterogeneous data. This data then undergoes outlier detection and labeling, cleaning, format conversion, and spatiotemporal alignment, followed by semantic unification processing. Finally, it is constructed into a graph database with a multi-dimensional index structure, generating a standardized engineering dataset. This step solves the problems of information fragmentation and inconsistent data standards in traditional engineering management. Based on the standardized engineering dataset, the system performs feature engineering, obtaining an optimized feature set through feature selection and dimensionality reduction, and dividing it into training, validation, and test datasets. The training dataset is used to train linear regression, support vector machine regression, and deep neural network models respectively. The validation and test datasets are used for parameter tuning and performance evaluation to obtain the prediction results of each model. Then, the prediction results of the three models, along with key features, are input into a gradient boosting decision tree algorithm for ensemble learning, yielding predicted values ​​for the three-dimensional indicators of engineering time, cost, and quality. This stacked ensemble method overcomes the shortcomings of insufficient prediction accuracy of a single model.

[0028] The complexity, risk level, and environmental impact of engineering tasks are quantified to form quantitative data on task characteristics. Based on this, the ergonomic risk index of engineering tasks is calculated using the occupational repetitive action method. Simultaneously, worker professional qualifications, years of service, physical fitness tests, and historical project score data are collected to establish a numerical profile of worker capabilities. The ergonomic risk index is combined with three-dimensional indicator predictions to calculate the predicted time, cost, and quality values ​​for each worker-task combination, forming the foundational data for task allocation calculations. Through constructed random search and simulated annealing optimization, the optimal resource allocation scheme is generated under the premise of satisfying the ergonomic risk threshold constraint. This solves the problem that traditional resource allocation does not consider ergonomic risks. The system establishes a database of engineering uncertainty factors, classifies and encodes external and internal uncertainty factors, and uses the α-cut fuzzy set method to build a mathematical model. This model is then combined with the resource allocation scheme to construct a three-dimensional indicator fuzzy mathematical model. Through a multi-objective symbiotic biological search algorithm, iterative optimization is performed by simulating three symbiotic relationships: mutual benefit, unilateral benefit, and parasitism. Pareto non-dominated sorting and crowding calculation are applied to select the optimal solution set under different uncertainty levels, forming an engineering construction management strategy. This solves the problem that traditional decision-making methods cannot effectively handle uncertainty.

[0029] In one specific embodiment, the process of performing step S101 may specifically include the following steps:

[0030] The system collects design parameter data, cost data, schedule data, quality data, safety data, and human resource data through a distributed Internet of Things system to obtain raw multi-source heterogeneous data.

[0031] Outlier detection is performed on the original multi-source heterogeneous data to obtain an outlier labeled dataset;

[0032] The outlier-labeled dataset is cleaned to obtain the cleaned dataset.

[0033] The cleaned dataset is subjected to format conversion and spatiotemporal alignment to obtain a format-normalized dataset.

[0034] Semantic unification is performed on the format-standardized dataset to obtain a semantically standardized dataset.

[0035] The semantically standardized dataset is constructed into a graph database, and a multi-dimensional index structure is established to obtain a standardized engineering dataset containing attribute features, temporal features, spatial features, and relational features.

[0036] Specifically, a distributed Internet of Things (IoT) system collects multi-source heterogeneous data. This system comprises a BIM data acquisition module that collects 3D design parameter data, a cost data acquisition module that collects cost data such as material and labor costs, RFID and video recognition technologies that acquire real-time progress data, non-destructive testing equipment and a sensor network that collects quality data such as structural and material quality, infrared cameras and gas detectors that monitor on-site safety conditions, and an intelligent positioning system and biometrics that collect human resource data such as worker location and work status, forming the raw multi-source heterogeneous data. Outlier detection is then performed on the raw multi-source heterogeneous data, and the density clustering algorithm (DBSCAN) is used to label the data. This algorithm divides data points according to density and automatically identifies whether a data point is a core point, boundary point, or noise point by setting two parameters: neighborhood radius and minimum number of points. For example, when detecting cost data, data exceeding a preset range (such as a sudden and significant fluctuation in the cost of a certain material) are marked as outliers, resulting in an outlier-labeled dataset.

[0037] The outlier-labeled dataset was then cleaned using a moving average filtering algorithm to remove noise. This algorithm smooths out data fluctuations by replacing the original value with the mean of several points before and after the data point. Labeled outliers were corrected (e.g., replaced with the mean) or removed, resulting in a cleaned dataset. The cleaned dataset underwent format conversion and spatiotemporal alignment, converting data from different sources to a unified JSON format and aligning it to a unified coordinate system using timestamps and spatial reference points. For example, IFC format design data exported from BIM software and cost data in Excel format were both converted to JSON format and linked by engineering component ID and time point to obtain a standardized dataset.

[0038] Then, semantic unification was performed on the format-standardized dataset, and semantic interoperability of data from different domains was achieved by establishing a unified ontology model. The ontology model defines the core concepts and relationships in the engineering domain, such as the association between components and concepts like materials, costs, and schedules, enabling data from different sources to communicate semantically, resulting in a semantically standardized dataset. The semantically standardized dataset was then constructed as a graph database with a multi-dimensional index structure. The graph database uses nodes and relationships to form a network structure, with engineering components as nodes and spatial and temporal relationships between components as edges, forming a mesh structure. Simultaneously, attribute feature indexes (by material type, cost range, etc.), temporal feature indexes (by planned schedule, actual progress, etc.), spatial feature indexes (by location coordinates, spatial relationships, etc.), and relational feature indexes (by dependency relationships, influence relationships, etc.) were established to obtain a standardized engineering dataset, solving the problems of information fragmentation and data sharding in traditional engineering management.

[0039] In one specific embodiment, the process of performing step S102 may specifically include the following steps:

[0040] Feature engineering is performed on the standardized engineering dataset, and an optimized feature set is obtained through feature selection and feature dimensionality reduction.

[0041] The optimized feature set is divided into a training set, a validation set, and a test set to obtain the model training dataset, the model validation dataset, and the model test dataset.

[0042] A linear regression model is trained using the model training dataset, and a regularization method is used to process the feature weights to obtain a linear prediction model.

[0043] The parameters of the linear prediction model were tuned using the model validation dataset, and the model performance was evaluated using the model test dataset to obtain the linear prediction results.

[0044] The model training dataset is used to train a support vector machine regression model and a deep neural network model, and the parameters are optimized using the model validation dataset to obtain the prediction results of the support vector machine and the neural network.

[0045] By using linear prediction results, support vector machine prediction results, and neural network prediction results as input features, and combining them with key features in the optimized feature set, the gradient boosting decision tree algorithm is used for ensemble learning to obtain the predicted values ​​of the three-dimensional indicators of engineering time, cost, and quality.

[0046] Specifically, feature selection techniques are employed to identify the features most strongly correlated with the three-dimensional indicators of project time, cost, and quality. By calculating the Pearson correlation coefficient between each feature and the target variable, features with an absolute correlation coefficient greater than 0.3 are selected. Simultaneously, variance inflation factor detection is used to remove multicollinear features; for example, design parameter features and cost features with similarity exceeding 80% are merged. Next, principal component analysis is used for feature dimensionality reduction, projecting high-dimensional features into a low-dimensional space and retaining principal components with a cumulative contribution rate of 95%, resulting in an optimized feature set. This optimized feature set is then divided chronologically, with the first 70% of the data used as the model training dataset, the middle 10% as the model validation dataset, and the last 20% as the model test dataset. This temporal division ensures that the model can learn the temporal evolution patterns during the project construction process, avoiding data leakage issues.

[0047] A linear regression model is trained using the model training dataset. Ridge regression-style regularization is employed to handle feature weights, and a penalty term is introduced for each feature to control model complexity. The objective function of the linear regression model includes a mean squared error term and an L2 regularization term. Feature weight values ​​are obtained by minimizing this objective function, thus forming a linear prediction model.

[0048] The linear prediction model was optimized using a model validation dataset, primarily by adjusting the regularization strength parameter λ. A series of candidate values, ranging from 0.001 to 100, were set according to a logarithmic scale, and the λ value with the smallest root mean square error on the validation set was selected as the optimal parameter. The model performance was then evaluated using a model test dataset, calculating the root mean square error, mean absolute error, and coefficient of determination between the predicted and actual values ​​to obtain the linear prediction results. A support vector machine (SVM) regression model and a deep neural network model were trained in parallel on the model training dataset. The SVM regression model used a radial basis function kernel to handle nonlinear relationships, and the optimal combination of the penalty factor C and the kernel parameter γ was determined through grid search. The deep neural network model constructed a four-layer network structure: the input layer corresponds to the feature dimension, the two hidden layers have varying node counts, and the output layer corresponds to the three-dimensional index. The LeakyReLU activation function and the AdamW optimizer were used, with a learning rate of 0.001. The learning rate, batch size, and number of hidden layer nodes were adjusted using the model validation dataset to obtain the prediction results for both the SVM and neural networks. Linear predictions, support vector machine predictions, and neural network predictions are used as first-order features. These are combined with key features from the optimized feature set (such as project size, construction process complexity, and material supply stability) and input into a gradient boosting decision tree meta-learner. This meta-learner iteratively trains multiple decision trees, with each tree learning the residuals from the previous prediction. Finally, the predictions from all trees are weighted and combined to obtain more accurate predictions of the project's time-cost-quality three-dimensional indicators.

[0049] In one specific embodiment, the process of performing step S103 may specifically include the following steps:

[0050] The complexity, risk level and environmental load of each engineering task are numerically quantified, the specific numerical scores of each item are calculated and weighted and summarized to obtain a task characteristic quantification data table.

[0051] Based on the task characteristic quantitative data table, the actual frequency, strength factor, posture factor, additional factor and recovery factor of technical movements are calculated using the occupational repetitive action method and then comprehensively calculated to obtain the ergonomic risk index matrix of engineering tasks.

[0052] Collect data on each worker's professional qualification certificate level, years of service, physical fitness test results, and historical project completion quality scores to form structured data records and obtain a numerical profile of the worker's abilities.

[0053] The ergonomic risk index matrix of engineering tasks is correlated with the predicted values ​​of three-dimensional indicators to generate the completion time, cost consumption and quality score prediction values ​​for each worker-task combination, thus obtaining the basic dataset for task allocation calculation.

[0054] A constructive random search operation is performed on the task allocation calculation base dataset. Workers and tasks are permuted and combined through iterative calculation to minimize the total project duration, total cost and maximize the total quality score. The boundary condition is that the total ergonomic risk score does not exceed the safety threshold, and an initial digital resource allocation table is obtained.

[0055] A local search optimization calculation is performed on the initial digital resource allocation table. The solution set is adjusted through a simulated annealing calculation process with gradually decreasing temperature parameters, and the optimal solution in each iteration is retained to obtain the resource allocation scheme.

[0056] Specifically, the complexity of actions is scored along two dimensions: the number of actions and physical load. The number of actions is calculated based on the total number of standard actions required for each task; for example, rebar tying includes actions such as bending, lifting, and twisting, with each action assigned a value based on frequency. Physical load is categorized into mild (1 point), moderate (3 points), and severe (5 points) based on the physical exertion required for each action. Risk level is determined by multiplying the probability of a safety accident by the degree of loss, with both probability and degree of loss using a 1-5 scale. For example, the probability score for high-altitude work is 4, and the degree of loss score is 5. Environmental load considers the impact of environmental factors such as noise, temperature, and humidity on workers, quantified using a 1-5 scale. The scores from the three dimensions are weighted and summed in a 4:4:2 ratio to form a comprehensive score, resulting in a task characteristic quantification data table. Based on this task characteristic quantification data table, the Occupational Repetitive Action Method (OCRA) is used to calculate the ergonomic risk index of the engineering task. First, the actual frequency of technical movements is calculated, i.e., the number of movements completed per unit time. Then, the strength factor is determined, scored based on the percentage of maximum voluntary strength required for the task. Next, the posture factor is evaluated, assigned a value based on the degree to which the work posture deviates from the ideal ergonomic position. Then, additional factors are determined, considering vibration, accuracy requirements, etc. Finally, the recovery factor is calculated based on the work-to-rest ratio. These five factors are then combined: the actual frequency of technical movements is multiplied by the strength factor, posture factor, and additional factors, and then divided by the recovery factor to obtain the OCRA index, which is used to construct an engineering task ergonomic risk index matrix. Data on each worker's professional qualification certificate level, years of service, physical fitness test results, and historical project completion quality scores are collected to form a numerical record of worker capabilities. Professional qualification certificate levels are divided into primary (1 point), intermediate (3 points), and advanced (5 points) according to national vocational qualification standards; years of service data are recorded according to actual years and converted into capability scores using an experience curve; physical fitness test results include a comprehensive score of three indicators: strength, endurance, and flexibility; historical project completion quality scores are calculated based on the average score of the worker's past project quality acceptance data.

[0057] The ergonomic risk index matrix of engineering tasks is correlated with the predicted values ​​of three-dimensional indicators to generate a basic dataset for task allocation calculation. Specifically, each worker's ability data is matched with the risk index of each task to determine the efficiency coefficient for each worker performing a specific task. This efficiency coefficient is then applied to standard time, standard cost, and standard quality indicators to obtain personalized time, cost, and quality predictions. For example, highly skilled workers performing high-risk tasks may achieve shorter time, slightly higher cost, and better quality; while novice workers performing the same task may experience longer time, slightly lower cost, but lower quality. A constructed random search operation is performed on the basic dataset for task allocation calculation, iteratively arranging and combining workers and tasks. The algorithm first generates an initial solution using a greedy strategy, prioritizing worker assignment to tasks they are most proficient in. Then, it randomly replaces some worker-task allocation relationships, checking whether the multi-objective optimization conditions of shortest total time, lowest total cost, and highest quality are met, while ensuring that the total ergonomic risk score does not exceed a safety threshold. The optimal solution is retained in each iteration, and an initial digital resource allocation table is obtained after multiple iterations. Local search optimization is performed on the initial digital resource allocation table, and simulated annealing is used to adjust the solution set. The algorithm sets an initial temperature parameter, accepting suboptimal solutions across a wide range during the high-temperature phase to escape local optima. As the temperature gradually decreases, the adjustment strategy becomes more conservative, ultimately accepting only better solutions during the low-temperature phase. Throughout the process, the optimal solution from each iteration is retained to form the final resource allocation scheme.

[0058] In one specific embodiment, the process of executing step S104 may specifically include the following steps:

[0059] Establish a database of uncertainties in engineering construction, identify external and internal uncertainties, classify and encode them to obtain a structured dataset of uncertainties;

[0060] The α-cut fuzzy set method is used to mathematically model each factor in the structured dataset of uncertainty factors, transforming each uncertainty factor into a fuzzy numerical value with a membership function, thus obtaining a fuzzy mathematical model of uncertainty factors.

[0061] By combining the fuzzy mathematical model of uncertainty factors with the resource allocation scheme, a fuzzy construction period calculation matrix, a fuzzy cost calculation matrix and a fuzzy quality calculation matrix are constructed to obtain a fuzzy mathematical model of engineering three-dimensional indicators that takes uncertainty into account.

[0062] Initialize the ecosystem of the multi-objective symbiotic biological search algorithm, create multiple candidate solutions and assign fitness values ​​to each solution to obtain an initial solution set family;

[0063] The initial solution set family is iterated through three symbiotic mechanisms: mutual benefit, unilateral benefit, and parasitic relationship. The mutual benefit relationship generates a new solution by combining the advantageous attributes of two solutions, the unilateral benefit relationship generates an improved solution by one solution influencing another, and the parasitic relationship replaces the existing solution through random mutation, resulting in the evolved solution set family.

[0064] Pareto nondominated sorting and congestion calculation are applied to the evolved solution set family to screen out the nondominated optimal solutions with the shortest construction period, lowest cost and highest quality at different α levels, forming multiple sets of optimal schemes and obtaining engineering construction management strategies.

[0065] Specifically, by collecting historical project data, key events leading to delays, increased costs, and decreased quality are extracted. These events are extracted from project anomaly records, change requests, and project summary reports using text mining techniques, forming a raw dataset of uncertainties. Hierarchical clustering analysis is then performed on this data, employing a fully connected method to calculate the similarity between different factors. Factors with similarity exceeding a preset threshold are grouped together, resulting in a classification of uncertainties. The factors are then divided into two main categories: external uncertainties and internal uncertainties. External uncertainties include extreme weather (such as the impact of heavy rain and high temperatures on open-air construction), policy changes (such as adjustments to environmental standards), and market fluctuations (such as price fluctuations in materials like steel and cement). Internal uncertainties include fluctuations in resource supply (such as delays in the delivery of key equipment), technological changes (such as adjustments to design schemes), and personnel turnover (such as the departure of skilled workers). The impact of each type of uncertainty is quantified by calculating the intensity, probability of occurrence, and duration of impact using expert scoring and historical data statistical analysis. A unique identifier is assigned to each factor, and a correlation graph between factors is established, ultimately forming a structured dataset of uncertainties.

[0066] Mathematical modeling of each factor in the structured dataset of uncertainty factors is performed using the α-cut fuzzy set method. This method represents each uncertainty factor as a fuzzy number with a membership function, which describes the membership degree of a given value to that uncertainty factor, ranging from 0 to 1. The minimum, most likely, and maximum likely values ​​of each factor are determined through historical data analysis, constructing a triangular membership function. Then, the interval values ​​of the fuzzy number are calculated at different α levels (α values ​​from 0 to 1). A larger α value results in a narrower interval, indicating lower uncertainty. For example, for the uncertainty factor of "heavy rain," α=0 may lead to a delay of 0-7 days, α=0.5 may lead to a delay of 2-5 days, and α=1 may result in a fixed delay of 3 days. In this way, all uncertainty factors are represented as fuzzy numerical values ​​with multi-level uncertainties, forming a fuzzy mathematical model of uncertainty factors. This fuzzy mathematical model of uncertainty factors is then combined with resource allocation schemes to construct a fuzzy mathematical model of engineering three-dimensional indicators that considers uncertainty. First, the project duration calculation in the resource allocation scheme is fuzzified. The deterministic project duration is combined with the project delay value affected by uncertain factors through fuzzy addition to form a fuzzy project duration calculation matrix. Similarly, fuzzy arithmetic operations are applied to cost and quality calculations to form fuzzy cost calculation matrices and fuzzy quality calculation matrices, respectively. For example, the deterministic project duration of a certain construction task is 10 days, but due to the uncertainty factor of "skilled worker departure," the actual project duration is a fuzzy number 10 + [2,5] days, i.e., 12-15 days, at an uncertainty level of α=0.3. Through similar calculations, a complete fuzzy mathematical model of the three-dimensional indicators of the project is obtained. This model reflects the possible range of values ​​for project time, cost, and quality under different uncertainty levels.

[0067] When initializing the multi-objective symbiotic biological search algorithm, the population size and iteration count parameters are first set. Then, multiple initial candidate solutions are generated in the solution space through Latin hypercube sampling. Each candidate solution represents a resource allocation and engineering implementation strategy. Its fitness is evaluated by calculating its fuzzy indices of duration, cost, and quality at different α levels, forming an initial solution set family. The multi-objective symbiotic biological search algorithm is an evolutionary algorithm that simulates biological symbiotic relationships, optimizing the solution set family by simulating three types of symbiotic relationships. In a mutually beneficial relationship, two solutions are randomly selected, and their midpoint is calculated as the new solution. If the new solution is better than the original solution, the original solution is replaced. In a unilateral benefit relationship, one solution learns from the other, but only the beneficiary updates. In a parasitic relationship, a solution is copied and its random dimensions are mutated. If the mutated solution is better than the original solution, the original solution is replaced. Through the alternating action of these three mechanisms, the solution set family gradually evolves towards the optimal direction.

[0068] Pareto non-dominated sorting and crowding calculation are applied to the evolved solution set. Pareto non-dominated sorting ranks solutions according to their dominance relationship; if a solution is no worse than another solution in all objectives and better in at least one objective, then the first solution dominates the second solution. A non-dominated solution is one that has no other solution that can dominate it. Crowding calculation measures the distance between solutions to maintain solution diversity. Through these two steps, the non-dominated optimal solutions with the shortest construction period, lowest cost, and highest quality at different α levels (e.g., α=0.2, 0.5, 0.8) are selected, forming multiple sets of optimal schemes, ultimately resulting in a complete engineering construction management strategy.

[0069] In one specific embodiment, the process of identifying external and internal uncertainties may specifically include the following steps:

[0070] Collect records of abnormal events and change requests from historical engineering projects, extract key events that lead to delays, increased costs, and decreased quality, and obtain the original dataset of uncertainty.

[0071] Cluster analysis was performed on the original dataset of uncertainty, and hierarchical clustering algorithm was used to classify similar factors to obtain the classification results of uncertainty factors;

[0072] The results of the classification of uncertainty factors are divided into external uncertainty and internal uncertainty. External uncertainty includes extreme weather, policy changes and market volatility, while internal uncertainty includes resource supply fluctuations, technological changes and personnel mobility, resulting in a two-level classification system.

[0073] The influence of various uncertain factors in the secondary classification system is quantified. The influence intensity, probability of occurrence and duration of each factor are calculated by expert scoring and historical data analysis to obtain the feature vector of uncertain factors.

[0074] Encoding rules are established based on the feature vectors of uncertainty factors, a unique identification code is assigned to each type of uncertainty factor, and a relationship graph between factors is established to obtain the network structure of uncertainty factors.

[0075] The network structure of uncertainty factors is converted into a structured data format, and a data table containing factor identifiers, classification information, influence characteristics, and correlations is established to obtain a structured dataset of uncertainty factors.

[0076] Specifically, by parsing the text of documents such as quality acceptance reports, progress comparison tables, cost settlement statements, and change requests for completed projects, key events affecting the construction period, cost, and quality are extracted. Data collection uses structured templates, labeling each event with information such as occurrence time, duration, scope of impact, cause description, and result description. For example, the event "Construction was delayed by 7 days due to water accumulation in the foundation pit caused by continuous heavy rain" is extracted from the progress report, coded as "Abnormal Event - Weather Category - 001," and associated with its specific impact data on the construction period. In this way, an uncertain raw dataset containing hundreds of typical events is accumulated from multiple historical projects. When performing cluster analysis on the uncertain raw dataset, a hierarchical clustering algorithm is used to classify similar factors. First, event feature vectors are constructed, including textual features of the event descriptions (vectorized using TF-IDF) and impact features (number of days of construction delay, percentage increase in cost, and degree of quality deviation). Then, a similarity matrix between events is calculated, using cosine similarity to calculate the similarity of textual features and Euclidean distance to calculate the similarity of impact features, and combining the two to obtain a comprehensive similarity. Based on the similarity matrix, the most similar event clusters are merged from bottom to top using the full connection method to form a clustering dendrogram. By setting a similarity threshold (e.g., 0.7), the dendrogram is truncated to obtain several event clusters, forming the classification results of uncertainty factors.

[0077] The classification results of uncertainty factors are further divided into external uncertainty and internal uncertainty categories. External uncertainties include factors beyond the control of the project owner, subdivided into extreme weather (e.g., heavy rain, high temperatures, typhoons), policy changes (e.g., adjustments to environmental standards, updates to regulations), and market fluctuations (e.g., material price fluctuations, inflation). Internal uncertainties include factors within the project that are partially controllable, subdivided into resource supply fluctuations (e.g., untimely material supply, equipment failures), technological changes (e.g., adjustments to design schemes, changes in construction techniques), and personnel turnover (e.g., changes in key personnel, labor disputes). This classification method forms a two-level classification system, facilitating the subsequent implementation of corresponding countermeasures for different categories of factors. The impact of each type of uncertainty factor in the two-level classification system is quantified using a dual-track assessment method: on the one hand, expert scores are collected through the Delphi method, with multiple rounds of anonymous evaluation allowing experts to score the impact of each factor in different project types; on the other hand, historical data statistical analysis is used to calculate the historical average impact value and standard deviation of each factor. By weighting and fusing the two evaluation results (historical data weighted at 0.7 and expert scores weighted at 0.3), a three-dimensional feature vector for each factor is obtained: influence intensity value (average deviation from the three-dimensional indicators, 1-5 points), occurrence probability value (historical frequency of occurrence, a decimal between 0 and 1), and duration value (average number of days the influence lasts). These three values ​​together constitute the feature vector of the uncertainty factor.

[0078] Encoding rules are established based on the feature vectors of uncertain factors, employing a hierarchical coding scheme in the form of "A-BB-CCC". Here, A represents the primary category (E for external, I for internal), BB represents the secondary category (e.g., WT for extreme weather, PC for policy changes), and CCC represents the factor sequence number. Simultaneously, a factor correlation matrix is ​​established based on the temporal, causal, and cumulative impact relationships between factors, recording the correlation strength (values ​​between 0 and 1) between each pair of factors. A correlation graph is drawn based on the correlation matrix, where nodes represent uncertain factors, edges represent correlation relationships, and edge thickness indicates correlation strength, forming a network structure of uncertain factors. This network structure is then converted into a structured data format, creating a set of relational data tables, including a basic factor information table (storing factor ID, name, description, and classification information), a factor feature table (storing feature values ​​such as impact strength, probability of occurrence, and duration), and a factor correlation table (storing the correlation type and strength between factors). These three tables are linked by factor IDs, forming a complete structured dataset of uncertain factors. This provides a data foundation for subsequent α-cut fuzzy set modeling and solves the problem of lacking systematic management of uncertain factors during engineering construction.

[0079] In one specific embodiment, the process of creating multiple candidate solutions and assigning a fitness value to each solution may specifically include the following steps:

[0080] Based on the fuzzy mathematical model of engineering three-dimensional index, the solution space and constraints of the multi-objective optimization problem are defined, and the population size parameter and iteration number parameter are set to obtain the algorithm control parameter set.

[0081] Based on the algorithm control parameter set and resource allocation scheme, multiple initial candidate schemes are generated in the solution space through Latin hypercube sampling, forming a diverse initial population and obtaining a candidate solution set;

[0082] For each solution in the candidate solution set, calculate the objective function values ​​for project duration, cost, and quality, and synthesize the comprehensive fitness value by weighted summation to obtain the initial solution set family.

[0083] Specifically, the optimization problem is defined based on a three-dimensional index fuzzy mathematical model. The solution space includes three types of decision variables: process sequencing, resource allocation, and risk response. Constraints include total resource constraints, technical logic constraints, and risk threshold constraints. Simultaneously, algorithm control parameters such as population size (typically 30-50), iteration count (100-300 times), mutation probability, and crossover probability are set. Based on these parameters and the resource allocation scheme, Latin hypercube sampling is used to generate initial candidate solutions. This method divides the value range of each decision variable into N equal intervals (N equals the population size), randomly selects a value from each interval, and generates candidate solutions by randomly combining values ​​from different dimensions. For example, multiple sequencing schemes satisfying logical relationships are generated for the process sequencing variable, and multiple allocation schemes satisfying resource constraints are generated for the resource allocation variable, ensuring that the diversity of the initial population covers the entire solution space.

[0084] For each candidate solution, three objective function values ​​are calculated: the schedule objective function calculates the completion time based on the critical path method; the cost objective function calculates the total cost including direct and indirect costs; and the quality objective function evaluates the quality level based on factors such as worker-task matching. Fuzzy objective function values ​​are calculated at multiple α levels (0.3, 0.5, 0.7), and finally, a weighted summation (e.g., schedule, cost, and quality weights of 0.4, 0.4, and 0.2 respectively) is used to synthesize a comprehensive fitness value, forming an initial solution set family. This provides a foundation for multi-objective optimization and effectively solves the problem that traditional methods struggle to simultaneously optimize three-dimensional indicators and handle uncertainty.

[0085] The above describes the digital management method for the entire construction process in the embodiments of this application. The following describes the digital management system for the entire construction process in the embodiments of this application. Please refer to [link / reference]. Figure 2 One embodiment of the digital management system for the entire construction process in this application includes:

[0086] Processing module 201 is used to collect multi-source heterogeneous data throughout the entire life cycle of engineering construction and perform standardization processing to obtain a standardized engineering dataset;

[0087] Integration module 202 is used to build a stacked ensemble prediction system using a standardized engineering dataset. It integrates the outputs of linear regression, support vector machine and neural network through gradient boosting decision tree to obtain predicted values ​​of three-dimensional indicators of engineering time-cost-quality.

[0088] The allocation module 203 is used to allocate engineering resources based on the predicted values ​​of three-dimensional indicators, combined with the engineering task ergonomic risk index calculated by the occupational repetitive action method, and to obtain a resource allocation scheme.

[0089] The quantification module 204 is used to quantify the uncertainties of the project based on the resource allocation scheme, and to calculate the set of schemes under multi-level uncertainties through a multi-objective symbiotic biological search algorithm to obtain the project construction management strategy.

[0090] above Figure 2 The digital management system for the entire process of engineering construction in this embodiment of the invention will be described in detail from the perspective of modular functional entities. The digital management equipment for the entire process of engineering construction in this embodiment of the invention will be described in detail from the perspective of hardware processing.

[0091] Figure 3 This is a schematic diagram of the structure of a digital management device for the entire construction process provided in an embodiment of the present invention. The digital management device 300 for the entire construction process can vary significantly due to different configurations or performance. It may include one or more central processing units (CPUs) 310 (e.g., one or more processors) and a memory 320, and one or more storage media 330 (e.g., one or more mass storage devices) for storing application programs 333 or data 332. The memory 320 and storage media 330 can be temporary or persistent storage. The program stored in the storage media 330 may include one or more modules (not shown in the diagram), each module may include a series of instruction operations on the digital management device 300 for the entire construction process. Furthermore, the processor 310 may be configured to communicate with the storage media 330 and execute the series of instruction operations in the storage media 330 on the digital management device 300 for the entire construction process to implement the steps of the aforementioned digital management method for the entire construction process.

[0092] The digital management equipment 300 for the entire construction process may also include one or more power supplies 340, one or more wired or wireless network interfaces 350, one or more input / output interfaces 360, and / or one or more operating systems 331, such as Windows Server, Mac OS X, Unix, Linux, FreeBSD, etc. Those skilled in the art will understand that... Figure 3 The illustrated structure of the digital management equipment for the entire construction process does not constitute a limitation on the digital management equipment for the entire construction process provided by this invention. It may include more or fewer components than illustrated, or combine certain components, or have different component arrangements.

[0093] The present invention also provides a computer-readable storage medium, which can be a non-volatile computer-readable storage medium or a volatile computer-readable storage medium, wherein the computer-readable storage medium stores instructions that, when the instructions are executed on a computer, cause the computer to perform the steps of the digital management method for the entire process of engineering construction.

[0094] Those skilled in the art will clearly understand that, for the sake of convenience and brevity, the specific working processes of the systems and units described above can be referred to the corresponding processes in the foregoing method embodiments, and will not be repeated here.

[0095] If the integrated unit is implemented as a software functional unit and sold or used as an independent product, it can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of the present invention, in essence, or the part that contributes to the prior art, or all or part of the technical solution, can be embodied in the form of a software product. This computer software product is stored in a storage medium and includes several instructions to cause a digital management device for the entire engineering construction process (which may be a personal computer, server, or network device, etc.) to execute all or part of the steps of the methods described in the various embodiments of the present invention. The aforementioned storage medium includes various media capable of storing program code, such as USB flash drives, portable hard drives, read-only memory (ROM), random access memory (RAM), magnetic disks, or optical disks.

[0096] The above embodiments are only used to illustrate the technical solutions of the present invention, and are not intended to limit it. Although the present invention has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that modifications can still be made to the technical solutions described in the foregoing embodiments, or equivalent substitutions can be made to some of the technical features. Such modifications or substitutions do not cause the essence of the corresponding technical solutions to deviate from the spirit and scope of the technical solutions of the embodiments of the present invention.

Claims

1. A digital management method for the entire process of engineering construction, characterized in that, include: Multi-source heterogeneous data is collected through a distributed Internet of Things (IoT) system. This system comprises a BIM data acquisition module for collecting 3D design parameters, a cost data acquisition module for collecting material and labor costs, RFID and video recognition technologies for real-time progress data, non-destructive testing equipment and sensor networks for collecting structural and material quality data, infrared cameras and gas detectors for monitoring on-site safety, and intelligent positioning and biometric technologies for collecting worker location and work status data, forming raw multi-source heterogeneous data. Outlier detection is performed on this raw data to obtain an outlier-labeled dataset. This outlier-labeled dataset is then cleaned to obtain a cleaned dataset. Finally, the cleaned dataset undergoes format conversion and spatiotemporal alignment to obtain a format-normalized dataset. The standardized dataset undergoes semantic unification to obtain a semantically standardized dataset. This dataset is then constructed into a graph database with a multi-dimensional index structure, resulting in a standardized engineering dataset containing attribute features, temporal features, spatial features, and relational features. A stacked ensemble prediction system is built using this standardized engineering dataset, integrating the outputs of linear regression, support vector machines, and neural networks through a gradient boosting decision tree to obtain predicted values ​​for three-dimensional indicators of engineering time, cost, and quality. Based on these three-dimensional indicator predictions, and combined with the engineering task ergonomic risk index calculated using the occupational repetitive action method, engineering resources are allocated to obtain a resource allocation scheme. Based on the resource allocation scheme, engineering uncertainty factors are quantified, and a multi-objective symbiotic biological search algorithm is used to calculate a set of schemes under multi-level uncertainty, resulting in an engineering construction management strategy. Based on the predicted values ​​of three-dimensional indicators, combined with the ergonomic risk index of engineering tasks calculated by the occupational repetitive action method, and the allocation of engineering resources, a resource allocation plan is obtained. This includes: quantifying the motion complexity, risk level, and environmental load of each engineering task, calculating the specific numerical scores of each item, and weighting and summarizing them to obtain a task characteristic quantification data table; based on the task characteristic quantification data table, using the occupational repetitive action method to calculate the actual frequency of technical movements, strength factor, posture factor, additional factor, and recovery factor, and performing comprehensive calculations to obtain the engineering task ergonomic risk index matrix; collecting the professional qualification certificate level, years of service, physical fitness test results, and historical project completion quality scores of each worker to form structured data records and obtain a numerical profile of worker capabilities. The ergonomic risk index matrix of engineering tasks is correlated with the predicted values ​​of three-dimensional indicators to generate the completion time, cost, and quality score prediction values ​​for each worker-task combination, thus obtaining the basic dataset for task allocation calculation. A constructed random search operation is performed on this dataset, iteratively arranging and combining workers and tasks to minimize the total project duration, total cost, and maximize the total quality score, with the ergonomic risk score not exceeding a safety threshold as the boundary condition, resulting in an initial digital resource allocation table. A local search optimization calculation is then performed on the initial digital resource allocation table, adjusting the solution set through a simulated annealing process with progressively decreasing temperature parameters, and retaining the optimal solution from each iteration to obtain the resource allocation scheme.

2. The digital management method for the entire construction process according to claim 1, characterized in that, A stacked ensemble prediction system is constructed using a standardized engineering dataset. The outputs of linear regression, support vector machine (SVM), and neural network are integrated using a gradient boosting decision tree to obtain predicted values ​​for the three-dimensional indicators of engineering time, cost, and quality. The process includes: performing feature engineering on the standardized engineering dataset, obtaining an optimized feature set through feature selection and dimensionality reduction; dividing the optimized feature set into training, validation, and test sets to obtain model training, validation, and test datasets; training a linear regression model using the training dataset, processing feature weights with regularization to obtain a linear prediction model; optimizing the parameters of the linear prediction model using the validation dataset, and evaluating model performance using the test dataset to obtain linear prediction results; training a support vector machine (SVM) regression model and a deep neural network model on the training dataset, and optimizing the parameters using the validation dataset to obtain SVM and neural network prediction results; using the linear, SVM, and neural network prediction results as input features, combined with key features from the optimized feature set, and performing ensemble learning using a gradient boosting decision tree algorithm to obtain predicted values ​​for the three-dimensional indicators of engineering time, cost, and quality.

3. The digital management method for the entire construction process according to claim 1, characterized in that, Based on resource allocation schemes, engineering uncertainties are quantified. A multi-objective symbiotic biological search algorithm is used to calculate a set of schemes under multi-level uncertainties, resulting in engineering construction management strategies. These strategies include: establishing a database of engineering construction uncertainties, identifying external and internal uncertainties, and classifying and coding them to obtain a structured dataset of uncertainties; mathematically modeling each factor in the structured dataset using the α-cut fuzzy set method, transforming each uncertainty factor into a fuzzy numerical value with membership functions, resulting in a fuzzy mathematical model of uncertainties; and combining this fuzzy mathematical model with resource allocation schemes to construct fuzzy duration calculation matrices, fuzzy cost calculation matrices, and fuzzy quality calculation matrices, thus obtaining a comprehensive strategy considering uncertainties. A fuzzy mathematical model of three-dimensional engineering indicators is constructed; the ecosystem of a multi-objective symbiotic biological search algorithm is initialized, multiple candidate solutions are created, and a fitness value is assigned to each solution to obtain an initial solution set family; the initial solution set family is iteratively evolved by simulating three symbiotic mechanisms: mutual benefit relationship, unilateral benefit relationship, and parasitic relationship. Among them, the mutual benefit relationship generates a new solution by combining the advantageous attributes of two solutions, the unilateral benefit relationship generates an improved solution by one solution influencing another solution, and the parasitic relationship replaces existing solutions through random mutation to obtain an evolved solution set family; Pareto non-dominated sorting and crowding degree calculation are applied to the evolved solution set family to screen out the non-dominated optimal solutions with the shortest construction period, lowest cost, and highest quality at different α levels, forming multiple sets of optimal schemes to obtain the engineering construction management strategy.

4. The digital management method for the entire construction process according to claim 3, characterized in that, A database of uncertainties in engineering construction is established to identify external and internal uncertainties, classify and code them, and obtain a structured dataset of uncertainties. This includes: collecting records of abnormal events and change requests from historical engineering projects, extracting key events that lead to delays, increased costs, and decreased quality, and obtaining the original dataset of uncertainties; performing cluster analysis on the original dataset of uncertainties, using hierarchical clustering algorithms to classify similar factors, and obtaining the classification results of uncertainties; dividing the classification results of uncertainties into external uncertainty categories and internal uncertainty categories, where external uncertainties include extreme weather, policy changes, and market fluctuations, and internal uncertainties include resource supply fluctuations, technological changes, and personnel mobility, resulting in a two-level classification system; quantifying the impact of various uncertainties in the two-level classification system, calculating the impact intensity, probability of occurrence, and duration of each factor through expert scoring and historical data analysis, and obtaining the feature vector of uncertainties; establishing coding rules based on the feature vectors of uncertainties, assigning a unique identification code to each type of uncertainty, and establishing a correlation graph between factors, resulting in the network structure of uncertainties; converting the network structure of uncertainties into a structured data format, establishing a data table containing factor identifiers, classification information, impact characteristics, and correlation relationships, and obtaining the structured dataset of uncertainties.

5. The digital management method for the entire construction process according to claim 4, characterized in that, The ecosystem of the multi-objective symbiotic biological search algorithm is initialized, multiple candidate solutions are created, and a fitness value is assigned to each solution to obtain an initial solution set family. This includes: defining the solution space and constraints of the multi-objective optimization problem based on a fuzzy mathematical model of engineering three-dimensional indicators, setting population size parameters and iteration number parameters to obtain the algorithm control parameter set; generating multiple initial candidate solutions in the solution space through Latin hypercube sampling based on the algorithm control parameter set and resource allocation scheme to form a diverse initial population and obtain a candidate solution set; calculating the project duration objective function value, cost objective function value, and quality objective function value for each solution in the candidate solution set, and synthesizing the comprehensive fitness value through a weighted summation method to obtain the initial solution set family.

6. A digital management system for the entire construction process, characterized in that, To implement the digital management method for the entire process of engineering construction as described in any one of claims 1-5, the digital management system for the entire process of engineering construction includes: a processing module for collecting multi-source heterogeneous data throughout the entire lifecycle of engineering construction and performing standardized processing to obtain a standardized engineering dataset; an integration module for constructing a stacked integrated prediction system using the standardized engineering dataset, integrating the output results of linear regression, support vector machine, and neural network through a gradient boosting decision tree to obtain predicted values ​​of three-dimensional indicators of engineering time, cost, and quality; an allocation module for allocating engineering resources based on the predicted values ​​of the three-dimensional indicators, combined with the engineering task ergonomic risk index calculated by the occupational repetitive action method, to obtain a resource allocation scheme; and a quantification module for quantifying the uncertainties of engineering based on the resource allocation scheme, calculating a set of schemes under multi-level uncertainties through a multi-objective symbiotic biological search algorithm to obtain an engineering construction management strategy.

7. A digital management device for the entire construction process, characterized in that, It includes a memory and a processor, the memory storing a computer program that can run on the processor, and the processor executing the computer program to implement the digital management method for the entire process of engineering construction as described in any one of claims 1 to 5.

8. A computer-readable storage medium having a computer program stored thereon, characterized in that, When a computer program is run by a processor, it causes the processor to execute the digital management method for the entire engineering construction process as described in any one of claims 1 to 5.

Images