Risk node evaluation system applied to construction project changes

Abstract

The application provides a risk node evaluation system applied to building engineering project changes, relates to the technical field of risk evaluation, and comprises the following steps: a simulation evaluation model of a project node in a building engineering scene is constructed; according to actual data requirements corresponding to building engineering project node changes and based on building engineering data experience fitting, change related data of a building engineering project node to be evaluated is obtained; according to the change related data of the building engineering project node to be evaluated and the simulation evaluation model, a change simulation risk quantitative value of the building engineering project node to be evaluated is calculated; a classification risk entropy estimation is performed on the change simulation risk quantitative value, a single change risk entropy and a joint change risk entropy of the building engineering project node to be evaluated are calculated; and according to the single change risk entropy and the joint change risk entropy of the building engineering project node to be evaluated, a risk comprehensive evaluation is performed to obtain an actual change risk quantitative value corresponding to the building engineering project node changes.

Description

Technical Field

[0001] This invention relates to the field of risk assessment technology, specifically a risk node assessment system for changes in construction engineering projects. Background Technology

[0002] In the implementation of construction projects, changes to project milestones are a frequent occurrence. Various changes to milestones, such as changes to drawings, construction techniques, materials and equipment, and schedule adjustments, can all lead to different levels of project risks. If the risks of these changes are not accurately and comprehensively quantified and assessed, it can easily lead to project delays, cost overruns, or even quality problems. Therefore, risk assessment of changes to construction project milestones has become a core aspect of project management.

[0003] In the existing risk assessment process, the data fitting stage lacks in-depth integration with the experience of the construction engineering industry. As a result, the final risk assessment results are difficult to provide accurate and practical quantitative basis for decision-making and risk management of changes in construction engineering project milestones.

[0004] Therefore, we now offer a risk node assessment system for changes in construction engineering projects, which provides a precise and comprehensive quantitative assessment of change risks. Summary of the Invention

[0005] To address the aforementioned technical problems, the present invention aims to provide a risk node assessment system applicable to changes in building engineering projects.

[0006] To achieve the above objectives, the present invention provides the following technical solution: a risk node assessment system for changes in building engineering projects, the system comprising a simulation assessment model construction module, a data fitting module, a change risk quantity calculation module, a single and joint risk entropy calculation module, and a change risk comprehensive assessment module; The simulation evaluation model building module is used to build simulation evaluation models that adapt to changes in project nodes in building engineering scenarios. The data fitting module obtains relevant change data for the nodes of the building project to be evaluated based on the actual data requirements of the corresponding building project node changes and on the empirical fitting of building project data. The change risk calculation module calculates the simulated risk quantification value of the change to be evaluated for the nodes of the building project to be evaluated based on the change-related data of the nodes and the simulation evaluation model. The single and combined risk entropy calculation module classifies the simulated risk quantification value to be changed into risk entropy estimation, and calculates the single change risk entropy and combined change risk entropy of the node of the building project to be evaluated. The change risk comprehensive assessment module is used to conduct a comprehensive risk assessment based on the single change risk entropy and the combined change risk entropy of the building project node to be assessed, and to obtain the actual change risk quantification value of the corresponding building project node change.

[0007] Furthermore, the process of constructing a simulation evaluation model using Bayesian networks and gradient boosting trees includes: Based on standardized data corresponding to historical building engineering scenarios, the prior probability of risk change for each node to be evaluated in the corresponding building engineering scenario is calculated in the corresponding training set, validation set, and test set. The standardized data includes the standardized engineering feature matrix and the historical risk quantification value after node change in the corresponding historical building engineering scenario. The prior risk probability of each node change to be evaluated in the corresponding building engineering scenario in the corresponding training set, validation set and test set is horizontally spliced ​​and fused with the corresponding engineering feature matrix to obtain the corresponding fused feature matrix. Based on the corresponding fusion feature matrix and the corresponding historical risk quantification value, a gradient boosting tree model is trained and adapted to the scenario, thereby constructing a simulation evaluation model and outputting the corresponding simulated risk quantification value.

[0008] Furthermore, the steps for calculating the prior probability of risk changes at each node to be evaluated in the corresponding building engineering scenario include: Core factors strongly correlated with node change risk are screened using Pearson correlation coefficients; risk factors exceeding the threshold are recorded as the core risk factor set of the Bayesian network based on threshold comparison. Based on the experience of construction engineering experts and the core factors in the core risk factor set of Bayesian networks, an initial causal edge set is constructed, thereby forming an initial topology; the initial topology is then optimized based on the PC algorithm. Based on maximum likelihood estimation, the conditional probabilities of nodes in the topology optimization structure of the corresponding training set, validation set, and test set under different states of the parent node are calculated; based on Bayesian inference, and according to the corresponding conditional probabilities, the corresponding prior risk probabilities are calculated.

[0009] Furthermore, the corresponding prior risk probabilities are fused with the engineering feature matrix by horizontal concatenation according to the horizontal concatenation function, thereby obtaining the corresponding input feature matrix.

[0010] Furthermore, the steps for training the gradient boosting tree model are as follows: The gradient boosting tree model is initialized with parameters, including initial learning rate, initial tree depth, initial number of iterations, L2 regularization coefficient, and leaf node splitting threshold; and the gradient boosting tree model after parameter initialization is configured for basic training, including setting the loss function, gradient boosting training, and 5-fold cross-validation optimization. Based on the dataset of the corresponding construction engineering scenario and the gradient boosting tree model after basic training configuration, a gradient boosting tree model specifically for the corresponding scenario is trained, which is denoted as the simulation evaluation model.

[0011] Furthermore, based on the application scenario of the corresponding construction project, the actual data requirements for changes in the project nodes are broken down, and then the corresponding basic engineering data and environmental and constraint data are defined. Based on experience fitting of corresponding types of construction project data, the data of the construction project node change to be evaluated is initially fitted. Then, based on industry experience in construction project risk assessment, the data after the initial fitting is optimized a second time to complete the correlation features between the data of the construction project node change to be evaluated, forming a fitted feature dataset of the construction project node change to be evaluated, which is recorded as change-related data.

[0012] Furthermore, based on the change-related data of the construction project nodes to be evaluated and the simulation evaluation model, the process of calculating the simulated risk quantification value of the change-prone nodes of the construction project to be evaluated includes: Input the standardized actual engineering feature matrix of the building project nodes to be evaluated into the trained simulation evaluation model; Using a Bayesian network, the prior probability of actual risk of node change in the construction project to be evaluated is first calculated; then, the prior probability of actual risk is horizontally concatenated and fused with the corresponding actual project feature matrix to obtain the fused actual input feature matrix; based on the gradient boosting tree model, the simulated risk quantification value of node change in the construction project to be evaluated is calculated and recorded as the simulated risk quantification value of the change.

[0013] Furthermore, the steps for classifying and estimating the risk entropy of the simulated risk quantification value to be changed, and calculating the single change risk entropy of the construction project node to be evaluated, are as follows: Based on the risk quantification value sample set, and using the kernel density estimation method, the risk probability distribution of changes between two adjacent nodes in the construction project to be evaluated is calculated; Based on the classic information entropy formula, a formula for calculating the entropy value of a single risk entropy specific to changes in building engineering nodes is constructed. Based on the entropy value calculation formula of a single risk entropy, the entropy values ​​of several types of core single change risks are calculated, and then the single change risk entropy set is obtained.

[0014] Furthermore, the steps for classifying and estimating the risk entropy of the simulated risk quantification value to be changed, and calculating the joint change risk entropy of the building project nodes to be evaluated, are as follows: Based on the topology optimization structure of Bayesian networks, the causal coupling relationship between several types of single change risks is identified. Then, based on the experience of construction engineering experts, the mutual information method is used to calculate the coupling degree between any two types of single change risks. Based on the degree of coupling, a node change risk coupling matrix is ​​constructed. The diagonal elements in the matrix represent the self-coupling degree of each individual risk, and the off-diagonal elements represent the mutual coupling degree between two types of risks. Based on the risk probability distribution and the node change risk coupling matrix, the joint probability distribution of several types of single change risks is calculated using the joint probability density function. Based on the joint entropy theory of information entropy and combined with the risk coupling matrix, the joint change risk entropy of several types of single change risks is calculated.

[0015] Furthermore, the process of conducting a comprehensive risk assessment based on the single change risk entropy and combined change risk entropy of the construction project node to be assessed includes: Based on the analytic hierarchy process, the weights of primary indicators and secondary indicators are assigned to the single change risk entropy and the joint change risk entropy. The comprehensive standardized entropy value of a single change risk is obtained by weighting the secondary indicator weights; the corresponding basic risk quantification value is calculated by weighted linear fusion based on the primary indicator weights and the comprehensive standardized entropy value. The basic risk quantification value is calibrated for engineering scenarios and matched with risk levels to obtain the actual change risk quantification value of the corresponding construction project node change, and then the final engineering risk level is matched.

[0016] Compared with the prior art, the beneficial effects of the present invention are: the risk node assessment system for changes in building engineering projects provided in this application constructs a complete risk assessment system with multi-module collaboration, and realizes intelligent processing of the entire process of risk assessment for changes in building engineering project nodes from data collection and fitting to final quantitative assessment.

[0017] By employing collaborative modeling with Bayesian networks and gradient boosting trees, this approach balances the interpretability of risk causal transmission logic with the quantitative accuracy of the nonlinear impact of risk factors, enabling the simulation assessment model to highly adapt to the actual characteristics of construction engineering scenarios. Through refined classification of change risks and risk entropy estimation, it not only accurately characterizes the inherent uncertainty of various individual change risks but also effectively quantifies the coupling correlation characteristics between multiple types of change risks, achieving multi-dimensional assessment of change risks and overcoming the shortcomings of traditional single-dimensional quantification methods. This significantly improves the accuracy and comprehensiveness of risk assessment for construction engineering project node changes, providing a reliable quantitative basis for scientific decision-making and risk management of construction engineering project node changes, effectively reducing engineering risks caused by node changes, and improving the project management level of construction engineering. Attached Figure Description

[0018] To more clearly illustrate the technical solutions in the embodiments of this application or the prior art, the drawings used in the embodiments will be briefly introduced below. Obviously, the drawings described below are only some embodiments recorded in this invention. For those skilled in the art, other drawings can be obtained based on these drawings.

[0019] Figure 1 This is a schematic diagram of a risk node assessment system applied to changes in building construction projects.

[0020] Figure 2 A schematic diagram illustrating the steps of collaborative processing using Bayesian Networks (BN) and Gradient Boosting Trees (XGBoost).

[0021] Figure 3 A schematic diagram illustrating the steps for calculating the prior probability of risk.

[0022] Figure 4 A schematic diagram illustrating the steps involved in training a gradient boosting tree model.

[0023] Figure 5 A schematic diagram illustrating the steps for calculating the entropy of a single change risk.

[0024] Figure 6 A schematic diagram illustrating the steps for calculating the joint change risk entropy.

[0025] Figure 7 This diagram illustrates the steps involved in conducting a comprehensive risk assessment based on both single change risk entropy and combined change risk entropy. Detailed Implementation

[0026] To make the objectives, technical solutions, and advantages of this invention clearer, the technical solutions of this invention will be described in detail below. Obviously, the described embodiments are merely some embodiments of this invention, and not all embodiments. Based on the embodiments of this invention, all other implementation methods obtained by those skilled in the art without creative effort are within the scope of protection of this invention.

[0027] It should be noted that the terms "first," "second," etc., 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 of this application described herein can be implemented in orders other than those illustrated or described herein. Furthermore, the terms "comprising" and "having," and any variations thereof, are intended to cover non-exclusive inclusion; for example, a process, method, system, product, or apparatus that comprises 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 apparatus.

[0028] like Figure 1 As shown, a risk node assessment system for changes in construction engineering projects is provided. The system includes, but is not limited to, a simulation assessment model construction module, a data fitting module, a change risk calculation module, a single and joint risk entropy calculation module, and a change risk comprehensive assessment module. The simulation evaluation model building module is used to build simulation evaluation models that adapt to changes in project nodes in building engineering scenarios.

[0029] It should be noted that by constructing a simulation evaluation model, it is possible to conduct a preliminary simulation evaluation of project nodes in a construction engineering scenario when changes occur, thereby providing data basis for assessing the risks of actual project nodes when changes occur.

[0030] In this embodiment, the simulation evaluation model is built based on the data of past node changes of projects in different construction engineering scenarios. The data of past node changes of projects in different construction engineering scenarios includes, but is not limited to, the corresponding historical change data, basic engineering data, and environmental and constraint data. To further clarify, the corresponding historical change data includes, but is not limited to, change types, such as changes to drawings / schemes, construction techniques, materials and equipment, and schedule adjustments; change triggering reasons, such as design defects, unsuitable on-site construction conditions, policy adjustments, and changes in actual construction needs; and change execution process data, such as the number of days of schedule delay and cost overrun percentage. The corresponding basic engineering data includes, but is not limited to, project type, such as building construction projects and bridge projects; project scale, such as building area, project cost, and construction period; personnel configuration, such as the number and qualifications of management, technical, and construction personnel; and equipment configuration, such as the type and quantity of construction machinery. The corresponding environmental and constraint data includes, but is not limited to, policy and regulatory data, such as new regulations related to environmental protection, safety, and quality; and natural environment data, such as weather conditions and geological features of the construction area. All of the above data primarily comes from historical industry case data provided by the construction project management platform and third-party engineering data service agencies, ensuring coverage of change scenarios for different project types and project milestones.

[0031] It should be noted that data obtained from different channels all need to undergo further data preprocessing, including data cleaning, feature engineering, and data partitioning, to improve data quality, so as to facilitate subsequent model training and thus improve accuracy.

[0032] In this embodiment, a simulation evaluation model is constructed using Bayesian Networks (BN) and Gradient Boosting Trees (XGBoost). Bayesian Networks (BN), with its probabilistic graphical model at its core, can clearly depict the causal relationships between variables, adapting to the risk transmission logic of node changes in construction engineering scenarios. For example, the causal chain of material change → construction process adjustment → schedule delay risk → cost overrun risk. Engineering technicians can intuitively understand the risk transmission path through the structure of the Bayesian Network, meeting the model's interpretability requirements. Furthermore, Bayesian Networks are highly adaptable to small sample data and can be combined with expert prior probabilities to compensate for the insufficient data volume in some new construction engineering scenarios, thereby improving the model's generalization ability. Gradient Boosting Trees (XGBoost) are an efficient and lightweight machine learning model that excels at capturing nonlinear relationships between features. They can accurately quantify the impact of various risk factors on node change risks, addressing the shortcomings of Bayesian Networks in risk quantification accuracy. The risk of changes in construction project nodes is affected by a variety of factors, and some of these factors have complex nonlinear relationships. For example, there is an interaction between geological conditions and construction difficulty. Gradient Boosting Tree (XGBoost) can gradually optimize the prediction error and improve the accuracy of the model's risk quantification by using gradient boosting.

[0033] It should be noted that the simulation evaluation model is constructed through the collaborative processing of Bayesian Network (BN) and Gradient Boosting Tree (XGBoost). Specifically, Bayesian Network (BN) is responsible for characterizing the causal relationship of risk and outputting the prior probability, while Gradient Boosting Tree (XGBoost) is responsible for quantifying the nonlinear effects and outputting the final risk value.

[0034] In this embodiment, the steps of the collaborative processing of Bayesian Network (BN) and Gradient Boosting Tree (XGBoost) are as follows: Step S10: Based on the standardized data corresponding to the historical building engineering scenarios, calculate the prior probability of risk changes for each node to be evaluated in the corresponding building engineering scenarios in the training set, validation set, and test set.

[0035] It should be noted that the standardized data is divided into training, validation, and test sets in a 7:2:1 ratio, and this divided data will be used for subsequent model training, validation, and testing. The standardized data includes the standardized engineering feature matrix for the corresponding historical building engineering scenario and the historical risk quantification values ​​after node changes.

[0036] Step S11: Horizontally concatenate and fuse the prior probability of risk change of each node to be evaluated in the corresponding building engineering scenario in the corresponding training set, validation set, and test set with the corresponding engineering feature matrix to obtain the corresponding fused feature matrix.

[0037] Step S12: Based on the corresponding fusion feature matrix and the corresponding historical risk quantification value, train the gradient boosting tree (XGBoost) model and implement scenario adaptation, thereby constructing a simulation evaluation model and outputting the corresponding simulated risk quantification value.

[0038] In step S10, the specific steps for calculating the prior probability of risk changes at each node to be evaluated for the corresponding building engineering scenario include: Step S101: Screen core factors strongly correlated with node change risk using Pearson correlation coefficient. The correlation coefficient calculation formula is as follows: ;in, This represents the engineering feature matrix corresponding to the training set, validation set, or test set. The Middle Each risk factor and its corresponding historical risk quantification value ; This represents the engineering feature matrix corresponding to the training set, validation set, or test set. The mean; This represents the historical risk quantification value corresponding to the training set, validation set, or test set. The mean.

[0039] Step S102: Calculate the correlation coefficient The risk factors are denoted as the core risk factor set of the Bayesian network: ;in, Indicates the first One core risk factor.

[0040] Step S103: Based on the experience of construction engineering experts and the core risk factor set of Bayesian networks The core factors are used to construct an initial set of causal edges, thereby forming an initial topological structure. Optimize the initial topology based on the PC algorithm By using the conditional independence test (chi-square test) and combining data from the training, validation, and test sets, the statistical significance of the initial causal edges is tested. Meaningless edges are removed and hidden edges are added. The test formula is as follows: ;in, This represents the actual number of observations for the corresponding node state combinations in the training, validation, and test sets. This represents the theoretical expected number of iterations, where k is the number of state combinations; if ,in, , For each degree of freedom, causal edges are retained; otherwise, they are deleted, thus obtaining the corresponding topology optimization structure: .

[0041] It should be noted that the PC algorithm (Peter-Clark algorithm) is a Bayesian network (BN) structure learning algorithm based on conditional independence test. Its core purpose is to optimize the topology of Bayesian networks, which meets the needs of risk assessment for node changes in construction engineering projects.

[0042] Step S104: Based on maximum likelihood estimation, calculate the corresponding topology optimization structure for the training set, validation set, and test set. Nodes in In the parent node Conditional probabilities under different states: ;in, Represents a statistical function based on data from the training set, validation set, and test set; Represents a node A certain state; The state combination of the parent node set.

[0043] To further explain, if In small sample scenarios, Laplace smoothing correction is used: ;in, This represents the smoothing coefficient, with a value of 1. For nodes The number of states.

[0044] Step S105: Based on Bayesian inference and according to the corresponding conditional probabilities, calculate the corresponding prior risk probabilities: ;in, , This represents the prior probability of the risk corresponding to the training set; , This represents the prior probability of the risk corresponding to the validation set; , This represents the prior probability of the risk corresponding to the test set; , as well as This indicates the corresponding data dimension.

[0045] In step S11, the specific process of horizontally concatenating and fusing the corresponding prior risk probabilities and the engineering feature matrix to obtain the corresponding fused feature matrix includes: The corresponding prior risk probabilities are fused with the engineering feature matrix using the horizontal concatenation function. The concatenation formula is as follows: ;in, This is a horizontal splicing function; , as well as These represent the XGBoost input feature matrices for the XGBoost training set, validation set, and test set, respectively.

[0046] It should be noted that, , as well as , The feature dimension corresponds to the engineering feature matrix; the XGBoost input feature matrix has BN causal probability features and combines engineering scene features.

[0047] In step S12, the steps for training the gradient boosting tree (XGBoost) model are as follows: Step S121: Initialize the parameters of the gradient boosting tree XGBoost model, including the initial learning rate, initial tree depth, initial number of iterations, L2 regularization coefficient, and leaf node splitting threshold.

[0048] Step S122: Perform basic training configuration on the gradient boosting tree XGBoost model after parameter initialization, including setting the loss function, gradient boosting training, and 5-fold cross-validation optimization. The specific basic training configuration process will not be described in detail.

[0049] Step S123: Based on the dataset of the corresponding construction engineering scenario and the gradient boosting tree XGBoost model after basic training configuration, train a gradient boosting tree XGBoost model specifically for the corresponding scenario, denoted as the simulation evaluation model, and make predictions based on the corresponding test set data, outputting the simulated risk quantification value: ;in, This represents the simulated risk quantification value under this construction project scenario; This represents the XGBoost gradient boosting tree model specific to this construction project scenario.

[0050] To further explain, the simulated risk quantification value under this construction engineering scenario... The error difference between the value and the corresponding historical risk quantification value is calculated, and a risk quantification error threshold is set. If the error exceeds the risk quantification error threshold, it needs to be fed back to BN and XGBoost for parameter optimization; otherwise, it does not.

[0051] The data fitting module obtains relevant change data for the nodes of the building project to be evaluated based on the actual data requirements of the corresponding building project node changes and on empirical fitting of building project data.

[0052] In this embodiment, based on the application scenario of the corresponding construction project, the actual data requirements for the changes in the nodes of the construction project are broken down, and then the corresponding basic engineering data and environmental and constraint data are defined. It should be noted that the data granularity should be determined based on the actual assessment accuracy requirements of the construction project. For example, cost data should be accurate to a percentage, and construction period data should be accurate to a day. At the same time, the data format should be standardized as structured numerical data and categorized data to ensure compatibility with the input format of the simulation assessment model.

[0053] In this embodiment, based on the empirical fitting of the corresponding type of building engineering data, the data of the building engineering project node change to be evaluated is initially fitted, and then the data after the initial fitting is optimized a second time by using the industry experience of building engineering risk assessment to complete the correlation features between the data of the building engineering project node change to be evaluated, forming a fitted feature dataset of the building engineering project node change to be evaluated, and recorded as change-related data, to ensure that the data completely matches the input feature dimensions of the simulation evaluation model.

[0054] Further data preprocessing is performed on the engineering basic data, environmental and constraint data, and change-related data of the construction project to be evaluated when the node changes are made. This includes data cleaning, feature engineering, etc., to obtain a standardized actual engineering feature matrix of the construction project node changes to be evaluated, so as to meet the data requirements of the simulation evaluation model and thus improve accuracy.

[0055] It should be noted that the basic engineering data, environmental and constraint data corresponding to the changes in the nodes of the construction project are predictable, while the change-related data are based on empirical fitting of the construction project data, and therefore contain a certain degree of error or deviation. However, in construction engineering, data errors are mostly random and irregular deviations, which are objective errors that cannot be completely eliminated and are affected by accidental factors such as measuring tools and operational precision. Data deviations, on the other hand, are mostly systematic and regular deviations, which are systematic errors that can be corrected by human intervention. They are caused by fixed factors such as data collection methods, fitting logic, and industry experience adaptability, and the direction and magnitude of the deviations are consistent. These deviations can be largely eliminated through optimization methods, scenario adaptation, and logical correction. This application reduces the amount of data deviation through secondary optimization.

[0056] The change risk calculation module calculates the simulated risk quantification value of the node of the building project to be evaluated based on the change-related data of the node and the simulation evaluation model, and records it as the simulated risk quantification value to be changed.

[0057] In this embodiment, based on the scenario type and node change type of the construction project to be evaluated, the simulation evaluation model specifically designed for the corresponding scenario type, which was trained during the simulation evaluation model construction phase, is invoked to ensure that the model is highly compatible with the engineering scenario of the node to be evaluated.

[0058] Input the standardized actual engineering feature matrix of the building project nodes to be evaluated into the trained simulation evaluation model; The prior probability of actual risk of node change in the construction project to be evaluated is calculated by using a Bayesian network; and the prior probability of actual risk is horizontally spliced ​​and fused with the corresponding actual project feature matrix to obtain the fused actual input feature matrix. By using the gradient enhancement reasoning logic of a dedicated simulation evaluation model, the nonlinear correlation of each risk factor in the actual input feature matrix is ​​captured, and the impact of various risk factors on the risk of node changes is accurately quantified. The simulation risk quantification value of the node change of the construction project to be evaluated is calculated by the simulation evaluation model and recorded as the simulation risk quantification value of the change.

[0059] The single and combined risk entropy calculation module classifies the simulated risk quantification value to be changed into risk entropy estimation, and calculates the single change risk entropy and combined change risk entropy of the building project node to be evaluated.

[0060] It should be noted that by assessing the risk of changes between two adjacent nodes in a construction project and the risk of progressive or parallel changes between multiple nodes, an accurate prediction of the overall change risk of the construction project to be assessed can be achieved.

[0061] In this embodiment, the simulated risk quantification value to be changed is classified into risk entropy estimation, including risk entropy estimation of single change risk and risk entropy estimation of joint change risk. To further explain, single change risk entropy corresponds to the risk entropy estimation of changes between two adjacent nodes in the construction project to be evaluated; joint change risk entropy corresponds to the risk entropy estimation of progressive or parallel changes between multiple nodes in the construction project to be evaluated.

[0062] It should be noted that the risks associated with changes at key points in construction projects are not singular. The risk formation mechanisms, impact scope, and transmission paths differ significantly among different change types. Directly estimating the entropy of the overall simulated risk value for the changes would fail to accurately depict the inherent characteristics of different risk types. Therefore, this step first uses the risk classification standards for changes at key points in construction projects to classify and refine the simulated risk value for the changes, laying the foundation for subsequent entropy estimation calculations.

[0063] In this embodiment, based on the core types of project node changes in architectural engineering and matching the core risk factor set used in the construction of the simulation evaluation model, the simulated risk quantification value to be changed is divided into four core single change risks: scheme design change risk (denoted as R1), construction process change risk (denoted as R2), material and equipment change risk (denoted as R3), and schedule adjustment change risk (denoted as R4). This classification not only covers the mainstream types of architectural engineering node changes but also highly matches the risk factors of the simulation evaluation model, ensuring the consistency between entropy value calculation and previous risk quantification.

[0064] In the further processing, let the output of the change risk calculation module be the simulated risk quantification value to be changed. Based on the risk contribution decomposition algorithm, the simulated risk value to be changed is quantified. Deconstructing the risk into its quantitative value corresponding to each individual change risk yields the set of simulated quantitative values ​​for each individual change risk: ;in, For the first The quantification value of the simulated risk to be changed for a single type of change risk. And satisfy ; For the first The contribution weight of a single type of change risk to the overall risk.

[0065] To eliminate the dimensional differences in the quantification values ​​of individual change risks, The engineering normalization process is performed, mapping it to the [0,1] interval to obtain the normalized quantified value of a single change risk. The formula is: ;in, For similar scenarios in construction engineering The minimum quantitative value for the risk of a single type of change; For similar scenarios in construction engineering The maximum quantifiable value of a single type of change risk.

[0066] Normalized Single Change Risk Quantification Indicates the first The relative risk level of a single type of change risk. The closer it is to 1, the higher the actual occurrence and impact of this type of risk.

[0067] Based on historical case data of similar node changes in the construction engineering industry, a sample set of risk quantification values ​​is constructed for each type of single change risk. ; To ensure the statistical significance of subsequent entropy value calculations, the sample set covers risk quantification values ​​under different engineering scenarios and different reasons for change, in order to represent the number of similar historical cases.

[0068] In this embodiment, the steps for classifying and estimating the risk entropy of the simulated risk quantification value to be changed, and calculating the single change risk entropy of the construction project node to be evaluated, are as follows: Step 21: Based on the risk quantification value sample set Based on the kernel density estimation method, the probability distribution of risk of change between two adjacent nodes in the construction project to be evaluated is calculated. ,in, Indicates the first Single change risk in the first The probability of occurrence under a sample. It should be noted that the kernel density estimation method does not require pre-setting the distribution type, making it more suitable for the non-normal distribution characteristics of construction engineering risk data; the risk probability distributions of the four types of single change risks are denoted as follows: .

[0069] Step 22: Based on the classic information entropy formula, construct a single risk entropy calculation formula specific to changes in building engineering nodes, and calculate the first... Entropy value of single change risk The formula is: ;in, This represents the risk impact coefficient, reflecting the first... The scope and severity of the impact of a single type of change risk on the overall construction project are determined by construction experts using the Delphi method in accordance with industry standards, with a range of values. ; The weight represents the probability of risk occurrence, reflecting the first... The actual probability of occurrence of a single type of change risk in the current project scenario to be evaluated is obtained by normalizing the prior probability of this type of risk output by the Bayesian network (BN), and its value ranges from [value range missing]. The higher the prior probability, The larger the value, the higher the likelihood that this type of risk will occur in the current project; This represents the classical information entropy term, used to indicate the mathematical dispersion of risk quantification values.

[0070] Step 23: Based on the entropy value The calculation formula is used to calculate the entropy value of the four core single change risks, thereby obtaining the single change risk entropy set. .

[0071] In this embodiment, the steps for classifying and estimating the risk entropy of the simulated risk quantification value to be changed, and calculating the joint change risk entropy of the building project nodes to be evaluated, are as follows: Step 31: Topology optimization structure based on Bayesian network (BN) The study identifies the causal coupling relationships among four types of single change risks, and then, based on the experience of construction engineering experts, uses the mutual information method to calculate the risk of any two types of single change risks. and Coupling between value range , The closer it is to 1, the greater the risk of a single change. and The stronger the coupling relationship, the easier it is for risk to be transmitted. The closer to 0, the higher the risk of a single change. and Basically independent.

[0072] Step 32: Based on Coupling Degree Construct a node change risk coupling matrix In the matrix, the diagonal elements represent the self-coupling degree (taken as 1) of each individual risk, and the off-diagonal elements represent the mutual coupling degree between the two types of risks.

[0073] Step 33: Based on the risk probability distribution and node change risk coupling matrix The joint probability distribution of four types of single change risks is calculated based on the joint probability density function. The joint probability distribution for: ;in, It represents the product of the individual risk probability distributions, i.e., the joint probability when there is no coupling relationship; For coupling correction terms, coupling degree The larger the value, the larger the correction term, indicating a stronger influence of the coupling relationship on the joint probability.

[0074] Step 34: Based on the joint entropy theory of information entropy, combined with the risk coupling matrix The combined change risk entropy of the four types of single change risks is calculated as follows: in, The coupling entropy correction coefficient is derived from the risk coupling matrix. The trace mean was calculated. The range of values ​​is This is used to quantify the overall impact of coupling relationships on the joint entropy value; This represents the summation of the joint probabilities of all possible risk combinations.

[0075] In this embodiment, the single change risk entropy and the combined change risk entropy are subjected to dual validity verification, specifically as follows: Do the single change risk entropy and the joint change risk entropy satisfy the basic properties of information entropy, namely, the non-negativity of information entropy (i.e., both the single change risk entropy and the joint change risk entropy are greater than or equal to 0) and the monotonicity (i.e., the joint change risk entropy is not less than any single change risk entropy)?

[0076] The risk entropy value of the construction project node change is compared with the preset benchmark library. If the corresponding entropy value exceeds the reasonable range of the industry, for example, the single change risk entropy is much higher than the industry average, or the ratio of the joint change risk entropy to the single change risk entropy exceeds the reasonable range of the industry, the actual scenario of the construction project to be evaluated is manually reviewed to confirm whether it is a high-coupling and high-uncertainty risk specific to the project.

[0077] If the data does not meet either of the two validity checks, it can be deduced that there are data anomalies or errors between certain project nodes, so that errors can be detected in a timely manner and more accurate data support can be provided for subsequent comprehensive risk assessment.

[0078] The change risk comprehensive assessment module is used to conduct a comprehensive risk assessment based on the single change risk entropy and the combined change risk entropy of the building project node to be assessed, and to obtain the actual change risk quantification value of the corresponding building project node change.

[0079] It should be noted that the change risk comprehensive assessment module converts the single change risk entropy and joint change risk entropy output by the single and joint risk entropy calculation modules into a quantitative risk value that fits the actual situation of the construction project, providing a direct quantitative basis for risk management and decision-making for changes at construction project nodes.

[0080] In this embodiment, entropy value hierarchical weighting, engineering scenario calibration, and risk quantification mapping are used to achieve this. This retains the accurate characterization of risk uncertainty and coupling by entropy values, while also solving the problem of connecting pure entropy value indicators with the actual risk management needs of engineering projects through a weighting system and calibration rules specific to building engineering. This ensures that the output actual change risk quantification value has both mathematical rigor and engineering practicality, as detailed below: Step S41: Based on the analytic hierarchy process, assign weights to primary and secondary indicators to the single change risk entropy and the joint change risk entropy; It should be noted that the primary indicator weight refers to the weight relationship between the single change risk entropy and the combined change risk entropy; the secondary indicator weight refers to the weight relationship between the four types of core single change risks in the single change risk entropy.

[0081] Step S42: Obtain the comprehensive standardized entropy value of the single change risk entropy by weighting according to the weights of the secondary indicators.

[0082] Step S43: Based on the weights of the primary indicators and the comprehensive standardized entropy value, calculate the corresponding basic risk quantification value through weighted linear fusion.

[0083] Step S44: Perform engineering scenario calibration and risk level matching on the basic risk quantification value to obtain the actual change risk quantification value of the corresponding construction project node change, and then match the final engineering risk level.

[0084] In step S44, the engineering scenario calibration of the basic risk quantification value includes introducing a scenario complexity coefficient and a node importance coefficient. The actual change risk quantification value is obtained by multiplying the basic risk quantification value with the scenario complexity coefficient and the node importance coefficient.

[0085] Based on the actual management and control needs of the industry, the calibrated actual change risk quantification values ​​are divided into 5 engineering risk levels, and the risk characteristics and control requirements of each level are clearly defined, enabling the hierarchical interpretation of the quantification values ​​and facilitating project managers to quickly formulate control strategies. The specific matching standards are as follows: Table 1: Optionally, in this embodiment, those skilled in the art will understand that all or part of the steps in the various methods of the above embodiments can be implemented by a program instructing the hardware related to the terminal device. The program can be stored in a computer-readable storage medium, which may include: flash drive, read-only memory (ROM), random access memory (RAM), disk or optical disk, etc.

[0086] The sequence numbers of the embodiments in this application are for descriptive purposes only and do not represent the superiority or inferiority of the embodiments.

[0087] If the integrated units in the above embodiments are implemented as software functional units and sold or used as independent products, they can be stored in the aforementioned computer-readable storage medium. Based on this understanding, the technical solution of this application, 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 one or more electronic devices to execute all or part of the steps of the methods described in the various embodiments of this application.

[0088] In the above embodiments of this application, the descriptions of each embodiment have different focuses. For parts not described in detail in a certain embodiment, please refer to the relevant descriptions of other embodiments.

[0089] In the several embodiments provided in this application, it should be understood that the disclosed application can be implemented in other ways. The device embodiments described above are merely illustrative; for example, the division of units is only a logical functional division, and in actual implementation, there may be other division methods. For example, multiple units or components may be combined or integrated into another system, or some features may be ignored or not executed. Furthermore, the coupling or direct coupling or communication connection shown or discussed may be through some interfaces; the indirect coupling or communication connection of units or modules may be electrical or other forms.

[0090] The units described as separate components may or may not be physically separate. The components shown as units may or may not be physical units; that is, they may be located in one place or distributed across multiple network units. Some or all of the units can be selected to achieve the purpose of this embodiment according to actual needs.

[0091] Furthermore, the functional units in the various embodiments of this application can be integrated into one processing unit, or each unit can exist physically separately, or two or more units can be integrated into one unit. The integrated unit can be implemented in hardware or as a software functional unit.

[0092] The above description is only a preferred embodiment of this application. It should be noted that for those skilled in the art, several improvements and modifications can be made without departing from the principle of this application, and these improvements and modifications should also be considered within the scope of protection of this application.

Claims

1. A risk node assessment system for changes in construction engineering projects, characterized in that, The system includes a simulation evaluation model construction module, a data fitting module, a change risk calculation module, a single and joint risk entropy calculation module, and a change risk comprehensive evaluation module. The simulation evaluation model building module is used to build simulation evaluation models that adapt to changes in project nodes in building engineering scenarios. The data fitting module obtains relevant change data for the nodes of the building project to be evaluated based on the actual data requirements of the corresponding building project node changes and on the empirical fitting of building project data. The change risk calculation module calculates the simulated risk quantification value of the change to be evaluated for the nodes of the building project to be evaluated based on the change-related data of the nodes and the simulation evaluation model. The single and combined risk entropy calculation module classifies the simulated risk quantification value to be changed into risk entropy estimation, and calculates the single change risk entropy and combined change risk entropy of the node of the building project to be evaluated. The change risk comprehensive assessment module is used to conduct a comprehensive risk assessment based on the single change risk entropy and the combined change risk entropy of the building project node to be assessed, and to obtain the actual change risk quantification value of the corresponding building project node change.

2. The risk node assessment system for changes in construction engineering projects according to claim 1, characterized in that, The process of constructing a simulation evaluation model using Bayesian networks and gradient boosting trees includes: Based on standardized data corresponding to historical building engineering scenarios, the prior probability of risk change for each node to be evaluated in the corresponding building engineering scenario is calculated in the corresponding training set, validation set, and test set. The standardized data includes the standardized engineering feature matrix and the historical risk quantification value after node change in the corresponding historical building engineering scenario. The prior risk probability of each node change to be evaluated in the corresponding building engineering scenario in the corresponding training set, validation set and test set is horizontally spliced ​​and fused with the corresponding engineering feature matrix to obtain the corresponding fused feature matrix. Based on the corresponding fusion feature matrix and the corresponding historical risk quantification value, a gradient boosting tree model is trained and adapted to the scenario, thereby constructing a simulation evaluation model and outputting the corresponding simulated risk quantification value.

3. The risk node assessment system for changes in construction engineering projects according to claim 2, characterized in that, The steps for calculating the prior probability of risk of change at each node to be evaluated in the corresponding building engineering scenario include: Core factors strongly correlated with node change risk are screened using Pearson correlation coefficients; risk factors exceeding the threshold are recorded as the core risk factor set of the Bayesian network based on threshold comparison. Based on the experience of construction engineering experts and the core factors in the core risk factor set of Bayesian networks, an initial causal edge set is constructed, thereby forming an initial topology; the initial topology is then optimized based on the PC algorithm. Based on maximum likelihood estimation, the conditional probabilities of nodes in the topology optimization structure of the corresponding training set, validation set, and test set under different states of the parent node are calculated; based on Bayesian inference, and according to the corresponding conditional probabilities, the corresponding prior risk probabilities are calculated.

4. The risk node assessment system for changes in construction engineering projects according to claim 3, characterized in that, The corresponding prior risk probabilities are fused with the engineering feature matrix by horizontal concatenation using the horizontal concatenation function, thereby obtaining the corresponding input feature matrix.

5. The risk node assessment system for changes in construction engineering projects according to claim 4, characterized in that, The steps for training a gradient boosting tree model are as follows: The gradient boosting tree model is initialized with parameters, including initial learning rate, initial tree depth, initial number of iterations, L2 regularization coefficient, and leaf node splitting threshold; and the gradient boosting tree model after parameter initialization is configured for basic training, including setting the loss function, gradient boosting training, and 5-fold cross-validation optimization. Based on the dataset of the corresponding construction engineering scenario and the gradient boosting tree model after basic training configuration, a gradient boosting tree model specifically for the corresponding scenario is trained, which is denoted as the simulation evaluation model.

6. The risk node assessment system for changes in construction engineering projects according to claim 5, characterized in that, Based on the application scenario of the corresponding construction project, the actual data requirements for the changes in the nodes of the construction project are broken down, and then the corresponding basic engineering data and environmental and constraint data are defined. Based on experience fitting of corresponding types of construction project data, the data of the construction project node change to be evaluated is initially fitted. Then, based on industry experience in construction project risk assessment, the data after the initial fitting is optimized a second time to complete the correlation features between the data of the construction project node change to be evaluated, forming a fitted feature dataset of the construction project node change to be evaluated, which is recorded as change-related data.

7. The risk node assessment system for changes in construction engineering projects according to claim 6, characterized in that, The process of calculating the simulated risk quantification value of the change-related nodes of the construction project to be evaluated, based on the change-related data of the nodes and the simulation evaluation model, includes: Input the standardized actual engineering feature matrix of the building project nodes to be evaluated into the trained simulation evaluation model; Using a Bayesian network, the prior probability of actual risk of node change in the construction project to be evaluated is first calculated; then, the prior probability of actual risk is horizontally concatenated and fused with the corresponding actual project feature matrix to obtain the fused actual input feature matrix; based on the gradient boosting tree model, the simulated risk quantification value of node change in the construction project to be evaluated is calculated and recorded as the simulated risk quantification value of the change.

8. The risk node assessment system for changes in construction engineering projects according to claim 7, characterized in that, The steps for classifying and estimating the risk entropy of the simulated risk quantification value to be changed, and calculating the single change risk entropy of the building project node to be evaluated, are as follows: Based on the risk quantification value sample set, and using the kernel density estimation method, the risk probability distribution of changes between two adjacent nodes in the construction project to be evaluated is calculated; Based on the classic information entropy formula, a formula for calculating the entropy value of a single risk entropy specific to changes in building engineering nodes is constructed. Based on the entropy value calculation formula of a single risk entropy, the entropy values ​​of several types of core single change risks are calculated, and then the single change risk entropy set is obtained.

9. The risk node assessment system for changes in construction engineering projects according to claim 8, characterized in that, The steps for classifying and estimating the risk entropy of the simulated risk quantification value to be changed, and calculating the joint change risk entropy of the building project nodes to be evaluated, are as follows: Based on the topology optimization structure of Bayesian networks, the causal coupling relationship between several types of single change risks is identified. Then, based on the experience of construction engineering experts, the mutual information method is used to calculate the coupling degree between any two types of single change risks. Based on the degree of coupling, a node change risk coupling matrix is ​​constructed. The diagonal elements in the matrix represent the self-coupling degree of each individual risk, and the off-diagonal elements represent the mutual coupling degree between two types of risks. Based on the risk probability distribution and the node change risk coupling matrix, the joint probability distribution of several types of single change risks is calculated using the joint probability density function. Based on the joint entropy theory of information entropy and combined with the risk coupling matrix, the joint change risk entropy of several types of single change risks is calculated.

10. The risk node assessment system for changes in construction engineering projects according to claim 9, characterized in that, The process of conducting a comprehensive risk assessment based on the single change risk entropy and combined change risk entropy of the construction project node to be assessed includes: Based on the analytic hierarchy process, the weights of primary indicators and secondary indicators are assigned to the single change risk entropy and the joint change risk entropy. The comprehensive standardized entropy value of a single change risk is obtained by weighting the secondary indicator weights; the corresponding basic risk quantification value is calculated by weighted linear fusion based on the primary indicator weights and the comprehensive standardized entropy value. The basic risk quantification value is calibrated for engineering scenarios and matched with risk levels to obtain the actual change risk quantification value of the corresponding construction project node change, and then the final engineering risk level is matched.

Images