Engineering building construction quality analysis system and method based on data processing
By constructing a dynamic causal relationship graph and combining multi-objective optimization and closed-loop feedback correction, the problem of insufficient in-depth mining of causal relationships in engineering construction quality management is solved. This enables the forward-looking identification and intelligent pre-control of quality risks, and improves the refinement of management and the data-driven level of decision-making.
Patent Information
- Application Number
- CN202511386816.7
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2025-09-26
- Publication Date
- 2026-02-17
AI Technical Summary
Existing technologies lack the ability to deeply mine the complex causal relationships between multiple data sources in engineering construction quality management. This leads to decisions on quality issues relying on experience, making it difficult to achieve forward-looking identification and integrated optimization, and making it difficult to effectively avoid major quality risks caused by the coupling of multiple factors.
By constructing a dynamic causal relationship graph and acquiring process flow networks and multi-source monitoring data from the building information model, the system performs causal relationship graph construction, risk profile construction, multi-objective optimization, and quality pre-control instruction generation. Combined with multi-objective optimization and closed-loop feedback correction, it achieves forward-looking identification and collaborative intelligent pre-control of engineering quality risks.
It enables the revelation of quantitative causal relationships between materials, processes, environment, and quality outcomes, improving the data-driven level of decision-making and the sophistication of management. It allows for the prediction and quantification of potential defects before they occur, reducing rework costs and project delays, and forming a sustainable intelligent management process.
Smart Images

Figure CN121544090A_ABST
Abstract
Description
TECHNICAL FIELD
[0001] The present application relates to the technical field of business data processing and intelligent decision-making, and particularly relates to an engineering construction quality analysis system and method based on data processing. BACKGROUND
[0002] Engineering construction quality analysis is a core link in project management, which is directly related to the safety, service life and investment benefit of buildings. In modern business practice, engineering project management increasingly relies on data-driven decision support systems, aiming to optimize resource allocation, control costs, shorten construction period and ensure engineering quality through processing and analyzing a large amount of data generated in the whole life cycle of the project, so as to maximize business value.
[0003] In the prior art, engineering construction quality management usually adopts a digital management platform based on a building information model, and combines a field sensor network to collect various monitoring data. Managers monitor the construction progress, material consumption and part of the quality detection indexes through these platforms. Some use statistical methods to analyze the collected data, and trigger an alarm when some indexes exceed the threshold value, to assist in quality control decision-making.
[0004] However, the prior art solutions still have obvious deficiencies in data processing and decision support. They mostly stay in isolated monitoring and post-response of single or a few data dimensions, and lack the ability to deeply mine the complex causal relationships among multi-source data. The data are often fragmented, and it is difficult to form a complete evidence chain to reveal the root cause of quality problems. Therefore, when quality defects occur, the decision often depends on the personal experience of engineers, lacks forward-looking risk prediction and integrated optimization decision-making ability, resulting in low quality management efficiency and difficulty in effectively avoiding potential major quality risks caused by multi-factor coupling. SUMMARY
[0005] To solve the above problems, the present application provides an engineering construction quality analysis system and method based on data processing, which uses a dynamic causal relationship map to perform risk deduction and simulation, and combines multi-objective optimization and closed-loop feedback correction, so as to realize forward-looking identification and collaborative intelligent pre-control of engineering quality risks.
[0006] The above object can be achieved by the following scheme: The engineering construction quality analysis system based on data processing comprises a causal relationship graph construction module, which is used for acquiring and aligning a process flow network in a building information model and multi-source monitoring data, the multi-source monitoring data comprising material parameters, process parameters, quality detection indexes and environmental parameters, taking process time sequence dependence and specification constraints as causal direction priors, statistically analyzing weight parameters of each edge for estimation and dynamic updating, and constructing a causal relationship graph; a risk profile construction module, which is used for performing reverse deduction of a causal path on the causal relationship graph, identifying a critical parameter combination, taking the critical parameter combination as simulation input, performing forward simulation of boundary conditions on the causal relationship graph, verifying and quantifying cascading failure effects and influence range, and constructing a forward-looking quality risk profile; a multi-objective collaborative optimization module, which is used for taking the cascading failure effects in the forward-looking quality risk profile as a core optimization objective function, and extracting resource and process dependence from the causal relationship graph as constraint conditions, to obtain a Pareto optimal balance solution set in a multi-dimensional decision space; a quality pre-control instruction generation module, which is used for selecting a target strategy from the Pareto optimal balance solution set, and combining a real-time state of a node in the causal relationship graph to instantiate the target strategy into a dynamic quality pre-control instruction; an efficiency evaluation and attribution correction module, which is used for executing the dynamic quality pre-control instruction, comparing field effect data with a prediction result of the forward-looking quality risk profile, and performing contribution degree traceability analysis on the causal relationship graph to update a corrected causal link weight to the causal relationship graph; an evolution audit and causal knowledge distillation module, which is used for performing version control and difference analysis on previous weight updates of the causal relationship graph, distilling a core causal rule set from causal links of the causal relationship graph, combining the core causal rule set with attribution correction records of each version graph to generate an evolution audit report.
[0007] Optionally, the causal relationship graph construction module comprises a process data fusion unit, which is used for taking a process flow network in the building information model as a structured skeleton, mounting the multi-source monitoring data as attribute information to the process flow network according to timestamps and spatial coordinates, and generating a time sequence attribute network; a causal structure learning unit, which is used for applying causal inspection on the time sequence attribute network with the process time sequence dependence and the specification constraints as hard constraint conditions, verifying and determining directed causal relationships among the material parameters, process parameters, quality detection indexes and environmental parameters, and generating an initial causal topology structure; and a dynamic edge weight estimation unit, which is used for establishing a dynamic regression algorithm with source node parameters as independent variables and target node parameters as dependent variables for the initial causal topology structure, performing online iterative updating according to the multi-source monitoring data, and obtaining a causal relationship graph.
[0008] Optionally, the risk profile construction module comprises: a quality failure definition unit configured to parse engineering acceptance specifications and material performance standards in the building information model, and convert quality requirement texts into a quality failure vector; a causal chain reverse solving unit configured to take the quality failure vector as a deduction endpoint, and perform reverse traversal of a causal path on the causal relationship graph to solve a critical parameter combination; and a boundary forward simulation unit configured to take the critical parameter combination as an initial disturbance condition, and perform time sequence forward propagation simulation on the causal relationship graph to verify and quantify cascading failure effects and influence ranges, aggregate all simulation results, and construct a forward-looking quality risk profile.
[0009] Optionally, the critical parameter combination comprises a process parameter set and a material parameter set, wherein: the process parameter set is used to define quality defects triggered under extreme non-standard operation, including construction time, maintenance condition, equipment operation threshold, and personnel operation process; and the material parameter set is used to define quality defects triggered under substandard material state, including material ratio, batch strength, environmental adaptability, and aging degree.
[0010] Optionally, the multi-objective collaborative optimization module comprises: a target and constraint definition unit configured to convert cascading failure effects in the forward-looking quality risk profile into a risk cost function, extract process logic dependency and resource time and space mutual exclusion relationship from the causal relationship graph, and formalize them into nonlinear inequality constraints to jointly constitute a multi-objective optimization space; and a Pareto front solving unit configured to collaboratively optimize the risk cost function, project duration, and resource consumption in a feasible region of the multi-objective optimization space, iteratively search, and generate a Pareto optimal balance solution set.
[0011] Optionally, the system further comprises: associating each balance solution in the Pareto optimal balance solution set with the critical parameter combination, and simulating to verify intervention effectiveness of the balance solution to generate a strategy risk cause mapping matrix.
[0012] Optionally, the quality pre-control instruction generation module comprises: a target strategy selection unit configured to calculate and select an optimal balance solution from the Pareto optimal balance solution set as an initial target strategy; an instruction refinement matching unit configured to match a refined target strategy with first intervention effectiveness from a strategy cluster of the initial target strategy based on the strategy risk cause mapping matrix; and an instruction instantiation generation unit configured to call a real-time state snapshot of the causal relationship graph, acquire current parameter values in the refined target strategy, and fill the real-time parameter values into the refined target strategy to instantiate and generate a dynamic quality pre-control instruction.
[0013] Optionally, the performance evaluation and attribution correction module comprises: a performance deviation quantification unit configured to continuously track field effect data after the dynamic quality pre-control instruction is issued, and compare the field effect data with predicted results in the prospective quality risk profile to quantitatively obtain performance deviation; a causal contribution degree attribution unit configured to take the performance deviation as a starting point of analysis, trace contribution degree on the causal relationship graph, and identify and calculate a corrected causal link; and a graph online correction unit configured to gradient correct the corrected causal link according to the contribution degree, and update to the causal relationship graph.
[0014] Optionally, the evolution audit and causal knowledge distillation module comprises: a graph evolution path audit unit configured to record and store weight update events of the causal relationship graph in different versions, and compare differences between graph structures and weight parameters of different versions to build a causal evolution path; and a core causal rule distillation unit configured to analyze convergence of the causal evolution path, extract and generalize a core causal rule set, and generate an evolution audit report in combination with the attribution correction record.
[0015] Based on the same inventive concept, the application further provides a data processing-based engineering construction quality analysis method, which comprises: acquiring and aligning a process flow network in a building information model and multi-source monitoring data, the multi-source monitoring data comprising material parameters, process parameters, quality detection indexes and environmental parameters, taking process time sequence dependence and specification constraints as causal direction priors, statistically analyzing weight parameters of each edge to estimate and dynamically update the weight parameters, and constructing a causal relationship graph; performing reverse deduction of a causal path on the causal relationship graph, identifying a critical parameter combination, taking the critical parameter combination as simulation input, performing forward simulation of boundary conditions on the causal relationship graph, verifying and quantifying cascade failure effects and influence ranges, and constructing a prospective quality risk profile; taking the cascade failure effects in the prospective quality risk profile as a core optimization objective function, extracting resource and process dependence relationships from the causal relationship graph as constraint conditions, and solving a Pareto optimal balance solution set in a multi-dimensional decision space; selecting a target strategy from the Pareto optimal balance solution set, and instantiating the target strategy into a dynamic quality pre-control instruction in combination with real-time states of nodes in the causal relationship graph; executing the dynamic quality pre-control instruction, comparing field effect data with predicted results of the prospective quality risk profile, performing contribution degree traceability analysis on the causal relationship graph, and updating a corrected causal link weight to the causal relationship graph; performing version control and difference analysis on weight updates of the causal relationship graph, distilling a core causal rule set from causal links of the causal relationship graph, combining the core causal rule set with attribution correction records of each version of the graph, and generating an evolution audit report.
[0016] Compared with the prior art, the present application has the following advantages: 1、The present application realizes the deep fusion and internal logic mining of multi-source heterogeneous data in the building construction process by constructing a dynamic evolution causal relationship map. Compared with the traditional method which relies on experience or isolated data analysis, the present application can reveal the quantitative causal relationship between materials, processes, environment and quality results, providing a scientific basis for the root tracing and accurate positioning of quality problems, greatly improving the data-driven level of decision-making and the fine degree of management; 2、The present application proposes a forward-looking risk assessment mechanism based on causal reasoning and forward simulation. By identifying the critical risk source through reverse tracing and simulating its cascading failure effect, the potential quality defects can be preformed and quantified before they occur, realizing the fundamental change from passive post-inspection to active pre-control. This forward-looking risk management mode significantly enhances the project's ability to resist uncertain factors and effectively reduces the rework cost and schedule delay caused by quality problems; 3、The present application establishes a complete intelligent management process from risk identification, multi-objective optimization decision to closed-loop feedback correction. By integrating quality risk, project duration and resource consumption into a unified optimization framework, a Pareto optimal strategy set is generated to provide scientific and multi-dimensional decision support for managers. At the same time, the performance evaluation and attribution correction module ensures continuous learning and self-evolution from actual execution effect, ensuring the long-term effectiveness and adaptability of the model and decision-making, forming a virtuous cycle of sustainable improvement.
[0017] Other features and advantages of the present application will be set forth in the following description, and in part will become apparent to those skilled in the art from the description, or can be learned by practice of the present application. The objects and other advantages of the present application will be realized and attained by the structure particularly pointed out in the description and claims hereof as well as the appended drawings. BRIEF DESCRIPTION OF DRAWINGS
[0018] In order to more clearly illustrate the technical solutions in the embodiments of the present application or the prior art, the following will briefly introduce the drawings needed to be used in the embodiment or prior art description. Obviously, the drawings in the following description are some embodiments of the present application, and for those skilled in the art, other drawings can also be obtained without creative labor on the basis of these drawings.
[0019] Figure 1 is the framework diagram of the engineering building construction quality analysis system based on data processing of the embodiment of the present application.
[0020] Figure 2 is the structural schematic diagram of the engineering building construction quality analysis system based on data processing of the embodiment of the present application.
[0021] Figure 3 is a causal relationship map structure schematic diagram of an embodiment of the present application.
[0022] Figure 4 is a time sequence probability distribution cloud rain map of risk evolution deduction of an embodiment of the present application.
[0023] Figure 5 is a multi-target collaborative optimization decision space schematic diagram of an embodiment of the present application. DETAILED DESCRIPTION
[0024] In order to make the purpose, technical scheme and advantages of the embodiments of the present application clearer, the technical scheme in the embodiments of the present application will be clearly and completely described below with reference to the drawings in the embodiments of the present application. Obviously, the described embodiments are part of the embodiments of the present application, rather than all the embodiments. Based on the embodiments in the present application, all other embodiments obtained by those skilled in the art without creative labor fall within the scope of protection of the present application.
[0025] With reference to Figure 1 , one embodiment of the present application proposes an engineering construction quality analysis system and method based on data processing, which adopts a dynamic causal relationship map to perform risk deduction and simulation, and combines multi-target optimization and closed-loop feedback correction, so as to realize forward-looking identification and collaborative intelligent pre-control of engineering quality risks.
[0026] As Figure 2 shown, the system of the embodiment specifically includes: A causal relationship map construction module is configured to acquire and align a process flow network in a building information model and multi-source monitoring data, the multi-source monitoring data including material parameters, process parameters, quality detection indexes and environmental parameters, to take process time sequence dependence and specification constraints as causal direction priors, to statistically analyze weight parameters of each edge for estimation and dynamic update, and to construct a causal relationship map. A risk profile construction module is configured to perform reverse deduction of causal paths on the causal relationship map, to identify a critical parameter combination, to take the critical parameter combination as simulation input, to perform forward simulation of boundary conditions on the causal relationship map, to verify and quantify cascading failure effects and influence range, and to construct a forward-looking quality risk profile. A multi-target collaborative optimization module is configured to take cascading failure effects in the forward-looking quality risk profile as a core optimization objective function, to extract resource and process dependence relationships from the causal relationship map as constraint conditions, and to solve a Pareto optimal balance solution set in a multi-dimensional decision space. a quality pre-control instruction generation module configured to select a target strategy from the Pareto optimal balance solution set and, in combination with real-time states of nodes in the causal relationship graph, instantiate the target strategy as a dynamic quality pre-control instruction; an effectiveness evaluation and attribution correction module configured to execute the dynamic quality pre-control instruction, compare field effect data with a prediction result of the prospective quality risk profile, perform a contribution degree traceability analysis on the causal relationship graph, and update a corrected causal link weight to the causal relationship graph; an evolution audit and causal knowledge distillation module configured to perform version control and difference analysis on previous weight updates of the causal relationship graph, distill a core causal rule set from causal links of the causal relationship graph, combine the core causal rule set with attribution correction records of each version of the graph, and generate an evolution audit report.
[0027] The construction of a dynamic causal relationship graph for risk deduction and simulation, in combination with multi-objective optimization and closed-loop feedback correction, can realize the prospective identification and collaborative intelligent pre-control of engineering quality risks.
[0028] Optionally, the causal relationship graph construction module comprises: a process data fusion unit configured to take a process flow network in the building information model as a structured skeleton, mount the multi-source monitoring data as attribute information to the process flow network according to timestamps and spatial coordinates, and generate a time-series attribute network; Specifically, the process data fusion unit first acquires a process flow network in the building information model (BIM), which defines the sequence and logical dependency of each construction process in the form of a directed graph as a structured skeleton for subsequent data fusion. Subsequently, the unit collects real-time multi-source monitoring data on site, which covers material parameters, process parameters, quality detection indexes and environmental parameters, and each piece of data contains a timestamp and spatial coordinates. The unit accurately mounts each piece of monitoring data to the corresponding process node in the process flow network as the dynamic attribute of the node. For example, the concrete temperature data collected at a specific time point and location will be associated with the concrete curing process node that is being performed at that time. This process converts the static process flow network into a time-series attribute network containing process and state information.
[0029] a causal structure learning unit configured to apply causal testing on the time-series attribute network with the process time-series dependency and the specification constraints as hard constraint conditions, test and determine the directed causal relationship between the material parameters, process parameters, quality detection indexes and environmental parameters, and generate an initial causal topology structure; Specifically, this causal structure learning unit performs causal discovery based on the temporal attribute network generated by the previous unit. This step uses recognized engineering process temporal dependencies and industry standard constraints as inviolable hard constraints, which predefine irreversible causal directions. For example, the quality of rebar tying inevitably affects the structural strength of subsequent concrete pouring, but the reverse is not true. Within this hard constraint framework, this unit applies causal testing algorithms, such as the Granger causality test, to analyze the attribute time series of each node in the temporal attribute network to test and determine whether there is a statistically significant directed causal relationship between the parameters. Through this testing process, the unit ultimately generates an initial causal topology in the form of a directed acyclic graph (DAG).
[0030] The dynamic edge weight estimation unit is used to establish a dynamic regression algorithm for the initial causal topology, with source node parameters as independent variables and target node parameters as dependent variables, and to perform online iterative updates based on the multi-source monitoring data to obtain a causal relationship map.
[0031] Specifically, the dynamic edge weight estimation unit establishes a dynamic regression algorithm for each directed edge in the initial causal topology to quantify its influence strength. In this embodiment, the algorithm is a Bayesian Dynamic Linear Model (BDLM). This model can not only estimate the magnitude of the causal effect but also quantify the uncertainty of the estimation results. For any target node, its state equation and observation equation can be expressed as: , In the formula, The value observed at the target node at time t is the quality inspection index. for The vector of source node parameter observations at time t, i.e., parameters such as materials, processes, and environment; for The state vector at time t is composed of the core element of the causal link weights to be estimated. To observe the noise, its variance Known; This is the state transition matrix, usually set as the identity matrix to represent the stationary evolution of the weights; The state evolution noise has a covariance matrix. The drastic change of weights over time is controlled, and these weights are preset based on historical data or expert experience. This unit utilizes the Kalman filter algorithm to adjust the state vector based on real-time multi-source monitoring data. The posterior distribution is updated iteratively online, and the updated posterior mean is taken as the weight parameter of the corresponding edge in the causal graph, as shown in Figure 3 The structure of the causal relationship graph is shown in the figure, which shows the coupling between the physical layer as the spatial reference and the causal layer representing the dependency relationship.
[0032] Optionally, the risk profile construction module comprises: A quality failure definition unit is configured to parse the engineering acceptance specifications and material performance standards in the building information model, and convert the quality requirement text into a quality failure vector. Specifically, the quality failure definition unit aims to convert the textual engineering specifications into machine-processable structured data. The unit automatically parses the engineering acceptance specifications and material performance standard documents contained in the BIM, extracts the key indicators and their quantitative thresholds related to quality requirements, such as concrete compressive strength, steel bar spacing error, etc. through natural language processing technology. Subsequently, the unit organizes these key indicators into a multi-dimensional vector, i.e. the quality failure vector. Each dimension of the vector corresponds to a quality indicator, and its domain is the allowed deviation range between the actual measured value and the standard required value of the indicator. When the quality detection result of a certain process makes any or multiple components of the vector exceed the allowed range, it is defined as a quality failure.
[0033] A causal chain reverse solving unit is configured to take the quality failure vector as the end of deduction, and perform reverse traversal of the causal path on the causal relationship graph to solve the critical parameter combination. Specifically, the causal chain reverse solving unit takes the final unqualified result represented by the quality failure vector as the target endpoint of deduction. The unit starts from the result node representing the final quality detection indicator on the previously constructed causal relationship graph, and traverses the path in the reverse direction of the directed edge. This traversal process will trace back all its direct and indirect upstream influencing factor nodes, i.e. process parameters, material parameters and environmental parameters nodes. By analyzing the weight parameters of each causal link in the graph, the unit can calculate the contribution of different upstream nodes to the final quality failure, and identify the specific value set of upstream input parameters that are most likely to cause quality failure. These sets are defined as critical parameter combinations.
[0034] A boundary forward simulation unit is configured to take the critical parameter combination as the initial disturbance condition, perform time-series forward propagation simulation on the causal relationship graph, verify and quantify the cascading failure effect, impact range, aggregate all simulation results, and construct a forward-looking quality risk profile.
[0035] Specifically, the boundary forward simulation unit takes the critical parameter combination solved by the previous unit as the initial disturbance condition of the simulation, and inputs it into the corresponding source node of the causal graph. Then, the unit performs a time-forward propagation simulation on the graph, that is, according to the directed link of the causal relationship, starting from the input end, gradually calculating and propagating the influence of parameter changes on downstream nodes. This process simulates how the quality problem evolves and amplifies with the progress of the construction process under critical conditions. Through this simulation, the unit can verify that the identified critical parameter combination indeed leads to the expected quality failure, and accurately quantify the cascading failure effect and its impact range. The unit simulates all identified critical parameter combinations and statistically aggregates all simulation results to finally build a comprehensive and forward-looking quality risk profile, as shown in Figure 4 The time sequence probability distribution cloud rain map of risk evolution and deduction in the present application, which shows the dynamic evolution process of the probability distribution of key risk indicators over time.
[0036] Optionally, the critical parameter combination includes a process parameter set and a material parameter set, wherein: The process parameter set is used to define quality defects that will be triggered under extreme non-standard operation, including construction time, curing condition, equipment operation threshold, and personnel operation process. Specifically, for the process parameter set, it aims to quantify the risks introduced by improper operation behavior or environmental control during the construction process. The parameters in this set are derived from the construction plan and operation procedures, and are defined by mapping the standard requirements and the boundary conditions of extreme non-standard operation. For example, the construction time parameter is used to determine whether the key time interval requirements have been followed, such as the minimum curing period before formwork removal; the curing condition includes a set of environmental monitoring values, such as temperature and humidity, with critical values set in the range that may cause abnormal concrete hydration; the equipment operation threshold involves key working parameters of construction machinery, such as the vibration frequency of a vibrator, with a critical value corresponding to an operation setting that may cause insufficient structural density; the personnel operation process is represented by a series of quantifiable step sequences, with a critical state corresponding to the omission or sequence disorder of key procedures.
[0037] The material parameter set is used to define quality defects that will be triggered under substandard material state, including material proportioning, batch strength, environmental adaptability, and aging degree.
[0038] Specifically, for the material parameter set, it is used to quantify the potential risks brought by the fact that the properties of the raw materials or semi-finished products do not meet the design requirements. The parameters in this set are derived from material performance standards and quality inspection reports. For example, the material ratio parameter directly reflects the proportion of each component of the mixture, and its critical value is set in the non-standard ratio range that will significantly affect the performance of the final product; the batch strength records the actual mechanical property test results of the material in a specific batch, and its critical value is the value lower than the design strength requirement; the environmental adaptability is used to describe the performance stability of the material in a specific site environment, and its critical value corresponds to the environmental condition under which the performance decreases sharply; and the aging degree quantifies the performance degradation of the material due to long storage time, and its critical value is defined as the state in which the material performance is lower than the lower limit of usability.
[0039] Optionally, the multi-objective collaborative optimization module comprises: a target and constraint definition unit, configured to convert the cascading failure effect in the prospective quality risk profile into a risk cost function, extract the logical dependency between processes and the resource time-space mutual exclusion relationship from the causal relationship graph, and formalize them into nonlinear inequality constraints, to jointly constitute a multi-objective optimization space; Specifically, the target and constraint definition unit is responsible for the mathematical formalization of the optimization problem. The unit first receives the quantified cascading failure effect in the prospective quality risk profile, and converts it into a calculable risk cost function through a risk valuation function. The risk cost function maps the cascading failure effect to the expected value of economic loss. At the same time, the unit automatically extracts the logical dependency between processes and the sharing or exclusive use demand of construction activities on resources from the previously constructed causal relationship graph. These relationships are formalized as a series of nonlinear inequality constraints. These constraints and the risk cost function jointly define the solution space of the multi-objective optimization problem.
[0040] a Pareto frontier solving unit, configured to collaboratively optimize the risk cost function, project duration and resource consumption in the feasible region of the multi-objective optimization space, and iteratively search and generate a set of Pareto optimal balance solutions.
[0041] Specifically, the Pareto frontier solving unit iteratively searches in the multi-dimensional objective space constituted by the risk cost function, the total project duration and the total resource consumption, and in the constraint feasible region defined by the previous unit. Since the three objectives are usually in conflict with each other, there is no single optimal solution. The unit uses the non-dominated sorting genetic algorithm (NSGA-II) to gradually approach and finally obtain a set of Pareto optimal balance solutions by constantly generating, evaluating and selecting new candidate solutions. Each solution in this solution set is a Pareto optimal solution, meaning that it is impossible to optimize any of the objectives without deteriorating any other objective, such as Figure 5As shown, it is a multi-objective collaborative optimization decision space diagram in the application, which shows the trade-off relationship between the risk cost, project duration and resource consumption three optimization objectives and the Pareto optimal frontier.
[0042] Optionally, the system further comprises: Each balance solution in the Pareto optimal balance solution set is associated with the critical parameter combination, and intervention effectiveness of the balance solution is simulated and verified, to generate a strategy risk cause mapping matrix.
[0043] Specifically, to establish a clear correspondence between the optimal strategy and the specific risk source that can be intervened, the process first performs association analysis on each balance solution in the Pareto optimal balance solution set, that is, the process matches the process parameters and resource allocation adjustment suggestions contained in each solution with the previously identified critical parameter combination one by one, to identify the risk cause that each balance solution mainly targets. Subsequently, for each associated “balance solution-critical parameter combination” pair, the process performs a new forward simulation to verify the intervention effectiveness, that is, in the simulation with the critical parameter combination as the initial condition, the intervention measures defined by the balance solution are applied synchronously, and by comparing the simulation results before and after the intervention, the inhibition effect of the balance solution on the specific risk is quantitatively evaluated. Finally, the process integrates all simulation verification results to generate a strategy risk cause mapping matrix. The rows of the matrix represent each balance solution, the columns represent all critical parameter combinations, and the element values in the matrix are the quantified intervention effectiveness values of the corresponding balance solution on the corresponding critical parameter combination, thereby clearly revealing the applicable scenarios and effectiveness of each alternative optimization strategy.
[0044] Optionally, the quality pre-control instruction generation module comprises: A target strategy selection unit is configured to calculate and select an optimal balance solution from the Pareto optimal balance solution set as an initial target strategy. Specifically, the target strategy selection unit receives the Pareto optimal balance solution set generated by the multi-objective collaborative optimization module. To select an optimal balance solution from the solution set, the unit adopts a decision preference function. The function allocates different weights to the three optimization objectives of risk cost, project duration and resource consumption according to the core management objectives of the current stage of the project. By calculating the weighted comprehensive score of each balance solution, the unit selects the solution with the optimal score as the initial target strategy, which provides a macroscopic guidance direction for subsequent instruction generation.
[0045] An instruction refinement matching unit is configured to match a refined target strategy with the first intervention effectiveness from a strategy cluster of the initial target strategy based on the strategy risk cause mapping matrix. Specifically, the instruction refinement matching unit first acquires the root critical parameter combination corresponding to the activated risk under the current working condition. Then, the unit queries the strategy risk cause mapping matrix generated in the previous step, in which the performance of the initial target strategy on all the critical parameter combinations that it can intervene is analyzed. The unit identifies the specific adjustment measure in the strategy cluster corresponding to the initial target strategy that has the highest intervention performance on the currently activated critical parameter combination, and determines it as the refined target strategy. This step ensures that the final generated instruction can accurately target the root cause of the current quality risk.
[0046] The instruction instantiation generation unit is configured to call the real-time state snapshot of the causal relationship graph, acquire the current parameter value in the refined target strategy, and fill the real-time parameter value into the refined target strategy to generate a dynamic quality pre-control instruction.
[0047] Specifically, the instruction instantiation generation unit receives the refined target strategy matched by the previous unit. To make it fully executable, the unit calls the causal relationship graph to acquire the real-time state snapshot of the nodes involved in the refined target strategy. For example, if the refined target strategy is "improve concrete curing temperature", the unit will query the temperature sensor readings in the corresponding location in real time. Then, the unit fills these real-time parameter values into the instruction template of the refined target strategy, and finally instantiates a dynamic quality pre-control instruction that contains specific execution objects, quantitative thresholds, and operation timing, which can be immediately executed.
[0048] Optionally, the performance evaluation and attribution correction module comprises: The performance deviation quantification unit is configured to continuously track the field effect data after the dynamic quality pre-control instruction is issued, and compare it with the predicted results in the prospective quality risk profile to quantitatively obtain a performance deviation. Specifically, the performance deviation quantification unit starts after the dynamic quality pre-control instruction is issued and executed. The unit continuously tracks the field effect data of each quality detection index and process parameter related to the instruction. Then, the unit accurately compares the collected field effect data with the simulation prediction results when constructing the prospective quality risk profile. By calculating the difference between the actual value and the predicted value, the unit quantitatively obtains a performance deviation, which truly reflects the gap between the actual effect and the expected effect of the pre-control measure.
[0049] The causal contribution degree attribution unit is configured to take the performance deviation as the starting point for analysis, trace the contribution degree on the causal relationship graph, and identify and calculate the corrected causal link. Specifically, the causal contribution attribution unit takes the performance deviation calculated by the previous unit as the starting point of analysis, traces the contribution degree on the causal relationship graph to identify the root cause of the deviation. The unit uses a contribution degree analysis algorithm to start from the node where the performance deviation occurs and propagates the deviation signal in reverse along the causal relationship graph. During the propagation process, the unit calculates the contribution degree of each causal link to the final deviation according to the weight of each causal link, thereby locating those causal links whose weight parameters are most inconsistent with the actual situation and calculating the corrected causal link weight.
[0050] The graph online correction unit is used to correct the corrected causal link according to the contribution degree and update the causal relationship graph.
[0051] Specifically, the graph online correction unit updates the causal relationship graph online according to the results obtained by the previous unit. The unit determines the step size and direction of weight update according to the contribution degree of each identified causal link to be corrected. This update process uses a gradient correction method, which fine-tunes the weight in the direction that reduces the prediction error. The corrected causal link weight is updated in real time to the causal relationship graph, replacing the original weight value, thereby completing an iteration learning based on actual feedback.
[0052] Optionally, the evolution audit and causal knowledge distillation module comprises: The graph evolution path audit unit is used to record and store the weight update events of the causal relationship graph in different versions, and to compare the differences between the graph structures and weight parameters of different versions to build a causal evolution path. Specifically, the graph evolution path audit unit records and stores each weight update event of the causal relationship graph when the weight is updated due to performance evaluation and attribution correction. Each update generates a new graph version, and records the time, triggering event, and weight change details before and after the update. By comparing the differences between these graph versions arranged in chronological order, the unit can clearly track the historical change trajectory of any causal link weight. Connecting these historical update events together, a complete causal evolution path that records the entire process of model cognition evolution over time is constructed.
[0053] The core causal rule distillation unit is used to analyze the convergence of the causal evolution path, extract and generalize a core causal rule set, and generate an evolution audit report in combination with the attribution correction record.
[0054] Specifically, the core causal rule distillation unit performs convergence analysis on the causal evolution path constructed by the previous unit, aiming to extract those causal relationships that tend to be stable and have significant impact after multiple iterations of correction from specific causal link weights with noise. To quantify this stability, the unit calculates a rule stability index for each causal link, and the formula is: , wherein, is the rule stability of the causal link from node to node ; is the average weight of the link in the last observation period ; is the standard deviation of the weight of the link in the period , representing its volatility; is a smoothing constant to prevent the denominator from being zero; is the total number of times the link is updated; is a decay factor, which is set according to expert experience to ensure that the rule is verified for a sufficient number of times. The unit identifies causal links with a rule stability higher than a pre-set threshold as core causal links, and generalizes them into a set of human-understandable core causal rules. Finally, the unit combines the extracted core causal rule set with the detailed record of the previous attribution correction to generate a structured evolution audit report.
[0055] Based on the same inventive concept, the present application also provides an engineering construction quality analysis method based on data processing, which comprises: Obtaining and aligning the process flow network in the building information model and the multi-source monitoring data, the multi-source monitoring data including material parameters, process parameters, quality detection indicators and environmental parameters, using process timing dependence and specification constraints as causal direction priors, statistically analyzing the weight parameters of each edge to estimate and dynamically update them, and constructing a causal relationship graph; Performing causal path reverse deduction on the causal relationship graph, identifying critical parameter combinations, using the critical parameter combinations as simulation inputs, performing boundary condition forward simulation on the causal relationship graph, verifying and quantifying the cascading failure effect and impact range, and constructing a forward-looking quality risk profile; Taking the cascading failure effect in the forward-looking quality risk profile as the core optimization objective function, and extracting resource and process dependence relationships from the causal relationship graph as constraint conditions, solving in a multi-dimensional decision space to obtain a set of Pareto optimal balance solutions; Selecting a target strategy from the set of Pareto optimal balance solutions, and combining the real-time state of the nodes in the causal relationship graph to instantiate the target strategy as a dynamic quality pre-control instruction; The dynamic quality pre-control instruction is executed, field effect data is compared with the prediction result of the prospective quality risk profile, contribution degree traceability analysis is performed on the causal relationship map, and the updated causal link weight is updated to the causal relationship map; The version control and difference analysis of the weight update of the causal relationship map are performed, the core causal rule set is distilled from the causal link of the causal relationship map, the core causal rule set is combined with the attribution correction record of each version map, and an evolution audit report is generated.
[0056] In order to verify the feasibility and effectiveness of the application, the application is applied to the main structure construction stage of a certain large complex construction project. The construction quality requirements of the project for concrete structure, steel bar engineering and the like are very high, and the traditional management mode is difficult to analyze the quality defects caused by multi-factor coupling.
[0057] In this embodiment, the causal relationship map construction module is started first. The process data fusion unit takes the process flow network defined in the building information model, such as "steel binding → formwork support → concrete pouring → maintenance", as a structured skeleton, and performs spatio-temporal matching and mounting on the multi-source monitoring data collected on site to generate a time series attribute network. Subsequently, the causal structure learning unit applies a causal test algorithm to analyze the statistical causal relationship between parameters under the hard constraint of industry specifications, and generates an initial causal topology structure. Finally, the dynamic edge weight estimation unit performs online iterative update to quantitatively estimate the edge weight of each causal link, and obtains a complete causal relationship map.
[0058] Based on the map, the risk profile construction module is started. The quality failure definition unit analyzes the engineering acceptance specification and defines "concrete compressive strength deficiency" as a quality failure vector. The causal chain reverse solving unit performs reverse traversal with this as the terminal point, and identifies a high-risk critical parameter combination, which is divided into a material parameter set and a process parameter set. The boundary forward simulation unit takes the critical parameter combination as the initial condition for forward simulation, and the result shows that the combination will cause a cascading failure effect, and finally the probability of the 28-day strength of the concrete in the region being lower than the design standard is as high as 85%, thereby constructing a prospective quality risk profile.
[0059] To deal with this risk, the multi-objective collaborative optimization module is started, the target and constraint definition unit converts the cascading failure effect into a risk cost function, and defines optimization targets and constraints such as construction period and resource consumption. The Pareto frontier solving unit iteratively searches to generate a Pareto optimal balance solution set containing multiple strategies. Subsequently, each balance solution is simulated and verified to generate a strategy risk cause mapping matrix.
[0060] The quality pre-control instruction generation module is then started. The target strategy selection unit selects the scheme A with the lowest risk cost as the initial target strategy from the Pareto optimal balance solution set according to the “quality first” principle. The instruction refinement matching unit queries the strategy risk cause mapping matrix and finds that the measure “replace cement” in the scheme A has the highest intervention efficiency on “insufficient cement stability”, and refines it as the refined target strategy. Finally, the instruction instantiation generation unit calls the real-time state snapshot of the causal relationship graph to generate the dynamic quality pre-control instruction.
[0061] After the instruction is executed, the performance deviation quantification unit of the performance evaluation and attribution correction module continuously tracks and finds that there is a 5% performance deviation between the actual strength and the prediction. The causal contribution degree attribution unit traces back and locates the deviation in the causal link of “cement hydration rate”. The graph online correction unit then performs gradient correction on the link weight and updates it to the causal relationship graph.
[0062] Finally, the evolutionary audit and causal knowledge distillation module performs version control and difference analysis on the graph updates of this and previous times, distills the core causal rule set of “the hydration rate of a certain brand of cement increases by 8% under specific temperature and humidity” from the long-term stable causal link, and generates an evolutionary audit report by combining the attribution correction record, providing valuable knowledge assets for future projects.
[0063] Table 1 Construction quality risk identification and simulation data representation example Table 2 Multi-objective optimization and pre-control instruction generation data representation example Table 3 Closed-loop feedback and causal relationship graph update data representation example From the above Tables 1-3, it can be seen that the present application exhibits excellent analysis and pre-control capability in complex construction scenarios. Table 1 clearly shows that through causal reasoning and simulation, hidden risks such as “R002” caused by multi-factor coupling are successfully identified. Table 2 records the multi-objective optimization process, providing a scientific trade-off basis for decision-makers, and the final generated instruction is specific and executable. Table 3 verifies the closed-loop learning capability, which can adaptively correct the causal relationship graph by comparing the deviation between prediction and reality, which makes the model accuracy continuously improve with the increase of application time.
[0064] It should be noted that the above formulae can be converted into unitless standard values or parameters of the same dimension that can be superimposed by using the dimensional consistency principle and mathematical standardization methods (for example, normalization processing, dimensionless parameter conversion, or unit system unification), so as to eliminate the interference of different dimensions on the operation logic, make the formula have mathematical operation rationality and objective law adaptability while preserving the original data distribution characteristics. It is a conventional technical means, and will not be described here. The electrical connection between the above-mentioned units does not necessarily mean direct connection, and indirect connection can also be used as long as the purpose of the application is achieved. The above-mentioned is only an exemplary embodiment of the application, and cannot limit the scope of the application.
[0065] That is, any equivalent changes and modifications made according to the teachings of the present application are still within the scope of the present application. Other embodiments of the present application will be readily apparent to those skilled in the art upon considering the description and practice of the principles disclosed herein. The present application is intended to cover any variations, uses, or adaptive changes of the present application that follow the general principles of the present application and include common knowledge or conventional technical means in the art that are not described in the present application.
Claims
1. A construction quality analysis system based on data processing, characterized in that, The system includes: The causal relationship graph construction module is used to acquire and align the process flow network and multi-source monitoring data in the building information model. The multi-source monitoring data includes material parameters, process parameters, quality inspection indicators and environmental parameters. The module uses process sequence dependence and specification constraints as causal direction priors, statistically analyzes the weight parameters of each side for estimation and dynamic updating, and constructs the causal relationship graph. The risk profile construction module is used to perform causal path backward deduction on the causal relationship graph, identify critical parameter combinations, use the critical parameter combinations as simulation input, perform boundary condition forward simulation on the causal relationship graph, verify and quantify cascade failure effects and impact range, and construct a forward-looking quality risk profile. The multi-objective collaborative optimization module is used to take the cascading failure effect in the forward-looking quality risk profile as the core optimization objective function, and extract the resource and process dependency relationship from the causal relationship map as the constraint condition, and solve for the Pareto optimal equilibrium solution set in the multi-dimensional decision space. The quality pre-control instruction generation module is used to select a target strategy from the Pareto optimal equilibrium solution set and, in combination with the real-time status of the nodes in the causal relationship graph, instantiate the target strategy into a dynamic quality pre-control instruction. The performance evaluation and attribution correction module is used to execute the dynamic quality pre-control instructions, compare the on-site effect data with the prediction results of the prospective quality risk profile, perform contribution tracing analysis on the causal relationship graph, and update the corrected causal link weights to the causal relationship graph. The evolution audit and causal knowledge distillation module is used to perform version control and difference analysis on the weight updates of the causal relationship graph, distill the core causal rule set from the causal links of the causal relationship graph, and combine the core causal rule set with the attribution correction records of each version of the graph to generate an evolution audit report.
2. The engineering construction quality analysis system based on data processing according to claim 1, characterized in that, The causal relationship mapping construction module includes: The process data fusion unit is used to use the process flow network in the building information model as a structured skeleton, and to attach the multi-source monitoring data as attribute information to the process flow network according to timestamps and spatial coordinates to generate a time-series attribute network. The causal structure learning unit is used to apply causal testing on the temporal attribute network with the process time dependence and the specification constraints as hard constraints, to test and determine the directed causal relationship between the material parameters, process parameters, quality inspection indicators and environmental parameters, and to generate an initial causal topology. The dynamic edge weight estimation unit is used to establish a dynamic regression algorithm for the initial causal topology, with source node parameters as independent variables and target node parameters as dependent variables, and to perform online iterative updates based on the multi-source monitoring data to obtain a causal relationship map.
3. The engineering construction quality analysis system based on data processing according to claim 1, characterized in that, The risk profile construction module includes: The quality failure definition unit is used to parse the engineering acceptance specifications and material performance standards in the building information model and convert the quality requirement text into a quality failure vector. The causal chain reverse solution unit is used to perform a reverse traversal of the causal path on the causal relationship graph with the quality failure vector as the deduction endpoint, and solve for the critical parameter combination. The boundary forward simulation unit is used to take the critical parameter combination as the initial perturbation condition, perform time-series forward propagation simulation on the causal relationship map, verify and quantify the cascade failure effect and the scope of influence, and aggregate all simulation results to construct a forward-looking quality risk profile.
4. The engineering construction quality analysis system based on data processing according to claim 3, characterized in that, The critical parameter combination includes: a set of process parameters and a set of material parameters, wherein: The process parameter set is used to define quality defects that may be triggered under extremely non-standard operation, including construction time, maintenance conditions, equipment operating thresholds and personnel operation procedures; The set of material parameters is used to define quality defects that will be triggered under substandard material conditions, including material ratio, batch strength, environmental adaptability, and degree of aging.
5. The engineering construction quality analysis system based on data processing according to claim 1, characterized in that, The multi-objective collaborative optimization module includes: The objective and constraint definition unit is used to transform the cascading failure effect in the prospective quality risk profile into a risk cost function, extract the process logic dependency and resource spatiotemporal mutual exclusion relationship from the causal relationship graph, and formalize them into nonlinear inequality constraints, which together constitute a multi-objective optimization space. The Pareto front solver unit is used to collaboratively optimize the risk cost function, project duration, and resource consumption within the feasible region of the multi-objective optimization space, and iteratively search and generate the Pareto optimal equilibrium solution set.
6. The engineering construction quality analysis system based on data processing according to claim 5, characterized in that, The system also includes: Each equilibrium solution in the Pareto optimal equilibrium solution set is correlated with the critical parameter combination, and the intervention effectiveness of the equilibrium solution is verified by simulation, generating a strategy risk cause mapping matrix.
7. The engineering construction quality analysis system based on data processing according to claim 6, characterized in that, The quality pre-control instruction generation module includes: The target strategy selection unit is used to calculate and select the optimal equilibrium solution from the Pareto optimal equilibrium solution set as the initial target strategy. The instruction refinement matching unit is used to match the refined target strategy with the highest intervention effectiveness from the strategy cluster of the initial target strategy based on the strategy risk cause mapping matrix. The instruction instantiation generation unit is used to call the real-time state snapshot of the causal relationship graph, obtain the current parameter value in the refinement target strategy, and fill the real-time parameter value into the refinement target strategy to instantiate and generate dynamic quality pre-control instructions.
8. The engineering construction quality analysis system based on data processing according to claim 1, characterized in that, The performance evaluation and attribution correction module includes: The performance deviation quantification unit is used to continuously track the on-site performance data after the dynamic quality pre-control command is issued, and compare it with the prediction results in the forward-looking quality risk profile to quantify the performance deviation. The causal contribution attribution unit is used to take the performance deviation as the starting point of analysis, trace the contribution on the causal relationship map, and identify and calculate the corrected causal links. The online graph correction unit is used to perform gradient correction on the corrected causal links according to the contribution level and update the causal relationship graph.
9. The engineering construction quality analysis system based on data processing according to claim 1, characterized in that, The evolution audit and causal knowledge distillation module includes: The graph evolution path auditing unit is used to record and store the weight update events of the causal relationship graph in a versioned manner, and to compare the differences in graph structure and weight parameters of different versions to construct the causal evolution path. The core causal rule distillation unit is used to perform convergence analysis on the causal evolution path, extract and generalize to form a core causal rule set, and generate an evolution audit report in combination with the attribution correction record.
10. A method for analyzing the quality of engineering construction based on data processing, applied to the engineering construction quality analysis system based on data processing as described in any one of claims 1-9, characterized in that, The method includes: Acquire and align the process flow network and multi-source monitoring data in the building information model. The multi-source monitoring data includes material parameters, process parameters, quality inspection indicators and environmental parameters. Using process sequence dependence and specification constraints as causal direction priors, statistically analyze the weight parameters of each side for estimation and dynamic updating, and construct a causal relationship graph. The causal path is reversed on the causal relationship graph to identify critical parameter combinations. The critical parameter combinations are used as simulation inputs to perform forward simulation of boundary conditions on the causal relationship graph to verify and quantify the cascade failure effects and their scope of influence, and to construct a forward-looking quality risk profile. The cascading failure effect in the forward-looking quality risk profile is used as the core optimization objective function, and the resource and process dependency relationship is extracted from the causal relationship graph as the constraint condition. The Pareto optimal equilibrium solution set is obtained by solving in the multi-dimensional decision space. The target strategy is selected from the Pareto optimal equilibrium solution set and, in combination with the real-time status of the nodes in the causal relationship graph, the target strategy is instantiated into a dynamic quality pre-control instruction. Execute the dynamic quality pre-control command, compare the on-site effect data with the prediction results of the forward-looking quality risk profile, perform contribution tracing analysis on the causal relationship graph, and update the corrected causal link weights to the causal relationship graph; Version control and difference analysis are performed on each weight update of the causal relationship graph. The core causal rule set is distilled from the causal links of the causal relationship graph. The core causal rule set is combined with the attribution correction records of each version of the graph to generate an evolution audit report.
Citation Information
Cited By
Full-process digital intelligent analysis system for environmental impact evaluation information
CN121810014A
Quality safety management method and system for water conservancy construction engineering
CN122155441A
