Quality safety control and engineering cost correlation analysis early warning method based on big data

Abstract

The application relates to the technical field of construction engineering quality safety management, and discloses a quality safety control and engineering cost correlation analysis early warning method based on big data. The method comprises the following steps: collecting concrete strength and steel bar protective layer thickness data through a sensor, combining artificial measurement verification to generate an original data set; standardizing treatment and cluster analysis are performed to identify a factor group related to structure stability and corrosion risk; a neural network model is used to predict potential rework cost and calculate a risk influence score; the quality safety factor weight is dynamically adjusted according to the construction stage, and the comprehensive index is optimized in combination with the project type; by extracting the trend change of the optimized index, a real-time cost risk early warning signal is generated by threshold comparison, and the early warning level is determined. The application dynamically adjusts the weight by fusing multi-source data, realizes the quantitative linkage and dynamic early warning of construction quality and cost risk, improves the accuracy and real-time performance of construction quality monitoring, and effectively reduces the cost risk.

Description

Technical Field

[0001] This application relates to the field of construction engineering quality and safety management technology, and in particular to a method for early warning analysis of the correlation between quality and safety control and engineering cost based on big data. Background Technology

[0002] With the continuous expansion of construction projects and the increasing complexity of construction environments, quality and safety control and cost risk management have become two core issues in the construction process. Traditional quality and safety assessment methods mostly rely on manual inspection and experience-based judgment, resulting in limited data sources and insufficient real-time information, making it difficult to comprehensively reflect the structural condition and potential risks. This leads to problems such as untimely identification of corrosion hazards and delayed early warning of cost risks in high-risk scenarios such as basement structures, resulting in rework, project delays, and increased costs.

[0003] In recent years, with the rapid development of big data technology, IoT sensors, and artificial intelligence methods, engineering quality and cost management has gradually entered a stage of intelligentization and data-driven approaches. By collecting key parameters such as concrete strength and rebar cover thickness in real time using sensors, and combining this with manual measurement and verification, high-quality raw datasets can be constructed, providing a reliable foundation for subsequent analysis. At the data processing level, standardization, cluster analysis, and correlation modeling methods are widely used to identify the deep-seated relationship between concrete strength and corrosion risk. However, most existing methods remain at the static analysis stage, lacking the dynamic adaptability to differences in construction stages and project type characteristics.

[0004] Meanwhile, traditional cost risk management methods are mostly based on historical experience and static models, which cannot reflect changes in risks during construction in real time and are difficult to issue early warning signals in a timely manner. This leads to delayed risk control measures and increases the possibility of project rework and cost overruns. Especially in engineering scenarios such as basements, which are greatly affected by factors such as groundwater level and environmental humidity, there is a significant coupling relationship between steel corrosion risk and concrete strength, and static weight allocation models often fail to accurately reflect the actual situation.

[0005] Based on the above problems, there is an urgent need to propose a quality and safety control and cost risk early warning method that integrates big data processing, artificial intelligence modeling and dynamic weight adjustment, so as to realize real-time monitoring, accurate evaluation and early warning of the construction process, thereby effectively reducing rework rate, optimizing resource allocation and ensuring project safety and economic benefits. Summary of the Invention

[0006] The purpose of this invention is to address the shortcomings in existing engineering quality, safety, and cost risk management, such as insufficient real-time performance, inadequate factor correlation analysis, lack of dynamic adaptability in weight adjustments, and delayed early warning. It provides a big data-based method for quality and safety control and engineering cost correlation analysis and early warning. This method enables real-time monitoring and data fusion of key quality factors during construction. Through intelligent algorithms, it establishes a dynamic correlation model between concrete strength and steel reinforcement cover thickness, thereby forming a risk assessment and cost early warning system adaptable to different project types and construction stages, improving the accuracy and real-time nature of construction quality, safety, and cost risk management.

[0007] Firstly, this application provides a method for early warning analysis of the correlation between quality and safety control and engineering cost based on big data, the method comprising: Step 1: Obtain concrete strength data and rebar cover thickness data collected by sensors, and verify the concrete strength data and rebar cover thickness data using manual measuring equipment to obtain the original dataset; Step 2: Standardize the original dataset to obtain a standardized dataset; Step 3: Use a clustering algorithm to group the standardized dataset and determine the association factor groups that include structural stability and corrosion risk; Step 4: Predict the risk impact score by using a neural network model to identify the associated factor groups. Step 5: Calculate the weights of the quality and safety factors based on the construction stage and risk impact score, and use a weighted summation method to integrate the risk impact score and stage differences to obtain a preliminary comprehensive index; Step 6: Adjust the weights of the preliminary comprehensive indicators according to the project type to obtain the optimized comprehensive indicators; Step 7: Extract the changing trend from the optimized comprehensive indicators, use threshold comparison to generate a real-time cost risk early warning signal, and determine the early warning level.

[0008] Compared with the prior art, the beneficial effects of the technical solution of this application are at least as follows: 1. By synchronously collecting and cross-validating sensor data and manual detection data, the accuracy and consistency of the original dataset were ensured. Through standardization and outlier removal, a high-quality and highly comparable data foundation was built, effectively improving the reliability of subsequent analysis.

[0009] 2. By establishing the linkage between clustering algorithms and thresholds, the coupling characteristics of concrete strength and steel reinforcement cover thickness under different construction scenarios were revealed. Deep correlation mining of risk factors can more accurately identify the potential relationship between corrosion risk and structural stability, providing a scientific basis for dynamic risk assessment.

[0010] 3. In the calculation of the preliminary comprehensive index, the construction stage and risk impact scores are introduced as dynamic adjustment factors. In the case of basement structure, the concrete strength factor is combined with the weight of the steel reinforcement protective layer thickness for linkage correction, realizing the dynamic adaptive allocation of factor weights and overcoming the limitations of the static weights of the traditional model.

[0011] 4. By introducing time series analysis and warning threshold comparison mechanisms, the dynamic changing trends of indicators can be captured, and early warnings can be triggered in the early stages of risk signals, which significantly improves the real-time and forward-looking nature of risk prevention and control.

[0012] 5. By combining the dynamic adjustment of quality and safety factors with cost risk early warning, the integration of quality control and cost management is achieved, which can effectively reduce rework rate and delay costs, and improve the overall economic benefits of the project while ensuring construction quality. Attached Figure Description

[0013] To more clearly illustrate the technical solutions of the embodiments of the present invention, the drawings used in the description of the embodiments will be briefly introduced below. Obviously, the drawings described below are some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0014] Figure 1 The flowchart is shown below for the big data-based quality and safety control and engineering cost correlation analysis and early warning method of this application. Figure 2 This is a flowchart of the method for obtaining the associated factor groups in this application. Detailed Implementation

[0015] The terms “first,” “second,” “third,” “fourth,” etc. (if present) in the specification, claims, and accompanying drawings of this application are used to distinguish similar objects and are not necessarily used to describe a particular order or sequence. It should be understood that such data can be interchanged where appropriate so that the embodiments described herein can be implemented in a sequence other than that illustrated or described herein. Furthermore, the terms “comprising” or “having,” and any variations thereof, are intended to cover a non-exclusive inclusion; for example, a process, method, system, product, or 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.

[0016] For ease of understanding, the specific process of the embodiments of this application is described below. Figure 1 The diagram shows a flowchart of an embodiment of the big data-based quality and safety control and engineering cost correlation analysis and early warning method provided by the present invention. The flowchart specifically includes the following steps: Step 1: Obtain concrete strength data and rebar cover thickness data collected by sensors. Verify the concrete strength data and rebar cover thickness data using manual measuring equipment to obtain the original dataset.

[0017] In one specific embodiment, the process of performing step 1 may specifically include the following steps: (1) Use manual measuring equipment to simultaneously verify the concrete strength data and the steel reinforcement cover thickness data to determine the consistency of the measurements; (2) Generate an original dataset containing strength correlation and thickness influence based on the values ​​collected by the sensor and the verification results of the manual measurement equipment; (3) Timestamp the original dataset to ensure the time consistency of the data.

[0018] Specifically, data fusion from sensors and manual measuring equipment enables dual verification of concrete strength and steel reinforcement cover thickness, ensuring the reliability and consistency of the original dataset.

[0019] Sensors continuously collect concrete strength data and rebar cover thickness data at key locations such as walls and floors, recording values ​​at fixed sampling intervals. To avoid the impact of sensor fluctuations in humid or high-load environments on data accuracy, manual measurement equipment is introduced for synchronous verification. Manual measurements are performed at the same locations at the sampling time points, and the obtained concrete strength and rebar cover thickness values ​​are compared with the sensor records. Measurement consistency is determined when the difference between the two types of data is within a set threshold range.

[0020] Based on this, combining sensor values ​​and manual verification results, a raw dataset containing strength correlation and thickness influence is generated. Strength correlation characterizes the relationship between concrete strength and the thickness of the concrete cover, while thickness influence characterizes the impact of insufficient thickness on strength performance. For example, strength correlation is achieved by calculating the statistical correlation between concrete strength data sequences and concrete cover thickness data sequences, such as using the Pearson correlation coefficient. A coefficient close to 1 indicates a positive correlation between increased thickness and increased strength; close to -1 indicates a negative correlation; and close to 0 indicates no significant linear relationship. Thickness influence is quantified by analyzing the regression influence coefficient of the cover thickness on concrete strength. This can be calculated using regression analysis methods, such as establishing a linear regression model with cover thickness as the independent variable and concrete strength as the dependent variable. The regression coefficient of this model is the thickness influence, representing the expected change in concrete strength for every unit increase in cover thickness. For example, a linear regression coefficient of 0.6 indicates that for every 1 mm increase in cover thickness, the concrete strength is expected to increase by 0.6 MPa. In this way, the original dataset not only contains a single numerical record, but also integrates the inherent relationship between the two, so that subsequent cluster analysis can accurately reflect the coupling characteristics between corrosion risk and structural stability when dividing factor groups.

[0021] To ensure data comparability across time, all collected and verified data are timestamped, and sampling times are annotated using a standardized format. This ensures data from different sources are aligned on the same timeline and provides a foundation for subsequent trend analysis, making the time series changes of the optimized comprehensive indicators complete and traceable.

[0022] Through the above process, the sensor hardware features, manual measurement techniques, and data processing algorithms form a functionally mutually supportive relationship. This not only solves the problem of insufficient accuracy caused by the heterogeneity of multi-source data, but also provides high-quality input for subsequent neural network prediction and weight adjustment through embedded correlation calculation methods, thereby achieving the technical effect of improving data reliability and risk identification accuracy.

[0023] Step 2: Standardize the original dataset to obtain a standardized dataset.

[0024] In one specific embodiment, the process of performing step 2 may specifically include the following steps: (1) Adjust the data sampling frequency to a uniform frequency based on the timestamp differences in the original dataset; (2) Convert the different units of the original dataset into a unified unit; (3) Identify the correlation between concrete strength and steel reinforcement cover thickness in the original dataset through cross-validation of data; (4) Remove abnormal data based on correlation and generate a standardized dataset containing intensity correlation and corrosion risk association.

[0025] Specifically, interpolation methods are used to resample data points with inconsistent timestamps, aligning all data at a uniform frequency of once per hour. This ensures data consistency across time during subsequent processing and avoids data bias caused by different sampling times. Furthermore, the different units in the original dataset are converted to a unified unit, such as converting concrete strength from psi to MPa, eliminating data incomparability caused by unit differences and enabling analysis of data from different sources under the same standard.

[0026] Cross-validation was used to identify the correlation between concrete strength and steel reinforcement cover thickness in the original dataset. The Pearson correlation coefficient between the two was calculated to quantify the degree of linear correlation. At the same time, cross-validation was applied to divide the dataset into training and validation sets. A regression model was fitted on the training set to verify the correlation, and the model accuracy was tested on the validation set to confirm the impact of cover thickness on corrosion risk when strength decreases. This step is closely related to subsequent processing and provides a basis for the removal of outlier data.

[0027] A deviation threshold is set based on the correlation coefficient. Data points that deviate from the threshold are removed. For example, if the correlation coefficient deviation of a data point exceeds 0.2, it is marked as an anomaly and removed, ensuring that the remaining data reflects the true correlation between strength and thickness. The removed data is then integrated, and strength correlation and corrosion risk scores are added to the remaining data points, forming a standardized dataset that is time-aligned, uses uniform units, has anomaly removed, and is labeled with strength correlation and corrosion risk association. The corrosion risk score is a composite index calculated based on the strength value and the protective layer thickness. Risk is negatively correlated with the strength value and also negatively correlated with the protective layer thickness (i.e., positively correlated with the inverse of the thickness).

[0028] The above processing solves problems related to data sampling, units, anomalies, and correlation analysis, improving the reliability and usability of the data. It provides accurate data support for subsequent steps such as cluster analysis and risk prediction, and helps to accurately assess the quality, safety, and cost risks of building structures. For example, in scenarios with different basement depths, it can detect potential structural weaknesses early, reduce noise interference, and improve the accuracy of predicting rework costs.

[0029] Step 3: Use a clustering algorithm to group the standardized dataset and determine the association factor groups that include structural stability and corrosion risk.

[0030] In one specific embodiment, the process of performing step 3 may specifically include the following steps: (1) Extract features from the quality and safety factors in the standardized dataset and generate feature vectors; (2) Clustering algorithm is used to group the feature vectors and identify the pattern recognition features between concrete strength and steel reinforcement cover thickness; (3) Based on pattern recognition features, determine the threshold linkage relationship between concrete strength and steel reinforcement cover thickness; (4) Based on the threshold linkage relationship, generate a group of related factors that include structural stability and corrosion risk.

[0031] The flowchart of the method for obtaining related factor groups is as follows: Figure 2 As shown.

[0032] Specifically, quality and safety factors refer to a set of key parameters that can reflect the direct or indirect impact of a building structure on its safety and durability during construction and use. These include, but are not limited to, concrete strength, steel reinforcement cover thickness, material performance indicators, construction process parameters, environmental factors, and structural testing indicators.

[0033] We selected concrete strength and steel reinforcement cover thickness as two core quality and safety factors. By calculating their statistical characteristics (such as mean and standard deviation), the original continuous data was transformed into quantifiable feature vectors. After standardization, these feature vectors eliminated dimensional differences, making the values ​​of different factors comparable and ensuring the effectiveness of the clustering process.

[0034] Clustering algorithms are used to group the generated feature vectors to identify pattern recognition features. By iteratively calculating the distance between vectors and the cluster centers, the data is automatically divided into several pattern groups. These pattern groups reflect the coupling relationship between concrete strength and concrete cover thickness. Specifically, the K-means algorithm is selected (which achieves clustering by iteratively calculating the distance from data points to the cluster centers). The K value is preset to 3 to 5 according to the dataset size, dividing the feature vectors into multiple clusters, such as stable patterns (high strength and sufficient thickness, e.g., strength > 28 MPa and thickness > 50 mm), medium-risk patterns (medium strength but insufficient thickness), and high-risk patterns (low strength and excessively thin thickness). These patterns directly reflect the correlation between concrete strength and concrete cover thickness. For example, in a basement structure monitoring scenario, 100 data points are divided into 3 clusters by K-means: Cluster 1 (average strength 35 MPa, thickness 60 mm) represents a high-stability pattern, Cluster 2 (average strength 25 MPa, thickness 40 mm) represents a medium-risk pattern, and Cluster 3 (average strength 15 MPa, thickness 20 mm) represents a high-risk pattern. This grouping method allows for the early detection of potential problems such as insufficient thickness leading to reduced strength, thereby improving the accuracy of structural assessments.

[0035] Based on the clustering results, pattern recognition features are further extracted. These features are numerical characteristics or statistical relationships extracted from the original engineering inspection data, capable of distinguishing different safety states (such as stable, medium risk, and high risk). For example, they specifically include numerical features, statistical relationships, and pattern labels. Numerical features include: the average concrete strength (e.g., 35 MPa), the average thickness of the steel reinforcement protective layer (e.g., 60 mm), and the statistical correlation between strength and thickness (e.g., Pearson correlation coefficient 0.9) within a given cluster. Statistical relationships include: the strength of the positive correlation between strength and thickness (a correlation coefficient close to 1 indicates a strong positive correlation), and the risk probability under different factor combinations (e.g., the probability of insufficient thickness when strength is below a threshold > 0.7). Pattern labels are: the interpretable safety state classification (e.g., "stable mode," "medium risk mode," and "high risk mode") ultimately formed based on the numerical features and statistical relationships.

[0036] The threshold linkage relationship is determined by calculating the correlation coefficient between strength and thickness in the clustering pattern, and then setting a lower limit for their linkage. For example, when the concrete strength is lower than a certain threshold, the corresponding protective layer thickness must not be lower than another threshold. If it is lower than the threshold, it indicates a potential corrosion risk. For each identified pattern (such as the stable pattern), a lower limit threshold for concrete strength (such as 28 MPa) is set and linked to a minimum threshold for the thickness of the steel reinforcement protective layer (such as 50 mm). By comparing data points within the pattern, when the strength is lower than the threshold, it is checked whether the thickness is synchronously insufficient (such as strength < 28 MPa and thickness < 50 mm), forming a linkage logic of "strength threshold triggering thickness threshold assessment". At the same time, the probability of insufficient thickness when the strength is lower than the threshold is calculated by conditional probability (if > 0.7, it is defined as strong linkage), quantifying the association between structural stability and corrosion risk (such as low strength is often accompanied by a thin protective layer, and the probability of corrosion increases), generating a group of correlation factors that include the relationship between structural stability and corrosion risk.

[0037] This technical solution not only achieves automatic clustering and pattern recognition of complex engineering data, but also transforms the results into actionable factor groups through linkage with thresholds. This realizes a closed-loop process from raw data collection and pattern recognition to risk group generation, highly integrating data-driven algorithm logic with the actual needs of the engineering field. This effectively improves the accuracy of risk identification and the timeliness of early warning, ensuring a dynamic balance between structural safety and cost control. It solves the technical problems of scattered parameters and isolated information in the correlation analysis between quality safety and cost, and has the beneficial effects of improving monitoring accuracy, reducing cost risks, and enhancing risk early warning capabilities.

[0038] Step 4: Predict the associated factor groups using a neural network model to obtain the risk impact score.

[0039] In one specific embodiment, the process of performing step 4 may specifically include the following steps: (1) if the concrete strength in the associated factor group is lower than a preset threshold, the associated factor group is input into a pre-trained neural network model; (2) the potential rework cost is predicted through the neural network model. (3) Calculate the initial risk impact score for assessing the overall risk level based on the ratio of rework cost to baseline cost and in combination with other risk factors in the associated factor group; (4) Normalize the initial risk impact score to generate a standardized risk impact score.

[0040] Specifically, the system first performs conditional judgment. When the concrete strength in any associated factor group is detected to be lower than a preset threshold, the neural network model prediction process is triggered. This ensures that only high-risk scenarios are deeply evaluated, avoiding invalid calculations. When the prediction process is triggered, the system automatically inputs the data of that group into a pre-trained neural network model. The input data includes not only concrete strength values ​​but also the thickness of the steel reinforcement protective layer and related time series parameters, forming an input vector that reflects the coupling relationship between structural stability and corrosion risk. The neural network model adopts a multilayer perceptron structure. The input layer receives a data vector containing elements such as strength values ​​and steel reinforcement protective layer thickness. The hidden layer uses the ReLU activation function to process nonlinear relationships to capture complex correlations. The output layer generates a predicted value for potential rework cost increases. During the training phase, the neural network model has undergone supervised learning based on a historical project dataset. This dataset covers the correspondence between different strength defects and rework costs. The model learns the nonlinear relationships between various factors through backpropagation, enabling it to capture the potential cost increase patterns caused by low strength and insufficient protective layer.

[0041] After the input data enters the model, it undergoes multi-layer feature extraction to output a predicted value of potential rework costs. Subsequently, the predicted rework costs are compared with the baseline cost (such as the total project budget or a single-stage budget) to calculate an initial score. For example, if the predicted cost is 1.2 times the baseline, the initial score is 0.2. Other risk factors from the associated factor group (such as thickness impact, reflecting the cumulative effect of insufficient steel reinforcement cover thickness on risk) are then incorporated. The score is then corrected through weighted summation, ensuring that the score reflects not only the direct risk of insufficient strength but also the indirect impact of structural corrosion risk on cost. For example, the formula for calculating the initial risk impact score is: Where R0 is the initial risk impact score, R1 is the preliminary score, and w f For the concrete strength factor weights, w t The protective layer thickness factor is weighted, with α1 and α2 being weighting coefficients. Alternatively, the initial risk impact score can be calculated using the following formula: Where R0 is the initial risk impact score, R1 is the preliminary score, and wf For the concrete strength factor weights, w t β1, β2, and β3 are the weights of the protective layer thickness factor.

[0042] Finally, the initial risk impact score is normalized by mapping it to the 0-1 range using the min-max normalization method, generating a standardized risk impact score. This standardized score eliminates the impact of cost benchmark differences between different projects or stages and can be directly used for subsequent early warning signal generation, forming a complete chain from prediction to alert.

[0043] The technical solution of this invention has a functionally mutually supportive relationship between the algorithm features and engineering inspection data. That is, the neural network relies on engineering monitoring indicators during the learning process, and the prediction results, in turn, serve engineering quality control and cost early warning, realizing a quantitative mapping between risk and cost, effectively solving the problem of risk identification lag, and having the beneficial effects of improving the accuracy of quality and safety control and reducing the risk of total engineering cost.

[0044] Step 5: Calculate the weights of the quality and safety factors based on the construction stage and risk impact score, and use a weighted summation method to integrate the risk impact score and stage differences to obtain a preliminary comprehensive index.

[0045] In one specific embodiment, the process of performing step 5 may specifically include the following steps: (1) Obtain information on the current construction stage; (2) Determine the initial weights of each quality and safety factor based on the current construction phase information; (3) The initial weights are dynamically adjusted based on the risk impact score to generate the adjusted weights; (4) Using a weighted summation method, the adjusted weights are combined with the risk impact score to generate a preliminary comprehensive index.

[0046] Specifically, the project management system allows for real-time access to information on the current construction stage, such as the foundation pouring stage or the main structure stage, ensuring the timeliness of subsequent weight calculations. This construction stage information serves as the basic input for determining subsequent weights, clarifying the focus of quality and safety factors at different construction stages.

[0047] The initial weights of each quality and safety factor are determined based on information from the current construction stage. For example, during the foundation pouring stage, the initial weight of the concrete strength factor is set to 0.4, the initial weight of the rebar cover thickness factor is set to 0.3, and other factors such as material uniformity are set to 0.3, for a total of 1. Based on the different requirements for quality and safety factors at different construction stages, an initial benchmark is provided for subsequent dynamic adjustments, ensuring that the weight allocation matches the key risks of each construction stage.

[0048] The initial weights are dynamically adjusted based on the risk impact score to generate adjusted weights. For example, if the risk impact score exceeds a preset value (e.g., 0.8), it is considered high. The initial weight of the concrete strength factor is adjusted from 0.4 to 0.5, the rebar cover thickness factor from 0.3 to 0.35, and the remaining factors are correspondingly reduced to emphasize potential corrosion risks. This adjustment can be achieved through linear interpolation, for example, the adjustment formula is: New weight = Initial weight × (1 + Risk impact score × ki), where ki is the weight coefficient of the i-th quality and safety factor, a stage-factor specific sensitivity coefficient derived from historical rework samples. The sensitivity coefficients for different quality and safety factors can be the same or different. Normalization is then applied to ensure the total weight sum remains 1. This dynamic adjustment improves sensitivity to high-risk factors, reduces structural stability hazards, and links the risk impact score to the initial weights, enabling the weights to reflect the actual risk level of different quality and safety factors at the current construction stage.

[0049] Finally, a weighted summation method is used to merge the adjusted weights with the risk impact score to generate a preliminary comprehensive index. For example, the preliminary comprehensive index = ∑(adjusted weights_i * factor values_i) + risk impact score × fusion coefficient, where the fusion coefficient is set to 0.5 based on engineering experience, thereby generating a preliminary comprehensive index that reflects the overall quality and safety level.

[0050] The acquisition of information during the construction phase provides a contextual environment for weight allocation, ensuring that the weight settings align with the actual project conditions. A dynamic adjustment mechanism based on risk impact scores enables the weights to respond in real-time to changes in risk, while the weighted fusion algorithm organically combines static quality indicators with dynamic risk prediction. For example, during the construction phase of the basement main structure, the initial weights are set to strength 0.35 and thickness 0.4. When the risk impact score for a certain area is 0.6, the weights are dynamically adjusted to strength 0.38 and thickness 0.42, with factor values ​​of 28 MPa for strength and 45 mm for thickness. The preliminary comprehensive index is calculated as (0.38*28 + 0.42*45) + 0.6*0.5 = 10.64 + 18.9 + 0.3 = 29.84. This process collectively solves the problem that static weight models are difficult to adapt to changes in the construction phase and the dynamic nature of risks. Through a phase-aware dynamic weight adjustment mechanism, the accuracy and timeliness of quality and safety assessments are improved.

[0051] Step 6: Adjust the weights of the preliminary comprehensive indicators according to the project type to obtain the optimized comprehensive indicators.

[0052] In one specific embodiment, the process of performing step 6 may specifically include the following steps: (1) Obtain information on the type of engineering project; (2) When the project type is a basement structure, first increase the weight of the concrete cover thickness to the preset value, and then adjust it in conjunction with the concrete strength factor; (3) Using the weighted summation method, the adjusted weights are applied to each quality and safety factor of the preliminary comprehensive index to obtain a new preliminary comprehensive index; (4) By using time series trend analysis, the changing trend of the new preliminary comprehensive index is extracted. When the changing trend exceeds the preset warning threshold, the new preliminary comprehensive index is corrected to generate the optimized comprehensive index.

[0053] Specifically, when a project type is identified as "basement," it means that the project is in a high-corrosion-risk environment, requiring strengthened monitoring and assessment of the specific risk of "corrosion risk." Since "reinforcement cover thickness" is a direct proxy variable for corrosion risk, its weight should be adjusted to give it a larger proportion in the comprehensive assessment, thereby making the final optimized comprehensive index more sensitive to corrosion risk.

[0054] Project type information is obtained by querying the engineering database or inputting an interface. When the project type is identified as a basement structure, the weight of the rebar cover thickness is increased from the base value to a preset value (e.g., from 0.3 to 0.5) according to the predefined project type-weight mapping rules. Simultaneously, this is adjusted in conjunction with the concrete strength factor; for example, when the strength is below 30 MPa, the thickness weight coefficient is further multiplied by 1.2 to emphasize the exacerbating effect of insufficient strength on corrosion risk. This coordinated adjustment comprehensively considers the synergistic impact of two key factors on structural safety, improving analysis accuracy. The increased weight helps identify risks early and reduces potential rework costs.

[0055] In the weighted comprehensive evaluation system, the sum of the weight coefficients of each factor is 1 (or 100%). When the weight of the concrete cover thickness is increased, the weights of other factors will decrease. After the weight adjustment is completed, the weighted summation method is used to apply the adjusted weights to each quality and safety factor of the preliminary comprehensive index, and the new preliminary comprehensive index value is recalculated. For example, suppose there are n factors with initial weights of W1, W2, ..., Wn, and a sum of 1. Now, the weight of the k-th factor (concrete cover thickness) is to be adjusted to Wk_new, then its weight change Δ = Wk_new - Wk, and the new weight of the remaining j-th factor is: Wj_new = Wj × (1 - Δ) / (SUM - Wk), where SUM is the sum of the weights of the remaining factors before adjustment.

[0056] The recalculated indicators are optimized using time series trend analysis. The moving average method or exponential smoothing method is used to extract the trend of the indicators over a number of past time stamps. When the rate of change of the trend is detected to exceed the preset warning threshold (e.g., 10%), the indicator value is corrected. For example, if the upward trend exceeds the threshold, the indicator value is increased by 5% to incorporate the potential risk growth. Finally, an optimized comprehensive indicator that reflects the specific risks of the project is generated.

[0057] For example, in a basement construction project, the initial comprehensive index was 75. After identifying the project type as a basement, the thickness weight increased from 0.3 to 0.5. Simultaneously, due to the concrete strength being 28 MPa (lower than 30 MPa), a linkage adjustment was triggered, further increasing the thickness weight to 0.6. The recalculated new index value was 82. Trend analysis showed that this index increased by 12% in the last 5 timestamps, exceeding the 10% warning threshold. Therefore, the index value was corrected to 82 × 1.05 = 86.1, ultimately generating the optimized comprehensive index.

[0058] By introducing project-type specific weight adjustments, the problem of a one-size-fits-all assessment standard is solved, making risk assessment more aligned with engineering realities. The weight linkage mechanism enhances the sensitivity of key risk factors and improves the targeted nature of early warnings. Time series trend analysis enables the capture of dynamic changes in risks, making the assessment results more forward-looking. These improvements collectively address the technical issues of insufficient flexibility in the assessment model and its failure to fully consider project-type specificity and the temporal changes in risks, providing a more accurate decision-making basis for engineering cost risk early warning.

[0059] Step 7: Extract the changing trend from the optimized comprehensive indicators, use threshold comparison to generate a real-time cost risk early warning signal, and determine the early warning level.

[0060] In one specific embodiment, the process of performing step 7 may specifically include the following steps: (1) Extract the time series change trend from the optimized comprehensive index; (2) Determine whether the trend of the time series changes exceeds the preset warning threshold. If so, generate a real-time cost risk warning signal. (3) Determine the final warning level based on the strength of the warning signal.

[0061] Specifically, the process begins by extracting time-series trends from optimized comprehensive indicators. This step arranges the comprehensive indicators in chronological order, forming a sequence reflecting the overall changes in concrete strength and steel reinforcement cover thickness. The least squares method is used to fit the curves of indicator values ​​over time, and their slopes are calculated to quantify the trend strength. The calculated trend slope is then compared with a preset warning threshold. When the trend slope exceeds the threshold, a real-time cost risk warning signal is generated, and the signal strength level is determined based on the degree of exceedance. Finally, the signal strength level is mapped to a predefined warning level table, fine-tuned according to the current construction stage, and the final warning level is output. For example, when the trend slope of 0.6 exceeds the threshold of 0.5, a medium-strength signal is generated, which is mapped to an orange warning level during the foundation construction stage, triggering an enhanced monitoring command.

[0062] Time series trend analysis fully utilizes the temporal characteristics of the optimized comprehensive index to capture the dynamic evolution of risks; the threshold comparison mechanism transforms continuous trend quantities into discrete early warning signals, while the intensity mapping rule directly links the engineering risk level with management response measures. For example, in basement wall monitoring, if the optimized comprehensive index drops from 85 to 70 over three consecutive time points with a trend slope of 0.75, exceeding the threshold of 0.5, a high-intensity signal is generated. Combined with the fact that the project is in a critical pouring stage, it is ultimately determined to be at the red warning level, triggering a comprehensive inspection.

[0063] This technical solution enables early detection and tiered response to cost risks through trend prediction and a tiered early warning mechanism, thereby improving the initiative and sophistication of risk management.

[0064] The above-described embodiments are only used to illustrate the technical solutions of this application, and are not intended to limit them. Although this application has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that modifications can still be made to the technical solutions described in the foregoing embodiments, or equivalent substitutions can be made to some of the technical features. Such modifications or substitutions do not cause the essence of the corresponding technical solutions to deviate from the spirit and scope of the technical solutions of the embodiments of this application.

Claims

1. A method for quality and safety control and engineering cost correlation analysis and early warning based on big data, characterized in that, include: Step 1: Obtain concrete strength data and rebar cover thickness data collected by sensors, and verify the concrete strength data and rebar cover thickness data using manual measuring equipment to obtain the original dataset; Step 2: Standardize the original dataset to obtain a standardized dataset; Step 3: Use a clustering algorithm to group the standardized dataset and determine the association factor groups containing structural stability and corrosion risk; the clustering algorithm selected is the K-means algorithm, and the K value is preset to 3 to 5 according to the size of the dataset; Step 3 includes: extracting features from the quality and safety factors in the standardized dataset to generate feature vectors; grouping the feature vectors using a clustering algorithm to identify pattern recognition features between concrete strength and steel reinforcement cover thickness; determining the threshold linkage relationship between concrete strength and steel reinforcement cover thickness based on the pattern recognition features; and generating the association factor group containing the correlation between structural stability and corrosion risk based on the threshold linkage relationship. Step 4: Predict the risk impact score by using a neural network model to analyze the associated factor group; Step 4 includes: if the concrete strength in the associated factor group is lower than a preset threshold, then inputting the associated factor group into a pre-trained neural network model; predicting potential rework costs through the neural network model; calculating an initial risk impact score for assessing the overall risk level based on the ratio of the rework cost to the baseline cost and in combination with other risk factors in the associated factor group; and normalizing the initial risk impact score to generate a standardized risk impact score. Step 5: Calculate the weights of the quality and safety factors based on the construction stage and the risk impact score, and use a weighted summation method to integrate the risk impact score and stage differences to obtain a preliminary comprehensive index; Step 6: Adjust the weights of the preliminary comprehensive indicators according to the project type to obtain the optimized comprehensive indicators; Step 6 includes: obtaining project type information; when the project type is a basement structure, first increasing the weight of the steel reinforcement protective layer thickness to a preset value, and then adjusting it in conjunction with the concrete strength factor; using a weighted summation method, applying the adjusted weights to each quality and safety factor of the preliminary comprehensive index to obtain a new preliminary comprehensive index; using a time series trend analysis method, extracting the changing trend of the new preliminary comprehensive index; when the changing trend exceeds a preset warning threshold, correcting the new preliminary comprehensive index to generate the optimized comprehensive index; Step 7: Extract the changing trend from the optimized comprehensive index, use threshold comparison to generate a real-time cost risk early warning signal, and determine the early warning level.

2. The method as described in claim 1, characterized in that, Step 1 includes: The concrete strength data and the steel reinforcement protective layer thickness data were simultaneously verified using manual measuring equipment to ensure measurement consistency. Based on the values ​​collected by the sensors and the verification results of the manual measurement equipment, the original dataset containing strength correlation and thickness influence is generated; The original dataset is timestamped to ensure data consistency over time.

3. The method as described in claim 2, characterized in that, Step 2 includes: Based on the timestamp differences in the original dataset, adjust the data sampling frequency to a uniform frequency; convert the different units of the original dataset into a uniform unit; By cross-validating the data, the correlation between concrete strength and steel reinforcement cover thickness in the original dataset was identified; Abnormal data are removed based on the correlation, and a standardized dataset containing intensity correlation and corrosion risk association is generated.

4. The method as described in claim 1, characterized in that, Step 5 includes: Obtain information on the current construction phase; Based on the current construction phase information, determine the initial weights of each quality and safety factor; The initial weights are dynamically adjusted based on the risk impact scores to generate adjusted weights. The weighted summation method is used to integrate the adjusted weights with the risk impact score to generate the preliminary comprehensive index.

5. The method as described in claim 1, characterized in that, Step 7 includes: Extract the time series change trend from the optimized comprehensive index; Determine whether the trend of the time series exceeds a preset warning threshold; if so, generate a real-time cost risk warning signal. The final warning level is determined based on the strength of the warning signal.

Images