Building construction quality prediction method and system based on data analysis

Abstract

The invention relates to the technical field of prediction analysis, in particular to a building construction quality prediction method and system based on data analysis, and the method comprises the following steps: collecting data of material hardness, water absorption rate, environment temperature and humidity from a construction site, cleaning the data, formatting the data into a unified standard, calculating an average value, a standard deviation and a skewness coefficient, and calculating a prediction result; and generating data statistical characteristics. According to the method, the deep insight of the complex data relation of the construction site is provided by collecting the data of the construction site in real time, using the Gaussian process model and adjusting the kernel function parameters, so that the accuracy and reliability of the model are optimized, the model is adjusted in real time through the incremental learning method, synchronous updating of the model and the construction site data is guaranteed, and the construction efficiency is improved. The real-time performance and adaptability of prediction are improved, in addition, more comprehensive risk assessment is provided for a construction team through the use of the confidence interval, early warning of potential quality problems is achieved, and the structural safety risk and maintenance cost are reduced.

Description

Technical Field

[0001] The present invention relates to the technical field of prediction and analysis, and in particular to a method and system for predicting building quality based on data analysis. Background Art

[0002] The field of predictive analytics involves using historical data, statistical algorithms, and machine learning techniques to predict the likelihood of future events. This technology field is widely used in multiple industries, including finance, healthcare, retail, manufacturing, and construction. The core goal is to make more accurate predictions and decisions by analyzing existing data sets to identify potential trends, patterns, and relationships. This analysis can help organizations optimize resource allocation, improve operational efficiency, and reduce risks.

[0003] Among them, the construction quality prediction method refers to the use of predictive analysis technology to estimate the quality problems that may occur during the construction of a construction project. This method can be based on past construction data, engineering material characteristics, construction environment conditions and other factors to analyze and predict the quality, safety and compliance of the structure. Its main purpose is to identify possible quality defects in advance, so that the construction team can take timely measures to prevent or reduce these problems, ensure the structural safety and engineering quality of the building, reduce maintenance costs and extend the service life of the building.

[0004] Existing prediction and analysis technologies usually rely on static models and historical data, lacking the ability to respond to immediate changes on the construction site, resulting in delayed and inaccurate prediction results. This static prediction model is difficult to adapt to dynamic changes on the construction site, such as rapid changes in material properties and environmental conditions, which affects the accurate assessment of building quality. In addition, the uncertainty of the prediction results cannot be effectively expressed, making the risk assessment incomplete, which may lead to safety accidents and increase maintenance costs, thus affecting the service life and economic benefits of the building. Summary of the invention

[0005] The purpose of the present invention is to solve the shortcomings existing in the prior art and to propose a construction quality prediction method and system based on data analysis.

[0006] In order to achieve the above object, the present invention adopts the following technical solution: a method for predicting construction quality based on data analysis, comprising the following steps:

[0007] S1: Collect data on material hardness, water absorption, ambient temperature, and humidity from the construction site, clean the data and format it to a unified standard, calculate the mean, standard deviation, and skewness coefficient, and generate data statistical features;

[0008] S2: Based on the statistical characteristics of the data, radial basis functions and Matérn functions of Gaussian processes are selected as kernel functions, the length scale and variance parameters of the kernel functions are modified, and five-fold cross validation is applied to determine the optimal parameter settings for constructing a Gaussian process model and generating a preliminary configuration of the model;

[0009] S3: using the preliminary configuration of the model, collecting the current parameter input in real time for online updating, adjusting the current parameters through the incremental learning method and testing the influence of the current parameters on the model, maintaining the synchronous update of the model parameters and the construction site data, and generating the model update status;

[0010] S4: updating the state of the model, performing quality prediction, calculating the predicted value and 95% confidence interval of each prediction point, expressing the uncertainty range of the prediction result with the confidence interval, and generating the construction prediction result;

[0011] S5: According to the construction prediction results, an abnormality detection threshold is set, the construction quality is monitored, data points exceeding the threshold are identified, warning signals are recorded and issued, and abnormality monitoring results are generated.

[0012] The data statistical characteristics specifically include the mean, standard deviation, and skewness coefficient; the model preliminary configuration specifically refers to the length scale and variance parameter of the kernel function; the model update status specifically includes the current parameter input, incremental learning method, and synchronous update status; the construction prediction results specifically include the predicted value and 95% confidence interval of each prediction point; the abnormal monitoring results specifically include the abnormal detection threshold, data points exceeding the threshold, and early warning signals.

[0013] As a further solution of the present invention, the step of acquiring the data statistical features is specifically:

[0014] S111: Collect material hardness data, water absorption data, ambient temperature and humidity data at the construction site, unify the format and create initial data records to obtain a standardized data set;

[0015] S112: Performing data verification on the standardized data set, removing outliers exceeding a preset threshold, and filling in missing values ​​to obtain a verified data set;

[0016] S113: Extract key data points from the verified data set, and calculate the mean, standard deviation and skewness coefficient of each data point using the formula:

[0017] ;

[0018] ;

[0019] ;

[0020] Get the statistical characteristics of the data;

[0021] in, Represents the total number of data points, indicating the number of elements in the data set, Represents a single data point value, which is any of hardness, water absorption, ambient temperature or humidity, and is used for statistical analysis. Represents the arithmetic mean, which is used to reflect the center position of the data set. represents the sample standard deviation, represents the coefficient of skewness.

[0022] As a further solution of the present invention, the steps of obtaining the preliminary configuration of the model are specifically as follows:

[0023] S211: Based on the statistical characteristics of the data, radial basis functions and Matérn functions of Gaussian processes are selected as core calculation models, initial parameters including length scale and variance are set, and an initial configuration of the kernel function is generated;

[0024] S212: Apply the kernel function initial configuration to perform a five-fold cross validation, perform performance evaluation on each parameter configuration, calculate the average validation error, determine the best performing parameter, and obtain a validation result;

[0025] S213: Based on the verification result, the kernel function configuration is updated using the optimal length scale and variance parameter settings to construct a Gaussian process model using the formula:

[0026] ;

[0027] Calculate the model kernel value and generate the preliminary configuration of the model;

[0028] in, and Represents two data points, used to calculate the kernel function value between each other, represents the variance parameter, which affects the degree of variation in the model output. Represents the length scale of the kernel function, determines the sensitivity of the distance between data points, and adjusts the matching of the model to the input space scale.

[0029] As a further solution of the present invention, the step of acquiring the model update status is specifically as follows:

[0030] S311: using the model preliminary configuration, collecting data from the construction site in real time, including material hardness, water absorption, ambient temperature and humidity, standardizing the data to adapt to the model input requirements, and generating real-time standardized data;

[0031] S312: inputting the real-time standardized data into the configured Gaussian process model, updating the model parameters using the incremental learning method, reflecting the impact of the new data on the model in real time, and obtaining the update result of the model parameters;

[0032] S313: Based on the update result of the model parameters, the performance test of the model is performed using the formula:

[0033] ;

[0034] Adjust and optimize model parameters, synchronize models with real-time data, and generate model update status;

[0035] in, Indicates the parameters before updating. represents the target parameter value derived based on the new data, is to adjust the rate to ensure a smooth transition, is the new parameter value after adjustment.

[0036] As a further solution of the present invention, the steps of obtaining the construction prediction results are specifically as follows:

[0037] S411: using the parameters of the model update state, performing quality prediction on each data point in the construction data set, calculating the predicted value in real time, and generating a predicted value list for each point based on the output of the model;

[0038] S412: Applying a statistical method to each prediction point in the prediction value list, calculating a 95% confidence interval, and estimating the upper and lower bounds of each prediction value, to generate a confidence interval list;

[0039] S413: Integrate the predicted value list with the corresponding confidence interval list, using the formula:

[0040] ;

[0041] Express the 95% confidence interval for each prediction point and generate construction prediction results;

[0042] in, represents the predicted value of each data point, is the standard error, is the sample size, is the z-value of the standard normal distribution corresponding to the confidence level.

[0043] As a further solution of the present invention, the step of obtaining the abnormal monitoring result is specifically:

[0044] S511: according to the construction prediction result, setting an abnormality detection threshold, determining a benchmark for monitoring construction quality, and generating a threshold setting result;

[0045] S512: Using the set threshold, each of the threshold setting results is monitored in real time, and each data point is calculated to see whether it exceeds the threshold, using the formula:

[0046] ;

[0047] Determine the severity of the data points exceeding the threshold and generate a list of data points exceeding the threshold;

[0048] in, represents the predicted value, is the set threshold, Indicates the magnitude of the superthreshold, calculated by and The difference between The ratio provides a standardized assessment of suprathreshold data points;

[0049] S513: Record and analyze the list of data points exceeding the threshold value, issue a warning signal for each data point exceeding the threshold value, integrate monitoring results and warning signals, mark and record events that require attention, and generate abnormal monitoring results.

[0050] A construction quality prediction system based on data analysis, wherein the construction quality prediction system based on data analysis is used to execute the above-mentioned construction quality prediction method based on data analysis, and the system comprises:

[0051] The data collection and feature construction module collects material hardness, water absorption, ambient temperature, and humidity data from the construction site, cleans the data, and calculates the mean, standard deviation, and skewness coefficient to generate a construction feature summary;

[0052] The model configuration and verification module sets the kernel function parameters of the Gaussian process model based on the construction feature summary, including the length scale and variance parameters of the radial basis function and the Matérn function, and adjusts the parameters using five-fold cross validation to generate the optimal model configuration;

[0053] The model online update module uses the optimal model configuration to process the newly collected construction data in real time, applies the incremental learning method to update the model parameters online, keeps synchronization with the construction site data, and generates the real-time status of the model;

[0054] The quality prediction and monitoring module predicts the construction quality through the real-time status of the model, calculates the predicted value and 95% confidence interval of each prediction point, records the uncertainty range of the prediction results, sets the anomaly detection threshold according to the prediction results, identifies and records the data points that exceed the threshold, issues a warning signal, and generates quality monitoring results.

[0055] Compared with the prior art, the advantages and positive effects of the present invention are:

[0056] In the present invention, the accuracy of predicting building quality is improved by collecting data from the construction site in real time and performing statistical feature analysis. The use of Gaussian process models and adjustment of kernel function parameters provide deep insights into the complex data relationships at the construction site, thereby optimizing the accuracy and reliability of the model. The real-time adjustment of the model achieved by the incremental learning method ensures that the model is updated synchronously with the construction site data, improving the real-time and adaptability of the prediction. In addition, through the use of confidence intervals, a more comprehensive risk assessment is provided to the construction team, achieving early warning of potential quality problems, and reducing structural safety risks and maintenance costs. BRIEF DESCRIPTION OF THE DRAWINGS

[0057] Figure 1 It is a schematic diagram of the workflow of the present invention;

[0058] Figure 2 A flow chart of the steps for obtaining the statistical features of data of the present invention;

[0059] Figure 3 A flowchart of the steps for obtaining the preliminary configuration of the model of the present invention;

[0060] Figure 4 A flowchart of the steps for obtaining the update status of the model of the present invention;

[0061] Figure 5 A flowchart of the steps for obtaining the construction prediction results of the present invention;

[0062] Figure 6 The figure is a flow chart of the steps for obtaining abnormal monitoring results of the present invention. DETAILED DESCRIPTION

[0063] In order to make the purpose, technical solution and advantages of the present invention more clearly understood, the present invention is further described in detail below in conjunction with the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are only used to explain the present invention and are not intended to limit the present invention.

[0064] In the description of the present invention, it should be understood that the terms "length", "width", "up", "down", "front", "back", "left", "right", "vertical", "horizontal", "top", "bottom", "inside", "outside" and the like indicate positions or positional relationships based on the positions or positional relationships shown in the drawings, and are only for the convenience of describing the present invention and simplifying the description, rather than indicating or implying that the device or element referred to must have a specific orientation, be constructed and operated in a specific orientation, and therefore cannot be understood as limiting the present invention. In addition, in the description of the present invention, "multiple" means two or more, unless otherwise clearly and specifically defined.

[0065] Embodiment 1

[0066] See also Figure 1 The present invention provides a technical solution: a method for predicting construction quality based on data analysis, comprising the following steps:

[0067] S1: Collect data on material hardness, water absorption, ambient temperature, and humidity from the construction site, clean the data and format it to a unified standard, calculate the mean, standard deviation, and skewness coefficient, and generate data statistical features;

[0068] S2: Based on the statistical characteristics of the data, the radial basis function and Matérn function of the Gaussian process are selected as the kernel function, the length scale and variance parameters of the kernel function are modified, and the optimal parameter settings are determined by applying five-fold cross validation to construct the Gaussian process model and generate the preliminary configuration of the model;

[0069] S3: Use the initial configuration of the model, collect the current parameter input in real time for online update, adjust the current parameters through the incremental learning method and test the impact of the current parameters on the model, maintain the synchronous update of the model parameters and the construction site data, and generate the model update status;

[0070] S4: Update the model status, make quality predictions, calculate the predicted value and 95% confidence interval of each prediction point, use the confidence interval to represent the uncertainty range of the prediction results, and generate the construction prediction results;

[0071] S5: According to the construction prediction results, set the anomaly detection threshold, monitor the construction quality, identify the data points that exceed the threshold, record and issue early warning signals, and generate abnormal monitoring results.

[0072] The data statistical characteristics specifically include the mean, standard deviation, and skewness coefficient. The model preliminary configuration specifically refers to the length scale and variance parameter of the kernel function. The model update status specifically includes the current parameter input, incremental learning method, and synchronous update status. The construction prediction results specifically include the predicted value and 95% confidence interval of each prediction point. The abnormal monitoring results specifically include the abnormal detection threshold, data points exceeding the threshold, and early warning signals.

[0073] See also Figure 2 , the specific steps for obtaining data statistical features are:

[0074] S111: Collect material hardness data, water absorption data, ambient temperature and humidity data at the construction site, unify the format and create initial data records to obtain a standardized data set;

[0075] When collecting material hardness data, water absorption data, ambient temperature and humidity data at the construction site, it is first necessary to determine the collection methods of multiple data, including material hardness which can be measured on-site by a hardness tester, water absorption which needs to be measured through a standard water absorption test, and ambient temperature and humidity which need to be recorded with a thermometer and hygrometer. Through these operations, the accuracy of the data is ensured and it meets the standardized requirements of subsequent processing. After completing the collection and recording of the data, the unified format processing is to standardize these differentiated types of data into the same data format, which can facilitate data storage, management and subsequent data analysis.

[0076] S112: Perform data verification on the standardized data set, remove outliers that exceed a preset threshold, fill in missing values, and obtain a verified data set;

[0077] After obtaining the standardized data set, data verification is performed. This step involves eliminating outliers in the data set that exceed the preset threshold and filling in missing values. During the data verification process, reasonable thresholds need to be set, which are defined based on industry standards or empirical data from previous projects. Hardness values ​​or water absorption rates that exceed the normal range are considered abnormal, and environmental data including abnormal fluctuations in temperature and humidity also need to be identified and processed. By eliminating outliers, the quality and reliability of the data set can be ensured. Filling in missing values ​​involves interpolation or other data filling techniques to ensure the integrity of the data set.

[0078] S113: Extract key data points from the validated data set and calculate the mean, standard deviation, and skewness coefficient of each data point using the formula:

[0079] ;

[0080] ;

[0081] ;

[0082] Get the statistical characteristics of the data;

[0083] in, Represents the total number of data points, indicating the number of elements in the data set, Represents a single data point value, which is any of hardness, water absorption, ambient temperature or humidity, and is used for statistical analysis. Represents the arithmetic mean, which is used to reflect the center position of the data set. represents the sample standard deviation, represents the coefficient of skewness.

[0084] formula:

[0085] ;

[0086] ;

[0087] ;

[0088] The benefit of the formula is that through these three formulas, the central tendency, dispersion and skewness of the data distribution of the multi-dimensional evaluation data set are critical for analyzing the overall characteristics of the data. For example, the mean μ provides the average level of the data points, the standard deviation σ measures the degree of dispersion of the data, and the skewness coefficient γ reveals the direction and degree of skewness of the data distribution.

[0089] Detailed explanation of the formula and the process of formula calculation and derivation:

[0090] Suppose there is an actual data set {5,6,8,9}, calculate the average:

[0091] ;

[0092] The standard deviation σ is then calculated as:

[0093] ;

[0094] The skewness coefficient γ is calculated as:

[0095] ;

[0096] The results show that the data center is located at 7, the data points are distributed closely around the center (with a small degree of dispersion), and the data as a whole is slightly positively skewed, indicating that most of the data are concentrated below the average value, but a few higher values ​​raise the average level.

[0097] See also Figure 3 , the steps to obtain the initial configuration of the model are as follows:

[0098] S211: Based on the statistical characteristics of the data, the radial basis function and Matérn function of the Gaussian process are selected as the core calculation model, the initial parameters including the length scale and variance are set, and the initial configuration of the kernel function is generated;

[0099] Kernel function selection is based on the statistical characteristics of the data. Kernel functions include radial basis function and Matérn function. Radial basis function is used to describe smoother change characteristics, while Matérn function is used to capture rougher change characteristics. The length scale and variance parameters in the initial configuration are set and adjusted. The length scale parameter is used to control the fitting influence range of the input data, and the variance parameter is determined according to physical properties or experimental conditions, so as to establish the initial configuration of the kernel function.

[0100] S212: Apply the kernel function initial configuration to perform five-fold cross validation, perform performance evaluation on each parameter configuration, calculate the average validation error, determine the best performing parameter, and obtain the validation result;

[0101] The kernel function initial configuration is applied, and the performance of the configuration is evaluated multiple times through five-fold cross validation. First, the data set is evenly divided into five groups, four of which are used as training sets and one as a test set. Each group is used as a test set in turn, and the remaining four groups are used as training sets. The average validation error of the kernel function configuration in each group is evaluated. The calculation results include the mean and standard deviation of each group error and the overall error. The validation error is calculated for the differentiated parameter combination. The configuration parameters with the best performance are judged and determined according to the error values ​​of the differentiated parameter settings, and the optimal parameters after verification are obtained.

[0102] S213: Based on the verification results, the kernel function configuration is updated using the optimal length scale and variance parameter settings to construct a Gaussian process model using the formula:

[0103] ;

[0104] Calculate the model kernel value and generate the preliminary configuration of the model;

[0105] in, and Represents two data points, used to calculate the kernel function value between each other, represents the variance parameter, which affects the degree of variation in the model output. Represents the length scale of the kernel function, determines the sensitivity of the distance between data points, and adjusts the matching of the model to the input space scale.

[0106] formula:

[0107] ;

[0108] The formula is useful because it passes parameters and The flexible setting of adjusts the smoothness and variation range of the kernel function, thus providing a more suitable kernel value calculation method when processing data with differentiated complexity.

[0109] Detailed explanation of the formula and the process of formula calculation: , , set the length scale ,variance , substituting into the formula we get:

[0110] ;

[0111] The results show that the nuclear value Representative data points and The higher the value, the stronger the fit.

[0112] See also Figure 4 , the steps to obtain the model update status are as follows:

[0113] S311: Using the initial configuration of the model, real-time data collection of the construction site, including material hardness, water absorption, ambient temperature and humidity, standardize the data to adapt to the model input requirements, and generate real-time standardized data;

[0114] Real-time data collection from the construction site, including detailed measurements of material hardness, water absorption, ambient temperature and humidity, is performed to standardize the data to adapt to the model input requirements. Data is acquired in real time through sensors and data acquisition systems, and the data is cleaned and standardized through preprocessing procedures, including removing outliers and filling missing values, to ensure data quality and adapt to the needs of the Gaussian process model. This process involves the use of data validation techniques and statistical analysis methods to evaluate the accuracy and completeness of the data. Real-time data collection and processing provides instant and accurate input for the model, ensuring the real-time and accuracy of the model output. The generation of real-time standardized data records is carried out by adjusting the data acquisition equipment and setting matching acquisition frequencies and acquisition parameters.

[0115] S312: input the real-time standardized data into the configured Gaussian process model, use the incremental learning method to update the model parameters, reflect the impact of the new data on the model in real time, and obtain the update result of the model parameters;

[0116] Real-time standardized data is input into the configured Gaussian process model, and the model is updated online using the incremental learning method, including dynamically adjusting the model parameters using the current data points. In this process, the input of real-time data is completed through a dedicated data interface, which is responsible for formatting the collected data and transmitting it to the model processing unit. The incremental learning method gradually adjusts and optimizes the model parameters according to the newly input data to ensure that the model can reflect the current data characteristics and changing trends. This process involves not only automatic adjustment of parameters, but also continuous training and verification of the model, and evaluation of the impact of new parameters on model performance. The generated model parameter update results are obtained by calculating the difference between the new and old parameters and applying the optimization algorithm to determine the optimal adjustment amount.

[0117] S313: Based on the update results of the model parameters, the performance test of the model is performed using the formula:

[0118] ;

[0119] Adjust and optimize model parameters, synchronize models with real-time data, and generate model update status;

[0120] in, Indicates the parameters before updating. represents the target parameter value derived based on the new data, is to adjust the rate to ensure a smooth transition, is the new parameter value after adjustment.

[0121] formula:

[0122] ;

[0123] The benefit of the formula is that it provides a method to dynamically adjust model parameters so that the model can match current data changes, thereby improving the accuracy and reliability of the model.

[0124] Detailed explanation of the formula and the process of formula calculation and derivation:

[0125] Setting initial parameters is 0.5, the target parameter The value obtained based on the current data analysis is 0.8, and the learning rate Set to 0.1. Then the parameter update calculation process is:

[0126] ;

[0127] The results show that after introducing new data, the model parameters are updated from 0.5 to 0.53. The incremental learning method effectively incorporates the current data features into the model, making the model more accurately reflect the current data status. This subtle adjustment of the parameters helps the model maintain the optimal state, thereby ensuring the accuracy and practicality of the output results.

[0128] See also Figure 5 , the specific steps for obtaining the construction prediction results are as follows:

[0129] S411: using the parameters of the model update state, perform quality prediction on each data point in the construction data set, calculate the predicted value in real time, and generate a list of predicted values ​​for each point based on the output of the model;

[0130] Using the updated parameters of the model, the predicted value of construction quality is calculated through a high-precision algorithm. The quality of each data point is evaluated according to its environmental characteristics and previous performance, and a list of predicted values ​​for all data points is generated. This list not only includes the predicted value of each point, but also adjusts the sensitivity of the prediction model according to the construction conditions of the differentiated engineering stages. In this way, the accuracy and practicality of the prediction are ensured.

[0131] S412: Apply statistical methods to each prediction point in the prediction value list, calculate a 95% confidence interval, and estimate the upper and lower bounds of each prediction value to generate a confidence interval list;

[0132] Based on the list of predicted values, a statistical method is used to calculate the 95% confidence interval for each predicted value. This method refers to the distribution type and sample size of the data to ensure that the calculated confidence interval reflects both the uncertainty of the prediction and the range of prediction error. Through this method, the prediction result of each data point is given a certain credibility evaluation and provided to project managers as a basis for construction quality control.

[0133] S413: Integrate the predicted value list with the corresponding confidence interval list using the formula:

[0134]

[0135] Express the 95% confidence interval for each prediction point and generate construction prediction results;

[0136] in, represents the predicted value of each data point, is the standard error, is the sample size, is the z-value of the standard normal distribution corresponding to the confidence level.

[0137] formula:

[0138] ;

[0139] The benefit of the formula is that by introducing confidence levels and standard errors, the uncertainty of the predicted values ​​is quantified, providing a scientific decision-making tool for project management and enhancing the ability to predict potential problems during the construction process.

[0140] Detailed explanation of the formula and the process of formula calculation and derivation:

[0141] Set a predicted value , standard error , sample size , and adopt (corresponding to a 95% confidence level)

[0142] ;

[0143] This means that the 95% confidence interval is [46.42, 53.58]. The results show that, given the sample size and standard error of the prediction, 95% of the predicted values ​​are between 46.42 and 53.58, which provides a quality control reference for construction management.

[0144] See also Figure 6 , the specific steps for obtaining abnormal monitoring results are:

[0145] S511: according to the construction prediction result, set an abnormality detection threshold, determine the benchmark for monitoring construction quality, and generate a threshold setting result;

[0146] According to the construction prediction results, in order to set the anomaly detection threshold First, it is necessary to conduct statistical analysis on the predicted data. Statistical analysis includes calculating the mean, standard deviation and distribution of data. Through these statistical indicators, we can understand the overall trend and fluctuation of construction quality. Based on the statistical results, the upper or lower bound of the 95% confidence interval is selected as the anomaly detection threshold. This setting ensures that the threshold can cover most of the normal data and effectively identify abnormal data. Then, the threshold is applied to the real-time monitoring system. The system will compare the construction data at each moment in real time to see if it exceeds the threshold. If it exceeds the threshold, it will be marked as abnormal and generate the threshold setting result.

[0147] S512: Using the set threshold, monitor each threshold setting result in real time and calculate whether each data point exceeds the threshold using the formula:

[0148] ;

[0149] Determine the severity of the data points exceeding the threshold and generate a list of data points exceeding the threshold;

[0150] in, represents the predicted value, is the set threshold, Indicates the magnitude of the superthreshold, calculated by and The difference between The ratio provides a standardized assessment of suprathreshold data points;

[0151] formula:

[0152] ;

[0153] The formula is useful in that it compares real-time data With threshold The relationship between the data points and the threshold value is calculated. This method can not only quantify the degree to which the data points exceed the threshold value, but also The function ensures the non-negativity of the result, making the monitoring result more intuitive and easy to analyze.

[0154] Detailed explanation of the formula and the process of formula calculation and derivation:

[0155] set up (predicted value), (Threshold), substitute into the formula to calculate:

[0156] ;

[0157] The results show that the predicted value exceeds the threshold by 20%, that is, Value ratio 20% higher, a result that can be used to iterate analysis or trigger an alarm system.

[0158] S513: Record and analyze the list of data points exceeding the threshold value, issue a warning signal for each data point exceeding the threshold value, integrate the monitoring results and the warning signal, mark and record events that require attention, and generate abnormal monitoring results.

[0159] After obtaining the super-threshold data points, the monitoring system records and analyzes the data. The recording process includes the data's timestamp, value, and its relative position to the threshold. The analysis involves the frequency, duration, and occurrence pattern of the data points. This information is crucial for predicting potential risks and problems. By analyzing the abnormal data points, the cause of the abnormality is determined, whether it is a fluctuation in construction quality, equipment failure, or operational error, and corresponding response measures are taken, including adjusting construction parameters or strengthening operating specifications to reduce risks in future time periods, ensure stable construction quality, generate abnormal monitoring results, and improve the controllability and safety of the construction process.

[0160] A construction quality prediction system based on data analysis, the construction quality prediction system based on data analysis is used to execute the above-mentioned construction quality prediction method based on data analysis, the system comprises:

[0161] The data collection and feature construction module collects material hardness, water absorption, ambient temperature, and humidity data from the construction site, cleans the data, and calculates the mean, standard deviation, and skewness coefficient to generate a construction feature summary;

[0162] The model configuration and verification module sets the kernel function parameters of the Gaussian process model based on the construction feature summary, including the length scale and variance parameters of the radial basis function and Matérn function, and uses five-fold cross-validation to adjust the parameters to generate the optimal model configuration;

[0163] The model online update module uses the optimal model configuration to process the newly collected construction data in real time, applies the incremental learning method to update the model parameters online, keeps in sync with the construction site data, and generates the real-time status of the model;

[0164] The quality prediction and monitoring module predicts the construction quality through the real-time status of the model, calculates the predicted value and 95% confidence interval of each prediction point, records the uncertainty range of the prediction results, sets the anomaly detection threshold according to the prediction results, identifies and records the data points that exceed the threshold, issues a warning signal, and generates quality monitoring results.

[0165] The above are only preferred embodiments of the present invention and are not intended to limit the present invention in other forms. Any technician familiar with the profession may use the technical contents disclosed above to change or modify them into equivalent embodiments with equivalent changes and apply them to other fields. However, any simple modification, equivalent change and modification made to the above embodiments based on the technical essence of the present invention without departing from the technical solution of the present invention still falls within the protection scope of the technical solution of the present invention.

Claims

1. A construction quality prediction method based on data analysis, characterized in that: The following steps are involved: Collect data on material hardness, water absorption, ambient temperature, and humidity from the construction site, clean the data and format it to a unified standard, calculate the mean, standard deviation, and skewness coefficient, and generate data statistical features; Based on the statistical characteristics of the data, radial basis functions and Matérn functions of Gaussian processes are selected as kernel functions, the length scale and variance parameters of the kernel functions are modified, and five-fold cross validation is applied to determine the optimal parameter settings for constructing a Gaussian process model and generating a preliminary configuration of the model; Using the preliminary configuration of the model, collecting current parameter input in real time for online updating, adjusting the current parameters through incremental learning methods and testing the impact of the current parameters on the model, maintaining the synchronous update of model parameters and construction site data, and generating a model update state; By updating the state of the model, quality prediction is performed, the predicted value and 95% confidence interval of each prediction point are calculated, the uncertainty range of the prediction result is expressed by the confidence interval, and the construction prediction result is generated; According to the construction prediction results, an abnormal detection threshold is set, the construction quality is monitored, data points exceeding the threshold are identified, early warning signals are recorded and issued, and abnormal monitoring results are generated.

2. The method for predicting building construction quality based on data analysis according to claim 1, characterized in that: The data statistical characteristics specifically include the mean, standard deviation, and skewness coefficient; the model preliminary configuration specifically refers to the length scale and variance parameter of the kernel function; the model update status specifically includes the current parameter input, incremental learning method, and synchronous update status; the construction prediction results specifically include the predicted value and 95% confidence interval of each prediction point; the abnormal monitoring results specifically include the abnormal detection threshold, data points exceeding the threshold, and early warning signals.

3. The method for predicting construction quality based on data analysis according to claim 2 is characterized in that: The steps for obtaining the data statistical features are specifically as follows: Collect material hardness data, water absorption data, ambient temperature and humidity data at the construction site, unify the format and create initial data records to obtain a standardized data set; Performing data verification on the standardized data set, removing outliers that exceed a preset threshold, and filling in missing values ​​to obtain a verified data set; Extract key data points from the validated data set and calculate the mean, standard deviation and skewness coefficient of each data point using the formula: ； ； ； Get the statistical characteristics of the data; in, Represents the total number of data points, indicating the number of elements in the data set, Represents a single data point value, which is any of hardness, water absorption, ambient temperature or humidity, and is used for statistical analysis. Represents the arithmetic mean, which is used to reflect the center position of the data set. represents the sample standard deviation, represents the coefficient of skewness.

4. The method for predicting construction quality based on data analysis according to claim 3 is characterized in that: The steps for obtaining the preliminary configuration of the model are specifically as follows: Based on the statistical characteristics of the data, radial basis functions and Matérn functions of Gaussian processes are selected as core calculation models, initial parameters including length scale and variance are set, and initial configuration of kernel functions is generated; Apply the kernel function initial configuration to perform five-fold cross validation, perform performance evaluation on each parameter configuration, calculate the average validation error, determine the best performing parameter, and obtain validation results; Based on the verification results, the kernel function configuration is updated using the optimal length scale and variance parameter settings to construct a Gaussian process model using the formula: ； Calculate the model kernel value and generate the preliminary configuration of the model; in, and Represents two data points, used to calculate the kernel function value between each other, represents the variance parameter, which affects the degree of variation in the model output. Represents the length scale of the kernel function, determines the sensitivity of the distance between data points, and adjusts the matching of the model to the input space scale.

5. The method for predicting construction quality based on data analysis according to claim 4 is characterized in that: The steps for obtaining the model update status are specifically as follows: Using the preliminary configuration of the model, real-time data from the construction site is collected, including material hardness, water absorption, ambient temperature and humidity, and the data is standardized to fit the model input requirements to generate real-time standardized data; Input the real-time standardized data into the configured Gaussian process model, use the incremental learning method to update the model parameters, reflect the impact of the new data on the model in real time, and obtain the update result of the model parameters; Based on the update results of the model parameters, the performance test of the model is performed using the formula: ； Adjust and optimize model parameters, synchronize models with real-time data, and generate model update status; in, Indicates the parameters before updating. represents the target parameter value derived based on the new data, is to adjust the rate to ensure a smooth transition, is the new parameter value after adjustment.

6. The method for predicting construction quality based on data analysis according to claim 5 is characterized in that: The steps for obtaining the construction prediction results are specifically as follows: Using the parameters of the model to update the state, perform quality prediction for each data point in the construction data set, calculate the predicted value in real time, and generate a list of predicted values ​​for each point based on the output of the model; Applying statistical methods to each prediction point in the prediction value list, calculating a 95% confidence interval, and estimating the upper and lower bounds of each prediction value to generate a confidence interval list; Combine the predicted values ​​list with the corresponding confidence interval list using the formula: ； Express the 95% confidence interval for each prediction point and generate construction prediction results; in, represents the predicted value of each data point, is the standard error, is the sample size, is the z-value of the standard normal distribution corresponding to the confidence level.

7. The method for predicting construction quality based on data analysis according to claim 6 is characterized in that: The steps for obtaining the abnormal monitoring results are specifically as follows: According to the construction prediction results, an abnormality detection threshold is set, a benchmark for monitoring construction quality is determined, and a threshold setting result is generated; Using the set threshold, each of the threshold setting results is monitored in real time to calculate whether each data point exceeds the threshold, using the formula: ； Determine the severity of the data points exceeding the threshold and generate a list of data points exceeding the threshold; in, represents the predicted value, is the set threshold, Indicates the magnitude of the superthreshold, calculated by and The difference between The ratio provides a standardized assessment of suprathreshold data points; The list of over-threshold data points is recorded and analyzed, a warning signal is issued for each over-threshold data point, the monitoring results and the warning signal are integrated, events that require attention are marked and recorded, and abnormal monitoring results are generated.

8. The construction quality prediction system based on data analysis is characterized by: According to the method for predicting construction quality based on data analysis according to any one of claims 1 to 7, the system comprises: The data collection and feature construction module collects material hardness, water absorption, ambient temperature, and humidity data from the construction site, cleans the data, and calculates the mean, standard deviation, and skewness coefficient to generate a construction feature summary; The model configuration and verification module sets the kernel function parameters of the Gaussian process model based on the construction feature summary, including the length scale and variance parameters of the radial basis function and the Matérn function, and adjusts the parameters using five-fold cross validation to generate the optimal model configuration; The model online update module uses the optimal model configuration to process the newly collected construction data in real time, applies the incremental learning method to update the model parameters online, keeps synchronization with the construction site data, and generates the real-time status of the model; The quality prediction and monitoring module predicts the construction quality through the real-time status of the model, calculates the predicted value and 95% confidence interval of each prediction point, records the uncertainty range of the prediction results, sets the anomaly detection threshold according to the prediction results, identifies and records the data points that exceed the threshold, issues a warning signal, and generates quality monitoring results.

Images