Premixed concrete quality early warning method and storage medium

Abstract

The invention relates to the technical field of quality early warning of premixed concrete. The method comprises the steps of analyzing the data distribution condition of a fusion data set in the same enterprise and project based on a K-Means clustering result, identifying an abnormal matching relationship, calculating a comparison abnormal value based on the abnormal matching relationship, and determining the quality of the premixed concrete based on the comparison abnormal value. The method comprises the steps of calculating abnormal scores of enterprises and projects, quantifying the abnormal degree of each enterprise and project in construction quality control to obtain a multi-dimensional quality score, performing comprehensive weighted calculation on a single-dimensional quality score, the multi-dimensional quality score and a supervision and detection score by adopting an analytic hierarchy process to obtain a comprehensive quality score, and calculating the quality of the enterprises and projects according to the comprehensive quality score. And comparing the comprehensive quality score with a preset quality score threshold, and if the comprehensive quality score is lower than the preset quality score threshold, generating early warning data. The method has the effect of ensuring stable concrete construction quality.

Description

Technical Field

[0001] This invention relates to the technical field of early warning of ready-mixed concrete quality, and in particular to a method and storage medium for early warning of ready-mixed concrete quality. Background Technology

[0002] Ready-mixed concrete is a key material in construction projects, and its quality directly affects the safety and durability of building structures. However, during the production, transportation, and construction of ready-mixed concrete, its quality fluctuates and may even exhibit abnormalities due to various factors such as production processes, environmental conditions, and construction quality. Traditional concrete quality testing methods typically rely on manual sampling or analysis of a single data source, making it difficult to achieve a comprehensive and real-time assessment of concrete quality. This leads to delays in the detection of quality problems, impacting project progress and safety. Summary of the Invention

[0003] To ensure stable concrete construction quality, this application provides a method for early warning of ready-mixed concrete quality and a storage medium.

[0004] The above-mentioned objective of this application is achieved through the following technical solution: A method for early warning of ready-mixed concrete quality, comprising: Acquire enterprise self-inspection data, third-party testing data, and rebound data; Using Z-Score analysis, anomaly scores were assigned to the enterprise's self-inspection data, third-party testing data, and rebound data, resulting in enterprise self-inspection anomaly scores, third-party testing anomaly scores, and rebound anomaly scores. Based on the principle of equal weight, the enterprise self-inspection anomaly scores, third-party inspection anomaly scores, and rebound anomaly scores are merged to form an anomaly score dataset. Based on the anomaly score dataset, a single-dimensional quality score is calculated. Using K-Means clustering analysis, with rebound data as the core, the self-inspection data, third-party testing data, and rebound data of enterprises are fused to form a fused dataset. K-Means clustering analysis is then performed on the fused dataset to obtain the K-Means clustering results. Based on the K-Means clustering results, the data distribution of the fused dataset in the same enterprise and project is analyzed, abnormal matching relationships are identified, and comparative outliers are calculated based on the abnormal matching relationships. Based on the comparison of outliers, the anomaly scores of enterprises and projects are calculated to quantify the degree of anomalies of each enterprise and project in construction quality control, and to obtain multi-dimensional quality scores. Obtain monitoring data from the main station; Based on the monitoring data from the main station, perform monitoring anomaly identification and processing, count the number of monitoring anomalies, and calculate the monitoring detection score based on the number of monitoring anomalies. The analytic hierarchy process is used to comprehensively and weightedly calculate the single-dimensional quality score, the multi-dimensional quality score, and the supervision and inspection score to obtain the comprehensive quality score. The overall quality score is compared with a preset quality score threshold. If the overall quality score is lower than the preset quality score threshold, an early warning is generated.

[0005] By adopting the above technical solution, it is possible to acquire enterprise self-inspection data, third-party testing data, and rebound data, and introduce central station supervision data as an authoritative data source. Based on Z-Score analysis, anomaly scores are applied to data from different sources to form comparable anomaly scales. This ensures that enterprise self-inspection anomaly scores, third-party testing anomaly scores, and rebound anomaly scores have consistent scoring criteria. Furthermore, based on the principle of equal weighting, enterprise self-inspection anomaly scores, third-party testing anomaly scores, and rebound anomaly scores can be merged in a single dimension to form an anomaly score dataset for calculating a single-dimensional quality score. This provides a clear quantitative basis for the degree of anomalies in the production process. Finally, K-Means clustering analysis can be used to fuse multi-source data with rebound data as the core to obtain clustering results. Based on clustering results, the data distribution of the same enterprise and project is analyzed, and abnormal matching relationships are identified to calculate comparative outliers. This enables multi-dimensional quality scores to reflect the degree of abnormality in construction quality control of enterprises and projects. It can also perform supervision anomaly identification based on the central station's supervision data and count the number of supervision anomalies to calculate supervision and inspection scores. This allows supervision and inspection scores to reflect abnormal situations under regulatory spot checks. Furthermore, it can use the analytic hierarchy process to comprehensively weight and calculate a comprehensive quality score by combining single-dimensional quality scores, multi-dimensional quality scores, and supervision and inspection scores. The comprehensive quality score is then compared with a preset quality score threshold to generate early warning data. This ensures that the early warning data covers the three dimensions of production, construction, and supervision and meets the business needs of subsequent storage, retrieval, and traceability.

[0006] In a preferred example, this application can be further configured as follows: based on the principle of equal weight, the enterprise's self-inspection anomaly score, the third-party inspection anomaly score, and the rebound anomaly score are fused to form an anomaly score dataset; based on the anomaly score dataset, a single-dimensional quality score is calculated, including: Based on the principle of equal weighting, the enterprise self-inspection anomaly score, the third-party inspection anomaly score, and the rebound anomaly score are weighted and calculated to obtain the weighted anomaly score result. The weighted anomaly scoring results are normalized to obtain standardized anomaly scoring results; Based on the standardized anomaly scoring results, the arithmetic mean method was used to calculate the single-dimensional quality score.

[0007] By adopting the above technical solution, the self-inspection anomaly scores, third-party testing anomaly scores, and rebound anomaly scores are weighted according to the principle of equal weight. This ensures that the weights of each data source are balanced in the production quality assessment, avoiding excessive influence of a single data source on the scoring results. This allows the single-dimensional quality score to more fairly reflect the quality level of different testing data, improving the credibility and impartiality of the score. By normalizing the weighted anomaly score results, anomaly scores from different data sources are mapped to a unified numerical range, eliminating the influence caused by differences in data sources or numerical scales. This makes different types of anomaly scores comparable, thus ensuring the high stability and consistency of the calculated single-dimensional quality score. Using the arithmetic mean method to calculate the single-dimensional quality score makes the scoring calculation process simpler and more efficient, avoiding the computational overhead and instability that complex nonlinear models may bring. At the same time, it ensures the comprehensive evaluation capability of the calculation results for different data sources, improving the stability and interpretability of the single-dimensional quality score, enabling it to effectively reflect the quality level of concrete production.

[0008] In a preferred example, this application can be further configured as follows: using K-Means clustering analysis, with rebound data as the core, to perform fusion processing on enterprise self-inspection data, third-party inspection data, and rebound data to form a fused dataset, and to perform K-Means clustering analysis based on the fused dataset to obtain K-Means clustering results, including: Using rebound data as the core, and matching it with enterprise self-inspection data and third-party testing data, a clustered dataset is formed; Based on the rebound data, and according to the distribution characteristics of the rebound data, the initial cluster centers of K-Means are set; Based on the clustering dataset, the K-Means clustering analysis method is used for iterative calculation. The initial cluster centers are adjusted according to the rebound data, enterprise self-inspection data and third-party detection data until the position change of the initial cluster centers is less than the preset threshold, and the converged K-Means cluster centers are obtained. Based on the converged K-Means cluster centers, the Euclidean distance between each sample in the clustered dataset and each cluster center is calculated. Then, according to the minimum distance principle, the cluster category to which each sample in the clustered dataset belongs is determined, and the K-Means clustering result is obtained.

[0009] By adopting the above technical solution, it is possible to perform matching processing on enterprise self-inspection data and third-party testing data with rebound data as the core and form a clustered dataset. This provides a unified data input standard for subsequent cluster analysis. It is possible to set the initial cluster centers of K-Means based on the distribution characteristics of rebound data, ensuring that the initial cluster centers correspond to the characteristics of rebound data. It is possible to perform iterative calculations based on the clustered dataset and continuously adjust the initial cluster centers until the convergence condition is met, so that the K-Means cluster centers can reflect the sample distribution structure of the clustered dataset. It is possible to calculate the Euclidean distance between the sample and the cluster center based on the converged K-Means cluster centers and determine the cluster category to which the sample belongs according to the minimum distance principle, thereby obtaining K-Means clustering results to support subsequent data distribution analysis and anomaly matching relationship identification.

[0010] In a preferred example, this application can be further configured to: analyze the data distribution of the fused dataset within the same enterprise and project based on K-Means clustering results, identify anomalous matching relationships, and calculate comparative outliers based on these anomalous matching relationships, including: Based on the K-Means clustering results, the statistical information of the fused dataset within its respective cluster category is calculated, and a data matching feature matrix is ​​constructed based on the statistical information; Based on the data matching feature matrix, calculate the deviation value of the fused dataset in the same enterprise and project; Based on the deviation value, calculate the contrast outliers of the fused dataset.

[0011] By adopting the above technical solution, statistical calculations can be performed on the fused dataset within its respective cluster category based on the K-Means clustering results. This allows the statistical information to reflect the distribution characteristics of the fused dataset within each cluster category. A data matching feature matrix can be constructed based on the statistical information, giving the matching relationship of the fused dataset in different statistical dimensions a structured expression. Deviation values ​​can be calculated based on the data matching feature matrix within the same enterprise and project, enabling the deviation values ​​to characterize the degree of difference between the fused dataset and the statistical characteristics of the cluster category within the same enterprise and project. Comparison outliers can be calculated based on the deviation values, allowing the comparison outliers to serve as input for subsequent identification of abnormal matching relationships and calculation of construction quality scores. This also facilitates consistent comparative analysis and record-keeping of anomalies in different enterprises and projects.

[0012] In a preferred example, this application can be further configured to: perform supervision anomaly identification processing based on the central station's supervision data and count the number of supervision anomalies, and calculate a supervision detection score based on the number of supervision anomalies, including: The monitoring data from the main station is analyzed to extract monitoring data fields, which include monitoring and detection intensity values, design strength requirement values, and rebound monitoring values. Using enterprises and projects as statistical objects, the monitoring data fields are aggregated and processed to obtain a set of monitoring data items corresponding to enterprises and projects; For each supervision data entry in the supervision data entry set, calculate the strength deviation value between the supervision detection strength value and the design strength requirement value, and determine the rebound supervision deviation value based on the rebound supervision value; A first comparison and judgment process is performed on the strength deviation value, and a second comparison and judgment process is performed on the rebound monitoring deviation value to obtain the first judgment result information and the second judgment result information. Based on the first judgment result information and the second judgment result information, the supervision anomaly judgment processing is performed, and when the preset supervision anomaly judgment rules are met, a supervision anomaly event is judged to have occurred. The number of abnormal monitoring events is accumulated and statistically analyzed to obtain the number of abnormal monitoring events for each enterprise or project. Based on the number of abnormal monitoring events and the preset scoring mapping rules, a linear mapping process is performed on the number of abnormal monitoring events to obtain the monitoring and detection score for each enterprise or project.

[0013] By adopting the above technical solutions, the monitoring data from the main station can be analyzed and the monitoring intensity value, design intensity requirement value, and rebound monitoring value can be extracted. This provides clear data field inputs for subsequent monitoring anomaly identification and processing. The monitoring data fields can be aggregated and processed by enterprise and project as statistical objects to obtain a set of monitoring data items. This allows the judgment and statistics of monitoring anomalies to establish a correspondence at the enterprise and project levels. The intensity deviation value can be calculated for the set of monitoring data items, and the rebound monitoring deviation value can be determined. This provides a calculable quantitative basis for monitoring anomaly judgment and processing. The intensity deviation value and the rebound monitoring deviation value can be compared and judged separately to obtain the first judgment result information and the second judgment result information. Based on the first judgment result information and the second judgment result information, monitoring anomaly judgment and processing can be performed to form monitoring anomaly events. The monitoring anomaly events can be cumulatively statistically analyzed to obtain the number of monitoring anomalies, and linear mapping processing can be performed based on the scoring mapping rules to obtain the monitoring detection score. This allows the monitoring detection score to reflect the monitoring anomaly situation of the enterprise or project under a unified scoring standard and to be used for subsequent comprehensive scoring calculation.

[0014] In a preferred embodiment, this application can be further configured to: employ the analytic hierarchy process (AHP) to perform a comprehensive weighted calculation of the single-dimensional quality score, the multi-dimensional quality score, and the supervision and inspection score to obtain a comprehensive quality score, including: A hierarchical analysis structure model is established based on single-dimensional quality scoring, multi-dimensional quality scoring, and supervision and inspection scoring. Based on the hierarchical analysis structure model, a judgment matrix is ​​constructed. Based on the judgment matrix, the eigenvector method is used to calculate the weight allocation result. Based on the weighting results, the overall quality score is calculated using the following formula: , where Q total This refers to the overall quality score. W1 refers to the weight of the single-dimensional quality score, Q1 is the single-dimensional quality score, W2 refers to the weight of the multi-dimensional quality score, Q2 is the multi-dimensional quality score, W3 refers to the weight of the supervision and inspection score, and Q3 is the supervision and inspection score.

[0015] By adopting the above technical solutions, a hierarchical analysis structure model can be established based on single-dimensional quality scoring, multi-dimensional quality scoring, and supervision and inspection scoring. This ensures a clear structured correspondence between the calculation objects, evaluation levels, and indicator sources of the comprehensive quality score. A judgment matrix can be constructed based on the hierarchical analysis structure model, reflecting the relative importance of production quality, construction quality, and supervision and inspection within the judgment matrix. This provides a verifiable data carrier for weight calculation, enabling the use of the eigenvector method to calculate weight allocation results based on the judgment matrix. The weights W1 (single-dimensional quality score), W2 (multi-dimensional quality score), and W3 (supervision and inspection score) have consistent solution rules and maintain correspondence with the judgment matrix. Finally, the comprehensive quality score Q can be calculated using a weighted summation formula based on the weight allocation results. total This makes the overall quality score Q total It can simultaneously consider Q1, Q2 and Q3 and maintain a one-to-one binding relationship between weights and scoring items, avoiding situations where the source of weights is unclear or the correspondence of scoring items is ambiguous during the formation of the comprehensive quality score. This allows the comprehensive quality score Q to serve as a unified input for subsequent quality score threshold comparison and early warning data generation, and facilitates horizontal comparison and record traceability between different enterprises and projects.

[0016] In a preferred example, this application can be further configured to include: a comprehensive quality score, followed by: Based on the comprehensive quality score, the quality control status of each enterprise is classified and marked. Enterprises with good quality control are marked in green, enterprises with medium quality control are marked in yellow, and enterprises with poor quality control are marked in red. For newly started construction projects, a quality tracking mechanism is established to obtain quality marking results, and early warning data is generated based on the quality marking results.

[0017] By adopting the above technical solution and classifying enterprises according to their comprehensive quality scores, the results of enterprise quality management become more intuitive and easier to understand. This helps construction supervision departments quickly identify enterprises with excellent quality management (green), those with average quality control (yellow), and those with quality management problems (red), thus improving the targeting of quality supervision. By monitoring the quality marking results of newly started projects, an early warning mechanism can be triggered as soon as a new project changes from green to yellow or red, allowing quality management departments to intervene in advance, improving the dynamic controllability of construction process quality, and preventing serious quality problems from occurring.

[0018] In a preferred embodiment, this application can be further configured to include: early warning data, which also includes: Based on the early warning data, the concrete quality early warning data is displayed through data visualization. The data visualization includes drawing early warning analysis bar charts, concrete strength distribution maps, and abnormal percentage maps of monitored objects. Based on the early warning data, early warning notifications are sent to managers via SMS, email, or APP. The early warning notifications include the type and severity of the anomaly.

[0019] By adopting the above technical solution, it is possible to visualize concrete quality early warning data based on early warning data, and draw early warning analysis bar charts, concrete strength distribution charts, and abnormal percentage charts of monitored objects according to preset chart generation rules. This allows early warning data to be presented in a graphical form and to correspond with the abnormal statistical results. It is also possible to generate early warning notifications based on the early warning data and send them to management personnel via SMS, email, or APP. This allows early warning notifications to be bound to early warning data and to meet the sending requirements of different notification channels. Furthermore, the early warning notifications include the abnormality category and degree, enabling the recipient to identify and record the early warning event based on the abnormality category and degree, and facilitating subsequent querying and tracing by enterprise and project dimensions.

[0020] In a preferred embodiment, this application can be further configured as: a pre-mixed concrete quality early warning method, which also includes: Construct a comprehensive database to store multi-dimensional quality scores, supervision and inspection scores, comprehensive quality scores, and early warning data. During a preset period, the early warning data in the comprehensive database is compared with the actual engineering quality data to obtain the comparison results. Based on the comparison results, the early warning error is calculated. If there is a deviation between the early warning data and the actual engineering quality data, the deviation value is obtained. Based on the deviation value, the parameters of Z-Score analysis, K-Means clustering analysis and analytic hierarchy process are adjusted. By adopting the above technical solutions, a comprehensive database can be constructed to uniformly store multi-dimensional quality scores, supervision and inspection scores, comprehensive quality scores, and early warning data. This allows the scoring results and early warning results to establish a correlation within the same data space and maintain a traceable record link. It enables the reading of early warning data from the comprehensive database at preset intervals and comparison processing with actual engineering quality data to obtain comparison results. This provides a clear basis for comparison between early warning data and actual engineering quality data. It allows the calculation of early warning errors based on the comparison results and the determination of deviation values ​​when discrepancies exist between early warning data and actual engineering quality data. These deviation values ​​can be used to characterize the magnitude and direction of the early warning error. Furthermore, it allows for the adjustment of parameters for Z-Score analysis, K-Means clustering analysis, and analytic hierarchy process based on the deviation values. This enables subsequent anomaly scoring, clustering analysis, and comprehensive weighted calculations to update parameters based on historical comparison results and form a closed-loop correction process. This facilitates iterative correction and record traceability of the early warning judgment logic during continuous operation.

[0021] The above-mentioned objective 2 of this application is achieved through the following technical solution: A computer-readable storage medium storing a computer program, which, when executed by a processor, implements the steps of the above-described ready-mixed concrete quality early warning method.

[0022] In summary, this application includes at least one of the following beneficial technical effects: 1. It can acquire enterprise self-inspection data, third-party testing data, and rebound data, and introduce central station supervision data as an authoritative data source. Based on Z-Score analysis, it performs anomaly scoring on data from different sources to form comparable anomaly scales, thus ensuring consistent scoring standards for enterprise self-inspection anomaly scores, third-party testing anomaly scores, and rebound anomaly scores. It can also fuse the three types of anomaly scores based on the principle of equal weight to form an anomaly score dataset for calculating single-dimensional quality scores, thereby providing clear quantitative evidence of the degree of anomalies in the production process. Furthermore, it can use K-Means clustering analysis, with rebound data as the core, to fuse multi-source data and obtain clustering results, and then analyze the same enterprise and project based on these clustering results. The system distributes data and identifies abnormal matching relationships to calculate and compare outliers, enabling multi-dimensional quality scores to reflect the degree of abnormality in construction quality control for enterprises and projects. It can identify supervision anomalies based on central station supervision data and count the number of supervision anomalies to calculate supervision and inspection scores, thus enabling supervision and inspection scores to reflect abnormal situations under regulatory spot checks. It can use the analytic hierarchy process to comprehensively and weight the single-dimensional quality scores, multi-dimensional quality scores, and supervision and inspection scores to obtain a comprehensive quality score, and compare the comprehensive quality score with a preset quality score threshold to generate early warning data. This allows the early warning data to cover the three dimensions of production, construction, and supervision and meet the business needs of subsequent storage, retrieval, and traceability. Attached Figure Description

[0023] Figure 1 This is a diagram illustrating the implementation of a pre-mixed concrete quality early warning method in one embodiment of the application. Detailed Implementation

[0024] The present application will be further described in detail below with reference to the accompanying drawings.

[0025] In one embodiment, such as Figure 1 As shown, this application discloses a method for early warning of ready-mixed concrete quality, which specifically includes the following steps: S1: Obtain enterprise self-inspection data, third-party testing data, and rebound data.

[0026] In this embodiment, enterprise self-inspection data refers to the self-inspection data of concrete production enterprises on concrete quality during the production process, including indicators such as slump, air content, unit water consumption, and compressive strength of the concrete mixture. Third-party testing data refers to data obtained by independent third-party testing institutions through laboratory testing of concrete samples, including concrete strength grade, flexural strength, and chloride ion content. Rebound data refers to data obtained by using a rebound hammer to test the poured concrete structure at the construction site, including rebound value, carbonation depth correction value, and corresponding strength assessment value.

[0027] Specifically, enterprise self-inspection data comes from test data recorded by the production enterprise during the concrete production process. This typically includes concrete slump, air content, unit water consumption, mixture temperature, sand and gravel moisture content, and compressive strength test results. This data is usually obtained through laboratory testing or online monitoring equipment and stored in the enterprise's quality management system. Third-party testing data comes from independent testing institutions and is usually based on laboratory testing after sampling. The test items include concrete cube compressive strength, flexural strength, chloride ion content, and other specified indicators. The data is stored in the third-party testing database and provided through a digital interface. Rebound data refers to the rebound value of the concrete structure surface detected by a rebound hammer at the construction site. The strength assessment value of the concrete is calculated by combining the carbonation depth correction value. Rebound data is usually recorded on a per-construction-area basis and stored in the on-site quality management system.

[0028] Furthermore, the on-site quality management system is a construction quality control platform based on real-time data acquisition, analysis, and early warning. It aims to dynamically monitor concrete production, construction, and environmental factors, promptly identify quality anomalies, assist project managers in optimizing the construction process, and improve the controllability of construction quality. This system achieves comprehensive quality control on the construction site by integrating data acquisition, processing, analysis, and early warning functions.

[0029] S2: Using Z-Score analysis, anomaly scores are assigned to the enterprise's self-inspection data, third-party testing data, and rebound data to obtain the enterprise's self-inspection anomaly score, third-party testing anomaly score, and rebound anomaly score.

[0030] In this embodiment, Z-Score analysis refers to a method that standardizes data based on standard scores (Z-Score) and identifies outliers.

[0031] Specifically, first, the mean and standard deviation of each data category are calculated, and then the standardization calculation is performed on individual data points using the following formula: Where X is the current data point, μ is the mean of historical data, and σ is the standard deviation, the calculated Z value represents the position of the data relative to the overall data distribution. Z values ​​are calculated for enterprise self-inspection data, third-party inspection data, and rebound data respectively, and it is determined whether they exceed the set abnormal threshold. For example, when |Z| is greater than 3, the data is marked as abnormal and a corresponding abnormal score is generated. Finally, normalization processing is performed so that the abnormal score falls within the set range, resulting in the enterprise self-inspection abnormal score, third-party inspection abnormal score, and rebound abnormal score.

[0032] S3: Based on the principle of equal weight, the enterprise's self-inspection anomaly score, third-party inspection anomaly score, and rebound anomaly score are merged to form an anomaly score dataset. Based on the anomaly score dataset, a single-dimensional quality score is calculated.

[0033] In this embodiment, the principle of equal weight means that when fusing anomaly scores from different data sources, it is assumed that the data from each source is of equal importance and is assigned the same weight.

[0034] Specifically, the principle of equal weighting means that when integrating anomaly scores from different data sources, it is assumed that the enterprise self-inspection anomaly scores, third-party detection anomaly scores, and rebound anomaly scores have the same importance. Therefore, a weighted average method is used for integration. First, the enterprise self-inspection anomaly scores, third-party detection anomaly scores, and rebound anomaly scores are normalized so that the scores from each data source are distributed within the same numerical range. Then, a weighted average is performed on each normalized anomaly score with the same weight to form an anomaly score dataset. Based on the anomaly judgment results corresponding to each checkpoint in the anomaly score dataset, statistics are performed to calculate the ratio between the number of checkpoints judged as valid anomalies and the total number of checkpoints participating in the statistics, thus obtaining the single-dimensional quality score, i.e.: Q1 = (number of valid anomalies / total number of checkpoints) × 1000, where the number of valid anomalies refers to the number of checkpoints in the anomaly score dataset that meet the preset anomaly judgment rules and are confirmed as anomalies, the total number of checkpoints refers to the total number of checkpoints in the anomaly score dataset participating in the anomaly statistics, and Q1 refers to the single-dimensional quality score.

[0035] S4: Using K-Means clustering analysis, with rebound data as the core, the enterprise's self-inspection data, third-party testing data, and rebound data are fused to form a fused dataset. K-Means clustering analysis is then performed on the fused dataset to obtain the K-Means clustering results.

[0036] Specifically, using rebound data as the core means using each rebound record in the rebound data as a matching benchmark and determining the matching key value. The matching key value includes the enterprise identifier, project identifier, and collection time identifier. Matching enterprise self-inspection data and third-party testing data means retrieving records in the enterprise self-inspection data and third-party testing data that match the matching key value and reading the corresponding field values. When there is no completely consistent record in the enterprise self-inspection data or third-party testing data at the collection time identifier, nearest neighbor matching is performed based on a preset time tolerance to determine the record to be fused. The read rebound data, enterprise self-inspection data, and third-party testing data are created into the same entry according to the enterprise identifier and project identifier, and the same entry is written into the fused dataset, so that the fused dataset contains records of the same type. K-Means clustering analysis based on a fused dataset and a set of multi-source fields corresponding to an enterprise and a project involves normalizing the numerical fields in the fused dataset and using the normalized field vectors as the clustering input vectors. Cluster centers are initialized according to a preset number of clusters K, and the Euclidean distance between each entry in the fused dataset and each cluster center is calculated. Each entry is assigned to the cluster category corresponding to the cluster center with the smallest distance. The cluster centers are then updated based on the mean of the entries within each cluster category. This process of Euclidean distance calculation, cluster category assignment, and cluster center update is repeated until the change in cluster centers is less than a preset threshold. Finally, the cluster category identifiers corresponding to each entry are output, and the set of cluster category identifiers is determined as the K-Means clustering result.

[0037] S5: Based on the K-Means clustering results, analyze the data distribution of the fused dataset within the same enterprise and project, identify abnormal matching relationships, and calculate comparative outliers based on these abnormal matching relationships.

[0038] Specifically, based on the K-Means clustering results, a clustering category identifier is assigned to each entry in the fused dataset. The fused dataset is then grouped according to enterprise identifier and project identifier to obtain a set of fused entries corresponding to the same enterprise and project. The distribution ratio of clustering category identifiers is statistically analyzed for the fused entry set, and the numerical distribution characteristics of the enterprise self-inspection data field, third-party inspection data field, and rebound data field in the fused entry set are statistically analyzed to form a data distribution. The numerical distribution characteristics include mean, standard deviation, and quantiles, which are used to characterize the central tendency and dispersion of the same enterprise and project across different fields. Based on the data distribution, the field bias between the enterprise self-inspection data field and the rebound data field is calculated. The system calculates the field deviation between the third-party testing data field and the rebound data field, and also calculates the field deviation between the enterprise's self-inspection data field and the third-party testing data field. Based on the field deviation values, it performs comparison and judgment processing with preset matching judgment rules to identify abnormal matching relationships. The preset matching judgment rules are rules used to determine whether the field deviation values ​​exceed the allowable range. Abnormal matching relationships refer to matching results that deviate from the consistency range between different fields within the same enterprise and project. Based on abnormal matching relationships, the system performs summary calculations on the field deviation values ​​to obtain comparative anomaly values. The summary calculation includes weighted summation or finding the maximum value of the field deviation values ​​that meet the abnormal matching relationships to obtain comparative anomaly values.

[0039] S6: Based on the comparison of outliers, calculate the anomaly score of enterprises and projects, quantify the degree of anomaly of each enterprise and project in construction quality control, and obtain a multi-dimensional quality score.

[0040] Specifically, risk clustering labeling is performed on the sample data of each enterprise and project based on the comparison of outliers. First, the risk judgment result of each sample is calculated according to the comparison of outliers and the preset risk judgment rules. The risk judgment result is input into the cluster analysis process to generate cluster category labels. Then, the cluster categories that meet the high-risk judgment conditions are selected from the cluster category labels as high-risk clusters. The ratio between the number of samples falling into high-risk clusters and the total number of samples participating in the statistics is calculated to obtain the multi-dimensional quality score, i.e.: Q2 = (number of samples falling into high-risk clusters / total number of samples) × 1000, where the number of samples falling into high-risk clusters refers to the number of samples whose cluster category in the cluster category label is marked as a high-risk cluster, the total number of samples refers to the total number of samples participating in cluster analysis and risk statistics, and Q2 refers to the multi-dimensional quality score.

[0041] S7: Obtain monitoring data from the main station.

[0042] In this embodiment, the central station's monitoring data refers to the set of monitoring and testing records generated and published by the central station. The monitoring and testing records include at least the enterprise identifier, project identifier, testing time, monitoring and testing intensity value, design intensity requirement value, and rebound monitoring value.

[0043] Specifically, the time range of the supervision data is determined according to the preset statistical period. A candidate supervision and inspection record sequence is obtained by reading the supervision and inspection records in the main station whose detection time falls within the time range of the supervision data. The supervision and inspection records corresponding to the target enterprise and target project are filtered by performing consistency matching processing on the enterprise identifier and project identifier in the candidate supervision and inspection record sequence. The filtered supervision and inspection records are organized in the order of detection time. Supervision and inspection records with missing fields are eliminated by checking the field integrity of supervision and inspection intensity value, design intensity requirement value and rebound supervision value. The unit of measurement is unified by checking the unit identifier of supervision and inspection intensity value and design intensity requirement value and performing unit conversion processing. The organized supervision and inspection records are summarized to form supervision data.

[0044] S8: Based on the monitoring data from the main station, perform monitoring anomaly identification and processing, count the number of monitoring anomalies, and calculate the monitoring detection score based on the number of monitoring anomalies.

[0045] Specifically, the monitoring and inspection record sequence is obtained by filtering the monitoring data from the main station according to enterprise and project identifiers. For each monitoring and inspection record in the sequence, the monitoring and inspection intensity value and the design strength requirement value are read, and the strength deviation value is calculated. The strength deviation value is calculated using difference operation, and the expression is as follows: ,in Indicates the intensity value of supervision and inspection and This indicates the design strength requirement value, in This is recorded as a monitoring anomaly event. This is recorded as a monitoring anomaly event, in which This represents the first strength deviation threshold and is used to limit the allowable deviation of the monitored strength value from the design strength requirement. For the same enterprise and project, the rebound monitoring values ​​are read from the monitoring and inspection record sequence to construct a rebound monitoring value sequence. The rebound center value is calculated using an arithmetic mean, and the expression is: ,in This represents the bounce supervision value corresponding to the k-th supervised detection record, where n represents the number of bounce supervision value sequences. The bounce supervision deviation value is calculated using the absolute difference operation, and the expression is: Where R represents the bounce supervision value corresponding to the current supervised detection record, in This is recorded as a monitoring anomaly event, in which This represents the rebound deviation threshold and is used to limit the allowable upper limit of the rebound monitoring value relative to the rebound center value. The cumulative count of monitoring anomalies corresponding to the monitoring detection record sequence is recorded as N. The number of monitoring anomalies represents the cumulative number of monitoring anomalies corresponding to the same enterprise and project within a preset statistical period, based on the number of monitoring anomalies N and the preset upper limit. Determine the mapping count and use Limits are imposed, with a preset maximum number of attempts. This represents the threshold number used to limit the range of a linear mapping, based on the mapping number value. With preset maximum number of attempts Calculate the normalized coefficient v and the expression is as follows: The normalization coefficient v represents the normalization ratio of the mapping frequency values ​​within a preset upper limit range. Based on the normalization coefficient v and the preset lower limit of the scoring interval... upper limit of the preset rating range Perform a linear mapping operation to obtain the supervised detection score and use it as Q3. The expression for calculating the supervised detection score is as follows: In the preset And preset When 45 of them came from ,exist When Pick .

[0046] S9: The analytic hierarchy process (AHP) is used to calculate the comprehensive quality score by weighting the single-dimensional quality score, the multi-dimensional quality score, and the supervision and inspection score.

[0047] In this embodiment, the analytic hierarchy process (AHP) refers to a method for calculating the weights of multiple indicators.

[0048] Specifically, a hierarchical analysis model is constructed around the comprehensive quality score, with the comprehensive quality score as the decision layer, production quality, construction quality, and supervision and inspection as the criterion layer, and single-dimensional quality scores, multi-dimensional quality scores, and supervision and inspection scores as the indicator layer. Based on preset pairwise comparison rules, the importance of production quality, construction quality, and supervision and inspection within the criterion layer is compared, and the comparison results are written into a judgment matrix. Each element of the judgment matrix is ​​represented by a proportional scale and represents the importance ratio of the corresponding two indicators. Based on the judgment matrix, a product operation is performed on each row of elements, and the square root operation is performed on the product result to obtain an initial weight vector. The normalization processing of the initial weight vector uses the sum of all initial weights as the normalization factor, and each initial weight is divided by the normalization factor to obtain the weight allocation result. The weight allocation result includes the weights corresponding to the single-dimensional quality score, the multi-dimensional quality score, and the supervision and inspection score. Based on the weight allocation result, a weighted summation operation is performed on the single-dimensional quality score, the multi-dimensional quality score, and the supervision and inspection score to obtain the comprehensive quality score. The expression for the weighted summation operation is as follows. , where Q total This refers to the overall quality score, where W1, W2, and W3 are the weights corresponding to Q1, Q2, and Q3, respectively, with W1 = 0.1, W2 = 0.2, and W3 = 0.7.

[0049] S10: Compare the overall quality score with the preset quality score threshold. If the overall quality score is lower than the preset quality score threshold, generate warning data.

[0050] In this embodiment, the quality scoring threshold refers to the critical value used to determine whether the concrete quality is abnormal.

[0051] Specifically, the quality scoring threshold is set based on historical engineering data or industry standards, typically referencing the quality scoring range of previously qualified projects and adjusting it in conjunction with safety requirements to ensure that the scoring threshold can accurately identify anomalies. First, the current overall quality score is compared with the quality scoring threshold. If the overall quality score is lower than the set threshold, it is determined that there may be an anomaly in the concrete quality, thus triggering an early warning mechanism to generate early warning data. This early warning data includes the anomaly category, the degree of anomaly, and suggested handling solutions, and is stored in a database, ultimately yielding the early warning data.

[0052] In one embodiment, in step S3, based on the principle of equal weight, the enterprise's self-inspection anomaly score, the third-party inspection anomaly score, and the rebound anomaly score are fused to form an anomaly score dataset. Based on the anomaly score dataset, a single-dimensional quality score is calculated, including: S301: Based on the principle of equal weight, the enterprise's self-inspection anomaly score, third-party inspection anomaly score, and rebound anomaly score are weighted and calculated to obtain the weighted anomaly score result.

[0053] Specifically, data on enterprise self-inspection anomaly scores, third-party testing anomaly scores, and rebound anomaly scores are organized to ensure that the time dimension of the three types of data is consistent with the enterprises and projects they belong to, making the data comparable under the same reference system. Then, according to the principle of equal weight, the three types of anomaly scores are numerically mapped so that each data type is within the same scoring range, and they are weighted and calculated. The calculated weighted anomaly score results reflect the overall trend of the three types of anomaly scores, while retaining the contribution ratio of each category of data, and finally obtain the weighted anomaly score results.

[0054] S302: Normalize the weighted anomaly score results to obtain standardized anomaly score results.

[0055] Specifically, normalization is used to adjust the numerical range of the weighted outlier scores for subsequent calculations. First, the distribution characteristics of the weighted outlier scores are analyzed, and their maximum and minimum values ​​are determined to set the normalization mapping range. Then, according to the set standard value range, all weighted outlier scores are converted to the same standard range, so that the data scores of different enterprises and different projects can be compared on the same scale, ensuring the consistency and comparability of the numerical distribution, and finally obtaining the standardized outlier score results.

[0056] S303: Based on the standardized anomaly scoring results, the arithmetic mean method is used to calculate the single-dimensional quality score.

[0057] Specifically, the standardized anomaly score results represent the performance of anomaly scores from different data sources under the same scoring system. To calculate the single-dimensional quality score, the standardized anomaly score needs to be further processed. First, the standardized anomaly score results of all enterprises and projects are collected and classified according to project affiliation so that data from the same project can be aggregated and analyzed. Then, the standardized anomaly scores under the same project are averaged to reflect the overall production quality of the project, and finally, the single-dimensional quality score is obtained.

[0058] In one embodiment, in step S4, K-Means clustering analysis is used to fuse the enterprise's self-inspection data, third-party inspection data, and rebound data, with rebound data as the core, to form a fused dataset. K-Means clustering analysis is then performed on the fused dataset to obtain the K-Means clustering results, including: S401: Using rebound data as the core, it matches enterprise self-inspection data and third-party testing data to form a clustered dataset.

[0059] Specifically, each rebound record is read from the rebound data, and the enterprise identifier, project identifier, and collection time identifier are extracted as matching keys. Matching keys are key information used to establish correspondences between different data sources. Based on the matching keys, self-inspection records matching the enterprise identifier and project identifier with a collection time identifier that meets the preset time tolerance are retrieved from the enterprise's self-inspection data, and the corresponding self-inspection field values ​​are read. Similarly, third-party testing records matching the enterprise identifier and project identifier with a collection time identifier that meets the preset time tolerance are retrieved from the third-party testing data, and the corresponding third-party field values ​​are read. When multiple records in the self-inspection data or third-party testing data meet the preset time tolerance, the matching record is determined and the field value is read based on the principle of minimizing the time difference between the collection time identifier and the rebound record. When no record in the enterprise self-inspection data or third-party testing data meets the preset time tolerance, the corresponding field value is marked as a missing value and the rebound record is retained. Based on the same matching key value, the rebound field value, self-inspection field value and third-party field value corresponding to the rebound record are combined to generate a cluster entry and the cluster entry is written into the cluster dataset. The matching key value extraction, retrieval matching and cluster entry generation are repeated until the rebound data traversal is completed and the cluster dataset is obtained.

[0060] S402: Based on the rebound data, set the initial cluster centers of K-Means according to the distribution characteristics of the rebound data.

[0061] Specifically, the rebound data is read and the rebound value fields are extracted to form a rebound value set. The rebound value set refers to the summary result of the rebound values ​​used to participate in the cluster initialization in the rebound data. Distribution statistical processing is performed on the rebound value set to obtain distribution characteristic parameters. Distribution statistical processing includes calculating the minimum, maximum, mean, standard deviation and quantile of the rebound value set. The distribution characteristic parameters are used to characterize the value range and dispersion of the rebound value set. According to the preset number of clusters K, the value range of the rebound value set is divided into K intervals and a representative value is determined in each interval. The determination of the representative value includes taking the median of the interval or taking the sample mean in the interval as the representative value. The representative value corresponding to each interval is used as the initial cluster center value of the rebound dimension. The initial cluster center value is established with the other field dimensions of the corresponding entries in the cluster dataset to form the initial cluster center of K-Means. The consistent initialization method includes searching in the cluster dataset for the cluster entry whose rebound value is closest to the initial cluster center value and using the full field vector of the cluster entry as the corresponding initial cluster center.

[0062] S403: Based on the clustering dataset, the K-Means clustering analysis method is used for iterative calculation. The initial cluster centers are adjusted according to the rebound data, enterprise self-inspection data and third-party detection data until the position change of the initial cluster centers is less than the preset threshold, and the converged K-Means cluster centers are obtained.

[0063] Specifically, for each cluster entry in the clustering dataset, the bounce data field, self-check field, and third-party field are extracted and combined to form a clustering feature vector. The clustering feature vector is a numerical vector used in K-Means distance calculation and center update. Normalization is performed on the clustering feature vector to eliminate differences in the dimensions of different fields. Normalization uses range normalization or standardized normalization and maps each field to a uniform value range according to a preset normalization rule. The initial K-Means cluster centers set in S402 are used as the current cluster centers, and iterative calculation is performed. Iterative calculation includes calculating the Euclidean distance between each clustering feature vector and each current cluster center, and assigning the clustering feature vector to the cluster category corresponding to the current cluster center with the smallest Euclidean distance. The Euclidean distance is obtained by summing the square roots of the sums of the squared differences between the clustering feature vector and the current cluster center in each field dimension. This completes the clustering process. After class assignment, the mean of the cluster feature vectors contained in each cluster category is calculated, and the mean result is used as the updated cluster center. The mean is calculated by summing the values ​​of each cluster feature vector in the same field dimension and dividing by the number of cluster feature vectors in that category. The position change magnitude between the updated cluster center and the current cluster center before the update is calculated and compared with a preset threshold. The position change magnitude is calculated by the Euclidean distance or the maximum dimensional difference of the corresponding cluster center vectors. When the position change magnitude of all cluster centers is less than the preset threshold, the updated cluster center is output as the converged K-Means cluster center and the iterative calculation process ends. When the position change magnitude of any cluster center is not less than the preset threshold, the updated cluster center is set as the current cluster center and the next round of iterative calculation process continues until the convergence condition is met.

[0064] S404: Based on the converged K-Means cluster centers, calculate the Euclidean distance between each sample in the clustered dataset and each cluster center, and determine the cluster category to which each sample in the clustered dataset belongs according to the minimum distance principle, thus obtaining the K-Means clustering results.

[0065] Specifically, the converged K-Means cluster centers are read and represented as cluster center vectors. The clustering dataset is read, and for each cluster entry in the dataset, the bounce data field, self-test field, and third-party field are extracted to form a sample feature vector. The sample feature vector is normalized using the same normalization rule as S403 to obtain a normalized sample feature vector. For each normalized sample feature vector, the Euclidean distance between it and each cluster center vector is calculated, and a set of Euclidean distances is obtained. The Euclidean distance is obtained by taking the square root of the sum of the squared differences between the normalized sample feature vector and the cluster center vector in each field dimension. The minimum distance principle means that the cluster center corresponding to the minimum Euclidean distance in the set of Euclidean distances is selected as the belonging cluster center. The cluster category identifier corresponding to the belonging cluster center is written into the cluster category field of the cluster entry to determine the cluster category to which the cluster entry belongs. The Euclidean distance calculation and cluster category identifier writing are repeated until the clustering dataset is traversed. The cluster category identifiers of each cluster entry in the clustering dataset are summarized to obtain the K-Means clustering result.

[0066] In one embodiment, step S5, namely, based on the K-Means clustering results, analyzes the data distribution of the fused dataset within the same enterprise and project, identifies anomalous matching relationships, and calculates comparative outliers based on these anomalous matching relationships, including: S501: Based on the K-Means clustering results, calculate the statistical information of the fused dataset within its respective cluster category, and construct a data matching feature matrix based on the statistical information.

[0067] Specifically, the cluster category identifiers from the K-Means clustering results are written into the cluster category field of the fused dataset to establish a correspondence between fused entries and cluster category identifiers. The fused dataset is then grouped according to the cluster category identifiers to obtain multiple fused subsets. For each fused subset, statistical calculations are performed on the enterprise self-inspection data field, the third-party inspection data field, and the rebound data field to obtain statistical information. The statistical calculations include calculating the mean, standard deviation, maximum value, minimum value, and quantiles. The statistical calculation results are then indexed and stored according to the field name and statistical item name to form intra-category statistical information. A data matching feature matrix is ​​constructed based on the intra-category statistical information. The data matching feature matrix is ​​a matrix structure with the cluster category identifier as the row index and the statistical items of the enterprise self-inspection data field, the third-party inspection data field, and the rebound data field as the column index. The intra-category statistical information corresponding to each cluster category identifier is filled into the matrix element positions of the data matching feature matrix according to the row index and column index, and the missing element verification of the matrix is ​​completed to obtain the data matching feature matrix.

[0068] S502: Based on the data matching feature matrix, calculate the deviation value of the fused dataset in the same enterprise and project.

[0069] Specifically, the fused dataset is aggregated using enterprise and project identifiers as indexes to obtain a set of fused entries corresponding to enterprises and projects. The clustering category identifiers corresponding to each fused entry in the fused entry set are read, and the corresponding category statistical benchmark vectors are retrieved from the data matching feature matrix based on these clustering category identifiers. The category statistical benchmark vector refers to the set of matrix elements in the data matching feature matrix corresponding to the clustering category identifier. For each fused entry in the fused entry set, the enterprise self-inspection data field, third-party detection data field, and rebound data field are extracted to form an entry feature vector. The difference between the entry feature vector and the category statistical benchmark vector is calculated in the corresponding field dimension to obtain a field deviation vector. The difference calculation is obtained by subtracting the corresponding field statistical benchmark value of the category statistical benchmark vector from the field value of the entry feature vector. An aggregation calculation is performed based on the field deviation vector to obtain the entry deviation value of the fused entry. The aggregation calculation includes summing or maximizing the absolute values ​​of the field deviation vectors to obtain the entry deviation value. A summary calculation is performed based on the entry deviation values ​​of each fused entry in the fused entry set to obtain the deviation value corresponding to the same enterprise and project. The summary calculation includes performing an arithmetic mean or weighted average on the entry deviation values ​​to obtain the deviation value.

[0070] S503: Calculate the contrast outliers of the fused dataset based on the deviation values.

[0071] Specifically, the fused dataset is aggregated using enterprise and project identifiers as indexes to obtain a set of fused entries corresponding to enterprises and projects. The feature vectors of each fused entry in the set of fused entries are read, and the deviation values ​​obtained in S502 are used as the comparison benchmark values. Deviation comparison processing is performed on each fused entry in the set of fused entries to obtain the entry comparison result information. Deviation comparison processing includes calculating the difference between the feature vector of the fused entry and the deviation value, and comparing the difference with a preset comparison anomaly threshold. When the difference is greater than the comparison anomaly threshold, the fused entry is marked with a comparison anomaly label. Based on the marking of the comparison anomaly labels, a summary calculation is performed on the fused entry set to obtain the comparison anomaly value. The summary calculation includes counting the number of occurrences of the comparison anomaly label or calculating the sum of the differences corresponding to the comparison anomaly labels to obtain the comparison anomaly value.

[0072] In one embodiment, step S8, namely, based on the central station's monitoring data, performs monitoring anomaly identification processing and counts the number of monitoring anomalies, and calculates a monitoring detection score based on the number of monitoring anomalies, includes: S801: Analyze the monitoring data from the main station and extract the monitoring data fields, which include the monitoring and detection intensity value, the design strength requirement value, and the rebound monitoring value.

[0073] Specifically, the monitoring and inspection records are read from the main station's monitoring data, and the monitoring and inspection intensity field, design strength requirement field, and rebound monitoring field are located according to the preset field mapping rules. The preset field mapping rules are used to determine the correspondence between field names, field positions, and field meanings. The monitoring and inspection intensity field, design strength requirement field, and rebound monitoring field are processed to obtain the monitoring and inspection intensity value, design strength requirement value, and rebound monitoring value. When the field content is in text format, the field content is converted to numerical format according to the preset numerical parsing rules, and abnormal character removal is performed. The monitoring and inspection intensity value and design strength requirement value are checked for consistency of measurement units, and if the measurement units are inconsistent, they are converted to consistent measurement units according to the preset unit conversion rules. The rebound monitoring value is checked for the range of values ​​to exclude missing values ​​and illegal values. The monitoring and inspection records corresponding to missing values ​​or illegal values ​​are removed or marked as missing. After completing the field location, value retrieval, parsing, verification, and validation, the monitoring and inspection intensity value, design strength requirement value, and rebound monitoring value are obtained.

[0074] S802: Using enterprises and projects as statistical objects, the monitoring data fields are aggregated to obtain a set of monitoring data entries corresponding to enterprises and projects.

[0075] Specifically, the system reads the central station's monitoring data and extracts the enterprise identifier and project identifier from each monitoring and inspection record, as well as the record time identifier corresponding to the monitoring and inspection record. The enterprise identifier is the identification information used to identify the enterprise to which the monitoring and inspection record belongs, the project identifier is the identification information used to identify the project to which the monitoring and inspection record belongs, and the record time identifier is the time information used to characterize the collection time or inspection time of the monitoring and inspection record. Based on the enterprise identifier and project identifier, the monitoring and inspection records are grouped and aggregated to generate multiple group sets. Each group set corresponds to a combination of enterprise identifiers and project identifiers. The monitoring and inspection intensity value, design intensity requirement value, and rebound monitoring value are extracted from the monitoring and inspection records in each group set, and time sorting is performed according to the record time identifier. Time sorting refers to arranging the monitoring and inspection records from earliest to latest according to the record time identifier. The time-sorted monitoring and inspection records in the same group set are written into the monitoring data entry set, and the enterprise identifier and project identifier are written into the index field of the monitoring data entry set to establish the correspondence between the monitoring data entry set and the enterprise and project. The grouping and aggregation, field extraction, time sorting, and writing processes are repeated until the central station's monitoring data is traversed to obtain the monitoring data entry set corresponding to the enterprise and project.

[0076] S803: For each supervision data entry in the supervision data entry set, calculate the strength deviation value between the supervision detection strength value and the design strength requirement value, and determine the rebound supervision deviation value based on the rebound supervision value.

[0077] Specifically, the set of supervision data entries is read and each supervision data entry is selected sequentially. A difference calculation is performed between the supervision and inspection intensity value and the design strength requirement value in the supervision data entry to obtain the strength deviation value. The difference calculation includes subtracting the design strength requirement value from the supervision and inspection intensity value and recording the difference as the strength deviation value. A unit consistency verification is performed on the strength deviation value, and if the verification fails, it is converted to a consistent unit according to a preset unit conversion rule before the difference calculation is performed again. The rebound supervision deviation value is determined based on the rebound supervision value in the supervision data entry, including determining the value used to characterize the rebound supervision. The center value of the value set is calculated, and the deviation of the rebound monitoring value relative to the center value is calculated. The center value refers to the statistical value used to characterize the clustering level of the rebound monitoring values. The center value is obtained by performing statistical calculations on the rebound monitoring values ​​corresponding to the same enterprise identifier and project identifier in the monitoring data entry set. The statistical calculation includes obtaining the center value by using the arithmetic mean method or the median method. The deviation is obtained by subtracting the center value from the rebound monitoring value and taking the absolute value of the difference. After completing the calculation of the intensity deviation value and the rebound monitoring deviation value, the intensity deviation value and the rebound monitoring deviation value are written into the deviation field of the monitoring data entry.

[0078] S804: Perform a first comparison and judgment process on the strength deviation value, and perform a second comparison and judgment process on the rebound monitoring deviation value to obtain the first judgment result information and the second judgment result information.

[0079] Specifically, the strength deviation value and rebound supervision deviation value corresponding to each supervision data entry in the supervision data entry set are read, and the strength anomaly judgment parameter corresponding to the first comparison judgment process and the rebound anomaly judgment parameter corresponding to the second comparison judgment process are obtained respectively. The strength anomaly judgment parameter refers to the parameter information used to compare and judge the strength deviation value and includes the strength deviation direction judgment parameter and the first strength deviation threshold. The rebound anomaly judgment parameter refers to the parameter information used to compare and judge the rebound supervision deviation value and includes the rebound deviation threshold. The first comparison judgment process includes performing a comparison operation between the strength deviation value and the strength anomaly judgment parameter and outputting the first judgment result information. The comparison operation includes judging whether the strength deviation value meets the strength deviation direction judgment parameter and judging whether the strength deviation value exceeds the first strength deviation threshold, and writing the judgment conclusion into the first judgment result information. The second comparison judgment process includes performing a comparison operation between the rebound supervision deviation value and the rebound anomaly judgment parameter and outputting the second judgment result information. The comparison operation includes judging whether the rebound supervision deviation value exceeds the rebound deviation threshold and writing the judgment conclusion into the second judgment result information. The first judgment result information and the second judgment result information are bound to the corresponding supervision data entry and written into the judgment field of the supervision data entry.

[0080] S805: Perform supervision anomaly judgment processing based on the first judgment result information and the second judgment result information, and determine that a supervision anomaly event has occurred when the preset supervision anomaly judgment rules are met.

[0081] Specifically, the first judgment result information and the second judgment result information corresponding to each supervision data entry in the supervision data entry set are read, and a preset supervision anomaly judgment rule is obtained. The preset supervision anomaly judgment rule refers to the rule information used to convert the first judgment result information and the second judgment result information into a supervision anomaly event judgment conclusion. The supervision anomaly judgment processing based on the first judgment result information and the second judgment result information includes performing rule matching judgment on the first judgment result information and the second judgment result information and generating supervision anomaly judgment result information. The rule matching judgment includes judging whether the first judgment result information indicates an intensity anomaly and judging whether the second judgment result information indicates a rebound anomaly. The supervision anomaly judgment result information is determined to be an anomaly according to the preset supervision anomaly judgment rule. When the supervision anomaly judgment result information is an anomaly, the supervision data entry is marked as a supervision anomaly event, and an event identifier and an event time identifier are generated for the supervision anomaly event. The event identifier refers to the identifier information used to identify the supervision anomaly event, and the event time identifier is determined by the record time identifier corresponding to the supervision data entry. The event identifier, the event time identifier and the supervision data entry are bound together and written into the event field of the supervision data entry.

[0082] S806: Accumulate statistics on abnormal monitoring events to obtain the number of abnormal monitoring events for each enterprise or project. Based on the number of abnormal monitoring events and the preset scoring mapping rules, perform linear mapping processing on the number of abnormal monitoring events to obtain the monitoring and detection score for each enterprise or project.

[0083] Specifically, the monitoring data entries are aggregated based on enterprise and project identifiers, and monitoring data entries marked as monitoring anomalies are extracted from the aggregation results. For each set of aggregation results corresponding to enterprise and project identifiers, the occurrence frequency of monitoring anomalies is counted, and the result is recorded as the monitoring anomaly occurrence frequency. The monitoring anomaly occurrence frequency refers to the number of monitoring data entries corresponding to either the enterprise or project identifier and judged as monitoring anomalies within a preset statistical period. A preset scoring mapping rule is obtained, and the linear mapping interval parameter in the scoring mapping rule is determined. The linear mapping interval parameter refers to the parameter information used to limit the mapping relationship between the monitoring anomaly occurrence frequency and the monitoring detection scoring, and includes the anomaly occurrence frequency. The system uses a set of lower and upper bounds for the number of monitored anomalies, anomaly counts, and scores. It then performs interval truncation on the monitored anomaly count to obtain a truncated anomaly count value. Interval truncation means that when the monitored anomaly count is less than the lower bound, the count is replaced with the lower bound; and when the count is greater than the upper bound, the count is replaced with the upper bound. Based on the truncated anomaly count value and the linear mapping interval parameters, a linear mapping operation is performed to obtain the monitored detection score. The linear mapping operation uses the following expression to obtain the original value of the monitored detection score, and then performs integer or decimal place rounding on the original value to obtain the monitored detection score. Where m is the number of anomalies truncated. This is the lower bound of the number of anomalies. This is the upper bound of the number of anomalies. This is the lower bound of the rating. Q3 is the upper limit of the score. The supervision and inspection score will be linked to the enterprise identifier or project identifier and written into the supervision and inspection score record.

[0084] In one embodiment, in step S9, the analytic hierarchy process (AHP) is used to perform a comprehensive weighted calculation of the single-dimensional quality score, the multi-dimensional quality score, and the supervision and inspection score to obtain a comprehensive quality score, including: S901: Establish a hierarchical analysis structure model based on single-dimensional quality scoring, multi-dimensional quality scoring, and supervision and inspection scoring.

[0085] Specifically, the modeling object of the hierarchical analysis structure model is clearly defined as the comprehensive quality score, which is set as the decision-making layer. Based on the composition of the comprehensive quality score, the criterion layer is determined to include production quality, construction quality, and supervision and inspection, and these are written into the criterion layer node set respectively. A unique node identifier is assigned to each criterion layer node for subsequent judgment matrix reference. A membership relationship is established between the criterion layer node set and the comprehensive quality score, and this membership relationship is represented as production quality pointing to the comprehensive quality score, construction quality pointing to the comprehensive quality score, and supervision and inspection pointing to the comprehensive quality score. Based on the score inputs corresponding to production quality, construction quality, and supervision and inspection, the indicator layer is determined to include single-dimensional quality scores, multi-dimensional quality scores, and supervision and inspection scores, and these are written into the indicator layer node set respectively. A membership relationship is established between the indicator layer node set and the criterion layer node set, and this membership relationship is represented as single-dimensional quality scores belonging to production quality, multi-dimensional quality scores belonging to construction quality, and supervision and inspection scores belonging to supervision and inspection. The structured representation results of the decision-making layer nodes, the criterion layer node set, the indicator layer node set, and the membership relationships are summarized to obtain the hierarchical analysis structure model.

[0086] S902: Based on the hierarchical analysis structure model, construct the judgment matrix, and use the eigenvector method to calculate the weight allocation result.

[0087] Specifically, based on the hierarchical analysis structure model, the criteria layer includes production quality, construction quality, and supervision and inspection, and the row and column order of the judgment matrix is ​​determined. The judgment matrix is ​​used to represent the pairwise relative importance of each indicator in the criteria layer. According to the preset pairwise comparison rules, the relative importance ratios of production quality and construction quality, production quality and supervision and inspection, and construction quality and supervision and inspection are given and written into the corresponding positions in the judgment matrix. The pairwise comparison rules refer to using a proportional scale to represent relative importance, and the value of the proportional scale is determined by a preset scaling table. The diagonal elements of the judgment matrix are all 1, and the matrix is ​​completed using a reciprocal rule. The rule that the remaining elements of the judgment matrix are reciprocals means that the element in the i-th row and j-th column of the judgment matrix is ​​the reciprocal of the element in the j-th row and i-th column. Based on the constructed judgment matrix, the eigenvector method is performed. The eigenvector method means finding the eigenvector corresponding to the largest eigenvalue of the judgment matrix and normalizing the eigenvector to obtain the weight allocation result. The normalization process of the eigenvector uses the sum of the vector elements as the normalization factor and divides each element by the normalization factor to obtain the weight allocation result. The weight allocation result includes the weight corresponding to the single-dimensional quality score, the weight corresponding to the multi-dimensional quality score, and the weight corresponding to the supervised detection score.

[0088] S903: Based on the weighting results, the overall quality score is calculated using the following formula: , where Q total This refers to the overall quality score. W1 refers to the weight of the single-dimensional quality score, Q1 is the single-dimensional quality score, W2 refers to the weight of the multi-dimensional quality score, Q2 is the multi-dimensional quality score, W3 refers to the weight of the supervision and inspection score, and Q3 is the supervision and inspection score.

[0089] Specifically, the weight allocation results are read and W1 corresponding to the single-dimensional quality score, W2 corresponding to the multi-dimensional quality score, and W3 corresponding to the supervised detection score are extracted. The single-dimensional quality score is read and recorded as Q1, the multi-dimensional quality score is read and recorded as Q2, and the supervised detection score is read and recorded as Q3. Multiplication operations are performed on each of these to obtain the results. , and The multiplication operation obtains the weighted score by multiplying the corresponding weight by the corresponding score, and then performs an addition operation to obtain Q. total The addition summation operation obtains the overall quality score by successively adding the three weighted scoring items.

[0090] In one embodiment, after step S9, i.e., the comprehensive quality score, the method further includes: S904: Based on the comprehensive quality score, the quality control status of each enterprise is classified and marked. Enterprises with good quality control are marked in green, enterprises with medium quality control are marked in yellow, and enterprises with poor quality control are marked in red.

[0091] Specifically, firstly, based on the companies' historical quality scores, construction compliance, and recent comprehensive quality scores, all companies involved in concrete production and construction are categorized to ensure that the scoring data covers the companies' quality performance in different projects. Then, quality scoring thresholds are set, and specific scoring ranges for green, yellow, and red markings are determined. If a company's comprehensive quality score is higher than the set high-quality threshold, it is marked as green, indicating that the company has a high level of quality management and stable construction quality. If the comprehensive quality score is in the normal range but lower than the high-quality threshold, it is marked as yellow, indicating that the company's quality control capabilities are average and require appropriate monitoring. If the comprehensive quality score is lower than the set low-quality threshold, it is marked as red, indicating that the company has significant problems with quality control and requires close attention.

[0092] S905: For newly started engineering projects, establish a quality tracking mechanism, obtain quality marking results, and generate early warning data based on the quality marking results.

[0093] In this embodiment, the quality tracking mechanism refers to a dynamic monitoring and feedback system based on comprehensive quality scores, which is used to continuously evaluate the quality control level of enterprises and projects during the construction process, and to take timely and targeted quality optimization measures based on the changing trends of quality scores.

[0094] Specifically, firstly, a preliminary quality assessment is conducted on the construction unit and the project itself for newly started projects. This assessment combines the construction company's historical quality scores with the current project's overall quality score to determine the project's initial quality status. Then, during construction, quality data is continuously monitored, tracking changes in single-dimensional quality scores, multi-dimensional quality scores, and supervision and inspection scores. If a downward trend in the quality score is detected or a warning threshold is reached during construction, the project's quality labeling results are dynamically updated. Based on this, quality warning data is generated according to the company's quality labeling and the project's quality score. If both the construction unit and the project are marked in red... A red quality warning is generated if the color is specified, indicating the highest quality risk. A red-yellow quality warning is generated if the construction unit's quality management is poor but the overall project is still under control. A yellow-red quality warning is generated if the project's quality score is low but the construction unit's management capabilities are acceptable. A red-blue quality warning is generated if the construction unit has quality risks but the project itself has a high quality score, indicating that the construction unit's quality management needs improvement but the project still maintains a good quality status. Finally, the warning data is stored in the concrete quality monitoring database and relevant management personnel are notified via SMS, email, or app push notifications to ensure that quality problems are detected and addressed promptly, ultimately yielding the warning data.

[0095] In one embodiment, step S10, i.e., the warning data, further includes: S101: Based on the early warning data, the concrete quality early warning data is displayed through data visualization. The data visualization includes drawing early warning analysis bar charts, concrete strength distribution maps, and abnormal percentage maps of monitored objects.

[0096] Specifically, based on early warning data, abnormal situations of different enterprises and projects are extracted, including abnormality categories, abnormality levels, affected construction areas, and enterprise information. The abnormal data is then grouped according to set classification standards to ensure data integrity and accurate visualization. Next, based on the abnormal data, an early warning analysis bar chart is created to compare and analyze abnormal situations of different enterprises or projects, highlighting enterprises or construction projects with high abnormality rates. Abnormality types are marked on the charts, enabling managers to quickly identify the source of quality problems. Furthermore, combined with concrete strength testing data, a concrete strength distribution map is created to show the strength variation trends and stability of different batches of concrete, ensuring managers can intuitively understand changes in concrete quality during construction. Simultaneously, the abnormality rate of different monitoring objects is calculated, categorized and statistically analyzed according to enterprise, project, or construction area, and presented through pie charts or bar charts to intuitively show the distribution of abnormal situations across different monitoring objects, ultimately completing the visualization of concrete quality early warning data.

[0097] S102: Based on early warning data, send early warning notifications to managers via SMS, email, or APP. The early warning notifications include the type and severity of the anomaly.

[0098] Specifically, based on the anomaly category and severity in the early warning data, the anomalies are categorized to ensure that management personnel can take appropriate measures according to the severity of the anomalies. Then, according to the management personnel information preset in the system, the target personnel who need to receive the early warning notification are identified, including quality supervisors, construction managers, and enterprise managers, and appropriate notification methods are selected. For example, for more urgent quality anomalies, real-time early warning notifications are sent via SMS or APP push, while for general quality fluctuations, detailed reports can be provided via email. Next, the early warning notification content is generated, including the anomaly category, severity, affected construction area or enterprise information, and suggested handling solutions, ensuring that management personnel can quickly understand the early warning information and take necessary measures, and finally, the early warning notification is sent.

[0099] In one embodiment, after step S10, i.e., the ready-mixed concrete quality early warning method, it further includes: S104: Construct a comprehensive database to store multi-dimensional quality scores, supervision and inspection scores, comprehensive quality scores, and early warning data.

[0100] In this embodiment, the integrated database refers to a structured data management system used to uniformly store and manage multi-source data related to concrete quality testing, so as to ensure data integrity, traceability and efficient retrieval capabilities.

[0101] Specifically, a database instance is created to store multi-dimensional quality scores, supervision and inspection scores, comprehensive quality scores, and early warning data. A data table structure is defined for the database instance, including a scoring data table and an early warning data table. The scoring data table stores multi-dimensional quality scores, supervision and inspection scores, and comprehensive quality scores bound to enterprise identifiers, project identifiers, and statistical period identifiers. Field constraint rules are set for the scoring data table to limit field types and value ranges. The early warning data table stores early warning data bound to enterprise identifiers, project identifiers, and early warning time identifiers. Field constraint rules are set for the early warning data table to limit the anomaly category field and the anomaly severity field. The system defines the value range, establishes a joint index for enterprise identifiers and project identifiers, and establishes time indexes for statistical period identifiers and early warning time identifiers to support retrieval by enterprise and project dimensions and time dimension. It sets data writing rules to ensure that multi-dimensional quality scores, supervision and inspection scores, and comprehensive quality scores are written to the scoring data table after generation, and that early warning data is written to the early warning data table after generation. It sets data consistency verification rules to verify the integrity of enterprise identifiers, project identifiers, and statistical period identifiers before writing, and performs missing mark or write rejection processing when the verification fails. After completing the data table structure setting, index creation, writing rule setting, and consistency verification rule setting, a comprehensive database is obtained.

[0102] S105: During a preset period, the early warning data in the comprehensive database is compared with the actual engineering quality data to obtain the comparison results. Based on the comparison results, the early warning error is calculated. If there is a deviation between the early warning data and the actual engineering quality data, the deviation value is obtained. Based on the deviation value, the parameters of Z-Score analysis, K-Means clustering analysis, and analytic hierarchy process are adjusted.

[0103] Specifically, the cycle for comparing early warning data is set, such as weekly, monthly, or periodically extracting early warning data from the comprehensive database at each construction stage, and obtaining actual engineering quality data to ensure the real-time nature and accuracy of the comparison data. Then, according to the construction project, construction area, or enterprise classification, historical early warning data and actual quality data are matched one-to-one to ensure the consistency of the comparison data. During the comparison process, error calculation methods are used to analyze the differences between early warning data and actual engineering quality data. For example, the early warning anomaly score is compared with the actual detected key quality indicators such as concrete strength and construction stability, and the deviation value between the two is calculated. If a significant deviation is found between the early warning data and the actual quality data, the cause of the deviation is further analyzed to determine whether it is due to data acquisition errors, unreasonable model parameter settings, or environmental factors that lead to a decrease in early warning accuracy. Based on the deviation value, the key parameters of Z-Score analysis, K-Means cluster analysis, and analytic hierarchy process are optimized and adjusted. For example, the standard deviation range of Z-Score is adjusted, the initialization strategy of K-Means cluster centers is optimized, and the weight allocation method in analytic hierarchy process is corrected to make subsequent early warning analysis more consistent with the actual engineering situation. Finally, the optimization and adjustment of model parameters are completed.

[0104] It should be understood that the sequence number of each step in the above embodiments does not imply the order of execution. The execution order of each process should be determined by its function and internal logic, and should not constitute any limitation on the implementation process of the embodiments of this application.

Claims

1. A method for early warning of the quality of ready-mixed concrete, characterized in that, The method for early warning of ready-mixed concrete quality includes: Acquire enterprise self-inspection data, third-party testing data, and rebound data; Using Z-Score analysis, anomaly scores were assigned to the enterprise's self-inspection data, third-party testing data, and rebound data, resulting in enterprise self-inspection anomaly scores, third-party testing anomaly scores, and rebound anomaly scores. Based on the principle of equal weight, the enterprise self-inspection anomaly scores, third-party inspection anomaly scores, and rebound anomaly scores are merged to form an anomaly score dataset. Based on the anomaly score dataset, a single-dimensional quality score is calculated. Using K-Means clustering analysis, with rebound data as the core, the self-inspection data, third-party testing data, and rebound data of enterprises are fused to form a fused dataset. K-Means clustering analysis is then performed on the fused dataset to obtain the K-Means clustering results. Based on the K-Means clustering results, the data distribution of the fused dataset in the same enterprise and project is analyzed, abnormal matching relationships are identified, and comparative outliers are calculated based on the abnormal matching relationships. Based on the comparison of outliers, the anomaly scores of enterprises and projects are calculated to quantify the degree of anomalies of each enterprise and project in construction quality control, and to obtain multi-dimensional quality scores. Obtain monitoring data from the main station; Based on the monitoring data from the main station, perform monitoring anomaly identification and processing, count the number of monitoring anomalies, and calculate the monitoring detection score based on the number of monitoring anomalies. The analytic hierarchy process is used to comprehensively and weightedly calculate the single-dimensional quality score, the multi-dimensional quality score, and the supervision and inspection score to obtain the comprehensive quality score. The overall quality score is compared with a preset quality score threshold. If the overall quality score is lower than the preset quality score threshold, an early warning data is generated.

2. The method for early warning of ready-mixed concrete quality according to claim 1, characterized in that, Based on the principle of equal weighting, the enterprise's self-inspection anomaly scores, third-party inspection anomaly scores, and rebound anomaly scores are fused to form an anomaly score dataset. Based on this dataset, a single-dimensional quality score is calculated, including: Based on the principle of equal weighting, the enterprise self-inspection anomaly score, the third-party inspection anomaly score, and the rebound anomaly score are weighted and calculated to obtain the weighted anomaly score result. The weighted anomaly scoring results are normalized to obtain standardized anomaly scoring results; Based on the standardized anomaly scoring results, the arithmetic mean method was used to calculate the single-dimensional quality score.

3. The method for early warning of ready-mixed concrete quality according to claim 1, characterized in that, The method employs K-Means clustering analysis, using rebound data as the core, to perform fusion processing on enterprise self-inspection data, third-party testing data, and rebound data to form a fused dataset. K-Means clustering analysis is then performed on this fused dataset to obtain the K-Means clustering results, including: Using rebound data as the core, and matching it with enterprise self-inspection data and third-party testing data, a clustered dataset is formed; Based on the rebound data, and according to the distribution characteristics of the rebound data, the initial cluster centers of K-Means are set; Based on the clustering dataset, the K-Means clustering analysis method is used for iterative calculation. The initial cluster centers are adjusted according to the rebound data, enterprise self-inspection data and third-party detection data until the position change of the initial cluster centers is less than the preset threshold, and the converged K-Means cluster centers are obtained. Based on the converged K-Means cluster centers, the Euclidean distance between each sample in the clustered dataset and each cluster center is calculated. Then, according to the minimum distance principle, the cluster category to which each sample in the clustered dataset belongs is determined, and the K-Means clustering result is obtained.

4. The method for early warning of ready-mixed concrete quality according to claim 1, characterized in that, The process of analyzing the data distribution of the fused dataset within the same enterprise and project based on K-Means clustering results, identifying anomalous matching relationships, and calculating comparative outliers based on these relationships includes: Based on the K-Means clustering results, the statistical information of the fused dataset within its respective cluster category is calculated, and a data matching feature matrix is ​​constructed based on the statistical information; Based on the data matching feature matrix, calculate the deviation value of the fused dataset in the same enterprise and project; Based on the deviation value, calculate the contrast outliers of the fused dataset.

5. The method for early warning of ready-mixed concrete quality according to claim 1, characterized in that, The process of identifying and processing supervisory anomalies based on the central station's monitoring data, counting the number of anomalies, and calculating a supervisory detection score based on the number of anomalies includes: The monitoring data from the main station is analyzed to extract monitoring data fields, which include monitoring and detection intensity values, design strength requirement values, and rebound monitoring values. Using enterprises and projects as statistical objects, the monitoring data fields are aggregated and processed to obtain a set of monitoring data items corresponding to enterprises and projects; For each supervision data entry in the supervision data entry set, calculate the strength deviation value between the supervision detection strength value and the design strength requirement value, and determine the rebound supervision deviation value based on the rebound supervision value; A first comparison and judgment process is performed on the strength deviation value, and a second comparison and judgment process is performed on the rebound monitoring deviation value to obtain the first judgment result information and the second judgment result information. Based on the first judgment result information and the second judgment result information, the supervision anomaly judgment processing is performed, and when the preset supervision anomaly judgment rules are met, a supervision anomaly event is judged to have occurred. The number of abnormal monitoring events is accumulated and statistically analyzed to obtain the number of abnormal monitoring events for each enterprise or project. Based on the number of abnormal monitoring events and the preset scoring mapping rules, a linear mapping process is performed on the number of abnormal monitoring events to obtain the monitoring and detection score for each enterprise or project.

6. The method for early warning of ready-mixed concrete quality according to claim 1, characterized in that, The analytic hierarchy process (AHP) is used to comprehensively and weightedly calculate the single-dimensional quality score, multi-dimensional quality score, and supervision and inspection score to obtain the comprehensive quality score, which includes: A hierarchical analysis structure model is established based on single-dimensional quality scoring, multi-dimensional quality scoring, and supervision and inspection scoring. Based on the hierarchical analysis structure model, a judgment matrix is ​​constructed. Based on the judgment matrix, the eigenvector method is used to calculate the weight allocation result. Based on the weighting results, the overall quality score is calculated using the following formula: , where Q total This refers to the overall quality score. W1 refers to the weight of the single-dimensional quality score, Q1 is the single-dimensional quality score, W2 refers to the weight of the multi-dimensional quality score, Q2 is the multi-dimensional quality score, W3 refers to the weight of the supervision and inspection score, and Q3 is the supervision and inspection score.

7. The method for early warning of ready-mixed concrete quality according to claim 6, characterized in that, The overall quality score then includes: Based on the comprehensive quality score, the quality control status of each enterprise is classified and marked. Enterprises with good quality control are marked in green, enterprises with medium quality control are marked in yellow, and enterprises with poor quality control are marked in red. For newly started construction projects, a quality tracking mechanism is established to obtain quality marking results, and early warning data is generated based on the quality marking results.

8. The method for early warning of ready-mixed concrete quality according to claim 1, characterized in that, The warning data also includes: Based on the early warning data, the concrete quality early warning data is displayed through data visualization. The data visualization includes drawing early warning analysis bar charts, concrete strength distribution maps, and abnormal percentage maps of monitored objects. Based on the early warning data, early warning notifications are sent to managers via SMS, email, or APP. The early warning notifications include the type and severity of the anomaly.

9. The method for early warning of ready-mixed concrete quality according to claim 1, characterized in that, The method for early warning of ready-mixed concrete quality also includes: Construct a comprehensive database to store multi-dimensional quality scores, supervision and inspection scores, comprehensive quality scores, and early warning data. During a preset period, the early warning data in the comprehensive database is compared with the actual engineering quality data to obtain the comparison results. Based on the comparison results, the early warning error is calculated. If there is a deviation between the early warning data and the actual engineering quality data, the deviation value is obtained. Based on the deviation value, the parameters of Z-Score analysis, K-Means clustering analysis, and analytic hierarchy process are adjusted.

10. A computer-readable storage medium storing a computer program, characterized in that, When the computer program is executed by the processor, it implements the steps of the ready-mixed concrete quality early warning method as described in any one of claims 1 to 9.

Images