Construction site multi-source data collaborative analysis and early warning system for supervision business

Abstract

The present application relates to the technical field of data processing, and more particularly to a construction site multi-source data collaborative analysis and early warning system for supervision business, comprising: a data processing module, which collects multi-source construction data through Internet of Things sensors and pre-processes the multi-source construction data to output construction characteristics; an analysis and early warning module, which trains an initial model according to the construction characteristics to obtain a deep learning model, uses the deep learning model to analyze the multi-source construction data to obtain early warning information; a collaborative interaction module, which is used to enable several users to share and collaborate information; an early warning judgment module, which judges whether the generation validity of the early warning information meets the requirements according to early warning characteristic values; an association adjustment module, which determines the time sequence association coupling threshold of the construction characteristics according to the noise proportion of the construction characteristics; and a confidence adjustment module, which determines the reordering confidence threshold of the multi-source construction data according to the redundancy rate of the multi-source construction data. The present application improves the management effectiveness of supervision business.

Description

Technical Field

[0001] This invention relates to the field of data processing technology, and in particular to a collaborative analysis and early warning system for multi-source data at construction sites for construction supervision business. Background Technology

[0002] In existing technologies, traditional construction site supervision relies heavily on manual inspections, single-point equipment monitoring, and simple threshold alarms, resulting in low levels of intelligence and collaboration. While some supervision systems have introduced robots, video, or sensor equipment, they generally suffer from limited data collection dimensions and incomplete coverage, making it difficult to achieve unified access and fusion analysis of multi-source data from personnel, equipment, environment, structure, and processes. Furthermore, inconsistent time bases and large fluctuations in transmission delays among various data sources easily lead to data time sequence misalignment, out-of-order distortion, and a lack of effective time sequence alignment, feature filtering, and quality self-verification mechanisms, failing to provide reliable data support for risk warnings. Most existing systems use fixed parameters and preset rules, lacking the ability to adaptively adjust based on warning effectiveness, data quality, and network conditions. They also lack a closed-loop optimization mechanism from warning effectiveness to feature filtering and data transmission, resulting in weak identification capabilities for potential and unknown risks, high rates of missed and false alarms, and insufficient management effectiveness in supervision operations.

[0003] Chinese Patent Publication No. CN110705917A discloses a robot supervision system for construction site applications, comprising an information acquisition system installed at the construction site and construction equipment, an information control center communicating with the information acquisition system, and a client connected to the information control center. The information acquisition system includes a data monitoring module installed on the construction equipment and a drone robot installed at the construction site. The data monitoring module includes a cement slurry specific gravity acquisition and measurement module and a radar ranging module installed on the aggregate hopper, a pressure acquisition module installed on the pile driver, and a drill rod drilling speed monitoring module and a drill rig rotation speed monitoring module installed on the drilling rig. Therefore, the robot supervision system for construction site applications suffers from insufficient management effectiveness due to its single data acquisition dimension, incomplete coverage, and lack of multi-source data collaboration mechanisms and parameter self-adjustment and closed-loop optimization mechanisms. Summary of the Invention

[0004] To address this, the present invention provides a multi-source data collaborative analysis and early warning system for construction site supervision, which overcomes the problems of insufficient management effectiveness in supervision due to the single data collection dimension, incomplete coverage, lack of multi-source data collaboration mechanism and parameter self-adjustment and closed-loop optimization mechanism in the existing technology.

[0005] To achieve the above objectives, this invention provides a multi-source data collaborative analysis and early warning system for construction sites, oriented towards supervision operations, comprising: The data processing module includes an acquisition unit for collecting multi-source construction data from the construction site via IoT sensors, and a preprocessing unit connected to the acquisition unit for preprocessing the multi-source construction data to output construction features. An analysis and early warning module, which is connected to the data processing module, includes a model training unit for training an initial model based on the construction characteristics to obtain a deep learning model, an analysis unit connected to the model training unit for analyzing the multi-source construction data based on the deep learning model to obtain analysis results, and an early warning unit connected to the analysis unit for generating early warning information based on the analysis results. A collaborative interaction module, which is connected to the analysis and early warning module, is used to enable several users to share information and collaborate. The early warning determination module, which is connected to the analysis and early warning module, is used to determine whether the validity of the early warning information generation meets the requirements based on the early warning representation value determined by the accuracy of the early warning information and the false negative rate of the deep learning model. The correlation adjustment module, which is connected to the early warning determination module, is used to determine the temporal correlation coupling threshold of the construction feature based on the noise ratio of the construction feature when the validity of the generated early warning information does not meet the requirements. A confidence adjustment module, which is connected to the association adjustment module, is used to determine the reordering confidence threshold of multi-source construction data based on the redundancy rate of multi-source construction data.

[0006] Furthermore, the early warning determination module determines the early warning representation value based on the ratio of the accuracy of the early warning information to the false negative rate of the deep learning model, in order to determine whether the validity of the generated early warning information meets the requirements.

[0007] Furthermore, the early warning determination module determines that the validity of the generated early warning information meets the requirements when the early warning characterization value is greater than or equal to the preset early warning characterization value. The early warning determination module determines that the validity of the generated early warning information does not meet the requirements when the early warning characterization value is less than the preset early warning characterization value.

[0008] Furthermore, in response to the condition that the validity of the generated early warning information does not meet the requirements, the correlation adjustment module determines whether the accuracy of the screening of construction features meets the requirements based on the noise ratio of the construction features.

[0009] Furthermore, the correlation adjustment module responds to the fact that the noise ratio of the construction feature is less than or equal to a preset first ratio, and determines that the screening accuracy of the construction feature meets the requirements; The correlation adjustment module responds to the fact that the noise ratio of the construction feature is greater than the preset first ratio, and determines that the screening accuracy of the construction feature does not meet the requirements.

[0010] Furthermore, the correlation adjustment module increases the temporal correlation coupling threshold of the construction feature in response to the noise ratio of the construction feature being greater than the preset first ratio and less than the preset second ratio. The correlation adjustment module responds to the fact that the noise ratio of the construction feature is greater than or equal to the preset second ratio, and initially determines that the collection validity of the multi-source construction data does not meet the requirements. It then determines whether the collection validity of the multi-source construction data meets the requirements based on the redundancy rate of the multi-source construction data.

[0011] Furthermore, the increase in the temporal correlation coupling threshold of the construction feature is determined by the difference between the noise ratio of the construction feature and the preset first ratio.

[0012] Furthermore, in response to the condition that the noise ratio of the construction feature is greater than or equal to the preset second ratio, the confidence adjustment module determines whether the collection validity of the multi-source construction data meets the requirements based on the redundancy rate of the multi-source construction data.

[0013] Furthermore, the confidence adjustment module determines that the collection validity of the multi-source construction data meets the requirements when the redundancy rate of the multi-source construction data is less than or equal to the preset redundancy rate. The confidence adjustment module determines that the collection validity of the multi-source construction data does not meet the requirements when the redundancy rate of the multi-source construction data is greater than the preset redundancy rate, and reduces the confidence threshold for reordering the multi-source construction data.

[0014] Furthermore, the reduction in the confidence threshold of the multi-source construction data reordering is determined by the difference between the redundancy rate of the multi-source construction data and the preset redundancy rate.

[0015] Compared with existing technologies, the beneficial effects of this invention are as follows: The system of this invention, by setting up a data processing module, an analysis and early warning module, a collaborative interaction module, an early warning judgment module, a correlation adjustment module, and a confidence adjustment module, determines the validity of the generated early warning information based on the early warning representation value determined by the accuracy of the early warning information and the false negative rate of the deep learning model. Because the system lacks an effective feature screening mechanism, it cannot remove redundant features from multi-source collaborative data or extract representative and discriminative key risk features. This results in the deep learning model relying solely on known risk features for training, making it difficult to automatically discover potential risk correlation patterns. Consequently, the deep learning model can only identify known risk types and cannot effectively perceive new types of risks. By determining the validity of the generated early warning information, the accuracy of construction feature screening can be verified and optimized in reverse, allowing for timely adjustment of the feature screening strategy and improving the model's ability to identify unknown risks. The temporal correlation coupling threshold of construction features is adjusted based on the noise ratio of the construction features. Due to preprocessing… The lack of a unified time base and insufficient time synchronization accuracy in multi-source construction data at different stages makes it difficult to establish a unified time base to align multi-source data streams. This results in misalignment and out-of-order data across time dimensions, making it impossible to accurately capture the inherent correlations between multi-source data at the same time during feature extraction. Consequently, an effective feature selection mechanism lacks a foundation for implementation. Increasing the temporal correlation coupling threshold for construction features can strengthen the screening intensity of temporally correlated features, eliminating invalid features with temporal misalignment or weak correlations. Adjusting the reordering confidence threshold for multi-source construction data based on its redundancy rate is also crucial. Due to the complex network environment at construction sites, fluctuating transmission paths, and different data sources using different time bases and transmission protocols, timestamp discrepancies and out-of-order arrival occur during the collection and transmission of multi-source data. Reducing the reordering confidence threshold for multi-source construction data can relax the reordering criteria for out-of-order data, improve the coverage of data reordering and temporal correction, reduce the redundancy rate of multi-source construction data, and enhance the management effectiveness of supervision services.

[0016] Furthermore, the system described in this invention determines the validity of the generated early warning information by setting preset early warning characterization values. Since the system lacks an effective feature screening mechanism, it cannot remove redundant features from multi-source collaborative data or extract representative and distinguishable key risk features. This results in the deep learning model relying solely on known risk features for training, making it difficult to automatically uncover potential risk correlation patterns. Consequently, the deep learning model can only identify known risk types and cannot effectively perceive new types of risks. By determining the validity of the generated early warning information, the system can reversely drive the verification and optimization of the accuracy of construction feature screening, adjust the feature screening strategy in a timely manner, improve the model's ability to identify unknown risks, and further enhance the management effectiveness of supervision business.

[0017] Furthermore, the system of the present invention adjusts the temporal correlation coupling threshold of construction features by setting a preset first proportion and a preset second proportion. Due to the inconsistent time base and insufficient time synchronization accuracy of multi-source construction data in the preprocessing stage, it is difficult to establish a unified time base to align multi-source data streams. The data is misaligned and disordered in the time dimension. When extracting features, it is impossible to accurately capture the inherent correlation between multi-source data at the same time, resulting in a lack of implementation basis for an effective feature screening mechanism. By increasing the temporal correlation coupling threshold of construction features, the screening intensity of temporal correlation features can be strengthened, and invalid features with temporal misalignment and weak correlation can be eliminated, thereby further improving the management effectiveness of supervision business.

[0018] Furthermore, the system described in this invention adjusts the confidence threshold for reordering multi-source construction data by setting a preset redundancy rate. Due to the complex network environment at construction sites, fluctuating transmission paths, and different data sources using different time bases and transmission protocols, timestamp deviations and out-of-order arrivals occur during the collection and transmission of multi-source data. By reducing the confidence threshold for reordering multi-source construction data, the reordering criteria for out-of-order data can be relaxed, the coverage of data reordering and time-series correction can be improved, the redundancy rate of multi-source construction data can be reduced, and the management effectiveness of supervision services can be further improved. Attached Figure Description

[0019] Figure 1 This is an overall structural block diagram of the construction site multi-source data collaborative analysis and early warning system for construction supervision business according to an embodiment of the present invention; Figure 2 This is a flowchart illustrating the process of determining the validity of early warning information generated by the multi-source data collaborative analysis and early warning system for construction sites, which is designed for supervision work, according to an embodiment of the present invention. Figure 3 This is a logical flowchart of the process for determining the temporal correlation coupling threshold of construction characteristics in the construction site multi-source data collaborative analysis and early warning system for construction supervision business according to an embodiment of the present invention; Figure 4 This is a flowchart illustrating the process of determining the confidence threshold for reordering multi-source construction data in a collaborative analysis and early warning system for construction site multi-source data for supervision purposes, as described in this embodiment of the invention. Detailed Implementation

[0020] To make the objectives and advantages of the present invention clearer, the present invention will be further described below with reference to embodiments; it should be understood that the specific embodiments described herein are merely for explaining the present invention and are not intended to limit the present invention.

[0021] Preferred embodiments of the present invention will now be described with reference to the accompanying drawings. Those skilled in the art should understand that these embodiments are merely illustrative of the technical principles of the present invention and are not intended to limit the scope of protection of the present invention.

[0022] Please see Figure 1 As shown, it is an overall structural block diagram of the construction site multi-source data collaborative analysis and early warning system for supervision business according to an embodiment of the present invention.

[0023] This invention provides a multi-source data collaborative analysis and early warning system for construction sites, oriented towards supervision operations, comprising: The data processing module includes an acquisition unit for collecting multi-source construction data from the construction site via IoT sensors, and a preprocessing unit connected to the acquisition unit for preprocessing the multi-source construction data to output construction features. An analysis and early warning module, which is connected to the data processing module, includes a model training unit for training an initial model based on the construction characteristics to obtain a deep learning model, an analysis unit connected to the model training unit for analyzing the multi-source construction data based on the deep learning model to obtain analysis results, and an early warning unit connected to the analysis unit for generating early warning information based on the analysis results. A collaborative interaction module, which is connected to the analysis and early warning module, is used to enable several users to share information and collaborate. The early warning determination module, which is connected to the analysis and early warning module, is used to determine whether the validity of the early warning information generation meets the requirements based on the early warning representation value determined by the accuracy of the early warning information and the false negative rate of the deep learning model. The correlation adjustment module, which is connected to the early warning determination module, is used to determine the temporal correlation coupling threshold of the construction feature based on the noise ratio of the construction feature when the validity of the generated early warning information does not meet the requirements. A confidence adjustment module, which is connected to the association adjustment module, is used to determine the reordering confidence threshold of multi-source construction data based on the redundancy rate of multi-source construction data.

[0024] Specifically, IoT sensors include tilt sensors, pressure sensors, and stress sensors.

[0025] Specifically, the multi-source construction data includes tower crane tilt angle data, foundation pit support pressure data, and foundation pit support structure stress data.

[0026] Specifically, preprocessing includes data cleaning, data alignment, data format standardization, and feature extraction.

[0027] Specifically, construction characteristics include the average tower crane tilt angle, the rate of change of foundation pit support displacement, and the stress fluctuation rate of foundation pit support structure.

[0028] Specifically, the process of training an initial model to obtain a deep learning model based on construction features involves dividing the construction features into a training set, a validation set, and a test set according to a preset ratio. The initial deep learning model is iteratively trained using the training set. The model performance is monitored in real time using the validation set, and hyperparameters are optimized and tuned to prevent overfitting and improve generalization ability. Finally, the trained model is evaluated using the test set. After passing the quality evaluation, a deployable deep learning model is generated.

[0029] Specifically, the initial model is an initial framework that has basic construction feature extraction capabilities and construction site risk perception capabilities, and is suitable for collaborative analysis of multi-source data at construction sites and intelligent early warning decision-making for supervision business. The initial framework can be selected by technical personnel in this field according to the actual application scenario to achieve collaborative analysis at the construction site.

[0030] Specifically, the deep learning model can be a long short-term memory network model, a temporal convolutional network model, or a graph neural network model, with the preferred embodiment being a long short-term memory network model.

[0031] Specifically, the process of analyzing multi-source construction data based on a deep learning model to obtain analysis results involves preprocessing the collected multi-source construction data to output construction features, inputting the construction features into a trained deep learning model, identifying and quantifying the data through the model's calculations, and finally outputting the analysis results.

[0032] Specifically, the process of generating early warning information based on the analysis results involves parsing the risk level, hazard type, location, and quantitative basis in the analysis results, matching the corresponding preset early warning template, automatically filling the early warning template with the parsed content, and finally obtaining a structured early warning message containing risk level, location description, and handling suggestions.

[0033] Specifically, the analysis results include the results of the tower crane tilt risk level assessment, the results of the foundation pit support displacement and settlement trend analysis, and the results of the foundation pit support structure stress safety assessment.

[0034] Specifically, the early warning information includes risk level information, location information of potential hazards, and disposal recommendations.

[0035] Specifically, the accuracy rate of early warning information is the ratio of the number of accurate early warnings to the total number of early warnings.

[0036] Specifically, accuracy means that the early warning information is consistent with the actual working conditions and the true state of potential hazards at the construction site.

[0037] Specifically, the false negative rate of a deep learning model is the ratio of the number of times a risk actually exists but is not identified by the deep learning model and triggers an alert to the total number of times the risk occurs.

[0038] Specifically, the noise proportion of construction features is the ratio of the number of noise features among construction features to the total number of construction features.

[0039] Specifically, noise refers to invalid, erroneous, or interference characteristics that fail to reflect the true operating conditions.

[0040] Specifically, the temporal correlation coupling threshold of construction features is the minimum critical value used to determine whether construction features from different sources are synchronized in the time dimension and whether they are closely related.

[0041] Specifically, the redundancy rate of multi-source construction data is the ratio of the amount of redundant data in multi-source construction data to the total amount of multi-source construction data.

[0042] Specifically, redundancy refers to duplicate, invalid, or data with no practical business value.

[0043] Specifically, the confidence threshold for reordering multi-source construction data is the minimum critical value for determining whether the time-series reordered multi-source construction data is usable data.

[0044] In implementation, the system of this invention sets up a data processing module, an analysis and early warning module, a collaborative interaction module, an early warning judgment module, a correlation adjustment module, and a confidence adjustment module. It judges the validity of the generated early warning information based on the early warning representation value determined by the accuracy of the early warning information and the false negative rate of the deep learning model. Because the system lacks an effective feature filtering mechanism, it cannot remove redundant features from multi-source collaborative data or extract representative and discriminative key risk features. This results in the deep learning model relying solely on known risk features for training, making it difficult to automatically discover potential risk correlation patterns. Consequently, the deep learning model can only identify known risk types and cannot effectively perceive new types of risks. By judging the validity of the generated early warning information, the accuracy of construction feature filtering can be verified and optimized in reverse, allowing for timely adjustment of feature filtering strategies and improving the model's ability to identify unknown risks. The temporal correlation coupling threshold of construction features is adjusted based on the noise ratio of the construction features. Since the preprocessing stage involves multi-source construction data... Due to inconsistent time bases and insufficient time synchronization accuracy, it is difficult to establish a unified time base to align multi-source data streams. Data exhibits misalignment and out-of-order behavior in the time dimension, making it impossible to accurately capture the inherent correlations between multi-source data at the same time during feature extraction. This results in a lack of a foundation for implementing effective feature selection mechanisms. By increasing the temporal correlation coupling threshold of construction features, the screening intensity of temporally correlated features can be strengthened, eliminating invalid features with temporal misalignment and weak correlation. The confidence threshold for reordering multi-source construction data can be adjusted based on the redundancy rate of the multi-source construction data. Because of the complex network environment at construction sites, fluctuating transmission paths, and different data sources using different time bases and transmission protocols, timestamp deviations and out-of-order arrivals occur during the collection and transmission of multi-source data. By reducing the confidence threshold for reordering multi-source construction data, the reordering admission conditions for out-of-order data can be relaxed, improving the coverage of data reordering and temporal correction, reducing the redundancy rate of multi-source construction data, and improving the management effectiveness of supervision services.

[0045] Please continue reading. Figure 2 The diagram shown is a logical flowchart illustrating the process of determining the validity of early warning information generation in the construction site multi-source data collaborative analysis and early warning system for construction supervision business according to an embodiment of the present invention.

[0046] Specifically, the early warning determination module determines the early warning representation value based on the ratio of the accuracy of the early warning information to the false negative rate of the deep learning model, in order to determine whether the validity of the generated early warning information meets the requirements.

[0047] Specifically, the early warning determination module determines that the validity of the generated early warning information meets the requirements when the early warning characterization value is greater than or equal to the preset early warning characterization value. The early warning determination module determines that the validity of the generated early warning information does not meet the requirements when the early warning characterization value is less than the preset early warning characterization value.

[0048] Understandably, in a construction site multi-source data collaborative analysis and early warning system for supervision business, the core logic of using preset early warning characterization values ​​to characterize the effectiveness of early warning information generation is to quantify the effectiveness of early warning information generation into a numerically calculable indicator. By comparing the early warning characterization value obtained by fusing the accuracy rate of early warning information and the false negative rate of deep learning models with the preset early warning characterization value, a scientific judgment can be made on the quality of multi-source data preprocessing at the construction site, the effectiveness of construction feature extraction, and the accuracy of early warning from deep learning models. When the warning indicator value is greater than or equal to the preset warning indicator value, it indicates that the multi-source data preprocessing at the construction site is sufficient, the correlation of construction features is strong, the core risk features are prominent, and the warning effect of the deep learning model is good. The effectiveness of the generated warning information meets the actual requirements of risk prevention and control and hidden danger warning in the supervision business, and can provide accurate and reliable warning support for supervisors, helping to carry out the supervision work efficiently. When the warning indicator value is less than the preset warning indicator value, it indicates that invalid data and noise data in the multi-source data at the construction site are seriously interfered with, the effectiveness of construction features is low, and the risk of missed reports by the deep learning model is high. It cannot achieve timely and accurate warning of risks at the construction site, and the effectiveness of the generated warning information does not meet the actual needs of the supervision business. It is necessary to adaptively adjust the data preprocessing parameters, construction feature screening strategies, and relevant thresholds of the deep learning model. The preset warning indicator value can be set according to the actual working conditions. The preset warning indicator value aims to ensure the management effectiveness and practicality of the supervision business. Optionally, the preset early warning characterization value is determined through a limited number of experiments by evaluating the management effect of different early warning characterization values ​​on the supervision business. The determined preset early warning characterization value should meet the requirement that it is neither too small nor will it cause excessive interference to the management process of the supervision business. For example, the preset early warning characterization value is generally selected in the range of [8, 20].

[0049] Preferably, the preferred embodiment of the preset early warning characterization value is 15.

[0050] In practice, the system described in this invention determines the validity of generated early warning information by setting preset early warning characterization values. However, since the system lacks an effective feature filtering mechanism, it cannot remove redundant features from multi-source collaborative data or extract representative and distinguishable key risk features. Consequently, the deep learning model relies solely on known risk features for training, making it difficult to automatically uncover potential risk correlation patterns. This results in the deep learning model only being able to identify known risk types and failing to effectively perceive new types of risks. By determining the validity of generated early warning information, the accuracy of construction feature filtering can be verified and optimized in reverse, allowing for timely adjustments to feature filtering strategies and enhancing the model's ability to identify unknown risks. This further improves the management effectiveness of supervision services.

[0051] Please continue reading. Figure 3 As shown, it is a logical flowchart of the process of determining the temporal correlation coupling threshold of construction characteristics in the construction site multi-source data collaborative analysis and early warning system for construction supervision business according to an embodiment of the present invention.

[0052] Specifically, in response to the condition that the validity of the generated early warning information does not meet the requirements, the correlation adjustment module determines whether the accuracy of the screening of construction features meets the requirements based on the noise ratio of the construction features.

[0053] Specifically, the correlation adjustment module responds to the fact that the noise ratio of the construction feature is less than or equal to a preset first ratio, and determines that the screening accuracy of the construction feature meets the requirements. The correlation adjustment module responds to the fact that the noise ratio of the construction feature is greater than the preset first ratio, and determines that the screening accuracy of the construction feature does not meet the requirements.

[0054] Specifically, the correlation adjustment module increases the temporal correlation coupling threshold of the construction feature in response to the noise ratio of the construction feature being greater than the preset first ratio and less than the preset second ratio. The correlation adjustment module responds to the fact that the noise ratio of the construction feature is greater than or equal to the preset second ratio, and initially determines that the collection validity of the multi-source construction data does not meet the requirements. It then determines whether the collection validity of the multi-source construction data meets the requirements based on the redundancy rate of the multi-source construction data.

[0055] It is understandable that the first preset percentage is less than the second preset percentage, and the three intervals divided by the first and second preset percentages correspond to three different scenarios: The first interval is where the proportion of noise with construction characteristics is less than or equal to the preset first proportion. The corresponding situation is that the accuracy of the screening of construction characteristics meets the requirements. The second interval is where the noise proportion of construction features is greater than the preset first proportion and less than the preset second proportion. The corresponding situation is: due to the inconsistent time base and insufficient time synchronization accuracy of multi-source construction data in the preprocessing stage, it is difficult to establish a unified time base to align multi-source data streams. The data is misaligned and out of order in the time dimension. When extracting features, it is impossible to accurately capture the inherent correlation between multi-source data at the same time, resulting in a lack of implementation basis for an effective feature screening mechanism. The third interval is where the noise proportion of construction characteristics is greater than or equal to the preset second proportion. The corresponding situation is: due to the complex network environment at the construction site, the transmission path has fluctuating delays, and different data sources use different time bases and transmission protocols, resulting in timestamp deviations and out-of-order arrival of multi-source data during the collection and transmission process.

[0056] Understandably, using preset first and second percentages to characterize the noise proportion level of construction features is based on the core logic of transforming the accuracy of construction feature selection into a quantifiable range of noise proportion values. The preset first percentage serves as the boundary between parameter adjustments required in the feature selection stage and anomalies in the data acquisition and preprocessing stage. The preset second percentage serves as the critical point for determining the failure of the feature selection stage and the failure of the data acquisition and preprocessing stage. This provides a quantitative basis for targeted identification of issues such as timing synchronization deviations, excessive noise features, disordered data acquisition, and timestamp misalignments, enabling layered problem tracing and precise control from data acquisition and preprocessing to feature selection and early warning information generation. The preset first and second percentages can be set according to actual working conditions. The preset first and second percentages aim to ensure the effectiveness and practicality of supervision management. Optionally, the preset first and second percentages are determined through a limited number of experiments by evaluating the management effect of different noise proportions on supervision management. The determined preset first and second percentages should be neither too small nor cause excessive interference to the management process of supervision management. For example, the preset first percentage is generally selected in the range of [10%, 20%], and the preset second percentage is generally selected in the range of [21%, 30%].

[0057] Preferably, the first percentage is 15% in the preferred embodiment, and the second percentage is 25% in the preferred embodiment.

[0058] Specifically, the increase in the temporal correlation coupling threshold of the construction feature is determined by the difference between the noise ratio of the construction feature and the preset first ratio.

[0059] Specifically, when the difference between the noise proportion of a construction feature and the preset first proportion is within 10%, the temporal correlation coupling threshold of the construction feature is increased to 1.1 times its original value. When the difference between the noise proportion of a construction feature and the preset first proportion exceeds 10%, in addition to increasing to 1.1 times its original value, the temporal correlation coupling threshold of the construction feature increases by 0.01 for every 1% exceeding the original value. For example, if the difference between the noise proportion of a construction feature and the preset first proportion is 12%, and the current temporal correlation coupling threshold of the construction feature is 0.7, then the increased temporal correlation coupling threshold of the construction feature is 0.7 × 1.1 + 0.01 × 2 = 0.79.

[0060] In practice, the system of the present invention adjusts the temporal correlation coupling threshold of construction features by setting a preset first proportion and a preset second proportion. Due to the inconsistent time base and insufficient time synchronization accuracy of multi-source construction data in the preprocessing stage, it is difficult to establish a unified time base to align multi-source data streams. The data is misaligned and out of order in the time dimension. When extracting features, it is impossible to accurately capture the inherent correlation between multi-source data at the same time, resulting in a lack of implementation basis for an effective feature screening mechanism. By increasing the temporal correlation coupling threshold of construction features, the screening intensity of temporal correlation features can be strengthened, and invalid features with temporal misalignment and weak correlation can be eliminated, further improving the management effectiveness of supervision business.

[0061] Please continue reading. Figure 4 As shown, it is a logical flowchart of the process of determining the reordering confidence threshold of multi-source construction data in the construction site multi-source data collaborative analysis and early warning system for construction site supervision business according to an embodiment of the present invention.

[0062] Specifically, the confidence adjustment module, in response to the condition that the noise ratio of the construction feature is greater than or equal to the preset second ratio, determines whether the collection validity of the multi-source construction data meets the requirements based on the redundancy rate of the multi-source construction data.

[0063] Specifically, the confidence adjustment module determines that the collection validity of the multi-source construction data meets the requirements when the redundancy rate of the multi-source construction data is less than or equal to the preset redundancy rate. The confidence adjustment module determines that the collection validity of the multi-source construction data does not meet the requirements when the redundancy rate of the multi-source construction data is greater than the preset redundancy rate, and reduces the confidence threshold for reordering the multi-source construction data.

[0064] It is understandable that the two intervals of the preset redundancy rate division correspond to two different scenarios: The first interval is when the redundancy rate of multi-source construction data is less than or equal to the preset redundancy rate. The corresponding situation is: the validity of the collection of multi-source construction data is determined to meet the requirements. The second interval is when the redundancy rate of multi-source construction data is greater than the preset redundancy rate. The corresponding situation is that due to the complex network environment at the construction site, the transmission path has fluctuating delays, and different data sources use different time bases and transmission protocols, which causes timestamp deviations and out-of-order arrival of multi-source data during the collection and transmission process.

[0065] Understandably, using a preset redundancy rate to characterize the effectiveness of multi-source construction data collection is based on the core logic of converting data collection quality into a quantifiable and comparable redundancy rate value. By comparing the redundancy rate of multi-source construction data with the preset redundancy rate, the extent to which duplicate, out-of-order, and redundant data are generated in the current data collection stage, as well as the level of control over the purity of the collected data, is determined. This provides core data support for subsequent adjustments to time-series reordering parameters, optimization of the accuracy of construction feature selection, and improvement of the effectiveness of early warning information generation. The redundancy rate of multi-source construction data is a key evaluation indicator in the data collection and transmission stages. An excessively high redundancy rate will lead to a large amount of duplicate, out-of-order, and invalid data entering the time-series alignment and feature extraction stages, causing effective construction features to be overwhelmed by noise and model computing power to be ineffectively used. This can result in problems such as insufficient accuracy of feature selection, false alarms and missed alarms, and delayed early warning generation. Especially given the diverse collection equipment, complex transmission environment, and wide range of data sources at construction sites, the effectiveness of collection directly determines the efficiency and accuracy of subsequent time-series alignment, feature fusion, and early warning analysis. The preset redundancy rate can be set according to actual working conditions. The setting of the preset redundancy rate aims to ensure the management effectiveness and practicality of supervision business. Optionally, the preset redundancy rate is determined through a limited number of tests by evaluating the management effect of redundancy rates of different multi-source construction data on the supervision business. The determined preset redundancy rate should be neither too small nor cause excessive interference to the management process of the supervision business. For example, the preset redundancy rate is generally selected in the range of [10%, 20%].

[0066] Preferably, the preset redundancy rate is 15% in the preferred embodiment.

[0067] Specifically, the reduction in the confidence threshold of the reordering of the multi-source construction data is determined by the difference between the redundancy rate of the multi-source construction data and the preset redundancy rate.

[0068] Specifically, when the difference between the redundancy rate of multi-source construction data and the preset redundancy rate is within 5%, the confidence threshold for reordering multi-source construction data is reduced to 0.95 times the original value. When the difference between the redundancy rate of multi-source construction data and the preset redundancy rate exceeds 5%, the confidence threshold for reordering multi-source construction data is reduced by 0.01 for every 1% exceeding the original value, in addition to being reduced to 0.95 times the original value. For example, if the difference between the redundancy rate of multi-source construction data and the preset redundancy rate is 7%, and the current confidence threshold for reordering multi-source construction data is 0.8, the reduced confidence threshold for reordering multi-source construction data is 0.8 × 0.95 - 0.01 × 2 = 0.74.

[0069] In implementation, the system of the present invention adjusts the confidence threshold for reordering multi-source construction data by setting a preset redundancy rate. Due to the complex network environment at the construction site, the transmission path has fluctuating delays, and different data sources use different time bases and transmission protocols, resulting in timestamp deviations and out-of-order arrival of multi-source data during collection and transmission. By reducing the confidence threshold for reordering multi-source construction data, the reordering admission conditions for out-of-order data can be relaxed, the coverage of data reordering and time sequence correction can be improved, the redundancy rate of multi-source construction data can be reduced, and the management effectiveness of supervision business can be further improved.

[0070] The technical solution of the present invention has been described above with reference to the preferred embodiments shown in the accompanying drawings. However, it will be readily understood by those skilled in the art that the scope of protection of the present invention is obviously not limited to these specific embodiments. Without departing from the principles of the present invention, those skilled in the art can make equivalent changes or substitutions to the relevant technical features, and the technical solutions after these changes or substitutions will all fall within the scope of protection of the present invention.

Claims

1. A multi-source data collaborative analysis and early warning system for construction sites, oriented towards supervision operations, characterized in that: include: The data processing module includes an acquisition unit for collecting multi-source construction data from the construction site via IoT sensors, and a preprocessing unit connected to the acquisition unit for preprocessing the multi-source construction data to output construction features. An analysis and early warning module, which is connected to the data processing module, includes a model training unit for training an initial model based on the construction characteristics to obtain a deep learning model, an analysis unit connected to the model training unit for analyzing the multi-source construction data based on the deep learning model to obtain analysis results, and an early warning unit connected to the analysis unit for generating early warning information based on the analysis results. A collaborative interaction module, which is connected to the analysis and early warning module, is used to enable several users to share information and collaborate. The early warning determination module, which is connected to the analysis and early warning module, is used to determine whether the validity of the early warning information generation meets the requirements based on the early warning representation value determined by the accuracy of the early warning information and the false negative rate of the deep learning model. The correlation adjustment module, which is connected to the early warning determination module, is used to determine the temporal correlation coupling threshold of the construction feature based on the noise ratio of the construction feature when the validity of the generated early warning information does not meet the requirements. A confidence adjustment module, which is connected to the association adjustment module, is used to determine the reordering confidence threshold of multi-source construction data based on the redundancy rate of multi-source construction data.

2. The construction site multi-source data collaborative analysis and early warning system for construction supervision business as described in claim 1, characterized in that, The early warning determination module responds to the early warning characterization value determined by the ratio of the accuracy of the early warning information to the false negative rate of the deep learning model, in order to determine whether the validity of the generated early warning information meets the requirements.

3. The construction site multi-source data collaborative analysis and early warning system for construction supervision business as described in claim 2, characterized in that, The early warning determination module determines that the validity of the generated early warning information meets the requirements when the early warning characterization value is greater than or equal to the preset early warning characterization value. The early warning determination module determines that the validity of the generated early warning information does not meet the requirements when the early warning characterization value is less than the preset early warning characterization value.

4. The construction site multi-source data collaborative analysis and early warning system for construction supervision business as described in claim 3, characterized in that, In response to the condition that the validity of the generated early warning information does not meet the requirements, the correlation adjustment module determines whether the accuracy of the screening of construction features meets the requirements based on the noise ratio of the construction features.

5. The construction site multi-source data collaborative analysis and early warning system for construction supervision business as described in claim 4, characterized in that, The correlation adjustment module determines that the accuracy of the construction feature screening meets the requirements when the noise ratio of the construction feature is less than or equal to a preset first ratio. The correlation adjustment module responds to the fact that the noise ratio of the construction feature is greater than the preset first ratio, and determines that the screening accuracy of the construction feature does not meet the requirements.

6. The construction site multi-source data collaborative analysis and early warning system for construction supervision business as described in claim 5, characterized in that, The correlation adjustment module increases the temporal correlation coupling threshold of the construction feature in response to the noise ratio of the construction feature being greater than the preset first ratio and less than the preset second ratio. The correlation adjustment module responds to the fact that the noise ratio of the construction feature is greater than or equal to the preset second ratio, and initially determines that the collection validity of the multi-source construction data does not meet the requirements. It then determines whether the collection validity of the multi-source construction data meets the requirements based on the redundancy rate of the multi-source construction data.

7. The construction site multi-source data collaborative analysis and early warning system for construction supervision business as described in claim 6, characterized in that, The increase in the temporal correlation coupling threshold of the construction feature is determined by the difference between the noise ratio of the construction feature and the preset first ratio.

8. The construction site multi-source data collaborative analysis and early warning system for construction supervision business as described in claim 7, characterized in that, The confidence adjustment module, in response to the condition that the noise ratio of the construction feature is greater than or equal to the preset second ratio, determines whether the collection validity of the multi-source construction data meets the requirements based on the redundancy rate of the multi-source construction data.

9. The construction site multi-source data collaborative analysis and early warning system for construction supervision business as described in claim 8, characterized in that, The confidence adjustment module determines that the collection validity of the multi-source construction data meets the requirements when the redundancy rate of the multi-source construction data is less than or equal to the preset redundancy rate. The confidence adjustment module determines that the collection validity of the multi-source construction data does not meet the requirements when the redundancy rate of the multi-source construction data is greater than the preset redundancy rate, and reduces the confidence threshold for reordering the multi-source construction data.

10. The construction site multi-source data collaborative analysis and early warning system for construction supervision business as described in claim 9, characterized in that, The reduction in the confidence threshold for reordering the multi-source construction data is determined by the difference between the redundancy rate of the multi-source construction data and the preset redundancy rate.

Images