Building construction real-time monitoring and early warning method based on big data

Abstract

The invention discloses a building construction real-time monitoring and early warning method based on big data, and relates to the technical field of construction management, and the method comprises the following steps: S1, determining a monitoring target and sensing parameters; s2, establishing a multi-dimensional data model; s3, determining a risk threshold and an early warning mechanism; s4, performing data processing on the real-time data stream; s5, constructing an intelligent anomaly detection model; s6, implementing dynamic adjustment and feedback circulation; and S7, integrating an external data source. According to the building construction real-time monitoring and early warning method based on the big data, by introducing a dynamic adjustment and feedback mechanism based on the big data and combining a real-time data flow and a machine learning algorithm, the building construction real-time monitoring and early warning method can significantly improve the response speed and accuracy of early warning; according to the method, the risk threshold and the monitoring index can be updated in real time, and the actual situation of the site can be reflected in time, so that the risk of accidents is reduced, and the safety and efficiency of construction management are improved.

Description

Technical Field

[0001] The present invention relates to the technical field of construction management, and in particular to a real-time monitoring and early warning method for building construction based on big data. Background Art

[0002] With the rapid development of the construction industry, construction safety and efficiency issues have become increasingly prominent. Traditional construction monitoring methods often rely on static monitoring technology and manual inspections, resulting in delays in data acquisition and analysis, and failure to identify potential risks in a timely manner. At the same time, the construction site environment is complex and changeable, involving multiple dynamic factors such as weather, material quality, and personnel operations.

[0003] The existing technology has deficiencies in the following aspects:

[0004] Static threshold limits: Many existing early warning systems use fixed risk thresholds. This approach cannot adapt to changes in different construction phases and environmental conditions and is prone to false alarms or missed alarms.

[0005] Data processing lags: Traditional data processing technologies often take time to clean and analyze data, and cannot achieve real-time response. The rapid changes at the construction site make it impossible for these delayed data processing results to provide effective guidance for the construction team.

[0006] Lack of feedback mechanism: Most existing systems lack effective communication and feedback mechanisms with construction personnel and are unable to obtain the needs and challenges faced by on-site staff in a timely manner, resulting in a disconnect between the monitoring system and actual needs.

[0007] Insufficient algorithm adaptability: Existing monitoring algorithms are mostly static models that lack the ability to self-learn and adapt, and cannot effectively handle changing construction environments and emergencies. Summary of the invention

[0008] In view of the deficiencies in the prior art, the present invention provides a real-time monitoring and early warning method for construction based on big data to solve the problems raised in the above background technology.

[0009] To achieve the above object, the present invention provides the following technical solution: a real-time monitoring and early warning method for construction based on big data, comprising the following steps:

[0010] S1. Clarify monitoring targets and sensing parameters;

[0011] Determine the monitoring targets during construction, such as the structural safety of the building, construction progress, and environmental impact. At the same time, determine the sensing parameters, such as vibration frequency, temperature, humidity, stress, and displacement. The selected monitoring targets and sensing parameters are used to reflect the construction status and potential risks.

[0012] S2. Establish a multidimensional data model;

[0013] After clarifying the monitoring objectives, a multidimensional data model is established, covering the spatial, temporal, physical and environmental dimensions, and a data framework is formed by integrating different types of data sources, such as sensor data, historical data, and external environmental data;

[0014] S3. Establish risk thresholds and early warning mechanisms;

[0015] Set the risk threshold for each sensor parameter and develop an early warning mechanism. When the monitoring data exceeds the set threshold, the early warning is automatically triggered. Combined with the dynamic adjustment algorithm, the threshold is continuously optimized according to the changes in field conditions to accurately reflect the actual risk level;

[0016] S4, real-time data stream for data processing;

[0017] Based on real-time data stream processing technology, real-time data is quickly processed through the stream processing framework;

[0018] S5. Build an intelligent anomaly detection model;

[0019] Based on historical data and normal operation data, build intelligent anomaly detection models to detect abnormal situations and identify potential abnormal patterns through unsupervised learning algorithms;

[0020] S6. Implement dynamic adjustments and feedback loops;

[0021] During the monitoring process, a dynamic adjustment and feedback mechanism is established to regularly evaluate and update risk thresholds and monitoring indicators based on real-time data and early warning results;

[0022] S7, integration of external data sources;

[0023] Introduce external data sources, such as weather forecasts, traffic conditions, and changes in the surrounding environment, and correlate them with the data at the construction site for analysis.

[0024] To further optimize the technical solution, in step S1, based on the determined sensing parameters, a sensor network is arranged at the construction site, including vibration frequency, temperature, humidity, stress, and displacement sensors. The sensors are connected through a wireless network to collect various environmental and structural data during the construction process in real time.

[0025] To further optimize the technical solution, in step S2, the dimensions in the multidimensional data model include:

[0026] Spatial dimension: represents the location of each monitoring point in the construction site;

[0027] Time dimension: The timestamp of the monitoring data records the specific time when the data was collected;

[0028] Physical dimension: physical parameters related to structural safety, such as stress and temperature;

[0029] Environmental dimension: Environmental factors that affect construction, such as wind speed, humidity, and rainfall.

[0030] Further optimizing the technical solution, in step S3, the dynamic adjustment algorithm automatically optimizes the risk threshold according to the changes in the real-time monitoring data, and adjusts the threshold through continuous iteration based on historical data and real-time data;

[0031] And build the following dynamic adjustment threshold model:

[0032] T new =T old +β·(DT old );

[0033] in:

[0034] T new is the adjusted risk threshold;

[0035] T old is the current risk threshold;

[0036] D is the average value of real-time monitoring data;

[0037] β is the adjustment coefficient, which ranges from [0,1] and determines the magnitude of the adjustment;

[0038] When this model is used,

[0039] Data collection and calculation: collect real-time monitoring data (such as stress, vibration, etc.) and calculate the average value D of these data, and update D every time new data arrives;

[0040] Threshold adjustment: In each monitoring cycle, a new risk threshold T is calculated based on real-time data. new , if the newly calculated D is equal to the current threshold T old If there are significant differences, adjustments will be made through the model;

[0041] Dynamic optimization: The adjustment coefficient β is set according to the volatility of historical data. If the historical data changes greatly, a smaller β value such as 0.1 is set to make the threshold adjustment smoother. On the contrary, if the data is relatively stable, β can be increased to 0.5 to speed up the threshold response speed.

[0042] Early warning trigger mechanism: When the real-time monitoring data exceeds the dynamically adjusted threshold T new When a warning occurs, it will automatically trigger the early warning mechanism and notify the construction team;

[0043] Feedback and iteration: After each warning is triggered, collect on-site feedback, evaluate the accuracy and timeliness of the warning, use the feedback information to further adjust the β value and threshold setting strategy, and regularly review and optimize the model to ensure that it matches the changes in the construction environment.

[0044] To further optimize the technical solution, the adjustment coefficient β also includes the following adjustment methods when dynamically adjusting:

[0045] Rule of thumb: Based on the specific conditions of the construction site and industry experience, professionals or the safety management team provide recommended values, which are adjusted based on the actual operations and historical events on site;

[0046] Adaptive adjustment: After setting the initial value, make adjustments through real-time feedback to monitor the accuracy and frequency of early warnings. If there are too many false alarms, gradually reduce the value of β; if the frequency of missed alarms is high, increase the value of β;

[0047] Adjustment based on optimization algorithms: Use machine learning methods such as genetic algorithms and particle swarm optimization to find the best β value through simulation experiments, train on historical data, and determine the best adjustment coefficient by optimizing objective functions such as reducing false positive and false negative rates;

[0048] Threshold backtesting: Backtest past monitoring data to simulate the impact of different beta values, analyze the warning effect under each setting, and select the adjustment coefficient with the best performance.

[0049] Further optimizing the technical solution, in step S4, the stream processing framework includes a distributed platform and a stream processing engine, the distributed platform is used to process real-time data streams and supports high throughput, scalability and persistence, and the stream processing engine is used to support stateful computing and process real-time data streams and batch data;

[0050] When data processing is performed based on real-time data streams, the following specific processes are included:

[0051] Data collection: Real-time data from the construction site, such as temperature, vibration, and humidity, is collected through a sensor network. The collected data is sent to a cluster on a distributed platform for storage;

[0052] Data stream forwarding: Use the topics of the distributed platform to classify different types of real-time data and forward them to the corresponding stream processing engine;

[0053] Real-time computing: Configure data stream processing jobs in the stream processing engine, use the stream computing engine to process data in real time, and analyze and calculate data by writing processing logic, such as window computing and state management;

[0054] Anomaly detection: Anomaly detection is performed based on sensing parameters. Assuming that we focus on vibration and temperature parameters, the anomaly detection formula is as follows:

[0055] A(t)=λ·(f(t)-μ f )+ν·(θ(t)-μ θ );

[0056] Among them, A(t) represents the abnormal value at time t, λ and v are weight coefficients, reflecting the influence of vibration and temperature on the abnormal value, f(t) and θ(t) are the vibration frequency and temperature monitored in real time, μ f and μ θ It is the average value of historical data and is used to determine the degree of abnormality of the current value;

[0057] Real-time warning mechanism: When the calculated abnormal value A(t) exceeds the set threshold, a real-time warning is triggered and an alarm message is sent to the construction team.

[0058] To further optimize the technical solution, in step S5, the intelligent anomaly detection model is as follows:

[0059]

[0060] Among them, A represents the degree of abnormality detected;

[0061] n is the sample size of the current monitoring data;

[0062] X i is the data point of the i-th sample;

[0063] μ is the mean of the sample, that is, the average value of normal operating data;

[0064] σ is the standard deviation of the sample, which is used to measure the dispersion of the data.

[0065] To further optimize the technical solution, the intelligent anomaly detection model includes the following specific processes when used:

[0066] Data preparation: Collect historical data and normal operation data during the construction process to ensure the quality and representativeness of the data. The data includes various monitoring parameters such as stress, vibration, and temperature.

[0067] Feature extraction: Extract features from monitoring data and calculate the mean μ and standard deviation σ. The features will be used for subsequent anomaly detection calculations.

[0068] Calculate the degree of anomaly: When new data arrives each time, calculate the degree of anomaly of the current data set. If the value of intelligent anomaly detection A exceeds the set threshold, it is considered an anomaly.

[0069] Model iteration and self-learning: As time goes by, the model is continuously updated, and the mean μ and standard deviation σ are recalculated whenever new normal operation data is added;

[0070] Abnormal pattern recognition: Combined with the isolation forest algorithm, the calculated abnormal degree is further analyzed to identify potential abnormal patterns. By clustering the samples, similar abnormal events are grouped;

[0071] Early warning trigger mechanism: When the model detects an anomaly and confirms its pattern, an early warning is automatically triggered to send an alert to the construction team.

[0072] To further optimize the technical solution, in step S6, the dynamic adjustment and feedback mechanism is based on an update model for updating the risk threshold T and the monitoring index M. The update model is as follows:

[0073] T new =T old +α·(DT old )+γ·F

[0074] M new =M old +β·(RM old )+δ·F;

[0075] in,

[0076] T new : Updated risk threshold;

[0077] T old : Current risk threshold;

[0078] D: average value of real-time monitoring data or other key indicators;

[0079] α: The adjustment coefficient of the risk threshold, which controls the amplitude of threshold update;

[0080] γ: Feedback influence coefficient, reflecting the influence of feedback from construction personnel;

[0081] M new : Updated monitoring indicators;

[0082] M old : Current monitoring indicators;

[0083] R: newly calculated monitoring indicator, adjusted based on field feedback;

[0084] β: adjustment coefficient of monitoring indicators;

[0085] δ: Feedback influence coefficient, reflecting the impact of construction personnel feedback on the adjustment of monitoring indicators;

[0086] F: Feedback value, reflecting the specific feedback of on-site construction personnel on thresholds and monitoring indicators, such as safety and practicality scores.

[0087] To further optimize this technical solution, the method also designs a real-time data visualization platform to present the monitored data in a graphical form. Through a variety of display methods such as dashboards, heat maps and trend charts, the construction team can quickly understand the data change trends and potential risks.

[0088] Compared with the prior art, the present invention provides a real-time monitoring and early warning method for construction based on big data, which has the following beneficial effects:

[0089] This real-time monitoring and early warning method for construction based on big data can significantly improve the response speed and accuracy of early warning by introducing a dynamic adjustment and feedback mechanism based on big data, combining real-time data streams and machine learning algorithms. This method can update risk thresholds and monitoring indicators in real time, and promptly reflect the actual situation on site, thereby reducing the risk of accidents and improving the safety and efficiency of construction management. BRIEF DESCRIPTION OF THE DRAWINGS

[0090] Figure 1 A flowchart of a method for real-time monitoring and early warning of construction based on big data proposed by the present invention;

[0091] Figure 2 A schematic diagram of the process of an intelligent anomaly detection model in a real-time monitoring and early warning method for construction based on big data proposed by the present invention;

[0092] Figure 3 A schematic diagram of the process of abnormal pattern recognition in a real-time monitoring and early warning method for construction based on big data proposed by the present invention;

[0093] Figure 4 A diagram showing a calculation method for abnormality scores in a real-time monitoring and early warning method for construction based on big data proposed by the present invention;

[0094] Figure 5 This is a flow chart of a dynamic adjustment algorithm in a real-time monitoring and early warning method for construction based on big data proposed by the present invention. DETAILED DESCRIPTION

[0095] The following will be combined with the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0096] Embodiment 1:

[0097] See also Figure 1 , a real-time monitoring and early warning method for building construction based on big data, comprising the following steps:

[0098] S1. Clarify monitoring targets and sensing parameters

[0099] Determine the monitoring targets during construction, such as the structural safety of the building, construction progress, and environmental impact. At the same time, determine the sensing parameters, such as vibration frequency, temperature, humidity, stress, and displacement. The selected monitoring targets and sensing parameters are used to reflect the construction status and potential risks.

[0100] In this embodiment, in step S1, based on the determined sensing parameters, a sensor network is arranged at the construction site, including vibration frequency, temperature, humidity, stress, and displacement sensors. The sensors are connected through a wireless network to collect various environmental and structural data during the construction process in real time.

[0101] S2. Establish a multidimensional data model

[0102] After clarifying the monitoring objectives, a multidimensional data model is established, covering the dimensions of space, time, physics and environment, and a data framework is formed by integrating different types of data sources, such as sensor data, historical data, and external environmental data. The model helps to understand the dynamic changes of the construction site more comprehensively, provides a basis for subsequent data analysis, and improves the relevance and comparability between data.

[0103] In this embodiment, in step S2, the dimensions in the multidimensional data model include:

[0104] Spatial dimension: represents the location of each monitoring point in the construction site;

[0105] Time dimension: The timestamp of the monitoring data records the specific time when the data was collected;

[0106] Physical dimension: physical parameters related to structural safety, such as stress and temperature;

[0107] Environmental dimension: Environmental factors that affect construction, such as wind speed, humidity, and rainfall.

[0108] S3. Establish risk thresholds and early warning mechanisms

[0109] Set the risk threshold for each sensing parameter and develop an early warning mechanism. When the monitoring data exceeds the set threshold, the early warning is automatically triggered. Combined with the dynamic adjustment algorithm, the threshold is continuously optimized according to changes in on-site conditions to accurately reflect the actual risk level, thereby improving the effectiveness of the early warning.

[0110] In this embodiment, in step S3, the dynamic adjustment algorithm automatically optimizes the risk threshold according to the changes in the real-time monitoring data, and adjusts the threshold through continuous iteration based on historical data and real-time data;

[0111] And build the following dynamic adjustment threshold model:

[0112] T new =T old +β·(DT old );

[0113] in:

[0114] T new is the adjusted risk threshold;

[0115] T old is the current risk threshold;

[0116] D is the average value of real-time monitoring data;

[0117] β is the adjustment coefficient, which ranges from [0,1] and determines the magnitude of the adjustment;

[0118] When this model is used,

[0119] Data collection and calculation: collect real-time monitoring data (such as stress, vibration, etc.) and calculate the average value D of these data, and update D every time new data arrives;

[0120] Threshold adjustment: In each monitoring cycle, a new risk threshold T is calculated based on real-time data. new , if the newly calculated D is equal to the current threshold T old If there are significant differences, adjustments will be made through the model;

[0121] Dynamic optimization: The adjustment coefficient β is set according to the volatility of historical data. If the historical data changes greatly, a smaller β value such as 0.1 is set to make the threshold adjustment smoother. On the contrary, if the data is relatively stable, β can be increased to 0.5 to speed up the threshold response speed.

[0122] Early warning trigger mechanism: When the real-time monitoring data exceeds the dynamically adjusted threshold T new When a warning occurs, it will automatically trigger the early warning mechanism and notify the construction team;

[0123] Feedback and iteration: After each warning is triggered, collect on-site feedback, evaluate the accuracy and timeliness of the warning, use the feedback information to further adjust the β value and threshold setting strategy, and regularly review and optimize the model to ensure that it matches the changes in the construction environment.

[0124] Furthermore, when the adjustment coefficient β is adjusted dynamically, it also includes the following adjustment methods:

[0125] Rule of thumb: Based on the specific conditions of the construction site and industry experience, professionals or the safety management team provide recommended values, which are adjusted based on the actual operations and historical events on site;

[0126] Adaptive adjustment: After setting the initial value, make adjustments through real-time feedback to monitor the accuracy and frequency of early warnings. If there are too many false alarms, gradually reduce the value of β; if the frequency of missed alarms is high, increase the value of β;

[0127] Adjustment based on optimization algorithms: Use machine learning methods such as genetic algorithms and particle swarm optimization to find the best β value through simulation experiments, train on historical data, and determine the best adjustment coefficient by optimizing objective functions such as reducing false positive and false negative rates;

[0128] Threshold backtesting: Backtest past monitoring data to simulate the impact of different beta values, analyze the warning effect under each setting, and select the adjustment coefficient with the best performance.

[0129] S4, real-time data stream for data processing

[0130] Based on real-time data stream processing technology, real-time data is quickly processed through the stream processing framework. The massive data generated at the construction site can be processed and analyzed in a timely manner to achieve instant warning. Real-time stream processing can significantly reduce data latency, ensuring that the construction team can quickly obtain the latest information and respond in a timely manner.

[0131] In this embodiment, in step S4, the stream processing framework includes a distributed platform and a stream processing engine, wherein the distributed platform is used to process real-time data streams and supports high throughput, scalability and persistence, and the stream processing engine is used to support stateful computing and process real-time data streams and batch data;

[0132] When data processing is performed based on real-time data streams, the following specific processes are included:

[0133] Data collection: Real-time data from the construction site, such as temperature, vibration, and humidity, is collected through a sensor network. The collected data is sent to a cluster on a distributed platform for storage;

[0134] Data stream forwarding: Use the topics of the distributed platform to classify different types of real-time data and forward them to the corresponding stream processing engine;

[0135] Real-time computing: Configure data stream processing jobs in the stream processing engine, use the stream computing engine to process data in real time, and analyze and calculate data by writing processing logic, such as window computing and state management;

[0136] Anomaly detection: Anomaly detection is performed based on sensing parameters. Assuming that we focus on vibration and temperature parameters, the anomaly detection formula is as follows:

[0137] A(t)=λ·(f(t)-μ f )+ν·(θ(t)-μ θ );

[0138] Among them, A(t) represents the abnormal value at time t, λ and ν are weight coefficients, reflecting the influence of vibration and temperature on the abnormal value, f(t) and θ(t) are the vibration frequency and temperature monitored in real time, μ f and μ θ It is the average value of historical data and is used to determine the degree of abnormality of the current value;

[0139] Real-time warning mechanism: When the calculated abnormal value A(t) exceeds the set threshold, a real-time warning is triggered and an alarm message is sent to the construction team.

[0140] Through real-time data stream processing technology, real-time data at the construction site can be efficiently managed and analyzed, which can greatly reduce data latency and improve the ability to respond to emergencies.

[0141] S5. Building an intelligent anomaly detection model

[0142] Based on historical data and normal operation data, an intelligent anomaly detection model is built to detect abnormal situations, and potential abnormal patterns are identified through unsupervised learning algorithms. Through continuous iteration and self-learning, the model can gradually adapt to the actual situation on site, improve the accuracy and real-time performance of detection, and thus achieve early warning of emergencies.

[0143] In this embodiment, in step S5, the intelligent anomaly detection model is as follows:

[0144]

[0145] Among them, A represents the degree of abnormality detected;

[0146] n is the sample size of the current monitoring data;

[0147] X i is the data point of the i-th sample;

[0148] μ is the mean of the sample, that is, the average value of normal operating data;

[0149] σ is the standard deviation of the sample, which is used to measure the dispersion of the data.

[0150] like Figure 2 As shown,

[0151] Furthermore, when the intelligent anomaly detection model is used, it includes the following specific processes:

[0152] Data preparation: Collect historical data and normal operation data during the construction process to ensure the quality and representativeness of the data. The data includes various monitoring parameters such as stress, vibration, and temperature.

[0153] Feature extraction: Extract features from monitoring data and calculate the mean μ and standard deviation σ. The features will be used for subsequent anomaly detection calculations.

[0154] Calculate the degree of anomaly: When new data arrives each time, calculate the degree of anomaly of the current data set. If the value of intelligent anomaly detection A exceeds the set threshold, it is considered an anomaly.

[0155] Model iteration and self-learning: As time goes by, the model is continuously updated. Whenever new normal operation data is added, the mean μ and standard deviation σ are recalculated to gradually adapt the model to the actual situation on site;

[0156] Abnormal pattern recognition: Combined with the isolation forest algorithm, the calculated abnormal degree is further analyzed to identify potential abnormal patterns. By clustering samples, similar abnormal events are grouped to help the construction team identify recurring risk sources;

[0157] Early warning trigger mechanism: When the model detects an anomaly and confirms its pattern, an early warning is automatically triggered to send an alert to the construction team.

[0158] Through formulated calculation methods and advanced machine learning techniques, the accuracy of anomaly detection can be continuously improved, effectively reducing safety risks at construction sites.

[0159] S6. Implement dynamic adjustments and feedback loops

[0160] During the monitoring process, a dynamic adjustment and feedback mechanism is established to regularly evaluate and update risk thresholds and monitoring indicators based on real-time data and early warning results. The importance of feedback is emphasized, and through communication with on-site construction personnel, their actual needs and challenges are understood, so as to better optimize the monitoring method and keep it efficient and practical.

[0161] In this embodiment, in step S6, the dynamic adjustment and feedback mechanism is based on an update model for updating the risk threshold T and the monitoring index M. The update model is as follows:

[0162] T new =T old +α·(DT old )+γ·F

[0163] M new =M old+β·(RM old )+δ·F;

[0164] in,

[0165] T new : Updated risk threshold;

[0166] T old : Current risk threshold;

[0167] D: average value of real-time monitoring data or other key indicators;

[0168] α: The adjustment coefficient of the risk threshold, which controls the amplitude of threshold update;

[0169] γ: Feedback influence coefficient, reflecting the influence of feedback from construction personnel;

[0170] M new : Updated monitoring indicators;

[0171] M old : Current monitoring indicators;

[0172] R: newly calculated monitoring indicator, adjusted based on field feedback;

[0173] β: adjustment coefficient of monitoring indicators;

[0174] δ: Feedback influence coefficient, reflecting the impact of construction personnel feedback on the adjustment of monitoring indicators;

[0175] F: Feedback value, reflecting the specific feedback of on-site construction personnel on thresholds and monitoring indicators, such as safety and practicality scores.

[0176] like Figure 5 As shown, the model, when used, includes:

[0177] Data collection and calculation: During the monitoring process, real-time data is continuously collected and its average value D and new monitoring indicators R are calculated.

[0178] Feedback acquisition: Communicate with on-site construction personnel to collect their feedback on current thresholds and monitoring indicators, and quantify them into feedback values ​​F (such as a score of 0-10) to reflect their satisfaction with the system and actual needs.

[0179] Threshold update: Update the risk threshold T using the formula new By setting appropriate α and γ values, the sensitivity of the update can be controlled. For example, if field feedback indicates that the current threshold is too high, α and γ can be appropriately increased to make the threshold adjust faster.

[0180] Monitoring indicator update: Similarly, the monitoring indicator M is updated through the formula newAccording to the changes in on-site needs, the monitoring indicators are adjusted through appropriate β and δ values ​​to ensure that they are more in line with the actual situation.

[0181] Periodic evaluation and adjustment: Evaluate the updated thresholds and monitoring indicators regularly (e.g., weekly or monthly) to ensure their effectiveness. Based on actual conditions and feedback, continue to adjust the α, β, γ, and β values ​​to optimize the entire monitoring process.

[0182] S7. Integrate external data sources

[0183] Introducing external data sources, such as weather forecasts, traffic conditions, and changes in the surrounding environment, and correlating them with data from the construction site for analysis can help to more comprehensively assess construction risks, especially when facing external environmental influences, and can improve the accuracy and reliability of early warnings. Using big data analysis, find the relationship between external factors and construction risks.

[0184] In this embodiment, the method also designs a real-time data visualization platform to present the monitored data in a graphical form. Through a variety of display methods such as dashboards, heat maps, and trend charts, the construction team can quickly understand data change trends and potential risks.

[0185] Embodiment 2:

[0186] like Figure 3 As shown, based on the real-time monitoring and early warning method for construction based on big data described in Example 1, the unsupervised learning algorithm used by the intelligent anomaly detection model includes the following specific processes:

[0187] Data Collection

[0188] Collect historical data: Collect various monitoring data from the construction site, including stress, vibration, temperature, humidity, etc. These data should be as comprehensive as possible and reflect the normal operating status.

[0189] Data preprocessing

[0190] Clean the data: remove noise and missing values ​​to ensure data quality. Mark outliers but do not remove them because subsequent models will identify these anomalies.

[0191] Feature standardization: Standardize the data to eliminate the dimensional effects between different features and ensure the effectiveness of model training.

[0192] Feature selection and extraction

[0193] Select key features: Select features relevant to anomaly detection from the raw data, such as vibration frequency, temperature, humidity, etc.

[0194] Extract features: Use methods such as sliding window technology to extract features of time series data, such as mean, standard deviation, maximum and minimum values.

[0195] Model Training

[0196] Build an isolation forest model: Initialize the isolation forest, set the number of trees (e.g. 100 trees) and the random state. Train the standardized feature data and build a decision tree by randomly selecting features and split points.

[0197] Anomaly score calculation

[0198] Calculate anomaly score: Use the trained isolation forest model to calculate the anomaly score for each sample data. The higher the score, the greater the possibility that the data point is considered anomaly. The calculation formula is as follows: Figure 4 As shown in the formula, E(h(x)) is the expected depth at which sample x is isolated in the forest, and c(n) is the normalization constant.

[0199] Threshold setting

[0200] Determine the anomaly threshold: Select an appropriate threshold by analyzing the distribution of anomaly scores. A common method is to select the percentile of the score, such as selecting the top 5% of the score as the anomaly threshold.

[0201] Anomaly detection and alarm

[0202] Detect anomalies: Input real-time data into the model, calculate the anomaly score and compare it with the set threshold. If the score exceeds the threshold, mark the sample as an anomaly.

[0203] Trigger alarm: Once an anomaly is detected, the early warning system is automatically triggered and a notification is sent to the construction team to ensure timely response.

[0204] Model update and iteration

[0205] Model feedback mechanism: Regularly review and analyze identified anomalies and feed new data back to the model for continuous optimization and updating.

[0206] Self-learning: As new data is continuously input, the isolation forest model is retrained, features and thresholds are adjusted to ensure that the model always adapts to changes in the construction site.

[0207] Intelligent anomaly detection is performed using an unsupervised learning algorithm based on isolation forests. Through systematic data collection, preprocessing, model training and real-time detection, it can effectively identify abnormal situations at construction sites, ensure construction safety and improve management efficiency.

[0208] The beneficial effects of the present invention are:

[0209] This real-time monitoring and early warning method for construction based on big data can significantly improve the response speed and accuracy of early warning by introducing a dynamic adjustment and feedback mechanism based on big data, combining real-time data streams and machine learning algorithms. This method can update risk thresholds and monitoring indicators in real time, and promptly reflect the actual situation on site, thereby reducing the risk of accidents and improving the safety and efficiency of construction management.

[0210] In the description of this specification, the description with reference to the terms "one embodiment", "some embodiments", "example", "specific example", or "some examples" etc. means that the specific features, structures, materials or characteristics described in conjunction with the embodiment or example are included in at least one embodiment or example of the present invention. In this specification, the schematic representations of the above terms do not necessarily refer to the same embodiment or example. Moreover, the specific features, structures, materials or characteristics described may be combined in any one or more embodiments or examples in a suitable manner. In addition, those skilled in the art may combine and combine the different embodiments or examples described in this specification and the features of the different embodiments or examples, without contradiction.

[0211] Although embodiments of the present invention have been shown and described, it will be appreciated by those skilled in the art that various changes, modifications, substitutions and variations may be made to the embodiments without departing from the principles and spirit of the present invention, and that the scope of the present invention is defined by the appended claims and their equivalents.

Claims

1. A real-time monitoring and early warning method for construction based on big data, characterized in that: The following steps are involved: S1. Clarify monitoring targets and sensing parameters; Determine the monitoring targets during construction, such as the structural safety of the building, construction progress, and environmental impact. At the same time, determine the sensing parameters, such as vibration frequency, temperature, humidity, stress, and displacement. The selected monitoring targets and sensing parameters are used to reflect the construction status and potential risks. S2. Establish a multidimensional data model; After clarifying the monitoring objectives, a multidimensional data model is established, covering the spatial, temporal, physical and environmental dimensions, and a data framework is formed by integrating different types of data sources, such as sensor data, historical data, and external environmental data; S3. Establish risk thresholds and early warning mechanisms; Set the risk threshold for each sensor parameter and develop an early warning mechanism. When the monitoring data exceeds the set threshold, the early warning is automatically triggered. Combined with the dynamic adjustment algorithm, the threshold is continuously optimized according to the changes in field conditions to accurately reflect the actual risk level; S4, real-time data stream for data processing; Based on real-time data stream processing technology, real-time data is quickly processed through the stream processing framework; S5. Build an intelligent anomaly detection model; Based on historical data and normal operation data, build intelligent anomaly detection models to detect abnormal situations and identify potential abnormal patterns through unsupervised learning algorithms; S6. Implement dynamic adjustments and feedback loops; During the monitoring process, a dynamic adjustment and feedback mechanism is established to regularly evaluate and update risk thresholds and monitoring indicators based on real-time data and early warning results; S7, integration of external data sources; Introduce external data sources, such as weather forecasts, traffic conditions, and changes in the surrounding environment, and correlate them with the data at the construction site for analysis.

2. According to the big data-based real-time monitoring and early warning method for construction according to claim 1, it is characterized in that: In step S1, based on the determined sensing parameters, a sensor network is arranged at the construction site, including vibration frequency, temperature, humidity, stress, and displacement sensors. The sensors are connected via a wireless network to collect various environmental and structural data during the construction process in real time.

3. The method for real-time monitoring and early warning of construction based on big data according to claim 1 is characterized in that: In step S2, the dimensions in the multidimensional data model include: Spatial dimension: represents the location of each monitoring point in the construction site; Time dimension: The timestamp of the monitoring data records the specific time when the data was collected; Physical dimension: physical parameters related to structural safety, such as stress and temperature; Environmental dimension: Environmental factors that affect construction, such as wind speed, humidity, and rainfall.

4. The method for real-time monitoring and early warning of construction based on big data according to claim 1 is characterized in that: In step S3, the dynamic adjustment algorithm automatically optimizes the risk threshold according to the changes in the real-time monitoring data, and adjusts the threshold through continuous iteration based on historical data and real-time data; And build the following dynamic adjustment threshold model: T new =T old +β·(D-T old ); in: T new is the adjusted risk threshold; T old is the current risk threshold; D is the average value of real-time monitoring data; β is the adjustment coefficient, which ranges from [0,1] and determines the magnitude of the adjustment; When this model is used, Data collection and calculation: collect real-time monitoring data, such as stress and vibration, and calculate the average value D of these data, updating D each time new data arrives; Threshold adjustment: In each monitoring cycle, a new risk threshold T is calculated based on real-time data. new , if the newly calculated D is equal to the current threshold T old If there are significant differences, adjustments will be made through the model; Dynamic optimization: The adjustment coefficient β is set according to the volatility of historical data. If the historical data changes greatly, a smaller β value such as 0.1 is set to make the threshold adjustment smoother. On the contrary, if the data is relatively stable, increase β to 0.5 to speed up the threshold response speed. Early warning trigger mechanism: When the real-time monitoring data exceeds the dynamically adjusted threshold T new When a warning occurs, it will automatically trigger the early warning mechanism and notify the construction team; Feedback and iteration: After each warning is triggered, collect on-site feedback, evaluate the accuracy and timeliness of the warning, use the feedback information to further adjust the β value and threshold setting strategy, and regularly review and optimize the model to ensure that it matches the changes in the construction environment.

5. The method for real-time monitoring and early warning of construction based on big data according to claim 4 is characterized in that: When the adjustment coefficient β is adjusted dynamically, the following adjustment methods are also included: Rule of thumb: Based on the specific conditions of the construction site and industry experience, professionals or the safety management team provide recommended values, which are adjusted based on the actual operations and historical events on site; Adaptive adjustment: After setting the initial value, make adjustments through real-time feedback to monitor the accuracy and frequency of early warnings. If there are too many false alarms, gradually reduce the value of β; if the frequency of missed alarms is high, increase the value of β; Adjustment based on optimization algorithms: Use genetic algorithms and machine learning methods such as particle swarm optimization to find the best β value through simulation experiments, train on historical data, and determine the best adjustment coefficient by optimizing objective functions such as reducing false positive and false negative rates; Threshold backtesting: Backtest past monitoring data to simulate the impact of different beta values, analyze the warning effect under each setting, and select the adjustment coefficient with the best performance.

6. The method for real-time monitoring and early warning of construction based on big data according to claim 1 is characterized in that: In step S4, the stream processing framework includes a distributed platform and a stream processing engine, wherein the distributed platform is used to process real-time data streams and supports high throughput, scalability and persistence, and the stream processing engine is used to support stateful computing and process real-time data streams and batch data; When data processing is performed based on real-time data streams, the following specific processes are included: Data collection: Real-time data from the construction site, such as temperature, vibration, and humidity, is collected through a sensor network. The collected data is sent to a cluster on a distributed platform for storage; Data stream forwarding: Use the topics of the distributed platform to classify different types of real-time data and forward them to the corresponding stream processing engine; Real-time computing: Configure data stream processing jobs in the stream processing engine, use the stream computing engine to process data in real time, and analyze and calculate data by writing processing logic, such as window computing and state management; Anomaly detection: Anomaly detection is performed based on sensing parameters. Assuming that we focus on vibration and temperature parameters, the anomaly detection formula is as follows: A(t)=λ·(f(t)-μ f )+v·(θ(t)-μ θ ); Among them, A(t) represents the abnormal value at time t, λ and v are weight coefficients, reflecting the influence of vibration and temperature on the abnormal value, f(t) and θ(t) are the vibration frequency and temperature monitored in real time, μ f and μ θ It is the average value of historical data and is used to determine the degree of abnormality of the current value; Real-time warning mechanism: When the calculated abnormal value A(t) exceeds the set threshold, a real-time warning is triggered and an alarm message is sent to the construction team.

7. The method for real-time monitoring and early warning of construction based on big data according to claim 1 is characterized in that: In step S5, the intelligent anomaly detection model is as follows: Among them, A represents the degree of abnormality detected; n is the sample size of the current monitoring data; X i is the data point of the i-th sample; μ is the mean of the sample, that is, the average value of normal operating data; σ is the standard deviation of the sample, which is used to measure the dispersion of the data.

8. The method for real-time monitoring and early warning of construction based on big data according to claim 7 is characterized in that: When the intelligent anomaly detection model is used, the following specific processes are included: Data preparation: Collect historical data and normal operation data during the construction process to ensure the quality and representativeness of the data. The data includes various monitoring parameters such as stress, vibration, and temperature. Feature extraction: Extract features from monitoring data and calculate the mean μ and standard deviation σ. The features will be used for subsequent anomaly detection calculations. Calculate the degree of anomaly: When new data arrives each time, calculate the degree of anomaly of the current data set. If the value of intelligent anomaly detection A exceeds the set threshold, it is considered an anomaly. Model iteration and self-learning: As time goes by, the model is continuously updated, and the mean μ and standard deviation σ are recalculated whenever new normal operation data is added; Abnormal pattern recognition: Combined with the isolation forest algorithm, the calculated abnormal degree is further analyzed to identify potential abnormal patterns. By clustering the samples, similar abnormal events are grouped; Early warning trigger mechanism: When the model detects an anomaly and confirms its pattern, an early warning is automatically triggered to send an alert to the construction team.

9. The method for real-time monitoring and early warning of construction based on big data according to claim 1 is characterized in that: In step S6, the dynamic adjustment and feedback mechanism is based on an update model, which is used to update the risk threshold T and the monitoring indicator M. The update model is as follows: T new =T old +α·(D-T old )+γ·F M new =M old +β·(RM old )+δ·F; in, T new : Updated risk threshold; T old : Current risk threshold; D: average value of real-time monitoring data or other key indicators; α: The adjustment coefficient of the risk threshold, which controls the amplitude of threshold update; γ: Feedback influence coefficient, reflecting the influence of feedback from construction personnel; M new : Updated monitoring indicators; M old : Current monitoring indicators; R: newly calculated monitoring indicator, adjusted based on field feedback; β: adjustment coefficient of monitoring indicators; δ: Feedback influence coefficient, reflecting the impact of construction personnel feedback on the adjustment of monitoring indicators; F: Feedback value, reflecting the specific feedback of on-site construction personnel on thresholds and monitoring indicators, such as safety and practicality scores.

10. The method for real-time monitoring and early warning of construction based on big data according to claim 1 is characterized in that: This method also designs a real-time data visualization platform to present the monitored data in a graphical form. Through multiple display methods such as dashboards, heat maps and trend charts, the construction team can quickly understand data change trends and potential risks.

Images