Construction safety risk real-time early warning system and method based on multi-source data fusion

Abstract

The invention discloses a construction safety risk real-time early warning system and method based on multi-source data fusion. The system comprises a data acquisition layer used for acquiring multi-source data of a construction site through various sensors; the data transmission and processing layer is used for preprocessing the data; the risk analysis engine is used for performing data fusion and dynamic risk calculation through a machine learning model based on the preprocessed multi-source data; and the early warning and feedback layer is used for performing graded early warning according to the risk assessment result. According to the invention, through fusion of multi-source information such as personnel positions, structure states, environmental factors and video monitoring, real-time, accurate and dynamic evaluation and early warning of construction safety risks are realized by using an intelligent algorithm, the hysteresis quality and one-sidedness of a traditional safety management method are effectively overcome, and the safety management level of a construction site is significantly improved.

Description

Technical Field

[0001] This invention relates to the field of construction engineering safety management technology, specifically to a real-time early warning system and method for construction safety risks based on multi-source data fusion. Background Technology

[0002] Safety management is a core component of project management during construction. Traditional safety management methods rely primarily on regular inspections by safety officers, workers' safety awareness, and post-accident analysis. However, this approach suffers from significant delays and subjectivity, failing to provide real-time and accurate warnings for sudden safety risks such as falls from heights, falling objects, collapses, and mechanical injuries.

[0003] With technological advancements, some single monitoring methods have begun to be applied, such as using cameras for video surveillance or using specific sensors to monitor the load on tower cranes. However, these systems are often information silos, capable of only alerting to a single risk, resulting in a high false alarm rate, and lacking the ability to comprehensively assess the combined effects of multiple risks in complex environments. For example, when a tower crane is performing hoisting operations near unstable scaffolding in strong winds, a single system cannot detect this complex and extremely high risk.

[0004] Therefore, there is an urgent need in this field for a security risk early warning solution that can integrate multiple information sources to achieve intelligent, real-time, and accurate early warning. Summary of the Invention

[0005] The technical problem to be solved by this invention is to overcome the defects of existing construction safety management, such as lag, one-sidedness and high false alarm rate, and to provide a construction safety risk real-time early warning system and method based on multi-source data fusion that can achieve comprehensive, real-time and accurate early warning.

[0006] To solve the above-mentioned technical problems, the present invention adopts the following technical solution: In a first aspect, the present invention provides a real-time early warning system for construction safety risks based on multi-source data fusion, comprising: Data acquisition layer: Composed of various sensor modules deployed on the construction site, used to comprehensively collect multi-source heterogeneous data such as physical environment, personnel and equipment location, and engineering structure status.

[0007] Data transmission and processing layer: responsible for receiving raw data from the data acquisition layer and performing preprocessing operations such as data cleaning, filtering, and format standardization.

[0008] Risk Analysis Engine: This is the core of the system, containing a built-in dynamic risk assessment model based on machine learning algorithms. The engine receives pre-processed multi-source data and uses data fusion techniques (such as DS evidence theory and Bayesian networks) to calculate a comprehensive and dynamically changing risk assessment result.

[0009] Early warning and feedback layer: Based on the results output by the risk analysis engine, and in accordance with the preset hierarchical early warning strategy, early warning information is sent to relevant personnel through terminals such as APP, web interface, and sound and light alarms.

[0010] Optional digital twin module: Constructs a virtual model that corresponds to the physical construction site at a 1:1 scale, and visualizes risk information to assist managers in making decisions.

[0011] Secondly, this invention provides a real-time early warning method for construction safety risks based on multi-source data fusion, comprising the following steps: S1: Real-time acquisition of multi-source data from the construction site; S2: Preprocess multi-source data; S3: Input the processed data into the risk assessment model for fusion calculation to generate risk assessment results; S4: Trigger a tiered early warning based on the risk assessment results.

[0012] The beneficial effects of this invention are as follows: ① Comprehensive perception: By integrating multiple sensors, it achieves comprehensive perception of personnel, machinery, environment, and structural status, breaking down information silos.

[0013] ② Precise early warning: By adopting a dynamic risk assessment model based on machine learning and multi-source data fusion technology, it can identify complex coupled risks that cannot be detected by a single sensor, significantly reducing the false alarm rate and the missed alarm rate.

[0014] ③ Real-time dynamics: The system can achieve 24 / 7 uninterrupted automatic monitoring and risk assessment, realizing the transformation from "post-event analysis" to "pre-event warning and in-event control".

[0015] ④ Decision support: By combining with digital twin technology, abstract risk data is transformed into intuitive and visual information, which greatly improves the efficiency and accuracy of managers' decision-making. Attached Figure Description

[0016] Figure 1 This is an overall architecture block diagram of the system described in one embodiment of the present invention.

[0017] Figure 2 This is a flowchart of the method described in one embodiment of the present invention.

[0018] Figure 3 This is a schematic diagram illustrating the principle of data fusion and risk analysis in one embodiment of the present invention. Detailed Implementation

[0019] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments.

[0020] refer to Figure 1 The system architecture in this embodiment includes four layers: ① Data acquisition layer 101: includes UWB positioning base stations and tags 102 for centimeter-level precise positioning of personnel and equipment; tilt / displacement sensors 103, installed on key support structures; smart safety helmets 104 with built-in IMU sensors; environmental monitoring stations 105 for monitoring meteorological data; and high-definition network cameras 106 for video surveillance and AI behavior analysis.

[0021] ② Data transmission and processing layer 107: Data is aggregated to the edge computing gateway or cloud server via wireless networks such as LoRa, 5G or Wi-Fi for data preprocessing.

[0022] ③ Risk Analysis Engine 108: Runs on a cloud server. It receives pre-processed data and calls a pre-trained machine learning model (e.g., an XGBoost or LSTM model trained using historical data) to calculate risk. Simultaneously, the engine employs DS evidence theory to fuse information from different confidence levels from location, structural monitoring, and video analysis to arrive at a final comprehensive risk value.

[0023] ④ Early warning and feedback layer 109: including digital twin visualization platform 110, which displays the three-dimensional model and risk heat map of the entire construction site in real time; mobile APP 111, which pushes early warning information to safety officers and workers; and on-site sound and light alarm 112, which is automatically triggered in extremely high risk situations.

[0024] refer to Figure 2 The workflow of this embodiment is as follows: Step S201: The system starts up, and all sensors begin to continuously collect data.

[0025] Step S202: The data transmission and processing layer cleans and standardizes the data to form a data stream with a unified format.

[0026] Step S203: The risk analysis engine extracts features from the data stream, such as "the rate of change of distance between personnel A and the tower crane hook", "the instantaneous value of the tilt angle of scaffolding B", and "the average wind speed of the current area".

[0027] Step S204: Input the feature vector into the risk assessment model. The model outputs a quantified risk value (e.g., 0-100) and the main risk types.

[0028] Step S205: Determine if the risk value exceeds the threshold. If it does not exceed the threshold, return to S201 to continue monitoring; if it exceeds the threshold, proceed to S206.

[0029] Step S206: Execute tiered warnings based on risk level (e.g., low, medium, high). Low-level risks are only marked on the digital twin platform; medium-level risks are pushed to the safety officer's APP; high-level risks trigger on-site audible and visual alarms and a company-wide broadcast simultaneously.

[0030] refer to Figure 3 This demonstrates the principle of data fusion. Raw information from location data 301, structural monitoring data 302, environmental data 303, and video analysis data 304, after feature extraction, becomes evidence with inherent uncertainty. The risk analysis engine 305 uses fusion algorithms (such as DS theory 306) to combine this evidence, ultimately obtaining a more confident comprehensive risk judgment 307, such as "There is a risk of object impact in the tower crane hoisting area, confidence level 92%".

[0031] The above description is merely a specific embodiment of the present invention, but the scope of protection of the present invention is not limited thereto. Any variations or substitutions that can be easily conceived by those skilled in the art within the scope of the technology disclosed in the present invention should be included within the scope of protection of the present invention. Therefore, the scope of protection of the present invention should be determined by the scope of the claims.

Claims

1. A real-time early warning system for construction safety risks based on multi-source data fusion, characterized in that, include: ① Data acquisition layer, including various sensor modules deployed at the construction site, used to collect physical environment data, personnel and equipment location data, and engineering structure status data; Claims ② Data transmission and processing layer, which is communicatively connected to the data acquisition layer, is used to receive and preprocess the data acquired by the various sensor modules; ③ Risk analysis engine, which is connected to the data transmission and processing layer, is used to perform data fusion and dynamic risk calculation based on preprocessed multi-source data through a preset risk assessment model to generate risk assessment results; ④ The early warning and feedback layer is connected to the risk analysis engine and is used to send graded early warning information to designated terminals based on the risk assessment results.

2. The system according to claim 1, characterized in that, The data acquisition layer includes: ①UWB or BLE positioning tags used to collect real-time location information of personnel and equipment; ② Inclination and displacement sensors used to monitor the stability of scaffolding and formwork support systems; ③ A smart safety helmet that integrates an inertial measurement unit and is used to detect falls or impacts; ④ Environmental sensors used to monitor wind speed, rainfall, temperature, and humidity; ⑤ Cameras used to collect on-site video stream data and identify safe and compliant behaviors based on computer vision algorithms.

3. The system according to claim 1, characterized in that, The system also includes: The digital twin module communicates with the risk analysis engine to build and maintain a virtual construction site model that is synchronized with the physical construction site, and to visualize the risk assessment results in the virtual construction site model.

4. The system according to claim 1, characterized in that: The risk assessment model in the risk analysis engine is a machine learning-based dynamic risk assessment model. It is trained using historical accident data, historical sensor data, and corresponding risk labels, and can output dynamic risk values ​​that change over time.

5. The system according to claim 4, characterized in that: When performing data fusion, the risk analysis engine uses DS evidence theory or Bayesian network algorithm to perform confidence fusion on heterogeneous data from different sensors in order to improve the accuracy of risk assessment.

6. A real-time early warning method for construction safety risks based on multi-source data fusion, characterized in that: Applied to the system as described in any one of claims 1-5, the method comprises: S1: Through various sensors deployed at the construction site, multi-source data on the physical environment, personnel and equipment locations, and engineering structure status are collected in real time; S2: Preprocess and standardize the collected multi-source data; S3: Input the preprocessed multi-source data into the preset risk assessment model, perform data fusion and dynamic risk calculation, and generate the risk assessment results for the current construction site; S4: Based on the risk assessment results, trigger the tiered early warning mechanism and send early warning information to the relevant terminal devices.

7. The method according to claim 6, characterized in that: The risk assessment model mentioned in step S3 is a machine learning model, and the dynamic risk calculation includes: S31: Based on the multi-source data, extract feature vectors related to security risks; S32: Input the feature vector into a trained machine learning model, and the model outputs the probability and risk level of one or more risk regions and risk types.

8. The method according to claim 6, characterized in that: The graded early warning mechanism described in step S4 triggers different intensity early warning methods according to different risk levels. The early warning methods include: highlighting in the digital twin model, sending APP push notifications to managers, and triggering on-site sound and light alarm devices.

Images