Railway construction safety intelligent monitoring system based on dynamic detection strategy

Abstract

The invention relates to a railway construction safety intelligent monitoring system based on a dynamic detection strategy, which comprises an intelligent safety helmet, a cloud platform and an alarm module, and is characterized in that the intelligent safety helmet is provided with a camera and an inertial sensor to shoot and collect construction scene data and environmental conditions in real time; the position of the intelligent mounting cap is positioned by arranging a positioning module; the cloud platform is provided with an edge calculation module, the edge calculation module is provided with a multi-module AI model and a dynamic detection strategy generation module, and the multi-mode AI model is provided with a target recognition module and a behavior recognition module. The dynamic detection strategy generation module automatically generates detection parameters according to construction scene data, environmental conditions and historical data collected by the intelligent safety helmet; the alarm module is connected with the cloud platform, and the cloud platform carries out monitoring according to the detection parameters and the target recognition module and the behavior recognition module operated by the multi-mode AI model. The method has the advantages that the detection precision and the potential hazard identification capability are remarkably improved.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of railway construction safety, and in particular to a railway construction safety intelligent monitoring system based on a dynamic detection strategy. BACKGROUND

[0002] The existing railway construction safety management relies on manual inspection, and has problems such as insufficient coverage and delayed response. The traditional AI scheme (single YOLO target detection or independent behavior recognition model) has the following defects: 1) Poor environmental adaptability: the model is not optimized for railway construction scenes (such as strong light, complex background); 2) Insufficient multi-source data fusion: lack of collaborative analysis of protection state, behavior action, and position information; 3) Real-time and precision contradiction: low-latency processing is not achieved through edge computing and cloud collaboration. SUMMARY

[0003] The purpose of the present application is to solve the problems of low efficiency of manual inspection, insufficient multi-source data fusion, and lack of closed-loop management in the prior art, and to realize active early warning and precise control of unsafe behavior, by providing a railway construction safety intelligent monitoring system based on a dynamic detection strategy, which combines real-time collected construction site data with a dynamic monitoring strategy based on historical data.

[0004] The purpose of the present application is achieved by the following technical solutions: A railway construction safety intelligent monitoring system based on a dynamic detection strategy, characterized by comprising an intelligent safety hat, a cloud platform, and an alarm module, wherein: The intelligent safety hat captures and collects construction scene data and environmental conditions in real time by setting up a camera and an inertial sensor, and the intelligent installation cap locates its position by setting up a positioning module; The cloud platform is provided with an edge computing module, the edge computing module is provided with a multi-module AI model and a dynamic detection strategy generation module, the multi-modal AI model has a target recognition module and a behavior recognition module, and the dynamic detection strategy generation module automatically generates detection parameters according to the construction scene data collected by the intelligent safety hat and the environmental conditions and historical data; The alarm module is connected to the cloud platform, the cloud platform monitors the target recognition module and the behavior recognition module running according to the detection parameters and the multi-modal AI model, and gives a warning prompt through the alarm module.

[0005] The dynamic monitoring strategy generation module comprises a construction progress sensing unit, an environment condition sensing unit and a strategy generation algorithm, wherein the construction progress sensing unit identifies the construction stage of railway construction through UWB, Beidou positioning and GIS dynamic heat map, the environment sensing unit collects light intensity and weather information through a camera and a tube-shaped sensor, and the strategy generation algorithm dynamically adjusts the detection parameters based on a migration learning model according to the construction stage and the environment condition.

[0006] The cloud platform is provided with an adaptive adjustment module, which dynamically detects the parameters of the automatically generated detection parameters according to the adaptive adjustment module, and adjusts the parameters of the edge computing unit in real time.

[0007] The adaptive adjustment module adopts a reinforcement learning algorithm, trains the model through historical data, and optimizes the model parameters of the edge computing unit in real time.

[0008] Combined with the change of construction progress and historical data, possible violations can be predicted and early warning can be given.

[0009] The alarm module comprises a local sound and light alarm, a team APP push and a monitoring center large screen early warning.

[0010] The advantages of the present application are that the problems of static detection standard, poor environmental adaptability and lack of closed-loop management in the prior art are solved, and the detection precision and potential hidden danger identification capability are significantly improved. BRIEF DESCRIPTION OF DRAWINGS

[0011] Figure 1 The flowchart of the present application is shown in the figure. Figure 2 The system architecture diagram of the present application is shown in the figure. DETAILED DESCRIPTION

[0012] The features and other related features of the present application are further described in detail below by embodiments combined with the drawings, so as to facilitate the understanding of the same by the same industry technical personnel: Embodiment: as shown in Figure 1 and Figure 2 The railway construction safety intelligent monitoring system based on dynamic detection strategy in the embodiment comprises the following technical solutions: 1) System architecture: Hardware layer: intelligent safety helmet integrated with multi-modal sensors (including camera, positioning module and inertial sensor).

[0013] Computing layer: local AI inference unit based on domestic processor.

[0014] Cloud platform: intelligent alarm cloud platform based on B / S architecture.

[0015] Data interaction layer: 4G / 5G+MQTT protocol communication.

[0016] The functions of the railway construction safety intelligent monitoring system in this embodiment include: (1) Multi-modal AI algorithm fusion Target recognition module: YOLOv5s-IR model (infrared enhancement) is adopted, the input size is optimized to 640x640, CBAM attention mechanism is introduced, and the detection accuracy of safety helmet and reflective vest in railway construction scene is ≥97%.

[0017] Support multi-scale feature fusion (C3 module), adapt to complex lighting conditions.

[0018] Behavior recognition module: Based on ResNet50-OpenPose, the skeleton points are extracted, the space-time feature vector (18 key points x 20 frames) is constructed, the bidirectional LSTM network is input, and the recognition accuracy of dangerous actions (climbing, edge protection, etc.) is ≥92%; Introduce Transformer time sequence attention mechanism to improve the action time sequence modeling ability.

[0019] Electronic fence module: Based on UWB+Beidou dual-mode positioning, combined with GIS dynamic heat map, set virtual boundary; Sliding time window algorithm (window size=3 seconds) is adopted to detect illegal intrusion behavior of personnel.

[0020] (2) Multi-source data fusion and decision engine: Data fusion algorithm: Through D-S evidence theory, the target recognition result, behavior analysis result and positioning data are fused to generate comprehensive risk score; Risk score formula:

[0021] Wherein α, β, γ are weight coefficients (determined by AHP hierarchical analysis method).

[0022] (3) Self-adaptive adjustment module: Function: Real-time optimization of model parameters of edge computing unit to adapt to environmental changes and construction needs.

[0023] Implementation: Reinforcement learning algorithm is adopted, model is trained through historical data, input size of YOLOv5s-IR is dynamically adjusted (such as from 640x640 to 480x480 to reduce computing load); Adjust the skeleton point extraction frequency of ResNet50-OpenPose-LSTM (such as from 20 frames / second to 15 frames / second to adapt to low light conditions).

[0024] (4) Potential hazard prediction Function: Based on construction progress changes and historical data, predict possible violations and give early warnings.

[0025] Implementation: Train an LSTM model with historical data to predict high-risk behaviors in the next phase (such as missing edge protection). Combine construction progress information to adjust detection strategies in advance (such as increasing the detection weight of tool falling during track laying).

[0026] Although the above embodiments have been described in detail with reference to the accompanying drawings for the purpose of illustrating the concepts and embodiments of the present application, those skilled in the art can recognize that various improvements and changes can be made to the present application without departing from the scope defined by the claims, and therefore detailed description is not repeated here.

Claims

1. An intelligent railway construction safety monitoring system based on a dynamic detection strategy, characterized by: It includes a smart helmet, a cloud platform and an alarm module, including: The smart helmet uses a camera and inertial sensors to capture construction scene data and environmental conditions in real time, and the smart installation cap locates its position through a positioning module. The cloud platform is equipped with an edge computing module, which is equipped with a multi-module AI model and a dynamic detection strategy generation module. The multimodal AI model has a target recognition module and a behavior recognition module. The dynamic detection strategy generation module automatically generates detection parameters based on the construction scene data, environmental conditions, and historical data collected by the smart helmet. The alarm module is connected to the cloud platform, which monitors the target recognition module and behavior recognition module running according to the detection parameters and the multimodal AI model, and issues early warning prompts through the alarm module.

2. A railway construction safety intelligent monitoring system based on a dynamic detection strategy according to claim 1, characterized in that: The dynamic monitoring strategy generation module includes a construction progress perception unit, an environmental condition perception unit, and a strategy generation algorithm. The construction progress perception unit uses UWB, Beidou positioning, and GIS dynamic heat maps to identify the construction stage of railway construction. The environmental perception unit uses cameras and tube sensors to collect light intensity and weather information. The strategy generation algorithm is based on a transfer learning model and dynamically adjusts detection parameters according to the construction stage and environmental conditions.

3. A railway construction safety intelligent monitoring system based on a dynamic detection strategy according to claim 1, characterized in that: The cloud platform is equipped with an adaptive adjustment module, which adjusts the parameters of the edge computing unit in real time based on the dynamic detection strategy of automatically generated detection parameters.

4. A railway construction safety intelligent monitoring system based on a dynamic detection strategy according to claim 2, characterized in that: The adaptive adjustment module uses a reinforcement learning algorithm to train the model through historical data and optimize the model parameters of the edge computing unit in real time.

5. A railway construction safety intelligent monitoring system based on a dynamic detection strategy according to claim 1, characterized in that: Combining construction progress changes with historical data, it predicts possible violations and provides early warnings.

6. A railway construction safety intelligent monitoring system based on a dynamic detection strategy according to claim 1, characterized in that: The alarm module includes local sound and light alarms, team APP push and monitoring center large-screen warnings.

Images