An intelligent safety helmet dynamic monitoring and early warning system based on multi-source data fusion

Abstract

The application discloses a kind of intelligent safety helmet dynamic monitoring and early warning system based on multi-source data fusion, it includes multi-source sensor module, for collecting human and environmental data;Edge computing and fusion processing unit, built-in multi-source data fusion algorithm and early warning decision logic, for analyzing and calculating whether the human and environmental data collected exceed alarm threshold;Local storage module, for storing the original human and environmental data collected, human and environmental data fusion results and video clip data;Alarm execution module, for early warning to human and environmental data exceeding alarm threshold;Wireless communication module, relevant data stored in local storage module is uploaded to management platform by wireless communication technology;Power management module, power supply for each module of the whole system.The application realizes personnel health status, environmental risk, position trajectory comprehensive evaluation and intelligent early warning by human data and dangerous gas concentration data.

Description

Technical Field

[0001] This invention belongs to the field of cross-monitoring technology of smart wearable devices and the Internet of Things. Specifically, it relates to a smart safety helmet dynamic monitoring and early warning system based on multi-source data fusion. Background Technology

[0002] Traditional industrial safety protection relies primarily on a combination of passive protection and manual supervision. Personal protective equipment (PPE) (such as ordinary safety helmets) only provides physical protection and cannot actively sense the physiological state of workers or environmental risks. Safety management mainly relies on manual methods such as team leader inspections and on-site supervision by monitors, which suffers from problems such as blind spots in supervision, delayed response, and incomplete data recording. In high-risk scenarios such as confined spaces and high-altitude operations, traditional methods are insufficient for real-time and continuous monitoring of personnel status and environmental risks. The maturity of IoT technology has made it possible to integrate various miniaturized sensors into wearable devices. Multi-sensor fusion technology uses algorithms to perform spatiotemporal synchronization and joint analysis of heterogeneous data sources such as accelerometers, gyroscopes, heart rate sensors, gas sensors, and positioning modules (GPS / BeiDou / UWB), enabling comprehensive perception of personnel posture, physiological parameters, and environmental information. Advances in edge computing technology allow smart safety helmets to complete data preprocessing and preliminary analysis locally, reducing cloud transmission latency and meeting real-time early warning requirements.

[0003] Existing smart safety helmets typically follow a three-layer IoT architecture: "perception layer - transmission layer - application layer". Perception layer: Integrates multiple sensors (six-axis accelerometer, heart rate sensor, gas sensor, temperature and humidity sensor, camera sensor, positioning module, etc.) to collect multi-dimensional data on the wearer and the environment; Transport layer: Uploads data to the management platform via wireless communication technologies such as 4G / 5G, Wi-Fi, and LoRa; Application layer: Data is stored, analyzed, and visualized by cloud platforms or local servers, supporting remote monitoring and decision-making by administrators.

[0004] The current technology suffers from a deficiency in safety monitoring systems used in construction sites, mines, or emergency operations. These systems typically separate human physiological monitoring, environmental gas monitoring, video recording, and location tracking into independent devices or unconnected subsystems. This architecture leads to the following causal chain of technical flaws: because human condition data (heart rate, body temperature, etc.) and hazardous gas concentration data are collected by different sensors and not fused at the local edge, the system cannot issue an early warning when gas concentrations rise but the human body has not yet shown obvious discomfort; conversely, when the human body exhibits abnormalities (such as heatstroke causing an increased heart rate) while gas concentrations are normal, existing systems often misjudge environmental hazards or ignore the health risks to personnel. This results in slow warning response times and low accuracy. Video recordings are usually only used for post-event playback, while location data (such as UWB or GPS) is transmitted independently. When an alert is triggered, the backend cannot automatically associate and push the personnel's location at the time of the alert with the video clips, requiring manual simultaneous review of multiple systems, resulting in long emergency response times and the potential loss of critical evidence in areas with signal interruptions. Due to the separation of functions, operators need to wear multiple independent devices such as gas detectors, smart bracelets, cameras and positioning tags at the same time. This not only significantly increases the cost of hardware purchase and maintenance, but also results in high overall power consumption and short battery life due to independent power supply and wireless transmission for multiple devices. Frequent charging affects the reliability of actual use. Summary of the Invention

[0005] To address the problems existing in the prior art, this invention provides an intelligent safety helmet dynamic monitoring and early warning system based on multi-source data fusion. By performing spatiotemporal alignment and joint analysis of human physiological / postural data, hazardous gas concentration data, video image data, and high-precision positioning data, it achieves comprehensive assessment and intelligent early warning of personnel health status, environmental risks, and location trajectory.

[0006] To achieve the above-mentioned technical objectives, the technical solution adopted by the present invention is as follows: A smart safety helmet dynamic monitoring and early warning system based on multi-source data fusion includes a multi-source sensor module, an edge computing and fusion processing module, a local storage module, an alarm execution module, a wireless communication module, and a power management module; The multi-source sensor module is communicatively connected to the edge computing and fusion processing unit and integrated into the helmet body to collect human and environmental data; The edge computing and fusion processing unit has a built-in multi-source data fusion algorithm and early warning decision logic, which is used to analyze and calculate whether the collected human body and environmental data exceed the alarm threshold. The local storage module is communicatively connected to the edge computing and fusion processing unit to store the collected raw human body and environmental data, the fusion results of human body and environmental data, and video clip data. The alarm execution module communicates with the edge computing and fusion processing unit to provide early warnings for human and environmental data that exceed the alarm threshold; it includes an audible and visual alarm and a vibration motor.

[0007] The wireless communication module is connected to the local storage module and uploads the relevant data stored in the local storage module to the management platform through wireless communication technology. The power management module supplies power to all modules in the entire system.

[0008] Furthermore, the multi-source sensor module includes a human body status sensor, a hazardous gas sensor, a video acquisition module, and a positioning module; Human body status sensors include a photoplethysmography (PPG) sensor, an infrared thermometer, and a three-axis accelerometer and gyroscope. The PPG sensor, installed inside the helmet where it contacts the forehead, is used to extract heart rate and heart rate variability. The non-contact infrared thermometer, pointed at the wearer's forehead, measures body temperature. The three-axis accelerometer and gyroscope (six-axis inertial measurement unit, IMU), mounted on the top of the helmet, detects falls, abnormal head movements, and other postures.

[0009] Hazardous gas sensors include electrochemical carbon monoxide sensors, catalytic combustion methane sensors, and metal oxide semiconductor hydrogen sulfide sensors; they are installed in the airflow channel at the front of the safety helmet to ensure smooth airflow.

[0010] The video capture module uses a low-power CMOS image sensor (5 megapixels) with a horizontal field of view of 120°, and is mounted on the front of the safety helmet. It is used to compress recorded video clips using H.264 hardware encoding to generate a video stream.

[0011] The positioning module employs a multi-mode Global Navigation Satellite System (GNSS, including GPS and BeiDou), supplemented by inertial navigation DR (dead reckoning) for use in tunnels or indoor areas without signal. The positioning data update frequency is 1Hz.

[0012] Furthermore, a conditioning circuit is provided between the human body state sensor and the edge computing and fusion processing unit; The output of the photoplethysmography (PPG) sensor is connected to a transimpedance amplifier (TIA) to convert it into a voltage signal. Then, it is filtered by a bandpass filter (0.5Hz~5Hz) to remove baseline drift and high-frequency noise, and finally sent to an analog-to-digital converter (ADC). The infrared temperature sensor outputs a tiny voltage signal, which is amplified 50 times by an instrumentation amplifier (INA) before entering the analog-to-digital converter. The triaxial accelerometer and gyroscope are directly connected to the edge computing and fusion processing unit via the I²C bus.

[0013] Furthermore, when the multi-source data fusion-based smart safety helmet dynamic monitoring and early warning system loses network access, the entire system operation steps include: S301, The edge computing unit detected a disconnection in the wireless communication link (no heartbeat packet was received from the background for 10 consecutive seconds).

[0014] S302, automatically switch to local independent operation mode, all sensors continue to collect data.

[0015] S303, the local early warning logic for human body status and gas concentration remains active; when an early warning is triggered, the audible and visual alarm and the vibration motor immediately activate.

[0016] S304. Video is continuously recorded and stored in a local circular buffer (capacity 32GB, can be overwritten cyclically).

[0017] S305: Location data is continuously recorded to local flash memory.

[0018] S306. After the network is restored, the system will automatically upload the alarm events, corresponding video clips, and location trajectories during the network outage to the backend.

[0019] Furthermore, the early warning process for human and environmental data collected by the multi-source sensor modules by the edge computing and fusion processing unit includes: S401, the edge computing unit receives the following characteristic values ​​in parallel: Human characteristics: Heart rate (HR), body temperature (Temp), fall signs (Fall). Gas characteristics: CO concentration, CH4 concentration, H2S concentration. Environmental motion characteristics: SVM combined acceleration, attitude angle.

[0020] S402. Construct the feature vector; X = [HR, Temp, Fall, C_CO, C_CH4, C_H2S, SVM] S403. Apply multi-source fusion rules based on decision trees; for example: Rule 1: If (C_CO>25ppm and HR>100), it is determined as a "gas toxicity + human stress" combined warning, and the "evacuation" command is output.

[0021] Rule 2: If (Fall = True and C_CO < 10ppm and CH4 < 0.5%VOL), then it is determined as "simple fall" and the "Fall Help" command is output.

[0022] Rule 3: If (Temp>38.5℃ and SVM lasts for 2 minutes <0.2g), it is determined as "heatstroke and exhaustion", and the "medical assistance" command is output.

[0023] Rule 4: If (C_CH4>1.5%VOL and Fall = False), then it is determined as "early stage of gas leak" and "early warning" is output.

[0024] S404, Integrated decision-making output comprehensive risk level (0: normal, 1: attention, 2: warning, 3: alarm).

[0025] S405, Alarm module driven according to hazard level: Level 0: No action. Level 1: Local vibration alert. Level 2: Yellow light flashing + buzzer sounding intermittently. Level 3: Red light flashing + continuous beeping + background forced upload of alarm event.

[0026] Compared with the prior art, the present invention has the following advantages: This invention employs a localized fusion architecture that integrates human physiological sensors, hazardous gas sensors, and video and positioning modules into a single helmet body. Furthermore, it deploys multi-source data fusion early warning rules based on decision trees within the edge computing unit. This overcomes the judgment delays and false alarms caused by the stacking of multiple devices and data silos in existing technologies. Specifically, existing systems cannot provide early warnings when gas concentrations rise but the human body has not yet shown obvious discomfort. However, this invention, by simultaneously analyzing the cross-changes in heart rate, body temperature, and multiple gas concentrations (e.g., triggering a composite evacuation command when CO exceeds 25 ppm and heart rate exceeds 100 beats / min), can issue warnings before environmental risks escalate into personal injury, while avoiding misinterpreting increased heart rate caused by heatstroke as gas poisoning. This fusion decision-making mechanism significantly improves the accuracy and response speed of early warnings, reducing false alarms and missed alarms.

[0027] Meanwhile, because this invention establishes a spatiotemporal correlation storage mechanism for video-positioning-attitude, and automatically binds all early warning events with their corresponding location coordinates and video clips, the backend can directly retrieve the on-site footage and trajectory information at that moment when an early warning is triggered, eliminating the need for manual simultaneous review across multiple separate systems. This significantly shortens emergency response time and reduces the difficulty of accident tracing. Furthermore, by setting up network outage detection and local independent operation modes, even if the wireless communication link is interrupted, the safety helmet can still independently complete all sensor data acquisition, local early warning decision-making, audible and visual vibration alarms, and video storage. Once the network is restored, it automatically retransmits key data, thus ensuring monitoring reliability in network outage environments. The above technical solution also reduces redundant hardware overhead by having multiple sensors share a power supply and communication unit, lowering the overall system cost and power consumption, and extending the effective battery life of the equipment. Attached Figure Description

[0028] Figure 1This is a block diagram of the overall structure of an intelligent safety helmet dynamic monitoring and early warning system based on multi-source data fusion, as described in an embodiment of the present invention. Figure 2 This is a flowchart of the system operation steps when the multi-source data fusion smart safety helmet dynamic monitoring and early warning system loses network connection in an embodiment of the present invention. Detailed Implementation

[0029] To facilitate understanding by those skilled in the art, the present invention will be further described below with reference to embodiments and accompanying drawings. The content mentioned in the embodiments is not intended to limit the present invention.

[0030] like Figure 1 As shown, this embodiment provides a smart safety helmet dynamic monitoring and early warning system based on multi-source data fusion, including a multi-source sensor module, an edge computing and fusion processing module, a local storage module, an alarm execution module, a wireless communication module, and a power management module; A multi-source sensor module, integrated into the helmet body, is used to collect data on the human body and the environment; The edge computing and fusion processing unit has a built-in multi-source data fusion algorithm and early warning decision logic, which is used to analyze and calculate whether the collected human body and environmental data exceed the alarm threshold. The local storage module is used to store the collected raw human body and environmental data, the fusion results of human body and environmental data, and video clip data; The alarm execution module is used to issue early warnings for human and environmental data that exceed the alarm threshold; it includes an audible and visual alarm and a vibration motor.

[0031] The wireless communication module uploads relevant data stored in the local storage module to the management platform through wireless communication technologies (4G / 5G, Wi-Fi, and LoRa).

[0032] The power management module supplies power to all modules in the system and manages low-power modes.

[0033] The multi-source sensor module includes a human body status sensor, a hazardous gas sensor, a video acquisition module, and a positioning module; Human body status sensors include a photoplethysmography (PPG) sensor, an infrared thermometer, and a three-axis accelerometer and gyroscope. The PPG sensor, installed inside the helmet where it contacts the forehead, is used to extract heart rate and heart rate variability. The non-contact infrared thermometer, pointed at the wearer's forehead, measures body temperature. The three-axis accelerometer and gyroscope (six-axis inertial measurement unit, IMU), mounted on the top of the helmet, detects falls, abnormal head movements, and other postures.

[0034] Hazardous gas sensors include electrochemical carbon monoxide sensors, catalytic combustion methane sensors, and metal oxide semiconductor hydrogen sulfide sensors; they are installed in the airflow channel at the front of the safety helmet to ensure smooth airflow.

[0035] The video capture module uses a low-power CMOS image sensor (5 megapixels) with a horizontal field of view of 120°, and is mounted on the front of the safety helmet. It is used to compress recorded video clips using H.264 hardware encoding to generate a video stream.

[0036] The positioning module employs a multi-mode Global Navigation Satellite System (GNSS, including GPS and BeiDou), supplemented by inertial navigation DR (dead reckoning) for use in tunnels or indoor areas without signal. The positioning data update frequency is 1Hz.

[0037] A conditioning circuit is provided between the human body status sensor and the edge computing and fusion processing unit; The output of the photoplethysmography (PPG) sensor is connected to a transimpedance amplifier (TIA) to convert it into a voltage signal. Then, it is filtered by a bandpass filter (0.5Hz~5Hz) to remove baseline drift and high-frequency noise, and finally sent to an analog-to-digital converter (ADC). The infrared temperature sensor outputs a tiny voltage signal, which is amplified 50 times by an instrumentation amplifier (INA) before entering the analog-to-digital converter. The triaxial accelerometer and gyroscope are directly connected to the edge computing and fusion processing unit via the I²C bus.

[0038] A conditioning circuit is also provided between the hazardous gas sensor and the edge computing and fusion processing unit; The electrochemical carbon monoxide sensor outputs a weak current signal (nA~μA level), which is converted into a voltage by a zero-bias current-to-voltage conversion circuit (TIA, feedback resistor 1MΩ), and then sent to the analog-to-digital converter through a second-order low-pass filter (cutoff frequency 10Hz).

[0039] The catalytic combustion sensor measures the resistance change via a Wheatstone bridge, and the differential output is amplified by an instrumentation amplifier (INA128) before being sent to an analog-to-digital converter.

[0040] All ADC outputs are connected to the edge computing unit via the SPI bus.

[0041] The data processing flow of the edge computing and fusion processing unit for human state sensors includes: S101: Acquire PPG signal at a sampling rate of 250Hz, calculate instantaneous heart rate by moving average filtering and peak detection, and take the average HR every 5 seconds.

[0042] S102. Simultaneously sample infrared temperature values ​​and obtain body temperature (Temp) through an ambient temperature compensation algorithm (lookup table correction).

[0043] S103 and IMU data are sampled at 100Hz. The fall detection algorithm calculates the synthetic acceleration vector length (SVM). If the SVM exceeds 2.5g for 0.5 seconds and the rate of change is less than 0.1g in the following second, it is determined to be a fall posture.

[0044] S104, the edge computing unit compares HR, Temp, and fall flags with pre-stored user-personalized thresholds.

[0045] Heart rate threshold: lower limit 50 beats / min, upper limit 120 beats / min (configurable).

[0046] Body temperature thresholds: lower limit 35.5℃, upper limit 38.5℃.

[0047] S105. If any parameter exceeds the limit or the fall flag is true, a human body status warning signal is generated and sent to the alarm execution module.

[0048] The data processing flow of the edge computing and fusion processing unit for hazardous gas sensors includes: S201, the edge computing unit reads the AD value of each gas channel every 200ms.

[0049] S202. Use the sensor's factory calibration curve (quadratic polynomial fitting) to convert the AD value into gas concentration values ​​C_CO, C_CH4, and C_H2S.

[0050] S203. Perform median filtering on 5 consecutive concentration values ​​to remove instantaneous spike noise.

[0051] S204. Compare the filtered concentration with the preset alarm threshold: CO: 24ppm - Level 1 warning; 50ppm - Level 2 alarm. CH4: 1% VOL Level 1 warning, 5% VOL Level 2 alarm. H2S: Level 1 warning at 10ppm, Level 2 alarm at 15ppm; S205. If the concentration of any gas exceeds the first-level warning value, a gas concentration warning signal will be generated; if it exceeds the second-level alarm value, an audible and visual alarm will be triggered immediately, and video recording will be forcibly started.

[0052] The video acquisition module and the positioning module are associated and stored in the local storage module; The acquisition time t_v (accurate to milliseconds) of each video frame, together with the positioning coordinates (longitude Lon, latitude Lat, altitude Alt) and attitude angle (from IMU) at the same time, form a data tuple.<t_v, Lon, Lat, Alt, Roll,Pitch, Yaw> .

[0053] This tuple, along with the corresponding H.264 video frame, is packaged together to form the recording file index header and stored in the local storage module (eMMC flash memory). This allows the system to quickly locate the corresponding position and recording segment based on the timestamp when the background requests playback via the wireless communication module.

[0054] When the multi-source data fusion-based smart safety helmet dynamic monitoring and early warning system loses network access, the entire system operation steps include: S301, The edge computing unit detected a disconnection in the wireless communication link (no heartbeat packet was received from the background for 10 consecutive seconds).

[0055] S302, automatically switch to local independent operation mode, all sensors continue to collect data.

[0056] S303, the local early warning logic for human body status and gas concentration remains active; when an early warning is triggered, the audible and visual alarm and the vibration motor immediately activate.

[0057] S304. Video is continuously recorded and stored in a local circular buffer (capacity 32GB, can be overwritten cyclically).

[0058] S305: Location data is continuously recorded to local flash memory.

[0059] S306. After the network is restored, the system will automatically upload the alarm events, corresponding video clips, and location trajectories during the network outage to the backend.

[0060] The early warning process for human and environmental data collected by multi-source sensor modules by the edge computing and fusion processing unit includes: S401, the edge computing unit receives the following characteristic values ​​in parallel: Human characteristics: Heart rate (HR), body temperature (Temp), fall signs (Fall). Gas characteristics: CO concentration, CH4 concentration, H2S concentration. Environmental motion characteristics: SVM combined acceleration, attitude angle.

[0061] S402. Construct the feature vector; X = [HR, Temp, Fall, C_CO, C_CH4, C_H2S, SVM] S403. Apply multi-source fusion rules based on decision trees; for example: Rule 1: If (C_CO>25ppm and HR>100), it is determined as a "gas toxicity + human stress" combined warning, and the "evacuation" command is output.

[0062] Rule 2: If (Fall = True and C_CO < 10ppm and CH4 < 0.5%VOL), then it is determined as "simple fall" and the "Fall Help" command is output.

[0063] Rule 3: If (Temp>38.5℃ and SVM lasts for 2 minutes <0.2g), it is determined as "heatstroke and exhaustion", and the "medical assistance" command is output.

[0064] Rule 4: If (C_CH4>1.5%VOL and Fall = False), then it is determined as "early stage of gas leak" and "early warning" is output.

[0065] S404, Integrated decision-making output comprehensive risk level (0: normal, 1: attention, 2: warning, 3: alarm).

[0066] S405, Alarm module driven according to hazard level: Level 0: No action. Level 1: Local vibration alert. Level 2: Yellow light flashing + buzzer sounding intermittently. Level 3: Red light flashing + continuous beeping + background forced upload of alarm event.

[0067] Compared with the prior art, the present invention has the following advantages: This invention employs a localized fusion architecture that integrates human physiological sensors, hazardous gas sensors, and video and positioning modules into a single helmet body. Furthermore, it deploys multi-source data fusion early warning rules based on decision trees within the edge computing unit. This overcomes the judgment delays and false alarms caused by the stacking of multiple devices and data silos in existing technologies. Specifically, existing systems cannot provide early warnings when gas concentrations rise but the human body has not yet shown obvious discomfort. However, this invention, by simultaneously analyzing the cross-changes in heart rate, body temperature, and multiple gas concentrations (e.g., triggering a composite evacuation command when CO exceeds 25 ppm and heart rate exceeds 100 beats / min), can issue warnings before environmental risks escalate into personal injury, while avoiding misinterpreting increased heart rate caused by heatstroke as gas poisoning. This fusion decision-making mechanism significantly improves the accuracy and response speed of early warnings, reducing false alarms and missed alarms.

[0068] Meanwhile, because this invention establishes a spatiotemporal correlation storage mechanism for video-positioning-attitude, and automatically binds all early warning events with their corresponding location coordinates and video clips, the backend can directly retrieve the on-site footage and trajectory information at that moment when an early warning is triggered, eliminating the need for manual simultaneous review across multiple separate systems. This significantly shortens emergency response time and reduces the difficulty of accident tracing. Furthermore, by setting up network outage detection and local independent operation modes, even if the wireless communication link is interrupted, the safety helmet can still independently complete all sensor data acquisition, local early warning decision-making, audible and visual vibration alarms, and video storage. Once the network is restored, it automatically retransmits key data, thus ensuring monitoring reliability in network outage environments. The above technical solution also reduces redundant hardware overhead by having multiple sensors share a power supply and communication unit, lowering the overall system cost and power consumption, and extending the effective battery life of the equipment.

[0069] The above provides a detailed description of a dynamic monitoring and early warning system for intelligent safety helmets based on multi-source data fusion. The specific embodiments described are merely for the purpose of helping to understand the method and core ideas of this application. It should be noted that those skilled in the art can make various improvements and modifications to this application without departing from its principles, and these improvements and modifications also fall within the protection scope of the claims of this application.

Claims

1. A smart safety helmet dynamic monitoring and early warning system based on multi-source data fusion, comprising a multi-source sensor module, an edge computing and fusion processing module, a local storage module, an alarm execution module, a wireless communication module, and a power management module; The multi-source sensor module is communicatively connected to the edge computing and fusion processing unit and integrated into the helmet body to collect human and environmental data; The edge computing and fusion processing unit has a built-in multi-source data fusion algorithm and early warning decision logic, which is used to analyze and calculate whether the collected human body and environmental data exceed the alarm threshold. The local storage module is communicatively connected to the edge computing and fusion processing unit to store the collected raw human body and environmental data, the fusion results of human body and environmental data, and video clip data. The alarm execution module is connected to the edge computing and fusion processing unit to provide early warnings for human and environmental data that exceed the alarm threshold. The wireless communication module is connected to the local storage module and uploads the relevant data stored in the local storage module to the management platform through wireless communication technology. The power management module supplies power to all modules in the entire system.

2. The intelligent safety helmet dynamic monitoring and early warning system based on multi-source data fusion according to claim 1, characterized in that, The multi-source sensor module includes a human body status sensor, a hazardous gas sensor, a video acquisition module, and a positioning module; Human body status sensors include photoplethysmography (PPG) sensors, infrared temperature sensors, and triaxial accelerometers and gyroscopes. Hazardous gas sensors include electrochemical carbon monoxide sensors, catalytic combustion methane sensors, and metal oxide semiconductor hydrogen sulfide sensors; The video acquisition module uses a low-power CMOS image sensor to compress video clip data using H.264 hard encoding and generate a video stream; The positioning module uses a multi-mode global navigation satellite system, supplemented by inertial navigation (DR) for use in tunnels or indoor areas without signal.

3. The intelligent safety helmet dynamic monitoring and early warning system based on multi-source data fusion according to claim 2, characterized in that, A conditioning circuit is provided between the human body status sensor and the edge computing and fusion processing unit; The output of the photoplethysmography (PPG) sensor is connected to a transimpedance amplifier to convert it into a voltage signal. Then, a bandpass filter is used to remove baseline drift and high-frequency noise before the signal is sent to an analog-to-digital converter. The infrared temperature sensor outputs a tiny voltage signal, which is amplified 50 times by an instrumentation amplifier before entering the analog-to-digital converter. The triaxial accelerometer and gyroscope are directly connected to the edge computing and fusion processing unit via the I²C bus.

4. The intelligent safety helmet dynamic monitoring and early warning system based on multi-source data fusion according to claim 3, characterized in that, A conditioning circuit is also provided between the hazardous gas sensor and the edge computing and fusion processing unit; The electrochemical carbon monoxide sensor outputs a weak current signal, which is converted into a voltage by a zero-bias current-to-voltage conversion circuit, and then sent to the analog-to-digital converter through a second-order low-pass filter. The catalytic combustion sensor measures the resistance change through a Wheatstone bridge, and the differential output is amplified by an instrumentation amplifier and then sent to an analog-to-digital converter. All analog-to-digital converter outputs are connected to the edge computing unit via the SPI bus.

5. The intelligent safety helmet dynamic monitoring and early warning system based on multi-source data fusion according to claim 4, characterized in that, The data processing flow of the edge computing and fusion processing unit for human state sensors includes: S101. Acquire PPG signal at a sampling rate of 250Hz, calculate instantaneous heart rate by moving average filtering and peak detection, and take the average HR every 5 seconds; S102. Simultaneously sample infrared temperature values ​​and obtain body temperature (Temp) through an ambient temperature compensation algorithm; S103 and IMU data are sampled at 100Hz. The fall detection algorithm calculates the synthetic acceleration vector length (SVM). If the SVM exceeds 2.5g for 0.5 seconds and the rate of change is less than 0.1g in the following second, it is determined to be a fall posture. S104, The edge computing unit compares HR, Temp, and fall flag with pre-stored user-personalized thresholds; S105. If any parameter exceeds the limit or the fall flag is true, a human body status warning signal is generated and sent to the alarm execution module.

6. The intelligent safety helmet dynamic monitoring and early warning system based on multi-source data fusion according to claim 5, characterized in that, The data processing flow of the edge computing and fusion processing unit for hazardous gas sensors includes: S201, the edge computing unit reads the AD value of each gas channel every 200ms; S202. Use the sensor's factory calibration curve to convert the AD value into a gas concentration value; S203. Perform median filtering on 5 consecutive concentration values ​​to remove instantaneous spike noise; S204. Compare the filtered concentration with the preset alarm threshold; S205. If the concentration of any gas exceeds the first-level warning value, a gas concentration warning signal will be generated; if it exceeds the second-level alarm value, an audible and visual alarm will be triggered immediately, and video recording will be forcibly started.

7. The intelligent safety helmet dynamic monitoring and early warning system based on multi-source data fusion according to claim 6, characterized in that, The video acquisition module and the positioning module are associated and stored in the local storage module; The acquisition time of each video frame, along with the positioning coordinates and attitude angle (from the IMU) at the same time, forms a data tuple; This tuple, along with the corresponding H.264 video frame, is packaged together to form the recording file index header and stored in the local storage module.

8. The intelligent safety helmet dynamic monitoring and early warning system based on multi-source data fusion according to claim 7, characterized in that, When the multi-source data fusion-based smart safety helmet dynamic monitoring and early warning system loses network access, the entire system operation steps include: S301, The edge computing unit detected a disconnection in the wireless communication link; S302, Automatically switch to local independent operation mode, all sensors continue to collect data; S303. The local early warning logic for human body status and gas concentration remains active. When the early warning is triggered, the audible and visual alarm and the vibration motor will immediately activate. S304. Video is continuously recorded and stored in a local circular buffer; S305, Location data is continuously recorded to local flash memory; S306. After the network is restored, the system will automatically upload the alarm events, corresponding video clips, and location trajectories during the network outage to the backend.

Images