A power tunnel operation and maintenance personnel safety intelligent early warning method and system

Abstract

The application discloses a kind of electric power tunnel operation and maintenance personnel safety intelligent early warning method and system, it is related to electric power system safety monitoring technical field.The method includes: obtaining the environmental state information and operation and maintenance personnel state information in the same monitoring window;Identify environmental disturbance event and personnel abnormal event, form disturbance event shadow result;Personnel state information is compared with disturbance shadow result and is time dislocation, determines the fitting result;According to the fitting result, peel off the interpretable fluctuation caused by environmental disturbance, obtain personnel independent response result;Finally determine the safety abnormal type and position and output early warning.The system includes information acquisition module, disturbance identification module, fitting analysis module, response stripping module and abnormality judging module.The application is distinguished by time dislocation comparison and fitting degree calculation, effectively distinguishes environmental disturbance and personnel abnormality, peels off the interference of environment to personnel state, greatly reduces false alarm rate, realizes accurate early warning.

Description

Technical Field

[0001] This invention relates to the field of power system safety monitoring technology, and in particular to a method and system for intelligent early warning of safety for power tunnel maintenance personnel. Background Technology

[0002] Power tunnels are an important component of urban power transmission and distribution networks. Their internal environment is complex, containing high-voltage cables and power transmission and distribution equipment, and is prone to environmental disturbances such as temperature rise, partial discharge, and electromagnetic interference during operation. At the same time, the narrow space and limited ventilation within the tunnels may lead to dangerous situations such as the accumulation of toxic gases and increased smoke concentration.

[0003] Currently, safety monitoring of power tunnel maintenance personnel mainly relies on independent environmental monitoring equipment (such as temperature and humidity sensors and gas detectors) and personnel monitoring equipment (such as positioning tags and wristbands). When environmental parameters exceed limits or personnel status becomes abnormal, the system triggers alarms accordingly. However, in actual operation and maintenance, environmental disturbances (such as sudden temperature rises and electromagnetic interference) directly affect personnel's physiological signals (such as heart rate and posture) and behavioral patterns, leading to "pseudo-abnormal" fluctuations in personnel status data. Traditional methods cannot distinguish whether abnormal personnel status is a passive response caused by environmental disturbances or an active abnormality caused by personnel accidents (such as falls or sudden illnesses), thus easily generating numerous false alarms and missed alarms, failing to achieve accurate early warning.

[0004] Therefore, how to isolate the impact of environmental disturbances on personnel status and accurately identify personnel's own safety risks is a technical problem that urgently needs to be solved in the field of power tunnel safety operation and maintenance. Summary of the Invention

[0005] This invention provides a method and system for intelligent early warning of safety for power tunnel maintenance personnel, aiming to solve the problem that existing technologies cannot distinguish between environmental disturbances and personnel abnormalities, resulting in poor early warning accuracy.

[0006] To achieve the above objectives, the present invention provides the following technical solution: On the one hand, the present invention provides a method for intelligent early warning of safety for power tunnel maintenance personnel, comprising the following steps: S1: Acquire environmental status information and maintenance personnel status information of the power tunnel within the same monitoring window. Synchronous data collection ensures that environmental and personnel data can be correlated and analyzed in both time and space.

[0007] S2: Based on the environmental status information and maintenance personnel status information, identify environmental disturbance events and personnel anomaly events, and generate disturbance event snapshots. By extracting the temporal characteristics of various disturbances, a complete record containing event type, start and end times, and affected area is formed, providing a foundation for subsequent correlation analysis.

[0008] S3: Compare the status information of the maintenance personnel with the image captured by the disturbance event using time misalignment to determine the fit between the personnel anomaly and the environmental disturbance event. Since there is a delay in the transmission of environmental disturbances to personnel, time misalignment comparison can accurately determine whether the personnel anomaly is related to the environmental disturbance and calculate the fit.

[0009] S4: Based on the alignment results, the explainable fluctuations caused by environmental disturbances are extracted from the personnel status information to obtain the personnel's independent response results. For highly aligned environmental disturbances, their contribution to the personnel status is deducted to restore the personnel's true status under conditions of no environmental interference.

[0010] S5: Based on the independent response results of the personnel, determine the type and location of the security anomaly of the maintenance personnel, and output intelligent early warning results. Based on the stripped-down independent response, determine whether the personnel themselves have a real security risk, avoiding misjudgments caused by environmental disturbances.

[0011] Furthermore, S1 specifically includes: synchronously collecting environmental status information and maintenance personnel status information of the power tunnel according to a unified monitoring time window; wherein, the environmental status information includes at least one of temperature information, humidity information, toxic gas concentration information, smoke concentration information, voltage information, current information, arc detection information, electromagnetic interference information, and ventilation status information; the maintenance personnel status information includes at least one of personnel positioning information, movement trajectory information, heart rate information, posture information, safety helmet wearing status information, protective equipment wearing status information, and vital sign information; and uniformly marking the collected environmental status information and maintenance personnel status information with time and location.

[0012] Furthermore, S2 specifically includes: extracting temperature change sequences, gas concentration change sequences, current change sequences, arc change sequences, and electromagnetic interference change sequences from environmental state information according to the time sequence within the monitoring window; determining a temperature rise disturbance event when the temperature change value exceeds a preset temperature threshold; determining a gas leak disturbance event when the toxic gas concentration exceeds a preset concentration threshold; determining a power abnormality disturbance event when the current fluctuation value exceeds a preset current fluctuation threshold; determining a partial discharge disturbance event when an arc signal is detected to be persistent; determining an electromagnetic interference event when the electromagnetic interference intensity exceeds a preset electromagnetic threshold; sorting each disturbance event according to its occurrence time, and recording the disturbance event type, start time, end time, duration, affected area, and corresponding monitoring window number to form a disturbance event recording result.

[0013] Furthermore, S3 specifically includes: comparing the status information of maintenance personnel with the image results of disturbance events to calculate the fit degree of disturbance trails and the fit degree of abnormal exit; adding the fit degree of disturbance trails and the fit degree of abnormal exit to obtain a safety fit index; comparing the safety fit index with a preset safety fit threshold, and determining the fit result between personnel abnormalities and environmental disturbance events based on the comparison result.

[0014] Furthermore, the calculation steps for the perturbation wake fit degree include: reading the start time, end time, and environmental delay response value during the duration of any environmental perturbation event; constructing a wake time corridor based on the environmental delay response value; reading the abnormal start time, abnormal peak time, and abnormal end time of a personnel abnormal event; mapping the abnormal start time, abnormal peak time, and abnormal end time to the wake time corridor; calculating the abnormal fitting value based on the mapping result; reading the intersection ratio between the area involved in the environmental perturbation event and the area involved in the personnel abnormality to obtain the area coverage value; and multiplying the abnormal fitting value by the area coverage value to obtain the perturbation wake fit degree.

[0015] Furthermore, the calculation steps for the abnormal exit fit degree include: reading the recovery time of each environmental disturbance event to the normal state and forming a disturbance exit node sequence; reading the abnormal recovery time of personnel corresponding to each environmental disturbance event and forming an abnormal exit node sequence; comparing the recovery order of the disturbance exit node sequence and the abnormal exit node sequence; when the recovery order is consistent, calculating the exit interval fit value; calculating the exit residue suppression value based on the abnormal residual amplitude of personnel and the maximum abnormal amplitude; multiplying the exit sequence closure value, the exit interval fit value, and the exit residue suppression value to obtain the abnormal exit fit degree.

[0016] Furthermore, S4 specifically includes: reading the target bonding source event, target safety bonding index, and abnormal time period from the bonding result; determining the fluctuation segment to be peeled off based on the wake time corridor corresponding to the target bonding source event; reading the baseline value corresponding to the fluctuation segment to be peeled off; deducting the environmental disturbance-explainable fluctuation from the fluctuation amount to be peeled off according to the proportion corresponding to the target safety bonding index; adding the peeled fluctuation amount to the baseline value to obtain the corrected detection value; and splicing the corrected detection value with the retained fluctuation segment to obtain the independent response result of the personnel.

[0017] Furthermore, S5 specifically includes: reading abnormal segments, abnormal locations, and abnormal durations from the individual response results; determining the corresponding location as an abnormal location when the residual amplitude of the abnormal segment is greater than a preset residual threshold, or the duration is greater than a preset duration threshold; determining the safety anomaly type based on the detection type corresponding to the abnormal segment; wherein, when the detection type is posture abnormality, the safety anomaly type is determined as fall risk anomaly; when the detection type is heart rate abnormality, the safety anomaly type is determined as vital sign abnormality; when the detection type is missing safety equipment, the safety anomaly type is determined as protection deficiency anomaly; when the detection type is lingering in a dangerous area, the safety anomaly type is determined as area intrusion anomaly; and generating intelligent early warning results based on the abnormal location, anomaly type, and abnormal duration.

[0018] On the other hand, the present invention provides a smart early warning system for the safety of power tunnel maintenance personnel, used to implement the above method, including: Information acquisition module: used to acquire information on the environmental status of the power tunnel and the status of maintenance personnel.

[0019] Disturbance recognition module: used to identify environmental disturbance events and abnormal personnel events, and generate a record of the disturbance events.

[0020] The alignment analysis module is used to compare the status information of maintenance personnel with the results of disturbance event recordings to determine the alignment result.

[0021] Response stripping module: Used to strip away interpretable fluctuations caused by environmental disturbance events based on the bonding results, and obtain independent response results from personnel.

[0022] Anomaly detection module: used to determine the type and location of safety anomalies based on the independent response results of personnel, and generate intelligent early warning results.

[0023] Furthermore, the information acquisition module includes an environmental sensor unit, a personnel positioning unit, a vital signs acquisition unit, and a video monitoring unit; the disturbance identification module includes an environmental disturbance analysis unit and a personnel behavior analysis unit; the fit analysis module includes a time matching unit and an anomaly correlation analysis unit; and the anomaly judgment module includes a risk level assessment unit and an alarm output unit.

[0024] Compared with the prior art, the beneficial effects of the present invention are as follows: Decoupling environmental and personnel impacts: By comparing time misalignments and calculating fit, it is possible to effectively distinguish whether abnormal personnel status is caused by environmental disturbances or by their own accidents, thus significantly reducing the false alarm rate.

[0025] Improve the targeting of early warnings: After stripping away the explainable fluctuations in the environment, the independent response results obtained by personnel are more likely to reflect the real safety risks, making it easier to accurately locate the type and location of anomalies.

[0026] Dynamic adaptability: By utilizing the wake time corridor and exit node sequence, the delay effect and residual impact of environmental disturbances are considered, adapting to the complex propagation environment inside the tunnel.

[0027] High system integration: It integrates multi-source data from environmental sensors, personnel positioning, vital signs, and video to achieve a closed-loop early warning process from perception to decision-making. Attached Figure Description

[0028] Figure 1 This is a flowchart illustrating a method for testing the performance of armored optical cables based on data co-transmission, provided as an embodiment of the present invention. Detailed Implementation

[0029] To make the objectives, technical solutions, and advantages of this invention clearer, the invention will be further described in detail below with reference to the accompanying drawings and specific embodiments. The following embodiments are only for explaining the invention and do not constitute a limitation on the scope of protection of this invention. Equivalent substitutions or modifications made by those skilled in the art based on their understanding of the inventive concept should all fall within the scope of protection of this invention.

[0030] Example 1: This example provides a method for intelligent early warning of safety for power tunnel maintenance personnel. Please refer to [link / reference]. Figure 1 The method includes the following steps S1 to S5.

[0031] Step S1: Obtain environmental status information and maintenance personnel status information of the power tunnel within the same monitoring window.

[0032] In a 220kV power tunnel, a monitoring window is set up every 200 meters along the tunnel's direction, with each monitoring window lasting 5 minutes. The system collects data synchronously in units of this window.

[0033] The environmental status information is collected as follows: thermocouple temperature sensors are deployed on the tunnel sidewalls and cable joints to collect ambient temperature and cable sheath temperature; capacitive humidity sensors are deployed to collect relative humidity; electrochemical gas sensors are deployed to detect the concentration of toxic, harmful, and combustible gases such as carbon monoxide, hydrogen sulfide, and methane; ionization smoke sensors are deployed to detect smoke concentration; Hall current transformers and Rogowski coils are deployed to collect cable current and transient waveforms; ultra-high frequency partial discharge sensors are deployed to detect partial discharge pulse signals; electromagnetic field probes are deployed to detect power frequency magnetic fields and high-frequency electromagnetic interference intensity; and wind pressure and air volume sensors are deployed to monitor the operating status of the ventilation system. The sampling frequency of all sensors is uniformly set to 1 Hz, and the time synchronization accuracy is controlled within 10 milliseconds.

[0034] The status information of maintenance personnel is collected in the following ways: Before entering the tunnel, maintenance personnel wear UWB positioning tags, which work in conjunction with positioning base stations deployed every 50 meters in the tunnel to achieve centimeter-level real-time positioning and record movement trajectories; they wear photoelectric heart rate wristbands with a sampling frequency of 2 Hz to obtain heart rate and heart rate variability parameters in real time; they wear nine-axis inertial measurement units to obtain attitude data such as pitch angle, roll angle, and yaw angle through attitude calculation, and combine the acceleration modulus to determine behaviors such as falls, crouching, and running; capacitive proximity switches and infrared photodiodes are installed inside safety helmets to determine whether the safety helmet is worn correctly; RFID sensing tags are affixed to protective equipment, and the system confirms whether the equipment is complete by reading the tag status; optional vital sign comprehensive acquisition patches are available to obtain parameters such as body temperature and blood oxygen saturation.

[0035] All collected environmental and personnel information is tagged with a unified timestamp and location label and stored in a real-time database.

[0036] Step S2: Identify environmental disturbance events and personnel abnormal events based on the environmental status information and the operation and maintenance personnel status information, and generate a disturbance event recording result.

[0037] The system processes data item by item within the current monitoring window every 5 minutes.

[0038] First, environmental disturbance events are identified. The system extracts temperature change sequences, gas concentration change sequences, current change sequences, arc detection sequences, and electromagnetic interference sequences in chronological order. For temperature data, if the temperature rise rate at three consecutive sampling points exceeds 5 degrees Celsius per second and the absolute temperature value exceeds 55 degrees Celsius, it is identified as a "temperature rise disturbance event"; if the temperature rises slowly above the threshold but at a low rate, it is classified as a "heat accumulation disturbance event." For toxic gases, if the concentration exceeds the first-level warning threshold and lasts for more than 2 seconds, it is identified as a "gas leak disturbance event"; if it exceeds the second-level warning threshold, it is identified as a "serious gas leak disturbance event." For current data, if the effective value of the current changes abruptly by more than 30% of the rated current within 0.1 seconds, it is identified as a "power abnormality disturbance event." For arc detection signals, when the UHF sensor detects a recurring pulse sequence with a pulse repetition frequency between 50 kHz and 1 MHz and an amplitude exceeding -65 dB / mW, it is identified as a "partial discharge disturbance event," which can be further distinguished as corona discharge, internal discharge, or surface discharge based on the phase resolution mode. For electromagnetic interference data, when the power frequency magnetic field measured by the broadband electromagnetic field probe exceeds 100 microtesla, or the high frequency electric field exceeds 10 volts per meter, it is determined to be an "electromagnetic interference event".

[0039] For each disturbance event identified, the system automatically generates a complete record containing the following fields: unique event identifier, event type, start time, end time, duration, affected area, and corresponding monitoring window number. All disturbance events are arranged in chronological order of their start times, forming a "Disturbance Event Record Result".

[0040] Simultaneously, the system analyzes personnel status information in parallel to identify abnormal events. For location and trajectory data, if a person's location enters a prohibited area marked by an electronic fence (such as inside a cable warehouse or a high-voltage test area), or if there is no change in movement for 30 consecutive seconds and they are not at a designated rest point, it is judged as an "abnormal area lingering" event; if the movement trajectory deviates from the predetermined inspection route by more than 5 meters, it is judged as a "abnormal path deviation" event. For heart rate data, upper and lower limits are set based on the person's age and baseline heart rate. For example, if an adult male's heart rate exceeds 120 beats per minute or is lower than 50 beats per minute for more than 10 seconds, it is judged as an "abnormal heart rate event". For posture data, if the absolute value of the pitch angle is greater than 60 degrees and the absolute value of the roll angle is greater than 30 degrees, and the acceleration modulus shows a momentary spike (reaching more than 3 times the acceleration due to gravity), it is judged as a "abnormal fall event"; if the person's posture remains horizontal for a long time (more than 30 seconds) without significant movement, it is judged as an "abnormal fainting event". For equipment wearing data, if the signal of not wearing a safety helmet lasts for more than 10 seconds, or the signal of missing protective equipment lasts for more than 30 seconds, it is judged as an "abnormal event of lack of protection". Each personnel abnormal event is also recorded, including event identifier, personnel identifier, event type, start time, peak time, end time, and location coordinates.

[0041] Step S3: Compare the status information of the maintenance personnel with the image of the disturbance event to determine the time misalignment and the fit between the personnel anomaly and the environmental disturbance event.

[0042] Because there is a physical delay in the propagation of environmental disturbances inside the tunnel, and there is also a lag in the physiological response of personnel, this step adopts the "time-displacement comparison" method, rather than a simple simultaneous comparison.

[0043] First, the system calculates the whetstone fit. For each potentially associated environmental disturbance event and personnel anomaly event, the system performs the following operations: reads the start and end times of the environmental disturbance event, as well as the corresponding environmental delay response value. The environmental delay response value is pre-calibrated based on the characteristics of the propagation medium—for thermal disturbances, the delay value equals the distance between the person and the heat source divided by the heat diffusion rate; for gas leaks, the delay value equals the distance divided by the wind speed; for electromagnetic or partial discharge disturbances, the delay value is approximately zero. Then, the system constructs a "whetstone time corridor," a time interval whose start time is the environmental disturbance start time plus the minimum possible delay, and whose end time is the environmental disturbance end time plus the maximum possible delay. The minimum and maximum delays are set based on factors such as airflow direction and obstacles within the tunnel; for example, the minimum delay is 0.5 times the nominal delay, and the maximum delay is 2 times the nominal delay. This whetstone time corridor represents the time window within which environmental disturbances may trigger physiological or behavioral abnormalities in personnel. Next, the system reads the abnormal start time, abnormal peak time, and abnormal end time of the personnel anomaly event and determines whether these three times fall within the whetstone time corridor. The system calculates the anomalous embedding value based on the number of occurrences: the highest embedding value is achieved if all three occurrences fall within the corridor; a medium embedding value is achieved if two occur within the corridor; a low embedding value is achieved if one occurs within the corridor; and a zero embedding value is achieved if none occur. Simultaneously, the system considers the temporal proximity to the corridor center at each moment—the closer to the center, the higher the embedding value. Next, the system calculates the area coverage value: the proportion of intersection between the affected area of ​​the environmental disturbance event and the location of the anomalous personnel event. If the personnel location is entirely within the affected area, the area coverage value is 1.0; if the personnel location is outside the area boundary but the distance to the boundary is less than the positioning error, the coverage value is calculated based on distance attenuation. Finally, the anomalous embedding value is multiplied by the area coverage value to obtain the disturbance wake fit.

[0044] Next, the abnormal exit fit is calculated. After both environmental disturbance events and personnel abnormal events enter the recovery phase, the system compares their exit processes. The specific steps are as follows: The recovery times of each environmental disturbance event to normal status are read and arranged chronologically to obtain a disturbance exit node sequence; the corresponding personnel abnormality recovery times (i.e., the times when personnel parameters fall back to within ±10% of the baseline) are read to form an abnormal exit node sequence. The system compares the recovery order of the two sequences: if the order is consistent, the exit sequence closure value is set to 1.0; otherwise, it is set to 0.5. Then, the exit interval fit value is calculated: for each pair of corresponding recovery times, the time difference is calculated; the smaller the time difference, the higher the fit value; the larger the time difference, the lower the fit value. If there are multiple pairs of events, the geometric mean of the fit values ​​for each pair is taken. Finally, the exit residual inhibition value is calculated: the residual amplitude and the maximum abnormal amplitude after the personnel abnormal event ends are read; the residual inhibition value is equal to 1 minus the ratio of the residual amplitude to the maximum abnormal amplitude. If the residual amplitude does not exceed the baseline, the residual inhibition value is set to 1.0. Finally, the abnormal exit fit is obtained by multiplying the exit sequence closure value, the exit interval fit value, and the exit residue inhibition value.

[0045] The system then adds the fit of the disturbance trail to the fit of the abnormal exit to obtain the safety fit index. The safety fit index ranges from 0 to 2. The system presets a safety fit threshold. If the safety fit index is greater than or equal to the threshold, it is judged as "high fit," meaning that the abnormal personnel event is highly likely to be caused by the environmental disturbance event; if the safety fit index is between 0.8 and the threshold, it is judged as "partial fit," meaning that the environmental disturbance partially contributes to the abnormal personnel event but other factors may exist; if the safety fit index is less than 0.8, it is judged as "no fit," meaning that the abnormal personnel event is independent of the environmental disturbance event.

[0046] When multiple environmental disturbances occur simultaneously, personnel anomalies may be affected by a combination of factors. In this case, the system calculates the fit index between each pair of disturbances and anomalies, and then uses a fuzzy fusion rule: the disturbance corresponding to the largest fit index is taken as the dominant fit source, and all fit indices are used as weights to proportionally allocate the explainable volatility in subsequent stripping steps.

[0047] Step S4: Based on the fitting results, extract the explainable fluctuations caused by environmental disturbance events from the status information of maintenance personnel to obtain the independent response results of personnel.

[0048] The goal of this step is to subtract the fluctuation components caused by environmental disturbances from the raw personnel signals, thus restoring the personnel's true state under conditions of no environmental interference.

[0049] First, the fluctuation section to be stripped is determined. The system reads the bonding results obtained in step S3, including the target bonding source event, the safe bonding index, and the abnormal period. Based on the wake time corridor corresponding to the target bonding source event, the fluctuation section to be stripped is determined to be the intersection of the abnormal period and the wake time corridor.

[0050] Then, the baseline and fluctuation amount are determined. The mean value of the personnel's state parameters within a stable window prior to the abnormal period is taken as the baseline value. For example, the baseline heart rate is 75 beats per minute. The raw fluctuation amount is obtained by subtracting the baseline value from the value of the raw signal at a certain moment.

[0051] Next, the explainable volatility of the environmental disturbance is estimated. Based on the fit index and disturbance intensity, the explainable volatility contributed by the environment is estimated. Specifically, if historical calibration data is pre-stored in the system, the transfer function can be obtained by looking up a table, and the explainable volatility can be calculated based on the current environmental parameter values, personnel location, and delay time. If no precise model is available, a linear scaling method can be used: the explainable volatility equals an empirical sensitivity coefficient multiplied by (the ratio of the fit index to the maximum possible fit index) and then multiplied by the environmental parameter deviation (the normalized result of the difference between the current environmental parameter value and the threshold). Another commonly used method is to use dynamic time warping to align the waveform, then take the correlation coefficient between the shape of the personnel volatility and the disturbance intensity curve during the disturbance's influence period, and multiply it by the volatility amplitude to obtain the explainable volatility.

[0052] Then, the stripping operation is performed. For each moment within the fluctuation range to be stripped, the corrected detection value is equal to the original signal value minus the amount of fluctuation explained by environmental disturbances. If the corrected detection value is below the baseline value or exceeds the reasonable physiological range, it is clamped to the baseline value or the physiological limit value.

[0053] Finally, signal splicing is performed. The corrected detection values ​​are spliced ​​with the original signals from the unaffected segments in chronological order to form a complete time series of independent personnel responses. For discrete state variables such as posture and equipment wearing, the stripping operation is as follows: if the anomaly is entirely explained by environmental disturbances, the anomaly state of that time period is marked as "environmentally induced" and not output as an independent personnel anomaly; otherwise, the original anomaly state is retained.

[0054] Step S5: Determine the type and location of the security anomaly for the operation and maintenance personnel based on the independent response results of the personnel, and output the intelligent early warning result.

[0055] First, extract anomalous segments. Analyze the individual's independent response results to identify segments that continuously exceed limits or exhibit abnormal states. For example, an independent heart rate response that consistently exceeds 120 beats per minute for more than 10 seconds, or a fall in the independent posture response that is not explained by environmental disturbances.

[0056] Then, anomaly detection is performed. The residual amplitude and duration of each anomalous segment are read. The residual amplitude is compared with a preset residual threshold, and the duration is compared with a preset duration threshold. Only when both the residual amplitude and duration are greater than the preset residual threshold are they considered valid anomalies.

[0057] Next, the anomaly type is determined. Based on the original detection type corresponding to the anomaly segment, it is mapped to the final safety anomaly type: if the original detection type is posture abnormality, it is mapped to "fall risk anomaly"; if the original detection type is heart rate abnormality, it is mapped to "vital signs abnormality"; if the original detection type is missing a safety helmet or protective equipment, it is mapped to "protection deficiency anomaly"; if the original detection type is lingering in a dangerous area or deviating from the path, it is mapped to "area intrusion anomaly"; if the original detection type is prolonged stillness without movement, it is mapped to "personnel missing or fainting anomaly"; if the original detection type is loss of location tag signal, it is mapped to "communication or location anomaly".

[0058] Then, locate the anomaly. Based on the personnel location information corresponding to the anomaly segments in the independent response results, accurately pinpoint the location of the anomaly. If the location error is large, it can be confirmed by combining video surveillance footage.

[0059] Finally, intelligent early warning results are generated and output. The early warning results include the following fields: warning level (divided into three levels: "Attention," "Warning," and "Emergency" based on the anomaly type and residual amplitude; for example, a residual heart rate exceeding 30 beats per minute for more than 30 seconds qualifies as "Emergency"); anomaly location (e.g., fire compartment C, at kilometer 1.2); anomaly type (e.g., abnormal vital signs—tachycardia); anomaly description (e.g., an individual's independent heart rate reaches 135 beats per minute for 18 seconds, environmental disturbances have been eliminated, indicating physical discomfort); recommended measures (e.g., immediately notify the individual to stop and rest via tunnel broadcast, and dispatch a rescue team); and timestamp. Early warning results are output through the following channels: tunnel audible and visual alarms, vibration and voice prompts on maintenance personnel's wristbands, pop-ups and SMS push notifications on the management platform, and on-site large screen displays.

[0060] Complete Application Example: Suppose that a partial discharge disturbance event occurs in tunnel B area within a certain monitoring window (start time 10:00:30, end time 10:01:45, delay response value 8 seconds). Maintenance personnel Zhang San is located in the center of B area. His original heart rate signal begins to rise at 10:00:38, peaks at 125 beats per minute at 10:01:20, and ends at 10:02:00. Calculation in step S3 shows a disturbance wake fit of 0.92, an abnormal exit fit of 0.71, and a safe fit index of 1.63, which is greater than the threshold of 1.2, thus indicating a high fit. Step S4 performs stripping: the baseline heart rate is 75 beats per minute. After deducting the environmental contribution, the corrected peak heart rate is 85 beats per minute, and the abnormal duration is shortened. Step S5 found that the residual amplitude was only 10 times per minute, which is less than the threshold of 20 times per minute, and the duration was 30 seconds, which is less than the threshold of 60 seconds. Therefore, it was determined that there was no independent anomaly, and only an environmental risk warning was output: "Partial discharge has occurred in fire compartment B. Please take protective measures." If another person, Li Si, experiences a heart rate spike to 140 beats per minute for 40 seconds during the same period without environmental disturbance, it is still considered a significant anomaly after separation, and the system outputs: "Emergency: Abnormal vital signs, location B, immediate rescue."

[0061] Example 2 provides a smart early warning system for the safety of power tunnel maintenance personnel, used to implement the method described in Example 1. The system includes the following modules.

[0062] Information Acquisition Module: Used to acquire environmental status information and maintenance personnel status information of the power tunnel. This module further comprises four units. The environmental sensor unit consists of a distributed sensor network, including a temperature and humidity sensor array, multi-gas sensors (electrochemical or infrared principles), current and voltage transformers, high-frequency current sensors, ultra-high frequency antennas, electromagnetic interference receivers, and anemometers. Each sensor connects to the edge gateway via RS485 or Modbus protocols, or wirelessly via ZigBee. The personnel positioning unit uses UWB positioning technology, deploying one positioning base station every 50 meters. The time synchronization accuracy between base stations reaches 0.1 nanoseconds, achieving centimeter-level real-time positioning in conjunction with positioning tags worn by personnel; optional backup solutions include Bluetooth AoA positioning or video plus AI positioning. The vital signs acquisition unit uses medical-grade wearable devices (such as customized smart bracelets) to upload data such as heart rate, blood oxygen, body temperature, and respiratory rate via low-power Bluetooth. The device has IP68 protection rating and intrinsically safe explosion-proof certification. The video surveillance unit uses explosion-proof infrared cameras with a resolution of 1080P and supports the ONVIF protocol. One camera is deployed every 100 meters at the top of the tunnel to assist in verifying personnel posture and behavior.

[0063] Disturbance Identification Module: Deployed on the edge computing gateway or tunnel monitoring center server. This module comprises two units. The environmental disturbance analysis unit analyzes the environmental data stream in real time, using a sliding window and adaptive threshold algorithm to identify disturbance events. It has a built-in rule engine and a machine learning classifier to further subdivide partial discharge types. The personnel behavior analysis unit obtains Euler angles based on posture calculation and uses a state machine model to identify behaviors such as falls, squatting, and running; abnormal heart rate detection uses an adaptive threshold; and equipment detection is achieved by reading changes in the status of RFID tags.

[0064] The fit analysis module comprises two units. The time matching unit maintains a global timeline, providing PTP-based time synchronization services; pre-calculates propagation delay tables for each sensor node based on the tunnel airflow and thermal diffusion models; implements time misalignment alignment; and constructs a data structure for the wake time corridor. The anomaly correlation analysis unit implements the fit calculation algorithm, maintains an event correlation graph, and outputs the correlation results.

[0065] The response stripping module comprises two units. The fluctuation decomposition unit implements signal decomposition algorithms, allowing selection of simple linear stripping, Kalman filter-based estimation, or independent component analysis to separate environmental interference components from mixed signals. Different transfer function model libraries are built-in for different types of personnel parameters. The signal reconstruction unit superimposes the stripped fluctuation components onto the baseline and smooths the boundaries to generate complete independent response curves for each person.

[0066] The anomaly detection module comprises two units. The risk level assessment unit dynamically calculates the risk level based on a predefined rule table, combined with the residual amplitude and duration of the anomaly after stripping, personnel historical health records, and current work content. An alert is triggered when the score exceeds 60 points. The alarm output unit distributes the alert message to multiple output channels via message middleware: tunnel site audible and visual alarms, maintenance personnel wristbands, monitoring center screens, mobile applications, and third-party emergency response systems. Simultaneously, alert logs are recorded on a blockchain ledger for traceability.

[0067] System workflow summary: The information acquisition module collects data every 50 milliseconds; the disturbance identification module processes data every 1 second, identifies disturbance events and pushes them to the event pool of the fitting analysis module; the fitting analysis module processes data in batches every 5 minutes, matches environmental disturbances with personnel anomalies, and calculates the fitting index; the response stripping module corrects the original personnel signals based on the fitting results; and the anomaly judgment module makes the final decision and outputs an early warning.

[0068] The above are merely preferred embodiments of the present invention and are not intended to limit the present invention in any way. Although the present invention has been disclosed above with reference to preferred embodiments, it is not intended to limit the present invention. Any person skilled in the art can make some modifications or alterations to the above-disclosed technical content to create equivalent embodiments without departing from the scope of the present invention. Any simple modifications, equivalent changes and alterations made to the above embodiments based on the technical essence of the present invention without departing from the scope of the present invention should still fall within the scope of the claims of the present invention.

Claims

1. A method for intelligent early warning of safety for power tunnel maintenance personnel, characterized in that, Includes the following steps: S1: Obtain environmental status information and maintenance personnel status information of the power tunnel within the same monitoring window; S2: Identify environmental disturbance events and personnel anomaly events based on the environmental status information and maintenance personnel status information, and generate disturbance event recording results; S3: Compare the status information of the maintenance personnel with the image of the disturbance event to determine the time misalignment and the fit between the personnel anomaly and the environmental disturbance event; S4: Based on the fitting results, extract the explainable fluctuations caused by environmental disturbance events from the status information of maintenance personnel to obtain the independent response results of personnel; S5: Determine the type and location of the security anomaly for the operation and maintenance personnel based on the independent response results of the personnel, and output the intelligent early warning results.

2. The intelligent early warning method for the safety of power tunnel maintenance personnel according to claim 1, characterized in that, According to claim 1, the intelligent early warning method for the safety of power tunnel maintenance personnel is characterized in that step S1 includes: Information on the environmental status of power tunnels and the status of maintenance personnel is collected synchronously within a unified monitoring time window. The environmental status information includes at least one of the following: temperature information, humidity information, toxic gas concentration information, smoke concentration information, voltage information, current information, arc detection information, electromagnetic interference information, and ventilation status information. The status information of the maintenance personnel includes at least one of the following: personnel location information, movement trajectory information, heart rate information, posture information, helmet wearing status information, protective equipment wearing status information, and vital sign information; The collected environmental status information and maintenance personnel status information are uniformly time-stamped and location-stamped.

3. The intelligent early warning method for the safety of power tunnel maintenance personnel according to claim 1, characterized in that, S2 includes: According to the time sequence within the monitoring window, extract the temperature change sequence, gas concentration change sequence, current change sequence, electric arc change sequence, and electromagnetic interference change sequence from the environmental status information; When the temperature change exceeds a preset temperature threshold, a temperature rise disturbance event is identified. When the concentration of toxic gas exceeds a preset concentration threshold, a gas leak disturbance event is identified. When the current fluctuation value exceeds the preset current fluctuation threshold, an abnormal power disturbance event is determined. When a persistent arc signal is detected, a partial discharge disturbance event is identified. When the intensity of electromagnetic interference exceeds a preset electromagnetic threshold, an electromagnetic interference event is identified. The disturbance events are sorted according to their occurrence time, and the event type, start time, end time, duration, affected area, and corresponding monitoring window number are recorded to form a disturbance event record.

4. The intelligent early warning method for the safety of power tunnel maintenance personnel according to claim 3, characterized in that, S3 includes: The system compares the status information of maintenance personnel with the results of disturbance event recordings to calculate the fit of disturbance trails and the fit of abnormal exits. The safety fit index is obtained by adding the perturbation wake fit and the abnormal exit fit. The safety fit index is compared with a preset safety fit threshold, and the fit result between personnel abnormality and environmental disturbance event is determined based on the comparison result.

5. The intelligent early warning method for the safety of power tunnel maintenance personnel according to claim 4, characterized in that, The calculation steps for the perturbation wake fit include: Read the start time, end time, and environmental delay response value during the duration of any environmental disturbance event; Construct a wake time corridor based on the environmental delay response value; Read the abnormal start time, abnormal peak time, and abnormal end time of abnormal personnel events; Map the anomaly start time, anomaly peak time, and anomaly end time to the wake time corridor; Calculate the abnormal chimerism value based on the mapping results; The area coverage value is obtained by reading the intersection ratio between the area affected by environmental disturbance events and the area affected by personnel anomalies. Multiplying the anomalous embedding value by the region coverage value yields the perturbation trail fitting degree.

6. The intelligent early warning method for the safety of power tunnel maintenance personnel according to claim 4, characterized in that, The calculation steps for the abnormal withdrawal fit include: Read the recovery time of each environmental disturbance event to the normal state and form a disturbance exit node sequence; Read the abnormal recovery time of personnel corresponding to each environmental disturbance event and form an abnormal exit node sequence; Compare the recovery order of the perturbed exit node sequence and the abnormal exit node sequence; When the recovery order is consistent, calculate the exit interval fitting value; The exit residue suppression value is calculated based on the abnormal residual amplitude and the maximum abnormal amplitude of personnel. The abnormal exit fit is obtained by multiplying the exit sequence closure value, the exit interval fit value, and the exit residual inhibition value.

7. The intelligent early warning method for the safety of power tunnel maintenance personnel according to claim 1, characterized in that, According to claim 1, the intelligent early warning method for the safety of power tunnel maintenance personnel is characterized in that step S4 includes: Read the target bonding source event, target safety bonding index, and abnormal time period from the bonding results; The fluctuation segment to be stripped is determined based on the tail time corridor corresponding to the source event of the target alignment. Read the baseline value corresponding to the fluctuation segment to be stripped; According to the proportion corresponding to the target safety fit index, the volatility that can be explained by environmental disturbances is deducted from the volatility to be stripped. The fluctuation value after stripping is added to the baseline value to obtain the corrected detection value; The corrected detection value is then combined with the retained fluctuation range to obtain the individual response results.

8. The intelligent early warning method for the safety of power tunnel maintenance personnel according to claim 1, characterized in that, S5 includes: Read out abnormal segments, abnormal locations, and abnormal durations from individual responses; When the residual amplitude of an abnormal segment is greater than a preset residual threshold, or the duration is greater than a preset duration threshold, the corresponding position is determined as an abnormal position. Determine the type of security anomaly based on the detection type corresponding to the abnormal segment; Among them, when the detection type is abnormal posture, the safety abnormality type is determined to be abnormal fall risk; When the detection type is abnormal heart rate, the safety abnormality type is determined as abnormal vital signs; When the detection type is "safety equipment missing", the safety anomaly type is determined to be "protection missing anomaly". When the detection type is "retention in dangerous areas", the security anomaly type is determined to be "area intrusion anomaly". Intelligent early warning results are generated based on the location, type, and duration of the anomaly.

9. A smart early warning system for the safety of power tunnel maintenance personnel, characterized in that, To implement the method according to any one of claims 1-8, comprising: Information acquisition module: used to acquire environmental status information and maintenance personnel status information of the power tunnel; Disturbance recognition module: used to identify environmental disturbance events and abnormal personnel events, and generate a record of the disturbance events; The alignment analysis module is used to compare the status information of maintenance personnel with the results of disturbance event recordings to determine the alignment result by comparing the time misalignment. Response stripping module: used to strip away interpretable fluctuations caused by environmental disturbance events based on the bonding results, and obtain independent response results from personnel; Anomaly detection module: used to determine the type and location of safety anomalies based on the independent response results of personnel, and generate intelligent early warning results.

10. A smart early warning system for the safety of power tunnel maintenance personnel according to claim 9, characterized in that, The information acquisition module includes an environmental sensor unit, a personnel positioning unit, a vital signs acquisition unit, and a video monitoring unit. The disturbance identification module includes an environmental disturbance analysis unit and a personnel behavior analysis unit; The bonding analysis module includes a time matching unit and an anomaly correlation analysis unit; The anomaly detection module includes a risk level assessment unit and an alarm output unit.

Images