Electric shock prevention grading early warning method and system based on artificial intelligence, and medium

Abstract

The present application relates to the technical field of work safety in the power scene, specifically is the power anti-electric shock grading early warning method, system and medium based on artificial intelligence, in the method part through setting 5 grades of voltage early warning position and matching corresponding reference electric field threshold value, can fully adapt to 380V to 230kV full voltage grade power operation scene, solve the industry problem that traditional single threshold early warning cannot adapt to different voltage grades, early warning distance is inaccurate, match the safety protection distance of each voltage grade, give consideration to power operation efficiency and personnel safety guarantee, in the system part through adding first battery, second battery double power supply structure, the first battery is powered for the basic module such as main control processing module, timing module, the second battery is powered for reminding warning unit, realize power supply loop completely independent separation from structure, completely avoid the interference of basic module power consumption, fault to warning function, guarantee anti-electric shock warning function continues effectively.

Description

Technical Field

[0001] This invention relates to the field of work safety technology in power scenarios, specifically to a graded early warning method, system, and medium for preventing electric shock based on artificial intelligence. Background Technology

[0002] The demand for safety protection in current power operation scenarios continues to increase. Smart wearable early warning devices for electric shock prevention for power workers have become an industry necessity. However, existing electric shock prevention watches and supporting early warning methods still have many structural and methodological defects. Most electric shock prevention early warning methods use a single fixed electric field threshold to directly determine the alarm, without setting appropriate early warning levels for different voltage levels such as 380V, 10kV, 35kV, 132kV, and 230kV. They also cannot match the safety protection distance corresponding to each voltage level, which can easily lead to problems such as the early warning distance being too far and interfering with normal operation, or the early warning distance being too short and failing to achieve the safety protection effect.

[0003] Existing early warning methods do not utilize electric field harmonic characteristics to distinguish interference, lack the intelligent correction capabilities of artificial intelligence deep learning, and rely heavily on direct computation by end-side hardware. Limited by insufficient end-side computing power, signal processing accuracy is low, making it impossible to effectively distinguish between real high-voltage electric field signals and environmental stray interference signals. False alarm and false alarm rates remain high, and scientific hierarchical early warning cannot be achieved. Overall, the accuracy, reliability, and intelligence of early warning are low, making it difficult to meet the precise, intelligent, and highly reliable requirements for power operation safety protection.

[0004] Meanwhile, wearable early warning devices generally adopt a single battery compartment with a single power supply structure. One battery powers all components such as timing, touch, display, communication, and alarm. When the single power supply structure is directly applied to electric shock prevention safety watches for power operations, there are unavoidable structural defects. It cannot safely adapt to the mandatory safety requirements of high-risk scenarios such as power operations, live-line inspections, and complex outdoor working conditions.

[0005] On the one hand, a single power supply provides a shared power circuit for all modules. When basic modules such as timing, touch control, positioning, and communication experience power consumption fluctuations, short circuits, overloads, or depletion of power, they will directly interrupt the power supply to the alarm module, causing multiple alarm functions such as buzzers, lights, and vibrations to completely fail. The alarm function of the electric shock-proof watch is a core life-saving function; if it fails, workers will be unable to perceive the danger of the high-voltage electric field, which can easily lead to electric shock accidents.

[0006] On the other hand, the single power supply circuit has no physical isolation, and circuit faults and electromagnetic interference in the basic module will be directly transmitted to the alarm module, causing alarm disorder and failure to operate; and there is no backup power supply structure, so once the main power supply is abnormal, the safety protection function of the whole machine will be completely paralyzed. Summary of the Invention

[0007] The purpose of this invention is to provide a graded early warning method, system, and medium for preventing electric shock based on artificial intelligence, addressing the aforementioned problems.

[0008] The technical solution adopted in this invention is as follows: a graded early warning method for electric shock prevention based on artificial intelligence, including the following steps: voltage warning level setting, the operator wears an early warning device to set the warning level according to the voltage level of the power operation scenario, and generates a matching early warning benchmark threshold E_base based on the warning level;

[0009] The electric field parameters in the area where the workers are located are collected and preprocessed. The electric field parameters in the power operation scenario where the workers are located are obtained, and the signals are amplified, converted from analog to digital and filtered to obtain a clean electric field signal E_filter.

[0010] Cloud data transmission and deep learning computing: The end-side early warning device transmits the clean electric field signal E_filter, the manually selected voltage level, the corresponding reference electric field threshold E_base, and GPS positioning data to the cloud. The cloud completes the interference correction and feature extraction of the electric field signal to obtain the real electric field strength E_real in the area where the operator is located.

[0011] The risk level determination and results are transmitted back down. The cloud-based artificial intelligence model uses the baseline electric field threshold E_base corresponding to the manually selected level as the judgment boundary, and combines the actual electric field strength E_real to divide the risk level. The risk level, the actual electric field strength E_real and the warning instructions are transmitted back down.

[0012] The system implements tiered early warning, whereby the end-side early warning device receives the early warning command and issues tiered warnings to the personnel wearing the device based on the command.

[0013] Furthermore, the voltage warning level settings include 5 voltage levels and corresponding warning reference thresholds E_base, which are as follows:

[0014] Gear 1, warning voltage 380V, warning reference threshold E_base=0.5kV / m;

[0015] Level 2, warning voltage 10kV, warning reference threshold E_base=1.0kV / m;

[0016] Level 3, warning voltage 35kV, warning reference threshold E_base=1.5kV / m;

[0017] Level 4, warning voltage 132kV, warning reference threshold E_base=2.0kV / m;

[0018] Level 5, warning voltage 230kV, warning reference threshold E_base=2.5kV / m.

[0019] Furthermore, in the acquisition and preprocessing of electric field parameters in the area where the workers are located, the planar induction electrode of the early warning device has an electrode area S of 1.2 cm² and an electrode distance d of 0.8 mm. It first captures the 50 Hz power frequency electric field in the space non-contactly through the principle of capacitive coupling, and converts the electric field intensity into the original induced electromotive force E. p This completes the physical conversion from electric field signal to electrical signal;

[0020] Then the original induced potential E p After amplification, the amplified signal E_amp is obtained, which has:

[0021] E_amp=E p ×G;

[0022] Where G is the magnification factor;

[0023] Next, a 12-bit analog-to-digital converter (ADC) circuit is used to process the amplified signal E_amp, converting the continuous analog electrical signal into a discrete digital signal to obtain the original electric field strength value E_raw in digital form.

[0024] Furthermore, in the acquisition and preprocessing of electric field parameters in the area where the workers are located, the raw electric field strength value E_raw in digital form is filtered using a second-order Butterworth low-pass filter algorithm to obtain a clean electric field signal E_filter, which is:

[0025] E_filter=α0•E_raw[i]+α1•E_raw[i-1]+α2•E_raw[i-2]-β1•E_filter[i-1]-β2•E_filter[i-2];

[0026] Where α0, α1, and α2 are the filter input coefficients, β1 and β2 are the filter feedback coefficients, i is the sequence number of the current sampled data, E_raw[i] is the original electric field intensity value collected at the current time, E_raw[i-1] and E_raw[i-2] are the original electric field intensity values ​​collected at the previous time and two time steps ago, respectively, E_filter[i] is the filtered electric field signal at the current time, and E_filter[i-1] and E_filter[i-2] are the filtered electric field signals at the previous time and two time steps ago, respectively.

[0027] Furthermore, in cloud data transmission and deep learning computation, the early warning device first integrates the filtered electric field signal E_filter, the manually selected voltage level, the corresponding reference electric field threshold E_base, GPS positioning data, and the unique identifier ID of the early warning device into a data packet, and then uploads it to the cloud through the communication unit;

[0028] Then, the data packet is retrieved from the cloud, and a multi-dimensional feature vector is extracted. This feature vector includes the filtered electric field signal E_filter, the manually set voltage level, the reference electric field threshold E_base, the 3rd harmonic feature H3, the 5th harmonic feature H5, and the 7th harmonic feature H7. Based on the 3rd, 5th, and 7th harmonic features, the interference confidence level P is calculated.

[0029] P=(|H3|+|H5|+|H7|) / (|H3|+|H5|+|H7|+E_filter);

[0030] Wherein, P ranges from 0 to 1. When P is less than 0.3, it is determined to be a real high-voltage electric field signal, and when P is greater than or equal to 0.3, it is determined to be an environmental stray interference signal.

[0031] Then, the harmonic characteristic factor F is obtained by weighted calculation of harmonic characteristics, which gives:

[0032] F = a·|H3| + b·|H5| + c·|H7|;

[0033] Where a, b, and c are fixed weighting coefficients that match the current voltage level, and the harmonic characteristic factor F is then normalized to obtain the dimensionless harmonic characteristic factor F_norm.

[0034] And based on the interference confidence level P and the dimensionless harmonic characteristic factor F_norm, the environmental interference correction coefficient K is obtained, which is:

[0035] K = 1 - (P × F_norm);

[0036] Used to correct deviations in the filtered electric field signal and eliminate errors caused by non-target electromagnetic interference from metals and cables in the environment, the true electric field strength E_real is calculated using the environmental interference correction coefficient K.

[0037] E_real = E_filter × K.

[0038] Furthermore, in the risk level determination and result downlink transmission, the risk level is divided based on the reference electric field threshold E_base corresponding to the manual selection, combined with the output real electric field strength E_real. These risk levels are classified as safety level, warning level, and danger level. The safety level is defined as the real electric field strength E_real being less than the reference electric field threshold E_base corresponding to the selected level.

[0039] The warning level is defined as follows: the actual electric field strength E_real is greater than or equal to the reference electric field threshold E_base corresponding to the selected range, and less than the sum of the reference electric field threshold E_base corresponding to the selected range and 1.0kV / m.

[0040] The danger level is defined as the sum of the actual electric field strength E_real and the reference electric field threshold E_base corresponding to the selected range, and 1.0 kV / m.

[0041] This application also provides an artificial intelligence-based graded early warning system for preventing electric shock, used to implement an artificial intelligence-based graded early warning method for preventing electric shock, including a cloud and an early warning device, and the cloud and the early warning device are wirelessly connected. The early warning device is an electric shock-proof watch, including a watch body, with both ends of the watch body connected to the watch strap respectively. The watch body is provided with a main control processing module, a timing module, a touch display module, a positioning communication module and an electrostatic induction early warning module. The watch body is also provided with a first battery, a second battery and an alert unit.

[0042] The first battery is used to power the main control processing module, the timing module, the touch display module, the positioning and communication module, and the electrostatic induction warning module;

[0043] The second battery is used to power the alert unit;

[0044] The reminder and warning unit is used to receive signals from the main control processing module and execute reminders and warnings.

[0045] Furthermore, the watch body is provided with a working cavity, in which the first battery, the second battery and the reminder and warning unit are all located. An insulating partition plate is also provided in the working cavity, which divides the working cavity into two independent cavities. The second battery and the reminder and warning unit are located in one independent cavity, and the first battery is located in the other independent cavity.

[0046] Furthermore, the reminder and warning unit includes a buzzer alarm module, a light warning module, and a vibration warning module;

[0047] The buzzer alarm module is used to emit a buzzer alarm sound;

[0048] The light warning module is used to emit a red light alarm;

[0049] The vibration alert module is used to trigger an alarm on the vibrating instrument body.

[0050] When the early warning device receives an early warning command from the cloud with a downlink transmission level of 500Hz, the buzzer alarm module emits a single 500Hz buzzer with a duration of 200ms and a sound intensity of ≥85dB; the light warning module flashes a single 1Hz red light for 200ms; and the vibration reminder module outputs a single 2G amplitude vibration for 200ms.

[0051] When the early warning device receives a warning command from the cloud with a danger level, the buzzer alarm module emits three consecutive 500Hz beeps, with each beep spaced 500ms apart; the light warning module flashes red light continuously at 3Hz; and the vibration reminder module outputs a 2G amplitude continuous vibration, with a single vibration duration of 500ms and a vibration interval of 300ms.

[0052] When the early warning device receives a downlink warning command from the cloud at the security level, the alert unit does not perform any alarm action, and the touch display module only displays the current voltage level and the actual electric field strength E_real calculated by artificial intelligence.

[0053] Furthermore, this application also provides an artificial intelligence-based graded early warning medium for electric shock prevention, including a computer program for storing a computer program that enables a computer to execute an artificial intelligence-based graded early warning method for electric shock prevention.

[0054] The beneficial effects of the present invention include at least one of the following;

[0055] 1. In the method section, by setting 5 voltage warning levels and matching the corresponding reference electric field threshold, it can fully adapt to power operation scenarios of all voltage levels from 380V to 230kV, solving the industry problem that traditional single threshold warning cannot adapt to different voltage levels and the warning distance is inaccurate. It matches the safety protection distance of each voltage level, taking into account both power operation efficiency and personnel safety.

[0056] 2. The electric field signal is acquired through hardware acquisition using planar induction electrodes with specific parameters. Combined with differential amplification, 12-bit ADC analog-to-digital conversion, and second-order Butterworth low-pass filtering, it can accurately acquire clean electric field signals after removing interference, providing stable and reliable raw data for subsequent artificial intelligence calculations. The electric field acquisition and preprocessing are highly accurate and have strong anti-interference capabilities.

[0057] 3. Based on the characteristics of the 3rd, 5th, and 7th harmonics of the electric field, the interference confidence level and harmonic characteristic factor are calculated. After normalization, the dimensionless interference correction coefficient is obtained. The real electric field strength is solved by cloud deep learning, which can effectively distinguish between the real high-voltage electric field and environmental stray interference, and significantly reduce the false alarm rate and missed alarm rate of the warning. The artificial intelligence correction logic is scientific and rigorous, and the edge-cloud collaborative computing perfectly solves the problem of insufficient computing power of the end watch.

[0058] 4. By adding a dual power supply structure of a first battery and a second battery inside the watch body, the first battery is dedicated to powering the main control processing module, timing module, and other basic modules, while the second battery is dedicated to powering the reminder and warning unit. This structurally achieves completely independent power supply circuits, thoroughly avoiding interference from power consumption or malfunctions of the basic modules on the warning function, ensuring the continuous effectiveness of the anti-electric shock warning function. A working cavity is set up inside the watch body, integrating the first battery, the second battery, and the reminder and warning unit within the working cavity, optimizing the internal space layout of the watch body and improving the watch's compactness and wearability. Furthermore, an insulating partition plate divides the working cavity into two independent chambers, allowing the second battery, the reminder and warning unit, and the first battery to be arranged separately, achieving physical and electrical isolation, eliminating electromagnetic interference and structural interference between modules, and improving the stability of device operation.

[0059] 5. The warning unit integrates a three-tiered structure module consisting of a buzzer alarm, a light warning, and a vibration alert. This multi-mode warning structure works in concert to adapt to noisy power operation environments, significantly improving the effectiveness and recognizability of warnings. Attached Figure Description

[0060] Figure 1 A flowchart of an AI-based graded early warning method for preventing electric shock in power systems;

[0061] Figure 2 This is a schematic diagram of an AI-based graded early warning system for preventing electric shock.

[0062] Figure 3 This is a schematic diagram of an early warning device.

[0063] Figure 4 This is a schematic diagram of the side structure of an early warning device;

[0064] Figure 5 This is a schematic diagram of the table's structure;

[0065] Figure 6 This is a schematic diagram of the internal structure of the working chamber;

[0066] Figure 7 This is a schematic diagram of the connection of a warning device module. Detailed Implementation

[0067] To make the objectives, technical solutions, and advantages of the embodiments of the present invention clearer, the technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. The components of the embodiments of the present invention described in the accompanying drawings can generally be arranged and designed in various different configurations.

[0068] Therefore, the following detailed description of the embodiments of the invention provided in the accompanying drawings is not intended to limit the scope of the claimed invention, but merely to illustrate selected embodiments of the invention. All other embodiments obtained by those skilled in the art based on the embodiments of the invention without inventive effort are within the scope of protection of the invention.

[0069] It should be noted that, unless otherwise specified, the embodiments and features described in this invention can be combined with each other.

[0070] It should be noted that similar labels and letters in the following figures indicate similar items. Therefore, once an item is defined in one figure, it does not need to be further defined and explained in subsequent figures.

[0071] like Figure 1 As shown, the AI-based graded early warning method for electric shock prevention in power applications is applicable to power work scenarios and includes the following steps:

[0072] Voltage warning level setting: Operators wear warning devices to set the warning level according to the voltage level of the power operation scenario, and generate a matching warning benchmark threshold E_base based on the warning level;

[0073] The electric field parameters in the area where the workers are located are collected and preprocessed. The electric field parameters in the power operation scenario where the workers are located are obtained, and the signals are amplified, converted from analog to digital and filtered to obtain a clean electric field signal E_filter.

[0074] Cloud data transmission and deep learning computing: The end-side early warning device transmits the clean electric field signal E_filter, the manually selected voltage level, the corresponding reference electric field threshold E_base, and GPS positioning data to the cloud. The cloud completes the interference correction and feature extraction of the electric field signal to obtain the real electric field strength E_real in the area where the operator is located.

[0075] The risk level determination and results are transmitted back down. The cloud-based artificial intelligence model uses the baseline electric field threshold E_base corresponding to the manually selected level as the judgment boundary, and combines the actual electric field strength E_real to divide the risk level. The risk level, the actual electric field strength E_real and the warning instructions are transmitted back down.

[0076] The system implements tiered early warning, whereby the end-side early warning device receives the early warning command and issues tiered warnings to the personnel wearing the device based on the command.

[0077] The purpose of this design is to comprehensively adapt to power operation scenarios of all voltage levels from 380V to 230kV by setting 5 voltage warning levels and matching corresponding reference electric field thresholds in the method part. This solves the industry problem that traditional single threshold warning cannot adapt to different voltage levels and the warning distance is inaccurate. It accurately matches the safety protection distance of each voltage level, taking into account both power operation efficiency and personnel safety.

[0078] Meanwhile, in this embodiment, the voltage warning level settings include 5 voltage levels and corresponding warning reference thresholds E_base, which are as follows:

[0079] Gear 1, warning voltage 380V, warning reference threshold E_base=0.5kV / m;

[0080] Level 2, warning voltage 10kV, warning reference threshold E_base=1.0kV / m;

[0081] Level 3, warning voltage 35kV, warning reference threshold E_base=1.5kV / m;

[0082] Level 4, warning voltage 132kV, warning reference threshold E_base=2.0kV / m;

[0083] Level 5, warning voltage 230kV, warning reference threshold E_base=2.5kV / m.

[0084] Meanwhile, in the acquisition and preprocessing of electric field parameters in the area where the workers are located, the planar induction electrode of the early warning device, with an electrode area S of 1.2 cm² and an electrode spacing d of 0.8 mm, first captures the 50 Hz power frequency electric field in the space non-contactly through the principle of capacitive coupling, and converts the electric field intensity into the original induced electromotive force E. p This completes the physical conversion from electric field signal to electrical signal;

[0085] Then the original induced potential E p After amplification, the amplified signal E_amp is obtained, which has:

[0086] E_amp=E p ×G;

[0087] Where G is the magnification factor;

[0088] Next, a 12-bit analog-to-digital converter (ADC) circuit is used to process the amplified signal E_amp, converting the continuous analog electrical signal into a discrete digital signal to obtain the original electric field strength value E_raw in digital form.

[0089] Furthermore, in the acquisition and preprocessing of electric field parameters in the area where the workers are located, the raw electric field strength value E_raw in digital form is filtered using a second-order Butterworth low-pass filter algorithm to obtain a clean electric field signal E_filter, which is:

[0090] E_filter=α0•E_raw[i]+α1•E_raw[i-1]+α2•E_raw[i-2]-β1•E_filter[i-1]-β2•E_filter[i-2];

[0091] Where α0, α1, and α2 are the filter input coefficients, β1 and β2 are the filter feedback coefficients, i is the sequence number of the current sampled data, E_raw[i] is the original electric field intensity value collected at the current time, E_raw[i-1] and E_raw[i-2] are the original electric field intensity values ​​collected at the previous time and two time steps ago, respectively, E_filter[i] is the filtered electric field signal at the current time, and E_filter[i-1] and E_filter[i-2] are the filtered electric field signals at the previous time and two time steps ago, respectively.

[0092] The purpose of this design is to complete the hardware acquisition of electric field signals through planar induction electrodes with specific parameters. Combined with differential amplification, 12-bit ADC analog-to-digital conversion and second-order Butterworth low-pass filtering, it can accurately acquire clean electric field signals after removing interference, providing stable and reliable raw data for subsequent artificial intelligence calculations. The electric field acquisition and preprocessing are highly accurate and have strong anti-interference capabilities.

[0093] In this embodiment, during cloud data transmission and deep learning computation, the early warning device first integrates the filtered electric field signal E_filter, the manually selected voltage level, the corresponding reference electric field threshold E_base, GPS positioning data, and the unique identifier ID of the early warning device into a data packet, and then uploads it to the cloud through the communication unit.

[0094] Then, the data packet is retrieved from the cloud, and a multi-dimensional feature vector is extracted. This feature vector includes the filtered electric field signal E_filter, the manually set voltage level, the reference electric field threshold E_base, the 3rd harmonic feature H3, the 5th harmonic feature H5, and the 7th harmonic feature H7. Based on the 3rd, 5th, and 7th harmonic features, the interference confidence level P is calculated.

[0095] P=(|H3|+|H5|+|H7|) / (|H3|+|H5|+|H7|+E_filter);

[0096] Wherein, P ranges from 0 to 1. When P is less than 0.3, it is determined to be a real high-voltage electric field signal, and when P is greater than or equal to 0.3, it is determined to be an environmental stray interference signal.

[0097] Then, the harmonic characteristic factor F is obtained by weighted calculation of harmonic characteristics, which gives:

[0098] F = a·|H3| + b·|H5| + c·|H7|;

[0099] Where a, b, and c are fixed weighting coefficients that match the current voltage level, and the harmonic characteristic factor F is then normalized to obtain the dimensionless harmonic characteristic factor F_norm.

[0100] And based on the interference confidence level P and the dimensionless harmonic characteristic factor F_norm, the environmental interference correction coefficient K is obtained, which is:

[0101] K = 1 - (P × F_norm);

[0102] Used to correct deviations in the filtered electric field signal and eliminate errors caused by non-target electromagnetic interference from metals and cables in the environment, the true electric field strength E_real is calculated using the environmental interference correction coefficient K.

[0103] E_real = E_filter × K.

[0104] In practical implementation, a lightweight convolutional neural network model is deployed on the cloud platform. This model is an embedded deep learning structure optimized for power scenarios, consisting of an input layer, two hidden layers, and an output layer. After receiving the uploaded data packet, the model extracts multi-dimensional feature vectors, including the filtered electric field signal E_filter, the manually set voltage level, the reference electric field threshold E_base, the 3rd harmonic feature H3, the 5th harmonic feature H5, and the 7th harmonic feature H7. The reason for selecting the above-mentioned third harmonic features is that the high-voltage electric field of the power system uses the 50Hz power frequency as the fundamental wave, and the real charged body mainly excites the 3rd, 5th, and 7th odd harmonics, which are the characteristic features of the high-voltage electric field. The stray interference in the field environment does not have stable harmonic components, which can effectively distinguish between real risks and interference signals. At the same time, the content of even harmonics in the power system is extremely low and has no identification value. Selecting the 3rd, 5th, and 7th harmonics can reduce the computational power consumption of the model while ensuring the recognition accuracy. The model fuses the aforementioned multidimensional features through two hidden layers, distinguishing between real high-voltage electric field signals and environmental stray interference signals, thus completing interference correction and feature extraction of the electric field signal.

[0105] In the part where the real electric field strength E_real is obtained, the model calculates the interference correction coefficient K based on the interference confidence P and the harmonic characteristic factor F. First, the maximum value of F is normalized to obtain the dimensionless normalized harmonic characteristic factor:

[0106] F_norm = F / F_max;

[0107] Where F_max is the preset maximum value of the harmonic characteristic factor at this voltage level, and F_norm ranges from 0 to 1 and is a dimensionless number;

[0108] The correction coefficient is then obtained using the following formula:

[0109] K = 1 - (P × F_norm);

[0110] Finally, K is constrained to the effective range of 0.8 to 1.2. This correction coefficient is a dimensionless value and can be directly used to solve for the actual electric field strength E_real = E_filter × K.

[0111] In this embodiment, in the risk level determination and result downlink transmission, the risk level is divided based on the reference electric field threshold E_base corresponding to the manual selection, combined with the output real electric field strength E_real. The risk levels are safety level, warning level and danger level, respectively. The safety level is when the real electric field strength E_real is less than the reference electric field threshold E_base corresponding to the selected level.

[0112] The warning level is defined as follows: the actual electric field strength E_real is greater than or equal to the reference electric field threshold E_base corresponding to the selected range, and less than the sum of the reference electric field threshold E_base corresponding to the selected range and 1.0kV / m.

[0113] The danger level is defined as the sum of the actual electric field strength E_real and the reference electric field threshold E_base corresponding to the selected range, and 1.0 kV / m.

[0114] like Figures 2 to 7 As shown, this application also provides an artificial intelligence-based graded early warning system for electric shock prevention, used to implement an artificial intelligence-based graded early warning method for electric shock prevention, including a cloud and an early warning device, and the cloud and the early warning device are wirelessly connected. The early warning device is an electric shock prevention watch, including a watch body 1, with both ends of the watch body 1 connected to the watch strap 2 respectively. The watch body 1 is provided with a main control processing module, a timing module, a touch display module, a positioning communication module and an electrostatic induction early warning module. The watch body 1 is also provided with a first battery 6, a second battery 7 and an alert unit.

[0115] The first battery 6 is used to power the main control processing module, the timing module, the touch display module, the positioning and communication module, and the electrostatic induction warning module;

[0116] The second battery 7 is used to power the reminder and warning unit;

[0117] The reminder and warning unit is used to receive signals from the main control processing module and execute reminders and warnings.

[0118] The purpose of this design is,

[0119] By adding a dual power supply structure of a first battery and a second battery inside the watch body, the first battery is dedicated to powering the main control processing module, timing module and other basic modules, while the second battery is dedicated to powering the reminder and warning unit. The power supply circuit is completely independent and separated from the basic module, which completely avoids the interference of power consumption and failure of the basic module on the warning function and ensures that the electric shock warning function is continuously effective.

[0120] It should also be noted that the main control processing module, timing module, touch display module, positioning communication module, and electrostatic induction warning module are all existing technologies. When implementing this solution, those skilled in the art can choose appropriate models. For example, the main control processing module can be an STM32R523 or Infineon CY9DF56x, which uses hardware interfaces such as GPIO and SP to realize signal conversion, level matching, and timing control. It only completes the signal reception of the electrostatic induction warning module, the driving of the reminder and warning unit, and the hardware coordination control of each module. There is no algorithm operation or logic judgment, and it is purely hardware-level circuit control.

[0121] The timing module can be selected from the DS3231M, which has a built-in MEMS resonator / quartz crystal oscillator and temperature compensation circuit. It generates a stable clock signal through hardware oscillation and independently completes hardware counting of year, month, day, hour, minute and second. It can maintain timing even when the power is off and directly outputs the hardware clock signal to the main control processing module.

[0122] For the touch display module, the capacitive touch FT6236U can be selected to be paired with the touch screen 3. The display part lights up the touch screen 3 through the hardware driver circuit to complete the output of data, gear, and time. The touch part collects the touch point signal through the capacitive sensing hardware circuit, converts the analog touch signal into a digital signal and transmits it to the main control processing module to realize hardware-level touch interaction.

[0123] For the positioning and communication module, you can choose the Hezhou Air780E, which receives satellite navigation signals through a built-in GPS hardware antenna, calculates location information, and completes data transmission and reception through cellular hardware circuitry, realizing the hardware functions of positioning signal acquisition and data transmission.

[0124] For electrostatic discharge (ESD) induction early warning modules, the ETCR1860 series power frequency electric field sensor can be selected. It adopts a planar sensing electrode and capacitive coupling hardware structure to capture 50Hz power frequency electric field signals in space without contact. After differential amplification, hardware filtering, and 12-bit ADC analog-to-digital conversion, the electric field strength is converted into a standard electrical signal and output to the main control processing module.

[0125] Meanwhile, in this embodiment, a working cavity 4 is provided inside the watch body 1. The first battery 6, the second battery 7, and the reminder and warning unit are all located in the working cavity 4. An insulating partition plate 5 is also provided inside the working cavity 4, and the insulating partition plate 5 divides the working cavity 4 into two independent cavities. The second battery 7 and the reminder and warning unit are located in one independent cavity, and the first battery 6 is located in the other independent cavity.

[0126] The purpose of this design is to optimize the internal space layout of the watch by integrating the first battery, the second battery, and the alert / warning unit within a working cavity inside the watch body, thereby improving the watch's structural compactness and wearability. Furthermore, an insulating partition plate divides the working cavity into two independent chambers, allowing the second battery, alert / warning unit, and first battery to be arranged separately, achieving physical and electrical isolation, eliminating electromagnetic interference and structural interference between modules, and improving the device's operational stability.

[0127] In practice, the insulating partition plate 5 divides the working cavity 4 into two independent cavities, which makes the structure more organized, facilitates the assembly, debugging and maintenance of internal components and reduces the difficulty of production and processing.

[0128] In this embodiment, a first charging port 11 and a second charging port 12 are provided on the side wall of the watch body 1, and the first charging port 11 and the second charging port 12 are arranged side by side. The first charging port 11 is used to charge the first battery 6, and the second charging port 12 is used to charge the second battery 7.

[0129] The purpose of this design is to set a first charging port and a second charging port on the side wall of the device, so that the first battery and the second battery can be charged independently, realizing independent charging and discharging management of the two batteries, avoiding charging imbalance, improving the convenience of device use and power supply flexibility, and the choice of charging port can be a Type-C protocol charging port.

[0130] In this embodiment, a specific composition of the reminder and warning unit is also provided, which...

[0131] Includes a buzzer alarm module, a light warning module, and a vibration alert module;

[0132] The buzzer alarm module is used to emit a buzzer alarm sound;

[0133] The device is used to emit a red light alarm.

[0134] The vibration alert module is used to trigger an alarm on the vibration meter 1.

[0135] Meanwhile, the buzzer alarm module includes a buzzer 8, which is located on one side of the body 1; the light warning module includes a red LED light 10, which is located at the bottom of the working chamber 4; and the vibration reminder module includes a rotor motor and an eccentric wheel, and the rotor motor can drive the eccentric wheel to rotate.

[0136] The purpose of this design is to integrate a buzzer alarm, light warning, and vibration alert into the warning unit, allowing for multi-mode warning structures to work collaboratively. This adapts to noisy environments during power operations, significantly improving the effectiveness and recognizability of the warnings. The buzzer is positioned on one side of the meter body, with an optimized sound structure layout to ensure effective alarm propagation and improve signal clarity. A red LED light is located at the bottom of the working chamber, ensuring a stable installation and providing a clear and easily identifiable warning, allowing workers to quickly perceive the danger warning signal. The vibration alert module uses a rotor motor driving an eccentric wheel, ensuring stable and reliable vibration drive and a clear tactile feedback, further enhancing the perception of electric shock prevention.

[0137] Furthermore, for the light warning module, its red LED light can be installed at the bottom of the watch body 1, so that the red light is more easily seen by the user through the watch body. For the vibration reminder module, it can be installed on the metal plate 9 at the bottom of the working cavity. The metal plate 9 is in contact with the user's wrist, making it easier to transmit the vibration effect.

[0138] Meanwhile, when the early warning device receives an early warning command from the cloud with a downlink transmission level of 500Hz, the buzzer alarm module emits a single 500Hz buzzer with a duration of 200ms and a sound intensity of ≥85dB; the light warning module flashes a single 1Hz red light with a duration of 200ms; and the vibration reminder module outputs a single 2G amplitude vibration with a duration of 200ms.

[0139] When the early warning device receives a warning command from the cloud with a danger level, the buzzer alarm module emits three consecutive 500Hz beeps, with each beep spaced 500ms apart; the light warning module flashes red light continuously at 3Hz; and the vibration reminder module outputs a 2G amplitude continuous vibration, with a single vibration duration of 500ms and a vibration interval of 300ms.

[0140] When the early warning device receives a downlink warning command from the cloud at the security level, the alert unit does not perform any alarm action, and the touch display module only displays the current voltage level and the actual electric field strength E_real calculated by artificial intelligence.

[0141] In this embodiment, a computer storage medium is also provided, which stores a computer program. The computer storage medium can be one of magnetic random access memory, read-only memory, programmable read-only memory, erasable programmable read-only memory, electrically erasable programmable read-only memory, flash memory, magnetic surface memory, and optical disk. It can also be various devices including one or any combination of the above-mentioned memories, such as mobile phones, computers, tablet devices, etc. The computer program can drive a system to resolve conflicts between log data of different formats, and the computer program processor can execute an artificial intelligence-based graded early warning method for preventing electric shock.

[0142] Although the present invention has been described in detail with reference to the foregoing embodiments, those skilled in the art can still modify the technical solutions described in the foregoing embodiments or make equivalent substitutions for some of the technical features. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of the present invention should be included within the protection scope of the present invention.

Claims

1. An AI-based graded early warning method for electric shock prevention in power operations, applicable to power work scenarios, and implemented using an AI-based graded early warning system for electric shock prevention, characterized in that... The early warning method includes the following steps: Voltage warning level setting: Operators wear warning devices to set the warning level according to the voltage level of the power operation scenario, and generate a matching warning benchmark threshold E_base based on the warning level; The electric field parameters in the area where the workers are located are collected and preprocessed. The electric field parameters in the power operation scenario where the workers are located are obtained, and the signals are amplified, converted from analog to digital and filtered to obtain a clean electric field signal E_filter. Cloud data transmission and deep learning computation: The edge early warning device transmits the clean electric field signal E_filter, the manually selected voltage level, the corresponding reference electric field threshold E_base, and GPS positioning data to the cloud. The cloud completes the interference correction and feature extraction of the electric field signal to obtain the real electric field strength E_real of the area where the worker is located. The early warning device first integrates the filtered electric field signal E_filter, the manually selected voltage level, the corresponding reference electric field threshold E_base, GPS positioning data, and the unique identifier ID of the early warning device into a data packet, and uploads it to the cloud through the communication unit. Then, the data packet is retrieved from the cloud, and a multi-dimensional feature vector is extracted. This feature vector includes the filtered electric field signal E_filter, the manually set voltage level, the reference electric field threshold E_base, the 3rd harmonic feature H3, the 5th harmonic feature H5, and the 7th harmonic feature H7. Based on the 3rd, 5th, and 7th harmonic features, the interference confidence level P is calculated. P=(|H3|+|H5|+|H7|) / (|H3|+|H5|+|H7|+E_filter); Wherein, P ranges from 0 to 1. When P is less than 0.3, it is determined to be a real high-voltage electric field signal, and when P is greater than or equal to 0.3, it is determined to be an environmental stray interference signal. Then, the harmonic characteristic factor F is obtained by weighted calculation of harmonic characteristics, which gives: F = a·|H3| + b·|H5| + c·|H7|; Where a, b, and c are fixed weighting coefficients that match the current voltage level, and the harmonic characteristic factor F is then normalized to obtain the dimensionless harmonic characteristic factor F_norm. And based on the interference confidence level P and the dimensionless harmonic characteristic factor F_norm, the environmental interference correction coefficient K is obtained, which is: K = 1 - (P × F_norm); Used to correct deviations in the filtered electric field signal and eliminate errors caused by non-target electromagnetic interference from metals and cables in the environment, the true electric field strength E_real is calculated using the environmental interference correction coefficient K. E_real = E_filter × K; The risk level determination and result are transmitted back down. The cloud-based artificial intelligence model uses the reference electric field threshold E_base corresponding to the manually selected gear as the judgment boundary, and combines the actual electric field strength E_real to divide the risk level. The risk level, the actual electric field strength E_real and the warning instruction are transmitted back down. Based on the reference electric field threshold E_base corresponding to the manually selected gear, the risk level is divided into three levels: safe level, warning level and dangerous level. The safe level is when the actual electric field strength E_real is less than the reference electric field threshold E_base corresponding to the selected gear. The warning level is defined as follows: the actual electric field strength E_real is greater than or equal to the reference electric field threshold E_base corresponding to the selected range, and less than the sum of the reference electric field threshold E_base corresponding to the selected range and 1.0kV / m. The danger level is defined as the sum of the actual electric field strength E_real and the reference electric field threshold E_base corresponding to the selected range, and 1.0 kV / m. The system implements tiered early warning, with the end-side early warning device receiving the warning command and issuing tiered warnings to the personnel wearing the device based on the command. The warning system includes a cloud and a warning device, and the cloud and the warning device are connected wirelessly. The warning device is an anti-electric shock watch, including a watch body (1). The two ends of the watch body (1) are respectively connected to the watch strap (2). The watch body (1) is equipped with a main control processing module, a timing module, a touch display module, a positioning communication module and an electrostatic induction warning module. The watch body (1) is also equipped with a first battery (6), a second battery (7) and a reminder and warning unit. The first battery (6) is used to power the main control processing module, timing module, touch display module, positioning communication module and electrostatic induction warning module; The second battery (7) is used to power the reminder and warning unit; The reminder and warning unit is used to receive signals from the main control processing module and execute reminders and warnings.

2. The method for graded early warning of electric shock prevention based on artificial intelligence according to claim 1, characterized in that, The voltage warning level settings include 5 voltage levels and corresponding warning baseline thresholds E_base, which are as follows: Gear 1, warning voltage 380V, warning reference threshold E_base=0.5kV / m; Level 2, warning voltage 10kV, warning reference threshold E_base=1.0kV / m; Level 3, warning voltage 35kV, warning reference threshold E_base=1.5kV / m; Level 4, warning voltage 132kV, warning reference threshold E_base=2.0kV / m; Level 5, warning voltage 230kV, warning reference threshold E_base=2.5kV / m.

3. The graded early warning method for electric shock prevention based on artificial intelligence according to claim 2, characterized in that, In the acquisition and preprocessing of electric field parameters in the area where the workers are located, the planar induction electrode of the early warning device has an electrode area S of 1.2 cm² and an electrode spacing d of 0.8 mm. It first captures the 50 Hz power frequency electric field in the space non-contactly through the principle of capacitive coupling, and converts the electric field intensity into the original induced electromotive force E. p This completes the physical conversion from electric field signal to electrical signal; Then the original induced potential E p After amplification, the amplified signal E_amp is obtained, which has: E_amp=E p ×G; Where G is the magnification factor; Next, a 12-bit analog-to-digital converter (ADC) circuit is used to process the amplified signal E_amp, converting the continuous analog electrical signal into a discrete digital signal to obtain the original electric field strength value E_raw in digital form.

4. The graded early warning method for electric shock prevention based on artificial intelligence according to claim 3, characterized in that, In the acquisition and preprocessing of electric field parameters in the area where the workers are located, the raw digital electric field strength value E_raw is filtered using a second-order Butterworth low-pass filter algorithm to obtain a clean electric field signal E_filter, as follows: E_filter=α0•E_raw[i]+α1•E_raw[i-1]+α2•E_raw[i-2]-β1•E_filter[i-1]-β2•E_filter[i-2]; Where α0, α1, and α2 are the filter input coefficients, β1 and β2 are the filter feedback coefficients, i is the sequence number of the current sampled data, E_raw[i] is the original electric field intensity value collected at the current time, E_raw[i-1] and E_raw[i-2] are the original electric field intensity values ​​collected at the previous time and two time steps ago, respectively, E_filter[i] is the filtered electric field signal at the current time, and E_filter[i-1] and E_filter[i-2] are the filtered electric field signals at the previous time and two time steps ago, respectively.

5. The artificial intelligence-based graded early warning method for electric shock prevention according to claim 1, characterized in that, The watch body (1) is provided with a working cavity (4), and the first battery (6), the second battery (7) and the reminder and warning unit are all located in the working cavity (4). The working cavity (4) is also provided with an insulating partition plate (5), and the insulating partition plate (5) divides the working cavity (4) into two independent cavities. The second battery (7) and the reminder and warning unit are located in one independent cavity, and the first battery (6) is located in the other independent cavity.

6. The graded early warning method for electric shock prevention based on artificial intelligence according to claim 5, characterized in that, The alert unit includes a buzzer alarm module, a light alert module, and a vibration alert module; The buzzer alarm module is used to emit a buzzer alarm sound; The light warning module is used to emit a red light alarm; The vibration alert module is used to trigger an alarm on the vibrating meter body (1); When the early warning device receives an early warning command from the cloud with a downlink transmission level of 500Hz, the buzzer alarm module emits a single 500Hz buzzer with a duration of 200ms and a sound intensity of ≥85dB; the light warning module flashes a single 1Hz red light for 200ms; and the vibration reminder module outputs a single 2G amplitude vibration for 200ms. When the early warning device receives a warning command from the cloud with a danger level, the buzzer alarm module emits three consecutive 500Hz beeps, with each beep spaced 500ms apart; the light warning module flashes red light continuously at 3Hz; and the vibration reminder module outputs a 2G amplitude continuous vibration, with a single vibration duration of 500ms and a vibration interval of 300ms. When the early warning device receives a downlink warning command from the cloud at the security level, the alert unit does not perform any alarm action, and the touch display module only displays the current voltage level and the actual electric field strength E_real calculated by artificial intelligence.

7. A graded early warning medium for electric shock prevention based on artificial intelligence, characterized in that, Used to store a computer program that causes a computer to perform the method as described in any one of claims 1 to 4.

Images