Working method of safety intelligent warning electronic fence special for maintenance

Abstract

The invention belongs to the technical field of maintenance operation safety protection of iron and steel plants, and relates to a working method of a safety intelligent warning electronic fence special for maintenance, four corners of a maintenance area are respectively provided with a composite detection assembly to form a cross detection net covering the whole area, and each composite detection assembly is connected with a central control terminal. Each set of composite detection assembly comprises two groups of active infrared detector modules, two face recognition cameras and an audible and visual alarm. The system adopts infrared signal time difference to calculate coordinates and is linked with face recognition to verify identities, so that the false alarm rate is remarkably reduced. And when an unauthorized person is detected, a grading alarm and expelling mechanism is triggered. Through high-precision face recognition, infrared dynamic positioning and Internet of Things early warning linkage technologies, real-time monitoring and personnel management of a maintenance area are realized, and personal injury and equipment damage caused by the fact that unauthorized personnel or devices enter the maintenance operation area by mistake are prevented.

Description

Technical Field

[0001] The present invention belongs to the technical field of safety protection for maintenance operations in steel plants, and particularly relates to a working method of a special safety intelligent warning electronic fence for maintenance. Background Art

[0002] In order to eliminate potential safety hazards at the operation site, steel enterprises have taken various measures to improve the maintenance operation environment and enhance the safety awareness of operators. However, there are still individual personnel who ignore safety regulations. For this reason, enterprises have developed a variety of preventive measures, such as physical fences, video monitoring, and single-type sensor alarm systems. However, these traditional means have many loopholes. Traditional physical fences rely on fixed boundaries and cannot be dynamically adjusted, making it difficult to meet the coverage requirements of complex terrains. The single-type sensor alarm system lacks an intelligent decision-making system for multi-modal alarms, and the false trigger rate reaches 20%-30%. It frequently gives false alarms in rainy, snowy weather or when animals accidentally break in. Therefore, it is urgent to design a safety protection system for maintenance operations that integrates multi-spectral perception, three-dimensional space positioning, and intelligent decision-making linkage.

[0003] Chinese Patent Application for Invention CN117250447A discloses a monitoring device, method, and storage medium for on-site high-voltage tests, which uses a double-layer infrared fence to detect intrusion behaviors, but cannot accurately locate the intrusion area and has no identity recognition mechanism. Chinese Patent Application for Utility Model CN209402635U discloses a substation video monitoring and intelligent analysis system that only relies on face recognition as an access condition and does not combine environmental dynamic perception to determine the specific location of the intruding person. Summary of the Invention

[0004] The purpose of the present invention is to provide a working method of a special safety intelligent warning electronic fence for maintenance, which realizes real-time monitoring and personnel management of the maintenance area through high-precision face recognition, infrared dynamic positioning, and Internet of Things early warning linkage technology, and prevents unauthorized personnel or devices from entering the maintenance operation area and causing personal injuries and equipment damage.

[0005] The technical solution adopted by the present invention to solve its technical problems is: a working method of a special safety intelligent warning electronic fence for maintenance, including the following steps:

[0006] (1) Install a set of composite detection components at the four corners of the maintenance area respectively to form a cross-detection network covering the entire area. Each set of composite detection components is connected to the central control terminal. Each set of composite detection components includes two groups of active infrared detector modules, two face recognition cameras, and an audible and visual alarm;

[0007] (2) Through the central control terminal, establish a feature database of authorized personnel, collect the facial information of maintenance personnel through the camera and store it in the database, and each database for each maintenance task is independently generated;

[0008] (3) When any composite detection component detects an interruption of the infrared signal, intrusion determination is performed; the camera closest to the intrusion point is called, and the real-time face feature vector is matched with the database for cosine similarity.

[0009] (4) If an unauthorized person breaks in, the coordinates (x, y) of the intrusion point are calculated through the signal reception time difference ΔT between two adjacent active infrared detector modules.

[0010] (5) Trigger hierarchical alarm sound and light warning, voice broadcast the intrusion direction, and send an alarm message containing the intrusion coordinates to the central control terminal.

[0011] Further, the two groups of active infrared detector modules in step (1) include the first group of active infrared transmitters and receivers and the second group of active infrared transmitters and receivers. The coverage angle of the first group of active infrared transmitters and receivers is 120° in the horizontal direction and 30° in the vertical direction; the second group of active infrared transmitters and receivers forms a 60° crossing angle with the first group of active infrared transmitters and receivers to form redundant detection; both the first group of active infrared transmitters and receivers and the second group of active infrared transmitters and receivers are installed on an aluminum shell with an IP67 protection level. The transmitting end includes an infrared laser diode with a wavelength of 940nm and a Fresnel lens to expand the horizontal coverage angle to 110°; the receiving end uses a photodiode array with a built-in high-pass filter.

[0012] Further, the two face recognition cameras in step (1) are installed at a 90° angle at each corner of the maintenance area. The face recognition cameras are high-resolution cameras with infrared fill light function, a field of view of 70°×50°, and a marked matching rate threshold ≥0.95, and can recognize normally even when wearing a safety helmet.

[0013] Further, the sound and light alarm in step (1) uses a directional horn and a warning light.

[0014] Further, the active infrared detector module and the sound and light alarm in step (1) are respectively connected to the central control terminal through RS-485 wired connection, and the face recognition camera transmits data through WiFi6.

[0015] Further, the hardware composition of the central control terminal in step (2) is as follows: an embedded processor based on the ARM Cortex-A72 architecture, supporting multi-threaded parallel computing; having a 512GB NVMe SSD storage function, capable of saving real-time data within 7 days; the interface type is 4×RJ45 gigabit Ethernet, 2×USB 3.0, LoRa module; the power supply and redundancy design are dual power inputs, with an automatic switching time <10ms, and the RAID 1 disk array ensures data security.

[0016] Further, the specific operation steps of step (3) are as follows: when any active infrared detector module detects a signal interruption, the system is triggered to run; at this time, the central control terminal calls the camera closest to the intrusion point, starts multi-frame image fusion recognition, with a sampling frequency ≥ 10 fps, and matches the real-time face feature vector with the database using cosine similarity, with a threshold ≥ 0.92; if the first pairing fails, the central control terminal calls other adjacent cameras for secondary matching.

[0017] Further, the calculation method of the coordinates (x, y) of the intrusion point in step (4) is to calculate the coordinates of the intrusion point through the signal reception time difference ΔT between two adjacent active infrared detector modules, and find the optimal solution for multiple distance differences to obtain the final coordinates (x, y). The distance differences from the intrusion point to the four corners are calculated through the following formula: Δd = c × (ΔT1 - ΔT2), where c is the propagation rate of infrared signals in the atmosphere, which is 3×10 8 m / s, and ΔT1 and ΔT2 are the signal delay time differences of different detectors;

[0018] Then, using the hyperbola intersection positioning method, according to the definition of a hyperbola, the locus of points with a fixed distance difference Δd from two fixed points is a branch of a hyperbola. Given that the coordinates of the two fixed points are (x A , y A ) and (x B , y B ), substitute into the formula to determine the final coordinates (x, y):

[0019]

[0020] Note: A single equation cannot determine the coordinates. Two sets of similar detection data are required to establish a system of equations for solution.

[0021] Further, the hierarchical alarm in step (5) is divided into a first-level alarm and a second-level alarm: if an unauthorized person breaks in, the first-level alarm is activated, and the local sound and light alarm is activated to broadcast the voice command "Unauthorized person has broken into area X, please exit immediately"; the second-level alarm is: if the intruder does not exit the area within 15 seconds, the central control terminal automatically sends this voice command to the on-site security management personnel, and the on-site security management personnel expel the intruder on-site;

[0022] When it is detected that multiple people break in simultaneously, the unauthorized person closest to the energized equipment is preferentially locked for tracking and alarm, and the on-site situation is transmitted to the central control terminal, and the central control terminal sends instructions to the on-site security management personnel for timely handling.

[0023] The present invention has the following beneficial effects: The present invention is a safety protection system for maintenance operations that integrates multi-spectral perception, three-dimensional spatial positioning, and intelligent decision-making linkage. By combining the layout of dual detectors at the four corners of the maintenance area with multi-modal biometric recognition technology, high-precision intrusion positioning and dynamic safety protection are achieved. The system calculates coordinates using the time difference of infrared signals and links with face recognition to verify identities, significantly reducing the false alarm rate. And a hierarchical alarm and expulsion mechanism is triggered when unauthorized personnel are detected. The present invention is applicable to scenarios such as equipment maintenance areas, substations, transmission line maintenance, and high-voltage equipment maintenance. Through high-precision face recognition, infrared dynamic positioning, and Internet of Things early warning linkage technology, real-time monitoring and personnel management of the maintenance area are realized, preventing unauthorized personnel or devices from entering the maintenance operation area and causing personal injury and equipment damage, effectively ensuring the safety of maintenance operations. BRIEF DESCRIPTION OF THE DRAWINGS

[0024] Figure 1 is the flowchart of the method of the present invention.

[0025] Figure 2 is the physical diagram of the active infrared detector module in the present invention.

[0026] Figure 3 is the installation layout diagram of the active infrared detector module of the present invention at the maintenance site of the high-voltage power distribution room in Embodiment 1. DETAILED DESCRIPTION OF THE INVENTION

[0027] The following are specific embodiments of the present invention, which further describe the technical solutions of the present invention. However, the protection scope of the present invention is not limited to these embodiments. Any changes or equivalent substitutions that do not deviate from the concept of the present invention are included in the protection scope of the present invention.

[0028] As Figure 1 shown, a working method of a safety intelligent warning electronic fence dedicated for maintenance includes the following steps:

[0029] (1) Install a set of composite detection components at the four corners of the maintenance area respectively to form a cross-detection network covering the entire area. Each set of composite detection components is connected to the central control terminal. Each set of composite detection components includes two groups of active infrared detector modules, two face recognition cameras, and an audible and visual alarm. The four corners of the maintenance area refer to the northeast corner, the southeast corner, the northwest corner, and the southwest corner.

[0030] Two groups of active infrared detector modules include the first group of active infrared transmitters and receivers and the second group of active infrared transmitters and receivers. The coverage angle of the first group of active infrared transmitters and receivers is 120° in the horizontal direction and 30° in the vertical direction. The second group of active infrared transmitters and receivers forms a 60° crossing angle with the first group of active infrared transmitters and receivers to form redundant detection. Both the first group of active infrared transmitters and receivers and the second group of active infrared transmitters and receivers are installed on an aluminum housing with an IP67 protection level. The transmitting end includes an infrared laser diode with a wavelength of 940nm and a Fresnel lens to expand the horizontal coverage angle to 110°. The receiving end uses a photodiode array with a built-in high-pass filter to filter out ambient light interference.

[0031] Two face recognition cameras are installed at a 90° angle at each corner of the maintenance area. The face recognition cameras are high-resolution cameras with infrared fill light function, and the field of view angle is 70°×50°. The marker matching rate threshold is ≥0.95, and it can be normally recognized even when wearing a safety helmet.

[0032] Considering the noisy environment at the maintenance site, the audible and visual alarm uses a directional horn and warning lights.

[0033] The active infrared detector module and the audible and visual alarm are respectively connected to the central control terminal through RS-485 wired connection, and the face recognition camera transmits data through WiFi6.

[0034] (2) Through the central control terminal, establish a feature database of authorized personnel. Collect the facial information of maintenance personnel through the camera and store it in the database. The database for each maintenance task is independently generated.

[0035] Collect and store multi-angle facial image data of authorized maintenance personnel, including the frontal face, the left 45° side face, and the right 45° side face. Use structured light three-dimensional imaging technology to generate a facial depth map to enhance the robustness to light. Generate a unique task code for each maintenance task, and set dynamic protection area parameters, including the coordinates of live equipment and the safety buffer distance. After each maintenance task is completed, automatically destroy the associated temporary feature database, and only retain the anonymized statistical data of the log files.

[0036] The central control terminal is the intelligent core of the electronic fence system, responsible for data integration, calculation and decision-making, alarm dispatching and dynamic management. The hardware composition of the central control terminal is as follows: an embedded processor based on the ARM Cortex-A72 architecture, supporting multi-threaded parallel computing (such as processing 8-channel camera data + 8-channel infrared signals); having a 512GB NVMe SSD storage function, capable of saving real-time data within 7 days; the interface types are 4×RJ45 (gigabit Ethernet), 2×USB 3.0, and LoRa module; the power supply and redundancy design are dual power inputs (220VAC + 48V DC lithium battery pack), with an automatic switching time < 10ms, and the RAID 1 disk array ensures data security.

[0037] Data processing: When the active infrared detector module is triggered, the ΔT signal is input to the terminal; the coordinate area is preliminarily calculated through the time difference of arrival algorithm; the corresponding camera is called for visual verification, and the YOLOv7 target detection model is used to identify the intruder's body shape and movement direction; the final coordinates are determined, and the positioning error < 0.5m.

[0038] (3) When any composite detection component detects an interruption of the infrared signal, intrusion determination is performed; the camera closest to the intrusion point is called, and the real-time face feature vector is matched with the database for cosine similarity.

[0039] The specific operation steps are as follows: When any active infrared detector module detects a signal interruption, the system operation is triggered; at this time, the central control terminal calls the camera closest to the intrusion point, starts multi-frame image fusion recognition, with a sampling frequency ≥ 10fps, and matches the real-time face feature vector with the database for cosine similarity, with a threshold ≥ 0.92; if the first pairing fails, the central control terminal calls other adjacent cameras for secondary matching.

[0040] (4) If an unauthorized person breaks in, the coordinates (x, y) of the intrusion point are calculated through the signal reception time difference ΔT between two adjacent active infrared detector modules.

[0041] The calculation method of the coordinates (x, y) of the intrusion point is to calculate the coordinates of the intrusion point through the signal reception time difference ΔT between two adjacent active infrared detector modules, and find the optimal solution for multiple groups of distance differences to obtain the final coordinates (x, y). The distance differences from the intrusion point to the four corners are calculated through the following formula: Δd = c×(ΔT1 - ΔT2), where c is the propagation rate of the infrared signal in the atmosphere, 3×10 8 m / s, and ΔT1 and ΔT2 are the signal delay time differences between different detectors;

[0042] Then, using the hyperbola intersection positioning method, according to the definition of a hyperbola, the locus of points with a fixed distance difference Δd from two fixed points is a branch of a hyperbola. Given that the coordinates of the two fixed points are (xA , y A ), and (x B , y B ), substitute into the formula to determine the final coordinates (x, y):

[0043]

[0044] Note: A single equation cannot determine the coordinates. Two sets of similar detection data are required to establish a system of equations for solution.

[0045] (5) Trigger hierarchical alarm sound and light warning, voice broadcast the intrusion direction, and send an alarm message containing the intrusion coordinates to the central control terminal.

[0046] The hierarchical alarm is divided into a first-level alarm and a second-level alarm: If an unauthorized person intrudes, the first-level alarm is activated, and the local sound and light alarm is activated, and the voice command "Unauthorized personnel have intruded into area X (such as the northeast area), please exit immediately" is broadcast; The second-level alarm is: If the intruder does not exit the area within 15 seconds, the central control terminal automatically sends this voice command to the on-site security management personnel, and the on-site security management personnel drive away the intruding personnel on site.

[0047] When it is detected that multiple people intrude at the same time, the unauthorized person closest to the energized equipment is preferentially locked for tracking and alarm, and the on-site situation is transmitted to the central control terminal, and the central control terminal sends instructions to the on-site security management personnel for timely handling.

[0048] Embodiment 1

[0049] Taking the maintenance of the high-voltage power distribution room as an example, the working method of the special maintenance safety intelligent warning electronic fence of the present invention is specifically as follows:

[0050] (1) Install a set of composite detection components at the four corners (northeast corner, southeast corner, northwest corner, southwest corner) of the maintenance area of the high-voltage power distribution room. Each set of composite detection components is connected to the central control terminal. Each set of composite detection components includes two groups of active infrared detector modules, two face recognition cameras, and a sound and light alarm; specifically as Figure 2 、 Figure 3 shown.

[0051] (2) Through the central control terminal, establish a feature database of authorized personnel, collect and store multi-angle facial image data of authorized maintenance personnel, including the front face, the left 45° side face, and the right 45° side face, and store them in the database. The database for each maintenance task is independently generated.

[0052] (3) When someone enters the southeast area, the active infrared detector module detects a signal interruption at this time, triggering the system to run; the central control terminal calls the camera closest to the intrusion point, starts multi-frame image fusion recognition, and performs cosine similarity matching between the real-time face feature vector and the database; it is found that the intruder is unauthorized.

[0053] (4) Calculate the coordinates of the intruder. Calculate the coordinates of the intrusion point through the signal reception time difference ΔT between two adjacent detectors, and find the optimal solution for multiple distance differences to obtain the final coordinates (x, y). The specific method is as follows:

[0054] Suppose two pairs of active infrared detector modules (such as the northeast corner - A and the southeast corner - B) simultaneously detect an intrusion event, and the time differences of the infrared signal interruption caused by the movement of the intruder are ΔT1 (from A to the target) and ΔT2 (from B to the target).

[0055] With the speed of light c = 3×10 8 m / s, the distance difference between the two to the target can be obtained:

[0056] Δd = c×(ΔT1 - ΔT2)

[0057] Then, using the hyperbola intersection positioning method, according to the definition of a hyperbola, the locus of points with a fixed distance difference Δd from two fixed points (such as: active infrared detector modules A and B) is a branch of a hyperbola. Given that the coordinates of the two active infrared detector modules A and B are (x A , y A ) and (x B , y B ), substitute into the formula to determine the final coordinates (x, y):

[0058]

[0059] According to the placement position, set the coordinates of the four active infrared detector modules as follows:

[0060] Northeast corner A: (x A , y A ) = (10, 10)m

[0061] Southeast corner B: (x B , y B ) = (10, 0)m

[0062] Northwest corner C: (x C , y C ) = (0, 10)m

[0063] Southwest corner D: (x D , y D ) = (0, 0)m

[0064] The measured delay time difference between two pairs of active infrared detector modules is as follows:

[0065] Pair A-D: ΔT AD = 100 ns, Δd AD = c × ΔT AD = 30 m

[0066] Pair B-C: ΔT BC = 50 ns, Δd BC = c × ΔT BC = 15 m

[0067] Substitute into the equation to solve:

[0068]

[0069] The solution is (x, y) ≈ (6.2, 4.8) m, corresponding to the southeast area.

[0070] (5) Trigger the first-level alarm sound and light warning, and broadcast the voice command "Unauthorized personnel have entered the southeast area. Please exit immediately". If the intruder leaves, stop the alarm; if the intruder does not exit the area within 15 seconds, the central control terminal automatically sends this voice command to the on-site security management personnel, and the on-site security management personnel drive away the intruders on the spot.

[0071] Embodiment 2

[0072] When an authorized person enters the maintenance area, the active infrared detector module detects that someone has entered, and the nearby face recognition camera takes a picture of the face information and compares it with the database data. After comparison, the entering person is an authorized person, and they can pass normally without triggering an alarm.

[0073] The present invention is not limited to the above embodiments. Any person should know that structural changes made under the inspiration of the present invention, as long as they have the same or similar technical solutions as the present invention, fall within the protection scope of the present invention.

[0074] The technologies, shapes, and structures not described in detail in the present invention are all well-known technologies.

Claims

1. A working method of a special safety intelligent warning electronic fence for maintenance, characterized in that, It includes the following steps: (1) Install a set of composite detection components at the four corners of the maintenance area respectively to form a cross-detection network covering the entire area. Each set of composite detection components is connected to the central control terminal. Each set of composite detection components includes two groups of active infrared detector modules, two face recognition cameras, and an audible and visual alarm; (2) Through the central control terminal, establish a feature database of authorized personnel. Collect the facial information of maintenance personnel through the camera and store it in the database. The database for each maintenance task is generated independently; (3) When any composite detection component detects an interruption of the infrared signal, perform intrusion determination; call the camera closest to the intrusion point, and perform cosine similarity matching between the real-time face feature vector and the database; (4) If an unauthorized person breaks in, calculate the coordinates (x, y) of the intrusion point through the signal reception time difference ΔT between two adjacent active infrared detector modules; (5) Trigger a hierarchical alarm with audible and visual warning, voice broadcast the intrusion direction, and send an alarm message containing the intrusion coordinates to the central control terminal.

2. The working method of the special maintenance safety intelligent warning electronic fence according to claim 1, characterized in that, In the step (1), the two groups of active infrared detector modules include the first group of active infrared transmitters and receivers and the second group of active infrared transmitters and receivers. The coverage angle of the first group of active infrared transmitters and receivers is 120° in the horizontal direction and 30° in the vertical direction; the second group of active infrared transmitters and receivers and the first group of active infrared transmitters and receivers form a 60° crossing angle to form redundant detection; both the first group of active infrared transmitters and receivers and the second group of active infrared transmitters and receivers are installed on an aluminum shell with an IP67 protection level. The transmitting end includes an infrared laser diode with a wavelength of 940nm and a Fresnel lens to expand the horizontal coverage angle to 110°; the receiving end uses a photodiode array with a built-in high-pass filter.

3. The working method of the special maintenance safety intelligent warning electronic fence according to claim 1, characterized in that, In the step (1), the two face recognition cameras are arranged at a 90° angle at each corner of the maintenance area. The face recognition camera is a high-resolution camera with an infrared fill light function, a field of view of 70°×50°, and a marked matching rate threshold ≥0.95, and can be normally recognized even when wearing a safety helmet.

4. The working method of the special maintenance safety intelligent warning electronic fence according to claim 1, characterized in that, In the step (1), the audible and visual alarm uses a directional horn and a warning light.

5. The working method of the special maintenance safety intelligent warning electronic fence according to claim 1, characterized in that, In the step (1), the active infrared detector module and the audible and visual alarm are respectively connected to the central control terminal through RS-485 wired connection, and the face recognition camera transmits data through WiFi6.

6. The working method of the special maintenance safety intelligent warning electronic fence according to claim 1, characterized in that, In the step (2), the hardware composition of the central control terminal is as follows: an embedded processor based on the ARM Cortex-A72 architecture, supporting multi-threaded parallel computing; having a 512GB NVMe SSD storage function, capable of saving real-time data within 7 days; the interface type is 4×RJ45 gigabit Ethernet, 2×USB 3.0, LoRa module; the power supply and redundancy design are dual power inputs, with an automatic switching time <10ms, and a RAID 1 disk array to ensure data security.

7. The working method of the special maintenance safety intelligent warning electronic fence according to claim 1, characterized in that, The specific operation steps of step (3) are as follows: When any active infrared detector module detects a signal interruption, the system is triggered to run; at this time, the central control terminal calls the camera closest to the intrusion point, starts multi-frame image fusion recognition, the sampling frequency ≥ 10fps, and performs cosine similarity matching between the real-time face feature vectors and the database, with the threshold ≥ 0.92; If the first pairing fails, the central control terminal calls other nearby cameras for secondary matching.

8. The working method of the special maintenance safety intelligent warning electronic fence according to claim 1, characterized in that, In step (4), the calculation method of the coordinates (x, y) of the intrusion point is as follows: the coordinates of the intrusion point are calculated through the signal reception time difference ΔT between two adjacent active infrared detector modules, and the optimal solution is obtained by finding the optimal solution for multiple sets of distance differences to obtain the final coordinates (x, y). The distance differences from the intrusion point to the four corners are calculated through the following formula: Δd = c×(ΔT1 - ΔT2), where c is the propagation rate of the infrared signal in the atmosphere, 3×10 8 m / s, and ΔT1 and ΔT2 are the signal delay time differences between different detectors; Using the hyperbola intersection location method again, according to the definition of a hyperbola, the locus of points with a fixed difference in distance Δd from two fixed points (e.g., active infrared detector modules A and B) is a branch of a hyperbola. Given that the coordinates of the two fixed points are (x A , y A ) and (x B , y B ), substitute into the formula to determine the final coordinates (x, y): Note: A single equation cannot determine the coordinates. Two sets of similar detection data are required to establish a system of equations for solution.

9. The working method of the special maintenance safety intelligent warning electronic fence according to claim 1, characterized in that, In step (5), the hierarchical alarm is divided into a first-level alarm and a second-level alarm: If an unauthorized person breaks in, the first-level alarm is activated, and the local sound and light alarm is activated, and the voice command "Unauthorized person has broken into area X, please exit immediately" is broadcast; the second-level alarm is: If the intruder does not exit the area within 15 seconds, the central control terminal automatically sends this voice command to the on-site security management personnel, and the on-site security management personnel drive away the intruder on the spot; When it is detected that multiple people break in at the same time, the unauthorized person closest to the live equipment is preferentially locked for tracking and alarm, and the on-site situation is transmitted to the central control terminal, and the central control terminal sends instructions to the on-site security management personnel for timely handling.

Images