Pumped storage power station dynamic safety area management and control system and method based on multi-source fusion positioning

Abstract

The invention discloses a pumped storage power station dynamic safety area management and control system and method based on multi-source fusion positioning, and the system comprises a sensing layer, a control layer and an execution layer which are in communication connection through an industrial network, and achieve the precise dynamic division and real-time response of a pumped storage power station safety area cooperatively. Wherein the sensing layer is used for collecting personnel positioning data and environment monitoring data and comprises a wearable UWB + IMU module, a non-contact millimeter wave radar and an environment sensor; the control layer is an edge computing gateway, runs a built-in multi-source data fusion algorithm, carries out fusion processing on the data acquired by the sensing layer, dynamically generates a partition instruction according to the fused positioning data and a preset security rule, and outputs a control signal in a JSON format; the execution layer is used for receiving the control signal of the control layer and executing a warning action, and comprises a laser projection array and a directional sound field system; according to the invention, accurate dynamic division and real-time response of the safety area are realized.

Description

Technical Field

[0001] This invention relates to the technical field of safety warning systems for pumped storage power stations, and in particular to a dynamic safety area control system and method for pumped storage power stations based on multi-source fusion positioning. Background Technology

[0002] The existing safety warning systems for pumped-storage power stations have significant shortcomings in addressing their unique operating environment. Firstly, due to the highly dynamic and periodic changes in unit operating status, frequent adjustments to safety boundaries and equipment maintenance areas are necessary. Static isolation systems based on physical fences and fixed signs are time-consuming to adjust, making it difficult to quickly respond to temporary expansion needs during maintenance. Secondly, most power stations are underground structures with complex spatial structures and narrow passageways. Physical fences themselves occupy a significant amount of effective passage width, hindering not only the transport of large equipment but also emergency evacuation. Furthermore, the visibility of fixed signs, which the system relies on, is limited in environments with equipment obstruction or complex lighting conditions, and they cannot dynamically adjust safety zone boundaries based on real-time equipment operating status. This makes it difficult to release or isolate the corresponding area promptly and accurately when maintenance tasks are completed ahead of schedule or when adjacent equipment is still running. Thirdly, the existing audible and visual alarms are largely ineffective in the high ambient noise levels prevalent in pumped-storage power stations, failing to effectively alert staff to danger and reducing the overall reliability of the system. Summary of the Invention

[0003] The purpose of this invention is to overcome the above-mentioned shortcomings and provide a dynamic safety zone management system and method for pumped storage power stations based on multi-source fusion positioning. This solves the problems of insufficient dynamic boundary adjustment, multi-environment adaptability and intelligent linkage in the safety management system of pumped storage power stations, and realizes accurate dynamic division and real-time response of safety zones.

[0004] To solve the above-mentioned technical problems, the technical solution adopted by this invention is: a dynamic safety area control system for pumped storage power stations based on multi-source fusion positioning, comprising a sensing layer, a control layer, and an execution layer. The sensing layer, control layer, and execution layer are connected via an industrial network communication to collaboratively achieve accurate dynamic division and real-time response of the safety area of ​​the pumped storage power station; wherein: The perception layer is used to collect personnel positioning data and environmental monitoring data, including wearable UWB+IMU modules, non-contact millimeter-wave radar and environmental sensors; The control layer, which is the edge computing gateway, runs a built-in multi-source data fusion algorithm to fuse the data collected by the perception layer. Based on the fused positioning data and preset security rules, it dynamically generates partitioning instructions and outputs control signals in JSON format. The execution layer, which receives control signals from the control layer and executes warning actions, includes a laser projection array and a directional sound field system.

[0005] Preferably, the wearable UWB+IMU module is integrated into a safety helmet or tool bag, and through ultra-wideband UWB and inertial measurement unit (IMU), it enables accurate perception of the position and movement status of the person wearing the terminal.

[0006] Preferably, the non-contact millimeter-wave radar detects the presence and location of personnel not wearing terminals without contact by transmitting and receiving millimeter-wave signals.

[0007] Preferably, the environmental sensor includes millimeter-wave radar and intelligent image recognition camera, used to monitor unauthorized personnel entering the warning area in real time.

[0008] Preferably, the multi-source data fusion algorithm is a tightly coupled fusion framework based on Kalman filter (KF) or unscented Kalman filter (UKF), used to fuse UWB, IMU, and radar data.

[0009] Preferably, the multi-source data fusion algorithm is as follows: To fuse UWB, IMU, and radar data, the state vector of the fusion system is first constructed. It covers core parameters of personnel movement and sensor error compensation, adapts to the multi-interference environment of underground powerhouses in pumped storage power stations, and includes state vectors. The definition is as follows: ; in: This refers to the location of personnel on the floor plan of the underground factory building; The velocity of the person's planar movement; To achieve zero bias in the IMU accelerometer, To achieve zero bias in the IMU gyroscope and compensate for sensor drift caused by factory vibration; When personnel move through the passage at a constant speed with minimal changes in posture, KF rapid fusion is used in two steps: first, the initial position is calculated using IMU data; then, based on the real-time acceleration a and angular velocity ω acquired by the IMU, combined with the sampling period Δt, the position and velocity are updated. a) Speed ​​update: ; ; b) Location update: ; ; After calculating the initial position using IMU data, UWB / radar is used to correct the error. When UWB or millimeter-wave radar outputs data, the position predicted by the IMU is corrected. a) Observed values: ; b) Error calculation: , representing the difference between the observed value and the predicted value; c) Final position: ; ; Kalman gain; When the personnel's posture changes drastically, the system automatically switches to UKF to handle nonlinear errors, as follows: a) Switching judgment logic: The edge computing gateway monitors the IMU attitude change rate in real time, if... If so, use UKF; otherwise, use KF. b) UKF Calculation: Nonlinearity is handled using three key sampling points: the current predicted position, the +0.5m offset position, and the -0.5m offset position. Then, the predicted positions of these three sampling points are calculated using IMU data. Finally, the three predicted positions are weighted and combined with UWB / radar observations to obtain the final fused position. , Prediction for 3 sampling points coordinate, The same applies to axes.

[0010] Preferably, the laser projection array consists of RGB laser modules, which can project clear safety zone boundary lines on the ground or wall to indicate safe and dangerous areas.

[0011] Preferably, the directional sound field system employs a parametric array loudspeaker, which can propagate sound in a directional manner and generate clear voice prompts or alarms within a specific area.

[0012] In addition, the present invention also discloses a control method for the above-mentioned dynamic safety area control system for pumped storage power stations based on multi-source fusion positioning, comprising the following steps: Step 1: On-site survey and rule formulation; Step 2: Hardware and software solution design; Step 3: System Deployment and Integration; Step 4: Software deployment and system integration; Step 5: Achieve dynamic boundary generation and multi-level response based on multi-source fusion positioning.

[0013] Furthermore, step 1 specifically includes the following sub-steps: a. Area definition: Jointly determine the list of key areas that need to be dynamically managed with the power plant operation and maintenance and safety departments. The key areas include the water pump turbine / generator motor body maintenance area, the inlet ball valve area, and the main transformer area. b. Rule Formulation: Define the security rules for each area, including the security distance under different device states, authorized personnel list, and multi-level response triggering conditions; c. On-site surveying: Use a 3D laser scanner or high-precision total station to create a high-precision three-dimensional point cloud model of the underground plant and main plant, generate a digital twin base, record the building structure, metal equipment distribution, existing network interfaces, power supply point locations in key areas, and assess the electromagnetic environment and potential areas of multipath interference. d. Existing system interface survey: Interact with SCADA system and personnel management system, determine data interface protocol and data format, and clarify the parameters that can be provided. The data interface protocol includes OPCUA, Modbus TCP / IP, and HTTPAPI. The parameters include unit speed, on / off status, and work order information.

[0014] Furthermore, the safety rules target three key areas of pumped storage power stations: the maintenance area of ​​the pump turbine / generator motor, the inlet ball valve area, and the main transformer area. Combining the dual discrete states of the unit, namely "500r / min operating state" and "0r / min shutdown maintenance state", the rules clarify quantitative standards and judgment logic from three aspects: "safe distance, authorized personnel list, and multi-level response triggering conditions" to ensure that the control is executable and verifiable. Core control equipment 1: Water pumps and turbines / generators / motors; The unit is in operating condition, i.e., the SCADA system detects the unit speed as RPM = 500 r / min; the safety distance algorithm formula is: ; ln(RPM) reflects the logarithmic relationship between rotational speed and the danger radius. The higher the rotational speed, the greater the risk of inertial collisions with rotating parts. The constant term 1.2 meters is the average rotor radius of the unit, covering the "rotor edge + personnel arm safety redundancy". Unit operating state triggering conditions: When the SCADA system outputs "unit speed = 500 r / min" and "generator motor outlet circuit breaker closing signal = 1", the safety distance is automatically triggered. The laser projection array projects a red boundary line in an area with a radius of 2.5 meters centered at the unit center (x0, y0), which is defined as the "no entry zone". At the same time, it covers the safety isolation range of electrical components such as live junction boxes and excitation devices around the unit. In the shutdown and maintenance state, i.e., when SCADA detects RPM=0r / min and "Anti-switching safety measure in place signal=1" + "Main circuit power failure signal=1" + "Grounding switch closed signal=1", the safety distance algorithm formula is: ; Parameter definitions and values: The outer diameter of the water pump turbine main shaft is taken as 1.75 meters; The operating space for maintenance tools is 1.0 meter. For redundancy in the protection of live compartments, a length of 0.8 meters is used to cover the live compartments of the unit, such as the safety isolation of the stator terminals, to prevent accidental electric shock. To provide a buffer for personnel collaboration and emergency evacuation, a distance of 0.5 meters is recommended. The shutdown maintenance state trigger conditions must be met simultaneously: "Speed ​​= 0 r / min" + "Governor lock engagement signal = 1" + "Main circuit power off = 1" + "Grounding switch closed = 1" + "Work order status = in execution" to trigger the maintenance work area boundary. Only authorized personnel are allowed to enter the boundary, and the boundary must avoid the energized intervals of adjacent equipment. Core control equipment 2: Inlet ball valve; In operation mode, SCADA detects "ball valve fully open signal = 1, safety distance is fixed at 2.5 meters"; the trigger condition is: when the "ball valve is fully open", the red boundary is triggered, prohibiting personnel from approaching the valve body flange, valve hydraulic mechanism and control box. In the shutdown and maintenance state, i.e., "ball valve fully closed signal = 1" + "working seal engaged signal = 1" + "maintenance seal engaged signal = 1", the safety distance is a fixed value of 1.8 meters. The triggering conditions are: "ball valve fully closed signal = 1" + "working seal engaged signal = 1" + "maintenance seal engaged signal = 1", triggering the yellow boundary. The boundary must clearly delineate the "mechanical maintenance area" and the "electrical maintenance area" to prevent accidental entry into the energized interval of the control circuit that has not been de-energized. Core control equipment 3: Main transformer; In the running state, i.e., when SCADA detects "main transformer energized signal = 1", the safe distance algorithm formula is: ; U is the rated voltage of the main transformer, in kV; 0.1×U is the safe distance of the high-voltage electric field, and 0.8 meters is the buffer between personnel accidental contact and energized intervals; Triggering condition: "Main transformer energized" signal triggers, projecting a red boundary, the boundary must completely cover the main transformer bushings, tap changers, high-voltage leads and other energized intervals, and non-electrical inspection personnel are prohibited from entering; In the shutdown and maintenance state, i.e., "main transformer de-energization signal = 1" and "grounding switch closed signal = 1" + "high voltage side voltage test negative = 1", the safety distance calculation formula is as follows: ; 0.05×U is the minimum electrical safety distance under maintenance conditions, and 1.0 meter is the tool operating space; the triggering conditions are: "power off" + "grounding switch closed" + "high voltage side voltage test no voltage = 1" + "electrical work permit approved", triggering the yellow boundary.

[0015] Furthermore, step 2 specifically includes the following sub-steps: a. Perception layer deployment design: Based on the 3D model, simulation software is used to plan the installation location, height and angle of UWB base stations, millimeter-wave radar and smart cameras to ensure full coverage without blind spots, and optimize the number of base stations to control costs; b. Network Design: Design an industrial ring network to power and transmit data to the front-end sensors, ensuring bandwidth and real-time performance; c. Control Layer Design: Design a fusion framework based on Kalman Filter (KF) or Unscented Kalman Filter (UKF) to tightly couple and fuse UWB, IMU, and radar data; develop a core algorithm module with device status and positioning data as input and dynamic boundary coordinates as output; design a web-based monitoring center interface to display personnel location, safe zone, device status, and alarm information in real time. d. Selection of execution unit: Laser projector and directional sound field speaker are selected as execution units.

[0016] Furthermore, step 3 specifically includes the following sub-steps: a. Sensor deployment and calibration: Accurately install UWB base stations and use a total station to measure and record the absolute coordinates of each base station; deploy millimeter-wave radar and cameras, and adjust the angles and coverage; perform joint calibration of sensors, unify timestamps, and ensure spatiotemporal consistency of data; b. Installation of the execution unit: Install the laser projector and directional speaker at a high position on the ceiling or wall to ensure that the projection surface and sound field cover the target area and avoid people looking directly at the laser source.

[0017] Furthermore, step 4 specifically includes the following sub-steps: Platform deployment: Deploy a customized operating system and environment on the edge gateway, and deploy microservices such as data fusion service, dynamic boundary engine, rule engine, and WebHMI service in the form of containers; b. System Integration: Develop a data interface client, debug it with the SCADA system to ensure normal status acquisition and control command issuance; configure the rule engine and enter the security rules defined in step 1.

[0018] Furthermore, step 5, dynamic boundary generation, specifically includes the following sub-steps: a. Data Input: The status signals of the unit are read in real time through the industrial network interface. The status signals are switch signals, including "running" and "stopped" status. The fusion positioning system tracks all personnel wearing positioning tags in real time to obtain accurate coordinate information. b. Boundary generation: The computing engine on the edge computing gateway receives the status signal from SCADA and selects the corresponding security rule according to the unit status. Rule A is used when the unit is in the "running" state, and rule B is used when the unit is in the "stopped" state. c. Command Issuance and Execution: The computing engine generates control commands and sends them to the laser projector and directional speakers via WiFi or Ethernet, projecting a halo around the ground around the unit and issuing a voice warning in the direction of the intruder; d. Real-time monitoring and dynamic adjustment: Monitor the unit status and personnel location in real time. If the unit status changes, query the rule base again to generate a new virtual boundary.

[0019] Furthermore, Rule A corresponds to the "running" state of the unit, that is, the SCADA detected speed = 500 r / min, the equipment is energized / rotating, and the core objective is to prevent collisions of rotating parts, prevent electric shock from high voltage, and prohibit unauthorized personnel from approaching; Rule B corresponds to the "stopped" state of the unit, that is, the SCADA detected speed = 0 r / min, the equipment is stopped + power off and grounded + anti-rotation safety measures are in place, and the core objective is to leave sufficient space for tool operation, prevent accidental entry into energized areas, and ensure the safe operation of maintenance personnel.

[0020] Furthermore, step 5, the multi-level response, specifically includes the following sub-steps: a. Warning Zone Response: The warning zone is an area 1m from the boundary. PWM dimming technology is used to achieve a slow flashing effect of the yellow laser to attract people's attention. The directional sound field system uses beamfront synthesis technology to directionally propagate voice prompts to the warning zone, ensuring that people in the area can clearly hear the prompts. b. Intrusion Zone Response: The intrusion zone is the area entering the boundary. The rotational scanning of the red laser is achieved through multi-laser head synchronous control technology. The sound field intensity adaptive algorithm automatically adjusts the sound pressure level according to the environmental noise of the intrusion zone to ensure clear transmission even in noisy environments.

[0021] Furthermore, the adaptive sound field intensity algorithm includes the following process: Step 1): Real-time acquisition and preprocessing of ambient noise: The miniature microphone built into the directional sound field system is installed near the speaker to simultaneously acquire ambient noise; Preprocessing logic: The acquired noise signal is subjected to "noise filtering" and "mean calculation"; Output parameter: Effective value of ambient noise. ; Step 2): Target sound pressure level calculation: Based on the reference sound pressure in the warning zone / intrusion zone, combined with environmental noise and scene compensation, calculate the real-time target sound pressure. The formula is: ; Real-time target sound pressure level; 15 represents the regional reference sound pressure level; 15 represents the target signal-to-noise ratio. This represents the current effective value of the pre-processed environmental noise. The scene compensation coefficient is 0-5dB. When the reverberation is severe in the underground factory, it is taken as 3-5dB, and when the passage is open, it is taken as 0-2dB to offset the sound reflection attenuation.

[0022] Beneficial effects of this invention: 1. This invention relies on multi-source fusion positioning and edge computing, enabling real-time response to unit status switching and temporary maintenance needs at the security boundary. The response speed is far faster than traditional static isolation, avoiding potential boundary lag issues. Layered perception using wearable modules, millimeter-wave radar, and environmental sensors covers personnel wearing and not wearing terminals, achieving full-scene, blind-spot-free monitoring and eliminating the risk of missed monitoring. Replacing physical fences with laser projection virtual boundaries frees up passageway space, facilitating the transportation of large equipment, eliminating emergency evacuation obstacles, and reducing fence maintenance costs. Laser projection (clear visibility) combined with directional sound fields (directional audio transmission) adapts to equipment obstruction and high-noise environments, ensuring accurate and effective warnings and improving warning reliability. Dynamically calculating safe distances using integrated equipment status data, coupled with multi-level responses (differentiated warnings for early warning zones and intrusion zones), avoids a "one-size-fits-all" approach, balancing control precision and maintenance efficiency. Supporting mainstream protocol integration with SCADA and personnel management systems, industrial ring network + containerized deployment facilitates future expansion and reduces system upgrade costs. Based on a digital twin platform and Web monitoring center, automatic execution of security rules and real-time global status visualization are achieved, reducing manual intervention, lowering costs, and preventing operational errors.

[0023] 2. This invention solves the problems of insufficient dynamic boundary adjustment, multi-environment adaptability and intelligent linkage in the safety management and control system of pumped storage power stations, and realizes accurate dynamic division of safety areas and real-time response. Attached Figure Description

[0024] Figure 1 This is a schematic diagram of a dynamic safety area control system architecture for a pumped storage power station based on multi-source fusion positioning. Figure 2 This is a flowchart for dynamic boundary generation based on multi-source fusion positioning. Detailed Implementation

[0025] The present invention will now be described in further detail with reference to the accompanying drawings and specific embodiments.

[0026] Example 1: As Figure 1 As shown, a dynamic safety area control system for pumped storage power stations based on multi-source fusion positioning includes: 1. Perception layer: a. Wearable UWB+IMU Module: This module is integrated into a safety helmet or tool kit and uses ultra-wideband (UWB) technology and an inertial measurement unit (IMU) to achieve accurate perception of a person's position and motion status.

[0027] b. Non-contact millimeter-wave radar: Used to detect personnel not wearing the terminal. By transmitting and receiving millimeter-wave signals, it detects the presence and location of personnel without contact, compensating for the shortcomings of wearable terminals and ensuring effective monitoring of all personnel within the power station.

[0028] c. Environmental sensors: Equipped with millimeter-wave radar and intelligent image recognition cameras to monitor unauthorized personnel entering the warning area in real time.

[0029] 2. Control Layer: The edge computing gateway runs a built-in multi-source data fusion algorithm to fuse data from multi-source positioning terminals and environmental sensors. Based on the fused positioning data and preset security rules, the gateway dynamically generates partitioning instructions and outputs control signals in JSON format to guide execution layer devices (such as laser projection arrays and directional sound field systems) to adjust the safety zone boundaries in real time. The multi-source data fusion algorithm is a tightly coupled fusion framework based on Kalman filter (KF) or unscented Kalman filter (UKF) for fusing UWB, IMU, and radar data.

[0030] Tightly coupled fusion framework based on Kalman filter (KF) or unscented Kalman filter (UKF): To fuse UWB, IMU, and radar data, the state vector of the fusion system is first constructed. It covers core parameters of personnel movement and sensor error compensation, adapts to the multi-interference environment of underground powerhouses in pumped storage power stations, and includes state vectors. The definition is as follows:

[0031] in: The location of personnel on the underground plant floor plan (unit: m, ignore) Axis (fixed floor height); The velocity of a person moving in a plane (unit: m / s); Zero bias of the IMU accelerometer (unit: m / s²). Zero bias (unit: rad / s) for IMU gyroscope, compensating for sensor drift caused by factory vibration.

[0032] When personnel move through the passage at a constant speed with minimal changes in posture (such as during normal inspections), KF rapid fusion is used in two steps: first, the initial position is calculated using IMU data; then, based on the real-time acceleration *a* and angular velocity *ω* acquired by the IMU, and combined with the sampling period Δt = 0.01s (a common frequency for IMUs), the position and velocity are updated. a) Speed ​​update: ; .

[0033] b) Location update: ; .

[0034] After calculating the initial position using IMU data, UWB / radar is used to correct the error. When UWB (positioning accuracy ±10cm) or millimeter-wave radar (±0.5m) outputs data, the position predicted by the IMU is corrected. a) Observed values:

[0035] b) Error calculation: , which represents the difference between the observed value and the predicted value.

[0036] c) Final position: ; . This is the Kalman gain, typically a fixed value of 0.8.

[0037] When personnel posture changes drastically (such as bending over or climbing during maintenance), the system automatically switches to UKF to handle nonlinear errors, with the following logic: a) Switching judgment logic: The edge computing gateway monitors the IMU attitude change rate in real time, if... (For the angular rate threshold when bending over / turning around), use UKF; otherwise, use KF.

[0038] b) UKF Calculation: Nonlinearity is handled using three key sampling points: the current predicted position, the +0.5m offset position, and the -0.5m offset position (covering the error range of personnel activity in the factory). Then, IMU data is used to calculate the predicted positions of the three sampling points. Finally, UWB / radar observations are combined, and the three predicted positions are weighted (0.6 for the midpoint and 0.2 for each offset point) to obtain the final fused position. , Prediction for 3 sampling points coordinate, The same applies to axes.

[0039] 3. Execution layer: a. Laser projection array: Composed of RGB laser modules. Using laser projection technology, it projects clear safety zone boundary lines onto the ground or walls, such as lines of different colors like yellow and red, to visually indicate safe and dangerous areas.

[0040] b. Directional sound field system: Uses parametric array loudspeakers to propagate sound in a direction, producing clear voice prompts or alarms within a specific area.

[0041] Example 2: This invention discloses a control method for the dynamic safety area control system of a pumped storage power station based on multi-source fusion positioning, comprising the following steps: Step 1: On-site survey and rule formulation: a. Area definition: Jointly determine the list of key areas that need to be dynamically managed with the power plant operation and maintenance and safety departments (such as: water pump turbine / generator motor maintenance area, inlet ball valve area, main transformer area, etc.).

[0042] b. Rule Formulation: Clearly define safety rules for each area (e.g., safety distances under different equipment conditions, authorized personnel lists, and multi-level response triggering conditions). Safety rules target three key areas of the pumped storage power station: the pump turbine / generator motor maintenance area, the inlet ball valve area, and the main transformer area. Considering the unit's dual discrete states of "500r / min operation" and "0r / min shutdown maintenance," quantitative standards and judgment logic are clearly defined from three aspects: "safety distances, authorized personnel lists, and multi-level response triggering conditions," ensuring that control measures are executable and verifiable.

[0043] Core control equipment 1: Water pumps and turbines / generators (rotary + electrical composite equipment) The safe distance algorithm formula is as follows: (for unit operation status, SCADA system detects unit speed RPM = 500 r / min): (Unit: meters) ln(RPM) reflects the logarithmic relationship between rotational speed and the danger radius—the higher the rotational speed, the greater the risk of inertial collisions with rotating parts; the constant term 1.2 meters represents the average rotor radius of the unit, covering "rotor edge + personnel arm safety redundancy". For example, when the RPM is 500, rice.

[0044] Unit operating state triggering conditions: When the SCADA system outputs "unit speed = 500 r / min" and "generator motor outlet circuit breaker closing signal = 1" (equipment is energized), the safety distance is automatically triggered. The laser projection array projects a red boundary line in an area with a radius of 2.5 meters centered at the unit center (x0, y0), which is defined as the "no entry zone". At the same time, it covers the safety isolation range of electrical components such as energized junction boxes and excitation devices around the unit.

[0045] In the shutdown and maintenance state (SCADA detection RPM=0r / min and "Anti-rotation safety measure in place signal=1" + "Main circuit power failure signal=1" + "Grounding switch closed signal=1"), the safety distance algorithm formula is: (Unit: meters) Parameter definitions and values: The outer diameter of the water pump turbine main shaft is taken as 1.75 meters; To provide operating space for maintenance tools, 1.0 meter is selected (to accommodate the movement range of a 1.5-meter long socket wrench and hoisting equipment). For redundancy in the protection of live compartments, a length of 0.8 meters is used to cover the live compartments of the unit, such as the safety isolation of the stator terminals, to prevent accidental electric shock. For personnel collaboration and emergency evacuation buffer, a depth of 0.5 meters is recommended.

[0046] The shutdown maintenance state trigger conditions must be met simultaneously: "Speed ​​= 0 r / min" + "Governor lock engagement signal = 1" + "Main circuit power off = 1" + "Grounding switch closed = 1" (electrical safety) + "Work order status = in execution" to trigger the maintenance work area boundary. Only authorized personnel are allowed to enter the boundary, and the boundary must avoid the energized intervals of adjacent equipment.

[0047] Core control equipment 2: Inlet ball valve In operation (SCADA detects "ball valve fully open signal = 1"), the safe distance is fixed at 2.5 meters. The trigger condition is: when the "ball valve is fully open", the red boundary is triggered, prohibiting personnel from approaching the valve body flange, valve hydraulic mechanism, and control box.

[0048] In the shutdown and maintenance state ("ball valve fully closed signal = 1" + "working seal engaged signal = 1" + "maintenance seal engaged signal = 1"), the safety distance is a fixed value of 1.8 meters (involving live-line protection, covering the maintenance isolation requirements of valve electrical junction boxes and actuators). The triggering condition is: "ball valve fully closed signal = 1" + "working seal engaged signal = 1" + "maintenance seal engaged signal = 1", triggering the yellow boundary. The boundary must clearly delineate the "mechanical maintenance area" and the "electrical maintenance area" to prevent accidental entry into the live-line interval of the control circuit that has not been de-energized.

[0049] Core control equipment 3: Main transformer In operation (SCADA detection "main transformer energized signal = 1"), the safe distance algorithm formula is: (Unit: meters) U is the rated voltage of the main transformer, in kV; 0.1×U is the safe distance from the high-voltage electric field, and 0.8 meters is for accidental contact by personnel plus buffer zone for energized areas. Triggering condition: Triggered by the "Main transformer energized" signal, a red boundary is projected. The boundary must completely cover the main transformer bushings, tap changers, high-voltage leads, and other energized areas. Entry by non-electrical inspection personnel is prohibited.

[0050] In the shutdown and maintenance state ("Main transformer power failure signal = 1" and "Grounding switch closed signal = 1" + "High voltage side voltage test negative = 1"), the safety distance calculation formula is as follows: (Unit: meters) 0.05×U is the minimum electrical safety distance under maintenance conditions, and 1.0 meter is the tool operating space. The triggering conditions are: "power off" + "grounding switch closed" + "high voltage side voltage test negative = 1" + "electrical work permit approved", triggering the yellow boundary.

[0051] c. On-site mapping: Using a 3D laser scanner or high-precision total station, high-precision 3D point cloud models are created for complex environments such as underground powerhouses and main powerhouses, generating a digital twin base. The building structure, distribution of metal equipment, existing network interfaces (RJ45 / fiber optic), and power supply locations in key areas are recorded, and the electromagnetic environment and potential areas of multipath interference are assessed.

[0052] d. Existing system interface survey: Interact with SCADA system and personnel management system (such as two-ticket system), determine data interface protocol (such as: OPC UA, Modbus TCP / IP, HTTP API) and data format, and clarify the parameters that can be provided (such as unit speed, switch status, work ticket information).

[0053] Step 2, Hardware and software solution design: a. Perception layer deployment design: Based on the 3D model, simulation software is used to plan the installation location, height, and angle of UWB base stations, millimeter-wave radar, and smart cameras to ensure full coverage without blind spots, and optimize the number of base stations to control costs.

[0054] b. Network Design: Design an industrial ring network (e.g., using IEEE 802.3bt / AT PoE++ switches) to power and transmit data to the front-end sensors, ensuring bandwidth and real-time performance.

[0055] c. Control Layer Design: Design a fusion framework based on Kalman filtering (KF) or unscented Kalman filtering (UKF) to tightly couple and fuse UWB, IMU, and radar data; develop a core algorithm module with device status (SCADA data) and positioning data as inputs and dynamic boundary coordinates (JSON format) as outputs; design a web-based monitoring center interface to display personnel location, safe zone, device status, and alarm information in real time.

[0056] d. Execution unit: Employs a laser projector and a directional sound field amplifier.

[0057] Step 3, System Deployment and Integration: a. Sensor deployment and calibration: Accurately install UWB base stations and use a total station to measure and record the absolute coordinates of each base station.

[0058] Deploy millimeter-wave radar and cameras, and adjust their angles and coverage. Perform joint calibration of the sensors and unify timestamps to ensure spatiotemporal consistency of data.

[0059] b. Installation of the execution unit: Install the laser projector and directional speaker at a high position on the ceiling or wall to ensure that the projection surface and sound field cover the target area and avoid people looking directly at the laser source.

[0060] Step 4: Software Deployment and System Integration a. Platform Deployment: Deploy a customized operating system and environment on the edge gateway. Deploy microservices such as data fusion services, dynamic boundary engines, rule engines, and Web HMI services in the form of containers.

[0061] b. System Integration: Develop a data interface client and debug it with the SCADA system to ensure normal status acquisition and control command issuance. Configure the rule engine and enter the safety rules defined in the first phase.

[0062] Step 5: Achieve dynamic boundary generation and multi-level response based on multi-source fusion localization: 1. Dynamic boundary generation mechanism for multi-source fusion positioning a. Data Input: The system reads the unit's status signals (switching signals, such as "unit_1_status": "running" or "unit_1_status": "stopped") in real time through the industrial network interface. The fusion positioning system (UWB+IMU) tracks all personnel wearing positioning tags in real time to obtain precise coordinate information.

[0063] b. Boundary Generation: The software (computing engine) running on the edge computing gateway receives the SCADA status signal and determines the rule to use based on the unit status (if the status is "running", rule A is used in the next step; if the status is "stopped", rule B is used in the next step). Rule A corresponds to the unit's "running" status (SCADA detected speed = 500 r / min, equipment is energized / rotating), and its core objectives are to prevent collisions with rotating parts, prevent electric shock from high voltage, and prohibit unauthorized personnel from approaching; Rule B corresponds to the unit's "stopped" status (SCADA detected speed = 0 r / min, equipment is stopped + power off and grounded + anti-rotation safety measures in place), and its core objectives are to leave sufficient space for tool operation, prevent accidental entry into energized areas, and ensure the safe operation of maintenance personnel.

[0064] The safety distances and triggering conditions specified by both for the three key areas (water pump turbine / generator motor, inlet ball valve, and main transformer) have been described above and will not be repeated here.

[0065] c. Command Issuance and Execution: The computing engine generates commands, which are instantly sent via network (such as WiFi or Ethernet) to laser projectors and directional speakers located on the roof of the plant, projecting halos of light onto the ground around the unit. A voice warning is issued in the direction of the intruder.

[0066] d. Real-time monitoring and dynamic adjustment: The system constantly monitors the status of the units and the location of personnel. Once the status changes (for example, a running unit stops), the loop starts from the beginning, the engine re-queries the rule base, and generates new virtual boundaries.

[0067] 2. Multi-level response principle a. Warning zone (e.g., 1m from the boundary): PWM dimming technology is used to achieve a slow flashing effect of the yellow laser to attract people's attention; the directional sound field system uses wavefront synthesis technology to propagate voice prompts directionally to the warning zone, ensuring that people in the area can clearly hear the prompts.

[0068] b. Intrusion Zone (Entry Boundary): Multi-laser head synchronous control technology enables the red laser to rotate and scan, forming a clear boundary warning; the sound field intensity adaptive algorithm automatically adjusts the sound pressure level according to the ambient noise in the intrusion zone, ensuring clear transmission even in noisy environments. The sound field intensity adaptive algorithm is based on "constant target signal-to-noise ratio" and uses a closed loop of "real-time acquisition of ambient noise → calculation of the required target sound pressure → dynamic adjustment of speaker output intensity" to counteract the effects of underground plant equipment operating noise (such as unit rotation, water pump operation, noise range 60-100dB) and reverberation (sound attenuation caused by concrete wall reflection).

[0069] Step 1): Real-time acquisition and preprocessing of ambient noise. The directional sound field system's built-in miniature microphone (sampling frequency 44.1kHz, range 40-120dB) is installed near the speaker (distance ≤0.5 meters) to simultaneously acquire ambient noise. Preprocessing logic: The acquired noise signal undergoes "noise filtering" (filtering low-frequency interference below 50Hz and high-frequency noise above 10kHz, focusing on the main noise frequency bands of human voices and equipment) and "mean calculation" (taking the effective noise value within 100ms to avoid erroneous adjustments due to instantaneous noise peaks). Output parameter: Effective value of ambient noise. (Unit: dB).

[0070] Step 2): Target sound pressure level calculation. Based on the baseline sound pressure level of the warning zone / intrusion zone, and combined with environmental noise and scene compensation, calculate the real-time target sound pressure level using the following formula:

[0071] Real-time target sound pressure level (unit: dB); 15 represents the regional reference sound pressure level (default warning zone 75dB, intrusion zone 105dB); 15 represents the target signal-to-noise ratio (dB, fixed threshold). This represents the current RMS value of the pre-processed ambient noise (dB). The scene compensation factor is 0-5dB (3-5dB for severe reverberation in underground factories, and 0-2dB for open passageways to offset sound reflection attenuation).

[0072] For example: Scenario 1: Warning zone, current ambient noise (During unit operation), the compensation coefficient K = 3dB, substitute into the calculation. The value is 103dB.

[0073] Scenario 2: Intrusion Zone, Current Ambient Noise (Maintenance bridge crane + unit operating simultaneously), compensation coefficient K=5dB. The value is 120dB.

[0074] The above embodiments have the following beneficial effects: 1. Resolve the problem of "lag in static isolation adjustment" and achieve real-time dynamic response of safety boundaries. Existing technologies using physical fences and fixed signs are time-consuming to adjust and cannot meet the needs of highly dynamic unit operation and temporary maintenance. This invention, through a collaborative architecture of "multi-source fusion positioning (UWB + IMU + millimeter-wave radar) + edge computing gateway + laser projection array," can read unit status signals (such as "running / stopped") from the SCADA system in real time. It dynamically generates safety boundary commands through a multi-source data fusion algorithm, and the laser projection array instantly updates the ground / wall boundary lines (e.g., switching from a yellow warning line to a red danger line). The response speed is far faster than traditional static isolation methods, allowing for rapid adaptation to scenarios such as maintenance area expansion and unit status switching, avoiding safety hazards caused by delayed boundary adjustments.

[0075] 2. Solve the problem of "incomplete personnel monitoring coverage" and achieve full-scene, blind-spot-free personnel monitoring. Existing technologies rely on wearable devices, which can easily miss individuals not wearing them. This invention achieves full coverage through a "layered sensing design": Wearable UWB+IMU modules accurately track the location and movement of maintenance personnel wearing the terminals; Non-contact millimeter-wave radar can detect people who are not wearing terminals (such as temporary visitors), making up for the coverage blind spots of wearable terminals; Environmental sensors (millimeter-wave radar + intelligent image recognition camera) further double-verify and alert on personnel entering the area, ensuring that all personnel in the power station are effectively monitored and avoiding safety accidents caused by unmonitored personnel.

[0076] 3. Solve the problem of "physical fences occupying space" and improve the space utilization and emergency safety of the power station. Existing technologies use physical fences that occupy narrow passageways in underground factories, hindering the transport of large equipment and emergency evacuation. This invention uses a laser projection array to generate a "virtual safety boundary," eliminating the need for physical fences and completely freeing up passageway space. This avoids spatial obstacles during the transportation of large maintenance equipment (such as water pump and turbine components), thereby improving maintenance efficiency. It also eliminates physical obstacles during emergency evacuation, reduces the risk of delays in personnel evacuation, and reduces the costs of installing, dismantling, and maintaining physical fences.

[0077] 4. Solves the problem of "low effectiveness of traditional warnings" and achieves accurate warnings in high-noise environments. Existing technologies suffer from poor visibility of fixed signs (due to equipment obstruction / complex lighting) and audible and visual alarms malfunction under high noise conditions. This invention solves this problem through a dual, precise warning solution combining visual and auditory signals. The laser projection array uses an RGB laser module, and the boundary lines of the projected yellow / red colors are clearly visible. It is not affected by equipment obstruction or changes in light, and personnel can intuitively identify safe / dangerous areas. The directional sound field system uses parametric array loudspeakers to directionally propagate voice prompts / alarms to the target area (such as the warning zone / intrusion zone), avoiding sound diffusion failure in high-noise environments, ensuring that people in the area can clearly receive warning information, and improving the reliability of warnings.

[0078] 5. Address the issue of "insufficient precision in safety management" and achieve precise safety management based on equipment status. Existing technologies cannot dynamically adjust security rules based on the real-time status of devices, easily leading to either "over-control" or "under-control." This invention achieves precise control through "data fusion + rule engine": The edge computing gateway uses Kalman filtering (KF) / unscented Kalman filtering (UKF) algorithms to integrate UWB, IMU, radar data and SCADA system equipment status (such as unit speed and switch status) and dynamically calculate the safe distance (such as increasing the safe distance when the unit is "running" and reducing it to adapt to maintenance needs when the unit is "stopped"). The multi-level response mechanism (slow flashing yellow laser + directional voice prompt in the warning zone, rotating red laser + adaptive sound pressure alarm in the intrusion zone) can accurately match the warning intensity according to the degree of personnel intrusion, avoiding a "one-size-fits-all" warning method, reducing the impact of false alarms on operation and maintenance efficiency, and ensuring a strong warning when dangerous intrusion occurs.

[0079] 6. Resolve the issue of "poor compatibility with existing systems" to achieve seamless integration and expansion with the existing power plant system. Existing technologies struggle to integrate with power plant SCADA and personnel management systems (such as the two-ticket system), resulting in data silos and low management efficiency. This invention addresses this by completing the interface survey of existing systems during the solution design phase. It supports mainstream protocols such as OPCUA, Modbus TCP / IP, and HTTPAPI, and can interact with the SCADA system in real time to exchange equipment status data and synchronize authorized personnel lists and work order information with the personnel management system. By adopting an industrial ring network (IEEE802.3bt / ATPoE++ switch) and containerized deployment (microservices such as data fusion service and rule engine), when expanding to new control areas or adding sensor types, there is no need to make large-scale modifications to the existing system, reducing system upgrade costs and improving long-term operation and maintenance flexibility.

[0080] 7. Address the issue of "high costs of manual operation and maintenance" and improve the automation and intelligence level of power plant safety management. Existing technologies rely on manual inspections to adjust fences and verify personnel permissions, resulting in high maintenance costs and a high risk of errors. This invention achieves automated management and control through a "digital twin platform + Web monitoring center." In the early stages, a digital twin model of the underground factory was generated using 3D laser scanning, providing a precise foundation for sensor placement and boundary planning; The web-based monitoring center can display personnel locations, security zone boundaries, equipment status, and alarm information in real time, allowing maintenance personnel to grasp the overall control situation without on-site inspections; Security rules are automatically executed (such as automatic boundary adaptation when authorized personnel enter the work area and automatic alerts when unauthorized personnel intrude), which greatly reduces manual intervention, lowers maintenance manpower costs, and avoids control loopholes caused by human error.

[0081] The above embodiments are merely preferred technical solutions of the present invention and should not be considered as limitations on the present invention. The scope of protection of the present invention should be limited to the technical solutions described in the claims, including equivalent substitutions of the technical features described in the claims. That is, equivalent substitutions and improvements within this scope are also within the scope of protection of the present invention.

Claims

1. A dynamic safety area control system for pumped storage power stations based on multi-source fusion positioning, characterized in that, It includes a perception layer, a control layer, and an execution layer, which are connected via an industrial network to collaboratively achieve precise dynamic delineation and real-time response of the safety zone of the pumped storage power station; wherein: The perception layer is used to collect personnel positioning data and environmental monitoring data, including wearable UWB+IMU modules, non-contact millimeter-wave radar and environmental sensors; The control layer, which is the edge computing gateway, runs a built-in multi-source data fusion algorithm to fuse the data collected by the perception layer. Based on the fused positioning data and preset security rules, it dynamically generates partitioning instructions and outputs control signals in JSON format. The execution layer, which receives control signals from the control layer and executes warning actions, includes a laser projection array and a directional sound field system.

2. The dynamic safety area control system for pumped storage power stations based on multi-source fusion positioning as described in claim 1, characterized in that, The wearable UWB+IMU module is integrated into a safety helmet or tool bag, and through ultra-wideband UWB and inertial measurement unit (IMU), it enables accurate perception of the position and movement status of the person wearing the terminal.

3. The dynamic safety area control system for pumped storage power stations based on multi-source fusion positioning according to claim 1, characterized in that, The non-contact millimeter-wave radar detects the presence and location of people not wearing terminals without contact by transmitting and receiving millimeter-wave signals.

4. The dynamic safety area control system for pumped storage power stations based on multi-source fusion positioning according to claim 1, characterized in that, The environmental sensors include millimeter-wave radar and intelligent image recognition cameras, used to monitor unauthorized personnel entering the warning area in real time.

5. The dynamic safety area control system for pumped storage power stations based on multi-source fusion positioning according to claim 1, characterized in that, The multi-source data fusion algorithm is a tightly coupled fusion framework based on Kalman filter (KF) or unscented Kalman filter (UKF), used to fuse UWB, IMU, and radar data.

6. The dynamic safety area control system for pumped storage power stations based on multi-source fusion positioning according to claim 5, characterized in that, The multi-source data fusion algorithm is specifically as follows: To fuse UWB, IMU, and radar data, the state vector of the fusion system is first constructed. It covers core parameters of personnel movement and sensor error compensation, adapts to the multi-interference environment of underground powerhouses in pumped storage power stations, and includes state vectors. The definition is as follows: ； in: This refers to the location of personnel on the floor plan of the underground factory building; The velocity of the person's planar movement; To achieve zero bias in the IMU accelerometer, To achieve zero bias in the IMU gyroscope and compensate for sensor drift caused by factory vibration; When personnel move through the passage at a constant speed with minimal changes in posture, KF rapid fusion is used in two steps: first, the initial position is calculated using IMU data; then, based on the real-time acceleration a and angular velocity ω acquired by the IMU, and combined with the sampling period Δt, the position and velocity are updated. a) Speed ​​update: ; ; b) Location update: ; ; After calculating the initial position using IMU data, UWB / radar is used to correct the error. When UWB or millimeter-wave radar outputs data, the position predicted by the IMU is corrected. a) Observed values: ; b) Error calculation: , representing the difference between the observed value and the predicted value; c) Final position: ; ; Kalman gain; When the personnel's posture changes drastically, the system automatically switches to UKF to handle nonlinear errors, as follows: a) Switching judgment logic: The edge computing gateway monitors the IMU attitude change rate in real time, if... If so, use UKF; otherwise, use KF. b) UKF Calculation: Nonlinearity is handled using three key sampling points: "current predicted position, +0.5m offset position, and -0.5m offset position". Then, the predicted positions of these three sampling points are calculated using IMU data. Finally, UWB / radar observations are combined to weight the three predicted positions, resulting in the final fused position. , Prediction for 3 sampling points coordinate, The same applies to axes.

7. The dynamic safety area control system for pumped storage power stations based on multi-source fusion positioning according to claim 1, characterized in that, The laser projection array consists of RGB laser modules, which can project clear safety zone boundaries onto the ground or walls to indicate safe and dangerous areas.

8. The dynamic safety area control system for pumped storage power stations based on multi-source fusion positioning according to claim 1, characterized in that, The directional sound field system uses parametric array loudspeakers, which can propagate sound in a specific direction and produce clear voice prompts or alarms within a specific area.

9. A control method for a dynamic safety area control system for pumped storage power stations based on multi-source fusion positioning as described in any one of claims 1 to 8, characterized in that, Includes the following steps: Step 1: On-site survey and rule formulation; Step 2: Hardware and software solution design; Step 3: System Deployment and Integration; Step 4: Software deployment and system integration; Step 5: Achieve dynamic boundary generation and multi-level response based on multi-source fusion positioning.

10. The method for dynamic safety area management of pumped storage power stations based on multi-source fusion positioning according to claim 9, characterized in that, Step 1 specifically includes the following sub-steps: a. Area definition: Jointly determine the list of key areas that need to be dynamically managed with the power plant operation and maintenance and safety departments. The key areas include the water pump turbine / generator motor body maintenance area, the inlet ball valve area, and the main transformer area. b. Rule Formulation: Define the security rules for each area, including the security distance under different device states, authorized personnel list, and multi-level response triggering conditions; c. On-site surveying: Use a 3D laser scanner or high-precision total station to create a high-precision three-dimensional point cloud model of the underground plant and main plant, generate a digital twin base, record the building structure, metal equipment distribution, existing network interfaces, power supply point locations in key areas, and assess the electromagnetic environment and potential areas of multipath interference. d. Existing system interface survey: Interact with SCADA system and personnel management system, determine data interface protocol and data format, and clarify the parameters that can be provided. The data interface protocol includes OPCUA, Modbus TCP / IP, and HTTPAPI. The parameters include unit speed, on / off status, and work order information.

11. The method for dynamic safety area management of pumped storage power stations based on multi-source fusion positioning according to claim 10, characterized in that, The safety rules target three key areas of pumped storage power stations: the maintenance area of ​​the pump turbine / generator motor, the inlet ball valve area, and the main transformer area. They combine the dual discrete states of the unit, namely "500r / min operating state" and "0r / min shutdown maintenance state", and clarify the quantitative standards and judgment logic from three aspects: "safe distance, authorized personnel list, and multi-level response triggering conditions" to ensure that the control is executable and verifiable. Core control equipment 1: Water pumps and turbines / generators / motors; The unit is in operating condition, i.e., the SCADA system detects the unit speed as RPM = 500 r / min; the safety distance algorithm formula is: ； ln(RPM) reflects the logarithmic relationship between rotational speed and the danger radius. The higher the rotational speed, the greater the risk of inertial collisions with rotating parts. The constant term 1.2 meters is the average rotor radius of the unit, covering "rotor edge + personnel arm safety redundancy". Unit operating state triggering conditions: When the SCADA system outputs "unit speed = 500 r / min" and "generator motor outlet circuit breaker closing signal = 1", the safety distance is automatically triggered. The laser projection array projects a red boundary line in an area with a radius of 2.5 meters centered at the unit center (x0, y0), which is defined as the "no entry zone". At the same time, it covers the safety isolation range of electrical components such as live junction boxes and excitation devices around the unit. In the shutdown and maintenance state, i.e., when SCADA detects RPM=0r / min and "Anti-rotation safety measure in place signal=1" + "Main circuit power off signal=1" + "Grounding switch closed signal=1", the safety distance algorithm formula is: ； Parameter definitions and values: The outer diameter of the water pump turbine main shaft is taken as 1.75 meters; The operating space for maintenance tools is 1.0 meter. For redundancy in the protection of live compartments, a length of 0.8 meters is used to cover the live compartments of the unit, such as the safety isolation of the stator terminals, to prevent accidental electric shock. To provide a buffer for personnel collaboration and emergency evacuation, a distance of 0.5 meters is recommended. The shutdown maintenance state trigger conditions must be met simultaneously: "Speed ​​= 0 r / min" + "Governor lock engagement signal = 1" + "Main circuit power off = 1" + "Grounding switch closed = 1" + "Work order status = in execution" to trigger the maintenance work area boundary. Only authorized personnel are allowed to enter the boundary, and the boundary must avoid the energized intervals of adjacent equipment. Core control equipment 2: Inlet ball valve; In operation mode, SCADA detects "ball valve fully open signal = 1", and the safe distance is fixed at 2.5 meters. The trigger condition is: when the "ball valve is fully open", the red boundary is triggered, prohibiting personnel from approaching the valve body flange, valve hydraulic mechanism and control box. In the shutdown and maintenance state, i.e., "ball valve fully closed signal = 1" + "working seal engaged signal = 1" + "maintenance seal engaged signal = 1", the safety distance is a fixed value of 1.8 meters. The triggering condition is: "ball valve fully closed signal = 1" + "working seal engaged signal = 1" + "maintenance seal engaged signal = 1", triggering the yellow boundary. The boundary must clearly delineate the "mechanical maintenance area" and the "electrical maintenance area" to prevent accidental entry into the energized interval of the control circuit that has not been de-energized. Core control equipment 3: Main transformer; In the running state, i.e., when SCADA detects "main transformer energized signal = 1", the safe distance algorithm formula is: ； U is the rated voltage of the main transformer, in kV; 0.1×U is the safe distance from the high-voltage electric field, and 0.8 meters is for personnel accidental contact + buffer of energized intervals; triggering condition: triggered by the "main transformer energized" signal, a red boundary is projected, and the boundary must completely cover the energized intervals such as the main transformer bushing, tap changer, and high-voltage leads. Non-electrical inspection personnel are prohibited from entering. In the shutdown and maintenance state, i.e., "main transformer de-energization signal = 1" and "grounding switch closed signal = 1" + "high voltage side voltage test negative = 1", the safety distance calculation formula is: ； 0.05×U is the minimum electrical safety distance under maintenance conditions. 1.0 meter is the tool operating space; the triggering conditions are: "power off" + "grounding switch closed" + "high voltage side voltage test no pressure = 1" + "electrical work permit approved", triggering the yellow boundary.

12. The method for dynamic safety area control of pumped storage power stations based on multi-source fusion positioning according to claim 9, characterized in that, Step 2 specifically includes the following sub-steps: a. Perception layer deployment design: Based on the 3D model, simulation software is used to plan the installation location, height and angle of UWB base stations, millimeter-wave radar and smart cameras to ensure full coverage without blind spots, and optimize the number of base stations to control costs; b. Network Design: Design an industrial ring network to power and transmit data to the front-end sensors, ensuring bandwidth and real-time performance; c. Control layer design: Design a fusion framework based on Kalman filter (KF) or unscented Kalman filter (UKF) to tightly couple and fuse UWB, IMU, and radar data; Develop the core algorithm module, which takes device status and location data as input and outputs dynamic boundary coordinates; design a web-based monitoring center interface to display personnel location, safe zone, device status and alarm information in real time. d. Selection of execution unit: Laser projector and directional sound field speaker are selected as execution units.

13. The method for dynamic safety area control of pumped storage power stations based on multi-source fusion positioning according to claim 9, characterized in that, Step 3 specifically includes the following sub-steps: a. Sensor deployment and calibration: Accurately install UWB base stations and use a total station to measure and record the absolute coordinates of each base station; deploy millimeter-wave radar and cameras, and adjust the angles and coverage; perform joint calibration of sensors, unify timestamps, and ensure spatiotemporal consistency of data; b. Installation of the execution unit: Install the laser projector and directional speaker at a high position on the ceiling or wall to ensure that the projection surface and sound field cover the target area and avoid people looking directly at the laser source.

14. The method for dynamic safety area control of pumped storage power stations based on multi-source fusion positioning according to claim 9, characterized in that, Step 4 specifically includes the following sub-steps: Platform deployment: Deploy a customized operating system and environment on the edge gateway, and deploy microservices such as data fusion service, dynamic boundary engine, rule engine, and WebHMI service in the form of containers; b. System Integration: Develop a data interface client and debug it with the SCADA system to ensure normal status acquisition and control command issuance; Configure the rule engine and enter the security rules defined in step 1.

15. The method for dynamic safety zone management of pumped storage power stations based on multi-source fusion positioning according to claim 9, characterized in that, Step 5, dynamic boundary generation, specifically includes the following sub-steps: a. Data Input: The status signals of the unit are read in real time through the industrial network interface. The status signals are switch signals, including "running" and "stopped" status. The fusion positioning system tracks all personnel wearing positioning tags in real time to obtain accurate coordinate information. b. Boundary generation: The computing engine on the edge computing gateway receives the status signal from SCADA and selects the corresponding security rule according to the unit status. Rule A is used when the unit is in the "running" state, and rule B is used when the unit is in the "stopped" state. c. Command Issuance and Execution: The computing engine generates control commands and sends them to the laser projector and directional speakers via WiFi or Ethernet, projecting a halo around the ground around the unit and issuing a voice warning in the direction of the intruder; d. Real-time monitoring and dynamic adjustment: Monitor the unit status and personnel location in real time. If the unit status changes, query the rule base again to generate a new virtual boundary.

16. The method for dynamic safety area management of pumped storage power stations based on multi-source fusion positioning according to claim 15, characterized in that, Rule A corresponds to the unit's "running" state, i.e., the SCADA-detected speed = 500 r / min, the equipment is energized / rotating, and the core objectives are to prevent collisions with rotating parts, prevent electric shock from high voltage, and prohibit unauthorized personnel from approaching; Rule B corresponds to the unit's "stopped" state, i.e., the SCADA-detected speed = 0 r / min, the equipment is stopped + power off and grounded + anti-rotation safety measures are in place, and the core objectives are to leave sufficient space for tool operation, prevent accidental entry into energized areas, and ensure the safe operation of maintenance personnel.

17. The method for dynamic safety zone management of pumped storage power stations based on multi-source fusion positioning according to claim 9, characterized in that, Step 5, the multi-level response, specifically includes the following sub-steps: a. Warning Zone Response: The warning zone is an area 1m from the boundary. PWM dimming technology is used to achieve a slow flashing effect of the yellow laser to attract people's attention. The directional sound field system uses beamfront synthesis technology to directionally propagate voice prompts to the warning zone, ensuring that people in the area can clearly hear the prompts. b. Intrusion Zone Response: The intrusion zone is the area entering the boundary. The rotational scanning of the red laser is achieved through multi-laser head synchronous control technology. The sound field intensity adaptive algorithm automatically adjusts the sound pressure level according to the environmental noise of the intrusion zone to ensure clear transmission even in noisy environments.

18. The method for dynamic safety zone management of pumped storage power stations based on multi-source fusion positioning according to claim 17, characterized in that, The sound field intensity adaptive algorithm includes the following process: Step 1): Real-time acquisition and preprocessing of ambient noise: The miniature microphone built into the directional sound field system is installed near the speaker to synchronously acquire ambient noise; Preprocessing logic: "Noise filtering" and "mean calculation" are performed on the acquired noise signal; Output parameter: RMS value of ambient noise ; Step 2): Target sound pressure level calculation: Based on the reference sound pressure in the warning zone / intrusion zone, combined with environmental noise and scene compensation, calculate the real-time target sound pressure. The formula is: ； Real-time target sound pressure level; 15 represents the regional reference sound pressure level; 15 represents the target signal-to-noise ratio. This represents the current effective value of the pre-processed environmental noise. The scene compensation coefficient is 0-5dB. When the reverberation is severe in the underground factory, it is taken as 3-5dB, and when the passage is open, it is taken as 0-2dB to offset the sound reflection attenuation.

Images