Acousto-optic alarm system for preventing misoperation of all-insulation inflatable ring main unit

Abstract

The invention relates to the technical field of ring main units, in particular to a sound-light alarm system for preventing misoperation of an all-insulation inflatable ring main unit. Comprising a multi-mode sensing module; an intelligent decision module; a sound-light alarm module; and a mechanical interlocking module. The sound-light-touch three-mode collaborative warning strategy is innovated, and the warning signal identification degree in a complex environment is enhanced through synchronous activation of directional sound waves, double-color stroboscopic light and vibration feedback; by combining dynamic audio adjustment, spectrum adaptive illumination and an operation behavior self-learning algorithm, intelligent dynamic adjustment of a warning strategy is realized, misjudgment of manual operation is effectively reduced, and an efficient risk intervention mechanism covering multiple senses is constructed; according to the invention, through a logic AND operation mechanism of high-voltage live detection and grounding knife switch position monitoring, a core high-risk scene that the power supply side is live and the grounding knife switch is closed is accurately captured, and the problem of missed judgment caused by single condition monitoring in the prior art is effectively solved.

Description

Technical Field

[0001] The invention relates to the technical field of ring network cabinets, in particular to an audible and visual alarm system for preventing misoperation of a fully insulated, pneumatic ring network cabinet. Background Art

[0002] As core equipment for power distribution and fault isolation in distribution networks, fully insulated, pneumatic ring main units (RMUs) ensure operational safety, crucial for reliable grid operation. Existing technologies primarily employ mechanical or electrical interlocks to prevent the risk of misoperation during power-on with a live grounding switch.

[0003] Chinese patent CN201721254288.0 discloses a voice alarm device for preventing the grounding switch of a ring network cabinet from being mistakenly closed. The voice alarm device for preventing the grounding switch of a ring network cabinet from being mistakenly closed includes a ring network cabinet body. The outside of the ring network cabinet body is provided with a live display and a grounding switch operation hole. The live display is connected to an inductive sensor. The inductive sensor is connected to the live side of the ring network cabinet body. The live display includes a three-phase live indicator light, an operation indicator light and a power indicator light. A cover plate is provided on the outside of the grounding switch operation hole. The cover plate covers the outside of the grounding switch operation hole and can slide on the outside of the grounding switch operation hole. A trigger alarm device is provided on the inside of the ring network cabinet body. The cover plate passes through the ring network cabinet body to connect the trigger alarm device. The trigger alarm device is connected to the power supply in the ring network cabinet body, can detect whether the ring network cabinet is energized, and promptly issue an audible and visual alarm, thereby solving the problems encountered in the prior art. Chinese patent CN201922217930.3 discloses a forced locking device for preventing the grounding knife switch of a distribution network ring network cabinet from being closed by mistake. The forced locking device for preventing the grounding knife switch of a distribution network ring network cabinet from being closed by mistake comprises an electromagnetic locking device (1), a baffle (3) and an alarm device (4); the utility model can detect whether the primary high-voltage equipment in the ring network cabinet is energized, and promptly issue an audible and visual alarm to implement forced locking. A baffle is installed on the operating hole of the grounding knife switch. When the ring network cabinet is in operation, a voice alarm is used to remind the operator, and forced locking is implemented to prevent misoperation; when the primary equipment is energized, once the baffle of the operating hole of the grounding knife switch is opened, the alarm device immediately issues an audible and visual alarm, and forced locking is implemented to prevent misoperation, thereby protecting the safety of equipment and personnel.

[0004] Although the above-mentioned existing technologies all have their own advantages, the above-mentioned schemes still have the following shortcomings in terms of risk identification logic, structural adaptability and warning effectiveness: 1. The environmental adaptability of the warning system is insufficient, and there is a lack of multi-dimensional and adjustable sound and light warning strategies. The warning signal is easily affected by environmental interference, resulting in delayed response by the operator; 2. The collaborative monitoring of dual risk conditions is missing. The existing technology only independently judges the energized state or the position of the knife switch, and has not established the logic and relationship of "the power supply side is energized and the grounding knife switch is closed", and cannot accurately identify the high-risk scenario of "power supply with grounding knife switch"; 3. Compatibility limitations of complex installation conditions: relying on mechanical interlocking or simple electrical control in the internal space of the circuit breaker, it cannot meet the anti-error requirements of ring network cabinets with miniaturized and high-insulation designs; in view of this, we propose an sound and light alarm system to prevent misoperation of fully insulated inflatable ring network cabinets. Summary of the Invention

[0005] The purpose of the present invention is to provide an audible and visual alarm system for preventing misoperation of a fully insulated pneumatic ring main unit, so as to solve the problems raised in the above-mentioned background technology.

[0006] In order to solve the above technical problems, the present invention aims to provide an audible and visual alarm system to prevent misoperation of a fully insulated pneumatic ring main unit, comprising: The multimodal sensing module integrates a high-voltage live detection unit, a gas pressure monitoring unit, a mechanical state sensing unit, and a temperature monitoring unit. It is used to collect high-voltage side voltage signals, SF6 gas pressure values, operating mechanism displacement, and contact temperature data in real time, and implements data credibility assessment through a redundant verification mechanism. The intelligent decision-making module has a built-in parallel verification algorithm based on multi-dimensional safety conditions and edge computing nodes. It is equipped with an operation behavior feature map library and a device status knowledge base to fuse perception data and generate operation permission signals or alarm control signals based on a dynamic weight allocation strategy; An audio-visual alarm module, electrically connected to the intelligent decision-making module, is used to perform graded audio-visual warnings, support audio-visual-tactile tri-modal coordinated warnings and environmental adaptive adjustment; A mechanical interlocking module is electrically connected to the intelligent decision-making module and integrates a dual locking structure of differential braking and worm gear to achieve physical locking and hot-swap maintenance of the operating mechanism; The communication linkage module is electrically connected to the intelligent decision-making module, is used to realize remote alarm and system linkage, and supports adaptive switching of multi-protocol communication gateways.

[0007] As a further improvement of this technical solution, the mechanical state sensing unit includes: The non-contact displacement sensor array arranged circumferentially around the grounding knife operating hole uses magnetoresistive displacement sensors to identify the insertion depth of the operating tool through electromagnetic induction and construct a three-dimensional spatial trajectory model; the three-dimensional insertion trajectory of the operating tool is constructed through the principle of electromagnetic induction with a resolution of 0.1mm, and can identify abnormal operations with tool insertion angle deviation greater than 15°.

[0008] The dynamic pressure sensor group integrated in the operating handle shaft detects changes in rotational torque based on a piezoelectric film array and generates a torque-time gradient curve. This design can collect rotational torque in real time and generate a "torque-time gradient curve" with an accuracy of 0.1N·m. When the torque sudden change value is greater than 20N·m, it will be marked as "suspected forced operation."

[0009] The intelligent decision-making module establishes an operation behavior feature map by analyzing the temporal relationship between insertion depth and torque, and combines a convolutional neural network to distinguish between normal operations and abnormal forced operations.

[0010] As a further improvement of this technical solution, the multimodal perception module is further integrated with a data credibility evaluation unit, which adopts DS evidence theory to fuse multi-sensor data and implements data verification through the following steps: Build the sensor confidence matrix: ,in For the The accuracy of historical data of each sensor; Build conflict coefficient threshold , when the data conflict The redundant sensor voting mechanism is triggered to ensure the collected data credibility is ≥99.5%. In this design, the DS evidence theory is used to build a data credibility assessment unit. By establishing a sensor confidence matrix, setting a conflict coefficient threshold, and implementing a redundant sensor voting mechanism, the problem of false positives and false negatives from a single sensor is resolved. This ensures the credibility of key data such as high-voltage side voltage and gas pressure is ≥99.5%, improving monitoring robustness in complex environments.

[0011] As a further improvement of this technical solution, the multi-dimensional security conditions include: First verification condition: the output value of the voltage transformer on the high-voltage side is lower than the preset lockout threshold and lasts for ≥500ms; Second verification condition: The SF6 gas pressure value is within the dynamic safety range of 0.42-0.55MPa, and the pressure trend prediction is performed based on the Kalman filter algorithm; Third verification condition: The cabinet door lock sensor returns a valid lock signal and the door gap displacement is ≤0.5mm; Fourth verification condition: The operation process complies with the preset "five-prevention" interlocking logic sequence, and the integrity of the operation instructions is verified by the hash algorithm; The intelligent decision-making module uses a weighted voting algorithm to make collaborative judgments on the four conditions, where the weight distribution formula is: , where For the The importance coefficient of each safety condition is dynamically optimized through a reinforcement learning framework. The design defines four verification conditions (powered state, gas pressure, cabinet door locking, and operating procedures) and uses a weighted voting algorithm for parallel verification. If any condition is not met, an interlock response is triggered.

[0012] As a further improvement of this technical solution, the intelligent decision-making module is configured with a dynamic self-learning mechanism, including: The weight optimization unit, based on a deep reinforcement learning framework and Bayesian network model, dynamically adjusts the weight distribution of multi-dimensional safety conditions by analyzing historical misoperation events and successful interception records, and generates a weight update confidence report. This design can adjust the safety condition weights based on historical misoperation data; The feature map update unit integrates a spatiotemporal clustering engine and adversarial training module to iteratively optimize the operational behavior feature map library and set version rollback thresholds. The feature map update unit iteratively optimizes the operational behavior model through spatiotemporal clustering and adversarial training, supporting the recognition of new abnormal patterns. The verification and rollback unit connects to the digital twin system for offline verification and automatically triggers historical version recovery when the error rate after an update exceeds a preset threshold. This design ensures the reliability of algorithm updates and reduces the error rate.

[0013] As a further improvement of this technical solution, the edge computing node of the intelligent decision-making module includes: FPGA accelerator, used for real-time processing of raw sensor signals and wavelet transform noise reduction, supporting parallel processing of 16 channels of sensor data; Local knowledge base, which stores device historical operation records and abnormal pattern characteristics, and configures the LRU cache elimination algorithm to optimize storage efficiency; The edge node executes a three-level decision-making process, including sensor-level anomaly filtering, feature-level risk identification, and system-level coordinated response. In this design, an FPGA accelerator processes 16 channels of sensor data in real time, using wavelet transforms for noise reduction (noise suppression ratio ≥ 20dB). This three-level decision-making process, consisting of sensor-level filtering, feature-level identification, and system-level coordinated response, achieves a response latency of ≤50ms, meeting real-time control requirements. The local knowledge base is configured with an LRU cache to optimize historical data storage efficiency and support rapid feature matching.

[0014] As a further improvement of this technical solution, the sound and light alarm module includes: Programmable strobe warning unit, including red and yellow LED array, supports Morse code mode and synchronized strobe strategy; Intelligent audio unit, including directional sound wave transmitter and voice interaction module, can generate 120dB pulse warning sound waves and multi-language voice prompts; The sound and light alarm module synchronizes and activates the tactile feedback unit based on the risk level signal, forming a multi-sensory coordinated warning mechanism. This design can form a tri-modal alarm of sound, light, and touch, covering visual, auditory, and tactile multi-sensory stimulation, improving the recognition of warning signals in complex environments.

[0015] As a further improvement of this technical solution, the sound and light alarm module further includes: The multimodal alarm coordination unit includes a dynamic audio adjustment circuit and a spectrum-adaptive LED array. The dynamic audio adjustment circuit uses PWM modulation technology to control the buzzer, supporting 85-120dB volume gradient adjustment. The integrated speech synthesis chip generates multilingual warning instructions. The spectrum-adaptive LED array features a red / yellow dual-color LED light group, automatically adjusts brightness via an ambient light sensor, and supports Morse code and synchronized strobe warning modes. The dynamic audio adjustment circuit supports 85-120dB volume gradient adjustment, and the spectrum-adaptive LED automatically adjusts brightness (200-1000cd / m²) via an ambient light sensor to maintain warning effectiveness in strong light and high noise environments (e.g., signal-to-noise ratio ≥15dB in 80dB noise). The intelligent learning and optimization unit includes an operation mode self-learning engine and an environmental noise suppression algorithm. The operation mode self-learning engine uses LSTM to analyze historical operation data, establish a normal operation behavior feature library, and trigger a pre-alarm in real time for actions that deviate from the feature threshold. The environmental noise suppression algorithm uses wavelet transform noise reduction technology. The operation mode self-learning engine uses LSTM to establish a normal behavior feature library, and triggers a pre-alarm in real time for actions that deviate from the threshold, reducing human misjudgment. As a further improvement of this technical solution, the mechanical interlocking module includes: The dual-redundant braking unit consists of electromagnetic brakes symmetrically arranged on both sides of the operating handle shaft. It uses a differential braking strategy to apply reverse braking torque, and the braking torque gradient is adjustable; A mechanical locking unit, comprising a worm gear transmission mechanism connected to a permanent magnet synchronous motor, integrating a torque limiter and a displacement feedback sensor to achieve closed-loop control of the operating lever locking position; The electromagnetic brake and worm gear mechanism form an electromechanical dual-locking structure, supporting a fail-safe mode and hot-swap maintenance mechanism. This design utilizes a symmetrically arranged electromagnetic brake, employing a differential braking strategy to apply a reverse braking torque (with an adjustable gradient). This dual-locking mechanism, combined with the worm gear mechanical locking unit (which integrates an integrated torque limiter and displacement feedback sensor), provides a locking force ≥ 200 N·m, preventing abnormal operating forces from breaking the lock. This supports a fail-safe mode (automatic locking upon power failure) and avoids the risk of mechanical interlock failure. As a further improvement of this technical solution, the communication linkage module includes: Multi-protocol communication gateway supports adaptive switching of IEC61850, ModbusTCP and MQTT protocols, and integrates time-sensitive network mechanism to ensure real-time communication; The intelligent linkage control unit automatically triggers the vortex forced exhaust device and sends a secondary alarm signal to the fire protection system when the laser spectral analysis unit detects that the SF6 gas concentration in the cabinet is greater than 1000ppm. It also packages and pushes the equipment location, fault type, and real-time infrared thermal imaging data to the operation and maintenance terminal. This design supports real-time early warning of the fire protection system and the operation and maintenance terminal, improving fault handling efficiency. The laser spectral analysis is combined with the edge AI acceleration chip to process gas concentration data in real time, output the leakage gradient distribution, and assist in accurately locating the leak point.

[0016] Compared with the prior art, the present invention has the following beneficial effects: 1. This invention innovates a trimodal collaborative sound-light-tactile warning strategy. By synchronizing directional sound waves, two-color strobe lighting, and vibration feedback, it enhances the discernibility of warning signals in complex environments. Combined with dynamic audio modulation, spectrally adaptive lighting, and a self-learning algorithm for operational behavior, it enables intelligent dynamic adjustment of warning strategies, effectively reducing human operator misjudgment and establishing an efficient risk intervention mechanism encompassing multiple senses. 2. This invention uses a logical AND operation mechanism combining high-voltage live detection and grounding switch position monitoring to accurately capture the core high-risk scenario of "live power on the power supply side and closed grounding switch," effectively resolving the problem of missed detections in existing single-condition monitoring technologies. Furthermore, it incorporates equipment operating status parameters (such as insulating gas pressure and contact temperature) into the risk assessment system, constructing a three-dimensional monitoring matrix of "electrical status, mechanical position, and equipment health." This significantly expands the risk identification dimension and enables the transition from single-condition monitoring to collaborative analysis of multi-dimensional safety parameters. 3. The present invention adopts a "differential braking + worm gear" dual locking structure, which does not require the modification of the internal space of the ring main unit and can be flexibly adapted to miniaturized, high-insulation ring main unit products. At the same time, through the integrated design of non-contact magnetoresistive displacement sensor and piezoelectric film pressure sensor, damage to the cabinet sealing structure is avoided, and non-invasive monitoring of operating behavior is achieved, effectively improving the reliability of the sensor and system compatibility under complex working conditions, and meeting the high protection requirements of fully insulated inflatable ring main units. BRIEF DESCRIPTION OF THE DRAWINGS

[0017] Figure 1 It is a system framework diagram of the present invention; Figure 2 This is a system framework diagram of the sound and light alarm module in the present invention. DETAILED DESCRIPTION

[0018] The following will provide a clear and complete description of the technical solutions in the embodiments of the present invention in conjunction with the accompanying drawings. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. All other embodiments obtained by ordinary technicians in this field based on the embodiments of the present invention without making any creative efforts shall fall within the scope of protection of the present invention.

[0019] like Figure 1 As shown, this embodiment provides an audible and visual alarm system for preventing misoperation of a fully insulated pneumatic ring main unit, comprising: a multimodal sensing module integrating a high-voltage live detection unit, a gas pressure monitoring unit, a mechanical state sensing unit, and a temperature monitoring unit, for real-time acquisition of high-voltage side voltage signals, SF6 gas pressure values, operating mechanism displacement, and contact temperature data, and implementing data credibility assessment through a redundant check mechanism; It's understandable that by collaborating with the high-voltage live detection unit (monitoring the live state on the incoming line) and the mechanical state sensing unit (monitoring the displacement of the grounding switch operating mechanism and identifying the switch's "closed" state), the logic and relationship of "live power on the power supply side and closed grounding switch" is directly established, resolving the core issue of the existing technology's "lack of dual risk condition monitoring" and precisely defining the triggering conditions for "power transmission with grounding switch." Simultaneously, the integration of the gas pressure monitoring unit (SF6 pressure) and the temperature monitoring unit (contact temperature) incorporates equipment operating status (such as insulating gas leakage and contact overheating) into the risk assessment system, forming a three-dimensional monitoring matrix of "electrical state + mechanical position + equipment health." Compared to existing technologies that focus solely on single or dual conditions, this approach provides a more comprehensive risk identification dimension.

[0020] In this embodiment, the high-voltage live detection unit uses a capacitive voltage sensor (model: CHV-25P). Based on the voltage division principle, it collects high-voltage side voltage signals in real time with a detection accuracy of ±0.5%FS. A detection value greater than 50V (the preset live threshold) indicates that the power supply side is live. The "power supply with grounding switch on" risk warning is triggered only when the "high-voltage live detection unit output value" is greater than 50V and the "mechanical state sensing unit detects that the grounding switch operating hole insertion depth is greater than 20mm (the closed position threshold)."

[0021] In this embodiment, SF6 gas pressure monitoring uses a microelectromechanical sensor (MEMS, model: MS5540C) with a detection range of 0-1MPa, an accuracy of ±1.5%FS, and a dynamic safety interval set to 0.42-0.55MPa. At the same time, contact temperature monitoring uses a distributed fiber optic temperature sensor (DTS, model: OFDR-1550), which is arranged along the contact conductive loop and has a temperature measurement accuracy of ±1°C. When the temperature is greater than 150°C and persists for 10 seconds, an over-temperature warning is triggered.

[0022] In this embodiment, the mechanical state sensing unit comprises a non-contact displacement sensor array circumferentially arranged around the grounding knife operating hole. These magnetoresistive displacement sensors use electromagnetic induction to identify the tool's insertion depth and construct a three-dimensional trajectory model. A dynamic pressure sensor group integrated into the operating handle's rotating shaft uses a piezoelectric thin film array to detect changes in rotational torque and generate a torque-time gradient curve. The intelligent decision-making module analyzes the temporal relationship between insertion depth and torque to establish a characteristic map of operating behavior and, combined with a convolutional neural network, distinguishes between normal and abnormal forced operation. The circumferentially arranged magnetoresistive sensors in this design use electromagnetic induction to identify the tool's insertion depth and construct a three-dimensional trajectory model. This accurately captures tool insertion angle and depth anomalies (such as incomplete insertion), avoids contact sensor wear, and adapts to the compact ring main unit sealing design. A second dynamic pressure sensor group uses a piezoelectric thin film array to detect the handle's rotational torque and generate a torque-time gradient curve. This monitors operating force, speed, and smoothness in real time (with a resolution of up to 0.1 N·m), identifying torque spikes characteristic of abnormal operations such as forced engagement. At the same time, the intelligent decision-making module combines the temporal relationship between insertion depth and torque, and establishes an operational behavior feature map through a convolutional neural network to distinguish normal operations (smooth curves) from abnormal forced operations (torque surges, trajectory deviations). The recognition accuracy rate is ≥98%, solving the problem of traditional fixed threshold algorithms missing new abnormal operations.

[0023] In this embodiment, the multimodal perception module is further integrated with a data credibility evaluation unit, which adopts DS evidence theory to fuse multi-sensor data and implements data verification through the following steps: establishing a sensor confidence matrix: ,in For the The accuracy of historical data of sensors; build conflict coefficient threshold , when the data conflict The redundant sensor voting mechanism is triggered to ensure that the credibility of collected data (such as high-voltage side voltage, gas pressure, and other key data) is ≥99.5%. This design uses DS evidence theory to construct a data credibility assessment unit. By establishing a sensor confidence matrix, setting a conflict coefficient threshold, and implementing a redundant sensor voting mechanism, it solves the problem of false positives and false negatives from single sensors, ensuring the credibility of key data such as high-voltage side voltage and gas pressure is ≥99.5%, and improving monitoring robustness in complex environments.

[0024] It should be added that the conflict coefficient threshold is constructed , through the formula , Calculate data conflict degree ,when The redundant sensor voting mechanism is triggered when a decision is made, and the multi-sensor data is subsequently fused using the Dempster synthesis rule to ensure that the collected data has a credibility of ≥99.5%. Furthermore, as the core algorithm of DS evidence theory, the Dempster synthesis rule is specifically designed to fuse the trust distribution of multiple independent evidence sources to resolve the information fusion problem when multi-sensor data conflicts. In this embodiment, this rule is deeply applied to the data credibility assessment unit of the multimodal perception module. Through precise mathematical modeling and logical judgment mechanisms, it effectively improves the reliability of key data such as high-voltage side voltage and gas pressure, providing solid data support for subsequent intelligent decision-making.

[0025] Furthermore, the intelligent decision-making module has a built-in parallel verification algorithm based on multi-dimensional safety conditions and edge computing nodes, and is equipped with an operation behavior feature map library and equipment status knowledge base, which is used to fuse perception data and generate operation permission signals or alarm control signals based on dynamic weight allocation strategies; In this embodiment, the multi-dimensional safety conditions include: a first verification condition: the output value of the high-voltage side voltage transformer is lower than the preset locking threshold and the duration is ≥500ms; a second verification condition: the SF6 gas pressure value is in the dynamic safety range of 0.42-0.55MPa, and the pressure trend is predicted based on the Kalman filter algorithm; a third verification condition: the cabinet door locking sensor returns a valid locking signal and the door gap displacement is ≤0.5mm; a fourth verification condition: the operation process complies with the preset "five-prevention" interlocking logic timing (i.e., preventing the accidental opening / closing of the circuit breaker, preventing the disconnector from pulling / closing under load, preventing the grounding wire from being connected under power, preventing the switch from closing with the grounding wire, and preventing the accidental entry into the energized interval), and the integrity of the operation instruction is verified by the hash algorithm; the intelligent decision-making module uses a weighted voting algorithm to make a collaborative judgment on the four conditions, where the weight distribution formula is: , where For the The importance coefficient of each safety condition is dynamically optimized through the reinforcement learning framework; for example, during normal operation, the high-voltage side voltage (live state) If the abnormal gas pressure causes misoperation frequently, the system will adjust through reinforcement learning. The decision weight of the gas pressure condition is increased to 0.3, adapting to the changing risk levels under different operating conditions and avoiding misjudgments based on a single condition. This design defines four verification conditions (power status, gas pressure, cabinet door lock, and operating procedures) and uses a weighted voting algorithm for parallel verification. Failure of any condition triggers an interlock response. The weights are dynamically optimized through reinforcement learning to adapt to the changing risk levels under different operating conditions and avoid misjudgments based on a single condition.

[0026] In this embodiment, the intelligent decision-making module is configured with a dynamic self-learning mechanism, including: a weight optimization unit, based on a deep reinforcement learning framework and a Bayesian network model, which dynamically adjusts the weight distribution of multi-dimensional security conditions by analyzing historical misoperation events and successful interception records, and generates a weight update confidence report; a feature map update unit, which integrates a spatiotemporal clustering engine and an adversarial training module, and is used for iterative optimization of the operational behavior feature map library, and sets a version rollback threshold; a verification and rollback unit, which is connected to the digital twin system for offline verification, and automatically triggers historical version recovery when the misjudgment rate after the update exceeds a preset threshold. In this design, the weight optimization unit adjusts the safety condition weights based on historical misoperation data based on deep reinforcement learning and Bayesian networks; the feature map update unit iteratively optimizes the operational behavior model through spatiotemporal clustering and adversarial training, and supports new abnormal pattern recognition; the verification and rollback unit is connected to the digital twin system to ensure the reliability of algorithm updates and reduce the misjudgment rate.

[0027] In this embodiment, the edge computing node of the intelligent decision-making module includes: an FPGA accelerator for real-time processing of raw sensor signals and wavelet transform noise reduction, supporting parallel processing of 16 channels of sensor data; a local knowledge base that stores historical device operation records and abnormal pattern characteristics, and configures an LRU cache elimination algorithm to optimize storage efficiency; the edge node executes a three-level decision-making process, including sensor-level anomaly filtering, feature-level risk identification, and system-level coordinated response. In this design, the FPGA accelerator processes 16 channels of sensor data in real time, performs wavelet transform noise reduction (noise suppression ratio ≥ 20dB), and executes a three-level decision-making process of "sensor-level filtering → feature-level identification → system-level coordinated response" with a response latency of ≤50ms, meeting real-time control requirements. The local knowledge base is configured with an LRU cache to optimize historical data storage efficiency and support fast feature matching.

[0028] The FPGA accelerator uses a Xilinx Zynq-7000 chip, supporting parallel processing of 16 channels of sensor data. It uses wavelet transform (DB4 wavelet basis) for noise reduction, achieving a noise suppression ratio of ≥20dB, and processes raw sensor signals in real time with a data processing latency of ≤10ms. The local knowledge base stores ≥10 years of historical operation records and abnormal pattern features, and uses a LRU cache eviction algorithm (1GB cache capacity) to optimize storage efficiency. Feature matching time is ≤5ms, supporting fast feature matching.

[0029] like Figure 2 As shown, further, the sound and light alarm module is electrically connected to the intelligent decision-making module, and is used to perform graded sound and light warnings, support sound-light-tactile tri-modal coordinated warnings and environmental adaptive adjustment; In this embodiment, the audio-visual alarm module includes a programmable strobe warning unit, comprising a red / yellow LED array that supports Morse coding and synchronized strobe patterns; an intelligent audio unit, comprising a directional acoustic wave transmitter and a voice interaction module, capable of generating 120dB pulsed warning sound waves and multilingual voice prompts; and a tactile feedback unit that synchronizes activation of the risk level signal, creating a multi-sensory coordinated alarm mechanism. In this design, the programmable strobe warning unit (red / yellow LED array that supports Morse coding and synchronized strobe patterns), the intelligent audio unit (120dB pulsed sound waves, multilingual voice prompts), and the tactile feedback unit (vibrating handle) are activated collaboratively to create a tri-modal audio-visual-tactile alarm. This provides visual, auditory, and tactile multi-sensory stimulation, enhancing the discernibility of warning signals in complex environments.

[0030] In this embodiment, the tactile feedback unit primarily consists of a vibration motor (Model: VM-20, adjustable frequency 50-200Hz) integrated into the operating handle. This motor is activated synchronously with the risk level signal. When the risk is low, the vibration frequency is 50Hz; when the risk is high, the vibration frequency increases to 200Hz. This tactile stimulation enhances the warning effect, forming a multi-sensory collaborative warning mechanism, significantly improving the recognition of warning signals in complex environments.

[0031] In this embodiment, the sound and light alarm module also includes: a multimodal alarm coordination unit, including a dynamic audio adjustment circuit and a spectrum-adaptive LED array; the dynamic audio adjustment circuit uses PWM modulation technology to control the buzzer, supports 85-120dB volume gradient adjustment, and an integrated speech synthesis chip generates multilingual warning instructions; the spectrum-adaptive LED array is equipped with a red / yellow dual-color LED light group, automatically adjusts the brightness through the ambient light sensor, and supports two warning modes: Morse code and synchronized strobe; an intelligent learning and optimization unit, including an operation mode self-learning engine and an environmental noise suppression algorithm: the operation mode self-learning engine uses LSTM to analyze historical operation data, establish a normal operation behavior feature library, and trigger a pre-alarm in real time for actions that deviate from the feature threshold; the environmental noise suppression algorithm uses wavelet transform noise reduction technology. In background noise above 80dB, the sound and light alarm signal signal-to-noise ratio can still be maintained at ≥15dB. In this design, the dynamic audio adjustment circuit supports 85-120dB volume gradient adjustment, and the spectrum-adaptive LED automatically adjusts brightness (200-1000cd / m²) through the ambient light sensor, maintaining a warning effect in strong light / high noise scenarios (for example, the signal-to-noise ratio in 80dB noise is ≥15dB). The operation mode self-learning engine establishes a normal behavior feature library based on LSTM, and provides real-time pre-alarms for operations that deviate from the threshold, reducing human error.

[0032] Furthermore, a mechanical interlocking module is electrically connected to the intelligent decision-making module, and integrates a differential braking and a worm gear dual locking structure to realize physical locking and hot-swap maintenance of the operating mechanism; in this embodiment, the mechanical interlocking module includes: a dual redundant braking unit, which is composed of electromagnetic brakes symmetrically arranged on both sides of the rotating shaft of the operating handle, and adopts a differential braking strategy to apply a reverse braking torque, and the braking torque gradient is adjustable; a mechanical locking unit, which includes a worm gear transmission mechanism connected to a permanent magnet synchronous motor, and integrates a torque limiter and a displacement feedback sensor to realize closed-loop control of the locking position of the operating lever; the electromagnetic brake and the worm gear mechanism form an electromechanical dual locking structure, which supports a fault safety mode and a hot-swap maintenance mechanism. Symmetrically arranged electromagnetic brakes employ a differential braking strategy to apply reverse braking torque (with adjustable gradient). Together with the worm gear mechanical locking unit (with integrated torque limiter and displacement feedback sensor), this creates a dual electromechanical locking mechanism with a locking force of ≥200 N·m, preventing abnormal operating forces from breaching the lock. A fail-safe mode (automatic locking upon power failure) is also supported, mitigating the risk of mechanical interlock failure. For example, a 10kV ring main unit (model: XGN15-12) in a chemical industrial park automatically entered a power-off lock state during a temporary power outage, preventing accidents caused by inadvertent lever operation. Upon power restoration, the system automatically released the lock in ≤100ms. This design also supports hot-swappable maintenance, enabling replacement of faulty components without powering down, reducing operational costs. The worm gear mechanism is driven by a permanent magnet synchronous motor, enabling closed-loop control of the locking position, enhancing the lifespan and accuracy of the mechanical structure.

[0033] Furthermore, the communication linkage module is electrically connected to the intelligent decision-making module to realize remote alarm and system linkage, and supports adaptive switching of multi-protocol communication gateways.

[0034] It is understandable that the multi-protocol communication gateway supports adaptive switching of IEC61850 (version 2.0), ModbusTCP (RTU mode), and MQTT (v3.1.1) protocols, integrates time-sensitive networking (TSN), has time synchronization accuracy ≤1μs, communication delay ≤10ms, and is compatible with power private networks, industrial buses, and IoT platforms, enabling seamless integration of local decision-making and remote monitoring.

[0035] In this embodiment, the communication linkage module includes: a multi-protocol communication gateway that supports adaptive switching of IEC61850, ModbusTCP, and MQTT protocols, and integrates a time-sensitive network mechanism to ensure real-time communication; an intelligent linkage control unit that automatically triggers the vortex-type forced exhaust device and sends a secondary alarm signal to the fire protection system when the laser spectrum analysis unit detects that the SF6 gas concentration in the cabinet is greater than 1000ppm, and at the same time packages and pushes the equipment location, fault type, and real-time infrared thermal imaging data to the operation and maintenance terminal. In this design, when the intelligent linkage control unit detects that the SF6 concentration is greater than 1000ppm, it automatically triggers the vortex-type forced exhaust device and pushes an alarm data packet containing infrared thermal imaging (including equipment location and fault type), supporting real-time early warning of the fire protection system linkage and the operation and maintenance terminal, thereby improving fault handling efficiency; the laser spectrum analysis is combined with the edge AI acceleration chip to process gas concentration data in real time, output the leakage gradient distribution, and assist in accurately locating the leak point.

[0036] Those skilled in the art will appreciate that the process of implementing all or part of the steps of the above embodiments may be accomplished by hardware, or by instructing related hardware through a program, which may be stored in a computer-readable storage medium.

[0037] The above shows and describes the basic principles, main features, and advantages of the present invention. Those skilled in the art should understand that the present invention is not limited to the above embodiments. The above embodiments and descriptions are merely preferred examples of the present invention and are not intended to limit the present invention. Various changes and improvements may be made to the present invention without departing from the spirit and scope of the present invention. Such changes and improvements fall within the scope of the present invention. The scope of protection claimed in the present invention is defined by the appended claims and their equivalents.

Claims

1. To prevent the misoperation of the fully insulated inflatable ring network cabinet, the sound and light alarm system is characterized by: include: The multimodal sensing module integrates a high-voltage live detection unit, a gas pressure monitoring unit, a mechanical state sensing unit, and a temperature monitoring unit. It is used to collect high-voltage side voltage signals, SF6 gas pressure values, operating mechanism displacement, and contact temperature data in real time, and implements data credibility assessment through a redundant verification mechanism. The intelligent decision-making module has a built-in parallel verification algorithm based on multi-dimensional safety conditions and edge computing nodes. It is equipped with an operation behavior feature map library and a device status knowledge base to fuse perception data and generate operation permission signals or alarm control signals based on a dynamic weight allocation strategy; An audio-visual alarm module, electrically connected to the intelligent decision-making module, is used to perform graded audio-visual warnings, support audio-visual-tactile tri-modal coordinated warnings and environmental adaptive adjustment; A mechanical interlocking module is electrically connected to the intelligent decision-making module and integrates a dual locking structure of differential braking and worm gear to achieve physical locking and hot-swap maintenance of the operating mechanism; The communication linkage module is electrically connected to the intelligent decision-making module, is used to realize remote alarm and system linkage, and supports adaptive switching of multi-protocol communication gateways.

2. The sound and light alarm system for preventing misoperation of a fully insulated pneumatic ring main unit according to claim 1 is characterized in that: The mechanical state sensing unit includes: A non-contact displacement sensor array arranged circumferentially around the grounding knife operating hole uses magnetoresistive displacement sensors to identify the insertion depth of the operating tool through electromagnetic induction and construct a three-dimensional spatial trajectory model; The dynamic pressure sensor group integrated into the operating handle shaft detects the change in rotational torque based on the piezoelectric film array and generates a torque-time gradient curve; The intelligent decision-making module establishes an operation behavior feature map by analyzing the temporal relationship between insertion depth and torque, and combines a convolutional neural network to distinguish between normal operations and abnormal forced operations.

3. The sound and light alarm system for preventing misoperation of a fully insulated pneumatic ring main unit according to claim 1 is characterized in that: The multimodal perception module is also integrated with a data credibility assessment unit, which adopts DS evidence theory to fuse multi-sensor data and implements data verification through the following steps: Build the sensor confidence matrix: ,in For the The accuracy of historical data of each sensor; Build conflict coefficient threshold , when the data conflict The redundant sensor voting mechanism is triggered to ensure the credibility of the collected data is ≥ 99.5%.

4. The sound and light alarm system for preventing misoperation of a fully insulated pneumatic ring main unit according to claim 1 is characterized in that: The multi-dimensional security conditions include: First verification condition: the output value of the voltage transformer on the high-voltage side is lower than the preset lockout threshold and lasts for ≥500ms; Second verification condition: The SF6 gas pressure value is within the dynamic safety range of 0.42-0.55MPa, and the pressure trend prediction is performed based on the Kalman filter algorithm; Third verification condition: The cabinet door lock sensor returns a valid lock signal and the door gap displacement is ≤0.5mm; Fourth verification condition: The operation process complies with the preset "five-prevention" interlocking logic sequence, and the integrity of the operation instructions is verified by the hash algorithm; The intelligent decision-making module uses a weighted voting algorithm to make collaborative judgments on the four conditions, where the weight distribution formula is: , where For the The importance coefficients of each safety condition are dynamically optimized through the reinforcement learning framework.

5. The sound and light alarm system for preventing misoperation of a fully insulated pneumatic ring main unit according to claim 1 is characterized in that: The intelligent decision-making module is equipped with a dynamic self-learning mechanism, including: The weight optimization unit, based on a deep reinforcement learning framework and Bayesian network model, dynamically adjusts the weight distribution of multi-dimensional security conditions by analyzing historical misoperation events and successful interception records, and generates a weight update confidence report; The feature map update unit integrates a spatiotemporal clustering engine and an adversarial training module to iteratively optimize the operational behavior feature map library and set the version rollback threshold. The verification and rollback unit connects to the digital twin system for offline verification and automatically triggers historical version recovery when the misjudgment rate after the update exceeds the preset threshold.

6. The sound and light alarm system for preventing misoperation of a fully insulated pneumatic ring main unit according to claim 1 is characterized in that: The edge computing nodes of the intelligent decision-making module include: FPGA accelerator, used for real-time processing of raw sensor signals and wavelet transform noise reduction, supporting parallel processing of 16 channels of sensor data; Local knowledge base, which stores device historical operation records and abnormal pattern characteristics, and configures the LRU cache elimination algorithm to optimize storage efficiency; The edge node executes a three-level decision-making process, including sensor-level anomaly filtering, feature-level risk identification, and system-level linkage response.

7. The sound and light alarm system for preventing misoperation of a fully insulated pneumatic ring main unit according to claim 1 is characterized in that: The sound and light alarm module comprises: Programmable strobe warning unit, including red and yellow LED array, supports Morse code mode and synchronized strobe strategy; Intelligent audio unit, including directional sound wave transmitter and voice interaction module, can generate 120dB pulse warning sound waves and multi-language voice prompts; The sound and light alarm module synchronously activates the tactile feedback unit according to the risk level signal, forming a multi-sensory collaborative alarm mechanism.

8. The sound and light alarm system for preventing misoperation of a fully insulated pneumatic ring main unit according to claim 1 is characterized in that: The sound and light alarm module also includes: The multimodal alarm coordination unit includes a dynamic audio adjustment circuit and a spectrum-adaptive LED array. The dynamic audio adjustment circuit uses PWM modulation technology to control the buzzer, supporting 85-120dB volume gradient adjustment, and an integrated speech synthesis chip to generate multilingual warning instructions. The spectrum-adaptive LED array is equipped with a red / yellow dual-color LED light group, which automatically adjusts brightness through an ambient light sensor and supports Morse code and synchronized strobe warning modes. The intelligent learning and optimization unit includes an operation mode self-learning engine and an environmental noise suppression algorithm: the operation mode self-learning engine analyzes historical operation data based on LSTM, establishes a normal operation behavior feature library, and triggers pre-alarms in real time for actions that deviate from the feature threshold; the environmental noise suppression algorithm uses wavelet transform noise reduction technology.

9. The sound and light alarm system for preventing misoperation of a fully insulated pneumatic ring main unit according to claim 1 is characterized in that: The mechanical interlocking module includes: The dual-redundant braking unit consists of electromagnetic brakes symmetrically arranged on both sides of the operating handle shaft. It uses a differential braking strategy to apply reverse braking torque, and the braking torque gradient is adjustable; A mechanical locking unit, comprising a worm gear transmission mechanism connected to a permanent magnet synchronous motor, integrating a torque limiter and a displacement feedback sensor to achieve closed-loop control of the operating lever locking position; The electromagnetic brake and the worm gear mechanism form an electromechanical double locking structure, supporting a fail-safe mode and a hot-swap maintenance mechanism.

10. The sound and light alarm system for preventing misoperation of a fully insulated pneumatic ring main unit according to claim 1, characterized in that: The communication linkage module includes: Multi-protocol communication gateway supports adaptive switching of IEC61850, Modbus TCP, and MQTT protocols, and integrates time-sensitive network mechanisms to ensure real-time communication; The intelligent linkage control unit automatically triggers the vortex-type forced exhaust device and sends a secondary alarm signal to the fire protection system when the laser spectrum analysis unit detects that the SF6 gas concentration in the cabinet is greater than 1000ppm. At the same time, it packages the equipment location, fault type and real-time infrared thermal imaging data and pushes them to the operation and maintenance terminal.

Images