Gas Leakage Monitoring Device for Power Distribution Equipment Applicable to Extremely Cold Environments

Abstract

This invention relates to the field of power distribution protection technology, specifically disclosing a gas leakage monitoring device for power distribution equipment suitable for extremely cold environments. The device includes: a data acquisition unit comprising multiple gas sensor probes deployed at vertical heights within the monitored area; a data processing unit that performs temperature compensation correction on the gas concentration signal to obtain corrected discrete-point concentration data; a fusion calculation unit that reconstructs the spatial concentration field of the monitored area based on the corrected discrete-point concentration data, performs dynamic data assimilation on the spatial concentration field to obtain an optimal estimated concentration field, calculates at least two characteristic parameters characterizing the gas state based on the optimal estimated concentration field, and outputs a decision output result; and a result application unit that identifies and warns of leakage status based on the decision output result. This invention can avoid false alarms and missed alarms caused by gas liquefaction or stratification, making it suitable for gas leakage monitoring of power distribution equipment in extremely cold environments.

Description

Technical Field

[0001] This application relates to the field of power distribution protection technology, and specifically discloses a gas leakage monitoring device for power distribution equipment suitable for extremely cold environments. Background Technology

[0002] Gas-insulated power distribution equipment, such as gas-insulated switchgear, high-voltage switchgear, and transformers, is widely used in power systems due to its compact structure and high reliability. These devices typically use sulfur hexafluoride or its mixtures as the insulating and arc-quenching medium. The internal gas pressure and density directly affect their insulation performance and operational safety. Gas leaks can not only degrade the equipment's insulation performance, potentially causing flashovers or explosions, but also pollute the environment. Therefore, real-time and accurate gas leak monitoring of power distribution equipment is a crucial step in ensuring the safe and stable operation of the power system.

[0003] In high-latitude and high-altitude regions, winter ambient temperatures are generally below -30°C, and can even reach below -40°C. Under such extremely cold conditions, monitoring gas leaks in power distribution equipment faces extremely severe technical challenges. Traditional monitoring devices or methods often fail to work effectively, specifically in the following aspects: Taking sulfur hexafluoride (SF6) gas as an example, its liquefaction temperature at 0.6 MPa pressure is approximately -25°C. When the ambient temperature is below its liquefaction point, SF6 gas liquefies, causing a sharp drop in gas pressure within the chamber. Traditional density relays or pressure gauges can detect pressure changes, but they cannot distinguish between gas loss due to leakage and physical liquefaction caused by low temperature. This liquefaction phenomenon frequently triggers low-pressure alarms and even interlock signals, resulting in numerous false alarms and causing problems for maintenance personnel.

[0004] Furthermore, under extremely cold conditions and when the equipment is stationary for extended periods, changes in gas density can cause the mixed gas of different components inside the equipment to stratify, resulting in significant differences between the gas concentration in local areas and the overall concentration. If the sensor's sampling point is poorly located, the measured values ​​will not represent the true gas state inside the equipment, thus increasing the risk of missed or false alarms.

[0005] Therefore, in view of this, the present invention provides a gas leakage monitoring device for power distribution equipment suitable for extremely cold environments, in order to solve the above-mentioned problems. Summary of the Invention

[0006] The purpose of this invention is to provide a gas leakage monitoring device for power distribution equipment that can avoid false alarms and missed alarms caused by gas liquefaction or stratification, and is suitable for use in extremely cold environments. This overcomes the comprehensive monitoring challenges caused by the combined effects of multiple factors such as sensor adaptability, gas state changes, and sampling interference in extremely cold environments.

[0007] To achieve the above objectives, the basic solution of the present invention provides a gas leakage monitoring device for power distribution equipment suitable for extremely cold environments, comprising: The data acquisition unit includes gas sensor probes deployed at multiple vertical heights within the monitored area, forming a multi-point sensor array to collect gas concentration signals at different heights. The data processing unit is connected to the data acquisition unit, receives the gas concentration signal, and performs temperature compensation correction on the gas concentration signal to obtain the corrected discrete point concentration data. The fusion computing unit is connected to the data processing unit. Based on the corrected discrete point concentration data, it reconstructs the spatial concentration field of the monitored area, performs dynamic data assimilation on the spatial concentration field to obtain the optimal estimated concentration field, and calculates at least two characteristic parameters for characterizing the gas state based on the optimal estimated concentration field, and outputs the decision output result. The result application unit is connected to the fusion computing unit and performs leakage status identification and early warning based on the decision output results.

[0008] Furthermore, each of the gas sensor probes is integrated with a temperature sensor for measuring the microenvironment temperature, and the temperature sensor is connected to the data processing unit.

[0009] Furthermore, the data processing unit includes: The adaptive temperature compensation module compensates for the original value of the gas concentration signal based on the real-time temperature collected by the temperature sensor on each gas sensor probe, and obtains the corrected physical concentration. The feature extraction module calculates the vertical concentration gradient based on the height of adjacent vertical sensor probes and their corresponding corrected physical concentrations.

[0010] Furthermore, the expression for the compensation algorithm is as follows: In the formula, i is the sensor number; , , These are the polynomial coefficients specified in the sensor's factory calibration. This is the temperature compensation coefficient; This is a reference temperature, typically 20℃; This represents the real-time temperature at the probe. This refers to the original voltage or current signal output by the sensor.

[0011] Furthermore, the formula for calculating the vertical concentration gradient is as follows: In the formula, This refers to the installation height.

[0012] Furthermore, the steps for reconstructing the spatial concentration field of the monitored area are as follows: First, calculate the experimental variability function: combine all sensors in pairs, calculate the square of their distance and concentration difference, and fit the result using a spherical model; For any grid point to be estimated in space, the following Kriging equations are established and solved to obtain the weights of each known point, taking into account distance and spatial structure. Calculate the concentration estimate for this grid point; Finally, by traversing all the grids, the measured concentration field at that moment is obtained.

[0013] Furthermore, the steps of dynamic data assimilation are as follows: Define the concentration values ​​of all grid points in the system at that moment, and extract key features or grid point concentrations; The prediction equation is established by setting the state transition matrix as the identity matrix, and the Kalman gain is calculated. By combining the predicted and measured values, the optimal estimated concentration field at that moment is obtained.

[0014] Furthermore, the characteristic parameters used to characterize the gas state include at least two of the following: total change rate, centroid vertical displacement rate, spatial variation coefficient, and second moment change rate.

[0015] Based on the same inventive concept, the present invention also provides a method for monitoring gas leakage in power distribution equipment suitable for extremely cold environments, comprising the following steps: Step S1: Real-time acquisition of gas concentration and temperature signals through a multi-point sensor array deployed at multiple vertical heights within the monitored area; Step S2: Based on the collected temperature signal, perform adaptive temperature compensation correction on the gas concentration signal to obtain the corrected discrete point concentration data; Step S3: Based on the corrected discrete point concentration data, reconstruct the spatial concentration field of the monitored area, perform dynamic data assimilation on the spatial concentration field, and output the optimal estimated concentration field; Step S4: Calculate at least two characteristic parameters for characterizing the gas state based on the optimal estimated concentration field, and identify the current gas state and execute the corresponding early warning strategy according to the combination of the characteristic parameters.

[0016] Furthermore, in step S1, when the ambient temperature sensor detects a temperature below -5°C, the sampling pipeline connected to the sensor probe is actively heated by a heating cable, and the sampled gas before entering the detection chamber is de-iced and de-dropletized by a heated cyclone separator.

[0017] The principle and effect of this solution are as follows: This invention completely solves the problem of false alarms and missed alarms caused by gas stratification. Traditional solutions use single-point sensors that can only represent a local area. If the leaked gas settles at the bottom due to low temperature, the sensor installed at a high position is completely ineffective. This invention uses a vertical array to sense the stratification and uses Kriging interpolation to reconstruct the whole picture. No matter where the gas drifts, the system can capture it, and the monitoring coverage is significantly improved.

[0018] This invention maintains high reliability and low maintenance in extremely cold environments. Ice blockage in traditional pump-suction pipelines is a major maintenance challenge in extremely cold regions, but the diffusion design of this invention physically eliminates ice blockage. Sensor signals are smoothed using Kalman filtering, ensuring stable system output even if a single sensor experiences momentary fluctuations due to extreme low temperatures, significantly reducing false alarm rates and the frequency of on-site maintenance.

[0019] This invention employs intelligent differentiation between liquefaction and leakage, eliminating invalid alarms. By using logical judgment that integrates temperature data and concentration change rate, the system can accurately identify liquefaction phenomena where the concentration decreases but the total amount remains unchanged, and lock the alarm, providing maintenance personnel with a true and reliable basis for decision-making. Attached Figure Description

[0020] To more clearly illustrate the technical solutions in the embodiments of this application, the accompanying drawings used in the description of the embodiments will be briefly introduced below. Obviously, the accompanying drawings described below are only some embodiments of this application. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0021] Figure 1 A system diagram of a gas leakage monitoring device for power distribution equipment suitable for extremely cold environments, according to an embodiment of this application, is shown. Detailed Implementation

[0022] To further illustrate the technical means and effects of the present invention in achieving its intended purpose, the following detailed description of the specific implementation methods, structures, features, and effects of the present invention, in conjunction with the accompanying drawings and preferred embodiments, is provided below.

[0023] A gas leak monitoring device for power distribution equipment suitable for extremely cold environments, implementing, for example... Figure 1 As shown: It includes a data acquisition unit, a data processing unit, a fusion computing unit, and a result application unit.

[0024] In extremely cold environments, heavy gases such as pure SF6, which are about five times denser than air, tend to accumulate close to the ground at low temperatures; while light gases or mixtures such as C4-FN mixtures are typically distributed in the upper middle or top. The data acquisition unit in this embodiment employs a multi-point sensor array. Multiple vertical profiles are selected within the monitored area, such as a GIS power distribution room or a switchgear cable interlayer. Three to five sensor probes are installed on each vertical profile to capture stable stratification interfaces caused by temperature differences.

[0025] In this embodiment, the monitoring object is a typical indoor GIS power distribution room with dimensions of 10m long, 8m wide, and 5m high. Five vertical monitoring poles are deployed in the center of the power distribution room and near the four corners. Each monitoring pole has four sensor probes installed at heights of 0.1m, 0.8m, 2.0m, and 3.5m above the ground, forming a layout with varying vertical spacing but denser coverage in key areas.

[0026] The sensor probe uses a gas sensor with a wide-temperature non-dispersive infrared sensor, a built-in reference gas chamber, and an automatic baseline calibration function. The operating temperature range is -40℃ to +60℃, and the measurement range is 0 to 3000ppm. In addition, a high-precision MEMS temperature sensor and a micro differential pressure sensor are integrated next to the gas probe to measure the temperature of the microenvironment and provide a basis for temperature compensation and stratification judgment.

[0027] The sensor probes are connected to the area data acquisition unit via shielded twisted-pair cables using an RS-485 bus. The acquisition unit is responsible for aggregating the data and performing preliminary encoding verification. The data is then uploaded to the data processing unit in the main control room via an industrial-grade fiber optic Ethernet switch. For remote monitoring points or those without fiber optic coverage, a 4G / LoRa wireless transmission module can be configured for redundancy.

[0028] The data processing unit and fusion computing unit are configured within an industrial control server. The server's memory is divided into a three-dimensional grid space, dividing the power distribution room into 0.5m × 0.5m × 0.2m grid cells for storing and calculating the concentration field. The results application unit consists of an LCD screen and an audible and visual alarm located in the monitoring center. The LCD screen runs a three-dimensional visualization interface, displaying real-time updates of the gas concentration distribution cloud map, historical trends, and alarm information within the power distribution room.

[0029] In extremely cold environments, sensor signals may drift or become noisy due to low temperatures. Therefore, preprocessing is required before data fusion and calculation. In this embodiment, the data processing unit includes an adaptive temperature compensation module and a feature extraction module. The adaptive temperature compensation module utilizes the temperature sensing element built into each sensor to compensate the concentration signal in real time. The compensation algorithm is modeled based on the relationship between temperature and reaction / diffusion rate to ensure that the original value output by the sensor can accurately reflect the physical concentration at different temperature gradients. The expression is as follows: In the formula, i is the sensor number; , , The polynomial coefficients, calibrated at the sensor's factory, are stored in the sensor's internal EEPROM. This is the temperature compensation coefficient, which is determined experimentally: by changing the ambient temperature of the sensor in a temperature chamber, recording the changes in the output value, and fitting the slope of the sensitivity versus temperature curve. A positive value indicates that the lower the temperature, the greater the gain needs to be. This is a reference temperature, typically 20℃; This represents the real-time temperature at the probe. This refers to the original voltage or current signal output by the sensor.

[0030] The feature extraction module extracts a more robust feature: the vertical concentration gradient, which is the concentration difference between two sensors. For adjacent sensors i and j on the same monitoring rod, the vertical concentration gradient is calculated using the following formula: In the formula, This refers to the installation height.

[0031] The fusion computing unit acquires the processed discrete point data and reconstructs the spatial concentration field of the entire power distribution room based on the Kriging method. Compared with traditional interpolation methods such as the inverse distance weighting method, the Kriging method not only considers the distance relationship between the point to be estimated and the known points, but also analyzes the spatial distribution law of the data in the entire monitoring area through the variogram function. It is particularly suitable for the scenario in this embodiment where the vertical concentration gradient is much greater than the horizontal gradient, and can more accurately reconstruct the gas stratification and accumulation phenomenon caused by the heavy gas effect, as detailed below: First, calculate the experimental variability function: pair all sensors and calculate the square of their distance and concentration difference. In this embodiment, due to the small spatial scale, a spherical model is used for fitting, as shown in the formula: In the formula, The distance between the two sensors; The nugget constant represents the measurement error and microscopic variation, and is determined by the sensor accuracy. The arch height represents structural variation, determined by gas diffusion characteristics; This is the range, representing the maximum distance at which spatial correlation exists. Beyond this distance, the concentrations at two points are no longer correlated. In extremely cold, confined spaces, this is typically... The value will decrease as airflow weakens, and needs to be configured based on on-site measured data.

[0032] In this embodiment, =0.1, =5.0m, =4.0m indicates that the concentration value has spatial correlation within a range of 4 meters.

[0033] Any grid point to be estimated in space Establish and solve the following Kriging equations to obtain the known points. It also considered the weights of distance and spatial structure. : In the formula, To find the variogram value between known point i and known point j based on the fitted theoretical model; Given points And the point to be estimated The values ​​of the variation function between them; is the Lagrange multiplier, used as an intermediate variable to solve for the minimum variance.

[0034] The concentration estimate for this grid point is then calculated using the following formula: Finally, by traversing all the grids, the measured concentration field at time k is obtained. .

[0035] Kriging provides the spatial distribution at the current moment, but the gas is flowing and the sensor is noisy. To obtain a smooth and accurate optimal estimate of the field over time, a Kalman filter is used to evaluate the measured field at time k. The following corrections have been made: Define the concentration values ​​of all grid points of the system at time k as follows: We are interested in the concentration distribution across the entire spatial field, but to simplify calculations, we typically extract key features or the concentration at grid points: That is, the concentration values ​​at N key grid points.

[0036] Considering that power distribution rooms are typically located in extremely cold environments with closed doors and windows and minimal airflow, this embodiment employs a random walk model for prediction. This assumes that the concentration field at the current moment is approximately equal to the optimal estimated field at the previous moment, and that uncertainty increases over time. Therefore, the state transition matrix is ​​set as the identity matrix. The prediction equation is as follows: In the formula, This is a priori estimate of time k based on the state at time k-1; P is the error covariance matrix, representing the uncertainty of the current estimate; The process noise covariance represents the degree of confidence in the diffusion equation itself. The coarser the model, the better. The larger it is set, the greater the process noise covariance in this embodiment. Set it as a diagonal matrix, with diagonal elements set to (0.5ppm). 2 This indicates a moderate level of confidence in the random walk model.

[0037] Measurement field As observed values, the observation matrix is ​​the identity matrix, and the noise covariance is measured. The system dynamically adjusts based on sensor performance and ambient temperature. This applies even in extremely cold conditions (T < -20℃). Set the diagonal element to (2.0ppm). 2 To reduce the reliance on a single measurement; at room temperature, it is set to (1.0 ppm). 2 .

[0038] The Kalman gain is calculated using the following formula: Kalman gain determines whether we place more trust in the predictive model. I still trust sensor measurements more. .like Very small (the sensor is very accurate). Larger values ​​tend to be more inclined towards measured values; if Very small (the model is very accurate). The smaller the value, the more likely it is to be a predicted value.

[0039] Then, by fusing the predicted and measured values, the optimal estimated concentration field at time k is obtained: This represents the measurement residual, which is the difference between the actual measurement and the predicted value.

[0040] Finally, update the error covariance matrix.

[0041] The optimal estimated concentration field output by the Kalman filter. For the final decision, including: Calculate the total concentration field M: The total change rate is: Calculate the centroid location: Calculate the centroid coordinates of the concentration field using a weighted average. : The vertical displacement rate of the center of mass is: In the formula, To monitor any grid point within the space, for Volume of a point for Point concentration estimates, , , for The coordinates of the geometric center of the point.

[0042] Calculation of higher-order statistical characteristics: To further improve the identification accuracy under complex flow fields, the second moment and spatial variation coefficient of the concentration field are calculated to determine the dispersion and uniformity of the concentration distribution. The expressions for the variance components in the three directions are as follows: The second moment of the overall space is the sum of the variance components in each direction, and its expression is as follows: The rate of change of the second moment is: The second moment is an absolute indicator, and its value is significantly affected by the total leakage amount. To eliminate the influence of total amount variations and more purely characterize the uniformity of the concentration field, a spatial variability coefficient is introduced, expressed as follows: In the formula, To measure the total effective volume of the area, This represents the spatial average concentration.

[0043] By incorporating these new features into the existing decision model, a more robust classifier can be constructed to distinguish complex working conditions, as detailed below: The fusion computing unit calculates four core discrimination parameters—total change rate, centroid vertical displacement rate, spatial coefficient of variation, and second moment change rate—for early warning judgment. Taking typical SF6 heavy gas as an example, the decision logic is as follows: Scenario 1: Static Layered Accumulation Characteristic features: ≈0: The total amount remains basically unchanged, excluding continuous leakage.

[0044] ≈0 and centroid height There is a strong negative correlation with the bottom temperature: the gas is locked at the bottom due to the low temperature.

[0045] A consistently high level (e.g., > 2.0) indicates that the gas is not diffused uniformly, but rather accumulates in a thin layer at the bottom.

[0046] ≈0: The aggregation pattern is stable.

[0047] Decision output: Determined as "static stratification". The system does not trigger a leak alarm, but the monitoring interface displays "Gas stratification and accumulation at the bottom, it is recommended to start ventilation or heating", and the alarm level is marked as warning level.

[0048] Scenario 2: Continuous micro-leakage from a point source Characteristic features: >0 and steadily increasing: There is a continuous gas supply.

[0049] The direction depends on the gas density; heavier gases sink, and lighter gases rise, but at a slower rate. Maintaining a high level (e.g., > 1.5): Because the leakage points are concentrated, the concentration field shows obvious "hot spots" and is extremely unevenly distributed.

[0050] >0: As the total leakage increases, the range of the accumulation area (second moment) also expands synchronously.

[0051] Decision Output: The system is judged to be "suspected continuous leakage". The system triggers a warning-level alarm and marks the possible leakage source area (the area near the centroid projection) on the 3D interface, prompting maintenance personnel to arrange inspection.

[0052] Scenario 3: Sudden Large-Scale Leakage Characteristic features: Rapid increase (step change): A large amount of gas is ejected instantaneously.

[0053] It manifests as high-speed directional movement (jet direction).

[0054] It exhibits a "high at first, then low" characteristic: Initially: The gas is jet-shaped and highly concentrated. Extremely high.

[0055] Later stage: The gas fills the space and tends to mix. It dropped rapidly.

[0056] Positive first, then negative: The initial diffusion range expands rapidly (the second moment increases), and after filling the space, there is no room for further increases, so the second moment tends to stabilize or decrease slightly.

[0057] Decision Output: Determined as "Serious Leakage". The system immediately triggers an emergency alarm (audio and visual alarm) and outputs prompts such as "Emergency Evacuation Recommended" and "Close the Corresponding Interval".

[0058] Scenario 4: Environmental disturbances, such as ventilation or temperature differences causing dust to rise. This is the scenario most prone to false alarms in extremely cold environments. For example, when the heater is turned on, the deposited SF6 is disturbed.

[0059] Characteristic features: ≈0: The total amount remains unchanged, which is the core characteristic that distinguishes disturbance from leakage.

[0060] The sudden change, with the heavy air suddenly moving upwards, indicates the intervention of an external force.

[0061] The concentration drops rapidly from a high value: the gas that was originally concentrated at the bottom is stirred up, and the concentration field tends to be uniform.

[0062] <0: The second moment decreases rapidly, indicating that the gas mass is "dispersed".

[0063] Decision output: The event is classified as an "environmental disturbance." The system locks all leakage alarms and records the event as an "airflow disturbance event" for subsequent environmental assessment.

[0064] Compared to traditional single-threshold alarm logic, this system constructs a decision tree based on multi-physics field characteristics such as total quantity, centroid, uniformity, and rate of change, possessing certain reasoning and diagnostic capabilities and achieving high-dimensional intelligent monitoring. Furthermore, through configurable model parameters, it can adapt to power distribution rooms of different sizes and various extreme cold conditions, rather than a rigid fixed mode.

[0065] The result application unit obtains the decision output results and displays them on the display. When a dynamic leakage state is determined, an audible and visual alarm is triggered.

[0066] In one possible embodiment, in special scenarios requiring pump-suction sampling, such as environments where sensors cannot be directly installed at potential leak points, the sampling pipeline employs a double-layer composite pipe structure. The inner layer is a PTFE tube with an inner diameter of 6mm, ensuring unobstructed gas flow and corrosion resistance; the outer layer is a 316L stainless steel braided mesh sheath. A self-regulating heating tape is evenly wound between the PTFE tube and the outer sheath.

[0067] The heat tracing cable is controlled by a solid-state relay within the processing layer. When the ambient temperature sensor detects a temperature below -5°C, the processing layer automatically switches on the power to the heat tracing cable. The heat tracing cable automatically adjusts its heating power based on the gas temperature inside the pipe, maintaining the temperature inside the pipe consistently 5–10°C higher than the ambient temperature and not lower than -5°C, thereby ensuring that no icing or condensation occurs in the gas path from the sampling port to the analysis cabinet.

[0068] A heated cyclone separator is installed before the gas enters the precision sensor. The sampling gas enters the separator tangentially, and under centrifugal force, any tiny ice crystals or droplets that may be entrained in the gas are thrown against the separator wall. The separator wall is heated to +5°C by a heating cable, causing the ice crystals to melt and flow down the wall, exiting through an automatic drain valve at the bottom. The clean gas then enters the sensor detection chamber from the top outlet, effectively protecting the core sensor.

[0069] Another embodiment of this application provides a method for monitoring gas leakage in power distribution equipment suitable for extremely cold environments, comprising the following steps: Step S1: Multi-dimensional data collection In an indoor GIS power distribution room measuring 10m long, 8m wide, and 5m high, five vertical monitoring poles were installed, located in the center of the room and near the four corners. Each monitoring pole was equipped with four sensor probes at heights of 0.1m, 0.8m, 2.0m, and 3.5m from the ground, forming a non-uniformly spaced, densely packed vertical layout, with a focus on the bottom and central areas.

[0070] Each sensor probe includes a core gas sensor and an auxiliary sensor, which collects raw gas concentration, temperature, and pressure signals at each location in real time at a sampling frequency of 1Hz.

[0071] Step S2: Temperature Adaptive Compensation Correction The adaptive temperature compensation module in the data processing unit uses the temperature sensing element built into each sensor to compensate the concentration signal in real time. The compensation algorithm is modeled based on the relationship between temperature and reaction and diffusion rate to ensure that the original value output by the sensor can truly reflect the physical concentration at that point under different temperature gradients. In addition, the feature extraction module extracts the vertical concentration gradient, which is the concentration difference between the upper and lower layers of sensors.

[0072] Step S3, Spatial Concentration Field Reconstruction The fusion computing unit acquires the processed discrete point data and reconstructs the spatial concentration field of the entire power distribution room based on the kriging method, as follows: First, calculate the experimental variability function: combine all sensors in pairs, calculate the square of their distance and concentration difference, and fit the result using a spherical model; For any grid point to be estimated in space, the following Kriging equations are established and solved to obtain the weights of each known point, taking into account distance and spatial structure. Calculate the concentration estimate for this grid point; Finally, by traversing all the grids, the measured concentration field at that moment is obtained.

[0073] Step S4: Dynamic Data Assimilation To obtain a smooth and accurate optimal estimated field over time, a Kalman filter is used to correct the measured field at that moment. This embodiment uses a random walk model for prediction, assuming that the concentration field at the current moment is approximately the same as the optimal estimated field at the previous moment. Specifically: Define the concentration values ​​of all grid points in the system at that moment, and extract key features or grid point concentrations; The prediction equation is established by setting the state transition matrix as the identity matrix, and the Kalman gain is calculated. By combining the predicted and measured values, the optimal estimated concentration field at that moment is obtained.

[0074] Step S5: Multi-feature fusion leakage identification Based on the optimal estimated concentration field at that moment, the total concentration field, the rate of change of the total concentration field, the position of the centroid, the vertical displacement rate of the centroid, the spatial coefficient of variation, the overall spatial second moment, and the rate of change of the second moment are calculated respectively, and the state is judged based on the preset decision logic.

[0075] Step S6: Early Warning and Visualization Output The result application unit obtains the decision output results and displays them on the display. When a dynamic leakage state is determined, an audible and visual alarm is triggered.

[0076] When the ambient temperature sensor detects a temperature below -5℃, the processing layer automatically activates the power supply to the heating cable, heating the double-layer composite sampling tube to maintain the temperature inside the tube consistently 5-10℃ higher than the ambient temperature and not lower than -5℃. Before entering the detection chamber, the sampling gas passes through a heated cyclone separator. The gas enters the separator tangentially, and under centrifugal force, any tiny ice crystals or droplets that may be entrained in the gas are thrown against the separator wall. The separator wall is heated to +5℃ by the heating cable, causing the ice crystals to melt and flow down the wall, exiting through the automatic drain valve at the bottom, while the clean gas enters the sensor detection chamber from the top outlet.

[0077] A simulated comparative experiment was conducted in an artificial climate chamber at -35℃ to verify the technical effectiveness of this method, as detailed below: Control group: The traditional single-point monitoring method was used, with only one sampling point set up at a height of 2.5 meters above the ground. The data was not subjected to temperature compensation and spatial reconstruction.

[0078] The present invention group adopts the method described above.

[0079] Test Scenario 1 Gas stratification: A small amount of SF6 gas is released at a height of 0.2 meters above the ground, and the cooling system is activated.

[0080] Results of the control group: No concentration change was detected in the single-point monitoring data at 2.5 meters, and the stratification phenomenon could not be identified.

[0081] Results of this invention group: After temperature compensation correction in step S2, the data response at 0.1 meters is obvious; Kriging interpolation in step S3 successfully reconstructs the concentration field of "high concentration at the bottom and low concentration at the top"; Kalman filtering in step S4 provides stable output; and multi-feature recognition in step S5 accurately determines the state as "layered accumulation" without triggering false alarms.

[0082] Test Scenario 2 Micro-leakage: Simulate a continuous micro-leakage source at a height of 1.0 meter above the ground.

[0083] Results of the control group: The sampling pipeline failed due to ice blockage after running for 2 hours at -35℃, and the monitoring was interrupted.

[0084] Results of this invention group: Anti-icing measures were effective, and the sampling pipeline was unobstructed; Kalman filtering effectively filtered out sensor noise, and the continuous upward trend of the total amount M was identified when the leakage rate was only 5 ml / min; step S5 identified... Exceeding the standard and It significantly increases the accuracy of warnings, issuing accurate alerts 15 minutes after a leak.

[0085] Experimental results show that the method of the present invention effectively solves technical problems such as gas stratification interference, sensor noise amplification, and ice blockage in sampling pipelines under extremely cold environments, and significantly improves the accuracy and reliability of leak monitoring.

[0086] The above description is merely a preferred embodiment of the present invention and is not intended to limit the present invention in any way. Although the present invention has been disclosed above with reference to preferred embodiments, it is not intended to limit the present invention. Any person skilled in the art can make some modifications or alterations to the above-disclosed technical content to create equivalent embodiments without departing from the scope of the present invention. Any indirect modifications, equivalent changes, and alterations made to the above embodiments based on the technical essence of the present invention without departing from the scope of the present invention shall still fall within the scope of the present invention.

Claims

1. A gas leakage monitoring device for power distribution equipment suitable for extremely cold environments, characterized in that, include: The data acquisition unit includes gas sensor probes deployed at multiple vertical heights within the monitored area, forming a multi-point sensor array to collect gas concentration signals at different heights. The data processing unit is connected to the data acquisition unit, receives the gas concentration signal, and performs temperature compensation correction on the gas concentration signal to obtain the corrected discrete point concentration data. The fusion computing unit is connected to the data processing unit. Based on the corrected discrete point concentration data, it reconstructs the spatial concentration field of the monitored area, performs dynamic data assimilation on the spatial concentration field to obtain the optimal estimated concentration field, and calculates at least two characteristic parameters for characterizing the gas state based on the optimal estimated concentration field, and outputs the decision output result. The result application unit is connected to the fusion computing unit and performs leakage status identification and early warning based on the decision output results.

2. The gas leakage monitoring device for power distribution equipment suitable for extremely cold environments according to claim 1, characterized in that, Each of the gas sensor probes is integrated with a temperature sensor for measuring the microenvironment temperature, and the temperature sensor is connected to the data processing unit.

3. The gas leakage monitoring device for power distribution equipment suitable for extremely cold environments according to claim 1, characterized in that, The data processing unit includes: The adaptive temperature compensation module compensates for the original value of the gas concentration signal based on the real-time temperature collected by the temperature sensor on each gas sensor probe, and obtains the corrected physical concentration. The feature extraction module calculates the vertical concentration gradient based on the height of adjacent vertical sensor probes and their corresponding corrected physical concentrations.

4. A gas leakage monitoring device for power distribution equipment suitable for extremely cold environments according to claim 3, characterized in that, The expression for the compensation algorithm is as follows: In the formula, i is the sensor number; , , These are the polynomial coefficients specified in the sensor's factory calibration. This is the temperature compensation coefficient; This is a reference temperature, typically 20℃; This represents the real-time temperature at the probe. This refers to the original voltage or current signal output by the sensor.

5. A gas leakage monitoring device for power distribution equipment suitable for extremely cold environments according to claim 4, characterized in that, The formula for calculating the vertical concentration gradient is as follows: In the formula, This refers to the installation height.

6. A gas leakage monitoring device for power distribution equipment suitable for extremely cold environments according to claim 1, characterized in that, The steps to reconstruct the spatial concentration field of the monitored area are as follows: First, calculate the experimental variability function: combine all sensors in pairs, calculate the square of their distance and concentration difference, and fit the result using a spherical model; For any grid point to be estimated in space, the following Kriging equations are established and solved to obtain the weights of each known point, taking into account distance and spatial structure. Calculate the concentration estimate for this grid point; Finally, by traversing all the grids, the measured concentration field at that moment is obtained.

7. A gas leakage monitoring device for power distribution equipment suitable for extremely cold environments according to claim 6, characterized in that, The steps of dynamic data assimilation are as follows: Define the concentration values ​​of all grid points in the system at that moment, and extract key features or grid point concentrations; The prediction equation is established by setting the state transition matrix as the identity matrix, and the Kalman gain is calculated. By combining the predicted and measured values, the optimal estimated concentration field at that moment is obtained.

8. A gas leakage monitoring device for power distribution equipment suitable for extremely cold environments according to claim 1, characterized in that, Characteristic parameters used to characterize the state of a gas include at least two of the following: total rate of change, centroid vertical displacement rate, spatial coefficient of variation, and second moment rate of change.

9. A method for monitoring gas leakage in power distribution equipment suitable for extremely cold environments, characterized in that, Includes the following steps: Step S1: Real-time acquisition of gas concentration and temperature signals through a multi-point sensor array deployed at multiple vertical heights within the monitored area; Step S2: Based on the collected temperature signal, perform adaptive temperature compensation correction on the gas concentration signal to obtain the corrected discrete point concentration data; Step S3: Based on the corrected discrete point concentration data, reconstruct the spatial concentration field of the monitored area, perform dynamic data assimilation on the spatial concentration field, and output the optimal estimated concentration field; Step S4: Calculate at least two characteristic parameters for characterizing the gas state based on the optimal estimated concentration field, and identify the current gas state and execute the corresponding early warning strategy according to the combination of the characteristic parameters.

10. A method for monitoring gas leakage in power distribution equipment suitable for extremely cold environments according to claim 9, characterized in that, In step S1, when the ambient temperature sensor detects that the temperature is below -5℃, the sampling pipeline connected to the sensor probe is actively heated by the heating cable, and the sampled gas before entering the detection chamber is treated by a heated cyclone separator to remove ice crystals and droplets.

Images