A monitoring method and system for a gas-protected medium voltage switchgear

Abstract

The application relates to the technical field of switch cabinet monitoring, in particular to a monitoring method and system for a gas-protected medium-voltage switch cabinet, which monitors the air pressure, temperature and busbar load in the gas-protected medium-voltage switch cabinet. A collaborative monitoring system fusing pressure, temperature and current multidimensional information is constructed, breaking through the limitation of single parameter monitoring. The Bayesian probability and D-S evidence theory are innovatively combined, the objectivity of statistical data is utilized, and the evidence conflict and cognitive uncertainty are properly handled, so that the diagnosis is upgraded from threshold alarm to probabilistic trust evaluation. The application can effectively monitor the current state type of the switch cabinet in real time, drive the operation and maintenance mode to shift from regular maintenance to predictive maintenance, ensure power supply reliability, and greatly reduce operation and maintenance cost and unplanned power outage risk.

Description

Technical Field

[0001] This application relates to the field of switchgear monitoring technology, specifically a monitoring method and system for gas-protected medium-voltage switchgear. Background Technology

[0002] Gas-insulated metal-enclosed switchgear, commonly known as gas-insulated switchgear (C-GIS), is widely used in medium-voltage power distribution fields such as urban power grids, rail transit, industrial plants, and data centers due to its compact structure, strong environmental adaptability, and high reliability. This type of equipment achieves full insulation and full enclosure by sealing high-voltage conductive components such as circuit breakers, disconnectors, and busbars within stainless steel or aluminum alloy chambers filled with low-pressure insulating gas (such as dry air, nitrogen, SF6, or mixtures thereof), greatly improving safety and maintenance-free operation.

[0003] In existing technologies, monitoring gas states commonly employs mechanical density relays or digital density transmitters. Alarms are triggered using fixed thresholds. Firstly, these alarms, based on fixed thresholds, are insensitive to slow, minute leaks, often triggering only after the leak has accumulated to a certain level, thus missing the opportunity for early warning. Secondly, existing technologies typically monitor gas parameters independently, failing to consider the non-uniform effect of localized overheating within the equipment on the overall temperature field of the gas chamber. This uneven temperature distribution can lead to deviations in pressure compensation based on single-point temperature measurements, potentially resulting in misjudgments.

[0004] Furthermore, for monitoring primary circuit connection points (such as busbar connections and circuit breaker contacts), traditional methods mainly rely on manual periodic inspections using infrared thermal imagers through observation windows, or the use of discretely installed wireless temperature sensors to monitor the absolute temperature of key points. The former cannot achieve real-time online monitoring and is limited by the inspection cycle, making it difficult to capture sudden or rapidly developing thermal defects; the latter, while enabling online monitoring, typically only sets a simple upper temperature limit alarm. These methods fail to effectively isolate the significant impact of load current fluctuations on contact temperature rise, potentially triggering false alarms due to increased load, or masking the hidden danger of substantially increased contact resistance due to low load. More importantly, existing contact point temperature monitoring systems and gas condition monitoring systems are independent of each other, with no correlation between data and alarm information. In actual operation, aging of gas chamber seals may lead to micro-leakage, and decreased sealing performance may alter local heat dissipation conditions or introduce moisture, exacerbating oxidation and corrosion of adjacent electrical connection points, forming a correlated fault of "leakage" and "overheating." The existing isolated monitoring model is completely unable to identify such complex failure modes, which is not conducive to root cause analysis and the development of accurate maintenance strategies. Summary of the Invention

[0005] In view of this, the purpose of this application is to provide a monitoring method and system for gas-protected medium-voltage switchgear to solve the problems in the background art.

[0006] To achieve the above objectives, this application adopts the following technical solution:

[0007] This application discloses a monitoring method for a gas-protected medium-voltage switchgear, comprising the following steps:

[0008] The overall temperature sequence, gas pressure sequence, bus load sequence, and actual temperature rise sequence of multiple contact points are acquired within a target time period in the gas-protected medium-voltage switchgear. The target time period includes multiple time points with a target duration prior to the current time point.

[0009] The target time period is divided into multiple time windows; the air pressure sequence of the multiple time windows is compensated based on the overall temperature sequence of the multiple time windows to obtain the standard air pressure sequence at the standard temperature; and the air pressure trend and the confidence level of the air pressure trend are extracted from the standard air pressure sequence.

[0010] The theoretical temperature rise of the contact point is calculated based on the pre-built contact point temperature rise prediction model and the bus load of multiple time windows. The residual between the theoretical temperature rise and the actual temperature rise of the contact point in multiple time windows is calculated, and the temperature rise anomaly index of the contact point corresponding to the maximum residual of each time window is extracted. The average actual temperature rise of multiple contact points corresponding to the current time window is calculated.

[0011] A pressure trend sequence is constructed based on pressure trends over multiple time windows; a temperature rise anomaly index sequence is constructed based on temperature rise anomaly indices over multiple time windows; and the correlation between the pressure trend sequence and the temperature rise anomaly index sequence is calculated.

[0012] A feature vector is constructed based on the air pressure trend, confidence level of the air pressure trend, temperature rise anomaly index, actual average temperature rise, and correlation of the current time window. The feature vector is then input into a pre-constructed fusion model to obtain the monitoring results. The fusion model integrates a Bayesian diagnostic model and DS evidence theory.

[0013] In one embodiment of this application, the air pressure sequence of multiple time windows is compensated based on the overall temperature sequence of multiple time windows to obtain a standard air pressure sequence at a standard temperature; and the air pressure trend and the confidence level of the air pressure trend are extracted from the standard air pressure sequence, including:

[0014] Aligning the overall temperature series and the air pressure series based on time points yields multiple sets of basic data. ,in, Indicates a point in time The corresponding air pressure Indicates a point in time The corresponding temperature;

[0015] For each time point Based on temperature air pressure Compensation is performed to obtain the equivalent air pressure representing 20 degrees Celsius. And the equivalent pressure sequence, wherein the mathematical expression for the equivalent pressure is:

[0016]

[0017] The slope of the equivalent pressure sequence for the current time window is extracted using the sliding window least squares method. and the coefficient of determination The slope As a pressure trend, and the determining coefficient As a confidence level.

[0018] In one embodiment of this application, the method for constructing the contact point temperature rise prediction model includes:

[0019] Acquire multiple temperature rise data samples when the switchgear is in a healthy state, wherein the temperature rise data samples include bus load and temperature rise value samples;

[0020] The basic mathematical expression for constructing the contact point temperature rise prediction model is as follows:

[0021]

[0022] In the formula, Indicates the first Temperature rise coefficient at each contact point Indicates the bus load. Indicates the first Temperature rise at each contact point Indicates the offset;

[0023] Substituting multiple temperature rise data samples into the basic mathematical expression and combining them with the least squares method for fitting, the temperature rise coefficient is obtained. And a model for predicting temperature rise at contact points.

[0024] In one embodiment of this application, the theoretical temperature rise of the contact point is calculated based on a pre-built contact point temperature rise prediction model and the bus load of multiple time windows. The residuals between the theoretical and actual temperature rises of the contact points in multiple time windows are also calculated. The temperature rise anomaly index of the contact point corresponding to the maximum residual in each time window is extracted, including:

[0025] For each time window, the bus load at each time point will be... Substituting the values ​​into the contact point temperature rise prediction model for multiple contact points, we obtain the theoretical temperature rise of multiple contact points at each time point. ;

[0026] For each contact point, calculate the average theoretical temperature rise within each time window. and the average of the actual temperature rise ;

[0027] For the same time window, calculate the average theoretical temperature rise. and the average of the actual temperature rise residual ,in:

[0028]

[0029] In the formula, Indicates the time window index;

[0030] At each time point, extract the maximum residual from references at multiple contact points. And based on the maximum residual Calculate the temperature rise anomaly index for each time window. The mathematical expression for the temperature rise anomaly index is:

[0031]

[0032] In the formula, Represents the maximum residual The standard deviation of the residuals at the corresponding contact points under healthy conditions.

[0033] In one embodiment of this application, the mathematical expression for the relevance is:

[0034]

[0035] In the formula, This represents a sequence of air pressure trends. This represents the mean of the air pressure trend series. This represents a temperature rise anomaly index sequence. This represents the mean of the temperature rise anomaly index series. The sequence length is given.

[0036] In one embodiment of this application, the method for constructing the fusion model includes:

[0037] Acquire historical operating data samples of the gas-protected medium-voltage switchgear for multiple historical time periods. The historical operating data samples include overall temperature sequence samples, gas pressure sequence samples, bus load sequence samples, and actual temperature rise sequence samples of multiple contact points.

[0038] Extract pressure trend samples, pressure trend confidence samples, temperature rise anomaly index samples, actual temperature rise average samples, and correlation samples from the historical operating data samples. Construct feature vector samples based on the pressure trend samples, pressure trend confidence samples, temperature rise anomaly index samples, actual temperature rise average samples, and correlation samples. Label the type of the historical operating data samples, wherein the type labels include normal, gas leak, overheating, and combined fault.

[0039] A Bayesian diagnostic model and a DS evidence model are constructed based on feature vector samples with various types of labels. The outputs of the Bayesian diagnostic model and the DS evidence model are then fused. The mathematical expression for the fused output is as follows:

[0040]

[0041] In the formula, This indicates the type label determination result of the output. As the first weight, As the second weight, As the third weight, Indicates the first Type tags, The features represented by the Bayesian diagnostic model output For type tags The posterior probability, This indicates that the evidence output by the DS evidence model is related to the type label. The minimum trust value, This indicates that the evidence output by the DS evidence model is for type labeling. The highest trust value.

[0042] In one embodiment of this application, a Bayesian diagnostic model is constructed based on feature vector samples of various types of labels, including:

[0043] Calculate the prior probability of each label type based on multiple historical operational data samples. The mathematical expression for the prior probability is:

[0044]

[0045] In the formula, This represents the total number of samples. Indicates the first The number of samples for each type of label. Indicates the first The probability of each type of label;

[0046] Construct a likelihood function for the feature vector samples corresponding to each type label. ;

[0047] Based on the prior probability and the likelihood function Construct a Bayesian diagnostic model, wherein the mathematical expression of the Bayesian diagnostic model is:

[0048]

[0049] In the formula, Represents the posterior probability. Indicates the first The likelihood function for each type label. Indicates the first The prior probability of each type of label.

[0050] In one embodiment of this application, a DS evidence model is constructed based on feature vector samples of various types of labels, including:

[0051] Construct the BPA function for pressure evidence, the BPA function for temperature evidence, and the BPA function for correlation evidence. The mathematical expressions for the BPA functions for pressure evidence, temperature evidence, and correlation evidence are as follows:

[0052]

[0053]

[0054]

[0055] In the formula, Indicates the level of confidence in the gas leak. Indicates the level of trust in contact overheating. Indicates the composite fault trust level. and All are weighting coefficients. This indicates the air pressure trend over a target time period. This is a threshold for air pressure trends generated based on feature vector samples. This indicates the temperature rise anomaly index for the target time period. This indicates that the threshold for generating the temperature rise anomaly index is based on feature vector samples. The correlation coefficient represents the target time period. It is an exponential function;

[0056] Trust level of the gas leak and the contact overheating confidence level Fusion was performed to obtain the gas-superheated confidence level. The gas-superheat confidence level With the aforementioned composite fault trust To achieve integration and gain integration trust. The fusion process takes into account the conflict coefficient.

[0057] Based on the aforementioned fusion trust level Calculate the trust function and likelihood function , wherein the trust function and the likelihood function The mathematical expression is:

[0058]

[0059]

[0060] In the formula, Type label The corresponding set of fault states, Type label A subset of the corresponding set of fault states This indicates the trust level of the subset.

[0061] In one embodiment of this application, it further includes:

[0062] When the monitoring results contain any type of fault, the monitoring results are sent to the target object.

[0063] This application also provides a monitoring system for a gas-protected medium-voltage switchgear, comprising:

[0064] The acquisition module is used to acquire the overall temperature sequence, gas pressure sequence, bus load sequence, and actual temperature rise sequence of multiple contact points collected within a target time period inside the gas-protected medium-voltage switchgear. The target time period includes multiple time points of a target duration prior to the current time point.

[0065] The barometric pressure analysis module is used to divide the target time period into multiple time windows; compensate the barometric pressure sequences of the multiple time windows based on the overall temperature sequence of the multiple time windows to obtain a standard barometric pressure sequence at a standard temperature; and extract the barometric pressure trend and the confidence level of the barometric pressure trend from the standard barometric pressure sequence.

[0066] The temperature rise analysis module is used to calculate the theoretical temperature rise of the contact point based on the pre-built contact point temperature rise prediction model and the bus load of multiple time windows, and to calculate the residual between the theoretical temperature rise and the actual temperature rise of the contact point in multiple time windows. It also extracts the temperature rise anomaly index of the contact point corresponding to the maximum residual in each time window, and calculates the average actual temperature rise of multiple contact points corresponding to the current time window.

[0067] The correlation extraction module is used to construct a pressure trend sequence based on pressure trends over multiple time windows; construct a temperature rise anomaly index sequence based on temperature rise anomaly indices over multiple time windows; and calculate the correlation between the pressure trend sequence and the temperature rise anomaly index sequence.

[0068] The fusion monitoring module is used to construct a feature vector based on the air pressure trend of the current time window, the confidence level of the air pressure trend of the current time window, the temperature rise anomaly index of the current time window, the actual average temperature rise of the current time window, and the correlation. The feature vector is then input into a pre-constructed fusion model to obtain the monitoring results. The fusion model integrates a Bayesian diagnostic model and DS evidence theory.

[0069] The beneficial effects of this application are as follows: This application discloses a monitoring method and system for gas-protected medium-voltage switchgear, which monitors the gas pressure, temperature, and bus load within the switchgear. It constructs a collaborative monitoring system integrating multi-dimensional information on pressure, temperature, and current, overcoming the limitations of single-parameter monitoring. It innovatively combines Bayesian probability and DS evidence theory, utilizing the objectivity of statistical data while properly handling evidence conflicts and cognitive uncertainties, upgrading diagnosis from threshold alarms to probabilistic trust assessment. This application can effectively monitor the current state type of the switchgear in real time, driving the operation and maintenance mode from periodic inspections to predictive maintenance, significantly reducing operation and maintenance costs and the risk of unplanned power outages while ensuring power supply reliability. Attached Figure Description

[0070] The present application will be further described below with reference to the accompanying drawings and embodiments:

[0071] Figure 1 This is a flowchart illustrating a monitoring method for a gas-protected medium-voltage switchgear in one embodiment of this application;

[0072] Figure 2 This is a topology diagram of the data acquisition hardware structure in one embodiment of this application;

[0073] Figure 3 This is a schematic diagram of the fusion model construction process in one embodiment of this application;

[0074] Figure 4 This is a structural diagram of a monitoring system for a gas-protected medium-voltage switchgear, as shown in one embodiment of this application. Detailed Implementation

[0075] The following specific examples illustrate the implementation of this application. Those skilled in the art can easily understand other advantages and effects of this application from the content disclosed in this specification. This application can also be implemented or applied through other different specific embodiments, and various details in this specification can also be modified or changed based on different viewpoints and applications without departing from the spirit of this application. It should be noted that, unless otherwise specified, the following embodiments and features in the embodiments can be combined with each other.

[0076] It should be noted that the illustrations provided in the following embodiments are only schematic representations of the basic concept of this application. Therefore, the drawings only show the layers related to this application and are not drawn according to the actual number, shape and size of the layers in the actual implementation. In the actual implementation, the form, number and proportion of each layer can be arbitrarily changed, and the layer layout may also be more complex.

[0077] Numerous details are explored in the following description to provide a more thorough explanation of embodiments of this application; however, it will be apparent to those skilled in the art that embodiments of this application may be practiced without these specific details.

[0078] Figure 1 This is a flowchart illustrating a monitoring method for a gas-protected medium-voltage switchgear according to one embodiment of this application, as shown below. Figure 1 As shown, a monitoring method for a gas-protected medium-voltage switchgear according to this application mainly includes the following steps:

[0079] S110, Basic Parameter Acquisition:

[0080] The overall temperature sequence, gas pressure sequence, bus load sequence, and actual temperature rise sequence of multiple contact points are acquired within a target time period in the gas-protected medium-voltage switchgear. The target time period includes multiple time points with a target duration prior to the current time point.

[0081] S120, Barometric Pressure Trend Feature Extraction:

[0082] The target time period is divided into multiple time windows; the air pressure sequence of the multiple time windows is compensated based on the overall temperature sequence of the multiple time windows to obtain the standard air pressure sequence at the standard temperature; and the air pressure trend and the confidence level of the air pressure trend are extracted from the standard air pressure sequence.

[0083] S130, Temperature Rise Anomaly Feature Extraction:

[0084] The theoretical temperature rise of the contact point is calculated based on the pre-built contact point temperature rise prediction model and the bus load of multiple time windows. The residual between the theoretical temperature rise and the actual temperature rise of the contact point in multiple time windows is calculated, and the temperature rise anomaly index of the contact point corresponding to the maximum residual of each time window is extracted. The average actual temperature rise of multiple contact points corresponding to the current time window is calculated.

[0085] S140, Relevance Feature Extraction:

[0086] A pressure trend sequence is constructed based on pressure trends over multiple time windows; a temperature rise anomaly index sequence is constructed based on temperature rise anomaly indices over multiple time windows; and the correlation between the pressure trend sequence and the temperature rise anomaly index sequence is calculated.

[0087] S150, a fusion analysis based on Bayesian and DS evidence theories:

[0088] A feature vector is constructed based on the air pressure trend, confidence level of the air pressure trend, temperature rise anomaly index, actual average temperature rise, and correlation of the current time window. The feature vector is then input into a pre-constructed fusion model to obtain the monitoring results. The fusion model integrates a Bayesian diagnostic model and DS evidence theory.

[0089] The following is a detailed explanation of the principles and implementation methods of the above overall process. In the following text, this application will describe the complete solution in logical order:

[0090] S110, Basic Parameter Acquisition:

[0091] Figure 2 This is a topology diagram of the data acquisition hardware structure in one embodiment of this application, such as... Figure 2 As shown, in this application, the bus load (current load) is collected by the current transformer 210 installed on the bus of the switchgear.

[0092] The ambient temperature of each sampling point in the switch cabinet is collected by temperature sensors 220 set at multiple sampling points, and the overall temperature is obtained by averaging.

[0093] Temperature values ​​at each contact point are collected using thermocouple 230;

[0094] The overall air pressure level inside the switch cabinet is collected by air pressure sensor 240;

[0095] The aforementioned current transformer 210, temperature sensor 220, thermocouple 230, and pressure sensor 240 are all connected to the edge terminal 250. The edge terminal 250 sends its output to the analysis host 260 for analysis and processing.

[0096] In this application, rolling data acquisition is adopted, and the basic data acquired includes the overall temperature sequence, air pressure sequence, bus load sequence and actual temperature rise sequence of multiple contact points for the most recent 1-2 weeks.

[0097] The overall temperature sequence, air pressure sequence, and bus load sequence were all acquired at a granularity of hourly / time.

[0098] The actual temperature rise sequence of multiple contact points needs to record the temperature of each measuring point every 15 minutes until the temperature rise change is less than 1K for 3 consecutive hours (considered stable). The temperature rise is obtained by subtracting the temperature values ​​before and after stabilization.

[0099] After processing, since the time granularity of the above-mentioned multiple sequences is not aligned, in order to facilitate subsequent processing, this application uses sliding window filtering to split multiple time points into multiple time windows (for example, one day as a time window).

[0100] S120, Barometric Pressure Trend Feature Extraction:

[0101] In this application, changes in air pressure can indicate whether a gas leak has occurred inside the switchgear. However, since the temperature inside the switchgear changes, the air pressure also changes with the temperature. Therefore, this application requires compensating the air pressure sequence using the overall temperature sequence before extracting the air pressure change trend characteristics at the standard temperature. Specifically, this includes:

[0102] S121, Align the overall temperature sequence and the air pressure sequence based on time points to obtain multiple sets of basic data. ,in, Indicates a point in time The corresponding air pressure Indicates a point in time The corresponding temperature (Kelvin, unit K);

[0103] In gas measurements, temperature and pressure are interrelated physical quantities and must be measured at the same time point to accurately reflect changes in the gas state. Alignment processing ensures the temporal consistency of temperature and pressure data, avoiding analytical biases caused by time asynchrony.

[0104] S122, for each time point Based on temperature air pressure Compensation is performed to obtain the equivalent air pressure representing 20 degrees Celsius. And the equivalent pressure sequence, wherein the mathematical expression for the equivalent pressure is:

[0105]

[0106] By using temperature compensation, the air pressure at each pressure collection point is standardized to 20 degrees Celsius. This eliminates the influence of temperature changes on air pressure measurement, ensuring that the air pressure value reflects the leakage of the gas itself, rather than the air pressure changes caused by temperature fluctuations.

[0107] S123, Extracting the slope of the equivalent pressure sequence for the current time window based on the sliding window least squares method. and the coefficient of determination The slope As a pressure trend, and the determining coefficient As a confidence level.

[0108] Specifically, in the context of a sliding window, the pressure change trend (slope) within a fixed time window is calculated by linearly fitting the equivalent pressure sequence within that window. ) and goodness of fit (coefficient of determination) ).

[0109] The sliding window method can analyze air pressure change trends in real time, avoiding the randomness of single-point measurements and improving the real-time performance and reliability of the detection. Slope It directly reflects the rate of change of gas pressure over time and is a key indicator for judging the severity of gas leaks. (Coefficient of Determination) As a confidence index, it reflects the degree of matching between the fitted straight line and the actual data, and can be used to judge the reliability of the air pressure trend. When A higher pressure indicates a significant trend in air pressure change and a high probability of leakage; when... A lower reading indicates that the pressure change may be affected by noise, requiring further verification. The characteristics of the least squares method allow this approach to effectively suppress the influence of measurement noise while preserving the pressure change trend, thus improving detection accuracy.

[0110] S130, Temperature Rise Anomaly Feature Extraction:

[0111] Before extracting abnormal temperature rise features, this application needs to construct a contact point temperature rise prediction model for each contact point. The core purpose is to establish a health status benchmark model for the switchgear contact points. By comparing the real-time temperature rise with the predicted values ​​of the health model, abnormalities such as poor contact can be accurately identified. The method for constructing the contact point temperature rise prediction model includes:

[0112] (1) Obtain multiple temperature rise data samples when the switchgear is in a healthy state, wherein the temperature rise data samples include bus load and temperature rise value samples;

[0113] Specifically, the temperature rise data samples need to cover typical operating load ranges (such as 20%, 50%, 80%, and 100% of rated load) to ensure the model's generalization ability. At least 5 sets of data should be collected for each load point (to avoid single-point errors), with a total sample size of ≥30 sets.

[0114] (2) Construct the basic mathematical expression for the contact point temperature rise prediction model, wherein the basic mathematical expression is:

[0115]

[0116] In the formula, Indicates the first Temperature rise coefficient at each contact point Indicates the bus load. Indicates the first Temperature rise at each contact point Indicates the offset;

[0117] The aforementioned model is based on Joule's law. A linear model built upon Joule's law possesses both physical interpretability and high-precision predictive capabilities. Furthermore, it avoids complex thermodynamic equations, simplifying calculations.

[0118] (3) Substitute multiple temperature rise data samples into the basic mathematical expression and combine them with the least squares method to obtain the temperature rise coefficient. And a model for predicting temperature rise at contact points.

[0119] Minimize the residuals using the least squares method, i.e. The constructed model can accurately predict the theoretical temperature rise of different contact points under different load levels, providing a basic reference for subsequent identification of abnormal temperature rises.

[0120] After constructing the contact point temperature rise prediction model, the process of extracting the temperature rise anomaly index includes:

[0121] S131, for each time window, the bus load at each time point. Substituting the values ​​into the contact point temperature rise prediction model for multiple contact points, we obtain the theoretical temperature rise of multiple contact points at each time point. ;

[0122] S132, For each contact point, calculate the average theoretical temperature rise within each time window. and the average of the actual temperature rise ;

[0123] Calculating the average value can suppress single-point fluctuations and achieve time alignment.

[0124] S133, for the same time window, calculate the average theoretical temperature rise. and the average of the actual temperature rise residual ,in:

[0125]

[0126] In the formula, Indicates the time window index;

[0127] residual This indicates that the actual temperature rise is higher than the predicted value under healthy conditions, which may be caused by abnormalities such as increased contact resistance or poor heat dissipation; residual This indicates that the actual temperature rise is lower than the predicted value, which may be caused by non-abnormal factors such as a decrease in ambient temperature or an improvement in heat dissipation conditions.

[0128] S134, at each time point, extract the maximum residual from the references of multiple contact points. And based on the maximum residual Calculate the temperature rise anomaly index for each time window. The mathematical expression for the temperature rise anomaly index is:

[0129]

[0130] In the formula, Represents the maximum residual The standard deviation of the residuals at the corresponding contact points under healthy conditions.

[0131] Normalization is performed by dividing the residuals by the standard deviation. As the residual standard deviation under healthy conditions, it reflects the range of fluctuation of the contact point under normal operating conditions. This indicates the multiple of abnormal residuals relative to healthy fluctuations, making the abnormality index comparable.

[0132] S140, Relevance Feature Extraction:

[0133] In the previous section, the pressure trend series and the temperature rise anomaly index series were time-aligned using a time window. This application uses the Pearson correlation coefficient to calculate the correlation between the two, and the mathematical expression for the correlation is as follows:

[0134]

[0135] In the formula, This represents a sequence of air pressure trends. This represents the mean of the air pressure trend series. This represents a temperature rise anomaly index sequence. This represents the mean of the temperature rise anomaly index series. The sequence length is given.

[0136] After the above feature extraction is completed, the feature vector for the current monitoring period is obtained. eigenvectors It can be represented as: .

[0137] The above process is a feature extraction process. After obtaining the feature vector, various methods can be used to map fault types. For example, AI models can be used, but due to the scarcity and imbalance of switchgear operation data, AI models easily learn to "always predict normal" to achieve high accuracy, but this is ultimately useless. Based on the above practical problems, this application adopts a fusion analysis scheme based on Bayesian + DS evidence theory, and the specific process is as follows:

[0138] S150, a fusion analysis based on Bayesian and DS evidence theories:

[0139] Before conducting the fusion analysis, this application pre-constructs a fusion model that integrates Bayesian and DS evidence theories. Figure 3 This is a schematic diagram of the fusion model construction process in one embodiment of this application, as shown below. Figure 3 As shown, the construction methods of the fusion model include:

[0140] (1) Obtain historical operating data samples of gas-protected medium-voltage switchgear for multiple historical time periods, wherein the historical operating data samples include overall temperature sequence samples, gas pressure sequence samples, bus load sequence samples and actual temperature rise sequence samples of multiple contact points;

[0141] The historical operating data samples obtained in this application cover the typical operating states of the switchgear under different operating conditions, including normal operation, gas leakage, overheating, and combined fault scenarios. The historical operating data samples are collected by the acquisition module described above and stored in the memory.

[0142] (2) Extract pressure trend samples, pressure trend confidence samples, temperature rise anomaly index samples, actual temperature rise average samples, and correlation samples from the historical operating data samples, and construct feature vector samples based on the pressure trend samples, the pressure trend confidence samples, the temperature rise anomaly index samples, the actual temperature rise average samples, and the correlation samples; and label the type of the historical operating data samples, wherein the type labels include normal, gas leak, overheating, and compound fault;

[0143] The process of extracting feature samples is the same as described above. Please refer to the previous text for understanding, and it will not be repeated here.

[0144] After data collection is complete, tags need to be manually assigned, including:

[0145]

[0146] After labeling, construct two feature vectors for each sample. Using the fewest dimensions, it distinguishes four states to the greatest extent possible (normal, leakage, overheating, and compound).

[0147] (3) Construct a Bayesian diagnostic model and a DS evidence model based on feature vector samples with multiple types of labels, and fuse the outputs of the Bayesian diagnostic model and the DS evidence model, specifically including:

[0148] (3-1) Construction of Bayesian diagnostic model:

[0149] (3-1-1) Calculate the prior probability of each type of label based on the label categories of multiple historical running data samples. The mathematical expression for the prior probability is:

[0150]

[0151] In the formula, This represents the total number of samples. Indicates the first The number of samples for each type of label. Indicates the first The probability of each type of label;

[0152] For example, if out of 1000 historical data entries, 970 are normal, 10 are leaking, 15 are overheating, and 5 are multiple faults, then:

[0153]

[0154] (3-1-2) Construct a likelihood function for the feature vector samples corresponding to each type label. The principle of the likelihood function is: for each fault hypothesis... We consider all the corresponding eigenvectors as a data cluster. We assume that the eigenvectors within each cluster follow a multivariate Gaussian distribution. Likelihood function The mathematical expression is:

[0155]

[0156] In the formula, The dimension of the feature vector; Representing state The mean vector under this fault mode represents the "typical center" of each feature value. Representing state The covariance matrix described below describes the range of fluctuations and correlations among the eigenvalues. Each eigenvalue can be directly calculated using historical data. The corresponding mean vector Covariance Matrix .

[0157] (3-1-3) Based on the aforementioned prior probability and the likelihood function Construct a Bayesian diagnostic model, wherein the mathematical expression of the Bayesian diagnostic model is:

[0158]

[0159] In the formula, Represents the posterior probability. Indicates the first The likelihood function for each type label. Indicates the first The prior probability of each type of label.

[0160] Obtain the prior probability and likelihood function Then, based on the standard form of Bayes' theorem, a Bayesian diagnostic model is constructed, and the output posterior probability is the current feature vector. Below, various types of tags The probability of.

[0161] (3-2) Construction of the DS evidence model:

[0162] (3-2-1) Construct the BPA function for pressure evidence, the BPA function for temperature evidence, and the BPA function for correlation evidence. The mathematical expressions for the BPA functions for pressure evidence, temperature evidence, and correlation evidence are as follows:

[0163]

[0164]

[0165]

[0166] In the formula, Indicates the level of confidence in the gas leak. Indicates the level of trust in contact overheating. Indicates the composite fault trust level. and All are weighting coefficients. This indicates the air pressure trend over a target time period. This is a threshold for air pressure trends generated based on feature vector samples. This indicates the temperature rise anomaly index for the target time period. This indicates that the threshold for generating the temperature rise anomaly index is based on feature vector samples. The correlation coefficient represents the target time period. It is an exponential function;

[0167] (3-2-2) Determine the confidence level of the gas leak. and the contact overheating confidence level Fusion was performed to obtain the gas-superheated confidence level. The gas-superheat confidence level With the aforementioned composite fault trust To achieve integration and gain integration trust. The fusion process takes into account the conflict coefficient.

[0168] Specifically, the mathematical expression for fusion is:

[0169]

[0170]

[0171] In the formula, and For two independent sources of evidence (trust level) that need to be merged, for example Evidence of stress; Temperature evidence.

[0172] and To represent a focal element, for example: , ; , ;

[0173] Indicates the target proposition;

[0174] This represents the conflict coefficient.

[0175] This indicates the level of trust after the integration.

[0176] (3-2-3) Based on the aforementioned fusion trust level Calculate the trust function and likelihood function , wherein the trust function and the likelihood function The mathematical expression is:

[0177]

[0178]

[0179] In the formula, Type label The corresponding set of fault states, Type label A subset of the corresponding set of fault states This indicates the trust level of the subset.

[0180] (3-3) Output fusion:

[0181] The outputs of the Bayesian diagnostic model and the DS evidence model are fused, and the mathematical expression for the fused output is as follows:

[0182]

[0183] In the formula, This indicates the type label determination result of the output. As the first weight, As the second weight, As the third weight, Indicates the first Type tags, The features represented by the Bayesian diagnostic model output For type tags The posterior probability, This indicates that the evidence output by the DS evidence model is related to the type label. The minimum trust value, This indicates that the evidence output by the DS evidence model is for type labeling. The highest trust value.

[0184] The weights of the fused output are determined using expert experience or cross-validation.

[0185] After constructing the above fusion model, the feature vector of the current monitoring period will be... By substituting the inputs into the Bayesian diagnostic model and the DS evidence model respectively, and fusing the inputs, the output type label determination result can be obtained. .

[0186] Finally, if the monitoring results contain any type of fault, the monitoring results will be sent to the target object to alert relevant personnel that a fault exists.

[0187] This application discloses a monitoring method for gas-protected medium-voltage switchgear, which monitors the gas pressure, temperature, and bus load within the switchgear. A collaborative monitoring system integrating multi-dimensional information on pressure, temperature, and current is constructed, overcoming the limitations of single-parameter monitoring. It innovatively combines Bayesian probability and DS evidence theory, utilizing the objectivity of statistical data while properly handling evidence conflicts and cognitive uncertainties, upgrading diagnosis from threshold alarms to probabilistic trust assessments. This application can effectively monitor the current state of the switchgear in real time, driving the operation and maintenance mode from periodic inspections to predictive maintenance, significantly reducing operation and maintenance costs and the risk of unplanned power outages while ensuring power supply reliability.

[0188] like Figure 4 As shown, this application also provides a monitoring system for a gas-protected medium-voltage switchgear, comprising:

[0189] The acquisition module is used to acquire the overall temperature sequence, gas pressure sequence, bus load sequence, and actual temperature rise sequence of multiple contact points collected within a target time period inside the gas-protected medium-voltage switchgear. The target time period includes multiple time points of a target duration prior to the current time point.

[0190] The barometric pressure analysis module is used to divide the target time period into multiple time windows; compensate the barometric pressure sequences of the multiple time windows based on the overall temperature sequence of the multiple time windows to obtain a standard barometric pressure sequence at a standard temperature; and extract the barometric pressure trend and the confidence level of the barometric pressure trend from the standard barometric pressure sequence.

[0191] The temperature rise analysis module is used to calculate the theoretical temperature rise of the contact point based on the pre-built contact point temperature rise prediction model and the bus load of multiple time windows, and to calculate the residual between the theoretical temperature rise and the actual temperature rise of the contact point in multiple time windows. It also extracts the temperature rise anomaly index of the contact point corresponding to the maximum residual in each time window, and calculates the average actual temperature rise of multiple contact points corresponding to the current time window.

[0192] The correlation extraction module is used to construct a pressure trend sequence based on pressure trends over multiple time windows; construct a temperature rise anomaly index sequence based on temperature rise anomaly indices over multiple time windows; and calculate the correlation between the pressure trend sequence and the temperature rise anomaly index sequence.

[0193] The fusion monitoring module is used to construct a feature vector based on the air pressure trend of the current time window, the confidence level of the air pressure trend of the current time window, the temperature rise anomaly index of the current time window, the actual average temperature rise of the current time window, and the correlation. The feature vector is then input into a pre-constructed fusion model to obtain the monitoring results. The fusion model integrates a Bayesian diagnostic model and DS evidence theory.

[0194] This application discloses a monitoring system for gas-protected medium-voltage switchgear, which monitors gas pressure, temperature, and bus load within the switchgear. It constructs a collaborative monitoring system integrating multi-dimensional information on pressure, temperature, and current, overcoming the limitations of single-parameter monitoring. Innovatively combining Bayesian probability and DS evidence theory, it utilizes the objectivity of statistical data while properly handling evidence conflicts and cognitive uncertainties, upgrading diagnosis from threshold alarms to probabilistic trust assessment. This application can effectively monitor the current state of the switchgear in real time, driving the operation and maintenance mode from periodic inspections to predictive maintenance, significantly reducing operation and maintenance costs and the risk of unplanned power outages while ensuring power supply reliability.

[0195] This embodiment also provides an electronic terminal, including: a processor and a memory;

[0196] The memory is used to store computer programs, and the processor is used to execute the computer programs stored in the memory so that the terminal performs any of the methods in this embodiment.

[0197] As will be understood by those skilled in the art, the computer-readable storage medium described in this embodiment allows for the implementation of all or part of the steps in the above method embodiments by computer program-related hardware. The aforementioned computer program can be stored in a computer-readable storage medium. When executed, the program performs the steps of the above method embodiments; and the aforementioned storage medium includes various media capable of storing program code, such as ROM, RAM, magnetic disks, or optical disks.

[0198] The electronic terminal provided in this embodiment includes a processor, a memory, a transceiver, and a communication interface. The memory and the communication interface are connected to the processor and the transceiver and complete communication between them. The memory is used to store computer programs, the communication interface is used to perform communication, and the processor and the transceiver are used to run the computer programs, so that the electronic terminal performs the steps of the above method.

[0199] In this embodiment, the memory may include random access memory (RAM) and may also include non-volatile memory, such as at least one disk storage device.

[0200] The processors mentioned above can be general-purpose processors, including central processing units (CPUs), network processors (NPs), etc.; they can also be digital signal processors (DSPs), application-specific integrated circuits (ASICs), field-programmable gate arrays (FPGAs), or other programmable logic devices, discrete gate or transistor logic devices, or discrete hardware components.

[0201] In the above embodiments, although the present application has been described in conjunction with specific embodiments thereof, many substitutions, modifications, and variations of these embodiments will be apparent to those skilled in the art based on the foregoing description. The embodiments of the present application are intended to cover all such substitutions, modifications, and variations falling within the broad scope of the appended claims.

[0202] The above embodiments are merely illustrative of the principles and effects of this application and are not intended to limit this application. Any person skilled in the art can modify or alter the above embodiments without departing from the spirit and scope of this application. Therefore, all equivalent modifications or alterations made by those skilled in the art without departing from the spirit and technical concept disclosed in this application should still be covered by the claims of this application.

Claims

1. A monitoring method for a gas-protected medium-voltage switchgear, characterized in that, Including the following steps: The overall temperature sequence, gas pressure sequence, bus load sequence, and actual temperature rise sequence of multiple contact points are acquired within a target time period in the gas-protected medium-voltage switchgear. The target time period includes multiple time points with a target duration prior to the current time point. The target time period is divided into multiple time windows; the air pressure sequence of the multiple time windows is compensated based on the overall temperature sequence of the multiple time windows to obtain the standard air pressure sequence at the standard temperature; and the air pressure trend and the confidence level of the air pressure trend are extracted from the standard air pressure sequence. The theoretical temperature rise of the contact point is calculated based on the pre-built contact point temperature rise prediction model and the bus load of multiple time windows. The residual between the theoretical temperature rise and the actual temperature rise of the contact point in multiple time windows is calculated, and the temperature rise anomaly index of the contact point corresponding to the maximum residual of each time window is extracted. The average actual temperature rise of multiple contact points corresponding to the current time window is calculated. A pressure trend sequence is constructed based on pressure trends over multiple time windows; a temperature rise anomaly index sequence is constructed based on temperature rise anomaly indices over multiple time windows; and the correlation between the pressure trend sequence and the temperature rise anomaly index sequence is calculated. A feature vector is constructed based on the air pressure trend, confidence level of the air pressure trend, temperature rise anomaly index, actual average temperature rise, and correlation of the current time window. The feature vector is then input into a pre-constructed fusion model to obtain the monitoring results. The fusion model integrates a Bayesian diagnostic model and DS evidence theory.

2. The monitoring method for a gas-protected medium-voltage switchgear according to claim 1, characterized in that, The pressure sequence of multiple time windows is compensated based on the overall temperature sequence of multiple time windows to obtain the standard pressure sequence at the standard temperature. And extract the pressure trend and the confidence level of the pressure trend from the standard pressure sequence, including: Aligning the overall temperature series and the air pressure series based on time points yields multiple sets of basic data. ,in, Indicates a point in time The corresponding air pressure Indicates a point in time The corresponding temperature; For each time point Based on temperature air pressure Compensation is performed to obtain the equivalent air pressure representing 20 degrees Celsius. And the equivalent pressure sequence, wherein the mathematical expression for the equivalent pressure is: The slope of the equivalent pressure sequence for the current time window is extracted using the sliding window least squares method. and coefficient of determination The slope As a pressure trend, and the determining coefficient As a confidence level.

3. The monitoring method for a gas-protected medium-voltage switchgear according to claim 1, characterized in that, The method for constructing the contact point temperature rise prediction model includes: Acquire multiple temperature rise data samples when the switchgear is in a healthy state, wherein the temperature rise data samples include bus load and temperature rise value samples; The basic mathematical expression for constructing the contact point temperature rise prediction model is as follows: In the formula, Indicates the first Temperature rise coefficient at each contact point Indicates the bus load. Indicates the first Temperature rise at each contact point Indicates the offset; Substituting multiple temperature rise data samples into the basic mathematical expression and combining them with the least squares method for fitting, the temperature rise coefficient is obtained. And a model for predicting temperature rise at contact points.

4. The monitoring method for a gas-protected medium-voltage switchgear according to claim 1, characterized in that, Based on a pre-built contact point temperature rise prediction model and bus loads over multiple time windows, the theoretical temperature rise of the contact point is calculated. The residuals between the theoretical and actual temperature rises of the contact point over multiple time windows are also calculated. The temperature rise anomaly index corresponding to the maximum residual in each time window is extracted, including: For each time window, the bus load at each time point will be... Substituting the values ​​into the contact point temperature rise prediction model for multiple contact points, we obtain the theoretical temperature rise of multiple contact points at each time point. ; For each contact point, calculate the average theoretical temperature rise within each time window. and the average of the actual temperature rise ; For the same time window, calculate the average theoretical temperature rise. and the average of the actual temperature rise residual ,in: In the formula, Indicates the time window index; At each time point, extract the maximum residual from references at multiple contact points. And based on the maximum residual Calculate the temperature rise anomaly index for each time window. The mathematical expression for the temperature rise anomaly index is: In the formula, Represents the maximum residual The standard deviation of the residuals at the corresponding contact points under healthy conditions.

5. The monitoring method for a gas-protected medium-voltage switchgear according to claim 1, characterized in that, The mathematical expression for the relevance is: In the formula, This represents a sequence of air pressure trends. This represents the mean of the air pressure trend series. This represents a temperature rise anomaly index sequence. This represents the mean of the temperature rise anomaly index series. The sequence length is given.

6. The monitoring method for a gas-protected medium-voltage switchgear according to claim 1, characterized in that, The method for constructing the fusion model includes: Acquire historical operating data samples of the gas-protected medium-voltage switchgear for multiple historical time periods. The historical operating data samples include overall temperature sequence samples, gas pressure sequence samples, bus load sequence samples, and actual temperature rise sequence samples of multiple contact points. Extract pressure trend samples, pressure trend confidence samples, temperature rise anomaly index samples, actual temperature rise average samples, and correlation samples from the historical operating data samples. Construct feature vector samples based on the pressure trend samples, pressure trend confidence samples, temperature rise anomaly index samples, actual temperature rise average samples, and correlation samples. Label the type of the historical operating data samples, wherein the type labels include normal, gas leak, overheating, and combined fault. A Bayesian diagnostic model and a DS evidence model are constructed based on feature vector samples with various types of labels. The outputs of the Bayesian diagnostic model and the DS evidence model are then fused. The mathematical expression for the fused output is as follows: In the formula, This indicates the type label determination result of the output. As the first weight, As the second weight, As the third weight, Indicates the first Type labels, The features represented by the Bayesian diagnostic model output For type tags The posterior probability, This indicates that the evidence output by the DS evidence model is related to the type label. The minimum trust value, This indicates that the evidence output by the DS evidence model is for type labeling. The highest trust value.

7. The monitoring method for a gas-protected medium-voltage switchgear according to claim 6, characterized in that, A Bayesian diagnostic model is constructed based on feature vector samples with various types of labels, including: Calculate the prior probability of each label type based on multiple historical operational data samples. The mathematical expression for the prior probability is: In the formula, This represents the total number of samples. Indicates the first The number of samples for each type of label. Indicates the first The probability of each type of label; Construct a likelihood function for the feature vector samples corresponding to each type label. ; Based on the prior probability and the likelihood function Construct a Bayesian diagnostic model, wherein the mathematical expression of the Bayesian diagnostic model is: In the formula, Represents the posterior probability. Indicates the first The likelihood function for each type label. Indicates the first The prior probability of each type of label.

8. The monitoring method for a gas-protected medium-voltage switchgear according to claim 6, characterized in that, A DS evidence model is constructed based on feature vector samples with various types of labels, including: Construct the BPA function for pressure evidence, the BPA function for temperature evidence, and the BPA function for correlation evidence. The mathematical expressions for the BPA functions for pressure evidence, temperature evidence, and correlation evidence are as follows: In the formula, Indicates the level of confidence in the gas leak. Indicates the level of trust in contact overheating. Indicates the composite fault trust level. and All are weighting coefficients. This indicates the air pressure trend over a target time period. This is a threshold for air pressure trends generated based on feature vector samples. This indicates the temperature rise anomaly index for the target time period. This indicates that the threshold for generating the temperature rise anomaly index is based on feature vector samples. The correlation coefficient represents the target time period. It is an exponential function; Trust level of the gas leak and the contact overheating confidence level Fusion was performed to obtain the gas-superheated confidence level. The gas-superheat confidence level With the aforementioned composite fault trust To achieve integration and gain integration trust. The fusion process takes into account the conflict coefficient. Based on the aforementioned fusion trust level Calculate the trust function and likelihood function The trust function and the likelihood function The mathematical expression is: In the formula, Type label The corresponding set of fault states, Type label A subset of the corresponding fault state set, This indicates the trust level of the subset.

9. A monitoring method for a gas-protected medium-voltage switchgear according to claim 1, characterized in that, Also includes: When the monitoring results contain any type of fault, the monitoring results are sent to the target object.

10. A monitoring system for a gas-protected medium-voltage switchgear, characterized in that, include: The acquisition module is used to acquire the overall temperature sequence, gas pressure sequence, bus load sequence, and actual temperature rise sequence of multiple contact points collected within a target time period inside the gas-protected medium-voltage switchgear. The target time period includes multiple time points of a target duration prior to the current time point. The barometric pressure analysis module is used to divide the target time period into multiple time windows; compensate the barometric pressure sequences of the multiple time windows based on the overall temperature sequence of the multiple time windows to obtain a standard barometric pressure sequence at a standard temperature; and extract the barometric pressure trend and the confidence level of the barometric pressure trend from the standard barometric pressure sequence. The temperature rise analysis module is used to calculate the theoretical temperature rise of the contact point based on the pre-built contact point temperature rise prediction model and the bus load of multiple time windows, and to calculate the residual between the theoretical temperature rise and the actual temperature rise of the contact point in multiple time windows. It also extracts the temperature rise anomaly index of the contact point corresponding to the maximum residual in each time window, and calculates the average actual temperature rise of multiple contact points corresponding to the current time window. The correlation extraction module is used to construct a pressure trend sequence based on pressure trends over multiple time windows; construct a temperature rise anomaly index sequence based on temperature rise anomaly indices over multiple time windows; and calculate the correlation between the pressure trend sequence and the temperature rise anomaly index sequence. The fusion monitoring module is used to construct a feature vector based on the air pressure trend of the current time window, the confidence level of the air pressure trend of the current time window, the temperature rise anomaly index of the current time window, the actual average temperature rise of the current time window, and the correlation. The feature vector is then input into a pre-constructed fusion model to obtain the monitoring results. The fusion model integrates a Bayesian diagnostic model and DS evidence theory.

Images