GIS bus cabin air leakage state early warning method and related device

Abstract

The invention belongs to the technical field of power transmission, and discloses a GIS bus cabin air leakage state early warning method and a related device, and the method comprises the steps: obtaining historical meteorological data, historical sensor data and equipment log data of a GIS bus cabin; performing feature extraction on the equipment log data of the GIS bus cabin to obtain an equipment log feature vector; and based on the historical meteorological data, the historical sensor data and the equipment log feature vector of the GIS bus cabin, obtaining an air leakage state early warning result of the GIS bus cabin through a pre-trained air leakage state early warning model of the GIS bus cabin. Through the innovatively designed GIS bus cabin gas leakage state early warning model, efficient fusion and deep mining of multi-source data are realized, accurate prediction of the gas leakage state of the GIS bus cabin is realized by taking the multi-source data as a support, the accuracy and timeliness of gas leakage state early warning are further remarkably improved, more valuable decision support is provided for operation and maintenance personnel, and the early warning efficiency of the gas leakage state of the GIS bus cabin is improved. And the adaptability and robustness of the model to complex environment changes are enhanced.

Description

Technical Field

[0001] The invention belongs to the technical field of power transmission, and relates to a GIS busbar compartment air leakage state early warning method and related devices. Background Art

[0002] At present, the gas leakage maintenance of GIS (Gas Insulated Switchgear) busbar compartment is mainly to check and repair the leakage point after the pressure in the GIS busbar compartment drops and alarms. Although this method of early warning after the leakage occurs has advanced the leakage treatment time to a certain extent, it still has certain risks. For example, the leakage of SF6 gas in the GIS busbar compartment will directly affect the insulation performance and operation stability of the equipment, and even threaten the life safety of on-site workers. At the same time, SF6 gas poses a potential threat to the environment. The leakage of SF6 gas into the atmosphere will aggravate global warming.

[0003] On this basis, researchers began to study early warning methods for SF6 gas leakage in GIS busbar compartments, in order to achieve early warning and preventive maintenance of SF6 gas leakage in GIS busbar compartments, thereby reducing the risk of leakage. The current early warning method only considers single factors such as the gas pressure or temperature of SF6 gas for prediction. However, the leakage state of the GIS busbar compartment is affected by many complex factors, and the single-factor model is difficult to fully capture the interaction of these factors, resulting in the gradual decrease in the accuracy of the early warning over time. In addition, the operating environment of the GIS busbar compartment is complex and changeable, with significant differences in different regions. Existing early warning models often lack sufficient flexibility to adapt to these changes, resulting in large differences in early warning effects in different scenarios. Summary of the invention

[0004] The purpose of the present invention is to overcome the above-mentioned shortcomings of the prior art and provide a GIS busbar compartment air leakage state early warning method and related devices.

[0005] In order to achieve the above object, the present invention adopts the following technical solutions:

[0006] In a first aspect, the present invention provides a GIS bus compartment gas leakage status warning method, comprising: obtaining historical meteorological data, historical sensor data and equipment log data of the GIS bus compartment; extracting features of the equipment log data of the GIS bus compartment to obtain equipment log feature quantities; based on the historical meteorological data, historical sensor data and equipment log feature vectors of the GIS bus compartment, obtaining a GIS bus compartment gas leakage status warning result through a pre-trained GIS bus compartment gas leakage status warning model; wherein the GIS bus compartment gas leakage status warning model comprises a convolutional neural network, a long short-term memory network and a Transformer network; the convolutional neural network is used to extract features of historical meteorological data and historical sensor data, obtain meteorological feature vectors and sensor feature vectors and send them to the long short-term memory network; the long short-term memory network is used to fuse the equipment log feature vector, the meteorological feature vector and the sensor feature vector to obtain a fused feature vector; the Transformer network is used to obtain a GIS bus compartment gas pressure prediction result according to the fused feature vector.

[0007] Optionally, the historical meteorological data includes one or more of the following: light intensity, temperature, air pressure, humidity and precipitation.

[0008] Optionally, the historical sensor data includes one or more of the following: gas pressure, gas temperature, humidity level, gas pressure change rate, gas temperature change rate, pressure difference between inside and outside of the GIS busbar compartment, and gas frequency domain data.

[0009] Optionally, the step of extracting features from the equipment log data of the GIS bus compartment to obtain a device log feature vector includes: performing data preprocessing on the equipment log data of the GIS bus compartment; and using a pretrained BERT model to extract features from the equipment log data of the GIS bus compartment after data preprocessing to obtain a device log feature vector.

[0010] Optionally, before obtaining the GIS bus compartment air leakage status warning result through a pre-trained GIS bus compartment air leakage status warning model based on the historical meteorological data, historical sensor data and equipment log feature vector of the GIS bus compartment, the method also includes: preprocessing the historical meteorological data and historical sensor data of the GIS bus compartment; aligning the historical meteorological data, historical sensor data and equipment log feature vector of the GIS bus compartment after data preprocessing at an hourly frequency according to timestamps; wherein the meteorological data and sensor data are converted into an hourly frequency by an average method; and the equipment log feature vector is converted into an hourly frequency by a repetition method.

[0011] Optionally, the GIS bus compartment gas leakage status warning result is obtained through a pre-trained GIS bus compartment gas leakage status warning model based on the historical meteorological data, historical sensor data and equipment log feature vector of the GIS bus compartment, including: inputting the meteorological data, sensor data and equipment log feature vector of the GIS bus compartment into the pre-trained GIS bus compartment gas leakage status warning model to obtain the GIS bus compartment gas pressure prediction result; and temperature compensating the GIS bus compartment gas pressure prediction result; obtaining the GIS bus compartment gas leakage status prediction situation according to the temperature-compensated GIS bus compartment gas pressure prediction result and a preset GIS bus compartment gas pressure threshold; and obtaining the GIS bus compartment gas leakage status warning result according to the GIS bus compartment gas leakage status prediction situation.

[0012] Optionally, the GIS bus compartment gas leakage status warning result is obtained through a pre-trained GIS bus compartment gas leakage status warning model based on the historical meteorological data, historical sensor data and equipment log feature vectors of the GIS bus compartment, including: based on the historical meteorological data, historical sensor data and equipment log feature vectors of the GIS bus compartment, according to a preset confidence level, through a Bayesian deep learning method, a GIS bus compartment gas pressure prediction result and a confidence interval of the GIS bus compartment gas pressure prediction result are obtained; the GIS bus compartment gas pressure prediction result is temperature compensated, and the GIS bus compartment gas leakage status prediction situation is obtained according to the temperature compensated GIS bus compartment gas pressure prediction result and a preset GIS bus compartment gas pressure threshold; and according to a preset confidence interval width threshold, in combination with the GIS bus compartment gas leakage status prediction situation, a GIS bus compartment gas leakage status warning result is obtained.

[0013] In a second aspect, the present invention provides a GIS busbar compartment gas leakage status warning system, comprising: a data acquisition module, used to acquire historical meteorological data, historical sensor data and equipment log data of the GIS busbar compartment; a feature extraction module, used to extract features from the equipment log data of the GIS busbar compartment to obtain equipment log feature vectors; a model prediction module, used to obtain a GIS busbar compartment gas leakage status warning result through a pre-trained GIS busbar compartment gas leakage status warning model based on the historical meteorological data, historical sensor data and equipment log feature vectors of the GIS busbar compartment; wherein the GIS busbar compartment gas leakage status warning model comprises a convolutional neural network, a long short-term memory network and a Transformer network; the convolutional neural network is used to extract features of historical meteorological data and historical sensor data, obtain meteorological feature vectors and sensor feature vectors and send them to the long short-term memory network; the long short-term memory network is used to fuse the equipment log feature vector, the meteorological feature vector and the sensor feature vector to obtain a fused feature vector; the Transformer network is used to obtain a GIS busbar compartment gas pressure prediction result based on the fused feature vector.

[0014] A third aspect of the present invention provides a computer device, comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor implements the steps of the above-mentioned GIS busbar compartment air leakage status warning method when executing the computer program.

[0015] According to a fourth aspect of the present invention, a computer-readable storage medium is provided, wherein the computer-readable storage medium stores a computer program, and when the computer program is executed by a processor, the steps of the above-mentioned GIS busbar compartment air leakage status warning method are implemented.

[0016] Compared with the prior art, the present invention has the following beneficial effects:

[0017] The GIS busbar compartment leakage status warning method of the present invention realizes efficient fusion and deep mining of multi-source data through an innovatively designed GIS busbar compartment leakage status warning model, and uses multi-source data as a support to accurately predict the leakage status of the GIS busbar compartment, thereby significantly improving the accuracy and timeliness of the leakage status warning, providing more valuable decision support for operation and maintenance personnel, and enhancing the model's adaptability and robustness to complex environmental changes. Specifically, the convolutional neural network, as the front-end processor of the model, can accurately capture subtle changes in meteorological data and sensor data with its excellent local feature extraction capabilities, which are often early signals of abnormal equipment status. Through the synergistic effect of the convolutional layer and the pooling layer, the convolutional neural network not only effectively filters out noise interference in the data, but also greatly reduces the data dimension, providing a clearer and more compact feature representation for subsequent processing, thereby accelerating the training process of the long short-term memory network and the Transformer network. In addition, the meteorological data and sensor data are processed by convolutional neural network before time series analysis, which can effectively strengthen the local correlation between the gas pressure in the GIS busbar compartment and meteorological conditions and equipment conditions, making the model more sensitive and reliable when facing equipment anomalies or meteorological changes, thereby further improving the model's adaptability to complex environmental changes and the accuracy of early warning. As the core of the model's time series analysis, the long short-term memory network has a unique memory mechanism that enables the model to deeply understand the long-term dependencies between data, especially the intertwined influence of short-term fluctuations and long-term trends. The long short-term memory network plays a key role in connecting the past and the future. It cleverly integrates the local features extracted by the convolutional neural network with the text information in the equipment log data to form a fused feature vector containing rich spatiotemporal information. This fusion process not only enhances the model's sensitivity to changes in equipment status, but also provides a solid foundation for subsequent global dependency modeling. The introduction of the Transformer network further enhances the model’s ability to capture complex global dependencies. Through its powerful self-attention mechanism, the Transformer network can deeply explore the deep correlations in the fused feature vectors, thereby achieving accurate prediction of the gas pressure in the GIS busbar compartment. This global dependency modeling capability enables the model to more accurately identify potential leakage risks, providing strong support for timely warnings. BRIEF DESCRIPTION OF THE DRAWINGS

[0018] Figure 1 The present invention is a flowchart of a GIS busbar compartment air leakage status early warning method according to an embodiment of the present invention.

[0019] Figure 2 It is a schematic diagram of the framework of the GIS busbar compartment air leakage status warning model according to an embodiment of the present invention.

[0020] Figure 3 Schematic diagram of data alignment principle according to an embodiment of the present invention.

[0021] Figure 4 The present invention is a block diagram of the structure of the GIS busbar compartment air leakage status warning system according to an embodiment of the present invention. DETAILED DESCRIPTION

[0022] In order to enable those skilled in the art to better understand the scheme of the present invention, the technical scheme in the embodiments of the present invention will be clearly and completely described below in conjunction with the drawings in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work should fall within the scope of protection of the present invention.

[0023] It should be noted that the terms "first", "second", etc. in the specification and claims of the present invention and the above-mentioned drawings are used to distinguish similar objects, and are not necessarily used to describe a specific order or sequence. It should be understood that the data used in this way can be interchanged where appropriate, so that the embodiments of the present invention described herein can be implemented in an order other than those illustrated or described herein. In addition, the terms "including" and "having" and any variations thereof are intended to cover non-exclusive inclusions, for example, a process, method, system, product or device that includes a series of steps or units is not necessarily limited to those steps or units that are clearly listed, but may include other steps or units that are not clearly listed or inherent to these processes, methods, products or devices.

[0024] The present invention is further described in detail below in conjunction with the accompanying drawings:

[0025] See also Figure 1 and Figure 2 In one embodiment of the present invention, a GIS busbar compartment air leakage status warning method is provided, specifically a GIS busbar compartment air leakage status warning method based on multi-source data fusion, which can greatly improve the accuracy and adaptability of GIS busbar compartment air leakage status warning.

[0026] Specifically, the GIS busbar compartment air leakage state early warning method of the present invention comprises the following steps:

[0027] S1: Obtain historical meteorological data, historical sensor data, and equipment log data of the GIS busbar compartment.

[0028] S2: Extract features from the equipment log data of the GIS busbar compartment to obtain equipment log feature vectors.

[0029] S3: Based on the historical meteorological data, historical sensor data and equipment log feature vector of the GIS bus compartment, the GIS bus compartment air leakage status warning result is obtained through the pre-trained GIS bus compartment air leakage status warning model.

[0030] Among them, the GIS busbar compartment gas leakage status warning model includes convolutional neural network, long short-term memory network and Transformer network; the convolutional neural network is used to extract the characteristics of historical meteorological data and historical sensor data, obtain meteorological feature vectors and sensor feature vectors and send them to the long short-term memory network; the long short-term memory network is used to fuse the equipment log feature vector, meteorological feature vector and sensor feature vector to obtain the fused feature vector; the Transformer network is used to obtain the GIS busbar compartment gas pressure prediction result according to the fused feature vector.

[0031] Explanatory, sensor data refers to the relevant parameter data of the GIS bus compartment detected by sensors installed inside and outside the GIS bus compartment. Historical meteorological data, historical sensor data and equipment log data refer to meteorological data, sensor data and equipment log data within a certain period of time before the prediction period.

[0032] Explanation: Convolutional neural network (CNN) is a type of feedforward neural network with deep structure and convolution operation, and is one of the representative algorithms of deep learning. The design of CNN is inspired by the ability of animal visual system to process information hierarchically. It extracts local features of input data through convolution operation, forms complex feature representation through multi-layer convolution and pooling operation, and finally performs classification or regression tasks through fully connected layer.

[0033] Long Short-Term Memory (LSTM) is a special type of recurrent neural network (RNN) designed to solve the problem of gradient vanishing and gradient exploding when traditional RNN cannot effectively process long sequences. By introducing unique gating mechanisms and memory units, LSTM is able to capture long-span dependencies and has become a core model in tasks such as natural language processing, speech recognition, and time series prediction. LSTM introduces three gating mechanisms: forget gate, input gate, and output gate, which allow the network to dynamically decide whether to retain or forget information.

[0034] The Transformer network is a sequence model based on the attention mechanism, which was originally proposed by Google's research team and applied to machine translation tasks. The Transformer network abandons the traditional loop and convolution structure and only uses the self-attention mechanism to process the input sequence and the output sequence. The self-attention mechanism allows the model to compare each element in the input sequence with other elements when processing the sequence, so as to correctly process each element in different contexts. The Transformer network has stronger parallel computing capabilities, stronger ability to process long sequences, and can directly process the entire sequence.

[0035] The GIS busbar compartment leakage status early warning method of the present invention realizes efficient fusion and deep mining of multi-source data through an innovatively designed GIS busbar compartment leakage status early warning model, and uses multi-source data as support to accurately predict the leakage status of the GIS busbar compartment, thereby significantly improving the accuracy and timeliness of the leakage status early warning, providing more valuable decision support for operation and maintenance personnel, and enhancing the model's adaptability and robustness to complex environmental changes. It can effectively solve the problem that the single factor model is difficult to fully capture the interaction of these factors, which leads to the gradual decrease in early warning accuracy over time.

[0036] In addition, it can also solve the problem that due to the complex and changeable operating environment of the GIS busbar compartment, the meteorological conditions, equipment types and operation and maintenance strategies in different regions vary significantly, resulting in the existing early warning models often lacking sufficient flexibility to adapt to these changes, causing the early warning effects to vary greatly in different scenarios.

[0037] Specifically, as the front-end processor of the model, the convolutional neural network can accurately capture subtle changes in meteorological data and sensor data with its excellent local feature extraction capabilities, which are often early signals of abnormal equipment status. Through the synergy of convolutional layers and pooling layers, the convolutional neural network not only effectively filters out noise interference in the data, but also greatly reduces the data dimension, providing a clearer and more compact feature representation for subsequent processing, thereby accelerating the training process of long short-term memory networks and Transformer networks. In addition, processing meteorological data and sensor data through convolutional neural networks before performing time series analysis can effectively strengthen the local correlation between the gas pressure in the GIS busbar compartment and meteorological conditions and equipment conditions, making the model more sensitive and reliable when facing equipment anomalies or meteorological changes, thereby further improving the model's adaptability to complex environmental changes and the accuracy of early warnings.

[0038] As the core of the model's time series analysis, the long short-term memory network has a unique memory mechanism that enables the model to deeply understand the long-term dependencies between data, especially the intertwined influence of short-term fluctuations and long-term trends. The long short-term memory network plays a key role in connecting the local feature extraction (convolutional neural network) and the global time dependency modeling (Transformer network). It cleverly combines the local features extracted by the convolutional neural network with the text information in the equipment log data, effectively capturing the long-term and short-term patterns in the time series data and forming a fused feature vector containing rich spatiotemporal information. This fusion process not only enhances the model's sensitivity to changes in equipment status and the timeliness and accuracy of leakage prediction, but also provides a solid foundation for subsequent global dependency modeling, enabling the model to better cope with the complex environmental changes of the GIS busbar compartment and provide strong support for equipment maintenance and early warning.

[0039] The introduction of the Transformer network is used to enhance the time series features extracted by the long short-term memory network and further improve the model's ability to capture complex global dependencies. Through its powerful self-attention mechanism, the Transformer network can deeply explore the deep associations in the fused feature vector and more effectively capture the complex patterns in sensor data and meteorological data, thereby achieving accurate prediction of the gas pressure in the GIS busbar. This global dependency modeling capability enables the model to more accurately identify potential leakage risks and provides strong support for timely warnings. The local feature extraction of the Transformer network and the convolutional neural network and the short-term memory capacity of the long short-term memory network complement each other, making the entire model more robust and more accurate in prediction.

[0040] In a possible implementation, the historical meteorological data includes one or more of the following: light intensity, air temperature, air pressure, humidity and precipitation. Optionally, the historical sensor data includes one or more of the following: gas pressure, gas temperature, humidity level, gas pressure change rate, gas temperature change rate, GIS busbar internal and external pressure difference and gas frequency domain data.

[0041] Explanatory, the premise of multi-source data fusion is to obtain diversified data from different sources. In the GIS busbar compartment leakage status monitoring, these diversified data include meteorological data, sensor data and equipment log data. Among them, meteorological data generally selects light intensity, temperature, air pressure, humidity and precipitation, etc., and sensor data generally selects gas pressure, gas temperature, humidity level, gas pressure change rate, gas temperature change rate, GIS busbar compartment internal and external pressure difference and gas frequency domain data; among them, gas frequency domain data is obtained by converting gas pressure, gas temperature, gas pressure change rate, gas temperature change rate and / or GIS busbar compartment internal and external pressure difference from time domain to frequency domain through Fourier transformation to reveal the periodic changes and potential anomalies of gas state. Equipment log data generally selects operation records and maintenance records, etc.

[0042] In a possible implementation, the feature extraction of the equipment log data of the GIS bus compartment to obtain the equipment log feature vector includes: preprocessing the equipment log data of the GIS bus compartment; using a pre-trained BERT model to extract features from the equipment log data of the GIS bus compartment after data preprocessing to obtain the equipment log feature vector.

[0043] Explanatory, meteorological data, sensor data and equipment log data have differences in format, unit, precision, etc., which brings challenges to data fusion. Among them, equipment log data is mainly text-type data, while meteorological data and sensor data are usually numerical data.

[0044] Therefore, in order to realize the fusion of meteorological data, sensor data and equipment log data, different processing methods are adopted for meteorological data, sensor data and equipment log data.

[0045] Among them, the pre-trained BERT model is used to extract features from the device log data to obtain the device log feature vector. However, due to factors such as the professional quality, work attitude and recording habits of the recorder, the device log data may have omissions and errors, so it is necessary to preprocess the device log data first, including processing abnormal data, completing missing data, deleting duplicate recorded data, and correcting the time format. Among them, for missing data in the device log data, if there are fewer missing items in the missing data, the missing symbol (such as 0) is used to complete it. If there are more missing items in the missing data, the missing data is deleted. After data preprocessing, the pre-trained BERT model is used to extract text embedding from the device log data, and the device log data is converted into a quantitative feature vector, namely the device log feature vector.

[0046] Explanation: The BERT (Bidirectional Encoder Representations from Transformers) model is a pre-trained language representation model based on the Transformer architecture, proposed by Google in 2018. The BERT model learns rich language knowledge and contextual information through unsupervised learning on large-scale text data, and can be used in various natural language processing (NLP) tasks, such as text classification, named entity recognition, sentiment analysis, and question-answering systems.

[0047] In one possible implementation, see Figure 3 , the method based on the historical meteorological data, historical sensor data and equipment log feature vector of the GIS bus compartment, before obtaining the GIS bus compartment gas leakage status warning result through the pre-trained GIS bus compartment gas leakage status warning model also includes: preprocessing the historical meteorological data and historical sensor data of the GIS bus compartment; aligning the historical meteorological data, historical sensor data and equipment log feature vector of the GIS bus compartment after data preprocessing according to the hourly frequency according to the timestamp; wherein the meteorological data and sensor data are converted into the hourly frequency by the average method; the equipment log feature vector is converted into the hourly frequency by the repetition method.

[0048] Exemplarily, for data preprocessing of meteorological data and sensor data, the mean method can be used to fill in missing values, and all non-time data can be uniformly normalized.

[0049] For example, the GIS busbar compartment gas leakage status warning method based on multi-source data fusion is generally designed to provide a long-term warning of 3 to 5 days, that is, combining the historical meteorological data, historical sensor data and equipment log data of the GIS busbar compartment, and obtaining the GIS busbar compartment gas pressure prediction results for the next 3 to 5 days at one time. Therefore, in order to achieve periodic warning, all source data are unified into hourly frequencies, sensor data and meteorological data are converted to hourly frequencies by the average method, and the equipment log feature vectors are converted to hourly frequencies by repeating the timestamps, which increases the data frequency without introducing artificial errors.

[0050] For example, the prediction result of the gas pressure in the GIS bus compartment in the next 3 to 5 days may be a predicted value of the gas pressure in the GIS bus compartment at an hourly frequency.

[0051] In a possible implementation, the GIS bus compartment gas leakage status warning result is obtained through a pre-trained GIS bus compartment gas leakage status warning model based on the historical meteorological data, historical sensor data and equipment log feature vector of the GIS bus compartment, including: inputting the meteorological data, sensor data and equipment log feature vector of the GIS bus compartment into the pre-trained GIS bus compartment gas leakage status warning model to obtain the GIS bus compartment gas pressure prediction result; and temperature compensating the GIS bus compartment gas pressure prediction result; obtaining the GIS bus compartment gas leakage status prediction according to the temperature-compensated GIS bus compartment gas pressure prediction result and a preset GIS bus compartment gas pressure threshold; and obtaining the GIS bus compartment gas leakage status warning result according to the GIS bus compartment gas leakage status prediction.

[0052] Explanatory, gas pressure changes with temperature, usually following the ideal gas law. In the context of a GIS bus compartment, although the volume and amount of material may be relatively fixed, changes in temperature will still affect the gas pressure in the GIS bus compartment. When performing temperature compensation for the GIS bus compartment gas pressure prediction results, first obtain real-time temperature data from the temperature sensor in the GIS bus compartment; then use the temperature compensation formula or model to calculate the temperature-compensated GIS bus compartment gas pressure prediction results based on the real-time temperature data and the GIS bus compartment gas pressure prediction results.

[0053] Explanatory, according to the normal operating conditions and historical data of the GIS bus compartment, one or more GIS bus compartment gas pressure thresholds are set, and these GIS bus compartment gas pressure thresholds should be able to distinguish different situations such as the normal state, slight leakage and severe leakage of the GIS bus compartment. The temperature-compensated GIS bus compartment gas pressure prediction result is compared with the set GIS bus compartment gas pressure threshold. If the temperature-compensated GIS bus compartment gas pressure prediction result is lower than a certain GIS bus compartment gas pressure threshold, it may indicate that there is a leakage. For example, if the temperature-compensated GIS bus compartment gas pressure prediction result is lower than the GIS bus compartment gas pressure threshold of slight leakage but higher than the GIS bus compartment gas pressure threshold of severe leakage, the GIS bus compartment gas leakage state prediction is slight leakage; if the temperature-compensated GIS bus compartment gas pressure prediction result is lower than the GIS bus compartment gas pressure threshold of severe leakage, the GIS bus compartment gas leakage state prediction is severe leakage; if the temperature-compensated GIS bus compartment gas pressure prediction result is higher than the GIS bus compartment gas pressure threshold of normal state, the GIS bus compartment gas leakage state prediction is normal state.

[0054] Explanatory, according to the GIS busbar compartment leakage state prediction, the GIS busbar compartment leakage state warning result is obtained. For example, if the GIS busbar compartment leakage state prediction is normal, no warning is issued. If the GIS busbar compartment leakage state prediction is slight leakage or severe leakage, a first-level warning or a second-level warning is issued.

[0055] In a possible implementation, the GIS bus compartment gas leakage status warning result is obtained through a pre-trained GIS bus compartment gas leakage status warning model based on the historical meteorological data, historical sensor data and equipment log feature vector of the GIS bus compartment, including: based on the historical meteorological data, historical sensor data and equipment log feature vector of the GIS bus compartment, according to a preset confidence level, through a Bayesian deep learning method, a GIS bus compartment gas pressure prediction result and a confidence interval of the GIS bus compartment gas pressure prediction result are obtained; the GIS bus compartment gas pressure prediction result is temperature compensated, and the GIS bus compartment gas pressure prediction result after temperature compensation and a preset GIS bus compartment gas pressure threshold are used to obtain a GIS bus compartment gas leakage status prediction situation; and according to a preset confidence interval width threshold, in combination with the GIS bus compartment gas leakage status prediction situation, a GIS bus compartment gas leakage status warning result is obtained.

[0056] Explanation: In traditional deep learning models (such as ordinary LSTM or CNN), the output is a point estimate, which means that the model only gives a specific prediction value without providing any information about the confidence level of the prediction result. In practical applications, high-risk scenarios such as GIS busbar compartment air leakage status warning require confidence intervals for prediction results to understand the model's confidence in its own predictions.

[0057] Furthermore, the confidence interval can be used to distinguish high-risk and low-risk prediction results, reducing false positives and false negatives. Operation and maintenance personnel can allocate inspection and maintenance resources based on the predicted confidence interval. At the same time, abnormal situations can be captured more sensitively, improving the accuracy of leak prediction.

[0058] Exemplarily, in this embodiment, the confidence interval of the prediction result is provided by Bayesian deep learning (Monte Carlo Dropout). Bayesian deep learning (Monte Carlo Dropout) still maintains Dropout in the test phase, that is, multiple forward propagations are performed during prediction, and some neurons are randomly discarded each time, which is equivalent to sampling from multiple different neural network models, thereby obtaining the uncertainty of the prediction.

[0059] Among them, the prediction result set formula is:

[0060]

[0061] Among them, f(x ； ω i ) is the prediction result of the i-th Dropout sampling; is the final prediction result, i.e., the average value of multiple Dropout samplings, and M is the number of Dropout samplings. The final prediction result here refers to the prediction result of the gas pressure in the GIS busbar compartment.

[0062] The confidence interval can be obtained by calculating the standard deviation σ for multiple prediction results. 2 To get:

[0063]

[0064] Use the standard deviation and confidence level to calculate the upper and lower limits of the confidence interval. For a normal distribution, you can use the Z score to find the boundaries of the confidence interval. For example, for a 95% confidence level, the Z score is usually 1.96 (this is based on the standard normal distribution table). The lower limit of the confidence interval = the final prediction result - Z score × standard deviation; the upper limit of the confidence interval = the final prediction result + Z score × standard deviation.

[0065] The width of the confidence interval reflects the accuracy of the estimate. The wider the confidence interval, the greater the uncertainty of the estimate; the narrower the confidence interval, the smaller the uncertainty of the estimate. In practical applications, the reliability of the prediction results can be evaluated based on the width of the confidence interval, and decisions can be made accordingly. For example, based on the preset confidence interval width threshold, when the width of the confidence interval is greater than the preset confidence interval width threshold, it means that the uncertainty of the prediction result is large, and further investigation or preventive measures may be required; when the width of the confidence interval is less than the preset confidence interval width threshold, the prediction result can be considered relatively reliable, and the GIS busbar compartment leakage status warning result can be generated based on the GIS busbar compartment leakage status prediction.

[0066] By introducing the Monte Carlo Dropout uncertainty estimation into the GIS busbar compartment air leakage status warning model, the model is made more sensitive and reliable when facing equipment anomalies or sudden meteorological changes. This will greatly improve the accuracy of leakage warnings and provide more valuable decision support for operation and maintenance personnel.

[0067] The following are device embodiments of the present invention, which can be used to implement the method embodiments of the present invention. For details not disclosed in the device embodiments, please refer to the method embodiments of the present invention.

[0068] See also Figure 4 In another embodiment of the present invention, a GIS busbar compartment air leakage status warning system is provided, which can be used to implement the above-mentioned GIS busbar compartment air leakage status warning method. Specifically, the GIS busbar compartment air leakage status warning system includes a data acquisition module, a feature extraction module and a model prediction module.

[0069] Among them, the data acquisition module is used to obtain the historical meteorological data, historical sensor data and equipment log data of the GIS bus compartment; the feature extraction module is used to extract features from the equipment log data of the GIS bus compartment to obtain the equipment log feature vector; the model prediction module is used to obtain the GIS bus compartment leakage status warning result through the pre-trained GIS bus compartment leakage status warning model based on the historical meteorological data, historical sensor data and equipment log feature vector of the GIS bus compartment; among them, the GIS bus compartment leakage status warning model includes a convolutional neural network, a long short-term memory network and a Transformer network; the convolutional neural network is used to extract the features of historical meteorological data and historical sensor data, obtain the meteorological feature vector and the sensor feature vector and send them to the long short-term memory network; the long short-term memory network is used to fuse the equipment log feature vector, the meteorological feature vector and the sensor feature vector to obtain the fused feature vector; the Transformer network is used to obtain the GIS bus compartment gas pressure prediction result according to the fused feature vector.

[0070] All relevant contents of each step involved in the embodiment of the aforementioned GIS busbar compartment air leakage status warning method can be referred to the functional description of the functional module corresponding to the GIS busbar compartment air leakage status warning system in the embodiment of the present invention, and will not be repeated here.

[0071] The division of modules in the embodiments of the present invention is schematic and is only a logical function division. There may be other division methods in actual implementation. In addition, each functional module in each embodiment of the present invention may be integrated into one processor, or may exist physically separately, or two or more modules may be integrated into one module. The above-mentioned integrated modules may be implemented in the form of hardware or in the form of software functional modules.

[0072] In another embodiment of the present invention, a computer device is provided, the computer device includes a processor and a memory, the memory is used to store a computer program, the computer program includes program instructions, and the processor is used to execute the program instructions stored in the computer storage medium. The processor may be a central processing unit (CPU), or other general-purpose processors, digital signal processors (DSP), application-specific integrated circuits (ASIC), field-programmable gate arrays (FPGA) or other programmable logic devices, discrete gates or transistor logic devices, discrete hardware components, etc. It is the computing core and control core of the terminal, which is suitable for implementing one or more instructions, and is specifically suitable for loading and executing one or more instructions in the computer storage medium to implement the corresponding method flow or corresponding function; the processor described in the embodiment of the present invention can be used for the operation of the GIS busbar compartment gas leakage state warning method.

[0073] In another embodiment of the present invention, the present invention also provides a storage medium, specifically a computer-readable storage medium (Memory), which is a memory device in a computer device for storing programs and data. It is understandable that the computer-readable storage medium here can include both the built-in storage medium in the computer device and the extended storage medium supported by the computer device. The computer-readable storage medium provides a storage space, which stores the operating system of the terminal. In addition, one or more instructions suitable for being loaded and executed by the processor are also stored in the storage space, and these instructions can be one or more computer programs (including program codes). It should be noted that the computer-readable storage medium here can be a high-speed RAM memory or a non-volatile memory, such as at least one disk memory. The processor can load and execute one or more instructions stored in the computer-readable storage medium to implement the corresponding steps of the GIS busbar compartment air leakage status warning method in the above embodiment.

[0074] Those skilled in the art will appreciate that embodiments of the present invention may be provided as methods, systems, or computer program products. Therefore, the present invention may take the form of a complete hardware embodiment, a complete software embodiment, or an embodiment combining software and hardware. Moreover, the present invention may take the form of a computer program product implemented on one or more computer-usable storage media (including but not limited to disk storage, CD-ROM, optical storage, etc.) containing computer-usable program code.

[0075] The present invention is described with reference to flowcharts and / or block diagrams of methods, devices (systems), and computer program products according to embodiments of the present invention. It should be understood that each process and / or block in the flowchart and / or block diagram, as well as the combination of processes and / or blocks in the flowchart and / or block diagram, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, a special-purpose computer, an embedded processor, or other programmable data processing device to produce a machine, so that the instructions executed by the processor of the computer or other programmable data processing device generate instructions for implementing the processes in the flowchart and / or block diagram. Figure 1 A process or multiple processes and / or boxes Figure 1 A device that provides the functions specified in a block or multiple blocks.

[0076] These computer program instructions may also be stored in a computer-readable memory capable of directing a computer or other programmable data processing device to operate in a specific manner, so that the instructions stored in the computer-readable memory produce an article of manufacture comprising an instruction device, which implements the process Figure 1 A process or multiple processes and / or boxes Figure 1 A function specified in one or more boxes.

[0077] These computer program instructions can also be loaded onto a computer or other programmable data processing device so that a series of operating steps are executed on the computer or other programmable device to produce a computer-implemented process, thereby providing instructions for implementing the process. Figure 1 A process or multiple processes and / or boxes Figure 1 The steps for the functions specified in one or more boxes.

[0078] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention rather than to limit it. Although the present invention has been described in detail with reference to the above embodiments, ordinary technicians in the relevant field should understand that the specific implementation methods of the present invention can still be modified or replaced by equivalents. Any modification or equivalent replacement that does not depart from the spirit and scope of the present invention should be covered within the scope of protection of the claims of the present invention.

Claims

1. A GIS busbar compartment air leakage status early warning method, characterized in that: include: Obtain historical meteorological data, historical sensor data and equipment log data from the GIS busbar compartment; Extract the features of the equipment log data in the GIS busbar compartment to obtain the equipment log feature quantity; Based on the historical meteorological data, historical sensor data and equipment log feature vector of the GIS busbar compartment, the GIS busbar compartment air leakage status warning result is obtained through the pre-trained GIS busbar compartment air leakage status warning model; Among them, the GIS busbar compartment gas leakage status warning model includes convolutional neural network, long short-term memory network and Transformer network; the convolutional neural network is used to extract the characteristics of historical meteorological data and historical sensor data, obtain meteorological feature vectors and sensor feature vectors and send them to the long short-term memory network; the long short-term memory network is used to fuse the equipment log feature vector, meteorological feature vector and sensor feature vector to obtain the fused feature vector; the Transformer network is used to obtain the GIS busbar compartment gas pressure prediction result according to the fused feature vector.

2. The GIS busbar compartment air leakage state early warning method according to claim 1 is characterized in that: The historical meteorological data includes one or more of the following: light intensity, temperature, air pressure, humidity and precipitation.

3. The GIS busbar compartment air leakage state early warning method according to claim 1 is characterized in that: The historical sensor data includes one or more of the following: gas pressure, gas temperature, humidity level, gas pressure change rate, gas temperature change rate, pressure difference between inside and outside of the GIS busbar compartment, and gas frequency domain data.

4. The GIS busbar compartment air leakage state early warning method according to claim 1 is characterized in that: The feature extraction of the equipment log data of the GIS busbar compartment to obtain the equipment log feature vector includes: Preprocess the equipment log data of the GIS busbar compartment; The pre-trained BERT model is used to extract features from the equipment log data of the GIS bus compartment after data preprocessing to obtain the equipment log feature vector.

5. The GIS busbar compartment air leakage state early warning method according to claim 1 is characterized in that: The method also includes: preprocessing the historical meteorological data and historical sensor data of the GIS bus compartment; aligning the historical meteorological data, historical sensor data and equipment log feature vectors of the GIS bus compartment after data preprocessing at an hourly frequency according to a timestamp; wherein the meteorological data and sensor data are converted into an hourly frequency by an average method; and the equipment log feature vectors are converted into an hourly frequency by a repetition method.

6. The GIS busbar compartment air leakage state early warning method according to claim 1 is characterized in that: The GIS busbar compartment gas leakage status warning result obtained by the pre-trained GIS busbar compartment gas leakage status warning model based on the historical meteorological data, historical sensor data and equipment log feature vector of the GIS busbar compartment includes: The meteorological data, sensor data and equipment log feature vector of the GIS bus compartment are input into the pre-trained GIS bus compartment gas leakage status warning model to obtain the GIS bus compartment gas pressure prediction result; and the GIS bus compartment gas pressure prediction result is temperature compensated; According to the temperature-compensated GIS busbar compartment gas pressure prediction result and the preset GIS busbar compartment gas pressure threshold, the GIS busbar compartment gas leakage state prediction is obtained; According to the prediction of the air leakage status of the GIS busbar compartment, the early warning result of the air leakage status of the GIS busbar compartment is obtained.

7. The GIS busbar compartment air leakage state early warning method according to claim 1 is characterized in that: The GIS busbar compartment gas leakage status warning result obtained by the pre-trained GIS busbar compartment gas leakage status warning model based on the historical meteorological data, historical sensor data and equipment log feature vector of the GIS busbar compartment includes: Based on the historical meteorological data, historical sensor data and equipment log feature vectors of the GIS busbar compartment, and according to the preset confidence level, the GIS busbar compartment gas pressure prediction results and the confidence interval of the GIS busbar compartment gas pressure prediction results are obtained through the Bayesian deep learning method; The GIS bus compartment gas pressure prediction result is temperature compensated, and the GIS bus compartment gas pressure prediction result after temperature compensation and the preset GIS bus compartment gas pressure threshold are used to obtain the GIS bus compartment gas leakage status prediction situation; and according to the preset confidence interval width threshold, combined with the GIS bus compartment gas leakage status prediction situation, the GIS bus compartment gas leakage status warning result is obtained.

8. A GIS busbar compartment air leakage status early warning system, characterized in that: include: Data acquisition module, used to obtain historical meteorological data, historical sensor data and equipment log data of the GIS busbar compartment; A feature extraction module is used to extract features from the equipment log data of the GIS busbar compartment to obtain a feature vector of the equipment log; The model prediction module is used to obtain the GIS busbar compartment gas leakage status warning result through the pre-trained GIS busbar compartment gas leakage status warning model based on the historical meteorological data, historical sensor data and equipment log feature vector of the GIS busbar compartment; Among them, the GIS busbar compartment gas leakage status warning model includes convolutional neural network, long short-term memory network and Transformer network; the convolutional neural network is used to extract the characteristics of historical meteorological data and historical sensor data, obtain meteorological feature vectors and sensor feature vectors and send them to the long short-term memory network; the long short-term memory network is used to fuse the equipment log feature vector, meteorological feature vector and sensor feature vector to obtain the fused feature vector; the Transformer network is used to obtain the GIS busbar compartment gas pressure prediction result according to the fused feature vector.

9. A computer device comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, characterized in that: When the processor executes the computer program, the steps of the GIS busbar compartment air leakage status early warning method as described in any one of claims 1 to 7 are implemented.

10. A computer-readable storage medium storing a computer program, characterized in that: When the computer program is executed by a processor, the steps of the GIS busbar compartment air leakage status early warning method as described in any one of claims 1 to 7 are implemented.

Images