Gas particulate matter adjusting method, system, medium, equipment and product for sulfur hexafluoride gas insulated switchgear

Abstract

The invention discloses a method, a system, a medium, equipment and a product for adjusting gas particulate matters of a sulfur hexafluoride gas insulated switchgear, and belongs to the technical field of adjustment of gas particulate matters of sulfur hexafluoride gas insulated switchgears. The method comprises the following steps: carrying out data preprocessing on obtained gas characteristic data of the sulfur hexafluoride gas insulated switchgear in a fixed time window to obtain preprocessed gas characteristic data, carrying out particulate matter concentration prediction by utilizing a trained gas characteristic prediction model to obtain predicted particulate matter concentration, and calculating a prediction slope; according to the obtained real-time gas flow rate and the real-time pressure of the sulfur hexafluoride gas insulated switchgear, a GPR online learning mechanism is utilized to obtain a self-adaptive particulate matter accumulation threshold value; and if the predicted slope is greater than or equal to a slope threshold value or the predicted particulate matter concentration is greater than or equal to a self-adaptive particulate matter accumulation threshold value, adjusting the gas flow rate of the sulfur hexafluoride gas insulated switchgear by using a GPR-based Bayesian optimization algorithm.

Description

Technical Field

[0001] This invention relates to the field of particulate matter adjustment technology for sulfur hexafluoride gas-insulated switchgear, and particularly to a method, system, medium, equipment, and product for adjusting particulate matter in sulfur hexafluoride gas-insulated switchgear. Background Technology

[0002] Gas-insulated switchgear (GIS) has become an indispensable core component in modern power systems due to its excellent insulation performance and compact design. However, particulate matter mixed in with sulfur hexafluoride gas can significantly reduce the dielectric insulation strength, increase the risk of insulation failure, and pose a potential threat to the safe and stable operation of the power grid.

[0003] The causes of particulate contamination in gas-insulated switchgear are multidimensional: during manufacturing, residual metal debris may remain inside the equipment; during assembly, operators may accidentally introduce hair or clothing fibers; during transportation, friction between equipment components can generate metal debris; and during operation, the friction of moving mechanical parts continuously generates metal particles. Furthermore, low-fluoride solid deposits produced by arc decomposition are also a significant source of particulate matter affecting equipment operation. Excessive accumulation of particulate matter not only leads to surface wear and corrosion but also shortens service life, increases maintenance costs, and even poses safety hazards.

[0004] Existing gas detection devices employ a pipeline design for their gas path systems, lacking the ability to filter particles in dusty airflows and requiring dedicated filtration devices. However, under dusty conditions, traditional filtration solutions suffer from drawbacks such as excessive pressure drop and non-reusability, easily leading to particle interference with the detection equipment. Current domestic research largely focuses on the source apportionment and pollution characteristic analysis of sulfur hexafluoride gas particles, or is limited to single-point parameter monitoring, exhibiting the limitation of substituting local data for overall assessment. Research on particle control technologies and comprehensive parameter evaluation systems is relatively weak. Summary of the Invention

[0005] The purpose of this invention is to overcome the shortcomings of the prior art and provide a method, system, medium, equipment and product for adjusting gas particulate matter in sulfur hexafluoride gas-insulated switchgear. By comprehensively considering multiple key structural indices of the circuit breaker, the operating status of the circuit breaker can be comprehensively evaluated, and the operating condition of the circuit breaker can be judged more accurately. This avoids misjudgment due to the limitations of a single indicator, ensures that the circuit breaker is in good working condition, and thus improves the power supply reliability of the power system.

[0006] To solve the above-mentioned technical problems, the present invention is implemented using the following technical solution:

[0007] In a first aspect, the present invention provides a method for adjusting particulate matter in a sulfur hexafluoride gas-insulated switchgear, comprising:

[0008] The gas characteristic data of sulfur hexafluoride gas-insulated switchgear obtained within a fixed time window are preprocessed to obtain preprocessed gas characteristic data.

[0009] Based on the preprocessed gas characteristic data, the trained gas characteristic prediction model is used to predict particulate matter concentration, obtain the predicted particulate matter concentration and calculate the prediction slope.

[0010] Based on the real-time gas flow rate and real-time pressure of the sulfur hexafluoride gas-insulated switchgear, an adaptive particulate matter accumulation threshold is obtained using the GPR online learning mechanism.

[0011] If the predicted slope is greater than or equal to the slope threshold, or the predicted particulate matter concentration is greater than or equal to the adaptive particulate matter accumulation threshold, the gas flow rate of the sulfur hexafluoride gas-insulated switchgear is adjusted using a GPR-based Bayesian optimization algorithm.

[0012] Optionally, the gas characteristic data includes gas flow rate, pressure, and particulate matter concentration;

[0013] The data preprocessing of the gas characteristic data of the sulfur hexafluoride gas-insulated switchgear obtained within a fixed time window includes:

[0014] Noise is removed from the gas feature data to obtain denoised gas feature data;

[0015] The denoised gas feature data is normalized to obtain preprocessed gas feature data.

[0016] Optionally, the gas feature prediction model includes an LSTM layer, a temporal attention mechanism layer, and an output layer;

[0017] The expression for the LSTM layer is as follows:

[0018] ,

[0019] in, express Forget Gate Output The weight matrix represents the forget gate. express The state vector is hidden at all times. express Preprocessed gas characteristic data at specific times The bias vector representing the forget gate; express Input gate output at all times; This represents the weight matrix of the input gate. This represents the bias vector of the input gate; express Candidate cell state at any given time. and This represents the activation function. The weight matrix representing the state of the candidate unit. A bias vector representing the state of a candidate cell; express The cell status is updated in real time. express The cell status is updated in real time. express Candidate cell status at any given time; express Output gate output at all times, This represents the weight matrix of the output gate. This represents the bias vector of the output gate. express The dynamic feature vector at any given time;

[0020] The expression for the temporal attention mechanism layer is as follows:

[0021] ,

[0022] in, express The feature vector of particulate matter concentration at time step, Represents the attention context vector. express transpose, Represents the weight matrix. This represents the bias vector. This represents the attention-weighted feature vector. Indicates the start time. Indicates the end time. express The feature vector of particulate matter concentration at time step, Represents the natural exponential function;

[0023] The expression for the output layer is as follows:

[0024] ,

[0025] in, Indicates the target time Predicted particulate matter concentration;

[0026] The predicted slope is obtained using the following formula:

[0027] ,

[0028] in, Indicates the predicted slope. express The measured particulate matter concentration at any given time.

[0029] Optionally, the GPR online learning mechanism employs a Gaussian process regression model, and the kernel function of the Gaussian process regression model uses the following covariance function:

[0030] ,

[0031] in, Represents the covariance function. and Indicates the sample point number. , , This represents the total number of sample points. This represents the variance of all sample points. This represents the natural exponential function. Represents the set of observation data The variance of independent and identically distributed noise in the data. Indicates sample gas flow rate, Indicates sample Pressure Indicates sample gas flow rate, Indicates sample Pressure Represents the characteristic scale of gas flow velocity. Indicates the pressure characteristic scale, This represents the Kronecker delta function. hour, ,otherwise ;

[0032] The hyperparameters of the kernel function The following log-marginal likelihood function is used to obtain:

[0033] ,

[0034] in, , Indicates sample Observational data, express transpose, Represents the set of sample points. , Represents the covariance matrix. express The inverse matrix, , Represents the covariance matrix The Middle Line 1 The data in the column.

[0035] Optionally, the step of obtaining an adaptive particulate matter accumulation threshold using a GPR online learning mechanism based on the real-time gas flow rate and real-time pressure of the sulfur hexafluoride gas-insulated switchgear includes:

[0036] The covariance matrix is ​​obtained by calculating the covariance between the real-time gas flow rate and real-time pressure and each sample point. ;

[0037] According to the covariance matrix Calculate the adaptive particulate matter accumulation threshold;

[0038] The covariance of the real-time gas flow rate and real-time pressure with respect to each sample point is obtained by the following formula:

[0039] ,

[0040] in, Indicates real-time gas flow rate. Indicates real-time pressure. express and covariance;

[0041] The adaptive particulate matter accumulation threshold is obtained by the following formula:

[0042] ,

[0043] in, Indicates the adaptive particulate matter accumulation threshold. express The transpose of .

[0044] Optionally, the slope threshold is obtained by the following formula:

[0045] ,

[0046] in, Indicates the slope threshold. It represents the limiting rate of change in the concentration of gaseous particulate matter per unit time. This represents the exponential decay characteristic of the rate of change of gas particulate matter concentration during the extraction process. Indicates the target time. Represents a constant;

[0047] The GPR surrogate model in the GPR-based Bayesian optimization algorithm follows the following normal distribution:

[0048] ,

[0049] in, Indicates the gas flow rate as The corresponding particulate matter concentration at that time This represents the sample mean of the GPR proxy model. This represents the sample variance of the GPR proxy model;

[0050] The expected improvement function of the GPR-based Bayesian optimization algorithm is as follows:

[0051]

[0052] in, Indicates the gas flow rate as The corresponding expected improvement value, This represents the normal distribution value of the GPR proxy model. , Indicates the adaptive particulate matter accumulation threshold. Indicates real-time gas flow rate. Indicates real-time pressure. This represents the sample standard deviation of the GPR surrogate model. The cumulative distribution function represents the standard normal distribution. The probability density function representing the standard normal distribution;

[0053] Based on the desired improvement function, the required gas flow rate is obtained. To maximize the desired improvement value, and adjust the gas flow rate as needed. Adjust the gas flow rate of the sulfur hexafluoride gas-insulated switchgear; among which, .

[0054] In a second aspect, the present invention provides a particulate matter adjustment system for sulfur hexafluoride gas-insulated switchgear, used to implement the particulate matter adjustment method for sulfur hexafluoride gas-insulated switchgear as described in any one of the first aspects, comprising:

[0055] The preprocessing module is used to: preprocess the gas characteristic data of the sulfur hexafluoride gas-insulated switchgear obtained within a fixed time window to obtain preprocessed gas characteristic data;

[0056] The particulate matter concentration prediction module is used to: predict particulate matter concentration based on the preprocessed gas feature data and a trained gas feature prediction model, obtain the predicted particulate matter concentration and calculate the prediction slope.

[0057] The GPR online learning module is used to: obtain an adaptive particulate matter accumulation threshold based on the real-time gas flow rate and real-time pressure of the sulfur hexafluoride gas-insulated switchgear.

[0058] The Bayesian optimization module is used to: adjust the gas flow rate of the sulfur hexafluoride gas-insulated switchgear using a GPR-based Bayesian optimization algorithm if the predicted slope is greater than or equal to a slope threshold, or the predicted particulate matter concentration is greater than or equal to an adaptive particulate matter accumulation threshold.

[0059] Thirdly, the present invention provides a computer-readable storage medium having computer instructions stored thereon, which, when executed by a processor, implement the steps of the method for adjusting particulate matter in a sulfur hexafluoride gas-insulated switchgear as described in any of the first aspects.

[0060] Fourthly, the present invention provides a computer device, comprising:

[0061] Memory, used to store computer instructions;

[0062] A processor for executing the computer instructions to implement the steps of the method for adjusting particulate matter in a sulfur hexafluoride gas-insulated switchgear as described in any of the first aspects.

[0063] Fifthly, the present invention provides a computer program product, including computer instructions, characterized in that, when executed by a processor, the computer instructions implement the steps of the method for adjusting particulate matter in a sulfur hexafluoride gas-insulated switchgear as described in any of the first aspects.

[0064] Compared with the prior art, the beneficial effects achieved by the present invention are as follows:

[0065] 1. The particulate matter adjustment method for sulfur hexafluoride gas-insulated switchgear provided by this invention establishes a flow-velocity-pressure adaptive dynamic threshold function for particulate matter accumulation by combining Gaussian process regression (GPR), predicts the particulate matter accumulation trend by combining an LSTM network with a temporal attention mechanism, and uses Bayesian optimization to adjust the charging / pumping speed in real time. This enables accurate prediction and intelligent closed-loop control of particulate matter concentration under complex operating conditions. It solves the problems of low detection sensitivity and high false alarm rate caused by the use of fixed thresholds in the particulate matter adjustment of existing sulfur hexafluoride gas-insulated switchgear, which cannot adapt to dynamic changes in flow rate and pressure, as well as the problems of traditional single-point monitoring failing to fully reflect the particulate matter distribution trend and lagging control response. This invention achieves real-time, accurate prediction and adaptive control of particulate matter concentration under complex operating conditions.

[0066] 2. The particulate matter adjustment system for sulfur hexafluoride gas-insulated switchgear provided by the present invention, through the setting of a pre-processing module, a particulate matter concentration prediction module, a GPR online learning module and a Bayesian optimization module, jointly realizes the adjustment of particulate matter in sulfur hexafluoride gas-insulated switchgear, which has practical significance and good application prospects.

[0067] 3. The computer media, equipment, and products provided by this invention can execute the steps of the sulfur hexafluoride gas-insulated combined electrical appliance gas particulate matter adjustment method provided by this invention. Attached Figure Description

[0068] Figure 1 This is a flowchart of a method for adjusting particulate matter in a sulfur hexafluoride gas-insulated switchgear according to an embodiment of the present invention.

[0069] Figure 2 This is a structural diagram of a particulate matter adjustment system for a sulfur hexafluoride gas-insulated combined electrical appliance according to an embodiment of the present invention;

[0070] Figure 3 This is a data acquisition flowchart provided according to an embodiment of the present invention;

[0071] Figure 4 This is a structural diagram of a gas feature prediction model provided according to an embodiment of the present invention;

[0072] Figure 5 This is a diagram of an LSTM layer structure provided according to an embodiment of the present invention. Detailed Implementation

[0073] The technical solution of the present invention will be described in detail below with reference to the accompanying drawings and specific embodiments. It should be understood that the embodiments and specific features in the embodiments are detailed descriptions of the technical solution of the present application, rather than limitations thereof. In the absence of conflict, the embodiments and technical features in the embodiments can be combined with each other.

[0074] It should be noted that the term "and / or" in this article is merely a description of the relationship between related objects, indicating that three relationships can exist. For example, A and / or B can represent: A existing alone, A and B existing simultaneously, or B existing alone. Additionally, the character " / " in this article generally indicates that the preceding and following related objects have an "or" relationship.

[0075] Example 1:

[0076] This invention discloses a method for adjusting particulate matter in sulfur hexafluoride gas-insulated switchgear, with reference to... Figure 1 As shown, the specific steps include the following:

[0077] S1, perform data preprocessing on the gas characteristic data of the sulfur hexafluoride gas-insulated switchgear obtained within a fixed time window to obtain preprocessed gas characteristic data;

[0078] S2, Based on the preprocessed gas characteristic data, use the trained gas characteristic prediction model to predict the particulate matter concentration, obtain the predicted particulate matter concentration and calculate the prediction slope.

[0079] S3. Based on the real-time gas flow rate and real-time pressure of the sulfur hexafluoride gas-insulated switchgear, the adaptive particulate matter accumulation threshold is obtained using the GPR online learning mechanism.

[0080] S4. If the predicted slope is greater than or equal to the slope threshold, or the predicted particulate matter concentration is greater than or equal to the adaptive particulate matter accumulation threshold, the gas flow rate of the sulfur hexafluoride gas-insulated switchgear is adjusted using a GPR-based Bayesian optimization algorithm.

[0081] In step S1, refer to Figure 3 As shown, this embodiment acquires gas characteristic data, including gas flow rate, pressure, and particulate matter concentration, through a data acquisition unit; specifically, a flow meter measures gas flow rate, a pressure sensor measures gas pressure, and a particulate matter sensor measures particulate matter content.

[0082] The data preprocessing of the gas characteristic data of the sulfur hexafluoride gas-insulated switchgear obtained within a fixed time window includes:

[0083] Noise is removed from the gas feature data to obtain denoised gas feature data;

[0084] The denoised gas feature data is normalized to obtain preprocessed gas feature data.

[0085] In step S2, refer to Figure 4 As shown, the gas feature prediction model includes an LSTM layer, a temporal attention mechanism layer, and an output layer;

[0086] refer to Figure 5 As shown, the expression for the LSTM layer is as follows:

[0087] ,

[0088] in, express Forget Gate Output The weight matrix represents the forget gate. express The state vector is hidden at all times. express Preprocessed gas characteristic data at specific times The bias vector representing the forget gate; express Input gate output at all times; This represents the weight matrix of the input gate. This represents the bias vector of the input gate; express Candidate cell state at any given time. and This represents the activation function. The weight matrix representing the state of the candidate unit. A bias vector representing the state of a candidate cell; express The cell status is updated in real time. express The cell status is updated in real time. express Candidate cell status at any given time; express Output gate output at all times, This represents the weight matrix of the output gate. This represents the bias vector of the output gate. express The dynamic feature vector at time step.

[0089] The temporal attention mechanism layer employs a temporal pattern attention mechanism. Based on the dynamic importance of the particulate matter concentration change trend at each time step, it assigns attention weights to corresponding time steps and then weights and fuses the particulate matter concentration values ​​using weight coefficients to ultimately generate an attention-weighted feature vector representing temporal correlation features. The expression for the temporal attention mechanism layer is as follows:

[0090] ,

[0091] in, express The feature vector of particulate matter concentration at time step, Represents the attention context vector. express transpose, Represents the weight matrix. This represents the bias vector. express The dynamic feature vector at time step, This represents the attention-weighted feature vector. Indicates the start time. Indicates the end time. express The feature vector of particulate matter concentration at time step, This represents the natural exponential function.

[0092] The expression for the output layer is as follows:

[0093] ,

[0094] in, Indicates the target time The predicted particulate matter concentration.

[0095] The predicted slope is obtained using the following formula:

[0096] ,

[0097] in, Indicates the predicted slope. express The measured particulate matter concentration at any given time.

[0098] The training process of the gas feature prediction model includes:

[0099] Constructing a training set: Acquire historical gas characteristic data during the gas operation of sulfur hexafluoride electrical equipment, covering particulate matter concentration, internal gas pressure, and gas flow parameters; preprocess the historical gas characteristic data, including dividing it into discrete time periods according to fixed time windows, removing sensor noise, and normalizing multi-dimensional parameters to eliminate dimensional differences; then, spatiotemporally align the preprocessed historical gas characteristic time series segments with the actual particulate matter concentration measurements of the corresponding future time periods to construct a training set containing input features (gas characteristic time series of the current time period) and target labels (actual particulate matter concentration values ​​of the future time period);

[0100] The constructed gas feature prediction model is trained using the training set to obtain the trained gas feature prediction model. The training process includes two core elements: one is the attention weight matrix dynamically allocated based on the gas feature change trend at each historical moment, and the other is the feature data sequence of particulate matter concentration in the future time period.

[0101] In other embodiments, information flow can be managed through the update gate and reset gate in a gated recurrent unit (GRU) or a bidirectional gated recurrent unit (BiGRU), i.e., using GRU or BiGRU to replace the LSTM layer. Replacing LSTM with BiGRU can improve computational efficiency, but its long-term memory and fine-grained control capabilities are slightly weaker. The key improvement measure is to retain and strengthen the temporal attention mechanism to compensate for the lack of long-range dependency processing. At the same time, the expressive power of the model can be enhanced by increasing the network depth and fine-tuning the parameters to ensure performance in real-time edge computing scenarios.

[0102] In other embodiments, convolutional neural networks (CNNs) and Transformers can be used instead of the temporal attention mechanism layer. While the Transformer model utilizes self-attention to achieve parallel computation, significantly improving training efficiency and effectively capturing global dependencies in temporal data, it has limitations in local feature extraction. Therefore, a hybrid architecture combining CNNs and Transformers can be introduced to balance both global and local feature representation capabilities.

[0103] In step S3, the adaptive particulate matter accumulation threshold is adaptively adjusted by continuously integrating historical particulate matter concentration data under different operating conditions through an online learning mechanism.

[0104] The GPR online learning mechanism employs a Gaussian process regression model, and the kernel function of the Gaussian process regression model uses the following covariance function:

[0105] ,

[0106] in, Represents the covariance function. and Indicates the sample point number. , , This represents the total number of sample points. This represents the variance of all sample points. This represents the natural exponential function. Represents the set of observation data The variance of independent and identically distributed noise in the data. Indicates sample gas flow rate, Indicates sample Pressure Indicates sample gas flow rate, Indicates sample Pressure Represents the characteristic scale of gas flow velocity. Indicates the pressure characteristic scale, This represents the Kronecker delta function. hour, ,otherwise ;

[0107] The hyperparameters of the kernel function The following log-marginal likelihood function is used to obtain:

[0108] ,

[0109] in, , Indicates sample Observational data, express transpose, Represents the set of sample points. , Represents the covariance matrix. express The inverse matrix, , Represents the covariance matrix The Middle Line 1 The data in the column.

[0110] The process of obtaining an adaptive particulate matter accumulation threshold based on the real-time gas flow rate and pressure of the sulfur hexafluoride gas-insulated switchgear, using a GPR online learning mechanism, includes:

[0111] The covariance matrix is ​​obtained by calculating the covariance between the real-time gas flow rate and real-time pressure and each sample point. ;

[0112] According to the covariance matrix Calculate the adaptive particulate matter accumulation threshold;

[0113] The covariance of the real-time gas flow rate and real-time pressure with respect to each sample point is obtained by the following formula:

[0114] ,

[0115] in, Indicates real-time gas flow rate. Indicates real-time pressure. express and covariance;

[0116] The adaptive particulate matter accumulation threshold is obtained by the following formula:

[0117] ,

[0118] in, Indicates the adaptive particulate matter accumulation threshold. express The transpose of .

[0119] In step S4, the slope threshold is obtained by the following formula:

[0120] ,

[0121] in, Indicates the slope threshold. It represents the limiting rate of change in the concentration of gaseous particulate matter per unit time. This represents the exponential decay characteristic of the rate of change of gas particulate matter concentration during the extraction process. Indicates the target time. Represents a constant.

[0122] In this embodiment, when the predicted slope is greater than or equal to the slope threshold, or the predicted particulate matter concentration is greater than or equal to the adaptive particulate matter accumulation threshold, the conditions of the Bayesian optimization algorithm are triggered, the gas flow rate is updated, the valve is activated, and the audible and visual alarm device is activated. After the gas flow rate stabilizes, the activation state of the audible and visual alarm device is deactivated. When the Bayesian optimization algorithm cannot find a satisfactory gas flow rate, the valve is closed, the pumping / filling operation is stopped, and the audible and visual alarm device enters a continuous alarm state, which is manually deactivated by maintenance personnel after inspecting the equipment.

[0123] The GPR surrogate model in the GPR-based Bayesian optimization algorithm follows the following normal distribution:

[0124] ,

[0125] in, Indicates the gas flow rate as The corresponding particulate matter concentration at that time This represents the sample mean of the GPR proxy model. This represents the sample variance of the GPR proxy model;

[0126] The expected improvement function of the GPR-based Bayesian optimization algorithm is as follows:

[0127]

[0128] in, Indicates the gas flow rate as The corresponding expected improvement value, This represents the normal distribution value of the GPR proxy model. , Indicates the adaptive particulate matter accumulation threshold. Indicates real-time gas flow rate. Indicates real-time pressure. This represents the sample standard deviation of the GPR surrogate model. The cumulative distribution function represents the standard normal distribution. The probability density function representing the standard normal distribution;

[0129] Based on the desired improvement function, the required gas flow rate is obtained. To maximize the desired improvement value, and adjust the gas flow rate as needed. Adjust the gas flow rate of the sulfur hexafluoride gas-insulated switchgear; among which, .

[0130] In other embodiments, PID control can replace Bayesian optimization. PID control maintains system stability through comprehensive adjustment of proportional, integral, and derivative parameters, offering advantages such as simple structure and high reliability. However, it has poor adaptability in nonlinear and time-varying systems. To address this issue, fuzzy PID or neural network-based adaptive PID strategies can be employed to enhance control performance.

[0131] In summary, the method for adjusting particulate matter in a sulfur hexafluoride gas-insulated switchgear proposed in this embodiment has the following beneficial effects:

[0132] 1. Achieved dynamic and accurate prediction and early warning of particulate matter concentration: By introducing an LSTM network with a temporal attention mechanism, the model can focus on the key periods of accelerated accumulation of particulate matter in historical data, which significantly improves the prediction accuracy and interpretability of particulate matter concentration change trends in the near future, and overcomes the limitations of traditional single-point monitoring that is biased and incomplete.

[0133] 2. An adaptive dynamic detection threshold for changing operating conditions was constructed: A nonlinear mapping relationship between multiple parameters such as flow velocity and pressure and the particulate matter accumulation threshold was established using a Gaussian process regression (GPR) model. This threshold can learn and adaptively adjust online according to real-time operating conditions (such as sudden changes in flow velocity and pressure fluctuations), solving the problem of insufficient sensitivity or high false alarm rate of fixed threshold schemes under varying operating conditions.

[0134] 3. Intelligent closed-loop control is formed, significantly improving response speed and control accuracy: A Bayesian optimization algorithm is adopted, using the predicted particulate matter change trend as the target, to dynamically and in real-time solve and adjust the optimal charging / evacuation speed parameters. This optimized control strategy overcomes the response lag problem of traditional control methods, and can significantly shorten the leakage response time and improve the control accuracy of particulate matter accumulation while ensuring stable equipment pressure.

[0135] 4. An integrated closed-loop system of monitoring, diagnosis and control was constructed to improve the level of intelligence: The dynamic threshold setting of GPR, the trend prediction of LSTM network with time-series attention mechanism and the optimization of control parameters by Bayesian method were deeply integrated to form a complete intelligent closed-loop management system; This system realizes full-process automation from state perception and intelligent diagnosis to automatic control, which significantly improves the level of intelligent management of sulfur hexafluoride gas operations.

[0136] 5. Enhanced reliability and safety of power equipment: Integrating audible and visual alarm modules and valves, when abnormal particulate matter accumulation or predicted values ​​exceed dynamic thresholds are detected, an alarm can be triggered immediately and dangerous operations can be automatically terminated, effectively ensuring the safe and stable operation of power equipment and reducing the risk of insulation failure.

[0137] Example 2:

[0138] Based on the same inventive concept as Embodiment 1, this embodiment of the invention discloses a particulate matter adjustment system for a sulfur hexafluoride gas-insulated combined electrical appliance, with reference to... Figure 2 As shown, it includes:

[0139] The preprocessing module is used to: preprocess the gas characteristic data of the sulfur hexafluoride gas-insulated switchgear obtained within a fixed time window to obtain preprocessed gas characteristic data;

[0140] The particulate matter concentration prediction module is used to: predict particulate matter concentration based on the preprocessed gas feature data and a trained gas feature prediction model, obtain the predicted particulate matter concentration and calculate the prediction slope.

[0141] The GPR online learning module is used to: obtain an adaptive particulate matter accumulation threshold based on the real-time gas flow rate and real-time pressure of the sulfur hexafluoride gas-insulated switchgear.

[0142] The Bayesian optimization module is used to: adjust the gas flow rate of the sulfur hexafluoride gas-insulated switchgear using a GPR-based Bayesian optimization algorithm if the predicted slope is greater than or equal to a slope threshold, or the predicted particulate matter concentration is greater than or equal to an adaptive particulate matter accumulation threshold.

[0143] The specific functions of each module described above are explained in the relevant content of the method in Embodiment 1, and will not be repeated here.

[0144] Example 3:

[0145] This embodiment provides a computer-readable storage medium storing computer instructions that, when executed by a processor, implement the steps of the sulfur hexafluoride gas-insulated switchgear gas particulate matter adjustment method as described in any of the embodiments in Example 1.

[0146] Example 4:

[0147] This embodiment provides a computer device, including:

[0148] Memory, used to store computer instructions;

[0149] A processor is configured to execute the computer instructions to implement the steps of the method for adjusting particulate matter in a sulfur hexafluoride gas-insulated switchgear as described in any one of Embodiment 1.

[0150] Example 5:

[0151] This embodiment provides a computer program product, including computer instructions, characterized in that, when executed by a processor, the computer instructions implement the steps of the sulfur hexafluoride gas-insulated combined electrical appliance gas particulate matter adjustment method as described in any one of Embodiment 1.

[0152] Those skilled in the art will understand that embodiments of the present invention can be provided as methods, systems, or computer program products. Therefore, the present invention can take the form of a completely hardware embodiment, a completely software embodiment, or an embodiment combining software and hardware aspects. Furthermore, the present invention can take the form of a computer program product embodied 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.

[0153] This invention is described with reference to flowchart illustrations and / or block diagrams of methods, apparatus (systems), and computer program products according to embodiments of the invention. It will be understood that each block of the flowchart illustrations and / or block diagrams, and combinations of blocks in the flowchart illustrations and / or block diagrams, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, special-purpose computer, embedded processor, or other programmable data processing apparatus to produce a machine, such that the instructions, which execute via the processor of the computer or other programmable data processing apparatus, generate instructions for implementing the flowchart illustrations and / or block diagrams. Figure 1 One or more processes and / or boxes Figure 1 A device that provides the functions specified in one or more boxes.

[0154] These computer program instructions may also be stored in a computer-readable storage medium that can direct a computer or other programmable data processing device to function in a particular manner, such that the instructions stored in the computer-readable storage medium produce an article of manufacture including instruction means, which are implemented in a process Figure 1 One or more processes and / or boxes Figure 1 The function specified in one or more boxes.

[0155] These computer program instructions may also be loaded onto a computer or other programmable data processing equipment to cause a series of operational steps to be performed on the computer or other programmable equipment to produce a computer-implemented process, thereby providing instructions that execute on the computer or other programmable equipment for implementing the process. Figure 1 One or more processes and / or boxes Figure 1The steps of the function specified in one or more boxes.

[0156] The embodiments of the present invention have been described above with reference to the accompanying drawings. However, the present invention is not limited to the specific embodiments described above. The specific embodiments described above are merely illustrative and not restrictive. Those skilled in the art can make many other forms under the guidance of the present invention without departing from the spirit and scope of the claims. All of these forms are within the protection scope of the present invention.

Claims

1. A method of adjusting gas particles of a sulfur hexafluoride gas-insulated combined electrical apparatus, characterized by, The method comprises the following steps: data preprocessing is performed on the acquired gas characteristic data of the sulfur hexafluoride gas insulated combined electric appliance in a fixed time window to obtain preprocessed gas characteristic data; particle concentration prediction is performed on the preprocessed gas characteristic data by using a trained gas characteristic prediction model to obtain predicted particle concentration and calculate a prediction slope; an adaptive particle accumulation threshold is obtained by using a GPR online learning mechanism according to the acquired real-time gas flow rate and real-time pressure of the sulfur hexafluoride gas insulated combined electric appliance; if the prediction slope is greater than or equal to a slope threshold or the predicted particle concentration is greater than or equal to the adaptive particle accumulation threshold, the gas flow rate of the sulfur hexafluoride gas insulated combined electric appliance is adjusted by using a GPR-based Bayesian optimization algorithm.

2. The gas particle conditioning method of a SF6 gas-insulated combined electrical apparatus according to claim 1, characterized by, The gas characteristic data comprises a gas flow rate, a pressure and a particle concentration. The data preprocessing performed on the acquired gas characteristic data of the sulfur hexafluoride gas insulated combined electric appliance in a fixed time window comprises the following steps: noise in the gas characteristic data is removed to obtain denoised gas characteristic data; the denoised gas characteristic data is normalized to obtain preprocessed gas characteristic data.

3. The method of claim 1, wherein the gas particle adjustment method is a gas particle adjustment method of a sulfur hexafluoride gas-insulated combined electrical apparatus, and The gas characteristic prediction model comprises an LSTM layer, a time sequence attention mechanism layer and an output layer. The expression of the LSTM layer is as follows: ， wherein, denotes the output of the forget gate at time step t, denotes a weight matrix of the forget gate, denotes the hidden state vector at time step t, denotes the pre-processed gas feature data at time step t, denotes a bias vector of the forget gate; denotes the output of the input gate at time step t; denotes a weight matrix of the input gate, denotes a bias vector of the input gate; denotes the candidate cell state at time step t, and denotes an activation function, denotes a weight matrix of the candidate cell state, denotes a bias vector of the candidate cell state; denotes the updated cell state at time step t, denotes the updated cell state at time step t, denotes the candidate cell state at time step t; denotes the output of the output gate at time step t, denotes a weight matrix of the output gate, denotes a bias vector of the output gate, denotes the dynamic feature vector at time step t; The expression of the time sequence attention mechanism layer is as follows: ， wherein, denotes the particulate matter concentration feature vector at time instant denotes the attention context vector, denotes the transpose of denotes the weight matrix, denotes the bias vector, denotes the attention weighted feature vector, denotes the start time instant, denotes the end time instant, denotes the particulate matter concentration feature vector at time instant denotes the natural exponential function; The expression of the output layer is as follows: ， wherein, represents a predicted particulate matter concentration at a target time instant ; The prediction slope is obtained by the following formula: ， wherein, represents a predicted slope, represents the measured particulate matter concentration at the time instant.

4. The method of claim 1, wherein the gas particle adjustment method is a gas particle adjustment method of a sulfur hexafluoride gas-insulated combined electrical apparatus, and The GPR online learning mechanism adopts a Gaussian process regression model, and a kernel function of the Gaussian process regression model adopts a covariance function as follows: ， wherein denotes the covariance function, and denotes the sample point index, , , denotes the total number of sample points, denotes the variance of all sample points, denotes the natural exponential function, denotes the set of observation data the independent identically distributed noise variance, denotes the gas flow rate of the sample , denotes the pressure of the sample , denotes the gas flow rate of the sample , denotes the pressure of the sample , denotes the gas flow rate characteristic scale, denotes the pressure characteristic scale, denotes the Kronecker delta function, , , otherwise ; hyperparameters of the kernel function by the log marginal likelihood function as follows: ， wherein , denotes the observation data of the sample , denotes the transpose of , denotes the set of sample points, , denotes the covariance matrix, denotes the inverse matrix of , , denotes the data in the i-th row and j-th column of the covariance matrix . .​ 5. The method of claim 4, wherein the gas particle adjustment method is a gas particle adjustment method of a sulfur hexafluoride gas-insulated combined electrical apparatus, and The adaptive particle accumulation threshold is obtained by using a GPR online learning mechanism according to the acquired real-time gas flow rate and real-time pressure of the sulfur hexafluoride gas insulated combined electric appliance, and comprises the following steps: calculating a covariance of the real-time gas flow rate and real-time pressure with each sample point, resulting in a covariance matrix ; computing an adaptive particulate matter accumulation threshold from the covariance matrix , computing an adaptive particulate matter accumulation threshold from the covariance matrix The covariance between the real-time gas flow rate and real-time pressure and each sample point is obtained by the following formula: ， wherein, represents the real-time gas flow rate, represents the real-time pressure, represents and the covariance of The adaptive particle accumulation threshold is obtained by the following formula: ， wherein, denotes an adaptive particulate matter accumulation threshold, denotes the transpose of 6. The method of claim 1, wherein the gas particle adjustment method is a gas particle adjustment method of a sulfur hexafluoride gas-insulated combined electrical apparatus, and The slope threshold is obtained by the following formula: ， wherein denotes a slope threshold value, denotes a limit rate of change of the concentration of the gas particulate matter per unit time, denotes an exponential decay characteristic of the rate of change of the concentration of the gas particulate matter during the gas extraction process, denotes a target time instant, denotes a constant; The GPR proxy model in the GPR-based Bayesian optimization algorithm is subject to the following normal distribution: ， wherein, represents the corresponding particulate matter concentration for a gas flow rate of represents the sample mean of the GPR proxy model, represents the sample variance of the GPR proxy model;​ The expected improvement function of the GPR-based Bayesian optimization algorithm is as follows: wherein, represents a corresponding expected improvement value for a gas flow rate of represents a normal distribution value of the GPR proxy model, represents an adaptive particulate matter accumulation threshold, represents a real-time gas flow rate, represents a real-time pressure, represents a sample standard deviation of the GPR proxy model, represents a cumulative distribution function of a standard normal distribution, represents a probability density function of a standard normal distribution;​​ from the desired improvement function, a flow rate of the gas to be adjusted is obtained so that the desired improvement value is maximized, and a flow rate of the gas to be adjusted is obtained adjusting the flow rate of the sulfur hexafluoride gas in a gas-insulated combined electrical apparatus; wherein .

7. A gas particle conditioning system for a sulfur hexafluoride gas- insulated electrical power apparatus, comprising: The method comprises the following steps: a preprocessing module is configured to perform data preprocessing on the acquired gas characteristic data of the sulfur hexafluoride gas insulated combined electric appliance in a fixed time window to obtain preprocessed gas characteristic data; a particle concentration prediction module is configured to perform particle concentration prediction on the preprocessed gas characteristic data by using a trained gas characteristic prediction model to obtain predicted particle concentration and calculate a prediction slope; a GPR online learning module is configured to obtain an adaptive particle accumulation threshold by using a GPR online learning mechanism according to the acquired real-time gas flow rate and real-time pressure of the sulfur hexafluoride gas insulated combined electric appliance; a Bayesian optimization module is configured to adjust the gas flow rate of the sulfur hexafluoride gas insulated combined electric appliance by using a GPR-based Bayesian optimization algorithm if the prediction slope is greater than or equal to a slope threshold or the predicted particle concentration is greater than or equal to the adaptive particle accumulation threshold.

8. A computer readable storage medium having stored thereon computer instructions, wherein, The computer instructions, when executed by the processor, implement the steps of the method for adjusting gas particles of a sulfur hexafluoride gas-insulated combined electric appliance according to any one of claims 1-6.

9. A computer device, comprising: The computer instructions, when executed by the processor, implement the steps of the method for adjusting gas particles of a sulfur hexafluoride gas-insulated combined electric appliance according to any one of claims 1-6. The computer instructions, when executed by the processor, implement the steps of the method for adjusting gas particles of a sulfur hexafluoride gas-insulated combined electric appliance according to any one of claims 1-6. The computer instructions, when executed by the processor, implement the steps of the method for adjusting gas particles of a sulfur hexafluoride gas-insulated combined electric appliance according to any one of claims 1-6.

10. A computer program product comprising computer instructions, characterized in that, ​

Images