Multi-source fusion GIS pipe bus flange integrated monitoring method, system, equipment and medium

Abstract

The invention relates to the technical field of GIS pipe bus flange monitoring, and discloses a multi-source fusion GIS pipe bus flange integrated monitoring method, system, device and medium, and the method comprises the steps: obtaining the multi-source operation monitoring data of a target GIS pipe bus flange, building a monitoring research model based on the multi-source operation monitoring data, and carrying out the multi-source operation monitoring data of the target GIS pipe bus flange; taking the monitoring research model as an intelligent agent, establishing a multi-source information fusion decision framework based on the intelligent agent, constructing a fusion algorithm, performing feature extraction and state judgment on the multi-source information fusion decision framework, and obtaining a comprehensive health state evaluation result of the GIS tubular bus flange. And the operation state of the GIS tubular bus flange can be comprehensively reflected. By establishing a multi-source information fusion decision framework and an optimal state recognition model, deep integration and analysis of multi-source data are realized, associated information among the data is mined, and the reliability and practicability of a monitoring result are improved.

Description

Technical Field

[0001] This invention relates to the field of GIS pipe flange monitoring technology, and in particular to a multi-source fusion integrated monitoring method, system, equipment and medium for GIS pipe flanges. Background Technology

[0002] Current GIS pipe main flange monitoring technology has certain limitations. Some traditional monitoring methods focus only on monitoring a single type of signal, such as vibration or temperature signals, which makes it difficult to comprehensively reflect the operating status of the GIS pipe main flange. This single monitoring method is prone to overlooking other potential faults, resulting in an inaccurate assessment of the overall health status of the equipment.

[0003] Furthermore, some existing monitoring systems lack effective information fusion mechanisms. Different types of monitoring data are independent of each other, without in-depth integration and analysis, failing to fully explore the correlation information between multi-source data, thus affecting the reliability and practicality of monitoring results.

[0004] Furthermore, existing condition recognition models are poorly adaptable to complex operating conditions. When GIS pipe flanges face various complex operating environments and changing conditions, these models struggle to accurately identify the true state of the equipment, easily leading to misjudgments or omissions, which pose risks to the safe and stable operation of the equipment.

[0005] Meanwhile, some monitoring methods lack the ability to provide early warnings of faults. They often only issue alarms when the fault has already developed to a relatively serious stage, failing to provide timely and effective early warnings in the early stages of a fault, and thus failing to meet the needs of preventive maintenance of equipment in actual production.

[0006] Furthermore, existing monitoring technologies also have limitations in data processing efficiency. As the volume of monitoring data continues to increase, traditional data processing methods struggle to process large amounts of data quickly and accurately, leading to untimely feedback of monitoring results and impacting equipment operation and maintenance decisions. Summary of the Invention

[0007] In view of the problems existing in the prior art, the present invention provides a multi-source fusion integrated monitoring method, system, equipment and medium for GIS pipe main flanges, which can solve the problems existing in the current GIS pipe main flange monitoring technology, such as focusing only on single signal monitoring, lack of effective information fusion mechanism, poor adaptability of state recognition model, insufficient fault early warning capability and low data processing efficiency.

[0008] To solve the above-mentioned technical problems, the present invention provides the following technical solution: In a first aspect, the present invention provides a multi-source fusion integrated monitoring method for GIS pipe main flanges, the method comprising: Acquire multi-source operation monitoring data of the target GIS pipe flange, and establish a first monitoring research model based on the multi-source operation monitoring data. The first monitoring research model is used to construct a state identification objective function and several constraints characterizing the normal operation state of the GIS pipe flange based on the collected operation data. The first monitoring and research model is used as an intelligent agent to establish a multi-source information fusion decision framework based on the intelligent agent. In the multi-source information fusion decision framework, the reward function is an arbitrary function that represents and guides the intelligent agent to accurately identify the status within the normal operating status domain of the target GIS pipe flange. A first fusion algorithm is constructed to extract features and determine the state of the multi-source information fusion decision framework to obtain the optimal state recognition model. The first fusion algorithm is used to fuse the vibration features extracted by multi-scale entropy and the temperature features extracted by principal component analysis to obtain the optimal state recognition model. The optimal state identification model is solved to obtain the comprehensive health status assessment results of the GIS pipe flange.

[0009] Furthermore, including: The state recognition objective function includes: Establish an arbitrary function to represent the accuracy of fault state identification of the target GIS main flange; Fault conditions include overheating due to poor contact, abnormal vibration due to mechanical loosening, and SF6 gas leakage or excessive moisture.

[0010] Furthermore, including: The construction of several constraints characterizing the normal operating status of GIS pipe main flanges includes: The first constraint is the mapping relationship between the conductor temperature rise and the shell temperature rise: ; Among them, settings The temperature rise of the conductor relative to the environment, The proportionality coefficient represents the temperature rise of the outer casing relative to the environment. Determined by the thermal conductivity characteristics of the equipment, a value typically between 2 and 3 is taken as an empirical value, and the deviation threshold is... Tolerance for measurement errors and short-term fluctuations; The second constraint is the energy constraint of the vibration signal spectrum: ; Among them, settings Vibration signal The Fourier transform results are used to analyze its frequency components and frequency bands. Designated as the sensitive frequency range for the device, abnormal energy increases in these frequency bands are typically associated with mechanical loosening or resonance, and the energy threshold... It is derived from historical normal data and is used to determine whether there is a significant vibration anomaly.

[0011] The third constraint is the safety threshold constraint for SF6 gas parameters: ; Among them, settings and This is the rated pressure range for SF6 gas. Exceeding this range may indicate a leak or overcharging. This represents the upper limit for trace moisture content. The fourth constraint is the correction model for the influence of environmental temperature and humidity: ; Among them, settings To be in standard ambient temperature and conductor reference temperature under rated load, The environmental temperature influence coefficient reflects the transfer effect of environmental temperature changes on the conductor's temperature rise; the residual term... This indicates an additional temperature rise that cannot be explained by environmental factors. When it remains excessively high, it is determined to be an overheating fault caused by poor internal contact. The above four types of constraints together constitute a constraint set. It defines all the physical and engineering boundary conditions that the GIS pipe flange should meet under normal operating conditions.

[0012] Furthermore, including: In the multi-source information fusion decision framework, the reward function is any function that characterizes the guiding agent's accurate state identification within the normal operating state domain of the target GIS pipe flange, including: The vibration characteristics, temperature characteristics, SF6 gas state characteristics, and environmental characteristics of the target GIS pipe flange are used as the states in the multi-source information fusion decision framework. The status judgment result of the target GIS pipe flange is used as a decision variable, and the decision variable is used as an action in the multi-source information fusion decision framework. The reward function includes a penalty function, and the penalty items of the penalty function include a penalty item for misjudging abnormal vibration state, a penalty item for misjudging excessive temperature state, and a penalty item for misjudging abnormal SF6 gas state. The penalty item for misjudging abnormal vibration state is used to suppress false alarms caused by external vibration interference, and the penalty item for misjudging excessive temperature state is used to address false alarms caused by sudden environmental changes.

[0013] Furthermore, including: Solving the optimal state recognition model includes: Real-time acquisition of vibration signals, temperature signals, SF6 gas state signals, and ambient temperature and humidity signals of the target GIS pipe flange; Based on the real-time acquired multi-source signals, inference calculations are performed in conjunction with the optimal state recognition model. The inference calculation results are used as the comprehensive health status assessment results for the GIS pipe flange.

[0014] Furthermore, including: The construction of the first fusion algorithm involves feature extraction and state discrimination of the multi-source information fusion decision framework to obtain the optimal state recognition model, including: Multi-scale entropy is used to extract vibration features, which emphasizes the analysis of the complex changes in vibration signals across multiple time scales, and is expressed as: ; Among them, coarse-grained sequences are set. for: ; Among them, scale factor The coarsening level of the time scale is controlled, ranging from 1 to 10, to cover a variety of dynamic behaviors from instantaneous shocks to long-term trends. For the embedding dimension, a similarity tolerance is set. Set to 0.2 times the standard deviation of the signal, the sample entropy SampEn is used to measure the regularity of the sequence. The lower the entropy value, the more regular the signal, which is suitable for detecting periodic enhancement caused by early loosening.

[0015] Furthermore, including: The construction of the first fusion algorithm, which performs feature extraction and state discrimination on the multi-source information fusion decision framework to obtain the optimal state recognition model, also includes: Principal component analysis was performed on the temperature data to reduce its dimensionality, which is represented as follows: ; Among them, settings for A temperature data matrix with p rows and p columns, where n is the number of samples and p is the dimension of the temperature-related variables. The projection results are constructed from the eigenvectors corresponding to the k largest eigenvalues ​​of the covariance matrix, where k is either 2 or 3 to retain more than 95% of the information. The principal component score after dimensionality reduction significantly reduces computational complexity and eliminates redundancy between variables; Finally, the four types of features are fused and input into the machine learning classifier: ; Among them, settings For the comprehensive state discrimination model, the model parameters It was trained using a large amount of labeled data, and the training objective was to maximize the objective function. Simultaneously satisfying the constraint set The final output model is the optimal state recognition model.

[0016] Secondly, the present invention provides a multi-source fusion integrated monitoring system for GIS pipe flanges, comprising: The first model building module is used to acquire multi-source operation monitoring data of the target GIS main flange and establish a first monitoring research model based on the multi-source operation monitoring data. The multi-source operation monitoring data includes vibration signals, temperature signals, SF6 gas state signals, and ambient temperature and humidity signals; The first monitoring and research model includes a state identification objective function and several constraints characterizing the normal operating state of the GIS pipe flange; The framework establishment module is used to establish a multi-source information fusion decision framework for the first monitoring and research model. The optimal model building module is used to construct the first fusion algorithm, which performs feature extraction and state discrimination on the multi-source information fusion decision framework to obtain the optimal state recognition model. The first fusion algorithm includes vibration feature extraction based on multi-scale entropy, temperature feature extraction based on principal component analysis, and comprehensive state discrimination based on machine learning; The solution module is used to solve the optimal state identification model and obtain the comprehensive health status assessment results of the GIS pipe flange.

[0017] Thirdly, the present invention provides an electronic device including a memory and a processor, wherein the memory stores a computer program, and the processor executes the computer program to implement the steps of the method described above.

[0018] Fourthly, the present invention provides a computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, implements the steps of the method described above.

[0019] Compared with existing technologies, the beneficial effects of this invention are as follows: This invention proposes a multi-source fusion integrated monitoring method for GIS pipe main flanges. This method integrates multi-source operational monitoring data, solving the problem that existing monitoring technologies only focus on single signal monitoring, and can comprehensively reflect the operating status of GIS pipe main flanges. By establishing a multi-source information fusion decision framework and an optimal state identification model, deep integration and analysis of multi-source data are achieved, the correlation information between data is mined, and the reliability and practicality of monitoring results are improved.

[0020] Under complex operating conditions, the state recognition model of this invention exhibits excellent adaptability, accurately identifying the true state of equipment and reducing misjudgments and omissions, thus ensuring the safe and stable operation of the equipment. Simultaneously, this method enhances the early warning capability for faults, issuing timely warnings at the early stages of a fault, meeting the needs of preventative maintenance in actual production.

[0021] In terms of data processing, this invention employs a highly efficient data processing method, capable of quickly and accurately processing large amounts of monitoring data and providing timely feedback on monitoring results, thus offering strong support for equipment operation and maintenance decisions. Through the application of this invention, the monitoring level of GIS main flanges can be significantly improved, reducing equipment failure risks and enhancing the operational efficiency and reliability of the power system. Attached Figure Description

[0022] To more clearly illustrate the technical solutions of the embodiments of the present invention, the drawings used in the following description of the embodiments will be briefly introduced. Obviously, the drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0023] Figure 1 This is a flowchart of a multi-source fusion integrated monitoring method for GIS pipe main flanges, provided as an embodiment of the present invention.

[0024] Figure 2 This is an internal structural diagram of an electronic device for a multi-source fusion integrated monitoring method for GIS pipe flanges, provided in one embodiment of the present invention. Detailed Implementation

[0025] To make the above-mentioned objects, features, and advantages of the present invention more apparent and understandable, specific embodiments of the present invention will be described in detail below with reference to the accompanying drawings. Obviously, the described embodiments are only a part of the embodiments of the present invention, and not all of them. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort should fall within the protection scope of the present invention.

[0026] Example 1 is the first embodiment of the present invention, which provides a multi-source fusion integrated monitoring method for GIS pipe main flanges, including: Existing technologies suffer from several problems. For instance, some monitoring methods suffer from insufficient sensor accuracy and stability during data acquisition, leading to errors in the acquired multi-source monitoring data and affecting the accuracy and reliability of subsequent models. Furthermore, some monitoring systems are susceptible to external interference during data transmission, causing data loss or errors, further reducing monitoring effectiveness. Additionally, the lack of unified standards and specifications for the storage and management of multi-source data makes data retrieval, analysis, and sharing difficult, hindering the overall operation and optimization of the monitoring system.

[0027] It should be noted in advance that the system mentioned in Example 1, which is a real-time operation with the system as the subject, refers to any system configured with this method.

[0028] This invention provides a method that can effectively solve the problems mentioned above. The following will describe in detail how to realize the multi-source fusion integrated monitoring method for GIS pipe flanges using multiple embodiments. Figure 1 A flowchart illustrating a multi-source fusion integrated monitoring method for GIS pipe flanges is shown, including: S101, Obtain multi-source operation monitoring data of the target GIS main flange, and establish a first monitoring research model based on the multi-source operation monitoring data, wherein: It should be noted that, in order to achieve integrated monitoring of GIS pipe main flanges through multi-source fusion, the underlying logic can be identified from different multi-source operational monitoring data, and key features that accurately reflect the operational status of the GIS pipe main flanges can be extracted from these data. Through extensive experiments and data analysis, the correlation between different types of data is determined, ultimately achieving integrated design.

[0029] In this embodiment of the invention, the multi-source operation monitoring data includes vibration signals, temperature signals, SF6 gas state signals, and ambient temperature and humidity signals. In one specific implementation, multi-source operational monitoring data can be acquired through various methods. For vibration signals, high-precision vibration sensors can be installed at key locations on the GIS pipe flange to collect vibration data in real time at a certain sampling frequency. Temperature signals can be acquired using distributed fiber optic temperature sensors arranged along the GIS pipe flange to accurately measure temperature changes at different locations. For SF6 gas status signals, specialized gas detection equipment is used to monitor parameters such as SF6 gas pressure and trace moisture content in real time. Ambient temperature and humidity signals can be acquired using environmental temperature and humidity sensors installed near the GIS equipment.

[0030] It should be noted that after acquiring these multi-source operational monitoring data, preprocessing is necessary due to the diversity and complexity of the data, as well as the potential correlations and redundancies between different data points. Since the data may be affected by noise interference and missing values ​​during actual acquisition, the vibration signals need to be filtered to remove high-frequency noise; for missing values ​​in the temperature and SF6 gas state signals, interpolation methods can be used to supplement them. These preprocessing steps improve data quality and provide a reliable data foundation for the subsequent establishment of the first monitoring research model.

[0031] In this embodiment of the invention, the first monitoring and research model includes a state identification objective function and several constraints characterizing the normal operating state of the GIS main flange. In this embodiment of the invention, the state recognition objective function includes: Establish an arbitrary function to represent the accuracy of fault state identification of the target GIS main flange; Fault conditions include overheating due to poor contact, abnormal vibration due to mechanical loosening, and SF6 gas leakage or excessive moisture.

[0032] It should be noted that overheating caused by poor contact refers not only to obvious temperature rise, but also to latent overheating caused by early increase in contact resistance. For example, when the conductor temperature rise is 3°C higher than the historical average under the same load, and the shell temperature rise does not rise synchronously, combined with infrared thermography and current data, it can be determined that it is an initial overheating caused by poor contact.

[0033] In this embodiment of the invention, several constraints characterizing the normal operating status of the GIS pipe flange include constraints on the mapping relationship between conductor temperature rise and shell temperature rise, constraints on vibration signal spectrum characteristics, constraints on SF6 gas pressure and micro-moisture content thresholds, and constraints on corrections for the influence of environmental temperature and humidity.

[0034] It should be noted that establishing the first monitoring research model does not simply refer to data fitting or empirical judgment, but rather to combining multi-source monitoring data with physical mechanisms to construct a mathematical model that includes optimization objectives and operational boundaries. For example, in the monitoring of the GIS main flange of a 500kV substation, by collecting its vibration and temperature data under different loads, and combining the heat conduction equation and mechanical dynamics model, an initial research model was constructed with the objective of "maximizing the fault identification rate" and the constraint that "the conductor temperature rise does not exceed 2.5 times the shell temperature rise," which serves as the basis for subsequent intelligent integration.

[0035] It should be noted that the mapping relationship between conductor temperature rise and shell temperature rise is constrained based on the theory of heat conduction and is used to identify internal contact anomalies. For example, under normal circumstances, the conductor temperature rise is about 2.2 times the shell temperature rise. If this ratio reaches 3.5 in a certain measurement and the shell temperature does not rise significantly, it can be determined that there is a high-resistance contact point in the internal connection.

[0036] It should be noted that the environmental temperature and humidity influence correction constraint is used to eliminate the interference of external climate on temperature judgment; for example, when the ambient temperature rises from 25°C to 35°C, the conductor temperature naturally rises by about 6°C. If the actual temperature rise reaches 10°C, the residual of 4°C should be attributed to internal factors, such as increased load or contact deterioration.

[0037] Specifically, this stage is the foundational construction phase of the entire monitoring system. Its core task is to collect comprehensive operational data and, based on physical mechanisms and operational experience, establish an initial research model that includes objective functions and constraints, providing structured input for subsequent information fusion and intelligent decision-making.

[0038] First, a high-precision sensor network is deployed at key locations on the target GIS pipe flange to achieve synchronous acquisition of various physical quantities.

[0039] The collected data constitutes a multidimensional time series vector, represented as: The present invention sets The multi-source input vector at time t is used to characterize the overall operating status of the GIS pipe flange at that time.

[0040] in, This indicates the vibration signal collected by the accelerometer, reflecting the tightness and dynamic stability of the mechanical connection; Indicates the temperature of the conductor. This indicates the outer casing temperature; both of these factors together reflect the thermal conductivity and changes in contact resistance. Indicates SF6 gas pressure. Indicates trace moisture content, used to assess insulation performance and seal integrity; and These represent ambient temperature and humidity, respectively, and are used to correct for the impact of external climate on equipment status.

[0041] Furthermore, after acquiring the aforementioned multi-source data, a first monitoring and research model is constructed. The model consists of two core parts: a state recognition objective function. With normal operating state constraint set .

[0042] Furthermore, the design objective of the state recognition objective function is to maximize the accuracy of identifying typical faults, and its mathematical expression is: ; In this invention, the following settings are provided: It is a comprehensive performance indicator used to measure the model's overall ability to identify three main types of faults.

[0043] in, This indicates the accuracy rate in identifying overheating faults caused by poor contact. This indicates the accuracy rate in identifying abnormal vibrations caused by mechanical loosening. This indicates the accuracy rate in identifying SF6 gas leaks or excessive moisture levels. Weighting coefficient. , , Used to adjust the importance of different types of faults in the objective function, satisfying... This ensures the normalization of the objective function. Optimize variables. This represents the set of all adjustable parameters in the model, including classifier weights, feature extraction thresholds, etc., which are continuously adjusted during the training process to maximize... .

[0044] Furthermore, in addition to the objective function, this invention introduces a set of physical rationality and operational boundary constraints to constitute the feasible region for normal operation. These constraints ensure that the model does not misinterpret normal fluctuations that conform to physical laws as faults.

[0045] The first constraint is the mapping relationship between the conductor temperature rise and the shell temperature rise: ; In this invention, the following settings are provided: The temperature rise of the conductor relative to the environment, This refers to the temperature rise of the casing relative to the environment; under normal operating conditions, the two should maintain a certain proportional relationship. Proportional coefficient. Determined by the thermal conductivity characteristics of the equipment, it is usually taken as an empirical value between 2 and 3. Deviation threshold To tolerate measurement errors and short-term fluctuations, this invention is configured as follows: If the value exceeds this range, it may indicate an abnormal contact resistance.

[0046] The second constraint is the energy constraint of the vibration signal spectrum: ; In this invention, the following settings are provided: Vibration signal The Fourier transform result is used to analyze its frequency components. Frequency band. Designated frequency ranges as equipment sensitivity zones, such as the 50Hz power frequency and its harmonics (100Hz, 150Hz, etc.), abnormal energy increases in these frequency bands are usually associated with mechanical loosening or resonance. Energy threshold. It is derived from historical normal data and is used to determine whether there is a significant vibration anomaly.

[0047] The third constraint is the safety threshold constraint for SF6 gas parameters: ; In this invention, the following settings are provided: and This refers to the rated pressure range of SF6 gas, for example, 0.4 MPa to 0.6 MPa. Exceeding this range may indicate a leak or overcharging. The upper limit for trace moisture content is usually set at 200 ppm. Exceeding this value will significantly reduce insulation strength and increase the risk of discharge.

[0048] The fourth constraint is the correction model for the influence of environmental temperature and humidity: The present invention sets To be in standard ambient temperature and conductor reference temperature under rated load, This is the ambient temperature influence coefficient, reflecting the transfer effect of ambient temperature changes on the conductor's temperature rise. Residual term. This indicates an additional temperature rise that cannot be explained by environmental factors. When this rise remains excessively high, it can be identified as an overheating fault caused by poor internal contact. The above four types of constraints together constitute the constraint set. It defines all the physical and engineering boundary conditions that the GIS pipe flange should meet under normal operating conditions.

[0049] It should be noted that acquiring multi-source operational monitoring data of the target GIS pipe flange and establishing a primary monitoring research model based on this data provides structured input for subsequent information fusion and intelligent decision-making. This primary monitoring research model allows for preliminary analysis and processing of the multi-source operational monitoring data, identifying potential fault conditions. In subsequent information fusion, this model can serve as a foundation for integration with other data and models, further improving the accuracy and reliability of fault identification.

[0050] S102, Establish a multi-source information fusion decision-making framework for the first monitoring research model, wherein: It should be noted that after obtaining the first monitoring research model, in order to more effectively process and utilize multi-source operational monitoring data, a multi-source information fusion decision-making framework needs to be established. The core of this framework is to organically integrate data from different sources and of different types, and to explore the potential correlations between the data, so as to achieve a more accurate assessment and decision-making regarding the operational status of GIS pipe flanges.

[0051] In this embodiment of the invention, establishing a multi-source information fusion decision framework for the first monitoring research model includes: The first monitoring research model is treated as an intelligent agent; Establish a multi-source information fusion decision-making framework based on intelligent agents; In the multi-source information fusion decision framework, the reward function is an arbitrary function that represents the guiding agent to accurately identify the state within the normal operating state domain of the target GIS pipe flange.

[0052] It should be noted that treating the first monitoring research model as an intelligent agent does not refer to a physical entity, but rather to abstracting it into a software agent with perception, decision-making, and learning capabilities. For example, in the system software, this model is encapsulated as a module that can receive feature inputs, output state discrimination, and adjust parameters based on feedback, similar to an agent in reinforcement learning, enabling it to continuously optimize the recognition strategy through historical discrimination results.

[0053] It should be noted that the reward function, which is any function that represents the agent's ability to accurately identify the normal state, is not unlimited. Rather, it means that its mathematical form can be flexibly designed while meeting the guidance objective. For example, a +1 reward is given when the agent correctly identifies a normal state, a -0.5 penalty is given when a normal state is mistakenly identified as a fault, and a -2 penalty is given when a real fault is missed. This reward mechanism guides the model to reduce false alarms while ensuring safety.

[0054] In embodiments of the present invention, establishing a multi-source information fusion decision-making framework based on intelligent agents includes: The vibration characteristics, temperature characteristics, SF6 gas state characteristics, and environmental characteristics of the target GIS pipe flange are used as the states in the multi-source information fusion decision framework. The status determination result of the target GIS pipe flange is used as a decision variable; The decision variables are treated as actions within a multi-source information fusion decision framework.

[0055] It should be noted that using the state discrimination result as a decision variable in the decision variable refers to an output variable that can be adjusted during the optimization process, rather than a fixed threshold judgment. For example, the "whether it is overheating" output by the model is not a simple "yes / no" switch, but a finely adjustable probability value, such as 0.85. This value participates in the objective function optimization as a decision variable, enabling the system to weigh the risks of false alarms and false alarms.

[0056] It should be noted that treating decision variables as actions is a term borrowed from reinforcement learning, emphasizing that state determination is an active decision-making process; for example, at a certain moment, the agent determines the state based on its current features. Select the action to execute "output overheat warning". The system will then provide feedback on the actual inspection results, including rewards or penalties, to help learn better judgment strategies.

[0057] In this embodiment of the invention, the reward function includes a penalty function, and the penalty terms of the penalty function include a penalty term for misjudging abnormal vibration state, a penalty term for misjudging excessive temperature state, and a penalty term for misjudging abnormal SF6 gas state.

[0058] It should be noted that the multi-source information fusion decision framework is not a simple data splicing or weighted averaging, but rather a structured decision system that uses multi-source information as state input and state discrimination as action output. For example, features such as vibration entropy value, temperature principal component score, and gas pressure change rate can be combined into a state vector and input into a reinforcement learning-like decision structure. The model can then autonomously learn which feature combination should output a discrimination result such as "normal" or "overheating warning".

[0059] It should be noted that the vibration abnormality misjudgment penalty term is used to suppress false alarms caused by external vibration interference. For example, when nearby construction causes ground vibration, which leads to an increase in sensor signals, but there is no loosening inside the equipment, if the model misjudges it as "mechanical loosening", this penalty term will be triggered, prompting the model to learn to distinguish between external interference and real faults.

[0060] It should be noted that the penalty for misjudging temperature exceeding limits pays special attention to misjudgments caused by sudden environmental changes. For example, direct sunlight at noon in summer causes the shell temperature to rise sharply, but the conductor temperature does not rise synchronously. If the model judges it as "overheating" based solely on the shell temperature, it should be penalized to strengthen its dependence on the conductor-shell temperature difference.

[0061] Specifically, after the first monitoring research model M1 is established, this stage abstracts it into an intelligent agent and constructs a multi-source information fusion decision-making framework based on reinforcement learning, realizing the leap from static model to dynamic discrimination.

[0062] In this invention, M1 is regarded as an intelligent agent with perception and decision-making capabilities. Its perception input is multi-source features, and its decision output is state discrimination action.

[0063] The core of this framework is defining the state space, action space, and reward function to form an optimizable decision-making loop. The state space S consists of four types of feature vectors: ; In this embodiment, the settings are as follows: Let be the system state perceived by the agent at time t, where . This is a multi-scale entropy feature vector extracted from vibration signals, reflecting the signal complexity and regularity degradation. The temperature characteristics after dimensionality reduction by principal component analysis retain the main thermal change patterns. The state characteristics of SF6 gas include pressure, micro-water content and its rate of change; It is an environmental feature used to compensate for external disturbances.

[0064] The state space is the input basis for an agent's decision-making. In reinforcement learning, the agent selects actions based on its current state and learns the mapping from state to action through a policy network to maximize cumulative reward. The size of the state space directly affects policy complexity and learning efficiency; high-dimensional state spaces require function approximation. Value functions and state evaluation: The state value function and action value function evaluate the long-term benefits of a state or state-action pair based on the state space, guiding policy optimization.

[0065] Furthermore, the action space A is defined as the state discrimination result: In this embodiment, the following settings are provided. This represents the action performed by the intelligent agent at time t, i.e., the judgment result of the device's health status. 0 indicates normal state, 1 indicates overheating risk, 2 indicates abnormal vibration, and 3 indicates abnormal gas conditions.

[0066] It should be noted that this invention treats state discrimination as an action, enabling the model to have proactive decision-making capabilities.

[0067] To further guide the agent to make accurate identifications within the normal operating domain, this invention designs a reward function: ; In this embodiment, the settings are as follows: Positive rewards are given for correct identification to incentivize the agent to make accurate judgments; This is a penalty for misjudging abnormal vibrations, to prevent normal vibrations from being mistakenly reported as faults. This is a penalty for misjudging temperature limits, to prevent overheating alarms from being triggered erroneously due to environmental fluctuations; This is a penalty for misjudging SF6 gas anomalies, used to improve the reliability of gas monitoring. Penalty weight. , , It can be adjusted according to the importance of the site to ensure that critical faults are identified first.

[0068] It should be noted that this decision-making framework not only achieves the structured organization of multi-source information, but also provides quantifiable learning objectives for subsequent model optimization.

[0069] S103, Construct the first fusion algorithm to perform feature extraction and state discrimination on the multi-source information fusion decision framework, and obtain the optimal state recognition model, wherein: It should be noted that after obtaining the multi-source information fusion decision framework, a suitable fusion algorithm needs to be constructed to process it in order to achieve feature extraction and state discrimination, and then obtain the optimal state recognition model.

[0070] In this embodiment of the invention, the first fusion algorithm includes vibration feature extraction based on multi-scale entropy, temperature feature extraction based on principal component analysis, and comprehensive state discrimination based on machine learning. It should be noted that vibration feature extraction based on multi-scale entropy emphasizes analyzing the complex changes in vibration signals across multiple time scales, rather than a single frequency domain analysis. For example, coarse-grained processing of the acquired vibration time-series signals at scales τ=1 to 10 is performed, and the sample entropy at each scale is calculated. A significant decrease in time entropy indicates a periodic enhancement of the equipment over a medium- to long-term timescale, which may foreshadow a loosening of mechanical connections.

[0071] It should be noted that the comprehensive state discrimination based on machine learning does not use only a single classifier, but refers to the use of an algorithm with nonlinear discrimination capability to make the final decision after fusing multi-source features. For example, features such as vibration multi-scale entropy, temperature principal component, and gas micro-moisture content are input into a random forest classifier to train it to identify four states: "normal", "overheated", "loose", and "gas leak". The model outputs the class with the highest probability and its confidence level.

[0072] Specifically, after the decision-making framework is established, this stage uses a fusion algorithm to perform in-depth processing of multi-source features, combined with machine learning methods, to train the optimal state recognition model. .

[0073] First, feature extraction based on multi-scale entropy is performed on the vibration signal: In this embodiment, a coarse-grained sequence is set. for: Among them, the scaling factor The coarsening degree of the time scale is controlled, ranging from 1 to 10 in this invention, to cover a variety of dynamic behaviors from instantaneous shocks to long-term trends.

[0074] Embedding dimension Set to 2, similarity tolerance Set to 0.2 times the signal standard deviation. Sample entropy (SampEn) is used to measure the regularity of a sequence; the lower the entropy value, the more regular the signal, and it is suitable for detecting periodic enhancements caused by early loosening.

[0075] Secondly, principal component analysis was performed on the temperature data for dimensionality reduction. ; In this embodiment, the settings are as follows: for A temperature data matrix with p rows and p columns, where n is the number of samples and p is the dimension of the temperature-related variables. Matrix W is composed of the eigenvectors corresponding to the first k largest eigenvalues ​​of the covariance matrix; k is typically 2 or 3 to retain more than 95% of the information. Projection results. The principal component score after dimensionality reduction significantly reduces computational complexity and eliminates redundancy between variables.

[0076] Finally, the four types of features are fused and input into the machine learning classifier: The present invention sets For comprehensive state discrimination models, ensemble learning algorithms such as Random Forest or XGBoost are employed, exhibiting strong generalization ability and robustness. Model parameters It was trained using a large amount of labeled data, and the training objective was to maximize the objective function. Simultaneously satisfying the constraint set The final output model is the optimal state recognition model. .

[0077] It should be noted that constructing the first fusion algorithm to extract features and determine the state of the multi-source information fusion decision framework, and obtaining the optimal state recognition model, can further improve the accuracy and reliability of the operational status assessment of GIS pipe flanges. Vibration feature extraction based on multi-scale entropy can capture the complex changes in vibration signals at different time scales, more sensitively detecting early mechanical connection loosening and other problems in equipment, providing more timely information for fault early warning. Temperature feature extraction based on principal component analysis effectively reduces the dimensionality of temperature data, reduces computational complexity, and eliminates redundant information between variables, enabling the model to process temperature data more efficiently and improve the ability to identify abnormal temperature conditions.

[0078] S104, Solve the optimal state identification model to obtain the comprehensive health status assessment result of the GIS pipe flange, where: It should be noted that after obtaining the optimal condition identification model, it needs to be solved to obtain the comprehensive health status assessment result of the GIS pipe flange. The solution process involves inputting the actual collected multi-source operation monitoring data into the optimal condition identification model and using the model's discrimination ability to derive the assessment result.

[0079] In this embodiment of the invention, solving the optimal state recognition model includes: Real-time acquisition of vibration signals, temperature signals, SF6 gas state signals, and ambient temperature and humidity signals of the target GIS pipe flange; Inference calculations are performed based on real-time acquired multi-source signals and combined with the optimal state recognition model. The inference calculation results are used as the comprehensive health status assessment results for the GIS pipe flange.

[0080] Specifically, in the optimal model After training is completed, the actual operation phase begins.

[0081] This stage enables real-time monitoring and dynamic evaluation of GIS pipe main flanges.

[0082] Whenever new data arrives, the system executes the following inference process: ; In this invention, the following settings are provided: The newly acquired multi-source signal vectors are used as model input. The system sequentially extracts vibration features. Temperature characteristics Gas characteristics With environmental characteristics And input it into the trained model: The present invention sets The comprehensive health status category output by the model is used as the final evaluation result. The system determined that the equipment was operating normally. like If a fault is detected, a corresponding warning will be triggered based on the category and pushed to the operation and maintenance platform to achieve early detection and preventive maintenance of the fault.

[0083] In summary, this invention proposes a multi-source fusion integrated monitoring method for GIS pipe main flanges. This method integrates multi-source operational monitoring data, solving the problem that existing monitoring technologies only focus on single signal monitoring, and can comprehensively reflect the operational status of GIS pipe main flanges. By establishing a multi-source information fusion decision framework and an optimal state identification model, deep integration and analysis of multi-source data are achieved, the correlation information between data is mined, and the reliability and practicality of monitoring results are improved.

[0084] Under complex operating conditions, the state recognition model of this invention exhibits excellent adaptability, accurately identifying the true state of equipment and reducing misjudgments and omissions, thus ensuring the safe and stable operation of the equipment. Simultaneously, this method enhances the early warning capability for faults, issuing timely warnings at the early stages of a fault, meeting the needs of preventative maintenance in actual production.

[0085] In terms of data processing, this invention employs a highly efficient data processing method, capable of quickly and accurately processing large amounts of monitoring data and providing timely feedback on monitoring results, thus offering strong support for equipment operation and maintenance decisions. Through the application of this invention, the monitoring level of GIS main flanges can be significantly improved, reducing equipment failure risks and enhancing the operational efficiency and reliability of the power system.

[0086] Example 2, refer to Figure 2 This embodiment also provides a multi-source fusion integrated monitoring system for GIS pipe flanges, including: The first model building module is used to acquire multi-source operation monitoring data of the target GIS main flange and establish the first monitoring research model based on the multi-source operation monitoring data. Multi-source operation monitoring data includes vibration signals, temperature signals, SF6 gas state signals, and ambient temperature and humidity signals; The first monitoring research model includes a state identification objective function and several constraints characterizing the normal operating state of the GIS pipe flange. The framework building module is used to build a multi-source information fusion decision framework for the first monitoring research model. The optimal model building module is used to construct the first fusion algorithm, which performs feature extraction and state discrimination on the multi-source information fusion decision framework to obtain the optimal state recognition model. The first fusion algorithm includes vibration feature extraction based on multi-scale entropy, temperature feature extraction based on principal component analysis, and comprehensive state discrimination based on machine learning; The solution module is used to solve the optimal state identification model and obtain the comprehensive health status assessment results of the GIS pipe flange.

[0087] The above-mentioned unit modules can be embedded in the processor of the electronic device in hardware form or independent of it, or they can be stored in the memory of the electronic device in software form, so that the processor can call and execute the corresponding operations of the above modules.

[0088] This embodiment also provides an electronic device, which can be a terminal, and its internal structure diagram can be as follows: Figure 2 As shown, the electronic device includes a processor, memory, communication interface, display screen, and input devices connected via a system bus. The processor provides computing and control capabilities. The memory includes a non-volatile storage medium and internal memory. The non-volatile storage medium stores the operating system and computer programs. The internal memory provides an environment for the operation of the operating system and computer programs in the non-volatile storage medium. The communication interface is used for wired or wireless communication with external terminals; wireless communication can be achieved through Wi-Fi, carrier networks, NFC (Near Field Communication), or other technologies. When the computer program is executed by the processor, it implements a multi-source fusion integrated monitoring method for GIS pipe flanges. The display screen can be an LCD screen or an e-ink screen. The input devices can be a touch layer covering the display screen, buttons, a trackball, or a touchpad on the device's casing, or an external keyboard, touchpad, or mouse.

[0089] This embodiment also provides a computer-readable storage medium on which a computer program is stored, and when the computer program is executed by a processor, it performs the following steps: Acquire multi-source operation monitoring data of the target GIS main flange and establish a first monitoring research model based on the multi-source operation monitoring data; Multi-source operation monitoring data includes vibration signals, temperature signals, SF6 gas state signals, and ambient temperature and humidity signals; The first monitoring research model includes a state identification objective function and several constraints characterizing the normal operating state of the GIS pipe flange. Establish a multi-source information fusion decision-making framework for the first monitoring research model; A first fusion algorithm is constructed to extract features and determine the state of the multi-source information fusion decision framework, thereby obtaining the optimal state recognition model. The first fusion algorithm includes vibration feature extraction based on multi-scale entropy, temperature feature extraction based on principal component analysis, and comprehensive state discrimination based on machine learning; The optimal state identification model is solved to obtain the comprehensive health status assessment results of the GIS pipe flange.

[0090] It should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention and are not intended to limit it. Although the present invention has been described in detail with reference to preferred embodiments, those skilled in the art should understand that modifications or equivalent substitutions can be made to the technical solutions of the present invention without departing from the spirit and scope of the technical solutions of the present invention, and all such modifications or substitutions should be covered within the scope of the claims of the present invention.

[0091] Although preferred embodiments of the invention have been described, those skilled in the art, upon learning the basic inventive concept, can make other changes and modifications to these embodiments. Therefore, the appended claims are intended to be interpreted as including both the preferred embodiments and all changes and modifications falling within the scope of the invention.

[0092] Obviously, those skilled in the art can make various modifications and variations to this invention without departing from its spirit and scope. Therefore, if these modifications and variations fall within the scope of the claims of this invention and their equivalents, this invention also intends to include these modifications and variations.

Claims

1. A multi-source fusion integrated monitoring method for GIS pipe main flanges, characterized in that, The method includes: Acquire multi-source operation monitoring data of the target GIS pipe flange, and establish a first monitoring research model based on the multi-source operation monitoring data. The first monitoring research model is used to construct a state identification objective function and several constraints characterizing the normal operation state of the GIS pipe flange based on the collected operation data. The first monitoring and research model is used as an intelligent agent to establish a multi-source information fusion decision framework based on the intelligent agent. In the multi-source information fusion decision framework, the reward function is an arbitrary function that represents and guides the intelligent agent to accurately identify the status within the normal operating status domain of the target GIS pipe flange. A first fusion algorithm is constructed to extract features and determine the state of the multi-source information fusion decision framework to obtain the optimal state recognition model. The first fusion algorithm is used to fuse the vibration features extracted by multi-scale entropy and the temperature features extracted by principal component analysis to obtain the optimal state recognition model. The optimal state identification model is solved to obtain the comprehensive health status assessment results of the GIS pipe flange.

2. The multi-source fusion integrated monitoring method for GIS pipe main flanges according to claim 1, characterized in that, The state recognition objective function includes: Establish an arbitrary function to represent the accuracy of fault state identification of the target GIS main flange; Fault conditions include overheating due to poor contact, abnormal vibration due to mechanical loosening, and SF6 gas leakage or excessive moisture.

3. The multi-source fusion integrated monitoring method for GIS pipe main flanges according to claim 2, characterized in that, The construction of several constraints characterizing the normal operating status of GIS pipe flanges includes: The first constraint is the mapping relationship between the conductor temperature rise and the shell temperature rise: Among them, setting The temperature rise of the conductor relative to the environment, The proportionality coefficient represents the temperature rise of the outer casing relative to the environment. Determined by the thermal conductivity characteristics of the equipment, a value typically between 2 and 3 is taken as an empirical value, with a deviation threshold. Tolerance for measurement errors and short-term fluctuations; The second constraint is the energy constraint of the vibration signal spectrum: ; Among them, settings Vibration signal The Fourier transform results are used to analyze its frequency components and frequency band. Designated as the sensitive frequency range for the device, abnormal energy increases in these frequency bands are typically associated with mechanical loosening or resonance, and the energy threshold... It is derived from historical normal data and is used to determine whether significant vibration anomalies have occurred; The third constraint is the safety threshold constraint for SF6 gas parameters: Among them, setting and This is the rated pressure range for SF6 gas. Exceeding this range may indicate a leak or overcharging. This represents the upper limit for trace moisture content. The fourth constraint is the correction model for the influence of environmental temperature and humidity: ; Among them, settings To be in standard ambient temperature and conductor reference temperature under rated load, The environmental temperature influence coefficient reflects the transfer effect of environmental temperature changes on the conductor's temperature rise; the residual term... This indicates an additional temperature rise that cannot be explained by environmental factors. When it remains excessively high, it is determined to be an overheating fault caused by poor internal contact. The above four types of constraints together constitute a constraint set. It defines all the physical and engineering boundary conditions that the GIS pipe flange should meet under normal operating conditions.

4. The multi-source fusion integrated monitoring method for GIS pipe main flanges according to claim 3, characterized in that, In the multi-source information fusion decision framework, the reward function is any function that characterizes the guiding agent's accurate state identification within the normal operating state domain of the target GIS pipe flange, including: The vibration characteristics, temperature characteristics, SF6 gas state characteristics, and environmental characteristics of the target GIS pipe flange are used as the states in the multi-source information fusion decision framework. The status judgment result of the target GIS pipe flange is used as a decision variable, and the decision variable is used as an action in the multi-source information fusion decision framework. The reward function includes a penalty function, and the penalty items of the penalty function include a penalty item for misjudging abnormal vibration state, a penalty item for misjudging excessive temperature state, and a penalty item for misjudging abnormal SF6 gas state. The penalty item for misjudging abnormal vibration state is used to suppress false alarms caused by external vibration interference, and the penalty item for misjudging excessive temperature state is used to address false alarms caused by sudden environmental changes.

5. The multi-source fusion integrated monitoring method for GIS pipe main flanges according to claim 4, characterized in that, Solving the optimal state recognition model includes: Real-time acquisition of vibration signals, temperature signals, SF6 gas state signals, and ambient temperature and humidity signals of the target GIS pipe flange; Based on the real-time acquired multi-source signals, inference calculations are performed in conjunction with the optimal state recognition model. The inference calculation results are used as the comprehensive health status assessment results for the GIS pipe flange.

6. The multi-source fusion integrated monitoring method for GIS pipe main flanges according to claim 5, characterized in that, The construction of the first fusion algorithm involves feature extraction and state discrimination of the multi-source information fusion decision framework to obtain the optimal state recognition model, including: Multi-scale entropy is used to extract vibration features, which emphasizes the analysis of the complex changes in vibration signals across multiple time scales, and is expressed as: ; Among them, coarse-grained sequences are set. for: ; Among them, scale factor The coarsening level of the time scale is controlled, ranging from 1 to 10, to cover a variety of dynamic behaviors from instantaneous shocks to long-term trends. For the embedding dimension, a similarity tolerance is set. Set to 0.2 times the standard deviation of the signal, the sample entropy SampEn is used to measure the regularity of the sequence. The lower the entropy value, the more regular the signal, which is suitable for detecting periodic enhancement caused by early loosening.

7. The multi-source fusion integrated monitoring method for GIS pipe main flanges as described in claim 5, characterized in that, The construction of the first fusion algorithm, which performs feature extraction and state discrimination on the multi-source information fusion decision framework to obtain the optimal state recognition model, also includes: Principal component analysis was performed on the temperature data to reduce its dimensionality, which is represented as follows: ; Among them, settings for A temperature data matrix with p rows and p columns, where n is the number of samples and p is the dimension of the temperature-related variables. Matrix W is composed of the eigenvectors corresponding to the first k largest eigenvalues ​​of the covariance matrix, where k is 2 or 3 to retain more than 95% of the information. Projection results. The principal component score after dimensionality reduction significantly reduces computational complexity and eliminates redundancy between variables; Finally, the four types of features are fused and input into the machine learning classifier: ; Among them, settings For the comprehensive state discrimination model, the model parameters It was trained using a large amount of labeled data, and the training objective was to maximize the objective function. Simultaneously satisfying the constraint set The final output model is the optimal state recognition model.

8. A multi-source fusion integrated monitoring system for GIS pipe main flanges, using the method described in any one of claims 1 to 7, characterized in that, include: The first model building module is used to acquire multi-source operation monitoring data of the target GIS pipe flange and establish a first monitoring research model based on the multi-source operation monitoring data. The first monitoring research model is used to construct a state identification objective function and construct several constraints that characterize the normal operation state of the GIS pipe flange based on the collected operation data. The framework establishment module is used to treat the first monitoring and research model as an intelligent agent and establish a multi-source information fusion decision framework based on the intelligent agent. In the multi-source information fusion decision framework, the reward function is an arbitrary function that represents and guides the intelligent agent to accurately identify the status within the normal operating status domain of the target GIS pipe flange. The optimal model building module is used to construct the first fusion algorithm to perform feature extraction and state discrimination on the multi-source information fusion decision framework to obtain the optimal state recognition model. The first fusion algorithm is used to fuse the vibration features extracted by multi-scale entropy and the temperature features extracted by principal component analysis to obtain the optimal state recognition model. The solution module is used to solve the optimal state identification model and obtain the comprehensive health status assessment results of the GIS pipe flange.

9. An electronic device comprising a memory and a processor, wherein the memory stores a computer program, characterized in that, When the processor executes the computer program, it implements the steps of the multi-source fusion GIS pipe flange integrated monitoring method according to any one of claims 1 to 7.

10. A computer-readable storage medium having a computer program stored thereon, characterized in that, When the computer program is executed by the processor, it implements the steps of the multi-source fusion GIS pipe flange integrated monitoring method according to any one of claims 1 to 7.

Images