Intelligent on-line monitoring device for transformer

Abstract

The invention discloses an intelligent online monitoring device for a transformer, and relates to the technical field of transformers. The intelligent on-line monitoring device for the transformer comprises a front-end acquisition layer, a data transmission layer, a data processing and analysis layer, a cloud monitoring and management layer and an emergency response layer, through the deployment of the high-precision sensor network and the edge computing nodes, the system can accurately collect and process the operation data of the transformer in real time, and the monitoring accuracy and efficiency are improved. And secondly, the data transmission layer adopts a self-adaptive communication protocol and a data encryption technology, so that real-time and reliable transmission and security protection of the data are ensured, and meanwhile, traceability and tampering prevention of the data are realized by utilizing a block chain technology. In the data processing and analysis layer, the system realizes intelligent early warning and continuous optimization of transformer data through an integrated learning and deep learning model, and improves the accuracy and reliability of early warning. And the emergency response layer is arranged, so that the system can quickly respond to faults.

Description

Technical Field

[0001] The invention relates to the technical field of transformers, and in particular to an intelligent online monitoring device for transformers. Background Art

[0002] In modern power systems, transformers are indispensable equipment. However, the operating environment of transformers is complex, and they are in a high-load state for a long time. In addition, factors such as equipment aging and improper maintenance lead to frequent transformer failures and even explosion accidents. These accidents not only endanger the lives of on-site workers, but may also cause large-scale power outages, causing serious impacts on social economy and public safety. In order to effectively prevent and reduce the occurrence of such accidents, after in-depth research and technological innovation, a transformer intelligent monitoring system has been launched, aiming to achieve real-time monitoring of the operating status of transformers through advanced technical means to ensure the stability and safety of the power system. Summary of the invention

[0003] In view of the deficiencies in the prior art, the present invention provides an intelligent online monitoring device for a transformer, which solves the safety problems existing in the operation of the existing transformer.

[0004] To achieve the above objectives, the present invention is implemented through the following technical solutions: A transformer intelligent online monitoring device, comprising: The front-end acquisition layer is used to deploy a high-precision sensor network to collect transformer oil temperature, pressure, liquid level and sound data in real time; and receive and preliminarily process the data through edge computing nodes, extract features, and adaptively adjust strategies to improve accuracy and efficiency; The data transmission layer ensures real-time and reliable data transmission through adaptive communication protocols. It also uses data compression to reduce bandwidth usage and uses the AES-256 encryption algorithm to encrypt and protect data. It combines blockchain technology to record key information during data transmission to ensure data traceability and tamper-proofing. The data processing and analysis layer is used for intelligent data cleaning to remove abnormal data, and to perform feature extraction and dimensionality reduction on the remaining data. At the same time, integrated learning and deep learning models are built to analyze transformer data for intelligent early warning, and adaptive learning mechanisms are used to continuously optimize early warning performance. The cloud monitoring and management layer uses a big data platform to store, process and analyze data, supporting real-time query and historical tracing; uses an intelligent operation and maintenance platform to provide intuitive and comprehensive information support and automatically generate operation and maintenance suggestions; and uses an AI-assisted decision-making system to generate the optimal operation and maintenance strategy; The emergency response layer sets up a multi-level early warning mechanism to trigger the corresponding emergency response process; and quickly responds to faults through an intelligent dispatching system, sets up backup power supplies and self-recovery systems to ensure continuous operation of the system; and ensures that data is not lost and the system can be recovered through a remote disaster recovery backup center; The high-precision sensor network is used to deploy multiple types of sensors for oil temperature, pressure, liquid level, and noise on the transformer to collect transformer operation data in real time; The edge computing node is used to receive sensor data, perform preliminary processing and feature extraction, and provide key information for subsequent analysis.

[0005] Preferably, the high-precision sensor specifically includes: Oil temperature sensor: used to collect oil temperature data in real time oil (t), where t represents time; Pressure sensor: monitors the internal pressure of the transformer, and the data is expressed as P int (t); Liquid level sensor: records the transformer insulating oil level signal V(t) to evaluate the transformer insulation status; Noise sensor: collects sound signal A(t) to detect abnormal sound; Preliminary processing: including filtering F(D)=f ilter (D) and denoising Z(D) = z enoise (D), where D represents the received sensor data; Feature extraction: including calculation of oil temperature change trend ; Adaptive learning mechanism: using algorithm A learn (D new ) According to the new data D new Automatically adjust feature extraction strategies to improve the accuracy and efficiency of data processing.

[0006] Preferably, the sensor self-diagnosis and status monitoring functions are introduced into the front-end acquisition layer. The self-diagnosis module monitors the operating status of the sensor in real time, including key parameters such as signal quality and response time. When the sensor status is abnormal, the self-diagnosis module triggers an alarm mechanism, notifies the operation and maintenance personnel through the system interface or SMS / email, and prompts the possible fault type of the sensor and the recommended replacement time. The front-end acquisition layer also formulates a calibration and maintenance plan, and formulates a detailed calibration and maintenance plan according to the characteristics and application scenarios of the sensor, including calibration cycle, calibration method and maintenance priority.

[0007] Preferably, the self-diagnosis module is integrated into the sensor during the sensor design phase, and the corresponding self-diagnosis frequency and alarm threshold are set, specifically including: Self-diagnosis frequency f diag : Define the frequency of the self-diagnosis module to perform status monitoring in seconds or minutes; Signal quality threshold Q thresh: Define the acceptable range of signal quality, expressed as a percentage; if the signal quality collected by the sensor is Q, then Q should satisfy Q≥Q thresh ; Response time threshold T thresh : Define the upper limit of the sensor response time, in milliseconds or seconds; if the sensor response time is T, then T should satisfy T ≤ T thresh ; Alert Threshold thresh :Define the conditions for triggering the alarm mechanism, set as a single parameter exceeding the threshold, that is, if Q< Q thresh or T > T thresh , the alarm mechanism is triggered; or multiple parameters can be set to exceed the threshold at the same time as required, that is, if Q< Q thresh And T > T thresh Among them, the alarm mechanism for the internal pressure of the transformer, in addition to the remote alarm, is also required to automatically open the electromagnetic drain valve installed on the transformer when it is detected that the internal pressure of the transformer exceeds the alarm threshold, and actively release the internal pressure without waiting for maintenance personnel to come and perform pressure relief treatment, so as to avoid explosion due to excessive pressure accumulation.

[0008] Preferably, the calibration and maintenance plan specifically includes: Calibration cycle T cal : Define the time interval for sensor calibration in months or years; Calibration error range E cal : Define the acceptable range of sensor error after calibration, expressed as a percentage or absolute value; let the error of the sensor after calibration be E, then E should satisfy |E| ≤ E cal ; Maintenance priority P: According to the importance and failure rate of the sensor, the sensor is assigned a maintenance priority, namely high, medium and low, represented by P1, P2 and P3 respectively; the priority P is calculated based on the failure rate R of the sensor. fail Importance I imp Factors, using the weighted sum method to calculate the maintenance priority: P = w1 × R fail + w2 × I imp , where w1 and w2 are weight coefficients, and w1+w2=1.

[0009] Preferably, the adaptive communication protocol has multiple communication protocols, and selects the optimal communication protocol according to the network environment to ensure real-time and reliable transmission of transformer data; Data compression and encryption: The transmitted data D is compressed by using compression algorithm C(D) to reduce bandwidth usage; at the same time, the AES-256 encryption algorithm E is applied. AES-256(D) , protect the confidentiality and integrity of transformer data; combined with blockchain technology, B(T trans ) Record key information during data transmission to ensure data traceability and tamper-proofing.

[0010] Preferably, the data processing and analysis layer work content specifically includes: A1. Intelligent data cleaning: using machine learning algorithm M clean (D) Identify and eliminate abnormal or invalid data to ensure the accuracy of analysis results, including: A1.1. Abnormal data identification: Use machine learning algorithms to identify abnormal points in transformer data; A1.2. Invalid data elimination: For the identified abnormal data, the system will eliminate or correct it according to the preset rules; at the same time, the system will record the cause of the abnormal data for subsequent analysis and improvement; Assume that the data collected by the sensor is D={D1, D2, ..., D n}, where D i Represents the data at the i-th moment, and the intelligent data cleaning process is expressed as: D cleaned ={D i |isValid(D i )}; Among them, isValid(D i ) is a Boolean function used to determine whether the data Di is valid; A2. Feature extraction and dimensionality reduction: Extract key features through mathematical methods, reduce data dimensions, and improve processing efficiency, including: A2.1. Feature extraction: Use mathematical methods to extract oil temperature features from raw data. Oil temperature features are specifically expressed as the change trend and fluctuation range of oil temperature. The goal is to extract abnormal oil temperature features from oil temperature features. The original data before extraction is the cleaned data D cleaned , the extracted features are F={f1, f2, ..., f m}, where f j represents the jth feature; A2.2, Dimensionality reduction: Based on feature extraction, dimensionality reduction technology is used to further reduce the data dimension to reduce the amount of calculation and improve processing speed; A3. Intelligent early warning model: Based on the processed data, an integrated learning and deep learning model is built to monitor the transformer status in real time and predict potential faults. The model adopts an adaptive learning mechanism to automatically adjust parameters according to the newly collected transformer data and continuously optimize the early warning performance. At the same time, combined with anomaly detection algorithms, the accuracy and sensitivity of fault identification are improved, including: A3.1. Model construction: Use the processed data to build an early warning model; A3.2. Real-time monitoring and early warning: Input the real-time collected transformer data into the early warning model to obtain the current status and future trend of the transformer; when the model predicts a potential fault, it will trigger the early warning mechanism and send early warning information to the operation and maintenance personnel; Assume that the early warning model is M and the real-time collected data is D real , the warning result is expressed as: Y warning =M(D real ); Among them, Y warning It is a Boolean value or warning level indicating whether there is a potential fault in the current transformer or the severity of the fault.

[0011] Preferably, the cloud monitoring and management layer specifically includes: Big data platform, used to store, process and analyze transformer data, supporting real-time query and historical tracing; using distributed storage system S dist (D) Store transformer data and support real-time query A real-time (D) and historical tracing H trace (D,T past ), where T past Indicates a point in history; Intelligent operation and maintenance platform, used to display transformer status S display (D) Warning information W display (W) and historical trends, providing intuitive and comprehensive information support for operation and maintenance personnel; based on real-time analysis of transformer data, the intelligent operation and maintenance platform automatically generates operation and maintenance suggestions R suggest (D analysis ), operation and maintenance recommendations are directly based on the analysis results of transformer data to ensure the pertinence and effectiveness of operation and maintenance operations; AI-assisted decision making, based on transformer data, using reinforcement learning algorithms R RL (D, H) Generate optimal operation and maintenance strategy, including predictive maintenance plan P pred and operation and maintenance resource scheduling schedule , where H represents historical operation and maintenance data.

[0012] Preferably, the emergency response layer specifically includes: Multi-level early warning mechanism: according to the abnormal situation of transformer data anomaly , set different levels of warning L warning (D anomaly ), triggering the corresponding emergency response process; Intelligent dispatching system: Automatically dispatching operation and maintenance resources R according to early warning information Wdispatch (W), fast response to transformer fault; Backup power supply and self-recovery system: ensure that the system can continue to operate and automatically recover when the main power fails, ensuring the continuity and integrity of transformer data; specifically, S backup (T fail ) During the main power failure time T fail When the backup power supply is started, S backup (T fail ) indicates the backup mechanism in the standby power supply and self-recovery system that automatically starts when the main power supply or main system fails; Remote disaster recovery backup center: using distributed storage S dist-backup (D) and data replication R replicate (D) Technology realizes disaster recovery backup of data, ensuring that transformer data is not lost and the system can be restored in extreme cases.

[0013] Preferably, the high-precision sensor is integrated into one body, including a valve body and a data conversion and processing module installed on the front side thereof, a numerical display meter being installed on the surface of the data conversion and processing module, a threaded cover being rotatably connected to the bottom end of the valve body for threaded connection with the oil filling port of the transformer, a detection tube being connected to the bottom end of the valve body and located inside the threaded cover, a pressure sensor being installed at the bottom end of the detection tube, and an oil temperature sensor being installed at the bottom end of the pressure sensor, a liquid level sensor being installed on the surface of the detection tube and located below the threaded cover, one end of the valve body being connected to an electromagnetic drain valve, and the other end being connected to a noise sensor and a data acquisition terminal through a wire, a hole being opened on the side of the detection tube and separated by two cavities inside, one cavity being connected to the side hole to guide the transformer oil to the electromagnetic drain valve for discharge, and the other cavity being used for passing the wire.

[0014] The present invention provides an intelligent online monitoring device for transformers. Compared with the prior art, it has the following beneficial effects: 1. The intelligent online monitoring device for transformers, through the deployment of high-precision sensor networks and edge computing nodes, the system can collect and process transformer operation data in real time and accurately, improving the accuracy and efficiency of monitoring. Secondly, the data transmission layer adopts adaptive communication protocols and data encryption technology to ensure the real-time, reliable transmission and security protection of data, and uses blockchain technology to achieve data traceability and tamper-proofing. At the data processing and analysis layer, the system realizes intelligent early warning and continuous optimization of transformer data through integrated learning and deep learning models, improving the accuracy and reliability of early warning. Finally, the setting of the emergency response layer enables the system to respond to faults quickly, and ensures that data is not lost and the system can be restored through the off-site disaster recovery backup center, further improving the stability and reliability of the system. These innovative improvements provide a strong guarantee for the safe operation of the transformer; 2. The intelligent online monitoring device for transformers can collect transformer operating data such as oil temperature, pressure, vibration and sound in real time and comprehensively by deploying a high-precision, multi-type sensor network. The introduction of edge computing nodes realizes the preliminary processing and feature extraction of data, significantly improving the accuracy and efficiency of data processing. In particular, the application of adaptive learning mechanism enables the system to automatically adjust the feature extraction strategy according to new data and continuously optimize data processing performance. It provides more accurate and efficient technical means for real-time monitoring and early warning of transformers, ensuring the stability and safety of the power system; 3. The intelligent online monitoring device for transformers introduces sensor self-diagnosis and status monitoring functions in the front-end acquisition layer, monitors the operating status of the sensor in real time, and formulates calibration and maintenance plans, effectively ensuring the accuracy of the data and the long-term stable operation of the sensor. Secondly, the selection of adaptive communication protocols ensures the real-time and reliable transmission of data in different network environments, improving the communication efficiency and stability of the system. In addition, the application of data compression and encryption technology not only reduces bandwidth usage, but also protects the confidentiality and integrity of the data, enhancing the security of the system. Finally, blockchain technology is used to record key information in the data transmission process, which realizes data traceability and tamper-proofing, and further improves the credibility of the data and the reliability of the system. The above content provides strong support for the intelligent, efficient and safe operation of the transformer monitoring system; 4. The intelligent online monitoring device for the transformer and intelligent data cleaning use machine learning algorithms to effectively identify and eliminate abnormal or invalid data, ensuring the accuracy of the analysis results. At the same time, it records the causes of abnormal data, providing valuable information for subsequent analysis and improvement. Secondly, feature extraction and dimensionality reduction technology uses mathematical methods to extract key features and reduce data dimensions, which not only improves processing efficiency, but also enhances the model's sensitivity and accuracy to the transformer status. Finally, the intelligent early warning model uses integrated learning and deep learning algorithms, combined with adaptive learning mechanisms and anomaly detection algorithms, to achieve real-time monitoring of the transformer status and accurate prediction of potential faults, greatly improving the accuracy and sensitivity of the early warning, and providing strong support for the intelligence, efficiency and precision of the transformer monitoring system, effectively improving the safety and reliability of transformer operation; 5. The intelligent online monitoring device for transformers realizes real-time query and historical tracing of transformer data through the big data platform, providing strong data support for operation and maintenance. The introduction of the intelligent operation and maintenance platform not only intuitively displays the transformer status and warning information, but also automatically generates targeted operation and maintenance suggestions, improving the efficiency and accuracy of operation and maintenance. In particular, the application of AI-assisted decision-making system generates the optimal operation and maintenance strategy based on reinforcement learning algorithm, further improving the intelligence level and decision-making efficiency of operation and maintenance, and providing strong guarantee for the safe and stable operation of transformers; 6. The intelligent online monitoring device for transformers can trigger the emergency response process more accurately and improve the timeliness of fault handling through a multi-level early warning mechanism. The introduction of the intelligent dispatching system realizes the rapid and automatic dispatching of operation and maintenance resources, further shortening the fault response time. The setting of the backup power supply and self-recovery system and the off-site disaster recovery backup center ensures the continuous operation and data security of the system in the event of a main power failure or extreme conditions, effectively improving the reliability and stability of the transformer monitoring system and providing a strong guarantee for the safe operation of the power system. BRIEF DESCRIPTION OF THE DRAWINGS

[0015] Figure 1 It is a system module block diagram of the present invention.

[0016] Figure 2 It is a schematic diagram of the main process of the present invention.

[0017] Figure 3 It is a schematic diagram of the system flow of data collection and early warning analysis of the present invention.

[0018] Figure 4 It is a schematic diagram of the structure of the high-precision sensor of the present invention.

[0019] In the figure, there are valve body 1, data conversion and processing module 2, numerical display meter 3, threaded cover 4, detection tube 5, pressure sensor 6, oil temperature sensor 7, liquid level sensor 8, electromagnetic drain valve 9, noise sensor 10, and data acquisition terminal 11. DETAILED DESCRIPTION

[0020] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0021] See also Figure 1-Figure 4 , the present invention provides six technical solutions: The first embodiment: a transformer intelligent online monitoring device, comprising: The front-end acquisition layer is used to deploy a high-precision sensor network to collect transformer oil temperature, pressure, liquid level and sound data in real time; and receive and preliminarily process the data through edge computing nodes, extract features, and adaptively adjust strategies to improve accuracy and efficiency; The data transmission layer ensures real-time and reliable data transmission through adaptive communication protocols. It also uses data compression to reduce bandwidth usage and uses the AES-256 encryption algorithm to encrypt and protect data. It combines blockchain technology to record key information during data transmission to ensure data traceability and tamper-proofing. The data processing and analysis layer is used for intelligent data cleaning to remove abnormal data, and to perform feature extraction and dimensionality reduction on the remaining data. At the same time, integrated learning and deep learning models are built to analyze transformer data for intelligent early warning, and adaptive learning mechanisms are used to continuously optimize early warning performance. The cloud monitoring and management layer uses a big data platform to store, process and analyze data, supporting real-time query and historical tracing; uses an intelligent operation and maintenance platform to provide intuitive and comprehensive information support and automatically generate operation and maintenance suggestions; and uses an AI-assisted decision-making system to generate the optimal operation and maintenance strategy; The emergency response layer sets up a multi-level early warning mechanism to trigger the corresponding emergency response process; it also responds to faults quickly through an intelligent scheduling system, and sets up a backup power supply and self-recovery system to ensure continuous operation of the system; and it ensures that data is not lost and the system can be recovered through an off-site disaster recovery backup center.

[0022] Through the deployment of high-precision sensor networks and edge computing nodes, the system can collect and process transformer operation data in real time and accurately, improving the accuracy and efficiency of monitoring. Secondly, the data transmission layer adopts adaptive communication protocols and data encryption technology to ensure real-time, reliable transmission and security protection of data, and uses blockchain technology to achieve data traceability and tamper-proof. In the data processing and analysis layer, the system realizes intelligent early warning and continuous optimization of transformer data through integrated learning and deep learning models, improving the accuracy and reliability of early warning. Finally, the setting of the emergency response layer enables the system to respond to faults quickly, and ensures that data is not lost and the system can be restored through the off-site disaster recovery backup center, further improving the stability and reliability of the system. These innovative improvements provide a strong guarantee for the safe operation of the transformer.

[0023] The second implementation mode is mainly different from the first implementation mode in that the high-precision sensor network is used to deploy multiple types of sensors for oil temperature, pressure, liquid level, and noise on the transformer to collect transformer operation data in real time; the high-precision sensors specifically include: Oil temperature sensor 7: used to collect oil temperature data in real time oil (t), where t represents time; Pressure sensor 6: monitors the internal pressure of the transformer, and the data is expressed as P int (t); Liquid level sensor 8: records transformer insulating oil liquid level signal V(t) for evaluating transformer insulation status; Noise sensor 10: collects sound signals A(t) and is used to detect abnormal sounds, specifically vibration noises, and is installed on the transformer housing; A high-precision sensor is integrated into one, including a valve body 1 and a data conversion and processing module 2 installed on its front side. A numerical display meter 3 is installed on the surface of the data conversion and processing module 2. The bottom end of the valve body 1 is rotatably connected to a threaded cover 4 for threaded connection with the oil filling port of the transformer. The bottom end of the valve body 1 and the inside of the threaded cover 4 are also connected to a detection tube 5. A pressure sensor 6 is installed at the bottom end of the detection tube 5, and an oil temperature sensor 7 is installed at the bottom end of the pressure sensor 6. A liquid level sensor 8 is installed on the surface of the detection tube 5 and below the threaded cover. One end of the valve body 1 is connected to an electromagnetic drain valve 9, and the other end is connected to a noise sensor 10 and a data acquisition terminal 11 through a wire. The detection tube 5 has a hole on the side and is separated by two cavities inside. One cavity is connected to the side hole to guide the transformer oil to the electromagnetic drain valve 9 for discharge, and the other cavity is used to pass the wire.

[0024] Taking into account the diversity and complexity of the on-site environment, the device supports both wired and wireless data transmission modes. Users can flexibly choose according to actual needs and on-site conditions. Whether through a stable wired connection or a convenient wireless network, the stability and reliability of data transmission can be guaranteed, ensuring that operation and maintenance personnel can grasp the operating status of the transformer in real time; and the high-precision sensor network uses a built-in 4G module to enable the device to be seamlessly connected to the cloud platform to achieve real-time data upload and remote monitoring. Users can check the operating status of the transformer anytime and anywhere through the mobile phone APP or computer-side operation and maintenance platform, receive alarm information, and perform remote configuration and control. This intelligent management method greatly improves operation and maintenance efficiency and reduces labor costs; The edge computing node is used to receive sensor data, perform preliminary processing and feature extraction, and provide key information for subsequent analysis; Preliminary processing: including filtering F(D)=f ilter (D) and denoising Z(D) = z enoise (D), where D represents the received sensor data; Feature extraction: including calculation of oil temperature change trend ; Adaptive learning mechanism: using algorithm A learn (D new ) According to the new data D new Automatically adjust feature extraction strategies to improve the accuracy and efficiency of data processing.

[0025] By deploying a high-precision, multi-type sensor network, transformer operation data such as oil temperature, pressure, vibration and sound can be collected in real time and comprehensively. The introduction of edge computing nodes enables preliminary data processing and feature extraction, significantly improving the accuracy and efficiency of data processing. In particular, the application of adaptive learning mechanisms enables the system to automatically adjust feature extraction strategies based on new data and continuously optimize data processing performance. It provides more accurate and efficient technical means for real-time monitoring and early warning of transformers, ensuring the stability and safety of the power system.

[0026] The third implementation mode is mainly different from the first implementation mode in that: the sensor self-diagnosis and status monitoring function is introduced into the front-end acquisition layer, and the self-diagnosis module monitors the operating status of the sensor in real time, including key parameters such as signal quality and response time; when the sensor status is abnormal, the self-diagnosis module triggers an alarm mechanism, notifies the operation and maintenance personnel through the system interface or SMS / email, and prompts the possible fault type of the sensor and the recommended replacement time; The front-end acquisition layer also formulates a calibration and maintenance plan, and formulates a detailed calibration and maintenance plan according to the characteristics and application scenarios of the sensor, including calibration cycle, calibration method and maintenance priority; The self-diagnosis module is integrated into the sensor during the sensor design phase, and the corresponding self-diagnosis frequency and alarm threshold are set, including: Self-diagnosis frequency f diag : Defines the frequency of the self-diagnostic module to perform status monitoring in seconds or minutes; for example, f diag = 60 min means self-diagnosis is performed every 60 minutes; Signal quality threshold Q thresh : Defines the acceptable range of signal quality, expressed as a percentage; for example, Q thresh =90% means that the signal quality must be maintained above 90%; if the signal quality collected by the sensor is Q, then Q should satisfy Q≥Q thresh ; Response time threshold T thresh : Defines the upper limit of the sensor response time in milliseconds or seconds; for example, T thresh = 100 ms means the sensor response time must not exceed 100 milliseconds; if the sensor response time is T, then T should satisfy T≤ T thresh ; Alert Threshold thresh :Define the conditions for triggering the alarm mechanism, set as a single parameter exceeding the threshold, that is, if Q< Q thresh or T > T thresh , the alarm mechanism is triggered; or multiple parameters can be set to exceed the threshold at the same time as required, that is, if Q< Q threshAnd T > T thresh Among them, for the alarm mechanism of the internal pressure of the transformer, in addition to the remote alarm, it is also necessary to automatically open the electromagnetic drain valve 9 installed on the transformer when it is detected that the internal pressure of the transformer exceeds the alarm threshold, and actively release the internal pressure without waiting for maintenance personnel to come and perform pressure relief treatment, so as to avoid explosion due to excessive pressure accumulation; The calibration and maintenance plan includes: Calibration cycle T cal : Defines the sensor calibration interval in months or years; for example, T cal = 6 month means calibration is performed every 6 months; Calibration error range E cal : Defines the acceptable range of sensor error after calibration, expressed as a percentage or absolute value; for example, E cal = ±1% means that the sensor error after calibration should be within ±1%; let the error of the sensor after calibration be E, then E should satisfy |E| ≤ E cal ; Maintenance priority P: According to the importance and failure rate of the sensor, the sensor is assigned a maintenance priority, namely high, medium and low, represented by P1, P2 and P3 respectively; the priority P is calculated based on the failure rate R of the sensor. fail Importance I imp Factors, using the weighted sum method to calculate the maintenance priority: P = w1 × R fail + w2 × I imp , where w1 and w2 are weight coefficients, and w1+w2=1.

[0027] Adaptive communication protocol has multiple communication protocols and selects the optimal communication protocol according to the network environment to ensure real-time and reliable transmission of transformer data; Data compression and encryption: The transmitted data D is compressed by using compression algorithm C(D) to reduce bandwidth usage; at the same time, the AES-256 encryption algorithm E is applied. AES-256 (D) , protect the confidentiality and integrity of transformer data; combined with blockchain technology, B(T trans ) Record key information during data transmission to ensure data traceability and tamper-proofing.

[0028] The front-end acquisition layer introduces sensor self-diagnosis and status monitoring functions, monitors the operating status of the sensor in real time, formulates calibration and maintenance plans, and effectively ensures the accuracy of the data and the long-term stable operation of the sensor. Secondly, the selection of adaptive communication protocols ensures the real-time and reliable transmission of data in different network environments, and improves the communication efficiency and stability of the system. In addition, the application of data compression and encryption technology not only reduces bandwidth usage, but also protects the confidentiality and integrity of the data, and enhances the security of the system. Finally, the key information in the data transmission process is recorded in combination with blockchain technology, which realizes data traceability and tamper-proofing, and further improves the credibility of the data and the reliability of the system. The above content provides strong support for the intelligent, efficient and safe operation of the transformer monitoring system.

[0029] The fourth implementation mode is mainly different from the first implementation mode in that the work contents of the data processing and analysis layer specifically include: A1. Intelligent data cleaning: using machine learning algorithm M clean (D) Identify and eliminate abnormal or invalid data to ensure the accuracy of analysis results, including: A1.1. Abnormal data identification: Use machine learning algorithms (such as density-based clustering algorithm DBSCAN, Isolation Forest, etc.) to identify abnormal points in transformer data; these abnormal points may be caused by sensor failure, data transmission errors, or abnormalities of the transformer itself; A1.2. Invalid data elimination: For the identified abnormal data, the system will eliminate or correct it according to the preset rules (such as data range, data change trend, etc.); at the same time, the system will record the cause of the abnormal data for subsequent analysis and improvement; Assume that the data collected by the sensor is D={D1, D2, ..., D n}, where D i Represents the data at the i-th moment, and the intelligent data cleaning process is expressed as: D cleaned ={D i |isValid(D i )}; Among them, isValid(D i ) is a Boolean function used to determine whether the data Di is valid; A2. Feature extraction and dimensionality reduction: Extract key features through mathematical methods, reduce data dimensions, and improve processing efficiency, including: A2.1. Feature extraction: Use mathematical methods (such as principal component analysis (PCA) and linear discriminant analysis (LDA)) to extract oil temperature features from the original data. The oil temperature features are specifically expressed as the change trend and fluctuation range of the oil temperature. The goal is to extract the abnormal oil temperature features from the oil temperature features. The original data before extraction is the cleaned data D cleaned , the extracted features are F={f1, f2, ..., f m}, where f j represents the jth feature; A2.2. Dimensionality reduction: Based on feature extraction, dimensionality reduction techniques (such as PCA, t-SNE, etc.) are used to further reduce data dimensions to reduce the amount of calculation and improve processing speed; A3. Intelligent early warning model: Based on the processed data, we build integrated learning and deep learning models to monitor the transformer status in real time and predict potential faults. The model uses an adaptive learning mechanism to automatically adjust parameters according to the newly collected transformer data and continuously optimize the early warning performance. At the same time, we combine anomaly detection algorithms such as Isolation Forest and One-Class SVM to improve the accuracy and sensitivity of fault identification, including: A3.1. Model construction: Use the processed data (including cleaned data and feature data after dimensionality reduction) to build an early warning model; the early warning model is built using random forests based on ensemble learning, gradient boosting trees, neural networks based on deep learning, recurrent neural networks (such as LSTM), etc. A3.2. Real-time monitoring and early warning: Input the real-time collected transformer data into the early warning model to obtain the current status and future trend of the transformer; when the model predicts a potential fault, it will trigger the early warning mechanism and send early warning information to the operation and maintenance personnel; Assume that the early warning model is M and the real-time collected data is D real , the warning result is expressed as: Y warning =M(D real ); Among them, Y warning It is a Boolean value or warning level indicating whether there is a potential fault in the current transformer or the severity of the fault.

[0030] Intelligent data cleaning uses machine learning algorithms to effectively identify and remove abnormal or invalid data, ensuring the accuracy of the analysis results. At the same time, it records the causes of abnormal data and provides valuable information for subsequent analysis and improvement. Secondly, feature extraction and dimensionality reduction technology uses mathematical methods to extract key features and reduce data dimensions, which not only improves processing efficiency, but also enhances the model's sensitivity and accuracy to transformer status. Finally, the intelligent early warning model uses integrated learning and deep learning algorithms, combined with adaptive learning mechanisms and anomaly detection algorithms, to achieve real-time monitoring of transformer status and accurate prediction of potential faults, greatly improving the accuracy and sensitivity of early warnings, providing strong support for the intelligence, efficiency and precision of transformer monitoring systems, and effectively improving the safety and reliability of transformer operation.

[0031] The fifth implementation method is different from the first implementation method in that the cloud monitoring and management layer includes: Big data platform, used to store, process and analyze transformer data, supporting real-time query and historical tracing; using distributed storage system S dist (D) Store transformer data and support real-time query A real-time (D) and historical tracing H trace (D,T past ), where T past Indicates a point in history; Intelligent operation and maintenance platform, used to display transformer status S display (D) Warning information W display (W) and historical trends, providing intuitive and comprehensive information support for operation and maintenance personnel; based on real-time analysis of transformer data, the intelligent operation and maintenance platform automatically generates operation and maintenance suggestions R suggest (D analysis ), such as replacing parts, adjusting operating parameters, etc., operation and maintenance suggestions are directly based on the analysis results of transformer data to ensure the pertinence and effectiveness of operation and maintenance operations; the platform provides an intuitive interface where users can view real-time alarm information, historical records and equipment status. In addition, the operation and maintenance platform also supports remote configuration and control functions, making operation and maintenance work more efficient and convenient; AI-assisted decision making, based on transformer data, using reinforcement learning algorithms R RL (D, H) Generate optimal operation and maintenance strategy, including predictive maintenance plan P pred and operation and maintenance resource scheduling schedule , where H represents historical operation and maintenance data; the AI-assisted decision-making system generates the optimal operation and maintenance strategy based on historical data and current status to improve operation and maintenance efficiency and reliability.

[0032] Through the big data platform, real-time query and historical tracing of transformer data are realized, providing strong data support for operation and maintenance. The introduction of the intelligent operation and maintenance platform not only intuitively displays the transformer status and warning information, but also automatically generates targeted operation and maintenance suggestions, improving the efficiency and accuracy of operation and maintenance. In particular, the application of the AI-assisted decision-making system generates the optimal operation and maintenance strategy based on the reinforcement learning algorithm, further improving the intelligence level and decision-making efficiency of operation and maintenance, and providing a strong guarantee for the safe and stable operation of the transformer.

[0033] The sixth implementation method is mainly different from the first implementation method in that the specific contents of the emergency response layer include: Multi-level early warning mechanism: according to the abnormal situation of transformer data anomaly , set different levels of warning L warning (D anomaly ), triggering the corresponding emergency response process; Intelligent dispatching system: Automatically dispatching operation and maintenance resources R according to early warning information W dispatch (W), fast response to transformer fault; Backup power supply and self-recovery system: ensure that the system can continue to operate and automatically recover when the main power fails, ensuring the continuity and integrity of transformer data; specifically, S backup (T fail ) During the main power failure time T fail When the backup power supply is started, S backup (T fail ) indicates the backup mechanism in the standby power supply and self-recovery system that automatically starts when the main power supply or main system fails; Remote disaster recovery backup center: using distributed storage S dist-backup (D) and data replication R replicate (D) Technology realizes disaster recovery backup of data, ensuring that transformer data is not lost and the system can be restored in extreme cases.

[0034] Through the multi-level early warning mechanism, the emergency response process can be triggered more accurately, improving the timeliness of fault handling. The introduction of the intelligent dispatching system realizes the rapid and automatic dispatching of operation and maintenance resources, further shortening the fault response time. The setting of the backup power supply and self-recovery system and the remote disaster recovery backup center ensures the continuous operation and data security of the system in the event of a main power failure or extreme conditions, effectively improving the reliability and stability of the transformer monitoring system, and providing a strong guarantee for the safe operation of the power system.

[0035] Meanwhile, the contents not described in detail in this specification belong to the prior art known to those skilled in the art, and the model parameters of each electrical appliance are not specifically limited, and conventional equipment can be used.

[0036] It should be noted that, in this article, relational terms such as first and second, etc. are only used to distinguish one entity or operation from another entity or operation, and do not necessarily require or imply any such actual relationship or order between these entities or operations. Moreover, the terms "include", "comprise" or any other variants thereof are intended to cover non-exclusive inclusion, so that a process, method, article or device including a series of elements includes not only those elements, but also other elements not explicitly listed, or also includes elements inherent to such process, method, article or device.

[0037] Although embodiments of the present invention have been shown and described, it will be appreciated by those skilled in the art that various changes, modifications, substitutions and variations may be made to the embodiments without departing from the principles and spirit of the present invention, and that the scope of the present invention is defined by the appended claims and their equivalents.

Claims

1. A transformer intelligent online monitoring device, characterized in that: include: The front-end acquisition layer is used to deploy a high-precision sensor network to collect transformer oil temperature, pressure, liquid level and sound data in real time; The edge computing nodes receive and preliminarily process the data, extract features, and adaptively adjust strategies to improve accuracy and efficiency; The data transmission layer ensures real-time and reliable data transmission through adaptive communication protocols. It also uses data compression to reduce bandwidth usage and uses the AES-256 encryption algorithm to encrypt and protect data. It combines blockchain technology to record key information during data transmission to ensure data traceability and tamper-proofing. The data processing and analysis layer is used for intelligent data cleaning to remove abnormal data, and to perform feature extraction and dimensionality reduction on the remaining data. At the same time, integrated learning and deep learning models are built to analyze transformer data for intelligent early warning, and adaptive learning mechanisms are used to continuously optimize early warning performance. The cloud monitoring and management layer uses a big data platform to store, process and analyze data, supporting real-time query and historical tracing; uses an intelligent operation and maintenance platform to provide intuitive and comprehensive information support and automatically generate operation and maintenance suggestions; and uses an AI-assisted decision-making system to generate the optimal operation and maintenance strategy; Emergency response layer, set up a multi-level early warning mechanism to trigger the corresponding emergency response process; The intelligent dispatching system can quickly respond to failures, and a backup power supply and self-recovery system can be set up to ensure continuous operation of the system; an off-site disaster recovery backup center can be used to ensure that data is not lost and the system can be recovered.

2. The intelligent online monitoring device for transformer according to claim 1, characterized in that: The high-precision sensor network is used to deploy multiple types of sensors for oil temperature, pressure, liquid level, and noise on the transformer to collect transformer operation data in real time; The edge computing node is used to receive sensor data, perform preliminary processing and feature extraction, and provide key information for subsequent analysis; High-precision sensors include: Oil temperature sensor: used to collect oil temperature data in real time oil (t), where t represents time; Pressure sensor: monitors the internal pressure of the transformer, and the data is expressed as P int (t); Liquid level sensor: records the transformer insulating oil level signal V(t) to evaluate the transformer insulation status; Noise sensor: collects sound signal A(t) to detect abnormal sound; Preliminary processing: including filtering F(D)=f ilter (D) and denoising Z(D) = z enoise (D), where D represents the received sensor data; Feature extraction: including calculation of oil temperature change trend ; Adaptive learning mechanism: using algorithm A learn (D new ) According to the new data D new Automatically adjust feature extraction strategies to improve the accuracy and efficiency of data processing.

3. The intelligent online monitoring device for transformer according to claim 1 is characterized in that: The sensor self-diagnosis and status monitoring functions are introduced into the front-end acquisition layer. The self-diagnosis module monitors the operating status of the sensor in real time, including key parameters such as signal quality and response time. When the sensor status is abnormal, the self-diagnosis module triggers the alarm mechanism, notifies the operation and maintenance personnel through the system interface or SMS / email, and prompts the possible fault type of the sensor and the recommended replacement time. The front-end acquisition layer also formulates a calibration and maintenance plan, and formulates a detailed calibration and maintenance plan according to the characteristics and application scenarios of the sensor, including calibration cycle, calibration method and maintenance priority.

4. The intelligent online monitoring device for transformer according to claim 3 is characterized in that: The self-diagnosis module is integrated into the sensor during the sensor design phase, and the corresponding self-diagnosis frequency and alarm threshold are set, including: Self-diagnosis frequency f diag : Define the frequency of the self-diagnosis module to perform status monitoring in seconds or minutes; Signal quality threshold Q thresh : Define the acceptable range of signal quality, expressed as a percentage; if the signal quality collected by the sensor is Q, then Q should satisfy Q≥Q thresh ; Response time threshold T thresh : Define the upper limit of the sensor response time, in milliseconds or seconds; if the sensor response time is T, then T should satisfy T ≤ T thresh ; Alert Threshold thresh :Define the conditions for triggering the alarm mechanism, set as a single parameter exceeding the threshold, that is, if Q thresh or T > T thresh , the alarm mechanism is triggered; or multiple parameters can be set to exceed the threshold at the same time as required, that is, if Q thresh And T > T thresh Among them, the alarm mechanism for the internal pressure of the transformer, in addition to the remote alarm, is also required to automatically open the electromagnetic drain valve installed on the transformer when it is detected that the internal pressure of the transformer exceeds the alarm threshold, so as to actively release the internal pressure.​​ 5. The intelligent online monitoring device for transformer according to claim 3 is characterized in that: The calibration and maintenance plan includes: Calibration cycle T cal : Define the time interval for sensor calibration in months or years; Calibration error range E cal : Define the acceptable range of sensor error after calibration, expressed as a percentage or absolute value; let the error of the sensor after calibration be E, then E should satisfy |E| ≤ E cal ; Maintenance priority P: According to the importance and failure rate of the sensor, the sensor is assigned a maintenance priority, namely high, medium and low, represented by P1, P2 and P3 respectively; the priority P is calculated based on the failure rate R of the sensor. fail Importance I imp Factors, using the weighted sum method to calculate the maintenance priority: P = w1 × R fail + w2 × I imp , where w1 and w2 are weight coefficients, and w1+w2=1.

6. The intelligent online monitoring device for transformer according to claim 1, characterized in that: The adaptive communication protocol has multiple communication protocols and selects the optimal communication protocol according to the network environment to ensure real-time and reliable transmission of transformer data; Data compression and encryption: The transmitted data D is compressed by using compression algorithm C(D) to reduce bandwidth usage; at the same time, the AES-256 encryption algorithm E is applied. AES-256 (D) , protect the confidentiality and integrity of transformer data; combined with blockchain technology, B(T trans ) Record key information during data transmission to ensure data traceability and tamper-proofing.

7. The intelligent online monitoring device for transformer according to claim 1, characterized in that: The data processing and analysis layer work specifically includes: A1. Intelligent data cleaning: using machine learning algorithm M clean (D) Identify and eliminate abnormal or invalid data to ensure the accuracy of analysis results, including: A1.

1. Abnormal data identification: Use machine learning algorithms to identify abnormal points in transformer data; A1.

2. Invalid data elimination: For the identified abnormal data, the system will eliminate or correct it according to the preset rules; at the same time, the system will record the cause of the abnormal data for subsequent analysis and improvement; Assume that the data collected by the sensor is D={D1, D2, ..., D n }, where D i Represents the data at the i-th moment, and the intelligent data cleaning process is expressed as: D cleaned ={D i ∣isValid(D i )}; Among them, isValid(D i ) is a Boolean function used to determine whether the data Di is valid; A2. Feature extraction and dimensionality reduction: Extract key features through mathematical methods, reduce data dimensions, and improve processing efficiency, including: A2.

1. Feature extraction: Use mathematical methods to extract oil temperature features from raw data. Oil temperature features are specifically expressed as the change trend and fluctuation range of oil temperature. The goal is to extract abnormal oil temperature features from oil temperature features. The original data before extraction is the cleaned data D cleaned , the extracted features are F={f1, f2, ..., f m }, where f j represents the jth feature; A2.2, Dimensionality reduction: Based on feature extraction, dimensionality reduction technology is used to further reduce the data dimension to reduce the amount of calculation and improve processing speed; A3. Intelligent early warning model: Based on the processed data, an integrated learning and deep learning model is built to monitor the transformer status in real time and predict potential faults. The model adopts an adaptive learning mechanism to automatically adjust parameters according to the newly collected transformer data and continuously optimize the early warning performance. At the same time, combined with anomaly detection algorithms, the accuracy and sensitivity of fault identification are improved, including: A3.

1. Model construction: Use the processed data to build an early warning model; A3.

2. Real-time monitoring and early warning: Input the real-time collected transformer data into the early warning model to obtain the current status and future trend of the transformer; when the model predicts a potential fault, it will trigger the early warning mechanism and send early warning information to the operation and maintenance personnel; Assume that the early warning model is M and the real-time collected data is D real , the warning result is expressed as: Y warning =M(D real ); Among them, Y warning It is a Boolean value or warning level indicating whether there is a potential fault in the current transformer or the severity of the fault.

8. The intelligent online monitoring device for transformer according to claim 1, characterized in that: The specific contents of the cloud monitoring and management layer include: Big data platform, used to store, process and analyze transformer data, supporting real-time query and historical tracing; using distributed storage system S dist (D) Store transformer data and support real-time query A real-time (D) and historical tracing H trace (D,T past ), where T past Indicates a point in history; Intelligent operation and maintenance platform, used to display transformer status S display (D) Warning information W display (W) and historical trends, providing intuitive and comprehensive information support for operation and maintenance personnel; based on real-time analysis of transformer data, the intelligent operation and maintenance platform automatically generates operation and maintenance suggestions R suggest (D analysis ), operation and maintenance recommendations are directly based on the analysis results of transformer data to ensure the pertinence and effectiveness of operation and maintenance operations; AI-assisted decision making, based on transformer data, using reinforcement learning algorithms R RL (D, H) Generate optimal operation and maintenance strategy, including predictive maintenance plan P pred and operation and maintenance resource scheduling schedule , where H represents historical operation and maintenance data.

9. The intelligent online monitoring device for transformer according to claim 1, characterized in that: The specific contents of the emergency response layer include: Multi-level early warning mechanism: according to the abnormal situation of transformer data anomaly , set different levels of warning L warning (D anomaly ), triggering the corresponding emergency response process; Intelligent dispatching system: Automatically dispatching operation and maintenance resources R according to early warning information W dispatch (W), fast response to transformer fault; Backup power supply and self-recovery system: ensure that the system can continue to operate and automatically recover when the main power fails, ensuring the continuity and integrity of transformer data; specifically, S backup (T fail ) During the main power failure time T fail When the backup power supply is started, S backup (T fail ) indicates the backup mechanism in the standby power supply and self-recovery system that automatically starts when the main power supply or main system fails; Remote disaster recovery backup center: using distributed storage S dist-backup (D) and data replication R replicate (D) Technology realizes disaster recovery backup of data, ensuring that transformer data is not lost and the system can be restored in extreme cases.

10. The intelligent online monitoring device for transformer according to claim 2, characterized in that: A high-precision sensor is integrated into one, including a valve body and a data conversion and processing module installed on the front side thereof, a numerical display meter is installed on the surface of the data conversion and processing module, a threaded cover is rotatably connected to the bottom end of the valve body for threaded connection with the oil filling port of the transformer, a detection tube is connected to the bottom end of the valve body and located inside the threaded cover, a pressure sensor is installed at the bottom end of the detection tube, and an oil temperature sensor is installed at the bottom end of the pressure sensor, a liquid level sensor is installed on the surface of the detection tube and located below the threaded cover, one end of the valve body is connected to an electromagnetic drain valve, and the other end is connected to a noise sensor and a data acquisition terminal through a wire, a hole is opened on the side of the detection tube and two cavities are separated inside, one cavity is connected to the side hole to guide the transformer oil to the electromagnetic drain valve for discharge, and the other cavity is used for passing the wire.

Images