An artificial intelligence-based on-line data testing method for a mutual inductor

Abstract

The application discloses an online data testing method for a mutual inductor based on artificial intelligence and particularly relates to the technical field of mutual inductor data testing, which comprises the following steps: synchronously collecting primary side data, secondary side data, standard device reference signals and working condition environment data in an operating field and encapsulating the data into a structured data package with a unified time scale; deploying digital twin agents on a data sending end and a data receiving end respectively, transmitting monitoring signaling through an independent safety channel, and realizing virtual and real synchronous verification of data integrity and service rationality; forming a structured abstract from verification results and key features, storing the abstract in a block chain network, and constructing an unalterable testing process log to realize accurate tracing. The application realizes real-time and highly credible online testing of a mutual inductor under non-power-off conditions, has full-link traceability and intelligent early warning capability, and significantly improves testing efficiency and intelligent operation and maintenance level.

Description

Technical Field

[0001] This invention relates to the field of current transformer data testing technology, and more specifically, to an artificial intelligence-based online data testing method for current transformers. Background Technology

[0002] Current transformers and voltage transformers are key sensing devices for metering, protection, and control in power systems. Their measurement accuracy and operational reliability directly affect the safety, stability, and economic operation of the power grid. Traditionally, the performance testing and accuracy calibration of transformers have relied heavily on offline verification after periodic power outages. This method is not only cumbersome and time-consuming, but also forces equipment out of operation, leading to a surge in power generation losses and maintenance costs. Especially for ultra-high voltage (UHV) power stations and distributed renewable energy plants, frequent power outage verification is no longer sufficient to meet the stringent requirements of real-time and reliable metering in modern power systems, becoming a prominent bottleneck restricting maintenance efficiency and metering reliability.

[0003] To reduce power outages, the industry has attempted to introduce online monitoring technology, but its development faces multiple challenges. First, most solutions only provide simple alarms for exceeding electrical parameter limits, lacking the ability to calculate and trace transformer errors (such as ratio difference and phase difference) in real time. Second, test data is susceptible to interference or tampering during remote transmission, and issues such as inconsistent interfaces and excessive latency exist during end-to-cloud collaborative verification, leading to "data silos" and questionable reliability. Third, existing methods fail to effectively integrate multi-physical field operating condition data such as temperature, vibration, and load, making it impossible to establish a dynamic model reflecting the true health status of equipment. Consequently, they lack the ability to predict and warn of latent performance degradation (such as error drift caused by extreme temperature differences and dust erosion), and the operation and maintenance model remains at the "post-event response" level.

[0004] With the penetration of next-generation information technologies such as digital twins, artificial intelligence, and blockchain into the energy and power sector, building an intelligent metering and testing system that is "perceptible, analyzable, decision-making-capable, and traceable" has become a clear trend. However, how to deeply integrate these technologies with online testing of instrument transformers to achieve highly reliable acquisition and transmission of test data, full-link traceability and evidence storage, and early fault warning based on multi-dimensional data-driven intelligence remains a technological gap that urgently needs to be overcome. Summary of the Invention

[0005] To overcome the aforementioned deficiencies of the prior art, embodiments of the present invention provide an online data testing method for current transformers based on artificial intelligence.

[0006] To achieve the above objectives, the present invention provides the following technical solution:

[0007] An online data testing method for current transformers based on artificial intelligence includes the following steps:

[0008] S1. Test data preprocessing: At the current transformer operating site, the primary side signal, secondary side output signal, standard reference signal, and operating condition and environmental data are collected synchronously and packaged into a structured test data package with a unified time scale.

[0009] S2. Simultaneous Monitoring of Virtual and Real Data: Digital twin agents are deployed at both the sending and receiving ends of the test data packets. The digital twin agent at the sending end extracts key features from the test data packets and generates monitoring signaling, which is then sent to the receiving end through an independent secure channel. Upon receiving the test data packets, the digital twin agent at the receiving end uses the monitoring signaling to verify data integrity and business rationality.

[0010] S3. Blockchain Evidence Storage and Traceability: The summary information containing the verification results and key features is structured and stored in the blockchain network to form an immutable test process log; when data anomalies are detected, traceability analysis is performed based on the records in the blockchain network.

[0011] S4. Intelligent Anomaly Detection: Based on historical normal test data, a dynamic baseline model reflecting the health status of the transformer is established using machine learning algorithms; by comparing the deviation between real-time test data and the dynamic baseline model, performance degradation trends are identified and early warnings are triggered.

[0012] Specifically, in S1:

[0013] The collected data is synchronously acquired at a sampling rate of no less than 12800Hz and encapsulated according to a preset standard protocol;

[0014] Operating condition and environmental data should include at least the transformer body temperature, the effective value of three-dimensional vibration acceleration, and the load rate expressed as a percentage.

[0015] Specifically, in S2:

[0016] The digital twin agent is a software-defined monitoring agent that embeds a business layer semantic understanding module. It is used to parse electrical performance parameters and environmental state quantities in test data packets, simulate preliminary error calculation logic at the sending end, and simulate a high-fidelity virtual transformer at the receiving end to output theoretical secondary side signals.

[0017] Specifically, in S2:

[0018] The key features extracted include: the true RMS values ​​of the primary and secondary currents, the fundamental phase difference, the preliminary values ​​of the ratio difference and angle difference calculated in real time based on the reference signal of the standard, the total harmonic distortion rate, and the specific harmonic content rate.

[0019] Specifically, in S2:

[0020] The independent secure channel is a communication channel encrypted using quantum key distribution technology. Monitoring signaling is transmitted in parallel through the encrypted communication channel and the service data channel carrying test data packets.

[0021] Specifically, in S3:

[0022] The blockchain network is a lightweight permissioned blockchain network, and the nodes include the sender and receiver of test data packets, as well as the key intermediate gateway nodes.

[0023] The structured summary should include at least the test action type, the associated error analysis result fingerprint, and the unique identification information of the standard used.

[0024] Specifically, in S4:

[0025] The dynamic baseline model is specifically defined as the instrument transformer health status baseline model. It is established by learning multi-dimensional features from historical normal test data through time series analysis or clustering algorithms. These multi-dimensional features include the time-series variation patterns and correlations of error values, environmental parameters, and load conditions.

[0026] Specifically, the trend of deterioration in recognition performance includes:

[0027] The feature values ​​of real-time test data within the current sliding time window are input into the dynamic baseline model to calculate a comprehensive anomaly deviation score, which is used to assess the degree of anomaly in the correlation between multiple indicators and trigger different levels of graded warnings accordingly.

[0028] Specifically, S4 also includes:

[0029] Feedback update steps: Confirmed abnormal data patterns and features are fed back to the dynamic baseline model to achieve adaptive updates of the model.

[0030] The technical effects and advantages of this invention are as follows:

[0031] This system achieves high efficiency, accuracy, and real-time performance in instrument transformer testing. By simultaneously acquiring primary, secondary, and standard reference signals, as well as operating environment data, and encapsulating them into structured data packets with a unified timescale at a high sampling rate of no less than 12800Hz, it overcomes the limitation of traditional offline testing requiring power outages, truly realizing continuous online monitoring during operation. Simultaneously, digital twin agents deployed at the transmitting and receiving ends perform virtual-real synchronization verification on an independent secure channel. This enables real-time extraction of key features and verification of data integrity and service rationality, significantly improving the reliability and accuracy of test data. It provides a credible basis for error calculation and status assessment, achieving a fundamental shift from post-inspection-oriented real-time sensing.

[0032] A smart operation and maintenance system with traceability and predictability has been constructed. This method innovatively integrates blockchain and machine learning technologies, storing key information from the testing process in structured summaries on the blockchain to form an immutable, end-to-end log. This supports rapid tracing of abnormal data and determination of responsibility, greatly enhancing the auditability and credibility of the testing process. Simultaneously, a dynamic baseline model built based on historical normal data can identify hidden degradation trends in transformer performance through multi-dimensional feature analysis and trigger tiered early warnings based on abnormal deviation scores. This achieves an upgrade from passive response to predictive maintenance, significantly improving the safe and stable operation of the power system and asset management efficiency. Attached Figure Description

[0033] Figure 1 This is a flowchart of the method of the present invention. Detailed Implementation

[0034] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0035] like Figure 1 As shown, the steps of an artificial intelligence-based online data testing method for current transformers are as follows:

[0036] Step 1: Test data preprocessing: At the current transformer operating site, simultaneously collect its primary side signal, secondary side output signal, standard reference signal, and operating condition and environmental data, and encapsulate them into a structured test data package with a unified time scale.

[0037] Online testing is performed on current transformers (CTs) or voltage transformers (PTs) in power systems. At the transformer operating site, the following key data are synchronously acquired using a high-precision acquisition unit and encapsulated into business data packets for subsequent processing:

[0038] Primary side current / voltage signal: Measured directly or indirectly from the primary circuit of the transformer under test, serving as the excitation source reference. Secondary side output signal: Measured from the secondary output terminals of the transformer under test, serving as the core output of the tested object. Standard reference signal: Obtained from the output of a higher-precision broadband standard transformer (e.g., 0.01 class) connected in series / parallel with the primary circuit, serving as a physical reference for online comparison and error calculation. Operating condition and environmental data: Including transformer body temperature, ambient temperature and humidity, vibration signal, load rate, etc.

[0039] All data is collected synchronously at a sampling rate of no less than 12800Hz and encapsulated into structured data packets with a unified and precise time stamp in accordance with power industry standard communication protocols (such as DL / T 645-2022 or IEC 61850-9-2). This data packet is the specific content carried by the "service data packet sent by the source end" in the subsequent steps, and its "service semantics" refers to the electrical performance parameters of the instrument transformer and the environmental conditions.

[0040] Step 2, Simultaneous Monitoring of Virtual and Real Data: Deploy digital twin agents at both the data sending and receiving ends, and transmit monitoring signaling containing key feature hashes through independent secure channels. At the receiving end, perform data integrity verification and business rationality checks to ensure that the uploaded test data is authentic and has not been tampered with.

[0041] Application scenarios for testing current transformers:

[0042] Source end: refers to the edge intelligent terminal installed on the substation or new energy power plant side, responsible for collecting and sending the instrument transformer test data packets described in step zero. Destination end: refers to the online verification and analysis system deployed at the provincial metrology center or cloud platform. Software-defined monitoring agent digital twin: At the source end, this twin not only simulates data packet transmission but also simulates the preliminary error calculation logic based on local standard data; at the destination end, this twin simulates a high-fidelity virtual instrument transformer, whose input is the primary side signal and operating condition data, and whose output is the theoretically ideal secondary side signal. In the monitoring signaling between the twins, the key business semantic fields specifically refer to the real-time error estimate, environmental parameters, and data feature summary calculated from the standard.

[0043] The specific steps are as follows:

[0044] Digital twin construction: At the source end of data transmission (such as a smart terminal at the edge of a power plant) and the destination end (such as a provincial metering center cloud platform), corresponding software-defined monitoring agents are created as digital twins for their data sending and receiving modules. These not only simulate the sending and receiving of data packets but also embed a business layer semantic understanding module (such as parsing the DL / T 645 protocol to understand electrical energy values).

[0045] Parallel processing and feature extraction: When the source sends a service data packet, its monitoring agent twin simultaneously captures the complete payload content of the data packet, key business semantic fields (such as meter readings and timestamps), and the precise sending timestamp. Subsequently, the twin calculates the cryptographic hash value of the data packet as its unique content fingerprint and generates a monitoring signaling that includes this hash value, a digest of the key business fields, and the timestamp.

[0046] Key business semantic fields, specific to the current transformer testing scenario, are defined as the following calculable electrical performance parameters and environmental state quantities, and encapsulated as a structured summary:

[0047] Basic electrical quantities: True RMS value of primary current The excitation reference is obtained from a broadband standard instrument transformer. The true RMS value of the secondary current is... The fundamental phase difference is measured from the secondary output of the current transformer under test. The phase lag of the secondary current fundamental wave of the tested transformer relative to the primary current fundamental wave is expressed in degrees (°). Real-time error characteristic value (preliminary calculation): Preliminary calculation is performed using local standard data to create an error snapshot.

[0048] Preliminary value of ratio difference (ratio error) :

[0049]

[0050] in, This is the rated transformation ratio of the current transformer.

[0051] Preliminary value of angular difference (phase error) :

[0052]

[0053] in, To measure the phase lag of the secondary current fundamental wave relative to the primary current fundamental wave of the current transformer, the unit is degrees (°), and it is a measured value. This is the ideal phase difference (usually close to 0°) measured by the standard at the same time.

[0054] Harmonic and distortion characteristics: characterize the performance of the tested transformer under non-ideal sinusoidal conditions.

[0055] Total Harmonic Distortion :

[0056]

[0057] in, This represents the effective value of the fundamental frequency of the secondary current. For the first RMS value of subharmonics, The highest harmonic order considered (e.g., 50th). Harmonic content for a specific order. (For example, the 5th and 7th harmonics).

[0058] Operating conditions and environmental parameters: Body temperature Real-time temperature (°C) of key points in the transformer core or windings. Effective value of vibration acceleration. Three-dimensional composite vibration acceleration (m / s²) 2 Load rate : The percentage of the current primary current to the rated current.

[0059] The feature extraction module normalizes and labels the above parameters to generate a fixed-dimensional feature vector and calculates its feature summary hash. .Should Content hash of the original data packet Together, they constitute a dual fingerprint in the monitoring signaling used for integrity comparison and business rationality verification.

[0060] Trusted channel transmission: Monitoring signaling is transmitted to the monitoring agent twin at the destination in near real-time through a highly secure, low-latency channel that is independent of the business data channel and encrypted using quantum key distribution technology.

[0061] Virtual-to-real comparison and semantic verification: After the destination monitoring proxy twin receives the actual data packet through the business channel, it immediately performs the following:

[0062] a) Calculate the content hash of the actual data packet and compare it with the hash value in the received monitoring signaling to verify data integrity.

[0063] b) Parse the service fields of the actual data packets and perform a reasonableness check against the digest in the monitoring signaling (e.g., whether the values ​​are within a reasonable range, whether the timing is continuous). Any inconsistency will immediately trigger an anomaly alarm.

[0064] Step 3, Blockchain Evidence Storage and Traceability: Generate structured summaries of key information from each round of testing (such as error indicators, environmental factors, confidence levels, etc.) and use blockchain technology for immutable evidence storage; when data anomalies are found, the source can be quickly traced based on the on-chain records to pinpoint whether the problem occurred in the collection, transmission, or analysis stage.

[0065] In the application scenario of mutual inductor testing, the blockchain network not only records the data transmission path, but its core function is to preserve key evidence and results of the testing process:

[0066] Monitoring signaling hash: This hash value is bound to a test snapshot that includes real-time error estimates, environmental conditions, and timestamps, ensuring that intermediate test results cannot be tampered with.

[0067] The documented transaction content is expanded to include "test action type" (e.g., periodic online testing, triggered diagnostic testing), "summary of associated error analysis results," and "standard instrument ID used in the test." This constitutes a complete and reliable digital log for each round of online testing.

[0068] The summary of the associated error analysis results is generated by standardizing and hashing the following elements:

[0069] Core error metric: Preliminary value of the ratio of feature extraction errors. Preliminary value of angle difference .

[0070] Environmental Correction Factor: Calculate a comprehensive environmental impact factor This is obtained through the following empirical model:

[0071]

[0072] in, Let τ be the temperature of the transformer body (°C). For reference temperature (e.g., 25℃). The effective value of vibration acceleration (Unit: m / s) 2 ), and These are the temperature sensitivity coefficient and vibration sensitivity coefficient obtained by fitting historical data, respectively, with units of % / ℃ and % / (m / s). 2 ), This represents the natural logarithm function, used to smooth vibration data and avoid excessive linear effects of extreme vibration values ​​on the model.

[0073] Confidence score The result is calculated based on the signal-to-noise ratio (SNR) of the test data packets and the consistency with the historical baseline, ranging from [0,1]. The specific calculation formula is as follows:

[0074]

[0075] in, This represents the normalized signal-to-noise ratio. The similarity with the historical baseline is calculated using Dynamic Time Warping (DTW) or cosine similarity, with a value ranging from 0 to 1. , These are the model weight parameters. for, This is the Sigmoid function.

[0076] Source fingerprint: Contains the unique device identifier of the broadband standard current transformer used in this test and the electronic hash of its current calibration certificate.

[0077] After the above elements are serialized into a structured string, their Merkle tree root hash value is calculated and stored in the blockchain as the final form of the "digest". This digest not only provides a compact digital fingerprint for the test results, ensuring their immutability, but also preserves key features (such as...) This provides a verifiable mathematical commitment for subsequent cross-validation, trend analysis, and auditing without exposing the complete raw data.

[0078] Application of traceability: When a test data anomaly is found during virtual-to-real comparison or subsequent analysis (such as a sudden error), the blockchain can be queried to irrefutably confirm whether the abnormal data originated from the on-site data collection process (such as sensor failure), the data transmission process (such as interference or packet loss), the cloud analysis process (such as model inaccuracy), or the actual degradation of the transformer's own performance, providing a technical basis for responsibility division and fault location.

[0079] Specific steps:

[0080] Chain-based notarization: A lightweight permissioned blockchain network is established, with nodes including source nodes, destination nodes, and key intermediate gateway nodes. When each node's monitoring agent completes its operation (e.g., source sending, intermediate node forwarding, destination receiving), it packages the action of this operation (send / forward / receive), the associated monitoring signaling hash, its own node ID, and a timestamp accurate to the microsecond into a notarization transaction, signs it, and uploads it to the blockchain.

[0081] An immutable log is created: These transactions are packaged into blocks in chronological order, and after consensus is reached among the consensus nodes, an immutable chain log is formed. This completely records the digital footprint of each data packet from its generation, through each node, to its final confirmation of receipt.

[0082] Source tracing: Once an anomaly is detected during the comparison between virtual and physical data (such as a hash mismatch), the blockchain is immediately queried. By comparing the characteristics of the abnormal data packet with the "monitoring signaling hash" stored on the blockchain, the corresponding evidence record chain for that data packet can be quickly located. By analyzing the timing and status of the records of each node on this chain, it is possible to accurately determine at which node the anomaly occurred (e.g., if the hash value changes after forwarding through an intermediate gateway, the problem may lie with that gateway or its subsequent links), achieving problem delimitation within minutes.

[0083] Step 4: Intelligent Anomaly Detection: Machine learning is used to model historical normal data, forming a dynamic baseline reflecting the health status of the transformer under various operating conditions. By comparing the deviation of current data from the baseline in real time, potential performance degradation trends can be identified, triggering tiered early warnings and enabling predictive maintenance.

[0084] In current transformer testing applications, the goal of the dynamic baseline learning model is to establish the behavioral patterns of current transformer errors and associated parameters under normal operating conditions.

[0085] Characteristics of business data packets: Specifically refers to the patterns contained in the test data packets of current transformers, such as: the normal fluctuation range of error values ​​under specific load rate and temperature conditions; the typical change law of phase angle under different harmonic contents; the error drift curve caused by temperature cycling within 24 hours, etc.

[0086] Normal flow behavior baseline model: This is specifically defined as the "current transformer health status baseline model". This model learns from historical normal test data to understand the error tolerance range and parameter correlation spectrum of the current transformer under different seasons, different load sections, and different environmental pressures (such as sandstorms and high temperatures).

[0087] The purpose of anomaly detection: By monitoring the degree to which the data stream deviates from the baseline in real time, the system not only determines network communication anomalies, but more importantly, identifies hidden degradation trends in transformer performance. For example, even if a single error does not exceed the standard, but there is a systematic shift in its correlation pattern with temperature, or a gradual change in the angle difference under a specific harmonic, the model can provide early warnings and achieve predictive maintenance.

[0088] Specific steps:

[0089] Multi-dimensional feature acquisition: The monitoring agent twin not only collects data packet content features, but also continuously collects context features related to business flow, including but not limited to: the sending cycle pattern of business data packets, the proportion of data packets of different business types (such as real-time data and configuration instructions), the trend of the total effective payload per unit time, and the cross-node transmission delay distribution from innovation point two.

[0090] Dynamic baseline modeling: During the initial learning phase of the system, machine learning algorithms (such as time series analysis and clustering algorithms) are used to learn the above multidimensional features and establish a "normal flow behavior baseline model" for different business scenarios (such as power plant full-load periods and low-load periods at night). This baseline model describes the normal range, periodicity, and correlation of each feature value.

[0091] Real-time anomaly scoring: During the real-time monitoring phase, the system inputs multi-dimensional feature values ​​within the current sliding time window into the baseline model to calculate a comprehensive "anomaly deviation" score. This score not only considers a single indicator exceeding a threshold, but also focuses on anomalies in the correlation between multiple indicators (for example, if the sending cycle suddenly becomes disordered while the data volume of a single packet increases abnormally, even if neither exceeds an independent threshold, it is still identified as a high-risk anomaly).

[0092] Tiered early warning and feedback: Different levels of early warning are triggered based on the anomaly deviation score. Simultaneously, confirmed abnormal traffic patterns and their characteristics are fed back to the baseline model, enabling adaptive model updates and improving the ability to detect new anomaly patterns in the future.

[0093] The above formulas are all dimensionless calculations. Dimensionless calculations can be performed using various methods such as standardization, which will not be elaborated here. The formulas are derived from software simulations based on a large amount of collected data, and the preset parameters in the formulas can be set by those skilled in the art according to the actual situation.

[0094] The above embodiments can be implemented, in whole or in part, by software, hardware, firmware, or any other combination thereof. When implemented using software, the above embodiments can be implemented, in whole or in part, as a computer program product. The computer program product includes one or more computer instructions or computer programs. When the computer instructions or computer programs are loaded or executed on a computer, all or part of the processes or functions described in the embodiments of this application are generated. The computer can be a general-purpose computer, a special-purpose computer, a computer network, or other programmable device. The computer instructions can be stored in a computer-readable storage medium or transmitted from one computer-readable storage medium to another. For example, the computer instructions can be transmitted from one website, computer, server, or data center to another website, computer, server, or data center via wired or wireless (e.g., infrared, wireless, microwave, etc.) means. The computer-readable storage medium can be any available medium that a computer can access or a data storage device such as a server or data center that includes one or more sets of available media. The available medium can be a magnetic medium (e.g., floppy disk, ATA hard disk, magnetic tape), an optical medium (e.g., DVD), or a semiconductor medium. The semiconductor medium can be a solid-state ATA hard disk.

[0095] It should be understood that in the various embodiments of this application, the order of the above-mentioned processes does not imply the order of execution. The execution order of each process should be determined by its function and internal logic, and should not constitute any limitation on the implementation process of the embodiments of this application.

[0096] Those skilled in the art will recognize that the units and algorithm steps of the various examples described in conjunction with the embodiments disclosed herein can be implemented in electronic hardware, or a combination of computer software and electronic hardware. Whether these functions are implemented in hardware or software depends on the specific application and design constraints of the technical solution. Those skilled in the art can use different methods to implement the described functions for each specific application, but such implementation should not be considered beyond the scope of this application.

[0097] In the several embodiments provided in this application, it should be understood that the disclosed systems, apparatuses, and methods can be implemented in other ways. For example, the apparatus embodiments described above are merely illustrative; for instance, the division of units is only a logical functional division, and in actual implementation, there may be other division methods. For example, multiple units or components may be combined or integrated into another system, or some features may be ignored or not executed. Furthermore, the coupling or direct coupling or communication connection shown or discussed may be through some interfaces; the indirect coupling or communication connection between apparatuses or units may be electrical, mechanical, or other forms.

[0098] The units described as separate components may or may not be physically separate. The components shown as units may or may not be physical units; they may be located in one place or distributed across multiple network units. Some or all of the units can be selected to achieve the purpose of this embodiment, depending on actual needs.

[0099] In addition, the functional units in the various embodiments of this application can be integrated into one processing unit, or each unit can exist physically separately, or two or more units can be integrated into one unit.

[0100] If the aforementioned functions are implemented as software functional units and sold or used as independent products, they can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of this application, in essence, or the part that contributes to the prior art, or a portion of the technical solution, can be embodied in the form of a software product. This computer software product is stored in a storage medium and includes several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) to execute all or part of the steps of the methods described in the various embodiments of this application. The aforementioned storage medium includes various media capable of storing program code, such as USB flash drives, portable ATA hard drives, read-only memory (ROM), random access memory (RAM), magnetic disks, or optical disks.

[0101] The above description is merely a specific embodiment of this application, but the scope of protection of this application is not limited thereto. Any variations or substitutions that can be easily conceived by those skilled in the art within the scope of the technology disclosed in this application should be included within the scope of protection of this application. Therefore, the scope of protection of this application should be determined by the scope of the claims.

Claims

1. An online data testing method for current transformers based on artificial intelligence, characterized in that, Includes the following steps: S1. Test data preprocessing: At the current transformer operating site, the primary side signal, secondary side output signal, standard reference signal, and operating condition and environmental data are collected synchronously and packaged into a structured test data package with a unified time scale. S2. Simultaneous Monitoring of Virtual and Real Data: Digital twin agents are deployed at both the sending and receiving ends of the test data packets. The digital twin agent at the sending end extracts key features from the test data packets and generates monitoring signaling, which is then sent to the receiving end through an independent secure channel. Upon receiving the test data packets, the digital twin agent at the receiving end uses the monitoring signaling to verify data integrity and business rationality. In S2: The digital twin agent is a software-defined monitoring agent that embeds a business layer semantic understanding module to parse electrical performance parameters and environmental state quantities in test data packets. It simulates preliminary error calculation logic at the sending end and simulates a high-fidelity virtual transformer at the receiving end to output theoretical secondary side signals. The key features extracted include: the true RMS values ​​of the primary and secondary currents, the fundamental phase difference, the preliminary values ​​of the ratio difference and angle difference calculated in real time based on the standard reference signal, the total harmonic distortion rate, and the specific harmonic content rate; The independent secure channel is a communication channel encrypted using quantum key distribution technology. Monitoring signaling is transmitted in parallel through the encrypted communication channel and the service data channel carrying test data packets. S3. Blockchain Evidence Storage and Traceability: The summary information containing the verification results and key features is structured and stored in the blockchain network to form an immutable test process log; when data anomalies are detected, traceability analysis is performed based on the records in the blockchain network. S4. Intelligent Anomaly Detection: Based on historical normal test data, a dynamic baseline model reflecting the health status of the transformer is established using machine learning algorithms; by comparing the deviation between real-time test data and the dynamic baseline model, performance degradation trends are identified and early warnings are triggered.

2. The method for online data testing of current transformers based on artificial intelligence according to claim 1, characterized in that, In S1: The collected data is synchronously acquired at a sampling rate of no less than 12800Hz and encapsulated according to a preset standard protocol; Operating condition and environmental data should include at least the transformer body temperature, the effective value of three-dimensional vibration acceleration, and the load rate expressed as a percentage.

3. The online data testing method for current transformers based on artificial intelligence according to claim 1, characterized in that, In S3: The blockchain network is a lightweight permissioned blockchain network, and the nodes include the sender and receiver of test data packets, as well as the key intermediate gateway nodes. The structured summary should include at least the test action type, the associated error analysis result fingerprint, and the unique identification information of the standard used.

4. The method for online data testing of current transformers based on artificial intelligence according to claim 1, characterized in that, In S4: The dynamic baseline model is specifically defined as the instrument transformer health status baseline model. It is established by learning multi-dimensional features from historical normal test data through time series analysis or clustering algorithms. These multi-dimensional features include the time-series variation patterns and correlations of error values, environmental parameters, and load conditions.

5. The online data testing method for current transformers based on artificial intelligence according to claim 4, characterized in that, The specific trends in the degradation of recognition performance include: The feature values ​​of real-time test data within the current sliding time window are input into the dynamic baseline model to calculate a comprehensive anomaly deviation score, which is used to assess the degree of anomaly in the correlation between multiple indicators and trigger different levels of graded warnings accordingly.

6. The online data testing method for current transformers based on artificial intelligence according to claim 1, characterized in that, S4 also includes: Feedback update steps: Confirmed abnormal data patterns and features are fed back to the dynamic baseline model to achieve adaptive updates of the model.

Images