Methods and related equipment for dynamic identification of abnormalities in converter transformers considering operating conditions

Abstract

This application provides a method and related equipment for dynamic identification of converter transformer anomalies considering operating conditions, belonging to the field of power equipment operation and maintenance technology. The method includes simultaneously acquiring online operating power data of the converter transformer and online grounding current data of the transformer's grounding system and aligning them with timestamps; inputting the online operating power data into a trained power-current benchmark model for current prediction to obtain the theoretical expected value of the grounding current under the current operating conditions; performing operating condition decoupling processing on the online grounding current data based on the theoretical expected value of the grounding current to obtain a state residual sequence stripped of load profiles; and performing adaptive anomaly identification on the state residual sequence according to preset anomaly diagnosis conditions to obtain a judgment result. This application embodiment can accurately strip away operating condition interference and achieve keen detection of early latent anomalies in converter transformers.

Description

Technical Field

[0001] This application relates to the field of power equipment operation and maintenance technology, and in particular to a method and related equipment for dynamic identification of abnormalities in converter transformers that takes into account operating conditions. Background Technology

[0002] As a core component of ultra-high voltage direct current (UHVDC) transmission systems, the reliability of converter transformers directly impacts the safety and stability of the power grid. Currently, online monitoring of the grounding current in the core and clamping components of converter transformers primarily relies on the "static threshold warning" method, which sets a fixed alarm upper limit based on relevant operating procedures. From a physical perspective, the leakage flux distribution within the converter transformer is heavily influenced by the coupling effect of its operating active power (load). When the equipment is operating under high load, the leakage flux increases dramatically, causing a non-fault-related natural surge in the induced grounding current in the core and clamping components. This static threshold, which ignores operating condition variables, is easily triggered by normal magnetic field enhancement during high-load periods, leading to frequent false warnings from the monitoring system and significantly increasing the workload of on-site maintenance personnel.

[0003] Conversely, when equipment is operating under low or extremely low load, even if latent insulation defects such as multi-point grounding have already occurred, the total absolute value of the fault current, after being superimposed with the relatively low base current, often remains below the static red line. This phenomenon, where weak fault signals are masked by the trough of low-level operating conditions, leads to a long-term risk of unreported latent defects, making it extremely easy to miss the optimal maintenance window to avoid unplanned outages. Summary of the Invention

[0004] The main objective of this application is to propose a method and related equipment for dynamic identification of converter transformer anomalies that takes into account operating conditions, aiming to accurately isolate operating condition interference and achieve keen detection of early latent abnormal states of converter transformers.

[0005] To achieve the above objectives, one aspect of this application proposes a dynamic identification method for converter transformer anomalies considering operating conditions, the method comprising the following steps: Simultaneously acquire online operating power data of the converter transformer and online grounding current data of the main grounding system, and align the timestamps. The online operating power data is input into the trained power current benchmark model to predict the current and obtain the theoretical expected value of the ground current under the current operating conditions. Based on the theoretical expected value of the grounding current, the online grounding current data is subjected to operating condition decoupling processing to obtain a state residual sequence stripped of the load profile; The state residual sequence is adaptively anomaly identified based on preset anomaly diagnosis conditions to obtain a judgment result.

[0006] In some embodiments, the power current reference model is obtained through the following steps: Obtain a historical health sample set of the converter transformer and the main grounding system in a healthy operating range, wherein the historical health sample set includes historical operating power data and historical grounding current data; The historical operating power data is used as an input variable to characterize the combined electromagnetic and thermal operating conditions, and the historical grounding current data is used as an output variable to characterize the overall health status of the system. Correlation features are extracted and modeled on the input variables to obtain the power-current benchmark model.

[0007] In some embodiments, the step of using the historical operating power data as an input variable characterizing the combined electromagnetic and thermal operating conditions, and the historical grounding current data as an output variable characterizing the overall health status of the system, and extracting and modeling the correlation features of the input variables to obtain a power-current benchmark model, includes the following steps: The historical operating power data is mapped onto the horizontal axis, and the historical grounding current data is mapped onto the vertical axis to obtain a scatter plot; The physical coupling law of the scatter plot is nonlinearly fitted using a polynomial regression algorithm to form a power current benchmark model, wherein the power current benchmark model is a characteristic curve that reflects the nonlinear increase of current with load.

[0008] In some embodiments, the step of performing operating condition decoupling processing on the online grounding current data based on the theoretical expected value of the grounding current to obtain a state residual sequence stripped of the load profile includes the following steps: The theoretical expected value of the grounding current is timestamped and sorted with the online grounding current data to obtain a pairing sequence, wherein the theoretical expected value of the grounding current and the online grounding current data with the same timestamp are a pair of paired data; The paired data is subjected to operating condition fundamental wave filtering to remove the natural current fluctuation component caused by load fluctuation, thereby obtaining a health offset characteristic quantity that reflects the internal insulation evolution and mechanical displacement of the equipment. The health offset features are sorted according to timestamp order to obtain the state residual sequence.

[0009] In some embodiments, the adaptive anomaly identification of the state residual sequence based on preset anomaly diagnosis conditions to obtain a judgment result includes the following steps: Based on the distribution characteristics of health conditions in the historical health sample set, mathematical statistics are performed to obtain the dynamic early warning upper and lower limit threshold bands. The preset abnormal diagnosis conditions are determined based on the upper and lower threshold bands of the warning and the state residual sequence. The abnormal diagnosis conditions are that the state residual sequence shows a monotonically divergent trend and breaks through the boundary of the dynamic warning upper and lower threshold bands. If the abnormal diagnosis conditions are not met, the judgment result is determined to be that the converter transformer is in the normal operating condition fluctuation range. If the abnormal diagnostic conditions are met, the judgment result is determined to be that the converter transformer is in a sub-healthy state.

[0010] In some embodiments, the step of performing mathematical statistics on the distribution characteristics of health conditions based on the historical health sample set to obtain dynamic early warning upper and lower limit threshold bands includes the following steps: Calculate the health residual sequence of health conditions based on the historical health sample set; Mathematical statistics were performed on the health residual sequence to obtain the health mean and health standard deviation; Based on the statistical principle of normal distribution, the boundaries of the health interval are calculated according to the health mean and the health standard deviation to obtain the dynamic early warning upper and lower limit threshold bands.

[0011] In some embodiments, the method further includes the following steps: When the judgment result indicates that the converter transformer is in a sub-healthy state, an early warning command is output, wherein the early warning command is used to perform predictive maintenance for latent defects.

[0012] To achieve the above objectives, another aspect of this application proposes a converter transformer anomaly dynamic identification system that takes into account operating conditions, the system comprising: The first module is used to synchronously acquire online operating power data of the converter transformer and online grounding current data of the main grounding system and perform timestamp alignment. The second module is used to input the online operating power data into the trained power current benchmark model to predict the current and obtain the theoretical expected value of the ground current under the current operating conditions. The third module is used to perform operating condition decoupling processing on the online grounding current data according to the theoretical expected value of the grounding current, and to obtain a state residual sequence stripped of the load profile. The fourth module is used to adaptively identify anomalies in the state residual sequence based on preset anomaly diagnosis conditions, and obtain a judgment result.

[0013] To achieve the above objectives, another aspect of this application provides an electronic device, which includes a memory and a processor. The memory stores a computer program, and the processor executes the computer program to implement the above-described method.

[0014] To achieve the above objectives, another aspect of the embodiments of this application proposes a computer-readable storage medium storing a computer program that, when executed by a processor, implements the above-described method.

[0015] To achieve the above objectives, another aspect of this application provides a computer program product, including a computer program that, when executed by a processor, implements the above-described method.

[0016] The embodiments of this application include at least the following beneficial effects: This application provides a method and related equipment for dynamic identification of converter transformer anomalies considering operating conditions. This scheme synchronously acquires online operating power data of the converter transformer and online grounding current data of the transformer's grounding system and aligns them with timestamps; inputs the online operating power data into a trained power-current benchmark model for current prediction to obtain the theoretical expected value of the grounding current under the current operating conditions; performs operating condition decoupling processing on the online grounding current data based on the theoretical expected value of the grounding current, accurately removes operating condition interference, and obtains a state residual sequence stripped of the load profile. This can cut off the interference transmission path from severe load fluctuations to monitoring alarms, fundamentally eliminating the chronic problem of false alarms under high full load conditions and improving the confidence of alarms; and performs adaptive anomaly identification on the state residual sequence according to preset anomaly diagnosis conditions, which can keenly detect weak pathological deviations hidden behind macroscopic operating fluctuations, obtain judgment results, and achieve keen capture of early latent anomalies in the converter transformer. Attached Figure Description

[0017] Figure 1 This is a flowchart of the converter transformer anomaly dynamic identification method considering operating conditions provided in the embodiments of this application; Figure 2 This is a schematic diagram showing the effect of comparing the measured core grounding current with the dynamic baseline based on operating conditions provided in the embodiments of this application; Figure 3 This is a schematic diagram illustrating the correlation characteristics between operating power and core grounding current, and the effect of regression modeling, provided in the embodiments of this application. Figure 4 This is a schematic diagram illustrating the effect of dynamic residual sequence and latent anomaly identification provided in the embodiments of this application; Figure 5 This is a flowchart of a converter transformer anomaly dynamic identification method considering operating conditions provided in another embodiment of this application; Figure 6 This is a schematic diagram of the structure of the converter transformer abnormal dynamic identification system considering operating conditions provided in the embodiments of this application; Figure 7 This is a schematic diagram of the hardware structure of the electronic device provided in the embodiments of this application. Detailed Implementation

[0018] To make the objectives, technical solutions, and advantages of this application clearer, the following detailed description is provided in conjunction with the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are merely illustrative of this application and are not intended to limit it. In the following description, when referring to the accompanying drawings, unless otherwise indicated, the same numbers in different drawings represent the same or similar elements. The embodiments described in the following exemplary embodiments do not represent all embodiments consistent with those of this application; they are merely examples of systems and methods consistent with some aspects of the embodiments of this application as detailed in the appended claims.

[0019] Unless otherwise defined, all technical and scientific terms used herein have the same meaning as commonly understood by one of ordinary skill in the art to which this application belongs. The terminology used herein is for the purpose of describing embodiments of this application only and is not intended to limit this application.

[0020] Before providing a detailed description of the embodiments of this application, the relevant technologies involved in the embodiments of this application will be described first.

[0021] In related technologies, static threshold monitoring technology suffers from technical biases and defects, such as "false alarms are prone to occur under high loads, while missed alarms are prone to occur under low loads." Because it lacks a dynamic health benchmark that can adaptively adjust to different operating conditions, it is impossible to accurately extract purely characteristic components reflecting the insulation degradation of the equipment itself from complex monitoring sequences that are severely contaminated by operating conditions.

[0022] In view of this, this application provides a method and related equipment for dynamic identification of converter transformer anomalies considering operating conditions. The scheme includes simultaneously acquiring online operating power data of the converter transformer and online grounding current data of the transformer's grounding system, and aligning them with timestamps; inputting the online operating power data into a trained power-current benchmark model for current prediction to obtain the theoretical expected value of the grounding current under the current operating conditions; performing operating condition decoupling processing on the online grounding current data based on the theoretical expected value of the grounding current to obtain a state residual sequence stripped of load profiles; and performing adaptive anomaly identification on the state residual sequence according to preset anomaly diagnosis conditions to obtain a judgment result. This application embodiment can accurately strip away operating condition interference and achieve keen detection of early latent anomalies in the converter transformer.

[0023] The method for dynamic identification of converter transformer anomalies considering operating conditions provided in this application relates to the field of power equipment operation and maintenance technology. This method can be applied to a terminal, a server, or software running on either a terminal or a server. In some embodiments, the terminal can be a smartphone, tablet, laptop, desktop computer, smart speaker, smartwatch, or vehicle terminal, but is not limited to these. The server can be configured as an independent physical server, a server cluster or distributed system composed of multiple physical servers, or a cloud server providing basic cloud computing services such as cloud services, cloud databases, cloud computing, cloud functions, cloud storage, network services, cloud communication, middleware services, domain name services, security services, CDN, and big data and artificial intelligence platforms. The server can also be a node server in a blockchain network. The software can be an application implementing the method for dynamic identification of converter transformer anomalies considering operating conditions, but is not limited to the above forms.

[0024] This application can be used in a wide variety of general-purpose or special-purpose computer system environments or configurations. Examples include: personal computers, server computers, handheld or portable devices, tablet devices, multiprocessor systems, microprocessor-based systems, set-top boxes, programmable consumer electronics devices, network PCs, minicomputers, mainframe computers, and distributed computing environments including any of the above systems or devices. This application can be described in the general context of computer-executable instructions executed by a computer, such as program modules. Generally, program modules include routines, programs, objects, components, data structures, etc., that perform specific tasks or implement specific abstract data types. This application can also be practiced in distributed computing environments where tasks are performed by remote processing devices connected via a communication network. In distributed computing environments, program modules can reside in local and remote computer storage media, including storage devices.

[0025] Figure 1 This is an optional flowchart of the converter transformer anomaly dynamic identification method considering operating conditions provided in the embodiments of this application. Figure 1 The method may include, but is not limited to, steps S101 to S104.

[0026] Step S101: Simultaneously acquire the online operating power data of the converter transformer and the online grounding current data of the main grounding system, and align the timestamps.

[0027] Step S102: Input the online operating power data into the trained power current benchmark model to predict the current and obtain the theoretical expected value of the grounding current under the current operating conditions.

[0028] Step S103: Perform operating condition decoupling processing on the online grounding current data according to the theoretical expected value of the grounding current to obtain the state residual sequence of stripped load profile.

[0029] Step S104: Adaptive anomaly identification is performed on the state residual sequence according to the preset anomaly diagnosis conditions to obtain the judgment result.

[0030] In this embodiment, "considering operating conditions" means that when monitoring and judging whether the grounding current is abnormal, the real-time operating conditions of the equipment are taken into account as an important factor. This is because the grounding current of the converter transformer is not constant; it naturally fluctuates with changes in operating power. When the operating power is high, the grounding current will naturally increase, making it easy to trigger false alarms when judging anomalies based on static thresholds. Conversely, when the operating power is low, the grounding current signal is weak, making it easy to miss latent defects. The converter transformer anomaly dynamic identification method considering operating conditions provided in this application establishes a benchmark model between power and current, eliminating the interference of natural current fluctuations caused by load fluctuations on the monitoring results, thereby achieving an unbiased assessment of the equipment's health status.

[0031] Specifically, the first step is to collect and align multi-source heterogeneous data. The multi-source heterogeneous data includes two types of key data: one is the online operating power data of the converter transformer, and the other is the online grounding current data of the main grounding system (core and clamps). By synchronously acquiring multi-source heterogeneous data and performing data fusion and cleaning based on a unified timestamp, including outlier removal and smooth interpolation of missing values, a fused state feature set with high time-series fidelity is constructed.

[0032] Next, the system continuously performs state expectation value prediction, inputting the currently measured online operating power data into a pre-trained and fixed power current benchmark model, and instantly calculating the theoretical expected value of the grounding current that the core or clamping component should exhibit under the current specific load condition, such as... Figure 2 As shown, Figure 2 The period from March 2024 to November 2024 is the modeling and training period. Figure 2 As can be seen from this, the theoretical expected value predicted by the power current benchmark model basically achieves perfect tracking and coverage of all fluctuations in the real current caused by the fluctuations in the healthy operating conditions.

[0033] It should be noted that the power current reference model aims to learn the variation of grounding current with operating power during the training process. It deeply couples the coupling law between input and output variables from a large amount of historical data, and can accurately characterize the physical influence of nonlinear leakage flux on induced current. This power current reference model can accurately calculate the theoretical expected value of the corresponding grounding current of the core or clamp under healthy operating conditions based on any given online operating power data.

[0034] Furthermore, the mathematical difference between the theoretical expected value of the grounding current and the online grounding current data is defined as the "state residual". The state residual is continuously calculated at each sampling time and arranged in chronological order to form a state residual sequence.

[0035] Through residual verification technology, the system can decouple the currently collected online grounding current data, which includes fluctuations in operating conditions, in real time. By utilizing the characteristic that the expected value of the grounding current represents the normal operating condition, the system filters out changes in grounding current caused by normal operation, such as load fluctuations and increased leakage flux, and retains only the state residuals that reflect the insulation degradation of the equipment itself. This cuts off the interference transmission path from severe load fluctuations to the monitoring and alarm system, fundamentally eliminating the chronic problem of false alarms under high full load conditions. It significantly improves the confidence of alarms from the centralized control cockpit and intelligent operation and maintenance system, and achieves highly robust condition-immune-level state monitoring.

[0036] Finally, by utilizing the fact that the state residuals of a converter transformer in a healthy state should exhibit a typical stationary random walk state with a mean of zero, an envelope band is constructed for the state residuals. Then, it is determined whether each state residual in the state residual sequence deviates from the reasonable interpretation range of the current load condition, and the judgment result is obtained.

[0037] In some embodiments, the training method for the power current reference model may include, but is not limited to, steps S201 to S202.

[0038] Step S201: Obtain a historical health sample set of the converter transformer and the main grounding system in the healthy operating range. The historical health sample set includes historical operating power data and historical grounding current data.

[0039] Step S202: Historical operating power data is used as an input variable to characterize the comprehensive electromagnetic and thermal operating conditions, and historical grounding current data is used as an output variable to characterize the health status of the system. Correlation features are extracted and modeled for the input variables and the input variables to obtain the power current benchmark model.

[0040] In this embodiment, the power current reference model is constructed through multi-source data and can accurately isolate operating condition interference.

[0041] Specifically, a historical health sample set is extracted from the converter transformer and the main grounding system within the absolute healthy operating range. The healthy operating range represents the time window during which the converter transformer and the main grounding system are free from any defects. Historical operating power data and historical grounding current data are collected within this time window to obtain the historical health sample set used for training.

[0042] After obtaining the historical health sample set, the power current benchmark model is trained and solidified. The absolute value of the operating active power is used as the input variable to characterize the electromagnetic and thermal integrated operating conditions, and the grounding current at the synchronization moment is used as the output variable to characterize the physical health status. The core objective of the power current benchmark model is to learn the nonlinear physical coupling relationship between the input and output variables based on the historical health sample set, thereby establishing a mapping function between the input and output variables.

[0043] In some embodiments, step S202 may include, but is not limited to, steps S301 to S302.

[0044] Step S301: Map the historical operating power data to the horizontal axis and the historical grounding current data to the vertical axis to obtain a scatter plot.

[0045] Step S302: Nonlinear fitting of the physical coupling law of the scatter plot is performed using a polynomial regression algorithm to form a power current benchmark model, wherein the power current benchmark model is a characteristic curve used to reflect the nonlinear increase of current with load.

[0046] In this embodiment, the data of the historical health sample set is visualized as a two-dimensional scatter plot, such as... Figure 3 As shown, historical operating power data is mapped onto the horizontal axis, and historical grounding current data is mapped onto the vertical axis. When all sample points from the historical healthy sample set are mapped onto the scatter plot, the overall distribution of these scatter points intuitively reveals a strict and significant positive correlation between operating power and grounding current. For example, according to... Figure 3 As can be seen from the scatter plot, within the normal healthy operating range, the grounding current shows an increasing trend as the operating power increases.

[0047] Furthermore, after determining the existence of a positive correlation between historical operating power data and historical grounding current data based on the scatter plot, this embodiment uses a polynomial regression algorithm to nonlinearly fit the physical coupling law of the scatter plot. Through mathematical modeling, a characteristic curve is found so that the curve can approximate the overall distribution trend of all sample points to the greatest extent, thus obtaining a characteristic curve that reflects the nonlinear increase of current with load.

[0048] For example, this embodiment uses a second-order polynomial regression algorithm for fitting. The fitting curve corresponding to the second-order polynomial regression algorithm is a parabola. The least squares method is used to minimize the sum of squared residuals between the grounding current values ​​of all sample points and the corresponding values ​​on the fitting curve, thereby obtaining the coefficients of each term of the polynomial.

[0049] In some embodiments, step S103 may include, but is not limited to, steps S401 to S403.

[0050] Step S401: The theoretical expected value of grounding current and the online grounding current data are timestamped and sorted to obtain a pairing sequence. The theoretical expected value of grounding current and the online grounding current data with the same timestamp are a pair of paired data.

[0051] Step S402: Perform operating condition fundamental wave filtering on the paired data to remove the natural current fluctuation component caused by load fluctuations, and obtain the health offset characteristic quantity reflecting the internal insulation evolution and mechanical displacement of the equipment.

[0052] Step S403: Sort the health offset features according to the timestamp order to obtain the state residual sequence.

[0053] In this embodiment, the online grounding current data includes both the natural fluctuation component of the grounding current determined by the operating power and the health-related characteristic quantity of the grounding current caused by defects. The theoretical expected value of the grounding current represents the level of grounding current that the equipment should theoretically present when it is in a healthy state, as calculated by the power current reference model based on the current operating power. Therefore, by performing a difference calculation between the theoretical value of the grounding current and the online grounding current data, the influence of the operating conditions on the grounding current can be filtered out.

[0054] Specifically, firstly, the theoretical expected value of the grounding current with the same realization stamp is paired with the online grounding current data to form a pair of paired data. Then, all the paired data are sorted according to the order of the timestamps to obtain the pairing sequence.

[0055] After data alignment and pairing are completed, since the online grounding current data can be regarded as the superposition of two components, and the theoretical expected value of the grounding current just represents the current level that should be in a healthy state under the current operating conditions, the state residual obtained by subtracting the theoretical expected value of the grounding current from the online grounding current data becomes a weak fault signal that was originally submerged in the macroscopic operating condition fluctuations, and a health offset characteristic quantity reflecting the internal insulation evolution and mechanical displacement of the equipment is obtained.

[0056] By sorting discrete health offset features over time to form a state residual sequence, the system can identify early latent defects in subsequent steps by detecting the mean change of the sequence within a sliding window, the monotonicity of the sequence, and the magnitude of the values. This avoids misjudgment due to occasional noise or disturbances at a single moment. Only when a persistent deviation trend appears is it determined that there is a substantial abnormality in the equipment, thus improving the confidence of the early warning.

[0057] In some embodiments, step S104 may include, but is not limited to, steps S501 to S504.

[0058] Step S501: Perform mathematical statistics on the distribution characteristics of health conditions based on the historical health sample set to obtain the upper and lower threshold ranges for dynamic early warning.

[0059] Step S502: Determine whether the preset abnormal diagnosis conditions are met based on the upper and lower limit threshold bands of the warning and the state residual sequence. The abnormal diagnosis condition is that the state residual sequence shows a monotonically divergent trend and breaks through the boundary of the upper and lower limit threshold bands of the dynamic warning.

[0060] Step S503: If the abnormal diagnosis conditions are not met, the judgment result is determined to be that the converter transformer is in the normal operating condition fluctuation range.

[0061] Step S504: When the abnormal diagnosis conditions are met, the judgment result is determined to be that the converter transformer is in a sub-healthy state.

[0062] In this embodiment, adaptive anomaly identification mainly utilizes historical healthy sample sets to statistically analyze the distribution characteristics of state residual sequences under healthy operating conditions. For example, it can adaptively generate dynamic early warning upper and lower limit threshold bands by using the statistical 3σ principle or kernel density estimation confidence interval.

[0063] In routine monitoring, the pre-set abnormal diagnosis conditions are determined based on the upper and lower threshold bands of the warning and the state residual sequence. These abnormal diagnosis conditions are not simple single-point limit judgments, but rather a comprehensive consideration of the overall trend of the state residual and the threshold breakthrough situation. Only when the state residual sequence shows a monotonically divergent trend and continuously breaks through the boundaries of the dynamic warning upper and lower threshold bands, are both sub-conditions, is the abnormal diagnosis condition determined to be valid.

[0064] Unlike traditional static red lines which are ineffective for weak fault currents in low-load areas, this embodiment completely removes the baseline of the underlying operating conditions, enabling it to detect subtle pathological deviations hidden behind macroscopic operational fluctuations with extreme sensitivity. It truly achieves a technological generational leap from "post-event limit alarms" to "predictive sub-health status," winning valuable decision-making time for predictive maintenance of heavy-asset equipment. Moreover, the dual accuracy measurement based on trend characteristics and threshold breakthroughs can further reduce the risk of false alarms.

[0065] If the state residuals are stably converged within this threshold band, such as Figure 4 As shown, during the period from July 2024 to September 2024, the state residual sequence of the converter transformer never exceeded the upper and lower limits of the threshold band. At this time, the abnormal diagnosis condition is not met, and the system determines that the converter transformer is in the normal operating condition fluctuation range.

[0066] If so Figure 4 During the period from September 2024 to November 2024, the state residual sequence exhibits a significant directional drift. If the abnormal diagnosis condition is met at this time, the system determines that the converter transformer is in a sub-healthy state.

[0067] In some embodiments, step S501 may include, but is not limited to, steps S601 to S603.

[0068] Step S601: Calculate the health residual sequence of health conditions based on the historical health sample set.

[0069] Step S602: Perform mathematical statistics on the health residual sequence to obtain the health mean and health standard deviation.

[0070] Step S603: Based on the statistical principle of normal distribution, the boundary of the health interval is calculated according to the health mean and health standard deviation to obtain the dynamic early warning upper and lower limit threshold bands.

[0071] In this embodiment, the dynamic early warning upper and lower limit threshold band adopts the statistical 3σ principle. First, the historical operating power data in the historical healthy sample set is input into the trained and solidified power current benchmark model to predict the theoretical expected value of the grounding current under healthy operating conditions. Then, the health residual sequence is obtained by subtracting the historical grounding current data and the theoretical expected value of the grounding current under healthy operating conditions point by point.

[0072] It should be noted that although the method of obtaining the health residual sequence is the same as that of the state residual sequence in the real-time detection stage, their physical meanings are not the same. The health residual sequence is the residual sequence calculated when the equipment is running in health. The fluctuation of this sequence is limited to random noise in the measurement process and modeling residuals of the benchmark model, and does not contain any abnormal components caused by insulation defects, etc. Therefore, the health residual sequence can provide data support for determining the allowable fluctuation range of residuals under healthy conditions.

[0073] Next, mathematical statistics are performed on the health residual sequence to calculate the mean and standard deviation of the health residual state sequence, thus obtaining the health mean and health standard deviation. Ideally, the power current reference model should predict the ground current without deviation, so the health mean of the health residual sequence should approach zero. The health standard deviation quantifies the dispersion of the health residual sequence, that is, the range of fluctuation of the residual around the mean. Based on the statistical principle of normal distribution, for a random variable that follows a normal distribution, the probability that its value falls within the range of the mean plus or minus three times the standard deviation is about 99.7%. Once the residual value exceeds this range, it means that the residual value is very likely to be superimposed with abnormal health offset characteristics. Therefore, the value of the health mean minus three times the health standard deviation is used as the lower boundary of the dynamic early warning upper and lower limit threshold band, and the value of the health mean plus three times the health standard deviation is used as the upper boundary of the dynamic early warning upper and lower limit threshold band.

[0074] In some embodiments, the converter transformer abnormal dynamic identification method that takes into account operating conditions may also include, but is not limited to, step S701.

[0075] Step S701: When the judgment result is that the converter transformer is in a sub-healthy state, an early warning command is output. The early warning command is used to perform predictive maintenance for latent defects.

[0076] In this embodiment, when the judgment result is that the converter transformer is in a sub-healthy state, it indicates that latent defects inside the converter transformer have begun to develop. The uncontrolled fault leakage current component in the measured current gradually accumulates, causing a significant directional drift in the residual sequence. However, since the equipment is under low load conditions at this time, the absolute amplitude of the measured current (only about 40mA) is extremely far from the static alarm limit of hundreds of milliamperes. This phenomenon of weak fault signals being masked by the low-level operating conditions will lead to the risk of latent defects being missed. Therefore, after judging that the converter transformer is in a sub-healthy state, the system immediately diagnoses that the equipment has experienced latent defects such as multi-point grounding or sensor feedback device failure, and drives the output of high-level early warning commands.

[0077] The solutions of the embodiments of the present invention will be described in detail and explained below with reference to specific application examples: Reference Figure 5 , Figure 5 A flowchart of a method for dynamic identification of converter transformer anomalies considering operating conditions, provided in another embodiment of this application.

[0078] In this embodiment, the actual operation and monitoring sample data of the low YDC converter transformer at pole 2 of a certain UHVDC converter station are used to conduct in-depth verification and explanation of the embodiments of this application.

[0079] Specifically, in the process of multi-source heterogeneous data fusion and modeling verification, the system fully extracted the synchronous sample clusters of the converter transformer that were in the healthy operating range (to the left of the dividing line).

[0080] like Figure 3 As shown, mapping the absolute value of operating power to the horizontal axis and the core grounding current to the vertical axis, the scatter plot clearly reveals a strong and significant positive correlation between the two. In this embodiment, a second-order polynomial regression algorithm is used to mathematically model this batch of healthy-state data samples (blue dots), accurately fitting a red normal operating condition correlation characteristic model curve that reflects the nonlinear increase of current with load as a power-current benchmark model. This benchmark curve not only has a high degree of fit but also perfectly follows the nonlinear physical criteria of core magnetization and leakage magnetic radiation, forming a healthy-state envelope that adaptively and dynamically changes with operating conditions.

[0081] Next, we enter the real-time status monitoring and expected baseline tracking phase, referring to... Figure 2During a continuous monitoring period spanning dozens of days, the system, based on the real-time input absolute value of the operating power (solid green line) on the secondary coordinate axis, drove the power current benchmark model to continuously calculate and output the theoretical expected current (dashed red line). Comparing this with the measured core grounding current (solid blue line), it can be clearly observed that during the modeling training period, the dynamic baseline perfectly tracked and covered all fluctuations in the real current caused by the fluctuations in operating conditions.

[0082] However, after crossing the monitoring period dividing line, the measured current began to show an abnormal upward trend, deviating from the theoretical benchmark. But because the equipment load was low at this time, the measured grounding current was extremely far from the static alarm limit of hundreds of milliamperes, and traditional monitoring methods were completely in a "blind zone" at this moment.

[0083] In view of this, this embodiment subsequently triggers a dynamic residual verification and very early anomaly identification mechanism, such as Figure 4 As shown, the system calculates the difference between the measured value and the expected value, generating a state residual sequence (brown solid line) that separates the load profile. This state residual sequence is essentially completely decoupled from and stripped of the natural current fluctuation component caused by drastic load fluctuations, becoming a high-purity, high-sensitivity health offset characteristic quantity that reflects the internal insulation evolution and mechanical displacement of the equipment.

[0084] Based on the residual samples during the healthy period, the system automatically establishes adaptive early warning upper and lower limits (red-green horizontal dashed envelope) based on the ±3σ principle.

[0085] Within the healthy operating range, the residual curve exhibits a typical stationary random walk with a mean of zero, tightly constrained within the normal envelope, confirming the method's ability to completely shield against fluctuations under normal operating conditions. However, when latent defects within the equipment begin to develop, the uncontrolled fault leakage current component in the measured current gradually accumulates, causing a significant directional drift in the residual sequence.

[0086] Ultimately, as Figure 4 As shown by the prominently marked red scatter plot, the residuals continuously and strongly exceeded the pre-set +3σ adaptive warning upper limit. At this moment, the system achieved precise tuning, successfully capturing early signs of latent defects from subtle abnormal deviations. This embodiment conclusively demonstrates that the present invention completely overturns the technical limitations of static thresholds, utilizing a dual-drive approach of data mining and physical constraints to achieve highly sensitive and robust intelligent identification of the status of heavy-asset equipment at extremely low cost.

[0087] In summary, the converter transformer anomaly dynamic identification method considering operating conditions proposed in this application accurately isolates operating condition interference by constructing a power-current physical correlation model driven by multi-source data, thereby enabling the keen capture of early latent abnormal states of the converter transformer.

[0088] Furthermore, the method presented in this application is highly feasible and edge computing-friendly, requiring no additional complex acoustic or optical sensing hardware in areas with strong electromagnetic interference. It simply leverages the existing SCADA power and online current data of the converter station. The multinomial regression and residual statistical algorithm matrix used has extremely low computational complexity, perfectly matching the current computing power constraints of the substation edge computing gateway, and has extremely broad prospects for large-scale promotion.

[0089] Reference Figure 6 This application also provides a converter transformer anomaly dynamic identification system that takes into account operating conditions, which can implement the above-mentioned method. The system includes: The first module is used to synchronously acquire online operating power data of the converter transformer and online grounding current data of the main grounding system and perform timestamp alignment.

[0090] The second module is used to input online operating power data into a trained power current benchmark model to predict the current and obtain the theoretical expected value of the grounding current under the current operating conditions.

[0091] The third module is used to perform operating condition decoupling processing on the online grounding current data based on the theoretical expected value of the grounding current, and obtain the state residual sequence of stripped load profile.

[0092] The fourth module is used to adaptively identify anomalies in the state residual sequence based on preset anomaly diagnosis conditions and obtain the judgment result.

[0093] It is understood that the content of the above method embodiments is applicable to this system embodiment. The specific functions implemented in this system embodiment are the same as those in the above method embodiments, and the beneficial effects achieved are also the same as those achieved in the above method embodiments.

[0094] This application also provides an electronic device, which includes a memory and a processor. The memory stores a computer program, and the processor executes the computer program to implement the above-described method. This electronic device can be any smart terminal, including tablet computers, in-vehicle computers, etc.

[0095] It is understood that the content of the above method embodiments is applicable to this device embodiment. The specific functions implemented by this device embodiment are the same as those of the above method embodiments, and the beneficial effects achieved are also the same as those achieved by the above method embodiments.

[0096] Reference Figure 7 , Figure 7 The hardware structure of an electronic device according to another embodiment is illustrated. The electronic device includes: The processor 901 can be implemented using a general-purpose CPU (Central Processing Unit), microprocessor, application-specific integrated circuit (ASIC), or one or more integrated circuits, and is used to execute relevant programs to implement the technical solutions provided in the embodiments of this application.

[0097] The memory 902 can be implemented as a read-only memory (ROM), static storage device, dynamic storage device, or random access memory (RAM). The memory 902 can store the operating system and other application programs. When the technical solutions provided in the embodiments of this specification are implemented through software or firmware, the relevant program code is stored in the memory 902 and is called and executed by the processor 901 using the methods described in the embodiments of this application.

[0098] The input / output interface 903 is used to implement information input and output.

[0099] The communication interface 904 is used to enable communication and interaction between this device and other devices. Communication can be achieved through wired means (such as USB, Ethernet cable, etc.) or wireless means (such as mobile network, WIFI, Bluetooth, etc.).

[0100] Bus 905 transmits information between various components of the device, such as processor 901, memory 902, input / output interface 903, and communication interface 904.

[0101] The processor 901, memory 902, input / output interface 903, and communication interface 904 are connected to each other within the device via bus 905.

[0102] This application also provides a computer-readable storage medium storing a computer program that, when executed by a processor, implements the above-described method.

[0103] It is understood that the content of the above method embodiments is applicable to this storage medium embodiment. The specific functions implemented in this storage medium embodiment are the same as those in the above method embodiments, and the beneficial effects achieved are also the same as those achieved in the above method embodiments.

[0104] This application also provides a computer program product, including a computer program that, when executed by a processor, implements the above-described method.

[0105] It is understood that the content of the above method embodiments is applicable to the embodiments of this program product. The specific functions implemented by the embodiments of this program product are the same as those of the above method embodiments, and the beneficial effects achieved are also the same as those achieved by the above method embodiments.

[0106] Memory, as a non-transitory computer-readable storage medium, can be used to store non-transitory software programs and non-transitory computer-executable programs. Furthermore, memory may include high-speed random access memory, and may also include non-transitory memory, such as at least one disk storage device, flash memory device, or other non-transitory solid-state storage device. In some embodiments, memory may optionally include memory remotely located relative to the processor, and these remote memories can be connected to the processor via a network. Examples of such networks include, but are not limited to, the Internet, intranets, local area networks, mobile communication networks, and combinations thereof.

[0107] The embodiments described in this application are for the purpose of more clearly illustrating the technical solutions of the embodiments of this application, and do not constitute a limitation on the technical solutions provided by the embodiments of this application. As those skilled in the art will know, with the evolution of technology and the emergence of new application scenarios, the technical solutions provided by the embodiments of this application are also applicable to similar technical problems.

[0108] Those skilled in the art will understand that the technical solutions shown in the figures do not constitute a limitation on the embodiments of this application, and may include more or fewer steps than shown, or combine certain steps, or different steps.

[0109] The system embodiments described above are merely illustrative. The units described as separate components may or may not be physically separate; that is, they may be located in one place or distributed across multiple network units. Some or all of the modules can be selected to achieve the purpose of this embodiment according to actual needs.

[0110] Those skilled in the art will understand that all or some of the steps in the methods disclosed above, as well as the functional modules / units in the systems and devices, can be implemented as software, firmware, hardware, or suitable combinations thereof.

[0111] The terms “first,” “second,” “third,” “fourth,” etc. (if present) in the specification and accompanying drawings of this application are used to distinguish similar objects and are not necessarily used to describe a specific order or sequence. It should be understood that such data can be interchanged where appropriate so that the embodiments of this application described herein can be implemented in orders other than those illustrated or described herein. Furthermore, the terms “comprising” and “having,” and any variations thereof, are intended to cover non-exclusive inclusion; for example, a process, method, system, product, or apparatus that comprises a series of steps or units is not necessarily limited to those steps or units explicitly listed, but may include other steps or units not explicitly listed or inherent to such processes, methods, products, or apparatus.

[0112] It should be understood that in this application, "at least one (item)" means one or more, and "more than" means two or more. "And / or" is used to describe the relationship between related objects, indicating that three relationships can exist. For example, "A and / or B" can represent three cases: only A exists, only B exists, and both A and B exist simultaneously, where A and B can be singular or plural. The character " / " generally indicates that the preceding and following related objects are in an "or" relationship. "At least one (item) of the following" or similar expressions refer to any combination of these items, including any combination of single or plural items. For example, at least one (item) of a, b, or c can represent: a, b, c, "a and b", "a and c", "b and c", or "a and b and c", where a, b, and c can be single or multiple.

[0113] In the embodiments provided in this application, it should be understood that the disclosed systems and methods can be implemented in other ways. For example, the system embodiments described above are merely illustrative; for instance, the division of the units described above 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 systems or units may be electrical, mechanical, or other forms.

[0114] The units described above as separate components may or may not be physically separate. The components shown as units may or may not be physical units; that is, 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 according to actual needs.

[0115] Furthermore, 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. The integrated unit can be implemented in hardware or as a software functional unit.

[0116] If the integrated unit is implemented as a software functional unit and sold or used as an independent product, it 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 all or part 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 multiple 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 of the various embodiments of this application. The aforementioned storage medium includes various media capable of storing programs, such as USB flash drives, portable hard drives, read-only memory (ROM), random access memory (RAM), magnetic disks, or optical disks.

[0117] The preferred embodiments of the present application have been described above with reference to the accompanying drawings, but this does not limit the scope of the claims of the present application. Any modifications, equivalent substitutions, and improvements made by those skilled in the art without departing from the scope and substance of the embodiments of the present application shall be within the scope of the claims of the present application.

Claims

1. A method for dynamic identification of converter transformer anomalies considering operating conditions, characterized in that, The method includes the following steps: Simultaneously acquire online operating power data of the converter transformer and online grounding current data of the main grounding system, and align the timestamps. The online operating power data is input into the trained power current benchmark model to predict the current and obtain the theoretical expected value of the ground current under the current operating conditions. Based on the theoretical expected value of the grounding current, the online grounding current data is subjected to operating condition decoupling processing to obtain a state residual sequence stripped of the load profile; The state residual sequence is adaptively anomaly identified based on preset anomaly diagnosis conditions to obtain a judgment result.

2. The method according to claim 1, characterized in that, The power current reference model is obtained through the following steps: Obtain a historical health sample set of the converter transformer and the main grounding system in a healthy operating range, wherein the historical health sample set includes historical operating power data and historical grounding current data; The historical operating power data is used as an input variable to characterize the combined electromagnetic and thermal operating conditions, and the historical grounding current data is used as an output variable to characterize the overall health status of the system. Correlation features are extracted and modeled on the input variables to obtain the power-current benchmark model.

3. The method according to claim 2, characterized in that, The process of using historical operating power data as an input variable characterizing the combined electromagnetic and thermal operating conditions, and historical grounding current data as an output variable characterizing the overall health status of the system, and extracting and modeling the correlation features of the input variables to obtain a power-current benchmark model, includes the following steps: The historical operating power data is mapped onto the horizontal axis, and the historical grounding current data is mapped onto the vertical axis to obtain a scatter plot; The physical coupling law of the scatter plot is nonlinearly fitted using a polynomial regression algorithm to form a power current benchmark model, wherein the power current benchmark model is a characteristic curve that reflects the nonlinear increase of current with load.

4. The method according to claim 1, characterized in that, The step of performing operating condition decoupling processing on the online grounding current data based on the theoretical expected value of the grounding current to obtain a state residual sequence stripped of the load profile includes the following steps: The theoretical expected value of the grounding current is timestamped and sorted with the online grounding current data to obtain a pairing sequence, wherein the theoretical expected value of the grounding current and the online grounding current data with the same timestamp are a pair of paired data; The paired data is subjected to operating condition fundamental wave filtering to remove the natural current fluctuation component caused by load fluctuation, thereby obtaining a health offset characteristic quantity that reflects the internal insulation evolution and mechanical displacement of the equipment. The health offset features are sorted according to timestamp order to obtain the state residual sequence.

5. The method according to claim 2, characterized in that, The adaptive anomaly identification of the state residual sequence based on preset anomaly diagnosis conditions to obtain a judgment result includes the following steps: Based on the distribution characteristics of health conditions in the historical health sample set, mathematical statistics are performed to obtain the dynamic early warning upper and lower limit threshold bands. The preset abnormal diagnosis conditions are determined based on the upper and lower threshold bands of the warning and the state residual sequence. The abnormal diagnosis conditions are that the state residual sequence shows a monotonically divergent trend and breaks through the boundary of the dynamic warning upper and lower threshold bands. If the abnormal diagnosis conditions are not met, the judgment result is determined to be that the converter transformer is in the normal operating condition fluctuation range. If the abnormal diagnostic conditions are met, the judgment result is determined to be that the converter transformer is in a sub-healthy state.

6. The method according to claim 5, characterized in that, The step of performing mathematical statistics on the distribution characteristics of health conditions based on the historical health sample set to obtain dynamic early warning upper and lower limit threshold bands includes the following steps: Calculate the health residual sequence of health conditions based on the historical health sample set; Mathematical statistics were performed on the health residual sequence to obtain the health mean and health standard deviation; Based on the statistical principle of normal distribution, the boundaries of the health interval are calculated according to the health mean and the health standard deviation to obtain the dynamic early warning upper and lower limit threshold bands.

7. The method according to claim 5, characterized in that, The method further includes the following steps: When the judgment result indicates that the converter transformer is in a sub-healthy state, an early warning command is output, wherein the early warning command is used to perform predictive maintenance for latent defects.

8. A converter transformer anomaly dynamic identification system considering operating conditions, characterized in that, The system includes: The first module is used to synchronously acquire online operating power data of the converter transformer and online grounding current data of the main grounding system and perform timestamp alignment. The second module is used to input the online operating power data into the trained power current benchmark model to predict the current and obtain the theoretical expected value of the ground current under the current operating conditions. The third module is used to perform operating condition decoupling processing on the online grounding current data according to the theoretical expected value of the grounding current, and to obtain a state residual sequence stripped of the load profile. The fourth module is used to adaptively identify anomalies in the state residual sequence based on preset anomaly diagnosis conditions, and obtain a judgment result.

9. An electronic device, characterized in that, The electronic device includes a memory and a processor, the memory storing a computer program, and the processor executing the computer program to implement the method according to any one of claims 1 to 7.

10. A computer-readable storage medium storing a computer program, characterized in that, When the computer program is executed by a processor, it implements the method of any one of claims 1 to 7.

Images