Method and system for detecting winding defects of a converter transformer considering dc bias and multi-physical field bifurcation

Abstract

This invention relates to the field of winding defect diagnosis technology for converter transformers in ultra-high voltage direct current (UHVDC) transmission systems. It discloses a method and system for detecting converter transformer winding defects that considers DC bias and multi-physics twins. The invention relies on a reference temperature cloud map constructed using a multi-physics twin model, providing an accurate and interference-free reference basis for thermal field analysis. Through operating condition identification and interference stripping, it effectively eliminates multi-physics distortion interference under special operating conditions of the converter transformer, avoiding the masking and interference of operating condition fluctuations on defect characteristics. Simultaneously, through multi-dimensional fusion detection of defect feature thermal field cloud maps and electrical operating data, it not only improves the sensitivity of identifying early latent defects but also accurately distinguishes between normal operating condition fluctuations and true defect characteristics, fundamentally reducing the probability of misjudgment and missed judgment, and significantly improving the accuracy and reliability of converter transformer winding defect detection.

Description

Technical Field

[0001] This invention relates to the field of winding defect diagnosis technology for converter transformers in ultra-high voltage direct current transmission systems, and particularly to a method and system for detecting winding defects in converter transformers that takes into account DC bias and multi-physics twins. Background Technology

[0002] High-voltage direct current (HVDC) transmission technology, with its significant advantages such as rapid and flexible power regulation and low transmission loss, has become the core technology path for modern inter-regional energy interconnection and clean energy transmission, occupying an increasingly important position in the global power system. As the core hub equipment of the HVDC transmission system, the converter transformer undertakes the crucial functions of providing commutation reactance, regulating AC voltage, and achieving AC-DC electrical isolation. Its operating conditions are more unique and demanding than those of conventional AC power transformers: it not only needs to adapt to greater short-circuit impedance design requirements but also needs to withstand the superposition of complex electrical stresses such as AC-DC composite voltage, polarity reversal transient voltage, and high-order harmonic currents over long periods. The equipment is constantly in an extreme service environment with multi-physics field coupling distortion.

[0003] Under long-term, complex operating conditions, converter transformers are prone to latent structural faults such as inter-turn short circuits in their windings. Furthermore, the evolution of these faults is highly insidious, with early, minute defects often going undetected. If these defects are not identified and addressed in advance, they will gradually deteriorate, eventually leading to serious accidents such as insulation breakdown, resulting in unplanned equipment outages and posing a significant threat to the safe and stable operation of the UHVDC power grid. Therefore, achieving accurate and efficient detection of early defects in converter transformer windings is a core technological requirement for preventing equipment failures and ensuring the reliable operation of the DC power grid. It is also a key research direction and engineering challenge in the field of power equipment condition monitoring.

[0004] Currently, the industry's methods for detecting defects in converter transformer windings largely follow the design concepts of conventional AC power transformers, primarily relying on single physical field characteristic comparison, post-event transient protection actions, and static offline monitoring. These methods are not adequately adapted to the unique and complex operating conditions of converter transformers. They lack sufficient sensitivity in identifying early latent defects, are susceptible to interference from multi-physical field distortions in the equipment's background, leading to misjudgments. They struggle to accurately distinguish between normal anomalies caused by operating condition fluctuations and genuine fault characteristics resulting from defects, failing to reliably identify and locate early-stage hidden dangers. This severely restricts the accuracy and operational safety of condition-based maintenance of converter transformers, and fails to meet the high reliability requirements of UHVDC transmission systems for equipment operation and maintenance. Summary of the Invention

[0005] This invention provides a method and system for detecting defects in converter transformer windings that takes into account DC bias and multi-physics twins, solving the technical problem of how to improve the accuracy and reliability of converter transformer winding defect detection.

[0006] The first aspect of this invention provides a method for detecting defects in converter transformer windings that considers DC bias magnetism and multi-physics twinning, comprising: A multi-physics twin model of the converter transformer windings was constructed and electromagnetic and thermal coupling simulations were performed to obtain a reference temperature cloud map under unbiased magnetic conditions. Based on the reference temperature cloud map, the thermal field temperature data of the converter transformer is obtained, and the electrical operation data of the converter transformer is collected simultaneously. Using the thermal field space temperature data, the converter transformer electrical operation data, and the reference temperature cloud map, operating conditions are identified and interference is removed to generate a defect feature thermal field cloud map without bias magnetic interference. The winding defect detection results of the converter transformer are generated by using the thermal field cloud map of the defect features and the electrical operation data of the converter transformer.

[0007] Optionally, the construction of a multi-physics twin model of the converter transformer windings and the performance of electromagnetic and thermal coupling simulations to obtain a reference temperature contour map under unbiased magnetic conditions include: Using a single copper conductor and its insulating paper wrapping in a converter transformer winding as the smallest calculation unit, a multi-physical twin model of the winding is constructed. The multi-physics twin model of the winding is meshed, and the local loss density of various currents in each mesh cell is solved by an electromagnetic frequency domain solver. The total loss density of each grid cell is obtained by spatially superimposing multiple local loss densities within the same grid cell. Obtain the winding electrical operation data and winding geometric structure data of the converter transformer, and obtain the magnetic induction intensity of the winding multi-physics twin model; The reference total loss of the winding is calculated using the winding electrical operating data, the winding geometric data, and the magnetic induction intensity. The winding electrical operating data includes the effective value of the winding current, the overall resistance value of the winding, the conductivity of the winding conductor, and the current angular frequency. The winding geometric data includes the longitudinal wire gauge dimensions, the transverse wire gauge dimensions, and the total winding volume. The total loss density of each grid cell is integrated over the entire winding domain to obtain the total winding simulation loss. Calculate the target difference between the reference total loss of the winding and the simulated total loss of the winding. When the target difference is within a preset error range, perform heat source mapping on the total loss density of each grid cell. Based on the heat source mapping results, the multi-physics twin model of the winding is subjected to bidirectional coupling iteration of multi-physics fields until the residual convergence is achieved, and the reference temperature cloud map under the unbiased magnetic condition is obtained.

[0008] Optionally, the step of acquiring the thermal field temperature data of the converter transformer based on the reference temperature cloud map and simultaneously collecting the electrical operation data of the converter transformer includes: The region with the largest temperature gradient is extracted from the reference temperature cloud map and designated as a high-risk thermal field monitoring zone. According to the high-risk monitoring zone of the thermal field, non-invasive temperature sensors are used to collect the thermal field temperature data of the converter transformer. Simultaneously collect electrical operation data of the converter transformer; The electrical operating data of the converter transformer includes the real-time current on the grid side, the real-time current on the valve side, the real-time voltage on the grid side, the real-time parameter voltage on the valve side, and the measured value of the neutral point DC current.

[0009] Optionally, the step of using the thermal field space temperature data, the converter transformer electrical operation data, and the reference temperature cloud map to perform operating condition identification and interference removal, and generating a defect feature thermal field cloud map without magnetic interference, includes: Spatiotemporal matching is performed between the thermal field space temperature data, the converter transformer electrical operation data, and the reference temperature cloud map; A measured thermal field cloud map is constructed using the spatial temperature data of the thermal field after spatiotemporal matching, and the measured thermal field cloud map is compared with the reference temperature cloud map after spatiotemporal matching to obtain the thermal field dimension determination result. Determine whether the electrical operation data of the converter transformer after spatiotemporal matching meets the preset electrical dimension judgment conditions, and obtain the electrical dimension judgment result; When the thermal field dimension determination result is that the thermal field conforms to the pure bias magnetic characteristics, and the electrical dimension determination result is that the electrical conforms to the pure bias magnetic characteristics, then the converter transformer is determined to be in a pure DC bias magnetic disturbance condition. If the thermal field dimension determination result is that the thermal field does not conform to the pure bias magnetic characteristics, or the electrical dimension determination result is that the electrical does not conform to the pure bias magnetic characteristics, then the converter transformer is determined to be in a combined bias magnetic and defect condition. The measured spatial temperature of each spatial coordinate is extracted from the measured thermal field cloud map, and the reference spatial temperature of each spatial coordinate is extracted from the reference temperature cloud map after spatiotemporal matching. The magnetic interference was stripped using the measured spatial temperature at each spatial coordinate and the reference spatial temperature that is consistent with the spatial coordinate, and the global residual temperature distribution was obtained. Noise extraction and feature verification are performed on the global residual temperature distribution to generate a defect feature thermal field cloud map without bias magnetic interference.

[0010] Optionally, the step of using the defect feature thermal field cloud map and the electrical operation data of the converter transformer to perform winding defect detection and generate the winding defect detection result of the converter transformer includes: The thermal field defect feature index is extracted from the defect feature thermal field cloud map. The thermal field defect feature index includes the peak temperature of the defect hot spot, the diffusion area of ​​the defect hot spot, and the thermal field temperature rise gradient. Electrical defect characteristic indicators are extracted from the electrical operation data of the converter transformer. The electrical defect characteristic indicators include the differential current step change value between the grid side and the valve side of the converter transformer, the asymmetric negative sequence current surge, and the voltage harmonic distortion rate. The thermal defect characteristic indicators and the electrical defect characteristic indicators are synchronized in time and space; Based on the preset fusion judgment rules, the thermal field defect characteristic indicators after spatiotemporal synchronization and the electrical defect characteristic indicators after spatiotemporal synchronization are coupled and judged to generate the winding defect detection results of the converter transformer. The winding defect detection results include the actual existence of the winding defect, the defect type, and the severity level; The preset fusion judgment rule is specifically that if the thermal field defect characteristic index and the electrical defect characteristic index have the same defect level and are synchronized in time and space, then it is judged that a winding defect actually exists. If either index is not satisfied, it is judged as an interference signal, and the defect warning is blocked.

[0011] Optionally, it also includes: When the severity level of the winding defect detection result is severe, a power outage electrical test is performed, and a converter transformer winding defect diagnosis report is generated based on the test results.

[0012] A second aspect of the present invention provides a converter transformer winding defect detection system that considers DC bias magnetism and multi-physics twinning, comprising: The module is used to build a multi-physics twin model of the converter transformer windings and perform electromagnetic and thermal coupling simulations to obtain a reference temperature cloud map under unbiased magnetic conditions. The acquisition module is used to acquire the thermal field temperature data of the converter transformer based on the reference temperature cloud map, and simultaneously collect the electrical operation data of the converter transformer. The processing module is used to identify operating conditions and remove interference by using the thermal field space temperature data, the electrical operation data of the converter transformer and the reference temperature cloud map, and generate a defect feature thermal field cloud map without bias magnetic interference. The detection module is used to perform winding defect detection using the defect feature thermal field cloud map and the electrical operation data of the converter transformer, and generate the winding defect detection result of the converter transformer.

[0013] The third aspect of the present invention provides an electronic device, including a memory and a processor, wherein the memory stores a computer program, and when the computer program is executed by the processor, the processor causes the processor to perform the steps of the converter transformer winding defect detection method considering DC bias and multi-physics twin as described above.

[0014] The fourth aspect of the present invention provides a computer-readable storage medium having a computer program stored thereon, which, when executed, implements the converter transformer winding defect detection method as described above, taking into account DC bias and multi-physics twins.

[0015] The fifth aspect of the present invention provides a computer program product, the computer program product comprising a computer program stored on a non-transitory computer-readable storage medium, the computer program comprising program instructions, wherein, when the program instructions are executed by a computer, the computer performs the converter transformer winding defect detection method as described above, which takes into account DC bias and multi-physics twins.

[0016] As can be seen from the above technical solutions, the present invention has the following advantages: This invention provides a method and system for detecting defects in converter transformer windings, taking into account DC bias and multi-physics twinning. By constructing a multi-physics twin model of the converter transformer winding and conducting bidirectional electromagnetic and thermal field coupling simulations, a reference temperature cloud map under bias-free operating conditions is accurately obtained. Based on this, corresponding thermal field temperature data and electrical operation data are collected. Then, through operating condition identification and interference removal processing, complex operating condition interferences are eliminated, generating a defect feature thermal field cloud map without bias interference. Finally, the defect feature thermal field cloud map and electrical operation data are fused to achieve accurate detection of winding defects. This invention relies on the reference temperature cloud map constructed using a multi-physics twin model, providing an accurate and interference-free reference basis for thermal field analysis. Through operating condition identification and interference removal… This method effectively eliminates multi-physics distortion interference under special operating conditions of converter transformers, avoiding the masking and interference of operating condition fluctuations on defect characteristics. At the same time, through multi-dimensional fusion detection of defect feature thermal field cloud maps and electrical operation data, it not only improves the sensitivity of early latent defects but also accurately distinguishes between normal operating condition fluctuations and real defect characteristics. This fundamentally reduces the probability of misjudgment and missed judgment, significantly improving the accuracy and reliability of converter transformer winding defect detection. It overcomes the shortcomings of existing detection methods, such as difficulty in adapting to the special and complex operating conditions of converter transformers, susceptibility to interference leading to misjudgment, and insufficient sensitivity for early defect identification. This provides a solid technical guarantee for the safe operation and maintenance of converter transformers and the stable operation of UHVDC power grids. Attached Figure Description

[0017] To more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0018] Figure 1 The flowchart illustrates the steps of a converter transformer winding defect detection method that takes into account DC bias and multi-physics twinning, as provided in Embodiment 1 of the present invention. Figure 2 This is a flowchart of the steps of a converter transformer winding defect detection method considering DC bias and multi-physics twinning provided in Embodiment 2 of the present invention; Figure 3 This is a schematic diagram of the converter transformer winding structure provided in Embodiment 2 of the present invention; Figure 4 This is a structural block diagram of a converter transformer winding defect detection system that takes into account DC bias and multi-physics twinning, as provided in Embodiment 3 of the present invention. Figure 5 This is a structural block diagram of an electronic device provided in Embodiment 4 of the present invention. Detailed Implementation

[0019] This invention provides a method and system for detecting defects in converter transformer windings that takes into account DC bias and multi-physics twins, in order to solve the technical problem of how to improve the accuracy and reliability of converter transformer winding defect detection.

[0020] To make the objectives, technical solutions, and advantages of the embodiments of the present invention clearer, 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, 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.

[0021] The converter transformer winding defect detection method and system proposed in this invention, based on the inversion of bias magnetic decoupling thermal characteristics and electrical features, consists of two core parts: online multi-physics field twin decoupling early warning and offline steady-state electrical test diagnosis, forming a complete diagnostic closed loop from early defect perception to physical damage assessment.

[0022] At the core technology implementation level, this invention first constructs a multi-physical twin model of the winding, using a single copper conductor and the insulating paper wrapping as the smallest computational unit. A heterogeneous mesh partitioning strategy is employed to balance computational accuracy and efficiency. Electromagnetic and thermal coupling simulations are used to obtain a reference temperature cloud map under unbiased magnetic conditions. This model simultaneously incorporates measured values ​​of the neutral point DC current of the converter transformer, outputting dynamic reference temperature spatial distribution characteristics that take into account the current bias level, providing a precise reference for subsequent thermal interference removal. Subsequently, based on the reference temperature cloud map, high-risk monitoring areas of the thermal field are identified. Non-invasive temperature sensors are used to collect targeted thermal field data, simultaneously acquiring electrical operating data such as grid-side and valve-side current, voltage, and neutral point DC current. Through spatiotemporal matching and operating condition identification of the thermal field data, electrical data, and reference cloud map, the global heating background caused by bias is removed, generating a residual thermal field that only reflects the winding's own defects. Local abnormal temperature rises are captured in the residual thermal field, and combined with electrical transient change characteristics, early warning of latent defects is achieved, while actively shielding false alarm signals under pure bias conditions. When the defect severity level is severe, further steady-state electrical tests such as DC resistance and voltage ratio of the grid-side / valve-side windings are performed. By inversely calculating the test data and winding structure parameters, physical defects are quantified and located, completing the logical closed loop from online multi-field sensing to offline physical damage assessment.

[0023] Compared with existing technologies, this invention has the following significant advantages: First, this invention is more suitable for the complex AC / DC combined operating conditions of converter transformers, breaking away from the traditional detection framework of AC transformers that relies on static reference feature comparison. By constructing a multi-physics twin model that considers DC bias boundary conditions, it provides a dynamic, interference-free reference for defect detection, effectively solving the problem that existing technologies easily misjudge bias heating as internal defects under bias conditions. Second, this invention is more sensitive to early latent defects in windings. By integrating the thermal field defect characteristics without bias interference with electrical transient change characteristics, it can detect defects even when insulation deteriorates, without waiting for the fault current to accumulate to the stage that triggers the traditional differential protection operation. Early defect warning is achieved, avoiding misjudgment and missed judgment based on single feature determination. Thirdly, this invention constructs a non-invasive online monitoring and offline electrical testing linkage mechanism in the physical defect diagnosis stage, abandoning the destructive practice of densely pre-embedding sensors inside the windings. Through the inversion calculation of steady-state electrical test data, it can not only qualitatively determine the existence of defects, but also quantify the degree of physical damage (such as the number of broken strands in the conductor) and accurately locate the defect location (grid-side / valve-side windings, tap changers, etc.). Compared with the existing online monitoring system that can only provide qualitative alarms, it achieves a deep integration from "early warning" to "damage assessment", providing reliable technical support for the safe operation and maintenance and fault prevention of converter transformers.

[0024] Please see Figure 1 , Figure 1The flowchart illustrates the steps of a converter transformer winding defect detection method that considers DC bias and multi-physics twinning, as provided in Embodiment 1 of the present invention.

[0025] This invention provides a method for detecting defects in converter transformer windings that considers DC bias and multi-physics twinning, comprising: Step 101: Construct a multi-physics twin model of the converter transformer windings and perform electromagnetic and thermal coupling simulation to obtain a reference temperature cloud map under unbiased magnetic conditions.

[0026] In this embodiment of the invention, based on the actual geometric structure, electrical parameters and operating characteristics of the converter transformer winding, a multi-physics twin model of the winding is constructed that can accurately characterize the electromagnetic and thermal field coupling relationship of the winding. This model can accurately map the actual structure and physical characteristics of the converter transformer winding. After the model is constructed, the normal operating condition without bias is set, and the electromagnetic and thermal coupling simulation process is started. The global temperature distribution of the converter transformer winding under the condition without bias is obtained through simulation calculation, and then a reference temperature cloud map under the condition without bias is generated.

[0027] Step 102: Obtain the thermal field temperature data of the converter transformer based on the reference temperature cloud map, and simultaneously collect the electrical operation data of the converter transformer.

[0028] In this embodiment of the invention, using the constructed multi-physical twin model of the converter transformer winding and the generated reference temperature cloud map under unbiased magnetic conditions as references, the peak region of the temperature gradient is first identified and extracted from the reference temperature cloud map to determine the high-risk monitoring area of ​​the thermal field of the converter transformer winding. Then, non-invasive temperature sensors are deployed in this high-risk monitoring area to collect the thermal field space temperature data of the converter transformer under actual operating conditions in real time. At the same time, the electrical operation data of the converter transformer is collected synchronously through the online monitoring device of the converter transformer to ensure that the thermal field space temperature data and the electrical operation data are synchronized in the time dimension.

[0029] Step 103: Use thermal field space temperature data, converter transformer electrical operation data and reference temperature cloud map to identify operating conditions and remove interference, and generate a defect feature thermal field cloud map without bias magnetic interference.

[0030] In this embodiment of the invention, the thermal field temperature data, synchronously acquired converter transformer electrical operation data, and a reference temperature cloud map are used as inputs. First, the three are spatiotemporally aligned to ensure that the thermal field data and electrical data are completely matched in time and space coordinates. Then, a measured thermal field cloud map is constructed based on the spatiotemporally matched data, and it is compared and analyzed with the reference temperature cloud map to determine whether there are abnormal temperature field characteristics caused by external interference in the current operating condition from the thermal field dimension. At the same time, based on the synchronously acquired electrical operation data and combined with the preset electrical dimension judgment rules, it is determined whether there are composite interferences or abnormal fluctuations in the current operating condition from the electrical dimension. Based on the dual judgment results of the thermal field dimension and the electrical dimension, the true type of the current operating condition is identified. When it is confirmed that there are specific interference characteristics in the operating condition, the corresponding interference components are removed from the measured thermal field cloud map one by one in space coordinates, based on the reference temperature cloud map, to achieve interference stripping processing of the entire temperature distribution. Finally, a thermal field cloud map with no bias magnetic interference defect characteristics that only reflects the defect characteristics of the winding itself is obtained.

[0031] Step 104: Use the thermal field cloud map of defect features and the electrical operation data of the converter transformer to perform winding defect detection and generate the winding defect detection results of the converter transformer.

[0032] In this embodiment of the invention, the core inputs are the generated thermal field cloud map of the defect features without biased magnetic interference and the synchronously acquired electrical operation data of the converter transformer. First, thermal field feature indicators that can characterize potential defects in the winding are extracted from the defect feature thermal field cloud map. At the same time, corresponding electrical defect feature indicators are extracted from the electrical operation data of the converter transformer. Then, the extracted thermal field defect feature indicators and electrical defect feature indicators are spatiotemporally synchronized to ensure that the two types of feature data are completely corresponding in time and space. Based on the preset fusion judgment rules, the spatiotemporally synchronized thermal field and electrical defect feature indicators are coupled and judged. Only when the defect levels corresponding to the two types of feature indicators are consistent and spatiotemporally synchronized are they determined to be true defects in the winding, and the specific type and severity level of the defect are further determined. If any indicator does not meet the judgment conditions, it is judged as an interference signal and is shielded. Finally, the converter transformer winding defect detection result containing the actual existence of the winding defect, the defect type and severity level are generated, realizing accurate and reliable detection of converter transformer winding defects.

[0033] The converter transformer winding defect detection method considering DC bias and multi-physics twin refers to a method adapted to the AC / DC combined operation of converter transformers, relying on a multi-physics twin model and bias decoupling logic to realize winding defect detection. The winding multi-physics twin model refers to a simulation model that can accurately map the actual structure and physical characteristics of the converter transformer windings and can perform multi-physics coupled calculations. Electromagnetic and thermal coupling simulation refers to a two-way iterative solution process that introduces electromagnetic losses as a heat source into the thermal field calculation and combines temperature field changes with dynamic feedback of electromagnetic parameters. The reference temperature cloud map under no bias condition refers to the converter transformer operating under normal conditions without internal defects and without considering DC bias interference. The following are the theoretical temperature spatial distribution of the entire winding obtained through simulation; thermal field temperature data refers to the temperature acquisition data of the winding spatial points under the actual operating conditions of the converter transformer; converter transformer electrical operation data refers to various electrical parameter data reflecting the real-time electrical operating status of the converter transformer; operating condition identification and interference removal refers to the process of identifying the current operating condition type of the converter transformer and removing interference components such as DC bias; defect feature thermal field cloud map without bias interference refers to the temperature field distribution that reflects only the abnormal temperature rise of the winding itself after removing bias interference; winding defect detection results refer to the detection conclusions that include the existence state, defect type and severity of winding defects.

[0034] In this invention, a multi-physical twin model of the converter transformer winding is first constructed, and electromagnetic and thermal coupling simulations are conducted to obtain a reference temperature cloud map under unbiased magnetic conditions. The twin model accurately maps the actual structure and physical characteristics of the winding. Combined with the reference temperature cloud map generated by the coupling simulation, a reliable reference standard is provided for thermal field analysis and interference removal, avoiding detection errors caused by reference deviations from the source, and laying a precise and interference-free foundation for the entire defect detection process. Subsequently, based on the reference temperature cloud map, high-risk monitoring areas of the thermal field are determined, and thermal field spatial temperature data is collected, while electrical operation data is collected simultaneously. This targeted selection of monitoring areas avoids data redundancy caused by blind collection, while ensuring the spatiotemporal synchronization of thermal field data and electrical data. This provides complete multi-dimensional data support for operating condition identification and defect judgment, enabling subsequent analysis to be conducted based on comprehensive and synchronous data, improving the rationality and efficiency of the detection process. Then, through the analysis of thermal field data... The system performs spatiotemporal alignment, operating condition identification, and interference removal between electrical data and reference temperature cloud maps. Through dual judgment of thermal field and electrical dimensions, it accurately identifies the type of operating condition and eliminates interference components spatially, effectively removing interference signals under complex operating conditions and preventing interference signals from masking defect features. The generated unbiased magnetic interference defect feature thermal field cloud map accurately reflects the winding's defect state, providing precise feature basis for defect detection and allowing for clear identification of defect features. Finally, by fusing the defect feature thermal field cloud map with electrical operating data, it extracts two types of feature indicators and performs spatiotemporal synchronization and coupling judgment. Through the collaborative verification of dual features, it avoids the misjudgment and missed judgment problems that are prone to occur with single feature judgment. A defect is only judged to exist when the two types of feature indicators are consistent and spatiotemporally synchronized, significantly improving the accuracy and reliability of defect identification. It can accurately distinguish between operating condition fluctuations and true defect features, effectively identify early latent defects, and make defect detection more targeted and reliable.

[0035] This invention optimizes the entire process from benchmark establishment and data acquisition to interference removal and defect judgment, ultimately achieving accurate and reliable detection of converter transformer winding defects. It effectively improves the sensitivity, accuracy, and reliability of defect detection, and successfully solves the problems of existing converter transformer winding defect detection methods, which are difficult to adapt to complex operating conditions, are susceptible to interference leading to false positives and false negatives, have insufficient accuracy in early defect identification, and cannot reliably distinguish between operating condition fluctuations and true defect characteristics. This improves the overall level of converter transformer winding defect detection and provides a solid technical guarantee for the safe operation and maintenance of converter transformers, fault prevention and control, and the stable operation of UHVDC power grids.

[0036] Please see Figure 2 , Figure 2 This is a flowchart illustrating the steps of a converter transformer winding defect detection method that considers DC bias and multi-physics twinning, as provided in Embodiment 2 of the present invention.

[0037] This invention provides a method for detecting defects in converter transformer windings that considers DC bias and multi-physics twinning, comprising: Step 201: Construct a multi-physics twin model of the converter transformer windings and perform electromagnetic and thermal coupling simulations to obtain a reference temperature cloud map under unbiased magnetic conditions.

[0038] Please see Figure 3 The diagram shows the winding structure of a converter transformer. It intuitively illustrates the core physical structure of the converter transformer, including the core and the grid-side main winding, valve-side winding, and voltage regulating winding arranged in layers around the core. Each winding forms a multi-layered structure with specific spatial positions, clearly showing the relative positional relationship between the winding and the core. This provides an intuitive structural reference basis for the construction of a refined multi-physical twin model.

[0039] Further, step 201 may include the following sub-steps: S11. Using a single copper conductor and insulating paper wrapping the winding of a converter transformer as the smallest calculation unit, a multi-physical twin model of the winding is constructed.

[0040] In this embodiment of the invention, the traditional equivalent method of treating the winding as a uniform heating block is abandoned. Instead, a single copper wire in the converter transformer winding and the insulating paper wrapped around it are used as the smallest refined calculation unit. Based on the actual winding structure parameters of the converter transformer, the smallest unit is arrayed and structurally spliced ​​to fully restore the actual structural features of the winding with multiple layers, multiple plates, and segmented transposition. This constructs a multi-physics twin model that can accurately map the physical characteristics of the converter transformer winding.

[0041] S12. Mesh the multi-physical twin model of the winding and solve the local loss density of various currents in each mesh cell using an electromagnetic frequency domain solver.

[0042] In this embodiment of the invention, a heterogeneous meshing strategy is used to mesh the multi-physics twin model of the winding: In the electromagnetic field solution domain, adaptive mesh refinement is performed for regions with large leakage magnetic gradients, such as the winding edges, to accurately capture the skin effect and eddy current loss caused by high-frequency harmonics; In the fluid-thermal field solution domain, mesh refinement is performed at the fluid-structure interaction boundary layer of the transformer oil passage, taking into account both computational accuracy and solution efficiency. This heterogeneous meshing strategy can effectively solve the problems of large computational load or local accuracy distortion caused by traditional unified meshes; Subsequently, separate excitation conditions are set sequentially, including only the fundamental current and only each high-frequency harmonic current. Each current is input into the electromagnetic frequency domain solver to solve the magnetic field distribution of each mesh cell under the corresponding condition. Then, combined with parameters such as the conductor conductivity, current angular frequency, and wire gauge size, the local loss density of each mesh cell under the separate excitation is calculated, thus completing the calculation of the local loss density under multiple separate current excitations.

[0043] S13. Spatially superimpose multiple local loss densities within the same grid cell to obtain the total loss density of each grid cell.

[0044] In this embodiment of the invention, based on the superposition principle, the local loss densities of the fundamental current and each high-frequency harmonic current under individual excitation within the same grid cell are spatially vector superimposed to obtain the total loss density of the grid cell under comprehensive current excitation, thereby achieving a refined characterization of the loss distribution across the entire winding.

[0045] S14. Obtain the winding electrical operation data and winding geometric structure data of the converter transformer, and obtain the magnetic induction intensity of the winding multi-physical twin model.

[0046] In this embodiment of the invention, the electrical operation data of the winding is obtained through the online monitoring system of the converter transformer, including the effective value of the winding current, the overall resistance value of the winding, the conductivity of the winding conductor, and the current angular frequency. At the same time, the measured value of the DC current at the neutral point of the converter transformer is introduced simultaneously. The geometric structure data of the winding is obtained according to the design drawings of the converter transformer, including the longitudinal wire gauge dimensions, the transverse wire gauge dimensions, and the total volume of the winding. The magnetic induction intensity distribution of the entire domain of the multi-physical twin model of the winding is extracted through the simulation calculation results of the electromagnetic frequency domain solver.

[0047] S15. Calculate the reference total loss of the winding using the winding electrical operation data, winding geometric structure data, and magnetic induction intensity. The winding electrical operation data includes the effective value of the winding current, the overall resistance value of the winding, the conductivity of the winding conductor, and the current angular frequency; the winding geometric structure data includes the longitudinal wire gauge dimensions, the transverse wire gauge dimensions, and the total volume of the winding.

[0048] In this embodiment of the invention, the reference total winding loss is calculated based on the winding loss composition formula: First, calculate the resistance loss:

[0049] in, For resistive loss, This is the effective value of the winding current. This represents the overall resistance value of the winding. Next, calculate the eddy current loss:

[0050] in, For eddy current losses, and These are longitudinal and transverse eddy current losses, respectively. The conductivity of the winding conductor. It is the angular frequency of the current. It represents the magnetic flux density. This refers to the total volume of the winding. and These are the longitudinal and transverse gauge dimensions, respectively; The final reference total winding loss is:

[0051] in, This represents the reference total loss of the winding.

[0052] S16. Integrate the total loss density of each grid cell over the entire winding domain to obtain the total loss of the winding simulation.

[0053] In this embodiment of the invention, the total loss density of all grid cells in the entire domain of the multi-physical twin model of the winding is integrated by volume, and the total simulated loss of the winding under the current operating condition is obtained by summing them up, thereby realizing the quantitative statistics of the model's calculated loss.

[0054] S17. Calculate the target difference between the reference total loss of the winding and the simulated total loss of the winding. When the target difference is within the preset error range, the total loss density of each grid cell is mapped to a heat source.

[0055] In this embodiment of the invention, the target difference between the reference total loss of the winding and the simulated total loss is calculated. When the target difference is within a preset error range (e.g., the target difference ≤ 5%), the accuracy of the model loss calculation is confirmed to meet the requirements. Subsequently, through node mapping, the total loss density of each grid cell is accurately and energy-conservedly mapped to the volume heat source excitation in the flow-heat model grid cell, thus completing the precise conversion of electromagnetic loss to thermal field heat source.

[0056] If the target difference is not within the preset error range (e.g., target difference > 5%), proceed to step S11.

[0057] S18. Based on the heat source mapping results, the multi-physics twin model of the winding is subjected to bidirectional coupling iteration of multi-physics fields until the residual convergence is achieved, and the reference temperature cloud map under the unbiased magnetic condition is obtained.

[0058] In this embodiment of the invention, a multi-physics bidirectional coupling iteration is initiated based on the excitation of a bulk heat source: the thermal solver calculates the winding temperature field according to the heat source distribution, and then feeds back the temperature field changes to the electromagnetic solver in real time, dynamically updating the kinematic viscosity of the insulating oil and the local resistivity of the winding, and iterating repeatedly until the temperature field residual converges (i.e., the rate of change of the temperature field is less than the preset convergence threshold); finally, the reference temperature cloud map of the non-biased magnetic operating condition, taking into account the current DC biased magnetic boundary conditions, is calculated and output. This cloud map reflects the theoretical temperature spatial distribution of the converter transformer under the current specific AC / DC operating conditions and when there are no defects in the internal structure, providing a comparison benchmark for defect feature extraction.

[0059] The smallest computational unit refers to the refined basic modeling unit based on a single copper conductor and insulating paper used in constructing a multi-physical twin model of a winding; mesh generation refers to the spatial meshing operation performed on the multi-physical twin model of the winding; electromagnetic frequency domain solver refers to the computational tool used to solve for electromagnetic field distribution and loss parameters in the frequency domain; multiple currents refer to the fundamental current and various high-frequency harmonic currents; local loss density refers to the loss distribution value within each mesh unit under a single current excitation; total loss density refers to the comprehensive loss distribution value after superimposing the local loss densities corresponding to multiple currents within the same mesh unit; winding electrical operation data refers to electrical parameters such as current, resistance, conductivity, and angular frequency used to calculate the reference total loss of the winding; winding geometric structure data refers to the winding wire gauge used for modeling and loss calculation. Structural parameters such as dimensions and overall volume; magnetic induction intensity, which refers to the magnetic field intensity distribution value of the entire domain of the multiphysics twin model of the winding; the reference total loss of the winding, which refers to the theoretical total loss of the winding calculated based on electrical, geometric, and magnetic field parameters; the simulated total loss of the winding, which refers to the simulated total loss value obtained by integrating the total loss density of the mesh cells over the entire domain; the target difference, which refers to the difference between the reference total loss of the winding and the simulated total loss of the winding; the preset error range, which refers to the pre-set error range used to determine the accuracy of the model calculation; heat source mapping, which refers to the process of converting the total loss density of the mesh cells into the volume heat source required for thermal field calculation; multiphysics bidirectional coupling iteration, which refers to the process of mutual feedback and cyclic calculation between the electromagnetic field and the thermal field until a stable solution is obtained; residual convergence, which refers to the judgment condition that the multiphysics coupling calculation result reaches a stable state.

[0060] Step 202: Obtain the thermal field temperature data of the converter transformer based on the reference temperature cloud map, and simultaneously collect the electrical operation data of the converter transformer.

[0061] Furthermore, step 202 may include the following sub-steps: S21. Extract the area with the largest temperature gradient from the baseline temperature cloud map as the high-risk monitoring area of ​​the thermal field.

[0062] In this embodiment of the invention, based on the generated reference temperature cloud map of the unbiased magnetic operating condition, the temperature field of the entire winding is traversed, and the temperature gradient amplitude is calculated for each spatial coordinate. The region where the temperature gradient peak is concentrated is screened out. Combining the characteristic that the leakage magnetic field of the higher harmonics will cause the winding hot spot to shift nonlinearly towards the end and radially inward, the specific coil area at both ends of the winding axis and the radially inward insulating oil channel area close to the core side are jointly identified as the high-risk monitoring area of ​​the thermal field, so as to accurately lock the key parts that are prone to defects.

[0063] S22. Based on the high-risk monitoring area of ​​the thermal field, non-invasive temperature sensors are used to collect the thermal field temperature data of the converter transformer.

[0064] In this embodiment of the invention, the destructive traditional approach of densely pre-embedding sensors throughout the winding is abandoned. Instead, based on the identified high-risk monitoring area of ​​the thermal field, non-invasive physical temperature sensors such as distributed fiber optic temperature probes are selected and precisely deployed in specific coils at both ends of the winding axis and in the radially inner insulating oil channels close to the core side. The physical temperature data of these targeted key nodes are read in real time to form the thermal field spatial temperature data under the actual operating state of the converter transformer. This data is then accurately compared with the theoretically predicted temperature values ​​of the corresponding nodes output in real time by the digital twin model to extract the characteristics of the spatial temperature gradient.

[0065] S23. Synchronously collect the electrical operation data of the converter transformer, including the real-time current on the grid side, the real-time current on the valve side, the real-time voltage on the grid side, the real-time parameter voltage on the valve side, and the measured value of the neutral point DC current.

[0066] In this embodiment of the invention, while collecting thermal field temperature data, the electrical operation data of the converter transformer is simultaneously collected through the online monitoring device of the converter transformer. The electrical operation data of the converter transformer includes the real-time current on the grid side, the real-time current on the valve side, the real-time voltage on the grid side, and the real-time voltage on the valve side. At the same time, the measured value of the neutral point DC current is simultaneously collected at the neutral point of the converter transformer. Considering the problem that early micro-turn short circuits or local insulation degradation are difficult to trigger the operation of traditional differential protection, the system deeply extracts extremely weak transient change features from the macroscopic electrical waveform, calculates and tracks the differential current trajectory and asymmetric negative sequence current component of the converter transformer in real time. This provides real-time boundary condition input for the digital twin model and core physical criteria for subsequent defect feature decoupling, ensuring that the thermal field temperature data and electrical operation data are completely synchronized in the time dimension.

[0067] High-risk monitoring area of ​​thermal field refers to the critical winding area selected from the reference temperature cloud map, which has the largest temperature gradient and is prone to defects; non-invasive temperature sensor refers to a temperature measurement device that can collect temperature data without damaging the winding structure; real-time current on the grid side of the converter transformer refers to the real-time current data collected from the grid side winding of the converter transformer; real-time current on the valve side of the converter transformer refers to the real-time current data collected from the valve side winding of the converter transformer; real-time voltage on the grid side of the converter transformer refers to the real-time voltage data collected from the grid side winding of the converter transformer; real-time voltage on the valve side of the converter transformer refers to the real-time voltage data collected from the valve side winding of the converter transformer; measured value of neutral point DC current refers to the real-time DC current value collected at the neutral point of the converter transformer.

[0068] Step 203: Using thermal field space temperature data, converter transformer electrical operation data and reference temperature cloud map, the operating condition is identified and interference is removed to generate a defect feature thermal field cloud map without bias magnetic interference.

[0069] Furthermore, step 203 may include the following sub-steps: S31. Perform spatiotemporal matching of thermal field space temperature data, converter transformer electrical operation data and reference temperature cloud map.

[0070] In this embodiment of the invention, the timestamps of the thermal field space temperature data, the converter transformer electrical operation data and the reference temperature cloud map are first aligned to ensure that the three types of data correspond under the same time sampling window; then the physical spatial coordinates of the sensor temperature measurement node are mapped to the grid cell coordinate system of the reference temperature cloud map to achieve a one-to-one correspondence of spatial positions.

[0071] S32. Construct a measured thermal field cloud map using the spatial temperature data of the thermal field after spatiotemporal matching, and compare the measured thermal field cloud map with the reference temperature cloud map after spatiotemporal matching to obtain the thermal field dimension determination result.

[0072] In this embodiment of the invention, based on the spatial temperature data of the thermal field after spatiotemporal matching, the measured thermal field cloud map of the converter transformer under actual operating conditions is obtained by interpolation reconstruction. The measured thermal field cloud map is compared with the reference temperature cloud map calculated after inputting the same DC bias current. The focus is on analyzing whether the spatial thermodynamic distribution characteristics (such as hot spot location, temperature rise gradient, and temperature amplitude distribution) of the two are basically consistent. If they are consistent, it is determined that the thermal field dimension conforms to the pure bias magnetic characteristics. Otherwise, it is determined that the thermal field does not conform to the pure bias magnetic characteristics.

[0073] S33. Determine whether the electrical operation data of the converter transformer after spatiotemporal matching meets the preset electrical dimension judgment conditions, and obtain the electrical dimension judgment result.

[0074] In this embodiment of the invention, the electrical operation data of the converter transformer after spatiotemporal matching is analyzed, with a focus on monitoring the differential current step change value and the asymmetric negative sequence current surge value between the grid side and valve side of the converter transformer. If no obvious differential current step or negative sequence component surge appears in the electrical data, it is determined that the electrical dimension conforms to the pure magnetic bias characteristics. If the above-mentioned transient change characteristics exist, it is determined that the electrical dimension does not conform to the pure magnetic bias characteristics, providing electrical dimension evidence for the existence of defects.

[0075] S34. When the thermal field dimension determination result is that the thermal field conforms to the pure bias magnetic characteristics, and the electrical dimension determination result is that the electrical conforms to the pure bias magnetic characteristics, then the converter transformer is determined to be in a pure DC bias magnetic disturbance condition.

[0076] In this embodiment of the invention, when the thermal field distribution is basically consistent with the reference cloud map of the input bias current and there is no obvious transient change in the electrical data, it is comprehensively determined that the current overheating is only caused by pure DC bias disturbance caused by external environment or operating conditions; at this time, the system only outputs DC bias overheating alarm and actively shields false alarm signals for physical defects inside the converter transformer winding, so as to avoid misjudgment of defects under pure bias conditions.

[0077] S35. When the thermal field dimension determination result is that the thermal field does not conform to the pure bias magnetic characteristics, or the electrical dimension determination result is that the electrical does not conform to the pure bias magnetic characteristics, the converter transformer is determined to be in a combined bias magnetic and defect condition.

[0078] In this embodiment of the invention, if the measured thermal field does not match the distribution characteristics of the reference cloud map, or if there is a sudden increase in the differential current step / negative sequence component in the electrical data, the current operating condition is determined to be a composite operating condition with both bias magnetic interference and winding defects. Further bias magnetic interference stripping operation needs to be performed to extract the true defect characteristics.

[0079] S36. Extract the measured spatial temperature of each spatial coordinate from the measured thermal field cloud map, and extract the reference spatial temperature of each spatial coordinate from the reference temperature cloud map after spatiotemporal matching.

[0080] In this embodiment of the invention, all corresponding spatial coordinate nodes of the measured thermal field cloud map and the reference temperature cloud map are traversed, and the sensor measured temperature data of each node are extracted respectively. Compared with the model's predicted baseline temperature data .

[0081] S37. Using the measured spatial temperature of each spatial coordinate and the reference spatial temperature that is consistent with the spatial coordinate, the bias magnetic interference is stripped to obtain the global residual temperature distribution.

[0082] In this embodiment of the invention, the inversion capability based on the multiphysics twin model is achieved through the formula:

[0083] Calculations were performed on all spatial coordinate nodes across the entire domain. The background of global core saturation and increased thermal field of structural components caused by bias magnetism was mathematically stripped away to obtain the global residual temperature distribution that only reflects defect-related anomalies. .

[0084] S38. Perform noise extraction and feature verification on the global residual temperature distribution to generate a defect feature thermal field cloud map without bias magnetic interference.

[0085] In this embodiment of the invention, noise filtering is first applied to the global residual temperature distribution to eliminate minor temperature fluctuations caused by environmental or measurement errors. Then, the residual thermal field is analyzed in depth to verify whether there are local abnormal temperature rises in highly sensitive areas that cannot be explained by normal load and bias magnetic effect. At the same time, it is verified whether the local abnormal temperature rise is synchronous with the transient differential current step or the asymmetric negative sequence current surge in time. If the verification is successful, the local abnormal temperature rise area is visualized to generate a defect feature thermal field cloud map without bias magnetic interference, so as to achieve accurate perception of latent defects such as early inter-turn short circuits or minor overheating strand breakage.

[0086] Spatiotemporal matching refers to the operation of aligning and matching thermal field temperature data, electrical operation data, and reference temperature cloud maps in terms of time and space coordinates; measured thermal field cloud map refers to the real-time temperature field distribution reconstructed based on actually collected thermal field temperature data; thermal field dimension determination result refers to the conclusion of thermal field characteristic determination obtained by comparing the measured thermal field cloud map with the reference temperature cloud map; preset electrical dimension determination conditions refer to the pre-set criteria used to determine whether electrical operation characteristics meet the pure bias magnetic condition; electrical dimension determination result refers to the conclusion of electrical characteristic determination obtained based on electrical operation data; pure DC bias magnetic disturbance condition refers to the condition caused solely by DC bias magnetic disturbance. The following are the operating conditions that cause abnormal operation but have no internal winding defects: The combined condition of bias magnetism and defects refers to the condition where DC bias magnetism interference and internal winding defects coexist; the measured spatial temperature refers to the actual temperature value corresponding to each spatial coordinate in the measured thermal field cloud map; the reference spatial temperature refers to the theoretical temperature value corresponding to the spatial coordinate in the reference temperature cloud map; bias magnetism interference removal refers to the process of removing the temperature rise component caused by DC bias magnetism from the measured temperature; the global residual temperature distribution refers to the distribution of the temperature difference across the entire winding domain obtained after bias magnetism interference removal; and noise extraction and feature verification refers to the processing operation of removing interference noise from the residual temperature and verifying the authenticity of defect features.

[0087] Step 204: Use the defect feature thermal field cloud map and the electrical operation data of the converter transformer to perform winding defect detection and generate the winding defect detection results of the converter transformer.

[0088] Furthermore, step 204 may include the following sub-steps: S41. Extract thermal field defect characteristic indicators from the defect characteristic thermal field cloud map. The thermal field defect characteristic indicators include the peak temperature of the defect hot spot, the diffusion area of ​​the defect hot spot, and the thermal field temperature rise gradient.

[0089] In this embodiment of the invention, the temperature data of all spatial coordinate nodes in the defect feature thermal field cloud map without magnetic interference are traversed, and the temperature values ​​of each node are compared one by one. The node with the highest temperature is directly selected and determined as the spatial coordinate of the defect hotspot. The highest temperature value corresponding to the coordinate is extracted as the peak temperature of the defect hotspot. Centered on the defect hotspot, the area where the temperature exceeds the reference normal temperature rise range is defined as the defect hotspot region. The number of spatial grids contained in this region is counted and converted into the defect hotspot diffusion area. Around the adjacent coordinate nodes of the defect hotspot, the ratio of the temperature change difference between each two adjacent nodes to the spatial distance is calculated to obtain the thermal field temperature rise gradient, thereby completing the quantitative extraction of thermal field defect feature indicators.

[0090] S42. Extract electrical defect characteristic indicators from the electrical operation data of the converter transformer. The electrical defect characteristic indicators include the differential current step change value between the grid side and valve side of the converter transformer, the asymmetric negative sequence current surge, and the voltage harmonic distortion rate.

[0091] In this embodiment of the invention, for the differential current time-series data of the converter transformer grid side and valve side, the difference between the differential current at each sampling time and the differential current at the previous sampling time is calculated sequentially. Among the differences at multiple consecutive sampling points, the largest difference is selected as the differential current step change value. Three-phase current data is extracted from the electrical operation data, and the difference between the current of one phase and the average value of the current of the other two phases is calculated to obtain the negative sequence current value at the current time. Then, the difference between this value and the negative sequence current reference value under normal operating conditions is obtained to obtain the asymmetrical negative sequence current surge. For the voltage sampling data of the converter transformer grid side and valve side, the ratio of the amplitude of each harmonic voltage to the amplitude of the fundamental voltage is calculated. All harmonic ratios are added together to obtain the voltage harmonic distortion rate, thus completing the extraction of electrical defect characteristic indicators.

[0092] S43. Spatiotemporally synchronize thermal defect characteristic indicators with electrical defect characteristic indicators.

[0093] In this embodiment of the invention, the sampling time corresponding to the thermal defect characteristic index and the sampling time corresponding to the electrical defect characteristic index are adjusted to the same moment to ensure that the two reflect the same operating state of the converter transformer; at the same time, the spatial location of the defect hot spot is matched with the winding associated area where abnormal changes occur in the electrical data to ensure that the location of the thermal defect and the electrical defect characteristics are completely matched in space, thus achieving spatiotemporal synchronization of the two types of indicators.

[0094] S44. Based on the preset fusion judgment rules, the thermal field defect characteristic indicators after spatiotemporal synchronization and the electrical defect characteristic indicators after spatiotemporal synchronization are coupled and judged to generate the winding defect detection results of the converter transformer. The winding defect detection results include the actual existence of the winding defect, the defect type, and the severity level; The preset fusion judgment rule is that if the defect level corresponding to the thermal defect characteristic index and the electrical defect characteristic index is consistent and synchronized in time and space, it is judged as a real winding defect. If either index is not satisfied, it is judged as an interference signal and the defect warning is blocked.

[0095] In this embodiment of the invention, a quantitative classification standard for thermal defect level and electrical defect level is preset: for example, based on the peak temperature of the defect hot spot, the diffusion area and the temperature rise gradient, thermal defects are divided into three levels: slight, general and severe; based on the differential current step change value, the asymmetric negative sequence current surge increment and the voltage harmonic distortion rate, electrical defects are divided into three levels corresponding to thermal defects. Based on preset fusion judgment rules, the two types of indicators after spatiotemporal synchronization are coupled for judgment: only when the defect level corresponding to the thermal field defect characteristic indicator and the electrical defect characteristic indicator is completely consistent, and the defect hot spot and the electrical abnormal change are completely synchronized in time and space, is it determined that there is a real defect in the converter transformer winding. Combining the local abnormal temperature rise characteristics of the defect characteristic thermal field cloud map and the abnormal change characteristics of electrical data, the defect type is further determined to be early inter-turn short circuit or minor overheating strand breakage, etc. At the same time, the severity level is determined according to the level classification results. If any indicator does not meet the above conditions, it is determined to be an interference signal caused by environmental or measurement error, and the defect warning is actively shielded. Finally, a winding defect detection result containing the real existence of the winding defect, the defect type and the severity level are generated, realizing accurate perception and reliable identification of early latent defects.

[0096] Thermal defect characteristic indicators refer to the quantitative indicators of the thermal field extracted from the defect characteristic thermal field cloud map to characterize defects. Specifically, they include the peak temperature of the defect hotspot, the diffusion area of ​​the defect hotspot, and the thermal field temperature rise gradient. The peak temperature of the defect hotspot refers to the temperature value of the highest temperature point in the defect characteristic thermal field cloud map. The diffusion area of ​​the defect hotspot refers to the spatial range of the abnormal temperature rise area around the defect hotspot. The thermal field temperature rise gradient refers to the drastic degree of temperature change around the defect hotspot. Electrical defect characteristic indicators refer to the quantitative electrical indicators extracted from electrical operation data to characterize defects. Specifically, they include the differential current step jump value, the asymmetric negative sequence current surge, and the voltage harmonic distortion rate. The differential current step jump value refers to the differential current between the grid side and valve side of the converter transformer. The instantaneous change value; the sudden increase in asymmetrical negative sequence current refers to the change in negative sequence current relative to the reference value; voltage harmonic distortion rate refers to the proportion of harmonic components in the voltage waveform; spatiotemporal synchronization refers to the processing operation that keeps the thermal field defect characteristic indicators and electrical defect characteristic indicators consistent in time and space; preset fusion judgment rules refer to the logical rules for defect judgment based on thermal field and electrical characteristics; the actual existence of winding defects refers to the conclusion that the winding has actual physical defects; defect type refers to the specific category of winding defects; severity level refers to the severity of winding defects; interference signal refers to abnormal signals caused by environmental or measurement errors that do not represent actual defects; defect warning refers to the warning prompts issued for winding defects.

[0097] Step 205: When the severity level in the winding defect detection result is severe, perform a power outage electrical test and generate a converter transformer winding defect diagnosis report based on the test results.

[0098] Power outage electrical testing refers to steady-state electrical testing conducted while the converter transformer is out of service; converter transformer winding defect diagnosis report refers to a diagnostic document that integrates detection and test results and includes defect information and maintenance suggestions.

[0099] In this embodiment of the invention, when the severity level of the defect in the generated converter transformer winding defect detection result is severe, the system automatically triggers the power outage maintenance process, performs three electrical tests: DC resistance of the grid-side winding, DC resistance of the valve-side winding, and voltage ratio, conducts physical logic deduction based on the test data, and finally generates a converter transformer winding defect diagnosis report, thus establishing a closed loop for diagnosis between online multi-field early warning and offline physical damage assessment.

[0100] To address grid-side winding defects, the system uses testing equipment to measure the DC resistance of the grid-side windings at all tap positions of the converter transformer. The measured resistance values ​​for each tap are compared with the factory reference resistance data. If the resistance deviations at all taps are uniform and exceed the preset standard limit, a preliminary judgment is made that the main grid-side winding is faulty. Then, the change in resistance between adjacent tap positions is calculated. If the change shows a uniform pattern, the possibility of faults in the voltage regulating winding and tap changer is ruled out. Based on this, combined with structural parameters such as the number of parallel strands in the internal conductors of the converter transformer, the resistance value of the pure main winding without the voltage regulating winding engaged is used for reverse calculation. By matching the theoretical resistance value of a single conductor with the measured fault resistance value, the specific number of broken conductor strands is accurately estimated and quantified. For example, assuming the resistance of a normal single conductor in the middle tap (which can be considered as only having the main winding) is... and around the total number of shares The measured fault resistance is Then it can be done through the formula Calculate the number of broken shares ,in The number of broken strands is used to determine the degree of physical damage to the grid-side winding.

[0101] To address defects in the valve-side winding, the system collects DC resistance data of the valve-side winding using testing equipment. Given that the valve-side winding of the converter transformer is directly connected to the converter valve and is usually not equipped with an on-load tap changer, if an abnormal increase in the measured DC resistance of the valve-side winding is detected, the defect is diagnosed as a blown wire, poor welded joint, or loose lead connection in the main winding of the valve-side winding itself, thus completing the precise location of the defect in the valve-side winding.

[0102] The system also incorporates voltage ratio testing features to verify inter-turn and tap changer defects: the voltage ratio of each winding of the converter transformer is measured using testing equipment, the low-voltage side voltage ratio data is extracted and compared with the nameplate rating. If the overall low-voltage side voltage ratio deviates from the nameplate rating, it means that the effective number of working turns of the grid-side or valve-side winding has decreased, further confirming an inter-turn short-circuit fault. Conversely, if the voltage ratio anomaly only appears in isolation at a specific tap position, it is identified as a mechanical fault of the tap changer or a connection defect of the voltage regulating winding, thus achieving further subdivision of the defect type.

[0103] After completing the above three types of electrical tests and logical deductions, the system integrates online defect detection results, various electrical test data, and physical damage inversion conclusions to generate a converter transformer winding defect diagnosis report. This report includes the actual existence of the defect, the specific location of the defect (grid side / valve side), the defect type (broken conductor strand / inter-turn short circuit / tap switch failure, etc.), the degree of physical damage (number of broken strands / degree of insulation degradation, etc.), and maintenance recommendations. This completes a full diagnostic closed loop from online multi-field sensing to offline physical damage assessment, providing accurate data support for converter transformer maintenance decisions.

[0104] Current research shows that existing detection systems largely follow the design concepts of conventional AC power transformers, while their detection schemes for converter transformers operating under complex AC / DC conditions have certain shortcomings. This stems from the unique operating environment of converter transformers. During operation, they not only endure nonlinear eddy current amplification caused by non-sinusoidal high-order harmonic currents, but are also frequently affected by complex DC bias magnetism under various operating conditions, leading to core half-wave saturation and leakage flux distortion. This makes traditional detection methods based on a single physical field or static benchmark difficult to directly adapt. The main deficiencies of existing technologies lie in the two core aspects of online detection of early latent defects and physical damage assessment during power outage maintenance. These deficiencies stand in stark contrast to the detection method and system proposed in this invention. The specific shortcomings and corresponding solutions of this invention are as follows: From the perspective of online defect perception and early warning, converter transformer winding fault diagnosis faces unique technical challenges. There is a complex interference between actual internal insulation degradation or partial strand breakage and the heating of the core and structural components caused by normal DC bias, resulting in significant background thermal field distortion at different bias levels. A more critical technical difficulty lies in the masking effect of fault characteristics—abnormal hot spots caused by early, minor winding faults are often masked by the temperature rise due to bias, and the resulting transient electrical characteristics are extremely weak. This directly leads to a high likelihood of false alarms using methods based on static multi-physics characteristic comparison, while traditional protection methods based on post-event differential current amplitude exceeding limits are ineffective for early warning. Therefore, it is necessary to construct diagnostic criteria that take into account DC bias interference. This invention proposes a targeted solution. In step 201, a multi-physics twin model of the winding considering bias boundary conditions is constructed. A heterogeneous mesh partitioning strategy is used to ensure simulation accuracy. A reference temperature cloud map under the condition of no bias is obtained through electromagnetic and thermal coupling simulation. Then, in step 203, the thermal field temperature data, electrical operation data and reference temperature cloud map are spatiotemporally matched, operating conditions are identified and interference is removed. The global thermal field background caused by bias is removed in real time. At the same time, the transient differential current step change value and asymmetric negative sequence current surge value extracted in step 204 are combined as reliable basis for early warning of latent inter-turn short circuit and thermal defects, effectively solving the problems of false alarm and delayed warning in traditional methods.

[0105] In the areas of defect location and power outage electrical testing and diagnosis, converter transformers have extremely stringent insulation design requirements. Their valve-side windings and oil-paper insulation systems directly bear extremely high AC and DC voltage stresses, and any damage to the physical structure can easily induce partial discharge or even breakdown. In pursuit of high-precision internal positioning, some traditional solutions employ invasive methods such as densely encapsulating sensing components in insulating cardboard. Due to implementation difficulties, this severely compromises the original insulation strength of the equipment. At the same time, existing online systems generally lack quantitative diagnostic mechanisms that are linked to subsequent power outage maintenance. To address this issue, this invention abandons destructive hardware pre-embedding. Relying on the multi-physical twin model constructed in step 201, it locates high-risk monitoring areas of the thermal field from the reference temperature cloud map in step 202. Non-invasive temperature sensors are deployed only in external high-risk sensitive areas to implement directional monitoring, which not only ensures the insulation integrity of the equipment but also achieves targeted data acquisition. After triggering a severe defect warning in step 204, a power outage electrical test is performed in step 205. The DC resistance of the grid-side and valve-side windings is measured step by step and compared with the factory reference data. Combined with the internal structural parameters of the windings, the degree of physical damage is calculated in reverse. At the same time, the voltage ratio test is used for cross-verification to confirm the defect types such as winding strand breakage and tap changer failure. This completes the closed loop from online multi-field warning to offline physical damage assessment, solving the shortcomings of existing solutions in insulation damage and lack of linkage quantitative diagnosis.

[0106] It should be noted that the distributed fiber optic temperature probe in space thermal field monitoring can be replaced by an infrared thermal imager, or a surface acoustic wave wireless temperature sensor or an external ultrasonic / acoustic emission sensor can be used to assist in monitoring abnormal acoustic signals caused by local overheating as a substitute or supplementary verification of thermal characteristics.

[0107] The DC resistance and voltage ratio tests in the power outage electrical tests can be verified by combining the frequency response curve offset analysis of the winding deformation test or the short-circuit impedance measurement data.

[0108] Please see Figure 4 , Figure 4 This is a structural block diagram of a converter transformer winding defect detection system that takes into account DC bias and multi-physics twinning, as provided in Embodiment 3 of the present invention.

[0109] This invention provides a converter transformer winding defect detection system that considers DC bias magnetism and multi-physics twinning, comprising: Module 401 is used to construct a multi-physics twin model of the converter transformer windings and perform electromagnetic and thermal coupling simulations to obtain a reference temperature cloud map under unbiased magnetic conditions. The acquisition module 402 is used to acquire the thermal field temperature data of the converter transformer based on the reference temperature cloud map, and simultaneously collect the electrical operation data of the converter transformer. Processing module 403 is used to identify operating conditions and remove interference by using thermal field space temperature data, converter transformer electrical operation data and reference temperature cloud map, and generate defect feature thermal field cloud map without bias magnetic interference. The detection module 404 is used to detect winding defects by using the thermal field cloud map of defect features and the electrical operation data of the converter transformer, and to generate the winding defect detection results of the converter transformer.

[0110] Furthermore, module 401 includes: The winding multi-physical twin model submodule is used to construct a winding multi-physical twin model using a single copper conductor and the insulating paper wrapping of the converter transformer as the smallest calculation unit. The local loss density submodule is used to mesh the multi-physics twin model of the winding and solve the local loss density of various currents in each mesh cell through an electromagnetic frequency domain solver. The total loss density submodule is used to spatially superimpose multiple local loss densities within the same grid cell to obtain the total loss density of each grid cell. The magnetic induction intensity submodule is used to acquire the winding electrical operation data and winding geometric structure data of the converter transformer, and to acquire the magnetic induction intensity of the winding multi-physical twin model. The winding reference total loss submodule is used to calculate the winding reference total loss using winding electrical operating data, winding geometric data, and magnetic induction intensity. The winding electrical operating data includes the effective value of winding current, the overall resistance value of winding, the conductivity of winding conductors, and the current angular frequency; the winding geometric data includes the longitudinal wire gauge dimensions, the transverse wire gauge dimensions, and the total winding volume. The winding simulation total loss submodule is used to perform volume integration of the total loss density of each grid cell over the entire winding domain to obtain the winding simulation total loss. The heat source mapping submodule is used to calculate the target difference between the reference total loss of the winding and the simulated total loss of the winding. When the target difference is within the preset error range, the total loss density of each grid cell is mapped to a heat source. The reference temperature cloud map submodule is used to perform bidirectional coupling iteration of the multi-physics twin model of the winding based on the heat source mapping results until the residual convergence is achieved, so as to obtain the reference temperature cloud map under the unbiased magnetic condition.

[0111] Furthermore, the acquisition module 402 includes: The high-risk thermal field monitoring area submodule is used to extract the area with the largest temperature gradient from the baseline temperature cloud map as the high-risk thermal field monitoring area. The thermal field space temperature data submodule is used to collect thermal field space temperature data of the converter transformer using non-invasive temperature sensors based on the high-risk monitoring area of ​​the thermal field. The converter transformer electrical operation data submodule is used to synchronously collect the converter transformer electrical operation data; The electrical operating data of the converter transformer includes the real-time current on the grid side, the real-time current on the valve side, the real-time voltage on the grid side, the real-time parameter voltage on the valve side, and the measured value of the neutral point DC current.

[0112] Furthermore, the processing module 403 includes: The spatiotemporal matching submodule is used to perform spatiotemporal matching of thermal field temperature data, converter transformer electrical operation data and reference temperature cloud map. The thermal field dimension determination result submodule is used to construct a measured thermal field cloud map using the spatiotemporally matched thermal field spatial temperature data, and compare the measured thermal field cloud map with the spatiotemporally matched benchmark temperature cloud map to obtain the thermal field dimension determination result. The electrical dimension judgment result submodule is used to determine whether the electrical operation data of the converter transformer after spatiotemporal matching meets the preset electrical dimension judgment conditions, and to obtain the electrical dimension judgment result. The first determination submodule is used to determine that the converter transformer is in a pure DC biased magnetic disturbance condition when the thermal field dimension determination result is that the thermal field conforms to the pure biased magnetic characteristics and the electrical dimension determination result is that the electrical conforms to the pure biased magnetic characteristics. The second determination submodule is used to determine that the converter transformer is in a combined condition of magnetic bias and defect when the thermal field dimension determination result is that the thermal field does not meet the characteristics of pure magnetic bias, or the electrical dimension determination result is that the electrical does not meet the characteristics of pure magnetic bias. The reference space temperature submodule is used to extract the measured space temperature of each spatial coordinate from the measured thermal field cloud map, and to extract the reference space temperature of each spatial coordinate from the reference temperature cloud map after spatiotemporal matching. The global residual temperature distribution submodule is used to remove magnetic interference by using the measured spatial temperature of each spatial coordinate and the reference spatial temperature that is consistent with the spatial coordinate, so as to obtain the global residual temperature distribution. The defect feature thermal field cloud map submodule is used to extract noise and verify features of the global residual temperature distribution, and generate defect feature thermal field cloud maps without bias magnetic interference.

[0113] Furthermore, the detection module 404 includes: The thermal field defect characteristic index submodule is used to extract thermal field defect characteristic indices from the defect characteristic thermal field cloud map. The thermal field defect characteristic indices include the peak temperature of the defect hotspot, the diffusion area of ​​the defect hotspot, and the thermal field temperature rise gradient. The electrical defect characteristic index submodule is used to extract electrical defect characteristic indexes from the electrical operation data of the converter transformer. The electrical defect characteristic indexes include the differential current step change value between the grid side and valve side of the converter transformer, the asymmetric negative sequence current surge, and the voltage harmonic distortion rate. The spatiotemporal synchronization submodule is used to perform spatiotemporal synchronization of thermal defect characteristic indicators and electrical defect characteristic indicators; The winding defect detection result submodule is used to couple and judge the thermal field defect characteristic indicators and the electrical defect characteristic indicators after time and space synchronization based on the preset fusion judgment rules, and generate the winding defect detection results of the converter transformer. The winding defect detection results include the actual existence of the winding defect, the defect type, and the severity level; The preset fusion judgment rule is that if the defect level corresponding to the thermal defect characteristic index and the electrical defect characteristic index is consistent and synchronized in time and space, it is judged as a real winding defect. If either index is not satisfied, it is judged as an interference signal and the defect warning is blocked.

[0114] Furthermore, it also includes: The test module is used to perform a power outage electrical test when the severity level in the winding defect detection results is severe, and generate a converter transformer winding defect diagnosis report based on the test results.

[0115] Since the above is a system corresponding to a converter transformer winding defect detection method that considers DC bias and multi-physical field twins, its implementation principle is the same as that of a converter transformer winding defect detection method that considers DC bias and multi-physical field twins. For the sake of convenience and brevity, those skilled in the art can clearly understand that the specific working process of the system and modules described above can be referred to the corresponding process in the aforementioned method embodiments, and will not be repeated here.

[0116] Please see Figure 5 , Figure 5 This is a structural block diagram of an electronic device provided in Embodiment 4 of the present invention.

[0117] An electronic device according to an embodiment of the present invention includes: a memory 501 and a processor 502. The memory 501 stores a computer program. When the computer program is executed by the processor 502, the processor 502 performs the converter transformer winding defect detection method considering DC bias and multi-physics twin as described in the above embodiment.

[0118] Memory 501 may be an electronic memory such as flash memory, EEPROM (Electrically Erasable Programmable Read-Only Memory), EPROM, hard disk, or ROM. Memory 501 has storage space 503 for program code 513 for performing any of the method steps described above. For example, storage space 503 for program code may include various program codes 513 for implementing the various steps in the methods described above. These program codes may be read from or written to one or more computer program products. These computer program products include program code carriers such as hard disks, CDs, memory cards, or floppy disks. The program code may be compressed, for example, in a suitable form. When run by a computing processing device, this code causes the computing processing device to perform the various steps in the methods described above. These program codes may be read from or written to one or more computer program products. These computer program products include program code carriers such as hard disks, CDs, memory cards, or floppy disks. The program code may be compressed, for example, in a suitable form. When these codes are run by a computing device, the computing device causes the device to perform the various steps in the converter transformer winding defect detection method described above, which takes into account DC bias and multiphysics twins.

[0119] Embodiment 5 of the present invention also provides a computer-readable storage medium storing a computer program thereon, which, when executed by a processor, implements the converter transformer winding defect detection method considering DC bias and multi-physics twins as described in the above embodiments.

[0120] Embodiment 6 of the present invention also provides a computer program product, which includes a computer program stored on a non-transitory computer-readable storage medium. The computer program includes program instructions, wherein when the program instructions are executed by a computer, the computer performs the converter transformer winding defect detection method considering DC bias and multi-physics twin as described in the above embodiments.

[0121] Those skilled in the art will clearly understand that, for the sake of convenience and brevity, the specific working processes of the systems, devices, and units described above can be referred to the corresponding processes in the foregoing method embodiments, and will not be repeated here.

[0122] 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, or indirect coupling or communication connection between apparatuses or units, and may be electrical, mechanical, or other forms.

[0123] 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; 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.

[0124] Furthermore, the functional units in the various embodiments of the present invention 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.

[0125] 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 the present invention, 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 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 of the various embodiments of the present invention. The aforementioned storage medium includes various media capable of storing program code, such as USB flash drives, portable hard drives, read-only memory (ROM), random access memory (RAM), magnetic disks, or optical disks.

[0126] The above embodiments are only used to illustrate the technical solutions of the present invention, and are not intended to limit it. Although the present invention has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that modifications can still be made to the technical solutions described in the foregoing embodiments, or equivalent substitutions can be made to some of the technical features. Such modifications or substitutions do not cause the essence of the corresponding technical solutions to deviate from the spirit and scope of the technical solutions of the embodiments of the present invention.

Claims

1. A method for detecting defects in the windings of a converter transformer taking into account DC bias and multi-physical field coupling, characterized in that, include: A multi-physics twin model of the converter transformer windings was constructed and electromagnetic and thermal coupling simulations were performed to obtain a reference temperature cloud map under unbiased magnetic conditions. Based on the reference temperature cloud map, the thermal field temperature data of the converter transformer is obtained, and the electrical operation data of the converter transformer is collected simultaneously. Using the thermal field space temperature data, the converter transformer electrical operation data, and the reference temperature cloud map, operating conditions are identified and interference is removed to generate a defect feature thermal field cloud map without bias magnetic interference. The winding defect detection results of the converter transformer are generated by using the thermal field cloud map of the defect features and the electrical operation data of the converter transformer.

2. The method for detecting defects in converter transformer windings considering DC bias and multi-physics twinning as described in claim 1, characterized in that, The construction of a multi-physics twin model of the converter transformer windings and the performance of electromagnetic and thermal coupling simulations yield a reference temperature contour map under unbiased magnetic conditions, including: Using a single copper conductor and its insulating paper wrapping in a converter transformer winding as the smallest calculation unit, a multi-physical twin model of the winding is constructed. The multi-physics twin model of the winding is meshed, and the local loss density of various currents in each mesh cell is solved by an electromagnetic frequency domain solver. The total loss density of each grid cell is obtained by spatially superimposing multiple local loss densities within the same grid cell. Obtain the winding electrical operation data and winding geometric structure data of the converter transformer, and obtain the magnetic induction intensity of the winding multi-physics twin model; The reference total loss of the winding is calculated using the winding electrical operating data, the winding geometric data, and the magnetic induction intensity. The winding electrical operating data includes the effective value of the winding current, the overall resistance value of the winding, the conductivity of the winding conductor, and the current angular frequency. The winding geometric data includes the longitudinal wire gauge dimensions, the transverse wire gauge dimensions, and the total winding volume. The total loss density of each grid cell is integrated over the entire winding domain to obtain the total winding simulation loss. Calculate the target difference between the reference total loss of the winding and the simulated total loss of the winding. When the target difference is within a preset error range, perform heat source mapping on the total loss density of each grid cell. Based on the heat source mapping results, the multi-physics twin model of the winding is subjected to bidirectional coupling iteration of multi-physics fields until the residual convergence is achieved, and the reference temperature cloud map under the unbiased magnetic condition is obtained.

3. The method for detecting defects in converter transformer windings considering DC bias and multi-physics twinning as described in claim 1, characterized in that, The process of acquiring the thermal field temperature data of the converter transformer based on the reference temperature cloud map and simultaneously collecting the electrical operation data of the converter transformer includes: The region with the largest temperature gradient is extracted from the reference temperature cloud map and designated as a high-risk thermal field monitoring zone. According to the high-risk monitoring zone of the thermal field, non-invasive temperature sensors are used to collect the thermal field temperature data of the converter transformer. Simultaneously collect electrical operation data of the converter transformer; The electrical operating data of the converter transformer includes the real-time current on the grid side, the real-time current on the valve side, the real-time voltage on the grid side, the real-time parameter voltage on the valve side, and the measured value of the neutral point DC current.

4. The method for detecting defects in converter transformer windings considering DC bias and multi-physics twinning as described in claim 1, characterized in that, The step of using the thermal field space temperature data, the converter transformer electrical operation data, and the reference temperature cloud map to identify operating conditions and remove interference, generating a defect feature thermal field cloud map without magnetic interference, includes: Spatiotemporal matching is performed between the thermal field space temperature data, the converter transformer electrical operation data, and the reference temperature cloud map; A measured thermal field cloud map is constructed using the spatial temperature data of the thermal field after spatiotemporal matching, and the measured thermal field cloud map is compared with the reference temperature cloud map after spatiotemporal matching to obtain the thermal field dimension determination result. Determine whether the electrical operation data of the converter transformer after spatiotemporal matching meets the preset electrical dimension judgment conditions, and obtain the electrical dimension judgment result; When the thermal field dimension determination result is that the thermal field conforms to the pure bias magnetic characteristics, and the electrical dimension determination result is that the electrical conforms to the pure bias magnetic characteristics, then the converter transformer is determined to be in a pure DC bias magnetic disturbance condition. If the thermal field dimension determination result is that the thermal field does not conform to the pure bias magnetic characteristics, or the electrical dimension determination result is that the electrical does not conform to the pure bias magnetic characteristics, then the converter transformer is determined to be in a combined bias magnetic and defect condition. The measured spatial temperature of each spatial coordinate is extracted from the measured thermal field cloud map, and the reference spatial temperature of each spatial coordinate is extracted from the reference temperature cloud map after spatiotemporal matching. The magnetic interference was stripped using the measured spatial temperature at each spatial coordinate and the reference spatial temperature that is consistent with the spatial coordinate, and the global residual temperature distribution was obtained. Noise extraction and feature verification are performed on the global residual temperature distribution to generate a defect feature thermal field cloud map without bias magnetic interference.

5. The method for detecting defects in converter transformer windings considering DC bias and multi-physics twinning as described in claim 1, characterized in that, The step of using the defect feature thermal field cloud map and the electrical operation data of the converter transformer to perform winding defect detection, and generating the winding defect detection result of the converter transformer, includes: The thermal field defect feature index is extracted from the defect feature thermal field cloud map. The thermal field defect feature index includes the peak temperature of the defect hot spot, the diffusion area of ​​the defect hot spot, and the thermal field temperature rise gradient. Electrical defect characteristic indicators are extracted from the electrical operation data of the converter transformer. The electrical defect characteristic indicators include the differential current step change value between the grid side and the valve side of the converter transformer, the asymmetric negative sequence current surge, and the voltage harmonic distortion rate. The thermal defect characteristic indicators and the electrical defect characteristic indicators are synchronized in time and space; Based on the preset fusion judgment rules, the thermal field defect characteristic indicators after spatiotemporal synchronization and the electrical defect characteristic indicators after spatiotemporal synchronization are coupled and judged to generate the winding defect detection results of the converter transformer. The winding defect detection results include the actual existence of the winding defect, the defect type, and the severity level; The preset fusion judgment rule is specifically that if the thermal field defect characteristic index and the electrical defect characteristic index have the same defect level and are synchronized in time and space, then it is judged that a winding defect actually exists. If either index is not satisfied, it is judged as an interference signal, and the defect warning is blocked.

6. The method for detecting defects in converter transformer windings considering DC bias and multi-physics twinning according to any one of claims 1-5, characterized in that, Also includes: When the severity level of the winding defect detection result is severe, a power outage electrical test is performed, and a converter transformer winding defect diagnosis report is generated based on the test results.

7. A converter transformer winding defect detection system considering DC bias magnetism and multi-physics twinning, characterized in that, include: The module is used to build a multi-physics twin model of the converter transformer windings and perform electromagnetic and thermal coupling simulations to obtain a reference temperature cloud map under unbiased magnetic conditions. The acquisition module is used to acquire the thermal field temperature data of the converter transformer based on the reference temperature cloud map, and simultaneously collect the electrical operation data of the converter transformer. The processing module is used to identify operating conditions and remove interference by using the thermal field space temperature data, the electrical operation data of the converter transformer and the reference temperature cloud map, and generate a defect feature thermal field cloud map without bias magnetic interference. The detection module is used to perform winding defect detection using the defect feature thermal field cloud map and the electrical operation data of the converter transformer, and generate the winding defect detection result of the converter transformer.

8. An electronic device, characterized in that, The device includes a memory and a processor, wherein the memory stores a computer program, and when the computer program is executed by the processor, the processor performs the steps of the converter transformer winding defect detection method as described in any one of claims 1-6, which takes into account DC bias and multiphysics twins.

9. A computer-readable storage medium having a computer program stored thereon, characterized in that, When the computer program is executed, it implements the converter transformer winding defect detection method as described in any one of claims 1-6, which takes into account DC bias and multi-physics twins.

10. A computer program product, characterized in that, The computer program product includes a computer program stored on a non-transitory computer-readable storage medium, the computer program including program instructions, wherein when the program instructions are executed by a computer, the computer performs the converter transformer winding defect detection method considering DC bias and multiphysics twins as described in any one of claims 1-6.

Images