A transformer winding deformation detection method, device, equipment and storage medium

Abstract

The application discloses a transformer winding deformation detection method, device and equipment and a storage medium, and belongs to the technical field of power equipment detection. The method comprises the following steps: obtaining an oscillation wave signal of any one phase winding of a transformer; setting a plurality of initial sliding windows, and sequentially performing recursive updating on the noise baseline of the oscillation wave signal of the initial sliding window; extracting a middle oscillation wave; setting a plurality of terminal sliding windows, and calculating the variance indicator of the middle oscillation wave corresponding to the terminal sliding window based on the variance baseline of the middle oscillation wave; extracting a final oscillation wave; extracting an effective oscillation wave from the oscillation wave signals of the remaining phase windings according to the time sequence index corresponding to the final oscillation wave; and calculating the correlation coefficient and the main frequency between the effective oscillation waves of the phase windings, and respectively detecting the transformer winding deformation. The application solves the technical problems in the prior art that the fixed time window method cannot cover different winding effective sections, the noise interference is large, and the method is easily interfered by waveform misplacement to cause misjudgment.

Description

Technical Field

[0001] This invention relates to the field of power equipment testing technology, and in particular to a method, apparatus, equipment and storage medium for detecting transformer winding deformation. Background Technology

[0002] Currently, in the condition monitoring and maintenance of power equipment, transformer winding deformation detection commonly employs methods such as impact testing and low-voltage pulse testing. After the test, the oscillation wave signal generated by the winding response is collected, and the winding operating status is analyzed and judged based on the characteristics of the oscillation waveform. Among these methods, the oscillation wave method can effectively reflect changes in winding structural parameters and equivalent electrical parameters, and has become an important technical means in the field of transformer winding deformation detection.

[0003] In transformer winding deformation detection, the extraction and analysis of oscillation wave signals currently commonly employs a fixed time window approach. However, this method lacks adaptability to the oscillation decay characteristics of different windings in practical applications. Due to differences in the equivalent inductance, capacitance, and other parameters of high, medium, and low-voltage windings, their oscillation frequencies and decay processes vary. A uniform fixed window approach is prone to problems: for windings with rapid decay, low-energy redundant data appears in the latter part of the window, interfering with feature extraction and similarity calculation; for windings with slow decay, the fixed window cannot cover the entire effective oscillation segment, leading to the loss of crucial information; in multi-phase detection, it can also cause deviations in the proportion of effective information for different phases, affecting the reliability of the analysis. Furthermore, some techniques determine winding deformation by comparing the similarity of oscillation waves, but interference factors such as phase misalignment during on-site sampling, trigger delay, and environmental fluctuations can cause waveform shifts, affecting the calculation results. Relying solely on a single indicator such as the correlation coefficient for judgment is prone to misjudgment and cannot meet the requirements for accurate detection. Summary of the Invention

[0004] This invention provides a method, apparatus, device, and storage medium for detecting transformer winding deformation, in order to solve the technical problems in the prior art where the fixed time window method cannot cover the effective segments of different windings, has large noise interference, and is easily affected by waveform misalignment interference, leading to misjudgment.

[0005] To address the aforementioned technical problems, embodiments of the present invention provide a method for detecting transformer winding deformation, comprising: Obtain the oscillation wave signal of any phase winding of a transformer; Set several initial sliding windows, and recursively update the noise baseline of the oscillation wave signal in the initial sliding window; extract the intermediate oscillation wave from the oscillation wave signal based on the updated noise baseline; Set up several end sliding windows, and calculate the variance indicator of the intermediate oscillation wave corresponding to the end sliding window based on the variance baseline of the intermediate oscillation wave; extract the final oscillation wave from the intermediate oscillation wave based on the variance indicator. Based on the time series index corresponding to the final oscillation wave, extract the corresponding effective oscillation wave from the oscillation wave signals of the remaining phase windings; The correlation coefficient and dominant frequency between the effective oscillation waves of each phase winding are calculated, and the deformation of the transformer winding is detected based on the correlation coefficient and dominant frequency.

[0006] As a preferred embodiment, the step of setting several initial sliding windows and recursively updating the noise baseline of the oscillation wave signal in the initial sliding windows specifically includes: Set up several initial sliding windows; Calculate the slope sequence of the oscillating wave signal, and based on the slope sequence, calculate the average slope amplitude and slope standard deviation for each initial sliding window. Initialize the noise slope mean and noise slope standard deviation, and set the noise baseline online update conditions and noise baseline recursive update function based on the average slope amplitude and slope standard deviation of each initial sliding window, as well as the noise slope mean and noise slope standard deviation. Based on the online noise baseline update conditions and the noise baseline recursive update function, the noise baseline of the oscillating wave signal in the initial sliding window is recursively updated one by one.

[0007] As a preferred embodiment, the step of recursively updating the noise baseline of the oscillating wave signal in the initial sliding window one by one based on the online noise baseline update conditions and the noise baseline recursive update function specifically includes: If the oscillation wave signal in the current initial sliding window meets the online noise baseline update condition, then the noise baseline of the oscillation wave signal in the current initial sliding window is updated based on the noise baseline recursive update function. If the oscillation signal in the current starting sliding window does not meet the online noise baseline update condition, then the noise baseline of the oscillation signal in the current starting sliding window is updated according to the previous starting sliding window.

[0008] As a preferred embodiment, the extraction of intermediate oscillation waves from the oscillation wave signal based on the updated noise baseline specifically includes: Based on the updated noise baseline, calculate the adaptive threshold corresponding to the current starting sliding window; The current starting sliding window is determined based on the adaptive threshold and used as the starting point of the oscillation wave. Based on the starting point of the oscillation wave, the intermediate oscillation wave in the oscillation wave signal is extracted.

[0009] As a preferred embodiment, the step of setting several end sliding windows and calculating the variance indicator of the intermediate oscillation wave corresponding to the end sliding window based on the variance baseline of the intermediate oscillation wave; extracting the final oscillation wave from the intermediate oscillation wave based on the variance indicator specifically includes: Set up several end-sliding windows; The fluctuation amplitude of each end sliding window is calculated based on the intermediate oscillation wave in each end sliding window. Based on the fluctuation amplitude of each end sliding window, the variance baseline of the intermediate oscillation wave is calculated, and a dynamic threshold is set based on the variance baseline. Based on the dynamic threshold, determine the variance indicator of the intermediate oscillation wave corresponding to each end sliding window; Determine the minimum index corresponding to the variance indicator, and use the sliding window at the end of the minimum index as the termination point of the oscillation wave; Based on the termination point of the oscillation wave, the final oscillation wave in the intermediate oscillation wave is extracted.

[0010] As a preferred embodiment, the calculation of the correlation coefficient and dominant frequency between the effective oscillation waves of each phase winding, and the detection of transformer winding deformation based on the correlation coefficient and dominant frequency, specifically includes: Calculate the correlation coefficient between the effective oscillation waves of each phase winding, and compare the correlation coefficient with a preset threshold. When the correlation coefficient is not less than a preset threshold, the transformer winding is normal and without deformation. When the correlation coefficient is less than the preset threshold, the main frequency of the oscillation wave of each phase winding of the transformer and its difference are calculated, and the difference of the main frequency of each phase winding is compared with the preset ratio. When the difference is less than the preset ratio, the transformer winding is normally undeformed. If the difference is not less than the preset ratio, then the transformer winding has a deformation fault.

[0011] As a preferred embodiment, the calculation of the dominant frequency of the oscillation wave of each phase winding of the transformer specifically includes: Set the raised cosine window function, and calculate the window function based on the raised cosine window function and the effective oscillation wave; Based on the length of the effective oscillation wave, the windowed signal of the raised cosine window function is calculated, and the zero-fill signal is determined in combination with the window function. The zero-fill signal is transformed, and the positive frequency part of the transformed zero-fill signal is extracted to obtain the initial peak point; Parabolic interpolation is performed on the initially selected peak points to obtain interpolation coefficients, and the vertex correction amount is calculated using the interpolation coefficients. Based on the vertex correction amount, the initial peak point and the length of the effective oscillation wave, combined with the preset sampling frequency, the main frequency of the oscillation wave of each phase winding is calculated.

[0012] Accordingly, the present invention also provides a transformer winding deformation detection device, comprising: The acquisition module is used to acquire the oscillation wave signal of any phase winding of the transformer. The front-end noise module is used to set several initial sliding windows and recursively update the noise baseline of the oscillation wave signal in the initial sliding windows; it determines the starting point of the oscillation wave based on the updated noise baseline and extracts the intermediate oscillation wave from the oscillation wave signal. The end noise module is used to set several end sliding windows and calculate the variance indicator of the intermediate oscillation wave corresponding to the end sliding window based on the variance baseline of the intermediate oscillation wave; determine the oscillation wave termination point based on the variance indicator, and extract the final oscillation wave from the intermediate oscillation wave. The extraction module is used to extract the corresponding effective oscillation wave from the oscillation wave signals of the remaining phase windings according to the time series index corresponding to the final oscillation wave; The detection module is used to calculate the correlation coefficient and main frequency between the effective oscillation waves of each phase winding, and to detect the deformation of the transformer winding based on the correlation coefficient and main frequency.

[0013] Accordingly, the present invention also provides a terminal device, including a processor, a memory, and a computer program stored in the memory and configured to be executed by the processor, wherein the processor executes the computer program to implement the transformer winding deformation detection method as described above.

[0014] Accordingly, the present invention also provides a computer-readable storage medium comprising a stored computer program, wherein, when the computer program is executed, it controls the device containing the computer-readable storage medium to perform the transformer winding deformation detection method as described above.

[0015] Compared with the prior art, the embodiments of the present invention have the following beneficial effects: The technical solution of this invention, after acquiring the oscillation wave signal of any phase winding of a transformer, removes pre-existing noise and environmental noise from the stable attenuation segment of the final oscillation wave signal by setting initial and final sliding windows respectively. This allows for the identification of valid oscillation wave data. Simultaneously, by combining the initial and final sliding windows, the noise baseline of the oscillation wave signal is recursively updated and the variance indicator is calculated. This avoids the problem of not being able to cover the effective segments of different windings due to fixed time windows, and also avoids the problems of large noise interference and susceptibility to waveform misalignment interference leading to misjudgment. This improves the extraction of oscillation wave signals from subsequent phase windings and ensures the accuracy and stability of transformer winding deformation diagnosis. Furthermore, by using the time series index corresponding to the final oscillation wave of the extracted front-end and rear-end noise, the valid oscillation waves of the corresponding remaining phase windings can be extracted, ensuring that the effective segments of the three phases are completely consistent, avoiding misjudgment caused by waveform misalignment interference, facilitating subsequent comparison of correlation coefficients and main frequencies, and improving the accuracy and precision of the judgment. Attached Figure Description

[0016] Figure 1 : A flowchart illustrating the steps of a transformer winding deformation detection method provided in an embodiment of the present invention; Figure 2 This is a flowchart illustrating the adaptive method for extracting effective oscillation waves and determining winding deformation provided in this embodiment of the invention. Figure 3 : This is a waveform diagram of the oscillation wave of the high-voltage and low-voltage windings of the transformer, measured on-site; Figure 3 (a): Waveform diagram of the oscillation wave of the transformer high-voltage winding measured on site; Figure 3 (b): Waveform diagram of the oscillation wave of the low-voltage winding of the transformer measured on site; Figure 4 : The effective oscillation waveforms of the high-voltage and low-voltage windings of the transformer extracted by the adaptive method provided in the embodiments of the present invention; Figure 4 (a): The effective oscillation waveform of the transformer high-voltage winding extracted by the adaptive method provided in the embodiment of the present invention; Figure 4 (b): The effective oscillation waveform of the low-voltage winding of the transformer extracted by the adaptive method provided in the embodiment of the present invention; Figure 5 The waveform of the oscillation wave of the low-voltage winding of the 220kV autotransformer was measured on-site. Figure 6 : The effective oscillation waveform of the low-voltage winding of the transformer extracted by the adaptive method provided in the embodiments of the present invention; Figure 7: This is a structural diagram of a transformer winding deformation detection device provided in an embodiment of the present invention. Detailed Implementation

[0017] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0018] Example 1 Please refer to Figure 1 The present invention provides a method for detecting transformer winding deformation, comprising the following steps S101-S105: S101: Obtain the oscillation wave signal of any phase winding of the transformer.

[0019] In this embodiment, the three-phase oscillation wave signal of the transformer is mainly obtained through offline oscillation wave testing or online high-frequency monitoring. By setting the sampling position at the port of the three-phase (or multi-winding) transformer under test, the discrete oscillation wave signal data of any phase winding of the corresponding transformer can be measured. The port can be the end of the transformer bushing or the end screen of the bushing, but for a single deformation detection, the port position measured for each corresponding oscillation wave signal is the same.

[0020] In this embodiment, by setting the sampling location, the total number of sampling points, and the time index corresponding to each sampling point, the corresponding discrete oscillation wave signal data can be obtained through sampling and measurement. s p [n] : in, p The transformer is labeled with its phase identifiers, including phases A, B, and C. n For sampling point index, N This represents the total number of sampling points.

[0021] S102: Set several initial sliding windows and recursively update the noise baseline of the oscillation wave signal of the initial sliding window in turn; extract the intermediate oscillation wave from the oscillation wave signal based on the updated noise baseline.

[0022] It should be noted that during on-site transformer winding oscillation wave testing, the signals acquired in the initial stage of the test are typically influenced by factors such as the triggering sequence of the high-voltage switch, transient disturbances of the switch, electromagnetic interference in the measurement circuit, and transient response of the system upon power-up. These signals usually only reflect the background noise of the measuring equipment, wiring circuit, and the tested winding under static conditions, and do not contain effective oscillation wave information that characterizes the winding's properties. To ensure the accuracy of subsequent signal analysis and feature extraction, this invalid noise signal needs to be removed.

[0023] Understandably, before the oscillating wave signal appears, the acquired waveform mainly exhibits a stable noise distribution. Due to differences in excitation triggering timing and loop response characteristics among different oscillating wave testing methods, the duration of the initial noise segment is not fixed. If a fixed-length data segment is used to calculate the noise baseline, the truncation range may include part of the initial oscillation signal, leading to an overestimation of the noise baseline and affecting the accuracy of subsequent signal processing and feature extraction.

[0024] In this embodiment, a sliding window-based adaptive noise baseline update method is employed. This method combines the mean slope change and the degree of slope fluctuation to comprehensively determine the effective starting point of the oscillation wave. By setting a sliding window and adaptively recursively updating the noise baseline corresponding to the oscillation wave signal within each sliding window, the starting point of the oscillation wave between the front-end noise and the oscillation wave signal is identified based on the updated noise baseline. The oscillation wave signal data after the starting point is then extracted as the intermediate oscillation wave.

[0025] S103: Set several end sliding windows, and calculate the variance indicator of the intermediate oscillation wave corresponding to the end sliding window based on the variance baseline of the intermediate oscillation wave; extract the final oscillation wave from the intermediate oscillation wave based on the variance indicator.

[0026] It should be noted that during the oscillation wave test, after the excitation-generated oscillation signal has fully decayed and entered the subsequent stable phase, the acquired waveform data mainly consists of environmental electromagnetic interference, inherent noise of the measurement system, and stray signals from the grounding loop. This stable segment data essentially does not contain effective information reflecting the internal structural characteristics, electrical parameter distribution, and insulation status of the transformer windings. Directly using this data in subsequent feature extraction and state analysis would not only significantly increase computational redundancy in data processing but may also introduce noise interference, leading to deviations in the analysis results and reducing the reliability of winding defect diagnosis. Therefore, in the signal processing stage, it is necessary to adaptively extract the effective oscillation wave data segments and accurately eliminate the stable noise segments after the oscillation has fully decayed, thereby ensuring that subsequent analysis focuses only on effective signals with diagnostic value.

[0027] Understandably, to improve the sensitivity and adaptability of effective signal segment identification and avoid identification deviations under different test conditions and winding parameters using fixed threshold criteria, this embodiment adopts an adaptive identification method based on sliding window variance. By setting a continuously sliding data window on the acquired waveform, the variance characteristics of the signal within the window are calculated segment by segment. Effective differentiation is achieved by utilizing the significant differences in signal fluctuation amplitude and energy distribution between oscillating segments and stationary noise segments. During periods of oscillation, the signal amplitude changes drastically and the variance is large, while after attenuation, the signal fluctuation in the stationary segment is weak and the variance approaches the background noise level. Based on the dynamic change law of the variance characteristics, the start and end times of the oscillating wave can be automatically located, enabling accurate extraction of effective data segments and reliable removal of invalid stationary segments. This improves signal identification sensitivity and enhances the robustness of the algorithm in variable field test environments, laying a high-quality data foundation for subsequent winding characteristic analysis and defect diagnosis.

[0028] In this embodiment, a sliding window variance identification method is used to adaptively identify the stable data interval after the oscillation wave decays, thereby extracting effective oscillation wave data. By setting several end sliding windows and simultaneously calculating the variance baseline of the intermediate oscillation wave, the variance indicator of the intermediate oscillation wave for the corresponding end sliding window is calculated based on the variance baseline. The final oscillation wave is then extracted from the intermediate oscillation wave using the variance indicator.

[0029] S104: Based on the time series index corresponding to the final oscillation wave, extract the corresponding effective oscillation wave from the oscillation wave signals of the remaining phase windings.

[0030] In this embodiment, based on the time series index corresponding to the final oscillation wave of one phase, the original oscillation wave signals collected from the remaining phases of the three-phase winding of the transformer are synchronously intercepted and effective data is extracted. Using a unified time start and end interval as a benchmark, i.e., the time series index of the corresponding sampling point, the effective time interval of the corresponding final oscillation wave is determined. The background noise segment before oscillation triggering and the stable noise segment after complete oscillation decay in each phase signal are removed, retaining the effective oscillation waveform that truly reflects the winding structure characteristics and insulation response characteristics.

[0031] In this embodiment, by using timing alignment and data extraction methods, the effective oscillation wave signals of the three-phase windings A, B, and C can be guaranteed to be consistent and comparable in the time dimension. This avoids analysis deviations caused by differences in the trigger delay and signal duration of each phase, thereby providing a regular and reliable effective data foundation for subsequent three-phase waveform comparison, frequency domain feature analysis, and winding state consistency assessment, and improving the accuracy and stability of winding defect diagnosis.

[0032] S105: Calculate the correlation coefficient and dominant frequency between the effective oscillation waves of each phase winding, and detect the deformation of the transformer winding based on the correlation coefficient and dominant frequency.

[0033] In this embodiment, after extracting and preprocessing the effective oscillation wave signals of each phase winding, the effective oscillation waveforms of phases A, B, and C are quantitatively characterized and calculated. The correlation coefficients between phase signals and the dominant frequency parameters corresponding to each phase waveform are then obtained sequentially. The correlation coefficient measures the similarity of the oscillation waveforms of different phases in terms of time-domain variation trends and amplitude response patterns. The dominant frequency reflects the inherent resonance characteristics of the winding under oscillating excitation and can intuitively represent the distribution of parameters such as winding inductance and capacitance. After obtaining the correlation coefficients and dominant frequencies between the effective oscillation waves of each phase winding, the difference in the three-phase correlation coefficients and the dominant frequency offset are used as core criteria to construct a winding deformation identification basis. By comparing the characteristic benchmark values ​​under normal operating conditions, it is determined whether the winding has structural defects such as displacement, torsion, or short circuits. Based on the joint analysis of correlation coefficients and dominant frequencies, the reliability and sensitivity of winding deformation detection can be effectively improved, avoiding misjudgments and omissions caused by single feature criteria, and achieving accurate assessment of the health status of the transformer windings.

[0034] Implementing the above embodiments has the following effects: The technical solution of this invention, after acquiring the oscillation wave signal of any phase winding of a transformer, removes pre-existing noise and environmental noise from the stable attenuation segment of the final oscillation wave signal by setting initial and final sliding windows respectively. This allows for the identification of valid oscillation wave data. Simultaneously, by combining the initial and final sliding windows, the noise baseline of the oscillation wave signal is recursively updated and the variance indicator is calculated. This avoids the problem of not being able to cover the effective segments of different windings due to fixed time windows, and also avoids the problems of large noise interference and susceptibility to waveform misalignment interference leading to misjudgment. This improves the extraction of oscillation wave signals from subsequent phase windings and ensures the accuracy and stability of transformer winding deformation diagnosis. Furthermore, by using the time series index corresponding to the final oscillation wave of the extracted front-end and rear-end noise, the valid oscillation waves of the corresponding remaining phase windings can be extracted, ensuring that the effective segments of the three phases are completely consistent, avoiding misjudgment caused by waveform misalignment interference, facilitating subsequent comparison of correlation coefficients and main frequencies, and improving the accuracy and precision of the judgment.

[0035] Example 2 Please see Figure 2 The present invention also provides several preferred embodiments of the transformer winding deformation detection method, as follows: In a preferred embodiment, the step of setting a plurality of initial sliding windows and recursively updating the noise baseline of the oscillation wave signal of the initial sliding windows specifically includes: Set up several initial sliding windows; Calculate the slope sequence of the oscillating wave signal, and based on the slope sequence, calculate the average slope amplitude and slope standard deviation for each initial sliding window. Initialize the noise slope mean and noise slope standard deviation, and set the noise baseline online update conditions and noise baseline recursive update function based on the average slope amplitude and slope standard deviation of each initial sliding window, as well as the noise slope mean and noise slope standard deviation. Based on the online noise baseline update conditions and the noise baseline recursive update function, the noise baseline of the oscillating wave signal in the initial sliding window is recursively updated one by one.

[0036] It should be noted that in on-site oscillation wave measurements, due to factors such as switch triggering, the initial measured signal is often the background noise of the equipment and windings in a static state, which does not contain effective oscillation wave information and therefore needs to be discarded. Before the oscillation wave appears, the waveform is in a stable noise state. Considering that the initial noise length is not the same in different oscillation wave test methods, if a fixed length of data is truncated in advance to calculate the noise baseline, part of the oscillation signal may be intercepted, resulting in an excessively high baseline. This embodiment adopts a noise baseline adaptive update method based on a sliding window, combining the mean slope change and the degree of slope fluctuation to comprehensively determine the effective starting point of the oscillation wave.

[0037] In this embodiment, the oscillation wave signal obtained from engineering measurements is s p [ n ] n=0,1,…,N-1, Sampling interval △ t ;in, p For the identification of each phase of the transformer, n This serves as the sampling point index. The first derivative of the discrete oscillating wave signal is then calculated to obtain the slope sequence. d[n] : In the noise zone d[n] The value fluctuates randomly, and in the rising phase of the oscillation wave, d[n] Then it increases significantly.

[0038] Set the initial sliding window and calculate the average slope magnitude. S mean This is done to smooth the slope and suppress noise spikes; the length of the initial sliding window is set to... W s : Set the starting sliding window i Calculate the standard deviation of the slope σd ( i ): in, i It is the starting position index of the initial sliding window. k Indicates the first sliding window within the initial sliding window k discrete points, This represents the mean slope of the initial sliding window; the initial noise slope mean is initialized using the waveform of the first window as a noise reference. u slp (0) and initial noise slope standard deviation initialization σ slp (0): The conditions for online noise baseline updates are set, and the specific criteria are as follows: in, R th_slp and R th_sigma The fold change factor for determining the characteristic mutation is 1.5 to 2.0.

[0039] As a preferred embodiment, the step of recursively updating the noise baseline of the oscillating wave signal in the initial sliding window one by one based on the online noise baseline update condition and the noise baseline recursive update function specifically includes: If the oscillation wave signal in the current initial sliding window meets the online noise baseline update condition, then the noise baseline of the oscillation wave signal in the current initial sliding window is updated based on the noise baseline recursive update function. If the oscillation signal in the current starting sliding window does not meet the online noise baseline update condition, then the noise baseline of the oscillation signal in the current starting sliding window is updated according to the previous starting sliding window.

[0040] As a preferred embodiment, the step of extracting the intermediate oscillation wave from the oscillation wave signal based on the updated noise baseline specifically includes: Based on the updated noise baseline, calculate the adaptive threshold corresponding to the current starting sliding window; The current starting sliding window is determined based on the adaptive threshold and used as the starting point of the oscillation wave. Based on the starting point of the oscillation wave, the intermediate oscillation wave in the oscillation wave signal is extracted.

[0041] In this embodiment, if both judgment criteria in the online noise baseline update conditions are met, it indicates that the current initial sliding window is still in a noisy state and can be included in the noise baseline update. The noise baseline recursive update formula is: Wherein, α is the update weight, which takes a value between 0 and 1. Preferably, α is between 0.05 and 0.2, used for smoothing estimation.

[0042] If the two judgment criteria in the online update conditions for the noise baseline are not met simultaneously, then the update conditions are not satisfied, and therefore: For each window position i The formula for calculating the adaptive threshold is: in, K u and K σ This is the threshold amplification factor, preferably between 3 and 5.

[0043] Finally, the criterion for the starting point of the oscillating wave is: in, j Therefore i The global index within the starting sliding window is used to traverse all consecutive child windows covered by the current window, achieving consistency determination across multiple consecutive windows; and this condition must be continuously satisfied. L s windows (L) s If the value is 2~3, then the first window starting point that meets the condition will be taken as the starting point of the oscillation wave. Simultaneously, the initial noise data of the oscillation wave is removed to obtain the intermediate oscillation wave. ss p [ n ].

[0044] In this embodiment, after constructing the online update conditions and recursive update function for the noise baseline, a dynamic update process for the noise baseline and an identification process for the oscillation starting point are performed for the oscillation signal within each initial sliding window. Specifically, the slope statistical characteristics (average slope amplitude, slope standard deviation) of the current initial sliding window are compared with the online update conditions for the noise baseline: if the current initial sliding window signal meets the update conditions, that is, its fluctuation characteristics are highly consistent with the noise baseline characteristics, it is determined that the initial sliding window is still in the noise stable segment. Then, based on the noise baseline recursive update function, the statistical characteristics of the current initial sliding window are fused to iteratively correct the noise baseline parameters to achieve adaptive baseline update; if the current initial sliding window signal does not meet the update conditions, that is, its fluctuation amplitude deviates significantly from the noise baseline, it is determined that the initial sliding window has entered the oscillation transition stage. Then, the noise baseline parameters of the previous initial sliding window are directly used as the noise baseline of the current initial sliding window to avoid interference of the oscillation signal on the noise baseline.

[0045] Furthermore, based on the updated noise baseline parameters, an adaptive threshold corresponding to the current starting sliding window is calculated. This adaptive threshold comprehensively considers the mean and fluctuation range of the noise baseline and can dynamically adapt to the noise levels of different field environments. Then, the statistical characteristics of the current window are compared with the adaptive threshold. When the characteristic value first exceeds the threshold, the starting sliding window is determined as the starting point of the oscillation wave. Finally, using the identified starting point of the oscillation wave as a benchmark, combined with the determination result of the stable segment after oscillation wave attenuation, the effective intermediate oscillation wave segment in the oscillation wave signal is accurately extracted, and the preceding and following noise segments are eliminated. This provides a high-quality effective signal input for subsequent winding deformation detection, ensuring the accuracy and reliability of feature extraction and state discrimination.

[0046] Understandably, in existing technologies, the determination of the initial segment of an oscillation wave typically employs fixed amplitude thresholds, fixed time windows, or manual experience-based selection methods. These methods struggle to adapt to variations in noise levels and trigger states under different testing conditions, easily leading to inaccurate identification of the effective oscillation segment's starting position. To address these issues, this embodiment proposes an adaptive identification method for the effective starting point of an oscillation wave based on the slope characteristics of a sliding window. By jointly analyzing the average slope and slope fluctuation within the sliding window, the method detects the characteristic changes in the oscillation wave signal as it transitions from the preceding noise region to the effective oscillation region, thereby achieving automatic determination of the effective starting point.

[0047] Furthermore, this embodiment adaptively updates the judgment threshold based on the statistical characteristics of the noise in the leading edge of the oscillating wave, enabling the starting point identification method to adapt to different noise levels, different triggering conditions, and different test site environments, thereby improving the accuracy and stability of the oscillating wave starting point identification.

[0048] In a preferred embodiment, the step of setting several end sliding windows and calculating the variance indicator of the intermediate oscillation wave corresponding to the end sliding window based on the variance baseline of the intermediate oscillation wave, and extracting the final oscillation wave from the intermediate oscillation wave based on the variance indicator, specifically includes: Set up several end-sliding windows; The fluctuation amplitude of each end sliding window is calculated based on the intermediate oscillation wave in each end sliding window. Based on the fluctuation amplitude of each end sliding window, the variance baseline of the intermediate oscillation wave is calculated, and a dynamic threshold is set based on the variance baseline. Based on the dynamic threshold, determine the variance indicator of the intermediate oscillation wave corresponding to each end sliding window; Determine the minimum index corresponding to the variance indicator, and use the sliding window at the end of the minimum index as the termination point of the oscillation wave; Based on the termination point of the oscillation wave, the final oscillation wave in the intermediate oscillation wave is extracted.

[0049] It should be noted that the intermediate oscillation wave is obtained by removing front-end noise from the initial oscillation wave signal. Furthermore, by determining the termination point of the oscillation wave, the corresponding final oscillation wave can be extracted from the intermediate oscillation wave. That is, the final oscillation wave is obtained by removing the useless wave signal after the termination point of the oscillation wave from the intermediate oscillation wave (i.e., after filtering the front-end noise). Finally, the front-end noise and the useless back-end signal are removed and filtered to obtain the oscillation wave that is actually effective for the corresponding phase winding.

[0050] In this embodiment, the stable segment data after oscillation wave attenuation mainly consists of environmental noise and contains almost no winding structure information. Therefore, it is necessary to adaptively extract effective oscillation wave data and eliminate the stable segment of oscillation wave attenuation. To improve sensitivity, a sliding window variance identification method is adopted to adaptively identify effective oscillation wave data. To achieve accurate definition of the effective time period of the oscillation wave, an end-sliding window mechanism is constructed in the signal processing flow. Specifically, based on the time-domain distribution characteristics of the intermediate oscillation wave, several sets of end-sliding windows are set to cover the tail region of the waveform to be analyzed, serving as the basic identification unit for determining oscillation termination. Furthermore, using each end-sliding window as an independent calculation unit, amplitude statistical analysis is performed on the intermediate oscillation wave waveform contained within the window to obtain the fluctuation amplitude characteristic quantity corresponding to each window, thereby quantifying the energy concentration and fluctuation intensity of the signal within the window. Based on this, the fluctuation amplitude data of all end-sliding windows are calculated, and the variance baseline of the intermediate oscillation wave is obtained by fitting and calculating. The variance baseline reflects the stable fluctuation range of the effective oscillation signal in this frequency band. A dynamic threshold determination model is constructed based on the variance baseline. The dynamic threshold can adaptively adjust the threshold boundary to effectively distinguish between the oscillation attenuation stage and the pure noise stage. Based on the dynamic threshold, the fluctuation amplitude of each sliding window at the end is quantified and the corresponding variance indicator is calculated, which serves as the basis for judging the effectiveness of the window.

[0051] Furthermore, the minimum index position of the variance indicator is retrieved, where this position corresponds to the critical point where the oscillation energy decays to the noise level. The sliding window corresponding to this index is then determined as the oscillation wave termination point. Finally, using the identified oscillation wave termination point as a benchmark, combined with the previously determined oscillation wave start point, the effective oscillation wave data segments in the intermediate oscillation wave are completely extracted, achieving accurate removal of invalid noise segments.

[0052] In this embodiment, the measured intermediate oscillation wave signal is ssp[n] , u(i) The mean of the window, the window length Wp : in, Var(i) The larger the value, the greater the fluctuation amplitude within the sliding window at the end, indicating a valid oscillating wave. Var (i) When the waveform is below the threshold, it indicates that the waveform has decayed to the level of noise.

[0053] Furthermore, a noise baseline estimate and a dynamic threshold are set, wherein the variance baseline of the background noise is estimated using the stable segment of the later part of the oscillation wave. ,Right now: Furthermore, a dynamic threshold can be set to... Vth : in, K th The safety margin factor is preferably 2 to 5.

[0054] Setting a criterion for a continuous low-variance stationary segment effectively avoids misjudgments caused by fluctuations within a single window, making the cutoff point more robust. Define the low-variance indicator: Find the minimum index i * ,satisfy: Among them, here j Therefore i The global index within the sliding window starting from the current window is used to traverse all consecutive child windows covered by the current window, thus achieving consistency determination for multiple consecutive windows.

[0055] The termination point of the oscillating wave is: If no matching criteria are found i *,but: It is understandable that existing methods for truncating the effective segment of oscillating waves typically use a fixed time length, which is difficult to adapt to different winding damping characteristics and different oscillation decay rates. This can easily lead to insufficient or excessive truncation of the effective oscillation segment, thus affecting the accuracy of subsequent feature calculations. To address these issues, this embodiment proposes an adaptive truncating method for effective oscillating waves based on the dynamic variation of the variance of a sliding window. By analyzing the variance variation law during the transition from the effective oscillation region to the stable noise region at the tail end of the oscillating wave, and combining this with the noise statistics of the final stable region, an adaptive judgment threshold is set, thereby automatically determining the termination position of the effective oscillating wave. Furthermore, this embodiment uses a continuous low variance criterion to constrain the truncation point, reducing the risk of misjudgment caused by accidental fluctuations in a single window, local noise anomalies, or short-term disturbances, and improving the reliability of the effective oscillating wave truncating results. For multiphase oscillating wave data, this embodiment uses a unified termination position for synchronous truncating after determining the termination position, ensuring that the data intervals used for each comparison are consistent. This provides a consistent data basis for subsequent correlation coefficient calculations and main frequency comparisons, improving the comparability and accuracy of multiphase analysis results.

[0056] Furthermore, once the termination point of the oscillation wave is determined, multiphase synchronous interception can be performed, that is, when the termination point of the oscillation wave... n stopOnce determined, the oscillation wave signal of each phase is synchronously captured to obtain the effective oscillation wave waveform of each phase. ss p,eff : in, ss p [ n Since it is an oscillating wave sequence aligned with the starting point, its index... n =0 corresponds to the original signal in n start Location.

[0057] The time series is: This ensures that the effective segments of the three phases are completely consistent, which facilitates the comparison of correlation coefficients and main frequencies in the future.

[0058] It should be noted that the determination of the oscillation wave starting point is also based on the same operation described above. That is, after determining the oscillation wave starting point, the oscillation wave signal of each phase is synchronously intercepted based on the corresponding time series index to obtain the effective oscillation wave waveform of each phase. It can be understood that the final oscillation wave obtained earlier is essentially the oscillation wave signal in the corresponding phase after removing front-end noise and back-end useless signals. Therefore, for this embodiment, the final oscillation wave of the phase is essentially its corresponding effective oscillation wave. For the winding of the phase, the final oscillation wave can be directly used as its corresponding effective oscillation wave. For example, if the aforementioned steps are to obtain the oscillation wave signal of the A-phase winding, extract the intermediate oscillation wave of the A-phase winding, and extract the corresponding final oscillation wave based on the intermediate oscillation wave, then the effective oscillation wave of the A-phase winding is essentially the final oscillation wave extracted from the A-phase winding.

[0059] As a preferred embodiment, the calculation of the correlation coefficient and dominant frequency between the effective oscillation waves of each phase winding, and the detection of transformer winding deformation based on the correlation coefficient and dominant frequency, specifically includes: Calculate the correlation coefficient between the effective oscillation waves of each phase winding, and compare the correlation coefficient with a preset threshold. When the correlation coefficient is not less than a preset threshold, the transformer winding is normal and without deformation. When the correlation coefficient is less than the preset threshold, the main frequency of the oscillation wave of each phase winding of the transformer and its difference are calculated, and the difference of the main frequency of each phase winding is compared with the preset ratio. When the difference is less than the preset ratio, the transformer winding is normally undeformed. If the difference is not less than the preset ratio, then the transformer winding has a deformation fault.

[0060] In this embodiment, the effective oscillation waves of each phase winding of the transformer are calculated respectively. ss p,eff The correlation coefficient between phases is used to determine the winding status. When the correlation coefficient of each phase winding is not less than a preset threshold (e.g., 0.9), the winding status can be determined to be normal. When the correlation coefficient is lower than the threshold, the main frequency auxiliary discrimination step is initiated. The preset threshold can be set according to the actual situation and specific needs.

[0061] In this embodiment, the dominant frequency of the oscillation wave of each phase winding of the transformer is calculated. When the difference in dominant frequency between phases is less than a preset ratio, the decrease in correlation coefficient can be considered to be mainly caused by waveform phase shift. When the difference in dominant frequency between phases is greater than a preset ratio, it can be determined that the winding has a deformation fault. Preferably, the preset ratio is 5%.

[0062] As a preferred embodiment, the calculation of the dominant frequency of the oscillation wave of each phase winding of the transformer specifically includes: Set the raised cosine window function, and calculate the window function based on the raised cosine window function and the effective oscillation wave; Based on the length of the effective oscillation wave, the windowed signal of the raised cosine window function is calculated, and the zero-fill signal is determined in combination with the window function. The zero-fill signal is transformed, and the positive frequency part of the transformed zero-fill signal is extracted to obtain the initial peak point; Parabolic interpolation is performed on the initially selected peak points to obtain interpolation coefficients, and the vertex correction amount is calculated using the interpolation coefficients. Based on the vertex correction amount, the initial peak point and the length of the effective oscillation wave, combined with the preset sampling frequency, the main frequency of the oscillation wave of each phase winding is calculated.

[0063] It should be noted that traditional FFT directly transforms truncated oscillatory signals, resulting in severe spectral leakage, especially for oscillatory signals with a period number less than an integer multiple, where peak energy diffuses. This embodiment employs a Hann window (raised cosine window), which significantly suppresses sidelobe leakage and improves the accuracy of main peak localization. Since the change in winding deformation must not exceed 5% when using the main frequency to assist in determining winding deformation, high-precision calculation of the main peak is crucial. This is particularly effective for waveforms with high oscillation frequencies and rapid short-term decay, such as those from transformer low-voltage windings. The frequency resolution of conventional FFT is... △f=Fs / N The estimation error of the main frequency of short signals is large. This invention fills the signal with zeros to 4 times or more of the original length, making the frequency grid denser and improving the positioning accuracy of the main frequency peak.

[0064] In this embodiment, the raised cosine window function (Hann function) is a cosine-weighted function used to reduce boundary discontinuities when the signal is truncated, and is defined as: in, N e For the effective oscillation wave length, This is a raised cosine window function. Multiplying the raised cosine window function by the signal point by point yields the windowed short-time signal sequence. H p [ n ]: Then, a zero-fill FFT is performed to improve frequency resolution. Specifically, the windowed signal is first extended to... N FFT point: in, K Zero fill ratio, preferably, K ≥ 4, to improve the frequency resolution of the main peak search. Define the zero-fill signal: Perform a Fast Fourier Transform on the above zero-filled signal: Among them, the above formula k The index represents the frequency domain sequence, indicating the nth element in the frequency domain. k The location of each frequency component is then determined; subsequently, the main peak is located and the frequency is refined using parabolic interpolation, taking the positive frequency components. Among them, the above formula k It is the first in the frequency domain k The position index of each frequency component in the positive frequency range corresponds to the frequency sampling point of the Discrete Fourier Transform. Then, the initial selection of the point of maximum amplitude is performed, ignoring the DC component. k =0), find the initial peak point k m : Perform parabolic interpolation, and set the interpolation coefficients of the parabola. They are respectively: Perform vertex correction The calculation formula is: Finally, the formula for calculating high-precision main frequency estimation is: in, f peak Main frequency, F sIt is the sampling frequency.

[0065] It is understandable that the oscillation wave discrimination of transformer winding conditions typically uses a single waveform similarity index, such as the correlation coefficient to reflect the consistency between oscillation waves of different windings. However, relying solely on the correlation coefficient is susceptible to testing factors such as waveform misalignment, delayed triggering, and phase drift, leading to misjudgments. To address these issues, this embodiment proposes a winding condition determination method based on the fusion of correlation coefficient and dominant frequency characteristics. Firstly, the correlation coefficient between the effective oscillation waves of each phase winding is used as the primary criterion. When the correlation coefficient is lower than a preset threshold, the tested winding is judged as potentially abnormal. Based on this, this embodiment further introduces dominant frequency characteristics as an auxiliary criterion to verify the correlation coefficient determination results.

[0066] Specifically, during the main frequency extraction process, a Hann window is used to window the effective oscillation wave to suppress spectral leakage. Simultaneously, a zero-fill fast Fourier transform and parabolic interpolation method are combined to precisely estimate the main peak frequency of the oscillation wave, thereby improving the accuracy of the main frequency extraction. When the relative deviation of the main frequency between the oscillation waves of each phase winding exceeds a preset proportional threshold, it can be determined that the abnormal correlation coefficient is mainly caused by changes in winding structure parameters. When the relative deviation of the main frequency does not exceed the preset proportional threshold, it can be considered that the abnormal correlation coefficient is more likely caused by test factors such as waveform misalignment, delayed triggering, or phase drift. Through the above dual-criteria fusion method, the reliability of the winding state determination results can be improved, and the adaptability of this method to on-site interference factors can be enhanced.

[0067] Example 3 This invention also provides another embodiment for adaptive acquisition of effective oscillation waves in transformer winding deformation diagnosis: Taking the oscillation waves of the high-voltage and low-voltage windings of a 220kV three-phase combined transformer as an example, the adaptive oscillation wave extraction method proposed in this invention is implemented. The self-excited oscillation wave directly measured on-site is as follows: Figure 3 As shown, where, Figure 3 (a) shows the oscillation waveform of the high-voltage winding. Figure 3 (b) shows the oscillation waveform of the low-voltage winding. From Figure 3 It can be seen that the measured transformer winding oscillation waveform contains invalid oscillation data at both the beginning and end, mainly composed of noise, and is unrelated to the transformer winding condition. Their presence would obviously affect the judgment of the winding condition. The attenuation rate of the oscillation wave in the low-voltage winding is significantly higher than that in the high-voltage winding; therefore, the current method of capturing the oscillation waveform at a fixed time is inaccurate. Furthermore, regarding... Figure 3 The oscillating wave waveform in the figure is extracted using the adaptive oscillating wave extraction method proposed in this invention. The extracted effective oscillating wave waveform is as follows: Figure 4 As shown, where, Figure 4(a) shows the oscillation waveform of the high-voltage winding. Figure 4 (b) shows the oscillation waveform of the low-voltage winding. From Figure 4 As can be seen, this embodiment can effectively extract the oscillation waveforms of each phase winding of the transformer, eliminate noise interference, and effectively improve the accuracy of winding status judgment.

[0068] Example 4 This invention also provides another embodiment of transformer winding deformation detection that adaptively acquires effective oscillation waves: Taking the oscillation wave of the low-voltage winding of a 220kV autotransformer as an example, this invention implements the adaptive oscillation wave extraction method and winding deformation judgment. The self-excited oscillation wave directly measured on-site is shown below. Figure 5 As shown. The effective oscillation waveform of the transformer high-voltage winding extracted using the adaptive method proposed in this invention is as follows. Figure 6 As shown in the figure, the present invention effectively extracts the oscillation waveform of the transformer. A comparative analysis of the self-excited oscillation waveforms of phases A, B, and C reveals a significant offset in phase C. Furthermore, the correlation coefficients between the self-excited oscillation waveforms of AB, BC, and AC are calculated to be 0.99, 0.86, and 0.85, respectively. According to current methods, the C-phase winding of the transformer may have undergone deformation. Further, using the oscillation wave dominant frequency calculation method proposed in this invention, the dominant frequencies of the self-excited oscillation waves of phases A, B, and C are 5.00 kHz, 4.86 kHz, and 4.95 kHz, respectively. The difference in dominant frequencies between each phase winding is less than 5%, indicating that the transformer winding has not experienced deformation faults, consistent with the transformer disintegration situation. This demonstrates that the transformer winding condition diagnosis method proposed in this embodiment effectively improves the accuracy of the diagnosis.

[0069] Example 5 Please see Figure 7 The present invention provides a transformer winding deformation detection device, comprising: The acquisition module 201 is used to acquire the oscillation wave signal of any phase winding of the transformer; The front-end noise module 202 is used to set several initial sliding windows and recursively update the noise baseline of the oscillation wave signal of the initial sliding window in turn; determine the starting point of the oscillation wave based on the updated noise baseline, and extract the intermediate oscillation wave from the oscillation wave signal. The end noise module 203 is used to set a plurality of end sliding windows, and calculate the variance indicator of the intermediate oscillation wave corresponding to the end sliding window based on the variance baseline of the intermediate oscillation wave; determine the oscillation wave termination point based on the variance indicator, and extract the final oscillation wave from the intermediate oscillation wave; Extraction module 204 is used to extract the corresponding effective oscillation wave from the oscillation wave signals of the remaining phase windings according to the time series index corresponding to the final oscillation wave; The detection module 205 is used to calculate the correlation coefficient and main frequency between the effective oscillation waves of each phase winding, and to detect the deformation of the transformer winding based on the correlation coefficient and main frequency.

[0070] As a preferred embodiment, the step of setting several initial sliding windows and recursively updating the noise baseline of the oscillation wave signal in the initial sliding windows specifically includes: Set up several initial sliding windows; Calculate the slope sequence of the oscillating wave signal, and based on the slope sequence, calculate the average slope amplitude and slope standard deviation for each initial sliding window. Initialize the noise slope mean and noise slope standard deviation, and set the noise baseline online update conditions and noise baseline recursive update function based on the average slope amplitude and slope standard deviation of each initial sliding window, as well as the noise slope mean and noise slope standard deviation. Based on the online noise baseline update conditions and the noise baseline recursive update function, the noise baseline of the oscillating wave signal in the initial sliding window is recursively updated one by one.

[0071] As a preferred embodiment, the step of recursively updating the noise baseline of the oscillating wave signal in the initial sliding window one by one based on the online noise baseline update conditions and the noise baseline recursive update function specifically includes: If the oscillation wave signal in the current initial sliding window meets the online noise baseline update condition, then the noise baseline of the oscillation wave signal in the current initial sliding window is updated based on the noise baseline recursive update function. If the oscillation signal in the current starting sliding window does not meet the online noise baseline update condition, then the noise baseline of the oscillation signal in the current starting sliding window is updated according to the previous starting sliding window.

[0072] As a preferred embodiment, the extraction of intermediate oscillation waves from the oscillation wave signal based on the updated noise baseline specifically includes: Based on the updated noise baseline, calculate the adaptive threshold corresponding to the current starting sliding window; The current starting sliding window is determined based on the adaptive threshold and used as the starting point of the oscillation wave. Based on the starting point of the oscillation wave, the intermediate oscillation wave in the oscillation wave signal is extracted.

[0073] As a preferred embodiment, the step of setting several end sliding windows and calculating the variance indicator of the intermediate oscillation wave corresponding to the end sliding window based on the variance baseline of the intermediate oscillation wave; extracting the final oscillation wave from the intermediate oscillation wave based on the variance indicator specifically includes: Set up several end-sliding windows; The fluctuation amplitude of each end sliding window is calculated based on the intermediate oscillation wave in each end sliding window. Based on the fluctuation amplitude of each end sliding window, the variance baseline of the intermediate oscillation wave is calculated, and a dynamic threshold is set based on the variance baseline. Based on the dynamic threshold, determine the variance indicator of the intermediate oscillation wave corresponding to each end sliding window; Determine the minimum index corresponding to the variance indicator, and use the sliding window at the end of the minimum index as the termination point of the oscillation wave; Based on the termination point of the oscillation wave, the final oscillation wave in the intermediate oscillation wave is extracted.

[0074] As a preferred embodiment, the calculation of the correlation coefficient and dominant frequency between the effective oscillation waves of each phase winding, and the detection of transformer winding deformation based on the correlation coefficient and dominant frequency, specifically includes: Calculate the correlation coefficient between the effective oscillation waves of each phase winding, and compare the correlation coefficient with a preset threshold. When the correlation coefficient is not less than a preset threshold, the transformer winding is normal and without deformation. When the correlation coefficient is less than the preset threshold, the main frequency of the oscillation wave of each phase winding of the transformer and its difference are calculated, and the difference of the main frequency of each phase winding is compared with the preset ratio. When the difference is less than the preset ratio, the transformer winding is normally undeformed. If the difference is not less than the preset ratio, then the transformer winding has a deformation fault.

[0075] As a preferred embodiment, the calculation of the dominant frequency of the oscillation wave of each phase winding of the transformer specifically includes: Set the raised cosine window function, and calculate the window function based on the raised cosine window function and the effective oscillation wave; Based on the length of the effective oscillation wave, the windowed signal of the raised cosine window function is calculated, and the zero-fill signal is determined in combination with the window function. The zero-fill signal is transformed, and the positive frequency part of the transformed zero-fill signal is extracted to obtain the initial peak point; Parabolic interpolation is performed on the initially selected peak points to obtain interpolation coefficients, and the vertex correction amount is calculated using the interpolation coefficients. Based on the vertex correction amount, the initial peak point and the length of the effective oscillation wave, combined with the preset sampling frequency, the main frequency of the oscillation wave of each phase winding is calculated.

[0076] Those skilled in the art will understand that, for the sake of convenience and brevity, the specific working process of the device described above can be referred to the corresponding process in the foregoing method embodiments, and will not be repeated here.

[0077] Implementing the above embodiments has the following effects: The technical solution of this invention, after acquiring the oscillation wave signal of any phase winding of a transformer, removes pre-existing noise and environmental noise from the stable attenuation segment of the final oscillation wave signal by setting initial and final sliding windows respectively. This allows for the identification of valid oscillation wave data. Simultaneously, by combining the initial and final sliding windows, the noise baseline of the oscillation wave signal is recursively updated and the variance indicator is calculated. This avoids the problem of not being able to cover the effective segments of different windings due to fixed time windows, and also avoids the problems of large noise interference and susceptibility to waveform misalignment interference leading to misjudgment. This improves the extraction of oscillation wave signals from subsequent phase windings and ensures the accuracy and stability of transformer winding deformation diagnosis. Furthermore, by using the time series index corresponding to the final oscillation wave of the extracted front-end and rear-end noise, the valid oscillation waves of the corresponding remaining phase windings can be extracted, ensuring that the effective segments of the three phases are completely consistent, avoiding misjudgment caused by waveform misalignment interference, facilitating subsequent comparison of correlation coefficients and main frequencies, and improving the accuracy and precision of the judgment.

[0078] Example 6 Accordingly, the present invention also provides a terminal device, comprising: a processor, a memory, and a computer program stored in the memory and configured to be executed by the processor, wherein the processor executes the computer program to implement the transformer winding deformation detection method as described in any of the above embodiments.

[0079] The terminal device in this embodiment includes a processor, a memory, and a computer program and computer instructions stored in the memory and executable on the processor. When the processor executes the computer program, it implements the various steps described in Embodiment 1 above, for example... Figure 1 The steps S101 to S105 are shown. Alternatively, when the processor executes the computer program, it implements the functions of each module / unit in the above device embodiment, such as the front-end noise module 202.

[0080] For example, the computer program can be divided into one or more modules / units, which are stored in the memory and executed by the processor to complete the present invention. The one or more modules / units can be a series of computer program instruction segments capable of performing specific functions, which describe the execution process of the computer program in the terminal device. For example, the front-end noise module 202 is used to set several initial sliding windows and recursively update the noise baseline of the oscillation wave signal of the initial sliding windows; determine the starting point of the oscillation wave based on the updated noise baseline, and extract the intermediate oscillation wave from the oscillation wave signal.

[0081] The terminal device may be a desktop computer, laptop, handheld computer, or cloud server, etc. The terminal device may include, but is not limited to, a processor and memory. Those skilled in the art will understand that the schematic diagram is merely an example of a terminal device and does not constitute a limitation on the terminal device. It may include more or fewer components than illustrated, or combine certain components, or different components. For example, the terminal device may also include input / output devices, network access devices, buses, etc.

[0082] The processor can be a Central Processing Unit (CPU), or other general-purpose processors, digital signal processors (DSPs), application-specific integrated circuits (ASICs), field-programmable gate arrays (FPGAs), or other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components, etc. A general-purpose processor can be a microprocessor or any conventional processor. The processor is the control center of the terminal device, connecting all parts of the terminal device via various interfaces and lines.

[0083] The memory can be used to store the computer programs and / or modules. The processor implements various functions of the terminal device by running or executing the computer programs and / or modules stored in the memory and by calling data stored in the memory. The memory may mainly include a program storage area and a data storage area. The program storage area may store the operating system, applications required for at least one function, etc.; the data storage area may store data created based on the use of the mobile terminal, etc. In addition, the memory may include high-speed random access memory, and may also include non-volatile memory, such as hard disk, RAM, plug-in hard disk, smart media card (SMC), secure digital card (SD card), flash card, at least one disk storage device, flash memory device, or other volatile solid-state storage device.

[0084] Wherein, if the modules / units integrated in the terminal device are implemented as software functional units and sold or used as independent products, they can be stored in a computer-readable storage medium. Based on this understanding, all or part of the processes in the methods of the above embodiments of the present invention can also be implemented by a computer program instructing related hardware. The computer program can be stored in a computer-readable storage medium, and when the computer program is executed by a processor, it can implement the steps of the various method embodiments described above. Wherein, the computer program includes computer program code, which can be in the form of source code, object code, executable file, or some intermediate form, etc. The computer-readable medium can include: any entity or device capable of carrying the computer program code, recording medium, USB flash drive, portable hard drive, magnetic disk, optical disk, computer memory, read-only memory (ROM), random access memory (RAM), electrical carrier signals, telecommunication signals, and software distribution media, etc. It should be noted that the content contained in the computer-readable medium can be appropriately added or removed according to the requirements of legislation and patent practice in the jurisdiction. For example, in some jurisdictions, according to legislation and patent practice, the computer-readable medium does not include electrical carrier signals and telecommunication signals.

[0085] Example 7 Accordingly, the present invention also provides a computer-readable storage medium comprising a stored computer program, wherein, when the computer program is executed, it controls the device containing the computer-readable storage medium to perform the transformer winding deformation detection method as described in any of the above embodiments.

[0086] The specific embodiments described above further illustrate the purpose, technical solution, and beneficial effects of the present invention. It should be understood that the above descriptions are merely specific embodiments of the present invention and are not intended to limit the scope of protection of the present invention. In particular, it should be noted that any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of the present invention should be included within the scope of protection of the present invention for those skilled in the art.

Claims

1. A method for detecting transformer winding deformation, characterized in that, include: Obtain the oscillation wave signal of any phase winding of a transformer; Set up several initial sliding windows; Calculate the slope sequence of the oscillating wave signal, and calculate the average slope amplitude and slope standard deviation for each starting sliding window based on the slope sequence; initialize the noise slope mean and noise slope standard deviation, and set the noise baseline online update conditions and noise baseline recursive update function based on the average slope amplitude and slope standard deviation for each starting sliding window, as well as the noise slope mean and noise slope standard deviation; Based on the online noise baseline update conditions and the noise baseline recursive update function, the noise baseline of the oscillating wave signal in the initial sliding window is recursively updated one by one. If the oscillation wave signal in the current initial sliding window meets the online noise baseline update condition, then the noise baseline of the oscillation wave signal in the current initial sliding window is updated based on the noise baseline recursive update function. If the oscillation wave signal in the current starting sliding window does not meet the online update condition of the noise baseline, then the noise baseline of the oscillation wave signal in the current starting sliding window is updated according to the previous starting sliding window; based on the updated noise baseline, the adaptive threshold corresponding to the current starting sliding window is calculated. The current starting sliding window is determined based on the adaptive threshold and used as the starting point of the oscillation wave. Based on the starting point of the oscillation wave, extract the intermediate oscillation wave from the oscillation wave signal; Set up several end sliding windows, and calculate the variance indicator of the intermediate oscillation wave corresponding to the end sliding window based on the variance baseline of the intermediate oscillation wave. Extract the final oscillation wave from the intermediate oscillation wave based on the variance indicator. Based on the time series index corresponding to the final oscillation wave, extract the corresponding effective oscillation wave from the oscillation wave signals of the remaining phase windings; The correlation coefficient and dominant frequency between the effective oscillation waves of each phase winding are calculated, and the deformation of the transformer winding is detected based on the correlation coefficient and dominant frequency.

2. The method for detecting transformer winding deformation as described in claim 1, characterized in that, The method involves setting several end sliding windows and calculating the variance indicator of the intermediate oscillation wave corresponding to the end sliding window based on the variance baseline of the intermediate oscillation wave. Extracting the final oscillation wave from the intermediate oscillation wave based on the variance indicator specifically includes: Set up several end-sliding windows; The fluctuation amplitude of each end sliding window is calculated based on the intermediate oscillation wave in each end sliding window. Based on the fluctuation amplitude of each end sliding window, the variance baseline of the intermediate oscillation wave is calculated, and a dynamic threshold is set based on the variance baseline. Based on the dynamic threshold, determine the variance indicator of the intermediate oscillation wave corresponding to each end sliding window; Determine the minimum index corresponding to the variance indicator, and use the sliding window at the end of the minimum index as the termination point of the oscillation wave; Based on the termination point of the oscillation wave, the final oscillation wave in the intermediate oscillation wave is extracted.

3. A method for detecting transformer winding deformation as described in any one of claims 1-2, characterized in that, The calculation of the correlation coefficient and dominant frequency between the effective oscillation waves of each phase winding, and the detection of transformer winding deformation based on the correlation coefficient and dominant frequency, specifically includes: Calculate the correlation coefficient between the effective oscillation waves of each phase winding, and compare the correlation coefficient with a preset threshold. When the correlation coefficient is not less than a preset threshold, the transformer winding is normal and without deformation. When the correlation coefficient is less than the preset threshold, the main frequency of the oscillation wave of each phase winding of the transformer and its difference are calculated, and the difference of the main frequency of each phase winding is compared with the preset ratio. When the difference is less than the preset ratio, the transformer winding is normally undeformed. If the difference is not less than the preset ratio, then the transformer winding has a deformation fault.

4. The method for detecting transformer winding deformation as described in claim 3, characterized in that, The calculation of the dominant frequency of the oscillation wave of each phase winding of the transformer specifically includes: Set the raised cosine window function, and calculate the window function based on the raised cosine window function and the effective oscillation wave; Based on the length of the effective oscillation wave, the windowed signal of the raised cosine window function is calculated, and the zero-fill signal is determined in combination with the window function. The zero-fill signal is transformed, and the positive frequency part of the transformed zero-fill signal is extracted to obtain the initial peak point; Parabolic interpolation is performed on the initially selected peak points to obtain interpolation coefficients, and the vertex correction amount is calculated using the interpolation coefficients. Based on the vertex correction amount, the initial peak point and the length of the effective oscillation wave, combined with the preset sampling frequency, the main frequency of the oscillation wave of each phase winding is calculated.

5. A transformer winding deformation detection device, characterized in that, include: The acquisition module is used to acquire the oscillation wave signal of any phase winding of the transformer. The front-end noise module is used to set several initial sliding windows; Calculate the slope sequence of the oscillating wave signal, and calculate the average slope amplitude and slope standard deviation for each starting sliding window based on the slope sequence; initialize the noise slope mean and noise slope standard deviation, and set the noise baseline online update conditions and noise baseline recursive update function based on the average slope amplitude and slope standard deviation for each starting sliding window, as well as the noise slope mean and noise slope standard deviation; Based on the online noise baseline update conditions and the noise baseline recursive update function, the noise baseline of the oscillating wave signal in the initial sliding window is recursively updated one by one. If the oscillation wave signal in the current initial sliding window meets the online noise baseline update condition, then the noise baseline of the oscillation wave signal in the current initial sliding window is updated based on the noise baseline recursive update function. If the oscillation wave signal in the current starting sliding window does not meet the online update condition of the noise baseline, then the noise baseline of the oscillation wave signal in the current starting sliding window is updated according to the previous starting sliding window; based on the updated noise baseline, the adaptive threshold corresponding to the current starting sliding window is calculated. The current starting sliding window is determined based on the adaptive threshold and used as the starting point of the oscillation wave. Based on the starting point of the oscillation wave, extract the intermediate oscillation wave from the oscillation wave signal; The end noise module is used to set several end sliding windows and calculate the variance indicator of the intermediate oscillation wave corresponding to the end sliding window based on the variance baseline of the intermediate oscillation wave. The termination point of the oscillation wave is determined based on the variance indicator, and the final oscillation wave is extracted from the intermediate oscillation waves. The extraction module is used to extract the corresponding effective oscillation wave from the oscillation wave signals of the remaining phase windings according to the time series index corresponding to the final oscillation wave; The detection module is used to calculate the correlation coefficient and main frequency between the effective oscillation waves of each phase winding, and to detect the deformation of the transformer winding based on the correlation coefficient and main frequency.

6. A terminal device, characterized in that, The method includes a processor, a memory, and a computer program stored in the memory and configured to be executed by the processor, wherein the processor, when executing the computer program, implements the transformer winding deformation detection method as described in any one of claims 1 to 4.

7. A computer-readable storage medium, characterized in that, The computer-readable storage medium includes a stored computer program, wherein, when the computer program is executed, it controls the device containing the computer-readable storage medium to perform the transformer winding deformation detection method as described in any one of claims 1 to 4.

Images