On-line monitoring method for dielectric loss of power capacitor bank

Abstract

The invention discloses a power capacitor bank dielectric loss on-line monitoring method, particularly relates to the technical field of power system equipment insulation state monitoring, and is used for solving the problems of insufficient signal synchronization precision and large dielectric loss calculation error under complex electromagnetic interference in the existing method. The method comprises the following steps: acquiring voltage and current signals in real time and dynamically tracking fundamental frequency; intercepting a main period waveform, constructing phase space trajectory analysis complexity, and generating distortion features in combination with symbol sequence transfer entropy; dynamically adjusting filtering parameters based on the distortion characteristics and the Lyapunov exponent gradient; calculating a high-precision phase difference through zero crossing point interpolation correction and sliding window filtering, and finally determining a dielectric loss value; through dynamic frequency tracking, nonlinear feature collaborative analysis and an adaptive filtering strategy, noise interference is effectively suppressed, phase synchronization precision and dielectric loss calculation reliability are improved, and accurate and real-time monitoring support is provided for insulation performance evaluation of the power capacitor bank.

Description

Technical Field

[0001] The present invention relates to the technical field of insulation status monitoring of power system equipment. More specifically, the present invention relates to an on-line monitoring method for dielectric loss of a power capacitor bank. Background Art

[0002] Power capacitor banks are key equipment for reactive power compensation and voltage regulation in power systems. Real-time monitoring of their insulation performance is crucial for ensuring the safe operation of the power grid. Online detection of the tangent value of the dielectric loss angle can directly reflect the internal insulation status of capacitors, usually by synchronously collecting voltage and current signals during capacitor operation and calculating based on the phase difference. In the prior art, dielectric loss monitoring of power capacitor banks is usually carried out in complex electromagnetic environments such as substations and transmission lines, and there are various strong electromagnetic interference sources in the on-site environment, which significantly affects the reliability of signal acquisition.

[0003] In the prior art, it is difficult to ensure the phase synchronization accuracy of voltage and current signals under low signal-to-noise ratio conditions. Since electromagnetic interference in the operating environment of power capacitor banks easily causes signal waveform distortion and noise superposition, traditional synchronization methods are prone to deviation in the process of signal feature extraction and phase alignment, directly affecting the calculation accuracy of the dielectric loss value, and further leading to the risk of misjudgment of the insulation status, making it difficult to meet the requirements of high-precision online monitoring. Summary of the Invention

[0004] In order to overcome the above-mentioned defects of the prior art, an embodiment of the present invention provides an on-line monitoring method for dielectric loss of a power capacitor bank to solve the problems proposed in the above background art.

[0005] To achieve the above object, the present invention provides the following technical solutions: An on-line monitoring method for dielectric loss of a power capacitor bank, comprising the following steps: S1. Real-time obtain the voltage signal and current signal of the power capacitor bank, and record the signal acquisition time series; S2. Dynamically track the fundamental frequencies of the voltage signal and current signal according to the distribution of waveform zero points and extreme points of the voltage signal and current signal; S3. Intercept voltage signal segments and current signal segments of a preset duration according to the fundamental frequency, and extract the main period waveforms corresponding to the fundamental frequency in the voltage signal segments and current signal segments; S4. Construct the phase space trajectory of the main period waveform and analyze its complexity. At the same time, convert the main period waveform into a symbol sequence and calculate the transfer entropy between symbol sequences, and generate signal distortion features according to the complexity and transfer entropy; S5. Filter the voltage signal segments and current signal segments according to the signal distortion features and the maximum Lyapunov exponent gradient of the voltage signal segments and current signal segments; S6. According to the zero-crossing timing difference of the main cycle waveforms of the filtered voltage signal and the current signal, the phase difference between the voltage signal and the current signal is calculated to determine the dielectric loss tangent value of the power capacitor bank.

[0006] In a preferred embodiment, the voltage signal and current signal of the power capacitor bank are acquired in real time, and the signal acquisition time series is recorded, including: The synchronous acquisition of the voltage signal and the current signal is started by a synchronous trigger signal to ensure that the acquisition time deviation of the voltage signal and the current signal is less than a preset time threshold; Aligning sampling timestamps of the voltage signal and the current signal to the same time base, generating a voltage signal sequence and a current signal sequence containing timestamps; Amplitude mutation detection is performed on the voltage signal sequence and the current signal sequence, and abnormal signal segments whose amplitude mutation exceeds a preset amplitude threshold are eliminated.

[0007] In a preferred embodiment, dynamically tracking the fundamental frequency of the voltage signal and the current signal according to the distribution of the waveform zero points and the extreme value points of the voltage signal and the current signal includes: Extracting waveform zero points and extreme value points of voltage signal sequence and current signal sequence within a preset time window; Determine the fundamental wave period of the voltage signal sequence and the current signal sequence according to the time interval between adjacent waveform zero points; The extreme point offset errors of the voltage signal sequence and the current signal sequence are corrected by the cubic spline interpolation method to obtain the corrected fundamental wave period; The corrected inverse of the fundamental wave period is taken as the fundamental wave frequency.

[0008] In a preferred embodiment, according to the fundamental frequency, a voltage signal segment and a current signal segment of a preset duration are intercepted, and a main cycle waveform corresponding to the fundamental frequency in the voltage signal segment and the current signal segment is extracted, including: Determine the preset time length as an integer multiple of the fundamental wave period according to the fundamental wave frequency, and extract the voltage signal segment and the current signal segment of the corresponding time length from the voltage signal sequence and the current signal sequence; In the intercepted voltage signal segment and current signal segment, the starting zero-crossing point of the main cycle waveform is located with the cycle length corresponding to the fundamental frequency as the unit; Starting from the initial zero-crossing point, a voltage signal segment and a current signal segment within a complete fundamental wave cycle are extracted as the main cycle waveform; The extracted main period waveform is processed by mean filtering to eliminate random noise interference.

[0009] In a preferred embodiment, the phase space trajectory of the main periodic waveform is constructed and its complexity is analyzed. At the same time, the main periodic waveform is converted into a symbol sequence and the transfer entropy between the symbol sequences is calculated. The signal distortion features are generated based on the complexity and the transfer entropy, including: The time delay embedding method is used to construct the phase space trajectory of the main periodic waveform; Calculate the recurrence quantification analysis parameters of the phase space trajectory. The recurrence quantification analysis parameters include the recurrence rate and the determinism. The phase space complexity index is generated based on the recurrence rate and the determinism; When converting the main periodic waveform into a symbol sequence, the symbol intervals are divided according to the amplitude distribution of the main periodic waveform, and each symbol interval corresponds to a preset amplitude range; Statistical the transition probabilities of adjacent symbols in the symbol sequence, and calculate the transfer entropy between the symbol sequences according to the transition probabilities; Generate the signal distortion features based on the product of the normalized phase space complexity index and the transfer entropy, or when the phase space complexity index exceeds a preset threshold set based on historical data statistics, directly use the transfer entropy as the signal distortion feature.

[0010] In a preferred embodiment, the delay time of the time delay embedding method is set according to the sampling rate of the main periodic waveform, and the embedding dimension is set according to the spectral characteristics of the main periodic waveform.

[0011] In a preferred embodiment, according to the signal distortion features and the maximum Lyapunov exponent gradient of the voltage signal segment and the current signal segment, filtering processing is performed on the voltage signal segment and the current signal segment, including: Calculate the maximum Lyapunov exponent gradient of the voltage signal segment and the current signal segment. The maximum Lyapunov exponent gradient is calculated by the change rate of the maximum Lyapunov exponent within adjacent time windows; According to the comparison result between the absolute value of the maximum Lyapunov exponent gradient and a preset gradient threshold, adjust the cut-off frequency or the order of the filter; If the signal distortion features exceed the preset distortion threshold and the absolute value of the maximum Lyapunov exponent gradient exceeds the preset gradient threshold, perform filtering on the voltage signal segment and the current signal segment using adaptive mean filtering; If the signal distortion features do not exceed the preset distortion threshold or the absolute value of the maximum Lyapunov exponent gradient does not exceed the preset gradient threshold, perform low-pass filtering with fixed parameters.

[0012] In a preferred embodiment, the filtering window size is dynamically set according to the absolute value of the maximum Lyapunov exponent gradient.

[0013] In a preferred embodiment, calculating the phase difference between the voltage signal and the current signal according to the zero-crossing timing difference of the main periodic waveforms of the filtered voltage signal and the current signal to determine the tangent value of the dielectric loss angle of the power capacitor bank includes: In the main periodic waveforms of the filtered voltage signal and the current signal, locate the zero-crossing points of the voltage waveform and the current waveform respectively, and the zero-crossing points are determined by symbol change detection and interpolation correction; Calculate the time difference between the zero-crossing point of the voltage waveform and the zero-crossing point of the current waveform as the zero-crossing timing difference; Convert the zero-crossing timing difference into a phase difference according to the fundamental frequency; Calculate the tangent value of the dielectric loss angle based on the phase difference, and the tangent value of the dielectric loss angle is the tangent function value of the phase difference.

[0014] Compared with the prior art, the present invention has the following beneficial effects: 1. Through the collaborative processing of dynamic tracking of the fundamental frequency and adaptive filtering, the signal synchronization accuracy in a complex electromagnetic environment is effectively improved; based on the fundamental frequency tracking mechanism of the waveform zero-crossing and extreme point distribution, the frequency fluctuations of the voltage and current signals can be sensed in real time, combined with the two-dimensional distortion feature extraction of the phase space trajectory complexity analysis and the symbol sequence transfer entropy, accurately quantify the interference intensity, and break through the limitations of traditional methods that rely on fixed frequency bands or single parameters; through the dynamic decision logic of the maximum Lyapunov exponent gradient and the distortion feature, adaptively adjust the filtering strategy, suppress high-frequency noise and chaotic interference while retaining the details of the fundamental waveform, significantly improve the signal purity, and provide a highly reliable data basis for subsequent phase difference calculation; 2. Adopt a multi-level error suppression system to form a closed-loop optimization from signal acquisition, feature extraction to filtering processing; systematically eliminate the timing jitter error caused by noise through zero-crossing interpolation correction and sliding window mean filtering, and combine the real-time fundamental frequency to dynamically calibrate the phase difference calculation benchmark to ensure the high-precision output of the dielectric loss value; through non-linear feature fusion and dynamic parameter adaptation, balance the anti-interference ability and calculation efficiency, and avoid the error accumulation problem caused by the time-varying characteristics of the environment in traditional methods, providing more stable and sensitive monitoring support for the insulation state assessment of the power capacitor bank. Description of the Drawings

[0015] Figure 1 It is a flowchart of an on-line monitoring method for the dielectric loss of a power capacitor bank according to the present invention. Detailed Embodiments

[0016] Next, the technical solutions in the embodiments of the present invention will be clearly and completely described in conjunction with the accompanying drawings in the embodiments of the present invention. Obviously, the described embodiments are only a part of the embodiments of the present invention, rather than all of the embodiments. All other embodiments obtained by those of ordinary skill in the art based on the embodiments of the present invention without creative efforts shall fall within the protection scope of the present invention.

[0017] Embodiment: Figure 1 A method for on-line monitoring of dielectric loss of a power capacitor bank according to the present invention is provided, including the following steps: S1. Obtain the voltage signal and current signal of the power capacitor bank in real time, and record the signal acquisition time series; S2. Dynamically track the fundamental frequencies of the voltage signal and current signal according to the distribution of waveform zero points and extreme points of the voltage signal and current signal; S3. Intercept the voltage signal segment and current signal segment of a preset duration according to the fundamental frequency, and extract the main period waveforms corresponding to the fundamental frequency in the voltage signal segment and current signal segment; S4. Construct the phase space trajectory of the main period waveform and analyze its complexity. At the same time, convert the main period waveform into a symbol sequence and calculate the transfer entropy between symbol sequences, and generate signal distortion features according to the complexity and transfer entropy; S5. Filter the voltage signal segment and current signal segment according to the signal distortion features and the maximum Lyapunov exponent gradient of the voltage signal segment and current signal segment; S6. Calculate the phase difference between the voltage signal and current signal according to the zero-crossing time difference of the main period waveforms of the filtered voltage signal and current signal to determine the tangent value of the dielectric loss angle of the power capacitor bank.

[0018] S1. Obtain the voltage signal and current signal of the power capacitor bank in real time, and record the signal acquisition time series, which can be specifically implemented as: Start the synchronous acquisition of the voltage signal and current signal through a synchronous trigger signal to ensure that the acquisition time deviation between the voltage signal and current signal is less than a preset time threshold. The synchronous trigger signal is generated by receiving the synchronous pulse signal of the power system through an external trigger device, or generated by a periodic square wave signal generated by an internal high-precision clock. The preset time threshold is set to one-thousandth of the power frequency period according to the power frequency period of the power capacitor bank. For example, when the power frequency is 50 Hz, the power frequency period is 20 ms, and the preset time threshold is set to 20 μs. This threshold ensures that the phase difference calculation error is less than 0.01° through sampling accuracy redundancy design. The specific logic is: when the sampling interval is less than one-thousandth of the fundamental period, the zero-crossing positioning time error ≤ 0.02 μs, and the corresponding phase difference error ≤ 0.00036° (far lower than 0.01°).

[0019] Align the sampling timestamps of the voltage signal and the current signal to the same time reference to generate a voltage signal sequence and a current signal sequence containing timestamps. The time reference is the Coordinated Universal Time output by the GPS timing module, or the internal high-precision clock is locked to the same oscillator through a phase-locked loop circuit. The specific method for timestamp alignment is as follows: If the deviation between the sampling timestamps of the voltage signal and the current signal is within the preset time threshold, calculate the instantaneous value at the aligned time point by the linear interpolation method. For example, the sampling value of the voltage signal at time point t1 is V(t1), and the sampling value of the current signal at time point t2 is I(t2). When the time deviation |t1 - t2| ≤ 20 μs, taking the intermediate time point t = (t1 + t2) / 2 as the reference, calculate the aligned value V(t) = V(t1) + (V(t1 + Δt) - V(t1)) × (t - t1) / Δt, where Δt is the sampling interval.

[0020] Perform amplitude mutation detection on the voltage signal sequence and the current signal sequence, and eliminate abnormal signal segments with amplitude mutations exceeding the preset amplitude threshold. The amplitude mutation detection is achieved by calculating the amplitude change rate of adjacent sampling points. If the change rate exceeds the preset amplitude threshold, it is determined as a mutation point. The initial value of the preset amplitude threshold is set according to 10% of the rated voltage or current value of the power capacitor bank. For example, when the rated voltage is 10 kV, the initial threshold is set to 1 kV, and the amplitude standard deviation σ within the last 100 power frequency cycles is statistically calculated through a sliding window and dynamically adjusted to 3σ. The elimination range of the abnormal signal segment is centered on the mutation point and extended half a power frequency cycle forward and backward. For example, for a power frequency of 50 Hz, it corresponds to an extension of 10 ms.

[0021] The synchronization accuracy between the synchronous trigger signal and the time reference is verified by measuring the delay of the standard signal in a non-interference environment. If the delay exceeds the preset time threshold, adjust the clock synchronization circuit or replace the trigger source. The error control of the linear interpolation method is achieved by limiting the timestamp deviation not to exceed half of the sampling interval. When it exceeds, discard the data and re-collect it. The dynamic adjustment of the amplitude mutation threshold is performed every 5 minutes to ensure adaptation to the real-time operating conditions.

[0022] The external trigger device uses an opto-isolated interface to avoid ground potential difference interference. The internal high-precision clock maintains the frequency stability through temperature control, and the temperature fluctuation is controlled within ±0.1 °C. When the synchronization error of the second pulse signal of the GPS timing module exceeds 1 μs, an alarm is triggered and switched to the internal clock. The abnormal segments detected by the amplitude mutation are marked as invalid data and skipped in subsequent processing.

[0023] S2. According to the distribution of the waveform zero points and extreme points of the voltage signal and the current signal, dynamically track the fundamental frequencies of the voltage signal and the current signal, which can be specifically implemented as: Extract the waveform zero points and extreme points of the voltage signal sequence and the current signal sequence within a preset time window. The length of the preset time window is set to an integer multiple of the power frequency period of the power capacitor bank. For example, when the power frequency is 50 Hz, the power frequency period is 20 ms, and the preset time window is set to 5 power frequency periods, that is, 100 ms. When extracting zero points, traverse the sampling points of the voltage signal sequence and the current signal sequence. If the signs of two adjacent sampling points are opposite (such as changing from positive to negative or from negative to positive), it is determined as a zero point; when extracting extreme points, traverse the sampling point sequence. If the amplitude of a certain sampling point is greater than the amplitudes of the previous point and the next point, it is determined as a maximum point, otherwise it is a minimum point.

[0024] Determine the fundamental periods of the voltage signal sequence and the current signal sequence according to the time intervals between adjacent waveform zero points. The calculation method of the fundamental period is as follows: count the time intervals of all adjacent zero points within the preset time window, and after excluding the abnormal intervals that deviate from the average value by ±20%, take the arithmetic average of the remaining intervals as the current fundamental period. For example, within a 100 ms time window, 6 zero points are detected, and the adjacent intervals are 19.8 ms, 20.1 ms, 20.3 ms, 19.5 ms, 20.5 ms respectively. After excluding 19.5 ms (lower than 80% of the average value of 20 ms), the average value of the remaining intervals is 20.1 ms, that is, the fundamental period is 20.1 ms.

[0025] Correct the extreme point offset error of the voltage signal sequence and the current signal sequence by the cubic spline interpolation method to obtain the corrected fundamental period. The correction method of the extreme point offset error includes: taking the extreme point as the center, selecting a total of five points, namely two sampling points before and after it, as the interpolation nodes, reconstructing the waveform curve of this interval by the cubic spline interpolation method, and repositioning the exact position of the extreme point. For example, the original position of a maximum point is the sampling point n. Through interpolation, it is found that the actual extreme value is between the nth point and the (n + 1)th point, and the offset is 0.3 sampling intervals. Then the corrected position of the extreme point is n + 0.3. The corrected fundamental period is recalculated by updating the zero point interval. For example, the corrected zero point interval is adjusted from 20.1 ms to 19.9 ms.

[0026] Take the reciprocal of the corrected fundamental period as the real-time fundamental frequency. The calculation formula of the fundamental frequency is fundamental frequency = 1 / fundamental period. For example, when the fundamental period is 19.9 ms, the fundamental frequency is 1 / 0.0199 ≈ 50.25 Hz. The real-time fundamental frequency is updated once every preset time window. For example, the frequency value is output every 100 ms to ensure dynamic tracking of the power frequency fluctuation of the power capacitor bank.

[0027] The integer multiple setting of the preset time window should cover at least three power frequency cycles to avoid zero - point detection errors caused by short - term interference. The rejection threshold for abnormal intervals (±20%) is determined based on the statistics of historical operation data. For example, when the capacitor bank is operating normally, the fundamental wave period fluctuation range does not exceed ±2%. Therefore, a ±20% threshold is set to exclude the influence of sudden interference. The number of nodes of the cubic spline interpolation method is set to five points to ensure that the interpolation curve is smooth and the calculation amount is controllable.

[0028] The boundary conditions of the cubic spline interpolation method are set as natural spline conditions, that is, the second derivative is zero at the endpoints, to ensure the smoothness of the interpolation curve at the boundaries. The selection range of the interpolation interval is centered on the extreme points and extends two sampling points forward and backward. For example, when the sampling rate is 10 kHz, each sampling interval is 0.1 ms, and the interpolation interval covers a range of 0.5 ms to ensure that the waveform characteristics near the extreme points are included.

[0029] The update logic of the dynamic fundamental wave frequency includes: if the deviation of the fundamental wave frequencies calculated in two adjacent preset time windows exceeds ±0.5 Hz, a frequency jump alarm is triggered, and the sliding average method is used to smooth the output frequency value. For example, the current window frequency is 50.25 Hz, the previous window is 49.8 Hz, and the deviation is 0.45 Hz, which does not exceed the threshold; if the frequency in the next window suddenly changes to 51.0 Hz, an alarm is triggered and the smoothed value (50.25 + 51.0) / 2 = 50.625 Hz is output.

[0030] The extraction operations of zero points and extreme points are realized through a sliding window in real - time signal processing. Every time a new sampling point is received, the window data is updated, and the zero - point and extreme - point detections are re - executed. The update step of the sliding window is one sampling interval. For example, when the sampling rate is 10 kHz, the window data is updated every 0.1 ms to ensure real - time performance.

[0031] S3. Intercept the voltage signal segment and current signal segment of the preset duration according to the fundamental wave frequency, and extract the main - period waveforms corresponding to the fundamental wave frequency in the voltage signal segment and current signal segment. It can be specifically implemented as: Determine that the preset duration is an integer multiple of the fundamental wave period according to the fundamental wave frequency, and intercept the corresponding voltage signal segment and current signal segment from the voltage signal sequence and the current signal sequence. The integer multiple of the fundamental wave period is set to include at least five complete fundamental wave periods. For example, when the fundamental wave frequency is 50 Hz, the fundamental wave period is 20 ms, and the preset duration is set to 5 fundamental wave periods, that is, 100 ms. When intercepting the voltage signal segment and the current signal segment, intercept the data of the preset duration forward based on the current moment to ensure that the intercepted segment contains a complete periodic waveform. For example, if the current moment is t, the intercepted time range is the voltage signal sequence and the current signal sequence within [t - 100 ms, t]. The number of integer multiples of the fundamental wave period is dynamically adjusted according to the operating stability of the power capacitor bank. For example, when the frequency fluctuation exceeds ±0.5 Hz, increase the integer multiple to 10 periods to improve the anti-interference ability of the intercepted segment.

[0032] In the intercepted voltage signal segment and current signal segment, locate the starting zero-crossing point of the main periodic waveform with the period length corresponding to the fundamental wave frequency as the unit. The method for locating the starting zero-crossing point is as follows: Starting from the starting point of the intercepted voltage signal segment and current signal segment, slide the search window with the fundamental wave period as the step length, and detect the first zero-crossing point within each window as the starting point of the main periodic waveform. For example, when the fundamental wave period is 20 ms, each 20 ms is a search window, and the zero-crossing point where the first voltage signal changes from negative to positive within the window is the starting zero-crossing point. If no zero-crossing point is detected within the search window, expand the search range to the adjacent window until successful positioning. The specific operation of expanding the search range is: Expand the window length to 1.5 times the fundamental wave period, for example, 30 ms, and re-detect the zero-crossing point within the expanded window. If still unable to locate, skip this window and continue the search from the next window.

[0033] Starting from the starting zero-crossing point, extract the voltage signal segment and current signal segment within a complete fundamental wave period as the main periodic waveform. The method for extracting the complete fundamental wave period is: Taking the starting zero-crossing point as the starting point, intercept the data of a fundamental wave period length backward. For example, when the fundamental wave period is 20 ms, intercept the voltage signal segment and current signal segment from the starting point to 20 ms after the starting point. If the end of the intercepted segment does not contain a complete period, adjust the starting point forward to ensure the period integrity. For example, the starting zero-crossing point is at t0, and the intercepted segment is [t0, t0 + 20 ms]. If t0 + 20 ms exceeds the signal segment range, adjust the starting point to t0 - Δt to make the intercepted segment complete. The calculation method of Δt for adjusting the starting point is: Δt = fundamental wave period - (t_max - t0), where t_max is the maximum time stamp of the signal segment. For example, when t_max is t0 + 15 ms, Δt = 20 - 15 = 5 ms, and the adjusted intercepted segment is [t0 - 5 ms, t0 + 15 ms].

[0034] Perform mean filtering on the extracted main periodic waveform to eliminate random noise interference. The specific method of mean filtering is as follows: for each sampling point in the main periodic waveform, calculate the arithmetic mean of it and the two adjacent sampling points before and after, and replace the original sampling value. For example, if the value of a sampling point n is V(n), then the filtered value V’(n)=[V(n-1)+V(n)+V(n+1)] / 3. The filtering operation is performed separately on the voltage signal segment and the current signal segment to ensure noise suppression while retaining the fundamental waveform characteristics. The boundary processing method of mean filtering is as follows: for the starting point and the ending point of the signal segment, the mirror extension method is used to generate virtual sampling points. For example, the value of the virtual previous point n=-1 at the starting point n=0 is V(-1)=V(0), and the value of the virtual next point n=N+1 at the ending point n=N is V(N+1)=V(N).

[0035] The starting point adjustment logic of the intercepted segment includes: if the end of the intercepted segment exceeds the signal segment range, move the starting point forward by the time length of the insufficient part. For example, if the intercepted segment requires 20ms but the remaining duration is only 15ms, then the starting point is moved forward by 5ms to ensure the interception of a complete cycle. The trigger condition for dynamically adjusting the number of integer multiple cycles is: if the standard deviation of the fundamental frequency exceeds 0.3Hz within three consecutive preset time windows, the number of integer multiples is automatically increased. For example, if the standard deviation of the fundamental frequency within three 100ms windows is 0.4Hz, then the integer multiple increases from 5 to 10.

[0036] The extraction and filtering operations of the main periodic waveform are implemented through a circular buffer in real-time processing. Each time a new sampling point is received, the buffer data is updated, and the interception, positioning, and filtering are re-executed. The length of the buffer is twice the preset duration. For example, when the preset duration is 100ms, the buffer length is 200ms to ensure that there is always enough data for interception. The update logic of the circular buffer is: when a new sampling point arrives, remove the oldest data point from the buffer and add the new point to the end of the buffer to maintain a constant buffer length. For example, the original buffer data is [t-200ms, t], and after adding a new sampling point, it is updated to [t-199.9ms, t+0.1ms].

[0037] The number of sampling points for mean filtering is set to three points to balance the noise suppression effect and the computational complexity. The specific operation of the mirror extension method is: copy the value of the first sampling point as the virtual previous point at the starting end of the signal segment, and copy the value of the last sampling point as the virtual next point at the ending end. For example, if the signal segment is [V(0), V(1),..., V(N)], after extension, it is [V(0), V(0), V(1),..., V(N), V(N)]. The upper limit of the dynamically adjusted number of integer multiple cycles is set to 20 cycles to avoid a decrease in real-time performance due to excessive extension of the intercepted segment.

[0038] S4. Construct the phase space trajectory of the main periodic waveform and analyze its complexity. At the same time, convert the main periodic waveform into a symbol sequence and calculate the transfer entropy between symbol sequences. Generate signal distortion features based on complexity and transfer entropy, which can be specifically implemented as follows: Use the time-delay embedding method to construct the phase space trajectory of the main periodic waveform; the delay time of the time-delay embedding method is set according to the sampling rate of the main periodic waveform, and the embedding dimension is set according to the spectral characteristics of the main periodic waveform. The method for setting the delay time is: take one-fourth of the sampling interval of the main periodic waveform as the delay time. For example, when the sampling rate is 10 kHz, the sampling interval is 0.1 ms, and the delay time is set to 0.025 ms. The method for setting the embedding dimension is: perform a fast Fourier transform on the main periodic waveform, extract the spectral amplitude corresponding to the fundamental frequency. If the fundamental amplitude accounts for more than 90% of the total spectral energy, the embedding dimension is set to 3; otherwise, it is set to 5. For example, if the fundamental amplitude of a certain main periodic waveform is 85% of the total energy, the embedding dimension is set to 5 to capture high-frequency components. The process of constructing the phase space trajectory by the time-delay embedding method is: combine each sampling point of the main periodic waveform with the point at the subsequent delay time interval to form a trajectory point in a high-dimensional space. For example, when the embedding dimension is 3, the trajectory point is composed of three consecutive delay points: V(t), V(t+τ), V(t+2τ), where τ is the delay time.

[0039] Calculate the recurrence quantification analysis parameters of the phase space trajectory. The recurrence quantification analysis parameters include recurrence rate and determinism. Generate a phase space complexity index based on the recurrence rate and determinism. The method for calculating the recurrence rate is: construct a recurrence plot of the phase space trajectory and count the proportion of all recurrence points in the recurrence plot as the recurrence rate. The method for constructing the recurrence plot is: calculate the Euclidean distance between trajectory points. If the distance is less than a preset radius (such as 10% of the average distance between trajectory points), it is marked as a recurrence point. For example, if the number of trajectory points is 1000 and the average distance is 0.5V, the preset radius is 0.05V. Count the number of recurrence points N_rec that meet the conditions, and the recurrence rate = N_rec / 1000². The method for calculating determinism is: count the proportion of the diagonal structure in the recurrence plot. The diagonal structure is defined as a line segment formed by two consecutive recurrence points and with a length ≥ 2. For example, if there are 200 valid diagonals in the recurrence plot and the total number of recurrence points is 500, then determinism = 200 / 500 = 0.4. The phase space complexity index is generated by the ratio of the recurrence rate to determinism. For example, when the recurrence rate is 0.6 and the determinism is 0.4, the complexity index is 0.6 / 0.4 = 1.5. If the determinism is zero, the complexity index is set to a preset maximum value (such as 10) to avoid division-by-zero errors.

[0040] When converting the main period waveform into a symbol sequence, the symbol intervals are divided according to the amplitude distribution of the main period waveform, and each symbol interval corresponds to a preset amplitude range. The method for dividing the symbol intervals is as follows: calculate the mean and standard deviation of the amplitudes of the main period waveform, and divide the amplitude range into three intervals: the low amplitude interval is below the mean - 1 times the standard deviation, the medium amplitude interval is the mean ± 1 times the standard deviation, and the high amplitude interval is above the mean + 1 times the standard deviation, corresponding to the symbols "0", "1", and "2" respectively. For example, if the mean amplitude of the main period waveform is 5V and the standard deviation is 1V, the symbol intervals are: low amplitude (<4V), medium amplitude (4V - 6V), high amplitude (>6V). If the waveform amplitude exceeds the historical statistical range, the interval boundaries are dynamically adjusted to cover the current amplitude. The dynamic adjustment method is: recalculate the mean and standard deviation of the current main period waveform, and update the symbol intervals based on the new parameters. For example, if a main period waveform has an amplitude of 8V (the original high amplitude interval was >6V), recalculate the mean to be 5.5V and the standard deviation to be 1.2V, and update the high amplitude interval to >6.7V.

[0041] Statistical symbol sequence adjacent symbol transfer probability, according to the transfer probability to calculate the transfer entropy between symbol sequences. The statistical method of the transfer probability is: traverse the symbol sequence, count the number of occurrences of each pair of adjacent symbols (such as "0→1", "1→2"), and divide by the total number of transfers to get the probability. For example, if the length of the symbol sequence is 1000 and the total number of symbol transfers is 999, and "0→1" appears 300 times, then the probability is 300 / 999≈0.3. The calculation formula for the transfer entropy is: take the logarithm of the transfer probability of each pair of adjacent symbols and sum them with weights, and the formula is the negative weighted sum value. For example, if the probability of "0→1" in the symbol sequence is 0.3 and the probability of "1→2" is 0.2, then the transfer entropy is -(0.3×log20.3 + 0.2×log20.2)≈0.521 + 0.464 = 0.985. If a pair of symbols does not appear (such as "2→0"), its probability is set to the minimum value (such as 1e - 6) to avoid logarithmic calculation errors.

[0042] Generate a signal distortion feature based on the product of the normalized phase space complexity index and transfer entropy, or directly use the transfer entropy as the signal distortion feature when the phase space complexity index exceeds a preset threshold set based on historical data statistics. The normalization method is as follows: Scale the phase space complexity index and transfer entropy to the range of 0 - 1 respectively. The normalization formula for the complexity index is: Normalized value = Original value / Historical maximum value. For example, if the historical maximum complexity is 10 and the current value is 8, then the normalized value is 0.8. The normalization formula for transfer entropy is: Normalized value = Original value / Theoretical maximum value. The theoretical maximum entropy is determined by the number of symbols (e.g., when there are 3 symbols, the theoretical maximum entropy is log23 ≈ 1.585). When the current entropy value is 1.0, the normalized value ≈ 0.63. The method for setting the preset threshold is: Statistically calculate the mean and standard deviation of the phase space complexity index in the historical normal operation data of the power capacitor bank, and set the threshold to the mean plus twice the standard deviation. For example, if the historical mean is 1.2 and the standard deviation is 0.3, then the threshold is 1.2 + 2×0.3 = 1.8. When the real-time complexity index exceeds 1.8, directly use the transfer entropy as the distortion feature; otherwise, use the normalized product. For example, the normalized complexity is 0.7 (original value 7), the transfer entropy is 0.4 (original value 2), and the product is 0.7×0.4 = 0.28; if the threshold is 0.8 (corresponding to the original value 8), then directly use the transfer entropy 0.4.

[0043] The method for adjusting the preset radius of the recurrence plot is: If the recurrence rate is lower than 5% or higher than 95%, then automatically adjust the radius to 5% or 20% of the average distance of the current trajectory points to balance the recurrence point density. For example, when the initial radius is 0.05V and the recurrence rate is 3%, then adjust the radius to 0.025V (5% of the average distance). The triggering condition for dynamically adjusting the symbol interval is: When the amplitudes of three consecutive main cycle waveforms exceed the current interval range, trigger the recalculation of the mean and standard deviation. For example, if the high amplitudes of three consecutive waveforms are 7V, 7.2V, 7.5V (the original high amplitude interval is >6V), then trigger the interval update.

[0044] The maximum value of the phase space complexity index is set to twice the maximum ratio observed in the historical data. For example, if the historical maximum ratio is 5, then the preset maximum value is 10. The processing logic for the minimum value of transfer entropy is: If all transfer probabilities are at the minimum value (such as all being 1e - 6), then determine that the signal is completely random noise and set the transfer entropy to 0. The logical execution order of the normalized product and threshold judgment is: First check whether the complexity index exceeds the threshold. If it does not exceed the threshold, then calculate the product. For example, the normalized complexity is 0.7 (original value 7), the transfer entropy is 0.4 (original value 2), and the product is 0.7×0.4 = 0.28; if the threshold is 0.8 (corresponding to the original value 8), then directly use the transfer entropy 0.4.

[0045] The calculation of the diagonal structure length in recursive quantification analysis includes: counting the length of the longest sequence of consecutive recurrence points. For example, if a diagonal contains 5 consecutive recurrence points, the length is 5. During the division of the symbol interval, if the amplitude distribution of the main periodic waveform seriously deviates from the normal distribution (such as skewness > 2), the equal-interval method is used to replace the standard deviation division. For example, the amplitude range is 0 - 10V, which is equally divided into three intervals: 0 - 3.3V, 3.3 - 6.6V, and 6.6 - 10V. The dynamically adjusted symbol interval needs to cover at least 95% of the amplitude points of the current main periodic waveform; otherwise, the interval boundary continues to be expanded.

[0046] In this step S4, the signal distortion feature is generated through the collaborative analysis of the phase space trajectory and the symbol sequence transfer entropy. The chaoticity of the waveform can be quantified through the recursive quantification parameters (recurrence rate and determinism) of the phase space trajectory, and the symbol sequence transfer entropy reflects the degree of temporal causal break. The two capture the interference features from the perspectives of dynamic systems and information theory respectively; compared with the prior art that only relies on frequency domain energy or fixed thresholds, through dynamic parameters (delay time, embedding dimension) and adaptive symbol interval division, the problem of insufficient sensitivity of traditional methods under non-stationary signals and sudden interferences is solved; the fusion analysis of non-linear features and information entropy breaks through the limitations of traditional linear filtering or single-parameter evaluation, significantly improving the distortion detection accuracy under low signal-to-noise ratio conditions and meeting the high-precision dielectric loss monitoring requirements.

[0047] S5. Filter the voltage signal segment and the current signal segment according to the signal distortion feature and the maximum Lyapunov exponent gradient of the voltage signal segment and the current signal segment, which can be specifically implemented as: Calculate the maximum Lyapunov exponent gradient of the voltage signal segment and the current signal segment. The maximum Lyapunov exponent gradient is calculated through the change rate of the maximum Lyapunov exponent within adjacent time windows. The calculation method of the maximum Lyapunov exponent is: construct the phase space trajectory for the voltage signal segment and the current signal segment respectively, and calculate the maximum Lyapunov exponent through the Wolf algorithm, which characterizes the sensitivity of the signal to the initial conditions. The parameter settings of the Wolf algorithm include: the initial perturbation radius is 1% of the average distance between phase space trajectory points, and the evolution step size is 1 / 10 of the fundamental wave period. For example, when the fundamental wave period is 20ms, the evolution step size is set to 2ms; when the average distance between trajectory points is 0.5V, the initial perturbation radius is set to 0.005V. The specific operation of gradient calculation is: subtract the maximum Lyapunov exponent value of the previous time window from that of the current time window, and divide by the time window interval length. For example, the time window interval is 100ms, the current window exponent is 0.5 / s, and the previous time window exponent is 0.3 / s, then the gradient is (0.5 - 0.3) / 0.1 = 2.0 / s.

[0048] Adjust the cut-off frequency or order of the filter according to the comparison result between the absolute value of the maximum Lyapunov exponent gradient and the preset gradient threshold. The method for setting the preset gradient threshold is as follows: statistically calculate the standard deviation σ of the maximum Lyapunov exponent gradient in the historical normal operation data of the power capacitor bank, and set the threshold to 3σ. For example, if the historical standard deviation is 0.5 / s, then the threshold is 1.5 / s. The dynamic update period of the gradient threshold is 24 hours to ensure adaptation to equipment aging or environmental changes. If the absolute value of the gradient exceeds the threshold, increase the cut-off frequency of the filter to twice the fundamental frequency to suppress high-frequency chaotic noise; for example, when the fundamental frequency is 50Hz, the cut-off frequency is set to 100Hz. If the absolute value of the gradient does not exceed the threshold, keep the cut-off frequency at 1.2 times the fundamental frequency (such as 60Hz).

[0049] If the signal distortion feature exceeds the preset distortion threshold and the absolute value of the maximum Lyapunov exponent gradient exceeds the preset gradient threshold, perform filtering on the voltage signal segment and current signal segment using adaptive mean filtering, and the size of the filtering window is dynamically set according to the absolute value of the maximum Lyapunov exponent gradient. The rule for setting the size of the adaptive mean filtering window is as follows: for every 0.5 / s increase in the absolute value of the gradient, the window size decreases by 1 sampling point. For example, when the absolute value of the gradient is 2.0 / s, the window size is 3 sampling points; when the gradient is 1.0 / s, the window size is 5 sampling points. The lower limit of the filtering window size is set to 3 points, and the upper limit is set to 9 points to avoid insufficient smoothing due to too small a window or loss of details due to too large a window. The method for setting the preset distortion threshold is as follows: determine it according to the 90th percentile of the signal distortion feature in the historical data. For example, if the historical 90th percentile is 0.8, then the threshold is 0.8. If the signal distortion feature exceeds 0.8 and the gradient exceeds 1.5 / s, start the adaptive filtering.

[0050] If the signal distortion feature does not exceed the preset distortion threshold or the absolute value of the maximum Lyapunov exponent gradient does not exceed the preset gradient threshold, perform low-pass filtering with fixed parameters. The cut-off frequency of the low-pass filtering with fixed parameters is set to 1.2 times the fundamental frequency, and the filter order is set to 4th order. For example, when the fundamental frequency is 50Hz, the cut-off frequency is 60Hz, and a 4th-order Butterworth filter is used for filtering. The basis for choosing the order is that the 4th-order Butterworth filter has an attenuation rate of -50dB / octave at the cut-off frequency, which can effectively suppress high-frequency noise. The coefficients of the low-pass filter are calculated by the bilinear transformation method to ensure the stability of the digital filter.

[0051] During the calculation of the maximum Lyapunov exponent, if the correlation dimension of the phase space trajectory is less than 2, the signal is determined to be strongly noise - interfered, the gradient calculation is skipped, and the fixed - parameter filtering is directly enabled. The calculation method of the correlation dimension is: calculating the slope of the logarithmic distance distribution of trajectory points through the Grassberger - Procaccia algorithm. For example, if the slope is less than 1.5, it is determined that the correlation dimension is insufficient. The boundary - handling method of the adaptive mean filtering is: for the starting or ending position of the signal segment, the mirror - extension method is used to generate virtual sampling points. For example, the virtual previous point of the starting point n = 0 takes the value of n = 1, and the virtual subsequent point of the ending point n = N takes the value of n = N - 1.

[0052] The dynamic update logic of the gradient threshold includes: re - collecting historical data every 24 hours and calculating the standard deviation. If the deviation between the new standard deviation and the original threshold exceeds 20%, the threshold is updated. For example, the original threshold corresponding to a standard deviation of 0.5 / s at 1.5 / s, if the new standard deviation is 0.6 / s, the new threshold is 1.8 / s. When the fundamental frequency fluctuation of the fixed - parameter low - pass filter exceeds ±0.5 Hz, the cut - off frequency is dynamically adjusted to 1.2 times the current fundamental frequency. For example, when the fundamental frequency becomes 51 Hz, the cut - off frequency is adjusted to 61.2 Hz. The threshold - triggering logic of the signal distortion feature is: if the distortion features of three consecutive time windows all exceed the threshold, it is determined as continuous interference and the adaptive filtering mode is locked until the distortion feature is lower than the threshold.

[0053] Step S5 dynamically adjusts the filtering strategy by jointly analyzing the signal distortion feature and the maximum Lyapunov exponent gradient. The signal distortion feature reflects the waveform distortion degree, while the maximum Lyapunov exponent gradient quantifies the change rate of the system's chaos. The two respectively characterize the interference intensity from the perspectives of static distortion and dynamic chaos. Compared with the prior art that only relies on a single parameter or a fixed filtering mode, through double - threshold judgment (distortion feature and gradient threshold) and dynamic adjustment of the adaptive filtering window, the problems of insufficient sensitivity and high misjudgment rate in the scenario of mixed sudden interference and steady - state noise of the traditional method are solved. The collaborative decision - making mechanism of non - linear features and dynamic parameters breaks through the fixed - parameter limitation of traditional filters, accurately distinguishes steady - state noise and transient interference, can effectively suppress high - frequency chaotic noise and retain the details of the fundamental - wave waveform, ensuring the accuracy of subsequent phase - difference calculation and meeting the dielectric loss monitoring requirements in complex electromagnetic environments.

[0054] S6. Calculate the phase - difference value between the voltage signal and the current signal based on the zero - crossing timing difference of the main - period waveforms of the filtered voltage signal and current signal to determine the tangent value of the dielectric loss angle of the power capacitor bank, which can be specifically implemented as: In the main period waveforms of the filtered voltage signal and current signal, locate the zero-crossing points of the voltage waveform and current waveform respectively. The zero-crossing points are determined through sign change detection and interpolation correction. The specific method of sign change detection is as follows: Traverse the sampling point sequences of the filtered voltage signal and current signal. If the sign (positive / negative) of a certain sampling point is different from the previous point, it is marked as a candidate zero-crossing point. The specific method of interpolation correction is as follows: Select the amplitudes of the previous point and the current point near the candidate zero-crossing point, and calculate the exact zero-crossing point position through linear interpolation. For example, if the value of the voltage signal at sampling point n is +0.1V and the value at point n + 1 is -0.2V, then the zero-crossing point position is n + 0.1 / (0.1 + 0.2) = n + 0.33 sampling intervals. The selection range of the interpolation interval is one sampling point before and after the candidate point. For example, when the sampling interval is 0.1ms, the interpolation interval is [t - 0.1ms, t + 0.1ms] to ensure the calculation accuracy.

[0055] Calculate the time difference between the zero-crossing point of the voltage waveform and the zero-crossing point of the current waveform as the zero-crossing point timing difference. The calculation method of the time difference is as follows: Take the absolute value of the difference between the exact zero-crossing point timestamps of the voltage waveform and the current waveform. For example, if the voltage zero-crossing point is at time t_v = 100.5ms and the current zero-crossing point is at t_i = 100.8ms, then the time difference is |100.5 - 100.8| = 0.3ms. If there are multiple zero-crossing points in the same period, take the time difference of the first zero-crossing point pair. The screening logic of the time difference includes: If the time interval between two adjacent zero-crossing points is less than 10% of the fundamental wave period, it is determined as noise interference and excluded. For example, when the fundamental wave period is 20ms, candidate points with an interval less than 2ms are regarded as invalid.

[0056] Convert the zero-crossing point timing difference to a phase difference according to the fundamental wave frequency. The phase difference is the ratio of the time difference to the fundamental wave period length multiplied by 360 degrees. The fundamental wave period length is calculated through the real-time fundamental wave frequency dynamically tracked in step S2. For example, when the fundamental wave frequency is 50Hz, the fundamental wave period is 20ms. If the time difference is 0.3ms, then the phase difference is (0.3 / 20)×360° = 5.4°. The calculation result of the phase difference needs to retain at least two decimal places to improve the accuracy. The calculation result of the phase difference needs to be processed by moving window mean filtering, and the window size is 5 periods to suppress the jumps caused by instantaneous interference. For example, if the phase differences of 5 consecutive periods are 5.4°, 5.6°, 5.2°, 5.5°, 5.3°, then the filtered value is (5.4 + 5.6 + 5.2 + 5.5 + 5.3) / 5 = 5.4°.

[0057] Calculate the tangent value of the dielectric loss angle based on the phase difference. The tangent value of the dielectric loss angle is the tangent function value of the phase difference. The calculation of the tangent function value is achieved through the look-up table method or the numerical approximation method. For example, when the phase difference is 5.4°, the tangent value of the dielectric loss angle is tan(5.4°)≈0.0945. The calculation result is stored in the monitoring system database and triggers the real-time alarm logic: if the tangent value exceeds the preset alarm threshold (such as 0.1), an insulation anomaly alarm signal is generated. The setting method of the alarm threshold is: comprehensively determined according to the insulation level and historical operation data of the power capacitor bank. For example, the threshold of the capacitor bank with a rated voltage of 10kV is set to 0.1, and that of 35kV is set to 0.08. After the alarm signal is generated, the historical data comparison is automatically called. If the tangent values exceed the threshold for three consecutive cycles, the alarm is confirmed and the fault type is recorded.

[0058] The candidate zero-crossing screening logic for symbol change detection includes: if no valid zero-crossing is detected within the same cycle, trace back to the nearest valid zero-crossing. For example, if there is no zero-crossing in the current cycle voltage, the time stamp of the last zero-crossing in the previous cycle is used for calculation. For the calculation of the linear interpolation coefficient for interpolation correction, the denominator needs to be restricted from being zero. If the denominator is zero, the candidate point is directly taken as the zero-crossing. For example, if the amplitudes of two adjacent points are the same (such as +0.1V and +0.1V), the interpolation is skipped and the candidate point is marked as the final zero-crossing.

[0059] The window size of the sliding window mean filtering is set to 5 cycles, and the window duration is dynamically adjusted according to the real-time fundamental frequency. For example, when the fundamental frequency is 50Hz, the window duration is 5×20ms = 100ms. The boundary processing method for mean filtering is: when the data at the start or end of the window is insufficient, the mirror extension method is used to supplement the virtual data points. For example, the virtual previous point at the start point n = 0 of the window takes the value of n = 1, and the virtual subsequent point at the end point n = N takes the value of n = N - 1.

[0060] The calculation result of the tangent value of the dielectric loss angle needs to be corrected by temperature compensation. The compensation coefficient is set according to the difference between the operating environment temperature of the power capacitor bank and the standard temperature (25°C). For example, when the ambient temperature is 30°C, the compensation coefficient is 0.98, and the corrected tangent value is the original value × 0.98. The temperature compensation coefficient is obtained by fitting the calibration data in the laboratory, specifically as a quadratic polynomial function. For example, the coefficient = a×ΔT² + b×ΔT + c, where ΔT is the temperature difference.

[0061] The confirmation mechanism of the real-time alarm logic includes: if there is no manual confirmation operation within 10 minutes after the alarm is triggered, the standby capacitor bank is automatically started and the faulty equipment is isolated. The calling rule for historical data comparison is: only compare the historical data under the same operating conditions (such as a load rate of 80% - 100%), excluding misjudgments caused by low loads. For example, if the current load rate is 90%, only compare the data in the historical load rate range of 85% - 95%.

[0062] Through cross - field technology integration and dynamic collaborative analysis, break through the linear processing framework of traditional dielectric loss monitoring: combine the phase - space trajectory complexity of chaos theory with the symbol transfer entropy of information theory, quantify signal distortion from the two dimensions of dynamic system stability and information - flow break, and replace the traditional frequency - domain energy or single - parameter threshold method; based on dynamically intercepting waveforms according to the real - time fundamental frequency, adaptively dividing symbol intervals, and triggering filtering with gradient thresholds, construct a closed - loop of "interference evaluation - dynamic filtering - precise calculation" to solve the problem of error accumulation of fixed parameters under time - varying interference; through multi - level error suppression of symbol change detection, interpolation correction, and sliding - window filtering, overcome the zero - crossing jitter caused by noise and improve the calculation accuracy of phase difference. Through the collaborative design of non - linear features and dynamic strategies: integrate chaos analysis, entropy quantization, and adaptive filtering across disciplines to form a closed - loop optimization system at the algorithm level, replace the rigid solutions of traditional hardware synchronization or fixed - band filtering, significantly enhance the anti - interference ability and calculation robustness in complex electromagnetic environments, and achieve high - precision dielectric loss monitoring.

[0063] The calculations involved in the embodiments are all dimensionless numerical calculations, and the preset parameters and threshold selections in the calculations are set by those skilled in the art according to the actual situation.

[0064] It should be noted that the present invention can be deployed on the device itself to achieve embedded applications, or can also run on a PC or other terminals with a user interface, so as to meet various hardware environments and usage requirements.

[0065] The above - mentioned embodiments can be implemented in whole or in part by software, hardware, firmware, or any other combination. When implemented using software, the above - mentioned embodiments can be implemented in whole or in part in the form of a computer program product. The computer program product includes one or more computer instructions or computer programs. When the computer instructions or computer programs are loaded or executed on a computer, the processes or functions described in the embodiments of the present application are generated in whole or in part. The computer can be a general - purpose computer, a special - purpose computer, a computer network, or other programmable devices. The computer instructions can be stored in a computer - readable storage medium, or transmitted from one computer - readable storage medium to another computer - readable storage medium. For example, the computer instructions can be transmitted from one website, computer, server, or data center to another website, computer, server, or data center by wired (such as infrared, wireless, microwave, etc.) means. The computer - readable storage medium can be any available medium that the computer can access, or a data storage device such as a server or data center that contains one or more collections of available media. The available medium can be a magnetic medium (such as a floppy disk, hard disk, magnetic tape), an optical medium (such as a DVD), or a semiconductor medium. The semiconductor medium can be a solid - state drive.

[0066] Those skilled in the art can clearly understand that for the convenience and conciseness of description, the specific working processes of the systems, devices, and modules described above can refer to the corresponding processes in the foregoing method embodiments and will not be elaborated herein.

[0067] In several embodiments provided in the present application, it should be understood that the disclosed systems, devices, and methods can be implemented in other ways. For example, the device embodiments described above are merely illustrative. For example, the division of the modules is only a logical function division, and there may be other division methods in actual implementation. For example, multiple modules or components can be combined or integrated into another system, or some features can be ignored or not executed. Another point is that the displayed or discussed couplings or direct couplings or communication connections to each other can be through some interfaces. The indirect couplings or communication connections of the devices or modules can be in electrical, mechanical, or other forms.

[0068] The modules described as separate components may or may not be physically separated. The components displayed as modules may or may not be physical modules. They can be located in one place or distributed to multiple network modules. Some or all of the modules can be selected according to actual needs to achieve the purpose of the solution of this embodiment.

[0069] In addition, in each embodiment of the present application, the functional modules can be integrated in a processing module, or each module can exist physically alone, or two or more modules can be integrated in one module.

[0070] If the functions are implemented in the form of software function modules and sold or used as independent products, they can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of the present application, in essence, or the part that contributes to the prior art, or this 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 enable a computer device (which can be a personal computer, a server, or a network device, etc.) to execute all or part of the steps of the methods described in each embodiment of the present application. The foregoing storage medium includes: various media such as USB flash drives, mobile hard disks, read-only memory (ROM), random access memory (RAM), magnetic disks, or optical discs that can store program codes.

[0071] As described above, it is only the specific implementation manner of the present application, but the protection scope of the present application is not limited thereto. Any person skilled in the art within the technical scope disclosed by the present application can easily think of changes or substitutions, which should all be covered within the protection scope of the present application. Therefore, the protection scope of the present application shall be subject to the protection scope of the claims.

[0072] Finally, the above are only the preferred embodiments of the present invention and are not used to limit the present invention. Any modifications, equivalent substitutions, improvements, etc. made within the spirit and principle of the present invention shall be included within the protection scope of the present invention.

Claims

1. An on-line monitoring method for dielectric loss of a power capacitor bank, characterized in that, The steps include: S1. Obtain voltage and current signals of the power capacitor bank in real time and record the signal acquisition time series; S2. Dynamically track the fundamental frequency of the voltage signal and the current signal according to the distribution of the waveform zero points and extreme points of the voltage signal and the current signal; S3, intercepting a voltage signal segment and a current signal segment of a preset duration according to the fundamental frequency, and extracting a main period waveform corresponding to the fundamental frequency in the voltage signal segment and the current signal segment; S4, constructing the phase space trajectory of the main cycle waveform and analyzing its complexity, converting the main cycle waveform into a symbol sequence and calculating the transfer entropy between the symbol sequences, and generating signal distortion characteristics according to the complexity and transfer entropy; S5. Filtering the voltage signal segment and the current signal segment according to the signal distortion characteristics and the maximum Lyapunov exponent gradients of the voltage signal segment and the current signal segment; S6. According to the zero-crossing timing difference of the main cycle waveforms of the filtered voltage signal and current signal, the phase difference between the voltage signal and the current signal is calculated to determine the dielectric loss tangent value of the power capacitor bank.

2. The on-line dielectric loss monitoring method for a power capacitor bank according to claim 1, characterized in that, Acquire the voltage and current signals of the power capacitor bank in real time and record the signal acquisition time series, including: The synchronous acquisition of the voltage signal and the current signal is started by a synchronous trigger signal to ensure that the acquisition time deviation of the voltage signal and the current signal is less than a preset time threshold; Aligning sampling timestamps of the voltage signal and the current signal to the same time base, generating a voltage signal sequence and a current signal sequence containing timestamps; Amplitude mutation detection is performed on the voltage signal sequence and the current signal sequence, and abnormal signal segments whose amplitude mutation exceeds a preset amplitude threshold are eliminated.

3. The on-line dielectric loss monitoring method for a power capacitor bank according to claim 1, characterized in that According to the distribution of zero and extreme points of the voltage and current waveforms, the fundamental frequency of the voltage and current signals is dynamically tracked, including: Extracting waveform zero points and extreme value points of voltage signal sequence and current signal sequence within a preset time window; Determine the fundamental wave period of the voltage signal sequence and the current signal sequence according to the time interval between adjacent waveform zero points; The extreme point offset errors of the voltage signal sequence and the current signal sequence are corrected by the cubic spline interpolation method to obtain the corrected fundamental wave period; The corrected inverse of the fundamental wave period is taken as the fundamental wave frequency.

4. A method for on-line monitoring of dielectric loss of a power capacitor bank according to claim 1, characterized in that, According to the fundamental frequency, a voltage signal segment and a current signal segment of a preset time length are intercepted, and a main cycle waveform corresponding to the fundamental frequency in the voltage signal segment and the current signal segment is extracted, including: Determine the preset time length as an integer multiple of the fundamental wave period according to the fundamental wave frequency, and extract the voltage signal segment and the current signal segment of the corresponding time length from the voltage signal sequence and the current signal sequence; In the intercepted voltage signal segment and current signal segment, the starting zero-crossing point of the main cycle waveform is located with the cycle length corresponding to the fundamental frequency as the unit; Starting from the initial zero-crossing point, a voltage signal segment and a current signal segment within a complete fundamental wave cycle are extracted as the main cycle waveform; The extracted main period waveform is processed by mean filtering to eliminate random noise interference.

5. A method for on-line monitoring of dielectric loss of a power capacitor bank according to claim 1, characterized in that, Construct the phase space trajectory of the main cycle waveform and analyze its complexity. At the same time, convert the main cycle waveform into a symbol sequence and calculate the transfer entropy between symbol sequences. Generate signal distortion features based on complexity and transfer entropy, including: Construct the phase space trajectory of the main periodic waveform using the time delay embedding method; Calculate the recurrence quantification analysis parameters of the phase space trajectory. The recurrence quantification analysis parameters include the recurrence rate and determinism, and generate the phase space complexity index according to the recurrence rate and determinism; When converting the main periodic waveform into a symbol sequence, divide the symbol intervals according to the amplitude distribution of the main periodic waveform, and each symbol interval corresponds to a preset amplitude range; Statistically analyze the transition probability between adjacent symbols in the symbol sequence, and calculate the transfer entropy between symbol sequences according to the transition probability; Generate the signal distortion feature according to the product of the normalized phase space complexity index and the transfer entropy, or when the phase space complexity index exceeds the preset threshold set based on historical data statistics, directly use the transfer entropy as the signal distortion feature.

6. The on-line dielectric loss monitoring method for a power capacitor bank according to claim 5, characterized in that, The delay time of the time delay embedding method is set according to the sampling rate of the main periodic waveform, and the embedding dimension is set according to the spectral characteristics of the main periodic waveform.

7. A method for on-line monitoring of dielectric loss of a power capacitor bank according to claim 1, characterized in that, Filter the voltage signal segment and the current signal segment according to the signal distortion feature and the maximum Lyapunov exponent gradient of the voltage signal segment and the current signal segment, including: Calculate the maximum Lyapunov exponent gradient of the voltage signal segment and the current signal segment. The maximum Lyapunov exponent gradient is calculated by the change rate of the maximum Lyapunov exponent within adjacent time windows; Adjust the cut-off frequency or order of the filter according to the comparison result between the absolute value of the maximum Lyapunov exponent gradient and the preset gradient threshold; If the signal distortion feature exceeds the preset distortion threshold and the absolute value of the maximum Lyapunov exponent gradient exceeds the preset gradient threshold, perform filtering on the voltage signal segment and the current signal segment using adaptive mean filtering; If the signal distortion feature does not exceed the preset distortion threshold or the absolute value of the maximum Lyapunov exponent gradient does not exceed the preset gradient threshold, perform low-pass filtering with fixed parameters.

8. The on-line dielectric loss monitoring method for a power capacitor bank according to claim 7, characterized in that The size of the filtering window is dynamically set according to the absolute value of the maximum Lyapunov exponent gradient.

9. The on-line dielectric loss monitoring method for a power capacitor bank according to claim 1, characterized in that, Calculate the phase difference between the voltage signal and the current signal according to the zero-crossing time sequence difference of the main periodic waveforms of the filtered voltage signal and the current signal to determine the tangent value of the dielectric loss angle of the power capacitor bank, including: In the main periodic waveforms of the filtered voltage signal and the current signal, respectively locate the zero-crossing points of the voltage waveform and the current waveform. The zero-crossing points are determined by symbol change detection and interpolation correction; Calculate the time difference between the zero-crossing point of the voltage waveform and the zero-crossing point of the current waveform as the zero-crossing time sequence difference; Convert the zero-crossing time sequence difference into a phase difference according to the fundamental frequency; Calculate the tangent value of the dielectric loss angle based on the phase difference. The tangent value of the dielectric loss angle is the tangent function value of the phase difference.

Images