Intelligent monitoring system and method for high-low voltage equipment

Abstract

The application relates to the technical field of power equipment monitoring, and particularly discloses an intelligent monitoring system and method for high and low voltage equipment, which comprises the following steps: collecting real-time current waveforms, separating power frequency and harmonics through adaptive filtering, obtaining real-time residual waveforms, and constructing a residual fluctuation reference band based on non-corrosion historical data; comparing the real-time residual waveforms with the residual fluctuation reference band, capturing low-frequency drift and extracting a direct current bias track; performing short window polarity verification: when there is one-way bias, combining a power frequency reconstructed waveform to remove transient disturbance and identifying effective corrosion events; when there is alternating bias and non-transient disturbance, feeding back adjustment of the reference band; then classifying events according to sections, converting corrosion charge based on energy difference, and generating a corrosion activity index through time-weighted accumulation; finally, scanning the index sequence through a sliding window, triggering a direct current corrosion early warning after transient disturbance space-time verification; and the system can adaptively suppress interference and improve the accuracy and robustness of corrosion discrimination.

Description

Technical Field

[0001] This invention relates to the field of power equipment monitoring technology, specifically to an intelligent monitoring system and method for high and low voltage equipment. Background Technology

[0002] In urban underground cable trenches, high-voltage cables and medium- and low-voltage cables are often laid in the same trench with limited spacing. After the metal sheath of the high-voltage cable is cross-interconnected and grounded, there are always power frequency circulating currents and harmonic components induced by the load current in the grounding wire. If there is a DC traction rail transit line running parallel in the trench, its ground return current will couple into the sheath grounding circuit through the soil, superimposed with stray DC and low-frequency fluctuations generated by the soil electrochemical process, complicating the composition of the sheath current. The amplitude of these stray DCs is much smaller than the power frequency circulating current, but when there are local weak points in the sheath insulation, continuous DC at the microampere level can drive local electrochemical corrosion, gradually damaging the sheath metal.

[0003] Existing online monitoring methods use RMS or total harmonic distortion (THD) as criteria, which have almost no resolution for minute DC offsets superimposed on them. Methods to detect stray DC either require pre-embedded reference electrodes or offline measurement with an external DC signal applied during a power outage, neither of which can meet the requirements for continuous monitoring without power interruption.

[0004] Furthermore, transient events such as DC bias and protective actions also generate low-frequency transient components in the sheath current, which overlap with the characteristics of corrosion-related components, easily leading to false alarms. How to separate non-stationary, intermittent weak DC and low-frequency corrosion components online using only existing current sensors, and control the false alarm rate under transient conditions, remains an unsolved algorithmic challenge in cable operation and maintenance.

[0005] Therefore, the present invention provides an intelligent monitoring system and method for high and low voltage equipment. Summary of the Invention

[0006] The purpose of this invention is to provide an intelligent monitoring system and method for high and low voltage equipment to solve the aforementioned background problems.

[0007] The objective of this invention can be achieved through the following technical solution: an intelligent monitoring method for high and low voltage equipment, comprising: The real-time current waveform of the cable sheath grounding wire is collected, and the power frequency and harmonics are separated to obtain the real-time residual waveform. The same processing is performed on the original current waveform of the historical running period that has been confirmed to be free of corrosion to obtain the historical residual waveform. The fluctuation characteristics are statistically analyzed to construct the residual fluctuation reference band. The real-time residual waveform is compared with the residual fluctuation reference band to capture low-frequency drift segments and extract the DC bias trajectory. If reference band adjustment feedback is received, the residual fluctuation reference band is adjusted, and the low-frequency drift segments are recaptured and the DC bias trajectory is extracted. A short-window polarity check is performed on the DC bias trajectory to distinguish between unidirectional bias and alternating bias. If it is unidirectional bias, the power frequency component is extracted to reconstruct the waveform, transient disturbance events are eliminated, and valid corrosion events are determined. If it is alternating bias, the reference band adjustment feedback is generated by combining transient disturbance events. Corrosion groups are categorized using effective corrosion events as an index, the corrosion charge of effective corrosion events is calculated, and the corrosion charge of effective corrosion events within the same corrosion group is accumulated with time weighting to generate a corrosion activity index. The corrosion activity index sequence is obtained by sliding the monitoring window to scan the corrosion activity index; the corrosion activity index sequence is combined with transient disturbance events for spatiotemporal verification analysis to generate DC corrosion early warning.

[0008] Furthermore, the process of constructing the residual fluctuation benchmark band is as follows: A set of historical residual waveforms is constructed based on the historical residual waveforms; for each sampling time, the current amplitude of all historical residual waveforms is extracted from the set of historical residual waveforms to form an amplitude sample group; Obtain the maximum and minimum current amplitude values ​​in the amplitude sample group. Connect the maximum current amplitude values ​​at all sampling times in sequence along the time axis to generate the upper limit boundary curve of residual fluctuation. Connect the minimum current amplitude values ​​at all sampling times in sequence along the time axis to generate the lower limit boundary curve of residual fluctuation. The residual fluctuation reference band is formed by the upper limit boundary curve of residual fluctuation and the lower limit boundary curve of residual fluctuation.

[0009] Furthermore, the process of capturing low-frequency drift segments is as follows: The real-time residual waveform is compared with the residual fluctuation reference band to mark the out-of-limit points. The out-of-limit points with consecutive positions on the time axis are grouped into a candidate drift segment and the time range is obtained. Using waveform segments that fall within the time range of candidate drift segments in the real-time residual waveform as objects, calculate the power spectral density, and integrate the power spectral integral values ​​in the frequency range below the power frequency and the power spectral integral values ​​across the entire frequency band. Calculate the proportion of the power spectrum integral value in the frequency range below the power frequency to the total power spectrum integral value of the entire frequency band; If the proportion is greater than the preset energy concentration threshold, the candidate drift segment is determined to be a low-frequency drift segment, and the DC bias trajectory is extracted based on the low-frequency drift segment.

[0010] Furthermore, the method for extracting the DC bias trajectory is as follows: The real-time residual waveform within the time range corresponding to the low-frequency drift segment is extracted to obtain the DC bias trajectory of the low-frequency drift segment.

[0011] Furthermore, the process of adjusting the residual fluctuation benchmark band is as follows: Extract the over-limit amplitude data based on the reference band adjustment feedback, and obtain the positive maximum over-limit amount and the negative maximum over-limit amount; The upper limit boundary curve of residual fluctuation is obtained by adding the positive maximum over-limit to the current amplitude at all sampling times; the lower limit boundary curve of residual fluctuation is obtained by subtracting the absolute value of the negative maximum over-limit from the current amplitude at all sampling times; thus forming the adjusted residual fluctuation reference band.

[0012] Furthermore, the process for determining a valid corrosion event is as follows: Based on the real-time current waveform, the power frequency component is extracted to reconstruct the waveform, and the real-time cycle peak sequence is obtained by analysis. The same processing method is used to obtain the historical cycle peak sequence of the original current waveform corresponding to the historical non-corrosion operation period. The real-time peak frequency sequence is compared with the historical peak frequency sequence in the time domain to identify transient disturbance intervals. If the time interval of the DC bias trajectory does not overlap with any of the transient disturbance intervals, then the DC bias trajectory is identified as a valid corrosion event.

[0013] Furthermore, the process of performing time-domain comparison is as follows: Extract the maximum value from the historical cycle peak sequence as the upper limit of the historical cycle peak and the minimum value as the lower limit of the historical cycle peak; If a certain peak value in the real-time peak value sequence is greater than the upper limit of the historical peak value or less than the lower limit of the historical peak value, then the power frequency period corresponding to the peak value is marked as a transient disturbance point. A series of transient disturbance points on the time axis are grouped into a transient disturbance event. The time interval defined by the start and end times of each transient disturbance event on the time axis constitutes a transient disturbance interval.

[0014] Furthermore, the process of generating the corrosion activity index is as follows: Set a corrosion accumulation assessment period, with the current monitoring time as the endpoint, and extract the time interval covered by the corrosion accumulation assessment period as the corrosion accumulation time window; The corrosion charge of all effective corrosion events occurring within the corrosion accumulation time window in the corrosion group is summed up to determine the corrosion activity index of the corrosion group.

[0015] Furthermore, the process of performing spatiotemporal verification analysis is as follows: The historical corrosion activity index is extracted from the original current waveform during the non-corrosion historical operating period. The maximum value of all historical corrosion activity indices is compared with the preset sensitivity threshold, and the maximum value of the two is taken as the upper limit of historical normal fluctuation. If the corrosion activity index within the monitoring window is greater than the historical normal fluctuation limit, a DC corrosion warning will be generated.

[0016] An intelligent monitoring system for high and low voltage equipment includes the following modules: The residual reference module collects the real-time current waveform of the cable sheath grounding wire, separates the power frequency and harmonics to obtain the real-time residual waveform; and performs the same processing on the original current waveform of the historical operating period that has been confirmed to be free of corrosion to obtain the historical residual waveform, and constructs the residual fluctuation reference band by statistically analyzing the fluctuation characteristics. Drift extraction module: compares the real-time residual waveform with the residual fluctuation reference band, captures low-frequency drift segments and extracts the DC bias trajectory; if reference band adjustment feedback is received, the residual fluctuation reference band is adjusted, and the low-frequency drift segments and DC bias trajectory are recaptured and extracted. Polarity discrimination module: Performs short-window polarity verification on DC bias trajectory to distinguish between unidirectional bias and alternating bias; if it is unidirectional bias, extracts power frequency component to reconstruct waveform to eliminate transient disturbance events and determine valid corrosion events; if it is alternating bias, combines transient disturbance events to generate reference band adjustment feedback. Charge quantification module: Classifies corrosion groups based on effective corrosion events, calculates the corrosion charge of effective corrosion events, and accumulates the corrosion charge of effective corrosion events within the same corrosion group using time weighting to generate a corrosion activity index; Corrosion early warning module: The corrosion activity index is obtained by sliding the monitoring window to scan the corrosion activity index. The corrosion activity index sequence is combined with transient disturbance events for spatiotemporal verification analysis to generate DC corrosion early warning.

[0017] The beneficial effects of this invention are as follows: 1. Collect the real-time current waveform of the cable sheath grounding wire, separate the power frequency and each harmonic, and obtain the real-time residual waveform; and perform the same processing on the original current waveform of the historical operation period confirmed to be without corrosion to obtain the historical residual waveform, and construct the residual fluctuation reference band by statistically analyzing the fluctuation characteristics; provide an adaptive reference that fits the actual noise level of the line for subsequent judgment, and avoid the risk of misjudgment caused by artificially setting fixed thresholds.

[0018] 2. The real-time residual waveform is compared with the residual fluctuation reference band to capture low-frequency drift segments and extract the DC bias trajectory. If reference band adjustment feedback is received, the residual fluctuation reference band is adjusted, and low-frequency drift segments and DC bias trajectory are recaptured and extracted. Through the synergy of low-frequency energy concentration criterion and closed-loop adaptive reference adjustment, the identification of intermittent, non-stationary weak DC drift is achieved, effectively avoiding misjudging gradual environmental changes or normal system fluctuations as abnormalities, and significantly improving the specificity of detection. Short-window polarity verification is performed on the DC bias trajectory to distinguish between unidirectional bias and alternating bias. If it is unidirectional bias, the power frequency component is extracted to reconstruct the waveform to eliminate transient disturbance events and determine valid corrosion events. If it is alternating bias, reference band adjustment feedback is generated in combination with transient disturbance events. Non-corrosive interference is eliminated from the mechanism level. At the same time, transient disturbance intervals can be identified based solely on the existing power frequency reconstructed waveform, without the need for additional sensing equipment or power outages, significantly reducing the false alarm rate while ensuring detection sensitivity.

[0019] 3. Corrosion groups are categorized using effective corrosion events as an index. The corrosion charge of each effective corrosion event is calculated, and the corrosion charge of effective corrosion events within the same corrosion group is accumulated using a time-weighted method to generate a corrosion activity index. This allows for a quantitative characterization of the severity and development trend of corrosion at weak points in the sheath insulation, providing comparable decision indicators for differentiated operation and maintenance. The corrosion activity index sequence is obtained by sliding the monitoring window and scanning. The corrosion activity index sequence is then combined with transient disturbance events for spatiotemporal verification analysis to generate a DC corrosion early warning. This effectively filters out occasional anomalies and ensures that the final alarm corresponds to a real and continuous corrosion process, reducing the unnecessary attendance of maintenance personnel for on-site verification. Attached Figure Description

[0020] The invention will now be further described with reference to the accompanying drawings.

[0021] Figure 1 This is a flowchart of the steps of an intelligent monitoring method for high and low voltage equipment according to the present invention; Figure 2 This is a logical thinking diagram that distinguishes between unidirectional bias and alternating bias in this invention; Figure 3 This is a functional block diagram of an intelligent monitoring system for high and low voltage equipment according to the present invention. Detailed Implementation

[0022] To make the technical means, creative features, objectives and effects of this invention easier to understand, the invention will be further described below in conjunction with specific embodiments.

[0023] Example 1 like Figures 1-2 As shown, an intelligent monitoring method for high and low voltage equipment includes: Step 1: Collect the real-time current waveform of the cable sheath grounding wire, separate the power frequency and each harmonic, and obtain the real-time residual waveform; and perform the same processing on the original current waveform of the historical operating period that has been confirmed to be without corrosion to obtain the historical residual waveform, and statistically analyze the fluctuation characteristics to construct the residual fluctuation reference band. The process of separating the power frequency and harmonics from the real-time current waveform of the cable sheath grounding wire to obtain the real-time residual waveform is as follows: The real-time current waveform is obtained from the cable sheath grounding wire using a current sensor, and anti-aliasing filtering and analog-to-digital conversion are performed on the real-time current waveform to obtain a real-time current waveform sequence. A digital phase-locked loop based on dual second-order generalized integrators is used to extract the real-time angular frequency and real-time phase angle of the power frequency component from the real-time current waveform sequence. It is understandable that the current signal in the power system contains the power frequency component (50Hz) and its harmonic components; Based on the extracted real-time angular frequency and real-time phase angle, a unit amplitude sine reference signal and a unit amplitude cosine reference signal synchronized with the power frequency component are generated. Based on the common harmonic distribution characteristics in power systems, for each integer harmonic from the 2nd to the 50th order, the sine reference signal and cosine reference signal of the corresponding harmonic are generated by multiplying the real-time phase angle by the harmonic order. The real-time current waveform sequence is used as the main input signal. The unit amplitude sine reference signal and unit amplitude cosine reference signal of the power frequency component, as well as the sine reference signal and cosine reference signal of each harmonic are sequentially combined to form the reference input vector. The reference input vector is fed into an adaptive filter, which contains a weight vector of the same dimension as the reference input vector, and is used to synthesize a linear combination of the power frequency component and each harmonic component. The adaptive filter uses the normalized least mean square algorithm to update the weight vector. The step size factor of the normalized least mean square algorithm is automatically adjusted according to the instantaneous power of the reference input vector, without the need to manually set a fixed step size value, thus ensuring convergence speed and stability. During the iterative process of the adaptive filter, the output of the adaptive filter generates a reconstructed waveform that approximates the power frequency component and each harmonic component in the real-time current waveform point by point. The reconstructed waveform is subtracted from the real-time current waveform sequence at each sampling time to obtain a residual sequence containing only stray DC components, low-frequency fluctuation components, and random noise components. This residual sequence is the real-time residual waveform. The process of performing the same processing on the original current waveform of the confirmed corrosion-free historical operating period to obtain the historical residual waveform, and constructing the residual fluctuation reference band by statistically analyzing the fluctuation characteristics is as follows: For historical operating periods confirmed to be free of corrosion, obtain the original current waveform recorded on the cable sheath grounding wire during that historical operating period; It should be noted that the basis for determining that there was no corrosion was that the cable sheath did not show any DC corrosion warning during the historical operation period in the operation and maintenance records, and no traces of electrochemical corrosion were found in the appearance inspection of the grounding wire. Using the positive zero-crossing point of the power frequency component in the original current waveform as the zero moment of the time axis, a waveform segment of a fixed duration (e.g., 200 ms) is extracted to form the aligned original current waveform. The aligned original current waveform is then subjected to the same digital phase-locked loop parameters and adaptive filter structure as the real-time residual waveform to eliminate the power frequency component and harmonic components in the original current waveform, thus obtaining the historical residual waveform. Repeat the above process to process the original current waveforms of multiple historical operating periods that are determined to be corrosion-free, forming a set of historical residual waveforms. This set of historical residual waveforms covers normal fluctuation samples under different load conditions and different environmental conditions. For each sampling moment, the current amplitude of all historical residual waveforms at that sampling moment is extracted from the historical residual waveform set and formed into an amplitude sample group. The maximum and minimum current amplitudes in the amplitude sample group are calculated. The maximum and minimum current amplitudes fully reflect the limit fluctuation range of the residual current at that sampling moment under non-corrosive conditions. By sequentially connecting the maximum current amplitude values ​​at all sampling times along the time axis, a residual fluctuation upper limit boundary curve that varies with time is generated. By sequentially connecting the minimum current amplitude values ​​at all sampling times along the time axis, a residual fluctuation lower limit boundary curve that varies with time is generated. The continuous band-shaped region enclosed by the upper limit boundary curve of residual fluctuation and the lower limit boundary curve of residual fluctuation is the residual fluctuation reference band; It should be noted that the residual fluctuation reference band characterizes the natural fluctuation range of the residual current caused by measurement system noise and normal electromagnetic induction interference under non-corrosive conditions. Any residual fluctuation exceeding the upper limit boundary curve of the residual fluctuation or falling below the lower limit boundary curve of the residual fluctuation is considered as abnormal drift. Step 2: Compare the real-time residual waveform with the residual fluctuation reference band, capture the low-frequency drift segment and extract the DC bias trajectory; if reference band adjustment feedback is received, adjust the residual fluctuation reference band, and recapture the low-frequency drift segment and extract the DC bias trajectory. The process of comparing the real-time residual waveform with the residual fluctuation reference band, capturing low-frequency drift segments, and extracting the DC bias trajectory is as follows: The real-time residual waveform obtained in step one is compared with the residual fluctuation reference band at each sampling time. At each sampling time, the real-time residual amplitude of the real-time residual waveform at that sampling time is read, and at the same time, the upper limit amplitude of the residual upper limit boundary curve of the residual fluctuation in the residual fluctuation reference band and the lower limit amplitude of the residual fluctuation at that sampling time are read. When the real-time residual waveform at a certain sampling time has a real-time residual amplitude greater than the upper limit amplitude of the residual at the corresponding sampling time, or when the real-time residual amplitude at a certain sampling time is less than the lower limit amplitude of the residual at the corresponding sampling time, then that sampling time is marked as an out-of-limit point. Cluster consecutive out-of-limit points on the time axis into a connected component, and each connected component constitutes a candidate drift segment. For each candidate drift segment, calculate the number of sampling times contained in the candidate drift segment, and use the number of sampling times as the duration of the candidate drift segment; Calculate the time interval between the start times of two adjacent candidate drift segments, and use this time interval as the occurrence interval between the previous candidate drift segment and the next candidate drift segment; Using waveform segments that fall within the time range of candidate drift segments in the real-time residual waveform as the object, the power spectral density of the waveform segments is calculated; and the power spectral density is integrated in the frequency domain to obtain the power spectral integral value in the frequency range below the power frequency and the power spectral integral value in the entire frequency band. Calculate the proportion of the power spectrum integral value in the frequency range below the power frequency to the total power spectrum integral value of the entire frequency band; If the proportion is greater than the preset energy concentration threshold, it is determined that the main energy of the waveform segment is concentrated in the frequency range below the power frequency, and the candidate drift segment is confirmed as a low-frequency drift segment. Based on the confirmed low-frequency drift segment, the real-time residual waveform within the corresponding time range of the low-frequency drift segment is directly extracted to obtain the DC bias trajectory of the low-frequency drift segment. The DC bias trajectory, duration, and occurrence interval between each low-frequency drift segment are combined into a single DC bias description record. Combine the DC bias description records corresponding to all low-frequency drift segments into a DC bias trajectory set, and pass the DC bias trajectory set to step three; Specifically, if reference band adjustment feedback is received, the process of adjusting the residual fluctuation reference band, re-capturing the low-frequency drift segment, and extracting the DC bias trajectory is as follows: If the reference band adjustment feedback generated in step three is received, the over-limit amplitude data corresponding to the specific low-frequency drift segment that triggered the adjustment is extracted from the reference band adjustment feedback; Among them, the over-limit amplitude data includes the positive maximum over-limit amount of the real-time residual waveform relative to the upper limit boundary curve of the residual fluctuation within a specific low-frequency drift segment, and the negative maximum over-limit amount of the real-time residual waveform relative to the lower limit boundary curve of the residual fluctuation within a specific low-frequency drift segment. The current amplitude at all sampling times is increased by the positive maximum over-limit to obtain the adjusted residual fluctuation upper limit boundary curve. The adjusted residual fluctuation lower limit boundary curve is obtained by subtracting the absolute value of the negative maximum over-limit from the current amplitude at all sampling times. The adjusted residual fluctuation reference band is formed by the adjusted upper limit boundary curve of residual fluctuation and the adjusted lower limit boundary curve of residual fluctuation. The original residual fluctuation reference band is replaced with the adjusted residual fluctuation reference band, and the entire process of sampling point comparison, over-limit point marking, candidate drift segment aggregation, low-frequency drift segment confirmation and DC bias trajectory extraction is re-executed with the same real-time residual waveform obtained in step one using the adjusted residual fluctuation reference band and the same real-time residual waveform, to generate an updated DC bias trajectory set. The updated DC bias trajectory set is then passed to step three again, thereby completing the closed-loop operation of step two, which adjusts the residual fluctuation reference band and re-extracts the DC bias trajectory based on the feedback signal. Step 3: Perform short-window polarity verification on the DC bias trajectory to distinguish between unidirectional bias and alternating bias; if it is unidirectional bias, extract the power frequency component to reconstruct the waveform, eliminate transient disturbance events, and determine the effective corrosion events; if it is alternating bias, combine transient disturbance events to generate reference band adjustment feedback. The process of performing short-window polarity verification on the DC bias trajectory to distinguish between unidirectional bias and alternating bias is as follows: Step 3: Receive the set of DC bias tracks transmitted in Step 2; For each DC bias trajectory in the DC bias trajectory set, perform a short-window polarity consistency check: The short-window polarity consistency check uses the duration of the low-frequency drift segment corresponding to the DC bias trajectory as the processing interval, and the period length calculated from the real-time angular frequency of the power frequency component extracted in step one as the unit duration. The processing interval is continuously divided into multiple short windows along the time axis. For each short window, the real-time residual amplitude of all sampling times of the DC bias trajectory within the short window is extracted and the arithmetic mean is calculated to obtain the DC mean of the short window; Compare the short-window DC mean with zero; If the DC mean of the short window is greater than zero, the polarity of the short window is marked as positive; if the DC mean of the short window is less than zero, the polarity of the short window is marked as negative. Traverse all short windows corresponding to the DC bias trajectory and check whether the polarity of all short windows is positive or negative. When all short windows are positive or negative, the DC bias trajectory is determined to be unidirectional bias. When positive and negative polarities coexist in all short windows, the DC bias trajectory is determined to be alternating positive and negative bias. In the case of unidirectional bias, the process of extracting the power frequency component to reconstruct the waveform and eliminating transient disturbance events is as follows: While receiving the DC bias trajectory set, the real-time current waveform and the power frequency component reconstructed waveform output by the adaptive filter in step one during the generation of the real-time residual waveform are obtained. Among them, the power frequency component reconstructed waveform is the power frequency component time-domain signal corresponding to the sampling time of the real-time current waveform sequence; For the reconstructed waveform of the power frequency component, the period duration obtained by converting the real-time angular frequency of the power frequency component extracted in step one is used as the observation unit. Waveform segments within each observation unit are extracted, and the instantaneous amplitude with the largest absolute value in the waveform segment is extracted as the peak value of the cycle of the observation unit. The peak values ​​of all observation units experienced by the reconstructed waveform of the power frequency component are arranged in time sequence to form a real-time peak value sequence; For the original current waveform corresponding to the historical corrosion-free operating period provided in step one, the same processing method as that used to generate the real-time cycle peak sequence is adopted to obtain the historical cycle peak sequence. The maximum value in the historical cycle peak sequence is extracted as the upper limit of the historical cycle peak, and the minimum value in the historical cycle peak sequence is extracted as the lower limit of the historical cycle peak. Each peak value in the real-time peak value sequence is compared sequentially with the historical peak value upper limit and the historical peak value lower limit. When a certain peak value is greater than the upper limit of the historical peak value, or when a certain peak value is less than the lower limit of the historical peak value, the power frequency cycle corresponding to that peak value is marked as a transient disturbance point. A series of transient disturbance points on the time axis are grouped into a transient disturbance event. The time interval defined by the start and end times of each transient disturbance event on the time axis constitutes a transient disturbance interval. All transient disturbance intervals are combined to form a transient disturbance mark set, which indicates the entire time interval in which cycle-level sudden changes occur in the real-time current waveform; For each DC bias trajectory in the DC bias trajectory set that is determined to be unidirectional bias, the time interval in which the DC bias trajectory is located is obtained, and the time interval is superimposed and compared with each transient disturbance interval in the transient disturbance mark set. When the time interval of the DC bias trajectory coincides with any transient disturbance interval, the DC bias trajectory is considered to be affected by the transient disturbance, and the DC bias trajectory is not identified as a valid corrosion event. When the time interval of the DC bias trajectory does not overlap with any of the transient disturbance intervals, the DC bias trajectory is considered to be unaffected by transient disturbances and is identified as a valid corrosion event. All identified valid corrosion events are combined into a valid corrosion event set, and the valid corrosion event set is passed to step four; In the case of alternating bias, the process of generating reference band adjustment feedback in conjunction with transient disturbance events is as follows: For each DC bias trajectory in the DC bias trajectory set that is determined to be an alternating positive and negative bias, a reference band adjustment feedback is generated for that DC bias trajectory: For each DC bias trajectory in the DC bias trajectory set that is determined to be alternating positive and negative bias, the time interval in which the DC bias trajectory is located is obtained, and the time interval is superimposed and compared with each transient disturbance interval in the transient disturbance mark set. When the time interval of the DC bias trajectory coincides with any transient disturbance interval, no reference band adjustment feedback is generated for the DC bias trajectory. A reference band adjustment feedback is generated only when the time interval of the DC bias trajectory does not overlap with any of the transient disturbance intervals. The generation process is as follows: Extract the real-time residual amplitude of the real-time residual waveform within the low-frequency drift segment corresponding to the DC bias trajectory that is determined to be alternating positive and negative bias at each sampling time; Read the upper limit boundary curve of the residual fluctuation in the residual fluctuation reference band provided in step one, and read the upper limit amplitude of the residual at each sampling time in the low frequency drift segment; Calculate the difference between the real-time residual amplitude at each sampling time and the upper limit amplitude of the residual at the same sampling time, and select the maximum value from all positive difference values ​​as the positive maximum over-limit; Read the residual lower limit boundary curve of the residual fluctuation in the residual fluctuation reference band at each sampling time of the low frequency drift segment; Calculate the difference between the lower limit amplitude of the residual at each sampling time and the real-time residual amplitude at the same sampling time. Select the maximum value from all positive difference values ​​and use it as the absolute value of the maximum negative over-limit. The absolute values ​​of the positive maximum over-limit and the negative maximum over-limit are combined to form the over-limit amplitude data corresponding to the DC bias trajectory, and the over-limit amplitude data constitutes a reference band adjustment feedback. Step 3 sends all the generated reference band adjustment feedback to Step 2 to trigger Step 2 to adjust the residual fluctuation reference band according to the reference band adjustment feedback, and uses the adjusted residual fluctuation reference band to recapture the low-frequency drift segment and extract the DC bias trajectory. If, during step three, the real-time residual amplitude of all sampling times within a short window is zero when performing short window polarity consistency verification, then the DC mean of the short window is zero, and the short window does not participate in polarity determination; the process skips the short window and proceeds directly to the next short window. In step three, during the polarity determination process, after skipping the short window where the DC mean is zero, if the remaining short windows still meet the condition that all polarities are consistent, then the DC bias trajectory is still determined to be unidirectional bias. The technical solution of this invention is as follows: Real-time current waveform of the cable sheath grounding wire is collected, and the power frequency and harmonics are separated to obtain a real-time residual waveform; the same processing is performed on the original current waveform of the historical operating period confirmed to be corrosion-free to obtain a historical residual waveform, and the fluctuation characteristics are statistically analyzed to construct a residual fluctuation reference band; this provides an adaptive reference that fits the actual noise level of the line for subsequent judgments, avoiding the risk of misjudgment caused by manually setting a fixed threshold; the real-time residual waveform is compared with the residual fluctuation reference band, low-frequency drift segments are captured, and DC bias trajectory is extracted; if reference band adjustment feedback is received, the residual fluctuation reference band is adjusted, and low-frequency drift segments are recaptured and DC bias trajectory is extracted; through low frequency... The synergy between energy concentration criteria and closed-loop adaptive benchmark adjustment enables the identification of intermittent, non-stationary, weak DC drift, effectively avoiding misjudging gradual environmental changes or normal system fluctuations as anomalies and significantly improving detection specificity. Short-window polarity verification of the DC bias trajectory distinguishes between unidirectional and alternating bias. For unidirectional bias, the power frequency component is extracted to reconstruct the waveform, eliminating transient disturbances and identifying valid corrosion events. For alternating bias, a benchmark band adjustment feedback is generated based on transient disturbance events, eliminating non-corrosive interference at the mechanistic level. Furthermore, transient disturbance intervals can be identified solely based on the existing power frequency reconstructed waveform, without the need for additional sensing equipment or power outages, significantly reducing false alarm rates while ensuring detection sensitivity.

[0024] Example 2 Please see Figure 1 As shown, an intelligent monitoring method for high and low voltage equipment includes: Step 4: Classify corrosion groups based on effective corrosion events, calculate the corrosion charge of effective corrosion events, and accumulate the corrosion charge of effective corrosion events within the same corrosion group using time weighting to generate a corrosion activity index. The process of classifying corrosion groups by effective corrosion events and calculating the corrosion charge for each effective corrosion event is as follows: For each effective corrosion event in the set of effective corrosion events, read the installation location identifier of the current sensor used when the effective corrosion event occurred. The installation location identifier indicates the cross-interconnection section where the cable sheath grounding wire monitored by the current sensor is located. The cross-interconnection area indicated by the installation location identifier shall be designated as the segment to which the effective corrosion event belongs; all effective corrosion events belonging to the same segment shall be assigned to the same corrosion group. For each corrosion group, extract the duration of the low-frequency drift segment corresponding to each effective corrosion event within the corrosion group; Obtain the set of historical residual waveforms. For each historical residual waveform in the set, use Discrete Fourier Transform to calculate the historical frequency band energy of the historical residual waveform in the frequency band below the power frequency. The historical frequency band energy is calculated as follows: A discrete Fourier transform is performed on the historical residual waveform to obtain its spectral sequence. Each element in the spectral sequence corresponds to a frequency component and its complex amplitude. The square of the modulus of the complex amplitude of each frequency component is taken as its square amplitude. All frequency components with frequencies lower than the power frequency are selected from the spectral sequence. The square amplitudes corresponding to each selected frequency component are accumulated to obtain the low-frequency square amplitude sum. The low-frequency square amplitude sum is multiplied by the sampling time interval between two adjacent sampling times in the historical residual waveform, and then divided by the number of sampling points in the historical residual waveform. The result is the historical frequency band energy. Calculate the arithmetic mean of the historical frequency band energy of all historical residual waveforms in the historical residual waveform set, and use the arithmetic mean as the non-corrosion reference frequency band energy; For a valid corrosion event within a corrosion group, a waveform segment within the time period of the valid corrosion event is extracted from the real-time residual waveform and used as the event residual segment. The discrete Fourier transform is used to calculate the event frequency band energy in the frequency band below the power frequency of the event residual segment. The calculation method of the event frequency band energy is consistent with the calculation method of the historical frequency band energy. Calculate the difference between the energy of the event frequency band and the energy of the non-corrosion reference frequency band, and mark it as the energy difference corresponding to the effective corrosion event; By combining the energy difference, the duration of the low-frequency drift segment, and Joule's law, the energy difference corresponding to the effective corrosion event is converted into an equivalent amount of corrosion charge. The formula for calculating the equivalent corrosion charge is as follows: multiply the energy difference by the duration of the low-frequency drift segment, and take the square root of the result of the product to obtain the equivalent corrosion charge. For each valid corrosion event within the corrosion group, a corresponding corrosion charge is calculated in the manner described above; The process of generating a corrosion activity index by time-weighted accumulation of the corrosion charge of effective corrosion events within the same corrosion group is as follows: The corrosion accumulation assessment cycle is determined by the median of the time interval between two consecutive DC corrosion events in the same type of section (i.e., cable sheath sections with the same rated voltage level, the same sheath grounding method, and the same laying environment conditions) in the historical operation and maintenance records of the cable sheath. Using the current monitoring time as the endpoint, the time interval covered by the corrosion accumulation assessment cycle is extracted backward as the corrosion accumulation time window; The corrosion charge of all effective corrosion events occurring within the corrosion accumulation time window in the corrosion group is accumulated, and the result is taken as the corrosion charge accumulation of the corrosion group. The amount of accumulated corrosion charge is determined as the corrosion activity index of the corrosion group. A corrosion activity index is calculated for each corrosion group, and the corrosion activity indices of all corrosion groups are combined into a corrosion activity index set. Step 5: Obtain the corrosion activity index sequence by sliding the monitoring window to scan the corrosion activity index; combine the corrosion activity index sequence with transient disturbance events for spatiotemporal verification analysis to generate a DC corrosion early warning. A monitoring window is set based on the duration of the corrosion accumulation assessment cycle in step four; Within the monitoring window, the corrosion activity index is continuously acquired in a sliding manner with a preset sliding step size. For each corrosion group, the data sequence of the corrosion activity index changing over time corresponding to that corrosion group is extracted and recorded as the corrosion activity index sequence. Preferably, the preset sliding step size is one-tenth to one-fifth of the corrosion accumulation assessment cycle; The process of generating a DC corrosion early warning by combining the corrosion activity index sequence with transient disturbance events for spatiotemporal verification analysis is as follows: From the historical residual waveform set provided in step one, select cable sheath sections that have the same rated voltage level, the same sheath grounding method, and the same laying environment conditions as the current cable sheath grounding wire cross-interconnection grounding section, and take them as the same type of section; obtain the original current waveform of the same type of section in the historical operating period where corrosion is confirmed to be non-corrosion. For the original current waveform, the following steps are executed in sequence: real-time residual waveform generation processing (Step 1), DC bias trajectory extraction processing (Step 2), effective corrosion event identification processing (Step 3), and corrosion activity index calculation processing (Step 4) to obtain the historical corrosion activity index corresponding to each confirmed corrosion-free historical running segment. Since there are no effective corrosion events in the confirmed corrosion-free historical running segment, the calculated historical corrosion activity index is zero. The maximum value among all obtained historical corrosion activity indices is compared with a preset sensitivity threshold (a positive value determined based on sensor background noise and normal environmental fluctuations); the larger of the two values ​​is taken as the upper limit of historical normal fluctuations; this upper limit of historical normal fluctuations characterizes the background level of the corrosion activity index under non-corrosive conditions. At each sliding position of the monitoring window, determine whether all values ​​of the corrosion activity index within that monitoring window are greater than the historical normal fluctuation limit. At the same time, obtain the set of transient disturbance markers generated in step three, and determine whether the time interval covered by the monitoring window overlaps with any transient disturbance interval in the set of transient disturbance markers; When all values ​​of the corrosion activity index within the monitoring window are greater than the historical normal fluctuation limit, and the time interval covered by the monitoring window does not overlap with any of the transient disturbance intervals in the transient disturbance marker set, the end time of the monitoring window is taken as the trigger time for the DC corrosion warning, and a DC corrosion warning is generated. The technical solution of this invention is as follows: Corrosion groups are categorized using effective corrosion events as an index; the corrosion charge of each effective corrosion event is calculated; and the corrosion charge of effective corrosion events within the same corrosion group is accumulated using a time-weighted method to generate a corrosion activity index. This allows for a quantitative characterization of the severity and development trend of corrosion at weak points in the sheath insulation, providing comparable decision indicators for differentiated operation and maintenance. A corrosion activity index sequence is obtained by sliding the monitoring window to scan the corrosion activity index. The corrosion activity index sequence is then combined with transient disturbance events for spatiotemporal verification analysis to generate a DC corrosion early warning. This effectively filters out occasional anomalies and ensures that the final alarm corresponds to a real and continuous corrosion process, reducing the ineffective attendance of maintenance personnel for on-site verification.

[0025] Example 3 Please see Figure 3 As shown, an intelligent monitoring system for high and low voltage equipment includes the following modules: The residual reference module collects the real-time current waveform of the cable sheath grounding wire, separates the power frequency and harmonics to obtain the real-time residual waveform; and performs the same processing on the original current waveform of the historical operating period that has been confirmed to be free of corrosion to obtain the historical residual waveform, and constructs the residual fluctuation reference band by statistically analyzing the fluctuation characteristics. Drift extraction module: compares the real-time residual waveform with the residual fluctuation reference band, captures low-frequency drift segments and extracts the DC bias trajectory; if reference band adjustment feedback is received, the residual fluctuation reference band is adjusted, and the low-frequency drift segments and DC bias trajectory are recaptured and extracted. Polarity discrimination module: Performs short-window polarity verification on DC bias trajectory to distinguish between unidirectional bias and alternating bias; if it is unidirectional bias, extracts power frequency component to reconstruct waveform to eliminate transient disturbance events and determine valid corrosion events; if it is alternating bias, combines transient disturbance events to generate reference band adjustment feedback. Charge quantification module: Classifies corrosion groups based on effective corrosion events, calculates the corrosion charge of effective corrosion events, and accumulates the corrosion charge of effective corrosion events within the same corrosion group using time weighting to generate a corrosion activity index; Corrosion early warning module: The corrosion activity index is obtained by sliding the monitoring window to scan the corrosion activity index. The corrosion activity index sequence is combined with transient disturbance events for spatiotemporal verification analysis to generate DC corrosion early warning.

[0026] The foregoing has provided a detailed description of one embodiment of the present invention, but this description is merely a preferred embodiment and should not be construed as limiting the scope of the invention. All equivalent variations and modifications made within the scope of the present invention should still fall within the scope of the present invention.

Claims

1. A method for intelligent monitoring of high and low voltage equipment, characterized in that: include: The real-time current waveform of the cable sheath grounding wire is collected, and the power frequency and harmonics are separated to obtain the real-time residual waveform. The same processing was performed on the original current waveform of the historical running period that was confirmed to be free of corrosion to obtain the historical residual waveform, and the statistical fluctuation characteristics were used to construct the residual fluctuation reference band. The real-time residual waveform is compared with the residual fluctuation reference band to capture low-frequency drift segments and extract the DC bias trajectory. If reference band adjustment feedback is received, the reference band for residual fluctuation is adjusted, and the low-frequency drift segment is recaptured and the DC bias trajectory is extracted. Short-window polarity verification is performed on the DC bias trajectory to distinguish between unidirectional bias and alternating bias. If it is a unidirectional bias, extract the power frequency component to reconstruct the waveform, eliminate transient disturbance events, and determine the valid corrosion events; If it is an alternating bias, a reference band adjustment feedback is generated by combining transient disturbance events; Corrosion groups are categorized using effective corrosion events as an index, the corrosion charge of effective corrosion events is calculated, and the corrosion charge of effective corrosion events within the same corrosion group is accumulated with time weighting to generate a corrosion activity index. The corrosion activity index sequence is obtained by sliding the monitoring window to scan the corrosion activity index; the corrosion activity index sequence is combined with transient disturbance events for spatiotemporal verification analysis to generate DC corrosion early warning.

2. The intelligent monitoring method for high and low voltage equipment according to claim 1, characterized in that: The process of constructing the residual fluctuation benchmark band is as follows: A set of historical residual waveforms is constructed based on the historical residual waveforms; for each sampling time, the current amplitude of all historical residual waveforms is extracted from the set of historical residual waveforms to form an amplitude sample group. Obtain the maximum and minimum current amplitude values ​​in the amplitude sample group. Connect the maximum current amplitude values ​​at all sampling times in sequence along the time axis to generate the upper limit boundary curve of residual fluctuation. Connect the minimum current amplitude values ​​at all sampling times in sequence along the time axis to generate the lower limit boundary curve of residual fluctuation. The residual fluctuation reference band is formed by the upper limit boundary curve of residual fluctuation and the lower limit boundary curve of residual fluctuation.

3. The intelligent monitoring method for high and low voltage equipment according to claim 1, characterized in that: The process of capturing low-frequency drift segments is as follows: The real-time residual waveform is compared with the residual fluctuation reference band to mark the out-of-limit points. The out-of-limit points with consecutive positions on the time axis are grouped into a candidate drift segment and the time range is obtained. Using waveform segments that fall within the time range of candidate drift segments in the real-time residual waveform as objects, calculate the power spectral density, and integrate the power spectral integral values ​​in the frequency range below the power frequency and the power spectral integral values ​​across the entire frequency band. Calculate the proportion of the power spectrum integral value in the frequency range below the power frequency to the total power spectrum integral value of the entire frequency band; If the proportion is greater than the preset energy concentration threshold, the candidate drift segment is determined to be a low-frequency drift segment, and the DC bias trajectory is extracted based on the low-frequency drift segment.

4. The intelligent monitoring method for high and low voltage equipment according to claim 3, characterized in that: The method for extracting the DC bias trajectory is as follows: The real-time residual waveform within the time range corresponding to the low-frequency drift segment is extracted to obtain the DC bias trajectory of the low-frequency drift segment.

5. The intelligent monitoring method for high and low voltage equipment according to claim 1, characterized in that: The process of adjusting the residual fluctuation benchmark band is as follows: Extract the over-limit amplitude data based on the reference band adjustment feedback, and obtain the positive maximum over-limit amount and the negative maximum over-limit amount; The upper limit boundary curve of residual fluctuation is obtained by adding the positive maximum over-limit to the current amplitude at all sampling times; the lower limit boundary curve of residual fluctuation is obtained by subtracting the absolute value of the negative maximum over-limit from the current amplitude at all sampling times; thus forming the adjusted residual fluctuation reference band.

6. The intelligent monitoring method for high and low voltage equipment according to claim 1, characterized in that: The process for determining a valid corrosion event is as follows: Based on the real-time current waveform, the power frequency component is extracted to reconstruct the waveform, and the real-time cycle peak sequence is obtained by analysis. The same processing method is used to obtain the historical cycle peak sequence of the original current waveform corresponding to the historical non-corrosion operation period. The real-time peak frequency sequence is compared with the historical peak frequency sequence in the time domain to identify transient disturbance intervals. If the time interval of the DC bias trajectory does not overlap with any of the transient disturbance intervals, then the DC bias trajectory is identified as a valid corrosion event.

7. The intelligent monitoring method for high and low voltage equipment according to claim 6, characterized in that: The process of performing time-domain comparison is as follows: Extract the maximum value from the historical cycle peak sequence as the upper limit of the historical cycle peak and the minimum value as the lower limit of the historical cycle peak; If a certain peak value in the real-time peak value sequence is greater than the upper limit of the historical peak value or less than the lower limit of the historical peak value, then the power frequency period corresponding to the peak value is marked as a transient disturbance point. A series of transient disturbance points on the time axis are grouped into a transient disturbance event. The time interval defined by the start and end times of each transient disturbance event on the time axis constitutes a transient disturbance interval.

8. The intelligent monitoring method for high and low voltage equipment according to claim 1, characterized in that: The process of generating the corrosion activity index is as follows: Set a corrosion accumulation assessment period, with the current monitoring time as the endpoint, and extract the time interval covered by the corrosion accumulation assessment period as the corrosion accumulation time window; The corrosion charge of all effective corrosion events occurring within the corrosion accumulation time window in the corrosion group is summed up to determine the corrosion activity index of the corrosion group.

9. The intelligent monitoring method for high and low voltage equipment according to claim 1, characterized in that: The process of performing spatiotemporal verification analysis is as follows: The historical corrosion activity index is extracted from the original current waveform during the non-corrosion historical operating period. The maximum value of all historical corrosion activity indices is compared with the preset sensitivity threshold, and the maximum value of the two is taken as the upper limit of historical normal fluctuation. If the corrosion activity index within the monitoring window is greater than the historical normal fluctuation limit, a DC corrosion warning will be generated.

10. An intelligent monitoring system for high and low voltage equipment, used to implement the intelligent monitoring method for high and low voltage equipment as described in any one of claims 1-9, characterized in that: Including the following: The residual reference module collects the real-time current waveform of the cable sheath grounding wire, separates the power frequency and harmonics to obtain the real-time residual waveform; and performs the same processing on the original current waveform of the historical operating period that has been confirmed to be free of corrosion to obtain the historical residual waveform, and constructs the residual fluctuation reference band by statistically analyzing the fluctuation characteristics. Drift extraction module: compares the real-time residual waveform with the residual fluctuation reference band, captures low-frequency drift segments and extracts the DC bias trajectory; If reference band adjustment feedback is received, the reference band for residual fluctuation is adjusted, and the low-frequency drift segment is recaptured and the DC bias trajectory is extracted. Polarity discrimination module: performs short-window polarity verification on DC bias trajectory to distinguish between unidirectional bias and alternating bias; If it is a unidirectional bias, extract the power frequency component to reconstruct the waveform, eliminate transient disturbance events, and determine the valid corrosion events; If it is an alternating bias, a reference band adjustment feedback is generated by combining transient disturbance events; Charge quantification module: Classifies corrosion groups based on effective corrosion events, calculates the corrosion charge of effective corrosion events, and accumulates the corrosion charge of effective corrosion events within the same corrosion group using time weighting to generate a corrosion activity index; Corrosion early warning module: The corrosion activity index is obtained by sliding the monitoring window to scan the corrosion activity index. The corrosion activity index sequence is combined with transient disturbance events for spatiotemporal verification analysis to generate DC corrosion early warning.

Images