Industrial data analysis and monitoring method for station low-voltage ac power supply system

Abstract

The present application relates to the technical field of power monitoring and industrial data analysis, in particular to an industrial data analysis and monitoring method for a station low-voltage alternating current power supply system, comprising: synchronously collecting voltage signals, current signals and environmental magnetic field change signals at a feeder outlet; performing adaptive noise cancellation on the voltage signals and the current signals based on the environmental magnetic field change signals to obtain voltage residual signals and current residual signals; performing envelope detection processing on the voltage residual signals to determine an envelope event frequency; performing resistance estimation processing based on the voltage residual signals and the current residual signals to determine a resistance change value; in a case where the envelope event frequency is greater than a preset event frequency threshold and the resistance change value is greater than a first preset resistance change threshold, determining that the current feeder has contact degradation, and obtaining a first early warning result; the present application can distinguish between external electromagnetic interference and real contact abnormalities, improve anti-interference capability and reduce misjudgment.

Description

Technical Field

[0001] This invention relates to the field of power monitoring and industrial data analysis technology, specifically to a method for industrial data analysis and monitoring of station low-voltage AC power supply systems. Background Technology

[0002] Feeders in low-voltage AC power supply systems for stations are prone to poor contact, increased resistance, or localized degradation during long-term operation under load. Existing monitoring methods mainly rely on the effective values ​​of voltage and current or single threshold exceedances for detection, and combine waveform disturbances or resistance changes to judge contact anomalies. Due to electromagnetic interference and load changes at the site, environmental noise interference is introduced into the collected signals, making it difficult for the above method to distinguish between external electromagnetic interference and feeder contact degradation. The anti-interference capability is lower than the preset standard. At the same time, the method relies on a single electrical parameter or a single threshold condition for judgment, without comprehensively analyzing the frequency of short-term abnormal events and the continuous changing trend of contact resistance. This results in the false alarm trigger probability exceeding the set threshold and the inability to extract the characteristics of early slow degradation. Furthermore, existing methods lack a comprehensive processing mechanism for signal filtering stability, early warning classification, and subsequent failure trends, making it impossible to accurately identify and track the feeder's operating status under complex scenarios such as sudden interference changes, light load fluctuations, and long-term degradation. Summary of the Invention

[0003] The purpose of this invention is to provide an industrial data analysis and monitoring method for station low-voltage AC power supply systems, solving the problems of insufficient anti-interference capability and difficulty in distinguishing external interference from actual degradation in existing technologies. It also prevents transient interference or single-feature distortion from being misjudged as contact anomalies, reducing the probability of false alarms while maintaining the sensitivity of target feature extraction. This invention can be achieved through the following technical solutions: The industrial data analysis and monitoring method for station low-voltage AC power supply system includes: synchronously collecting voltage signals, current signals and ambient magnetic field change signals by voltage sensors, current sensors and ambient magnetic field sensors set at the feeder outlet within a preset sampling period; Adaptive noise cancellation is performed on voltage and current signals based on environmental magnetic field variation signals to obtain residual voltage and residual current signals, respectively. Envelope detection processing is performed on the residual voltage signal to determine the envelope event frequency; resistance estimation processing is performed based on the residual voltage signal and the residual current signal to determine the resistance change value; when the envelope event frequency is greater than a preset event frequency threshold and the resistance change value is greater than a first preset resistance change threshold, it is determined that there is contact degradation in the current feeder, and a first warning result is obtained.

[0004] Preferably, adaptive noise cancellation is performed on the voltage and current signals based on the environmental magnetic field change signal to obtain the residual voltage signal and residual current signal, respectively, including: A voltage channel adaptive filter for processing the voltage signal and a current channel adaptive filter for processing the current signal are constructed respectively, and both the voltage channel adaptive filter and the current channel adaptive filter have weight vectors; For the voltage channel adaptive filter, the voltage signal is used as the target input signal, the environmental magnetic field change signal is used as the reference input signal, the filter output value is determined based on the reference input signal and the weight vector of the voltage channel adaptive filter, the residual voltage signal is obtained, and the weight vector of the voltage channel adaptive filter is updated. For the current channel adaptive filter, the current signal is used as the target input signal, the ambient magnetic field change signal is used as the reference input signal, the filter output value is determined based on the reference input signal and the weight vector of the current channel adaptive filter, the residual current signal is obtained, and the weight vector of the current channel adaptive filter is updated.

[0005] Preferably, after updating the weight vector of the voltage channel adaptive filter, the method further includes: Within a preset sliding time window, the sliding variance is determined for the historical update values ​​of the weight vector of the voltage channel adaptive filter in the time series, and the increase in the difference of the sliding variance within adjacent consecutive sliding time windows is determined. If the increase in the difference exceeds the preset increment threshold, the weight vector at the end of the previous sliding time window remains unchanged, and invalid data markers are added to the remaining voltage signal in the current sliding time window. If the increase in the difference of the sliding variance within adjacent consecutive sliding time windows does not exceed a preset increment threshold, the updated weight vector is used for subsequent processing.

[0006] Preferably, envelope detection processing is performed on the residual voltage signal to determine the envelope event frequency, including: The power frequency and its characteristic harmonics in the residual voltage signal are bandpass filtered to obtain a bandpass voltage signal. An amplitude envelope is extracted from the bandpass voltage signal to obtain the envelope signal. Within multiple consecutive preset time windows, a threshold value is determined based on the root mean square value and standard deviation of the envelope signal. The number of independent out-of-limit events in the envelope signal with amplitudes greater than the threshold value and time intervals greater than a preset time interval is counted. Based on the number of independent out-of-limit events and the length of the preset time window, the envelope event frequency is determined.

[0007] Preferably, the resistance estimation process based on the residual voltage signal and the residual current signal to determine the resistance change value includes: Based on the residual voltage signal and the residual current signal, a fast Fourier transform is used to extract the fundamental voltage RMS value and the fundamental current RMS value within the current preset time window and the adjacent previous preset time window in the continuous preset time window. After performing phase compensation on the fundamental effective value of the voltage and the fundamental effective value of the current, the estimated value of the contact resistance in the current preset time window is determined; based on the estimated value of the contact resistance in the current preset time window and the estimated value of the contact resistance in the previous preset time window, the value of the resistance change is determined.

[0008] Preferably, the method further includes: if the envelope event frequency is greater than a preset event frequency threshold and the resistance change value is not greater than a first preset resistance change threshold, it is determined to be an external magnetic field interference event, and no warning result is generated; If the envelope event frequency is not greater than a preset event frequency threshold and the resistance change value is greater than a second preset resistance change threshold in multiple consecutive preset time windows and the resistance change value in the later preset time window is greater than the resistance change value in the previous preset time window, it is determined that the current feeder has resistive degradation, and a second warning result is obtained. If the envelope event frequency is not greater than the preset event frequency threshold and the resistance change value is not greater than the second preset resistance change threshold, the current feeder status is determined to be normal; wherein, the first preset resistance change threshold is greater than the second preset resistance change threshold.

[0009] Preferably, after obtaining the first warning result or the second warning result, the method further includes: constructing an exponential degradation model based on a sequence of historical contact resistance estimates composed of multiple preset time windows and the mapping relationship between time nodes and contact resistance estimates in the sequence; Based on the exponential degradation model and the preset contact resistance failure limit, the predicted failure time is determined, and the difference between the predicted failure time and the current time is taken as the remaining safe operating time.

[0010] Preferably, after obtaining the first warning result, the method further includes: when the preset switching conditions are met, performing a light load switching process on the current feeder to obtain the remaining current signal and the remaining voltage signal after the switch; Based on the remaining current signal and the remaining voltage signal after switching, the estimated contact resistance during the light load period is recalculated. Determine the peak fluctuation range of the estimated contact resistance during the light load period, and the decrease range of the estimated contact resistance compared to the value before switching; and take the maximum peak value of the fluctuation of the estimated contact resistance within a preset steady-state time period as the first amplitude threshold, and take the lower limit of the preset decrease range of the estimated contact resistance as the second amplitude threshold. If the decrease is greater than the second magnitude threshold, the first warning result is marked as a lower warning level, and the sampling window length when performing the fast Fourier transform is increased. Based on the increased sampling window length, the step of resistance estimation based on the residual voltage signal and the residual current signal is performed again. If the peak fluctuation amplitude is greater than the first amplitude threshold and the decrease amplitude is not greater than the second amplitude threshold, maintain the first warning result and record the fluctuation anomaly flag.

[0011] The beneficial effects of this invention are: 1. This invention configures independent parallel adaptive filter instances for corresponding voltage and current channels respectively, and performs dimensionless processing on the target input signal and reference input signal before the input filter according to the full-scale calibration benchmark value of their respective channels. At the same time, the weight vector is updated based on the residual voltage signal as the error signal. This can avoid cross-contamination of weights between different physically coupled channels and gradient anomalies and numerical overflows caused by differences in dimensions, thereby improving the convergence stability and data processing reliability of the adaptive noise cancellation process. 2. This invention calculates the sliding variance of the weight vector sequence within a preset sliding time window, and keeps the weight vector unchanged at the end of the previous sliding time window when the difference in sliding variance between adjacent consecutive sliding time windows increases by more than a preset increment threshold. It also adds invalid data markers to the voltage residual signal in the current sliding time window. This can prevent the filter from being updated incorrectly when interference changes suddenly, reduce false cancellation and voltage residual signal distortion caused by weight drift, and improve the reliability of subsequent envelope detection results. 3. This invention performs bandpass filtering, Hilbert transform to construct an analytical signal, and calculates the envelope signal in the power frequency and power frequency characteristic harmonic band of the residual voltage signal. After constructing a threshold value based on the root mean square value and standard deviation within a preset time window, the number of independent out-of-limit events that meet the time interval requirements is counted and converted into the envelope event frequency. This can extract the frequency of amplitude anomalies caused by partial discharge at the contact point or mechanical vibration of the contact surface, avoid repeatedly counting continuous fluctuations in the same disturbance, and thus enhance the ability to identify short-term contact anomalies. Attached Figure Description

[0012] Other features, objects, and advantages of the invention will become more apparent from the following detailed description of non-limiting embodiments with reference to the accompanying drawings: Figure 1This is a flowchart illustrating an industrial data analysis and monitoring method for a station low-voltage AC power supply system, as described in an embodiment of this application. Detailed Implementation

[0013] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0014] Please see Figure 1 The industrial data analysis and monitoring method for station low-voltage AC power supply system includes: synchronously collecting voltage signals, current signals and ambient magnetic field change signals by voltage sensors, current sensors and ambient magnetic field sensors set at the feeder outlet within a preset sampling period; Adaptive noise cancellation is performed on voltage and current signals based on environmental magnetic field change signals to obtain residual voltage and current signals, respectively; envelope detection processing is then performed on the residual voltage signals to determine the envelope event frequency; Resistance estimation is performed based on residual voltage and residual current signals to determine the resistance change value. If the envelope event frequency is greater than a preset event frequency threshold and the resistance change value is greater than a first preset resistance change threshold, it is determined that there is contact degradation in the current feeder, and a first warning result is obtained. Adaptive noise cancellation is performed on voltage and current signals based on the environmental magnetic field change signal to obtain residual voltage and residual current signals, respectively. This includes: constructing a voltage channel adaptive filter for processing voltage signals and a current channel adaptive filter for processing current signals, respectively. Both the voltage channel adaptive filter and the current channel adaptive filter have weight vectors. For the voltage channel adaptive filter, the voltage signal is used as the target input signal and the ambient magnetic field change signal is used as the reference input signal. The filtered output value is determined based on the reference input signal and the weight vector of the voltage channel adaptive filter, the residual voltage signal is obtained, and the weight vector of the voltage channel adaptive filter is updated. For the current-channel adaptive filter, the current signal is used as the target input signal and the ambient magnetic field change signal is used as the reference input signal. The filtered output value is determined based on the reference input signal and the weight vector of the current-channel adaptive filter, and the residual current signal is obtained and the weight vector of the current-channel adaptive filter is updated. After updating the weight vector of the voltage channel adaptive filter, the method further includes: within a sliding time window of a preset length, determining the sliding variance of the historical update values ​​of the weight vector of the voltage channel adaptive filter in the time series, and determining the increment of the difference in sliding variance within adjacent consecutive sliding time windows. If the increase in the difference exceeds the preset increment threshold, the weight vector at the end of the previous sliding time window remains unchanged, and invalid data markers are added to the remaining voltage signal in the current sliding time window. If the increase in the difference of the sliding variance within adjacent consecutive sliding time windows does not exceed a preset increment threshold, the updated weight vector is used for subsequent processing.

[0015] Based on the above implementation, this embodiment is for online monitoring of feeders in station low-voltage AC power supply systems. First, voltage signals, current signals and ambient magnetic field change signals are collected synchronously. Then, the ambient magnetic field change signals are used to adaptively cancel the residual voltage and residual current signals to obtain voltage and residual current signals with high signal-to-noise ratio. Envelope detection is performed on the residual voltage signal to extract the envelope event frequency. The resistance is estimated by combining the residual voltage signal and the residual current signal to obtain the resistance change value. When the envelope event frequency and the resistance change value simultaneously meet the preset criteria, it is determined that there is contact degradation in the current feeder and the first warning result is output. By introducing changes in the ambient magnetic field into the reference channel, external electromagnetic interference is separated from the degradation of the feeder itself, thus preventing transient interference from being misjudged as contact abnormalities. The preset sampling period refers to the repetition period for synchronously sampling the feeder signal. For example, the voltage, current and ambient magnetic field are synchronously acquired once within each system sampling period so that the three types of initial signals can be synchronously analyzed based on a unified time reference. The residual voltage signal and residual current signal are signal components after adaptive noise cancellation, used to preserve the actual operating changes of the feeder, such as changes in contact point voltage drop and load current fluctuations, and to reduce external magnetic field disturbances. The envelope event frequency is the statistical frequency of short-time out-of-limit events in the residual voltage signal, used to characterize the frequency of amplitude anomalies caused by partial discharge at the contact point or mechanical vibration of the contact surface; the resistance change value is the difference between the resistance estimation results within adjacent time windows, used to reflect whether the contact condition has continuously deteriorated. By using the ambient magnetic field change signal for reference cancellation, voltage and current characteristic components with a signal-to-noise ratio below a preset threshold can be effectively extracted and retained; at the same time, the envelope event frequency and resistance change value are jointly judged to avoid false alarms caused by a single threshold. This embodiment can identify contact degradation based on synchronous acquisition, adaptive cancellation, and dual-condition judgment, avoiding misjudgment caused by fluctuations in a single electrical parameter, and solving the problems of insufficient anti-interference capability and difficulty in distinguishing external interference from actual degradation in the prior art. In this embodiment, adaptive noise cancellation employs an adaptive filter for channel-by-channel processing. The target input signal and the reference input signal are divided by the full-scale calibration reference value of their respective channels before being input to the filter, and converted into a dimensionless digital sequence with a consistent range for iterative calculation. The residual voltage signal is obtained by subtracting the filtered output value from the target input signal; the residual voltage signal is used as the error signal and the ambient magnetic field change signal is used as the reference input signal to update the weight vector; the calculation process for updating the weight vector is as follows: in, This is the updated weight vector for the next time step; This represents the weight vector at the current moment; The preset update step size factor is dimensionless. This is the dimensionless residual voltage signal at the current moment, i.e., the error signal; This refers to the discrete sampling time sequence number; The reference input vector is composed of the dimensionless environmental magnetic field change signals at the current and historical moments; the same processing procedure is repeated on the current signal to obtain the residual current signal; The preset update step size factor is used to control the convergence speed and stability of the algorithm. The specific calibration method is as follows: In an environment where the amplitude of the change signal of the ambient magnetic field is lower than the preset steady-state threshold, different step size factors are input into the adaptive filter, and the step size value that minimizes the mean square error of the remaining voltage signal within a preset time of 10 power frequency cycles is selected as the preset update step size factor, and the value is greater than 0. Among them, the adaptive filter is used to estimate the coupling effect of the environmental magnetic field change signal on the voltage residual signal and the current residual signal in real time, and the weight vector is used to represent the current influence of the reference channel on the target channel, so as to fit the coupling relationship parameters of the current electromagnetic interference and obtain the steady-state operating electrical parameters of the feeder. The filtered output value is the interference estimate of the reference input signal after the current weight mapping, while the voltage residual signal is the result after subtracting the interference from the original voltage, preserving the voltage change of the feeder body; the same adaptive noise cancellation process is performed independently on the current signal; Since the physical transmission paths of the environmental magnetic field to the voltage loop and the current loop are inconsistent, adaptive filter instances corresponding to the voltage channel and the current channel and independent of each other are configured, and each maintains and updates its corresponding weight vector independently. When the current signal is processed repeatedly, the weight vector corresponding to the current channel is used for initialization and inner product operation, thereby avoiding cross-contamination of weights between different physical coupling channels. This embodiment uses the same type of reference signal to collaboratively cancel two target signals, captures the coupling relationship between changes in the ambient magnetic field and the feeder signal, and converts common-mode interference into a calculable cancellation term; without changing the original sampling structure, it reduces the impact of transient electromagnetic interference on subsequent envelope detection and resistance estimation, thereby improving the stability of subsequent degradation identification. In this embodiment, the stability judgment of the weight vector is used to prevent the filter from being updated incorrectly when there is a sudden change in interference. Within a sliding time window of a preset length, the variance of the weight vector in the time series is calculated to obtain the sliding variance. When the difference of the sliding variance in adjacent consecutive sliding time windows increases by more than a preset increment threshold, the weight vector at the end of the previous sliding time window is kept unchanged, and invalid data is added to the voltage remaining signal in the current sliding time window. When the increase in the difference of the sliding variance within adjacent consecutive sliding time windows is not greater than the preset increment threshold, the updated weight vector is used for subsequent processing; where the sliding time window is a fixed-length segment used to observe the short-term change trend of the weight vector, such as a time period covering a preset number of consecutive sampling points, in order to separate transient disturbances from stable changes. The sliding variance reflects the degree of dispersion of the weight vector within a window, and physically corresponds to whether the coupling relationship of the environmental magnetic field changes abruptly. The formula for calculating the sliding variance is: in, The sliding variance within the sliding time window; This represents the dimension of the weight vector, i.e., the total number of coefficients. For the first time window within the sliding time window The coefficient in the first... The value at each sampling time; This is the mean of the coefficient over the sliding time window; This represents the number of sampling points included within the sliding time window; The preset incremental threshold is a boundary condition for judging whether the weight change is abrupt. It can be obtained from historical sample statistics and is used to distinguish between normal environmental fluctuations and abnormal interference mutations. The specific threshold is determined as follows: in historical operating data where the estimated contact resistance is within the preset normal range and the rate of change of the ambient magnetic field signal is lower than the preset rate of change threshold, the maximum change of the sliding variance under different time windows is extracted, and the maximum change is multiplied by a redundancy coefficient of 1.2 to 1.5 to serve as the preset incremental threshold, so as to ensure that the normal environmental fluctuations of the system are included within the allowable threshold. Invalid data markers are used to indicate that the remaining voltage signal in the current window has not reached the preset reliability requirement, so as to limit the output data during the filter instability period from participating in subsequent identification calculations; This embodiment establishes a stability judgment mechanism through sliding variance and incremental threshold. When disturbances suddenly occur, the weights are frozen to prevent the filter from incorrectly updating abnormal coupling relationships to the model parameters; updates continue when the system is stable, thereby achieving adaptive adjustment to changing environments. This embodiment can reduce false cancellation caused by weight drift, avoid distortion of residual voltage signals, and improve the reliability of envelope event frequency extraction.

[0016] Envelope detection processing is performed on the residual voltage signal to determine the envelope event frequency. This includes: bandpass filtering of the power frequency and its characteristic harmonics in the residual voltage signal to obtain a bandpass voltage signal; extracting the amplitude envelope from the bandpass voltage signal to obtain the envelope signal; determining a threshold value based on the root mean square value and standard deviation of the envelope signal within multiple consecutive preset time windows; counting the number of independent out-of-limit events in the envelope signal whose amplitude is greater than the threshold value and whose time interval is greater than a preset time interval; and determining the envelope event frequency based on the number of independent out-of-limit events and the length of the preset time window. Resistance estimation is performed based on residual voltage and residual current signals to determine the resistance change value. This includes: using Fast Fourier Transform to extract the fundamental voltage and fundamental current RMS values ​​within the current preset time window and the adjacent previous preset time window within a continuous preset time window. After performing phase compensation on the fundamental effective values ​​of voltage and current, the estimated contact resistance value for the current preset time window is determined; based on the estimated contact resistance value for the current preset time window and the estimated contact resistance value for the previous preset time window, the resistance change value is determined.

[0017] This embodiment performs envelope detection on the residual voltage signal to extract amplitude abnormality events when the feeder contact is abnormal; at the same time, it performs fundamental resistance estimation on the residual voltage signal and the residual current signal to extract the change in contact state. Bandpass filtering is performed on the frequency band containing the power frequency and its characteristic harmonics in the residual voltage signal to obtain a bandpass voltage signal; then Hilbert transform is performed on the bandpass voltage signal to construct an analytic signal. The envelope signal is obtained by calculating the absolute value of the analytical signal; the root mean square value and standard deviation of the envelope signal are calculated within a preset time window, and the sum of the products of the root mean square value and the preset multiple and standard deviation is used as the threshold value. The number of independent out-of-limit events in the envelope signal whose amplitude is greater than the threshold and whose time interval is greater than the preset time interval is counted, and the ratio of this number to the preset time window length is taken as the envelope event frequency. Meanwhile, the effective values ​​of the fundamental voltage and the fundamental current are extracted using Fast Fourier Transform. After the residual voltage signal and the residual current signal are in phase or after phase compensation, the effective value of the fundamental voltage is divided by the effective value of the fundamental current to obtain the estimated value of the contact resistance. Then, the difference between adjacent time windows is calculated to obtain the resistance change value. The bandpass filtering process is used to retain the frequency band components related to contact anomalies near the power frequency and its characteristic harmonics, while filtering out noise components outside the set frequency band range. The specific frequency band boundary setting rules are as follows: A bandpass filter band is constructed by setting the lower cutoff frequency below the system power frequency and avoiding DC components, and the upper cutoff frequency as the frequency band boundary covering up to the 5th harmonic. The analytic signal is used to convert a real signal into a complex signal form that supports envelope calculation. It is constructed by using the obtained bandpass voltage signal as the real part of the analytic signal and the orthogonal component obtained after the bandpass voltage signal undergoes Hilbert transformation as the imaginary part of the analytic signal. The envelope signal is used to characterize the outer contour of the amplitude change of the residual voltage signal. Specifically, the absolute value operation involves calculating the square root of the sum of the squares of the real and imaginary parts to obtain the instantaneous amplitude envelope on the time axis. The threshold value is calculated using the following formula: in, Threshold value, unit: V; The root mean square value of the envelope signal within a preset time window; The standard deviation of the envelope signal within a preset time window, in units of: ; This is a dimensionless preset multiple; The preset multiplier is used to adjust the sensitivity to abnormal peaks. The value can be set based on sample statistics. The specific calibration criteria are as follows: Extract the maximum deviation amplitude of the envelope signal from the historical samples of the feeder's normal operation. Subtract the root mean square value of the normal envelope signal obtained from the statistical analysis of historical normal steady-state operation data from the maximum deviation amplitude. Divide the difference by the standard deviation of the normal envelope signal obtained from the statistical analysis of historical normal steady-state operation data. Use the upper limit or average value of the obtained ratio as a preset multiple. The preset multiple ranges from 3 to 5. Independent out-of-limit events require that the amplitude exceeds the limit and that the time interval between events is kept to prevent consecutive fluctuations in the same disturbance from being counted repeatedly. The fundamental effective values ​​of voltage and current correspond to the voltage drop and current intensity of the feeder on the main power frequency component, respectively, and their ratio reflects the changing trend of contact resistance. Phase compensation is used to handle measurement deviations when there is a phase difference between voltage and current. The calculation process is as follows: The fundamental phases of the residual voltage and current signals are extracted using a Fast Fourier Transform, and their phase difference is calculated. The formula for estimating the contact resistance is: in, This is an estimated value of the contact resistance for the current preset time window, in units of: ; This is the effective value of the fundamental voltage, in V. This is the effective value of the fundamental current, in amperes (A). The phase difference between the fundamental phase of the voltage and the fundamental phase of the current, in rad; the voltage values ​​used in the calculation only include the active voltage drop component that is in phase with the residual current signal. The characteristics of the purely resistive contact resistance of the feeder are obtained by separating the reactive voltage drop component generated by the inductive or capacitive load. In the calculation step of the contact resistance estimation, a lower limit constraint is introduced: when the obtained effective value of the fundamental current is not greater than the system's preset dead zone current threshold, the calculation of the contact resistance estimation value for the preset time window is stopped. The dead zone current threshold is calibrated based on the system's current transformer measurement accuracy baseline and the on-site steady-state background noise level, and the steady-state calculation result of the previous time window or the output invalid flag is retained, so as to avoid division overflow and sudden distortion of results caused by the denominator being zero or lower than the calculation accuracy lower limit. This embodiment extracts short-term out-of-limit events through envelope detection and characterizes contact anomalies as resistance change parameters through fundamental effective value calculation. This allows the frequency of frequent abnormal events and whether the resistance continues to increase to mutually corroborate each other. In field environments with multiple interference sources, both event frequency and resistance change criteria are obtained simultaneously, avoiding misjudgments caused by distortion of a single feature. If the envelope event frequency is greater than the preset event frequency threshold and the resistance change value is greater than the first preset resistance change threshold, it is determined that the current feeder has contact degradation and the first warning result is output. The preset event frequency threshold and the first preset resistance change threshold are obtained based on historical sample statistics to establish a joint judgment benchmark for envelope event frequency and resistance change value; Among them, the preset event frequency threshold is used to determine whether the envelope anomaly meets the warning triggering conditions, and the first preset resistance change threshold is used to determine whether the resistance change exceeds the normal fluctuation range. When both the envelope event frequency and the resistance change value meet the preset limit conditions, it is determined that the contact degradation feature has been extracted; when only the preset event frequency threshold limit condition is met, it is determined that the external magnetic field interference feature has been extracted; when only the resistance change value increases within a continuous preset time window condition is met, it is determined that the resistive degradation feature has been extracted. This embodiment establishes causal constraints by simultaneously satisfying two conditions, thereby effectively distinguishing between external interference and actual degradation; while extracting contact degradation characteristics, it reduces the probability of false alarms and improves the reliability of on-site monitoring results.

[0018] In a preferred embodiment of the present invention, the method further includes: when the envelope event frequency is greater than a preset event frequency threshold and the resistance change value is not greater than a first preset resistance change threshold, determining it as an external magnetic field interference event and not generating an early warning result; If the envelope event frequency is not greater than the preset event frequency threshold and the resistance change value is greater than the second preset resistance change threshold in multiple consecutive preset time windows and the resistance change value in the later preset time window is greater than the resistance change value in the previous preset time window, it is determined that there is resistive degradation in the current feeder, and a second warning result is obtained. If the envelope event frequency is not greater than the preset event frequency threshold and the resistance change value is not greater than the second preset resistance change threshold, the current feeder status is determined to be normal. Among them, the first preset resistance change threshold is greater than the second preset resistance change threshold; the specific statistical calculation rules for the preset event frequency threshold, the first preset resistance change threshold and the second preset resistance change threshold are as follows: extract the time series of envelope event frequency and resistance change value respectively from the historical steady-state operation dataset without anomalies; Calculate the average of the sequence and standard deviation and will The calculation results are used as the benchmark thresholds for the corresponding parameters; for the preset event frequency threshold and the second preset resistance change threshold, their corresponding benchmark thresholds are directly adopted. For the first preset resistance change threshold, multiply its corresponding reference threshold by a preset constant. The result is then obtained, where the preset constant is... To characterize the resistance step change margin coefficient caused by contact degradation, and ; After obtaining the first or second warning result, the method further includes: constructing an exponential degradation model based on a sequence of historical contact resistance estimates consisting of multiple preset time windows and the mapping relationship between time nodes and contact resistance estimates in the sequence; Based on the exponential degradation model and the preset contact resistance failure limit, the predicted failure time is determined, and the difference between the predicted failure time and the current time is taken as the remaining safe operating time. After obtaining the first warning result, the method further includes: when the preset switching conditions are met, performing a light load switching process on the current feeder to obtain the remaining current signal and the remaining voltage signal after the switch; Based on the remaining current signal and voltage signal after switching, the estimated contact resistance during the light load period is recalculated; the peak fluctuation range of the estimated contact resistance during the light load period and the decrease range of the estimated contact resistance compared to the pre-switching value are calculated; the first amplitude threshold is the maximum peak value of the fluctuation of the estimated contact resistance during the preset steady-state time period, and the second amplitude threshold is the lower limit of the preset decrease range of the estimated contact resistance. If the decrease is greater than the second magnitude threshold, the first warning result is marked as a lower warning level, and the sampling window length when performing the fast Fourier transform is increased. Based on the increased sampling window length, the step of resistance estimation based on the residual voltage signal and the residual current signal is performed again. If the peak fluctuation amplitude is greater than the first amplitude threshold and the decrease amplitude is not greater than the second amplitude threshold, maintain the first warning result and record the fluctuation anomaly flag.

[0019] Based on the above implementation, after obtaining the first or second warning result, when the envelope event frequency is greater than the preset event frequency threshold and the resistance change value is not greater than the first preset resistance change threshold, it is determined to be an external magnetic field interference event, and only the current analysis data is recorded without generating a warning. When the envelope event frequency is not greater than the preset event frequency threshold, and the resistance change value is greater than the second preset resistance change threshold and shows an upward trend in multiple consecutive preset time windows, it is determined that the current feeder has resistive degradation and a second warning result is obtained. When the envelope event frequency is not greater than the preset event frequency threshold and the resistance change value is not greater than the second preset resistance change threshold, the current feeder status is determined to be normal; wherein, the first preset resistance change threshold is set to be greater than the second preset resistance change threshold to achieve the separation and identification of contact deterioration and resistive deterioration. Among them, external magnetic field interference events are characterized by the frequency of envelope events exceeding the preset event frequency threshold caused by changes in the environmental magnetic field, and the feeder body resistance does not show continuous changes. The resistivity degradation characterizes the envelope event frequency as not exceeding the preset event frequency threshold, and the estimated resistance value shows an increasing trend within a continuous time window; the normal state characterizes the envelope event frequency and resistance change value as not reaching the preset limit conditions; historical sample statistics are used to generate relevant thresholds to ensure that the judgment conditions are consistent with the on-site sample distribution and reduce the deviation caused by human experience thresholds. This embodiment achieves effective classification and identification of external interference, contact degradation, resistivity degradation and normal state by branch judgment under different conditions, thereby improving the field applicability of monitoring and analysis results; This embodiment improves the accuracy of identifying interference events and resistance degradation characteristics through a classification logic model; After the first or second warning result is output, a time-variable structure is constructed based on the contact resistance estimation value sequence. For independent variables, the estimated value of contact resistance Exponential degradation model with dependent variable: in, These are the initial contact resistance fitting coefficients, in units of... ; The constant characterizing the degradation rate has the following dimensions: Due to the exponential term It must be a dimensionless pure numerical value, and the time variable... The dimensions are Therefore, the dimension of this constant must be . To offset the dimension of the time variable; using a sequence of historical contact resistance estimates consisting of multiple preset time windows, downsampling and reconstruction of the historical sequence is performed based on preset long-period time nodes, with the preset long period as the extraction node; Using this time point as the independent variable and the representative data as the dimensionless numerical value of the estimated contact resistance, the natural logarithm of both sides of the resistance equation is taken to transform it into a linear mapping relationship. The calculation formula is as follows: The intercept term is calculated using historical time window series data and the least squares method. and slope term The parameters of the exponential degradation model are obtained; the preset contact resistance failure limit is set. Substituting the target value into the transformed linear model, the predicted failure time is calculated using the following formula: And the predicted failure time The difference between the current time and the current time is taken as the remaining safe operating time; Among them, the contact resistance estimation value sequence is used to reflect the resistance change trajectory of the same feeder in multiple historical time windows; the exponential degradation model is used to describe the trend of resistance gradually increasing over time. The least squares method is used for accurate fitting of curve parameters based on historical sample data; preset contact resistance failure limits. The specific determination logic is as follows: extract the reference value of steady-state contact resistance of the target feeder in the initial stage of commissioning. ,unit: and the benchmark value Multiply by the preset degradation factor tolerance The preset contact resistance failure limit is obtained. : in, The value ranges from 1.5 to 2.0; this limit is used to determine whether the feeder has reached the threshold of being unusable; the remaining safe operating time is the length of time that can continue to operate before expected failure; This embodiment fits the degradation trend through historical sequence fitting, extends the current warning to subsequent risk assessment, and realizes the output of warning signals for the subsequent operating status of the feeder; This embodiment can further provide a failure time estimate after the early warning, thereby improving the reliability of the failure time prediction feature. After obtaining the first early warning result, the station low-voltage AC power supply system of this embodiment also includes a feeder switching control loop. The feeder switching control loop is used to execute the load switching operation of the feeder in response to the control command. When the preset switching conditions are met, the feeder switching control loop performs a light load switching process on the current feeder to obtain the remaining current signal and the remaining voltage signal after the switching. Based on the remaining current signal and voltage signal after switching, the estimated contact resistance during the light load period is recalculated; the peak fluctuation range of the estimated contact resistance during the light load period and the decrease range of the estimated contact resistance compared to the pre-switching value are calculated; the first amplitude threshold is the maximum peak value of the fluctuation of the estimated contact resistance during the preset steady-state time period, and the second amplitude threshold is the lower limit of the preset decrease range of the estimated contact resistance. If the decrease is greater than the second magnitude threshold, the first warning result is downgraded, and the sampling window length when performing the fast Fourier transform is increased. Based on the increased sampling window length, the resistance estimation processing step is performed again. If the peak fluctuation amplitude is greater than the first amplitude threshold and the decrease amplitude is not greater than the second amplitude threshold, maintain the first warning result and record the fluctuation anomaly flag. Among them, the preset switching conditions are used to define the triggering scenario of light load verification. It is triggered when the system load meets the preset light load conditions and responds to the short-time switching command. The specific judgment conditions are: the effective value of the feeder current monitored in real time is less than 20% of the rated current and there are no sensitive loads at the back end that do not allow short-time switching interruption. Light load switching is used to observe whether the estimated contact resistance remains relatively stable after the load is reduced. After light load switching is performed, the system will set a preset transient shielding delay. After the waveform reaches a steady state again, the steady state interval data will be extracted and recalculated to prevent distortion of the fundamental frequency extraction caused by sudden current changes or line oscillations during switching. Peak fluctuation amplitude is used to characterize the fluctuation range of the estimated resistance value during light load; decrease amplitude is used to characterize the degree of change of the estimated resistance value after light load compared with before switching. The first and second amplitude thresholds are used to distinguish between normal light load fluctuations and abnormal drops. The specific value selection rule is as follows: extract the maximum peak value of impedance fluctuation within a preset time period during the initial stage of normal operation as the first amplitude threshold. The minimum lower limit of impedance drop in historically confirmed contact degradation cases is extracted as the second amplitude threshold. Increasing the sampling window length improves the stability of fundamental frequency extraction when resistance estimation is significantly affected by load harmonics. The corresponding quantization adjustment step size rule is as follows: The original sampling window length is extended by multiples or increased by a preset number of power frequency cycles. The extended time series data is used as new input parameters to re-execute the calculation operation, so as to reduce the spectral leakage distortion caused by interharmonic interference. This embodiment verifies the stability of the first warning result through light load switching to eliminate the resistance estimation deviation introduced by short-term load changes; for cases where the decrease is greater than the second amplitude threshold, the sampling window length is increased to re-estimate the resistance and reduce the impact of interference on the fundamental frequency extraction. This embodiment adds an on-site verification after the warning, which improves the accuracy of the warning results and reduces the output of false alarm data; This method is deployed in the online monitoring terminal of the station's low-voltage AC power supply system; it synchronously collects three sensor signals within a preset sampling period, and inputs the synchronously collected signals into the adaptive noise cancellation module for filtering, and then outputs them to the envelope detection module and the resistance estimation module. When contact degradation, resistivity degradation, or external magnetic field interference events are detected, the system generates different types of analysis records or early warning results. It can also write the contact resistance estimation sequence, envelope event frequency, and resistance change value into the historical sample library for threshold updates, trend fitting, and light load verification. Input data flows continuously based on the same processing link to generate early warning, record, and verification results.

[0020] The foregoing detailed an embodiment of the present invention, but this content is merely a specific embodiment of the present invention and should not be considered as limiting the scope of the present invention. All equivalent variations and improvements made within the scope of the present invention should still fall within the patent coverage of the present invention.

Claims

1. An industrial data analysis and monitoring method for station low-voltage AC power supply systems, characterized in that, include: Within a preset sampling period, voltage signals, current signals, and ambient magnetic field change signals are synchronously acquired by voltage sensors, current sensors, and ambient magnetic field sensors installed at the feeder outlet. Based on the environmental magnetic field change signal, adaptive noise cancellation is performed on the voltage signal and the current signal respectively to obtain the voltage residual signal and the current residual signal; The remaining voltage signal is subjected to envelope detection processing to determine the envelope event frequency; Based on the residual voltage signal and the residual current signal, resistance estimation processing is performed to determine the resistance change value; If the envelope event frequency is greater than a preset event frequency threshold and the resistance change value is greater than a first preset resistance change threshold, it is determined that there is contact degradation in the current feeder, and a first warning result is obtained.

2. The industrial data analysis and monitoring method for station low-voltage AC power supply system according to claim 1, characterized in that, Based on the environmental magnetic field change signal, adaptive noise cancellation is performed on the voltage signal and the current signal respectively to obtain the residual voltage signal and the residual current signal, including: A voltage channel adaptive filter for processing the voltage signal and a current channel adaptive filter for processing the current signal are constructed, both of which have weight vectors; For the voltage channel adaptive filter, the voltage signal is used as the target input signal, the environmental magnetic field change signal is used as the reference input signal, the filter output value is determined based on the reference input signal and the weight vector of the voltage channel adaptive filter, the residual voltage signal is obtained, and the weight vector of the voltage channel adaptive filter is updated. For the current channel adaptive filter, the current signal is used as the target input signal, the ambient magnetic field change signal is used as the reference input signal, the filter output value is determined based on the reference input signal and the weight vector of the current channel adaptive filter, the residual current signal is obtained, and the weight vector of the current channel adaptive filter is updated.

3. The industrial data analysis and monitoring method for station low-voltage AC power supply systems according to claim 2, characterized in that, After updating the weight vector of the voltage channel adaptive filter, the following is also included: Within a preset sliding time window, the sliding variance is determined for the historical update values ​​of the weight vector of the voltage channel adaptive filter in the time series, and the increase in the difference of the sliding variance within adjacent consecutive sliding time windows is determined. If the increase in the difference exceeds a preset increment threshold, the weight vector at the end of the previous sliding time window remains unchanged, and invalid data markers are added to the remaining voltage signal in the current sliding time window. If the increase in the difference of the sliding variance within adjacent consecutive sliding time windows is not greater than a preset increment threshold, the updated weight vector is used for subsequent processing.

4. The industrial data analysis and monitoring method for station low-voltage AC power supply system according to claim 3, characterized in that, The remaining voltage signal is subjected to envelope detection processing to determine the envelope event frequency, including: The power frequency and the frequency band containing the power frequency characteristic harmonics in the residual voltage signal are subjected to bandpass filtering to obtain a bandpass voltage signal; The amplitude envelope is extracted based on the bandpass voltage signal to obtain the envelope signal; Within multiple consecutive preset time windows, a threshold value is determined based on the root mean square value and standard deviation of the envelope signal; The number of independent out-of-limit events in the envelope signal whose amplitude is greater than the threshold value and whose time interval is greater than a preset time interval is counted. Based on the number of independent out-of-limit events and the length of the preset time window, the envelope event frequency is determined.

5. The industrial data analysis and monitoring method for a station low-voltage AC power supply system according to claim 4, characterized in that, Based on the residual voltage signal and the residual current signal, resistance estimation processing is performed to determine the resistance change value, including: Based on the residual voltage signal and the residual current signal, a fast Fourier transform is used to extract the fundamental voltage RMS value and the fundamental current RMS value within the current preset time window and the adjacent previous preset time window in the continuous preset time window. After performing phase compensation on the fundamental effective value of the voltage and the fundamental effective value of the current, the estimated value of the contact resistance in the current preset time window is determined; based on the estimated value of the contact resistance in the current preset time window and the estimated value of the contact resistance in the previous preset time window, the change value of the resistance is determined.

6. The industrial data analysis and monitoring method for a station low-voltage AC power supply system according to claim 5, characterized in that, Also includes: If the frequency of the envelope event is greater than the preset event frequency threshold and the resistance change value is not greater than the first preset resistance change threshold, it is determined to be an external magnetic field interference event, and no warning result is generated. If the frequency of the envelope event is not greater than the preset event frequency threshold, and the resistance change value is greater than the second preset resistance change threshold in multiple consecutive preset time windows, and the resistance change value in the later preset time window is greater than the resistance change value in the previous preset time window, it is determined that the current feeder has resistive degradation, and a second warning result is obtained. If the envelope event frequency is not greater than the preset event frequency threshold and the resistance change value is not greater than the second preset resistance change threshold, the current feeder status is determined to be normal. Wherein, the first preset resistance change threshold is greater than the second preset resistance change threshold.

7. The industrial data analysis and monitoring method for station low-voltage AC power supply systems according to claim 6, characterized in that, After obtaining the first warning result or the second warning result, the process further includes: Based on the historical contact resistance estimation sequence consisting of multiple preset time windows, and the mapping relationship between time nodes and contact resistance estimation values ​​in the historical contact resistance estimation sequence, an exponential degradation model is constructed. Based on the exponential degradation model and the preset contact resistance failure limit, the predicted failure time is determined, and the difference between the predicted failure time and the current time is taken as the remaining safe operating time.

8. The industrial data analysis and monitoring method for a station low-voltage AC power supply system according to claim 6, characterized in that, After receiving the first warning result, the following is also included: When the preset switching conditions are met, the current feeder is subjected to light load switching processing to obtain the remaining current signal and the remaining voltage signal after switching. Based on the remaining current signal and the remaining voltage signal after the switch, the estimated contact resistance during the light load period is recalculated. Determine the peak fluctuation range of the estimated contact resistance during the light load period, and the decrease range of the estimated contact resistance compared to the value before switching; and take the maximum peak value of the fluctuation of the estimated contact resistance within a preset steady-state time period as the first amplitude threshold, and take the lower limit of the preset decrease range of the estimated contact resistance as the second amplitude threshold. If the decrease is greater than the second magnitude threshold, the first warning result is marked as a warning level reduction, and the sampling window length when performing the fast Fourier transform is increased. Based on the increased sampling window length, the step of resistance estimation based on the residual voltage signal and the residual current signal is performed again. If the peak fluctuation amplitude is greater than the first amplitude threshold and the decrease amplitude is not greater than the second amplitude threshold, the first warning result is maintained and the fluctuation anomaly flag is recorded.

Images