IoT-based remote monitoring method and system for power surge protectors

Abstract

This invention discloses a remote monitoring method and system for power surge protectors based on the Internet of Things (IoT), specifically relating to the field of overvoltage protection circuit technology. It addresses the problem of missed detection of latent failures caused by the inability of existing remote monitoring systems to assess the residual voltage characteristics of surge protectors online. The method involves real-time acquisition of high-frequency current waveforms via an IoT monitoring terminal; triggering surge events and extracting waveform segments based on the phase deviation between the fundamental and third harmonic waves and the coherence of the oscillation envelope; separating the rising edge of the waveform from the oscillation decay stage, and constructing an energy absorption efficiency factor based on the ratio of their time constants; identifying the dominant oscillation frequency shift and correcting the efficiency factor through attenuation mode spectrum; calculating the energy transfer imbalance by combining upstream waveform empirical mode decomposition; and generating an alarm signal when the corrected efficiency factor continuously falls below a critical value or the imbalance exceeds a threshold. This achieves online assessment of the residual voltage characteristics of the surge protector, eliminating overvoltage protection blind spots caused by performance degradation.

Description

Technical Field

[0001] This invention relates to the field of overvoltage protection circuit technology, and in particular to a method and system for remote monitoring of power surge protectors based on the Internet of Things. Background Technology

[0002] In the field of industrial power protection, surge protectors (SPDs) are key devices for protecting against lightning strikes and operational overvoltages. Currently, IoT-based remote monitoring systems for SPDs can achieve real-time monitoring of leakage current, temperature, and switch status, and perform degradation trend analysis through cloud platforms. These systems primarily rely on electrical parameter thresholds to identify explicit faults (such as short circuits or open circuits) and predict remaining lifespan based on long-term data accumulation. Existing technologies generally compare monitoring data with preset thresholds to generate early warnings, while some solutions combine historical data to train degradation models to improve prediction accuracy.

[0003] However, existing technologies lack online assessment of the residual voltage characteristics of SPDs, leading to missed detection of latent failure risks. When residual voltage rises (i.e., protection performance degrades), the SPD still maintains a normal conducting state, and conventional parameters such as leakage current do not exceed thresholds. Existing monitoring methods cannot capture such failure modes, causing maintenance personnel to misjudge the health status of the SPD, exposing the protected equipment to overvoltage risks. The root cause of this problem is that IoT monitoring systems can only obtain the electrical operating parameters of the SPD, while residual voltage detection requires measurement under high-voltage impact. Traditional methods must be offline testing with power off, which contradicts the fundamental goal of online monitoring. This deficiency creates a significant safety blind spot in existing remote monitoring systems and cannot meet the real-time verification requirements of SPD protection performance in high-reliability scenarios. Summary of the Invention

[0004] This invention addresses the technical problems existing in the prior art by providing a method and system for remote monitoring of power surge protectors based on the Internet of Things.

[0005] The technical solution of the present invention to solve the above-mentioned technical problems is as follows:

[0006] This invention provides the following technical solution:

[0007] The IoT-based remote monitoring method for power surge protectors includes:

[0008] S1. Real-time acquisition of high-frequency current waveforms of power surge protectors via IoT monitoring terminals;

[0009] S2. When the phase deviation between the fundamental wave and the third harmonic in the high-frequency current waveform is less than the critical angle and the coherence coefficient of the oscillation envelope at the preset discrete frequency point is greater than the coherence threshold, the surge event is triggered and the complete current waveform segment is extracted.

[0010] S3. Separate the rising edge stage and the oscillation decay stage from the complete current waveform segment, calculate the rising edge time width and the oscillation decay time constant, and calculate the ratio of the two as the energy absorption efficiency factor.

[0011] S4. The extreme point interval sequence of the high-frequency current waveform during the statistical oscillation decay stage generates the decay mode spectrum. The energy absorption efficiency factor is corrected by identifying whether the dominant oscillation frequency has shifted to the preset critical frequency band.

[0012] S5. Obtain the high-frequency current waveform of the upstream power surge protector of the same power distribution branch during the oscillation decay stage of the same surge event, perform empirical mode decomposition on the upstream and downstream waveforms respectively, extract the energy distribution probability density of the intrinsic mode function, and calculate the KL divergence between the two as the energy transfer imbalance.

[0013] S6. When the corrected energy absorption efficiency factor is continuously lower than the critical value or the energy transfer imbalance exceeds the preset transfer threshold, an alarm signal is generated.

[0014] Furthermore, the high-frequency current waveform of the power surge protector is collected in real time through an IoT monitoring terminal, including:

[0015] The high-frequency current signal of the power surge protector grounding wire is collected by the high-frequency current sensor of the Internet of Things monitoring terminal.

[0016] Perform anti-aliasing filtering on high-frequency current signals;

[0017] The filtered high-frequency current signal is converted into a digital waveform using an analog-to-digital converter.

[0018] The digital waveform is uploaded to the edge computing node via the Internet of Things (IoT) communication module;

[0019] The digital waveform is normalized at the edge computing node to generate a high-frequency current waveform.

[0020] Furthermore, when the phase deviation between the fundamental wave and the third harmonic in the high-frequency current waveform is less than the critical angle and the coherence coefficient of the oscillation envelope at the preset discrete frequency point is greater than the coherence threshold, a surge event marker is triggered and a complete current waveform segment is extracted, including:

[0021] The fundamental phase angle is obtained by extracting the fundamental component of the high-frequency current waveform;

[0022] The third harmonic phase angle is obtained by extracting the third harmonic component from the high-frequency current waveform.

[0023] The difference between the fundamental phase angle and the third harmonic phase angle is calculated as the phase deviation;

[0024] Bandpass filtering is performed on the high-frequency current waveform at preset discrete frequency points to obtain the oscillation component signal at each frequency point;

[0025] Envelope detection is performed on each oscillation component signal to generate an envelope signal sequence;

[0026] The cross-correlation coefficients between the envelope signal sequences are calculated as the oscillating envelope coherence coefficients;

[0027] The start time of a surge event is marked when the phase deviation is less than the critical angle and the oscillation envelope coherence coefficient is greater than the coherence threshold.

[0028] The current waveform data of a preset time length is extracted from the starting moment as a complete current waveform segment.

[0029] Furthermore, the rising edge phase and the oscillation decay phase are separated from the complete current waveform segment. The rising edge time width and the oscillation decay time constant are calculated, and their ratio is calculated as the energy absorption efficiency factor, including:

[0030] The moment when the current amplitude first reaches its peak value in a complete current waveform segment is taken as the end point of the rising edge.

[0031] The waveform segment before the end of the rising edge is divided into the rising edge phase;

[0032] The waveform segment after the end of the rising edge is divided into the oscillation decay stage;

[0033] The time difference between the start of the rising edge phase and the end of the rising edge phase is calculated as the rising edge time width.

[0034] The current waveform during the oscillation decay phase is fitted with an exponential decay curve.

[0035] Extract the decay time constant of the exponential decay curve as the oscillation decay time constant;

[0036] The energy absorption efficiency factor is generated by calculating the ratio of the rise time width to the oscillation decay time constant.

[0037] Furthermore, the extreme point interval sequence of the high-frequency current waveform during the statistical oscillation decay phase generates a decay mode spectrum. The energy absorption efficiency factor is corrected by identifying whether the dominant oscillation frequency has shifted to a preset critical frequency band, including:

[0038] Local maxima and minima are detected in the high-frequency current waveform during the oscillation decay phase.

[0039] Record the time intervals between adjacent maxima and minima to form an extreme point interval sequence;

[0040] Convert the extreme point interval sequence into an instantaneous frequency sequence;

[0041] The attenuation mode spectrum is generated by statistically analyzing the distribution histogram of the instantaneous frequency sequence;

[0042] The frequency with the highest probability density in the decay mode spectrum is identified as the dominant oscillation frequency.

[0043] Determine whether the dominant oscillation frequency is within the preset critical frequency band;

[0044] When the dominant oscillation frequency is within the preset critical frequency band, the energy absorption efficiency factor is reduced according to the preset correction coefficient.

[0045] Furthermore, the high-frequency current waveform of the upstream power surge protector on the same distribution branch during the oscillation decay phase of the same surge event is obtained. Empirical mode decomposition is performed on the upstream and downstream waveforms respectively, and the energy distribution probability density of the intrinsic mode function is extracted. The KL divergence between the two is calculated as the energy transfer imbalance, including:

[0046] The high-frequency current waveform of the upstream power surge protector during the oscillation decay phase under the same surge event identifier is obtained through the Internet of Things data bus.

[0047] Empirical mode decomposition is performed on the high-frequency current waveform during the oscillation decay phase of the current power surge protector to obtain the set of inherent mode functions of the current device.

[0048] Empirical mode decomposition is performed on the high-frequency current waveform during the oscillation decay phase of the upstream power surge protector to obtain the set of inherent mode functions of the upstream equipment;

[0049] Calculate the energy value of each intrinsic mode function in the current device intrinsic mode function set;

[0050] Calculate the energy value of each intrinsic mode function in the intrinsic mode function set of the upstream equipment;

[0051] Generate the energy distribution probability density function of the current device based on the energy values ​​of all intrinsic mode functions of the current device;

[0052] Generate the energy distribution probability density function of the upstream equipment based on the energy values ​​of all intrinsic mode functions of the upstream equipment;

[0053] The KL divergence between the current energy distribution probability density function and the upstream energy distribution probability density function is calculated as the energy transfer imbalance.

[0054] Furthermore, generating the current device energy distribution probability density function includes:

[0055] Calculate the energy value of each intrinsic mode function in the current device intrinsic mode function set. The energy value is the square integral of the amplitude of the corresponding intrinsic mode function during the entire oscillation decay phase.

[0056] The energy values ​​of all intrinsic mode functions are normalized so that the sum of all energy values ​​equals one.

[0057] The normalized energy value is used as the probability value of the corresponding intrinsic mode function to generate the current device energy distribution probability density function.

[0058] Furthermore, calculating the KL divergence as the degree of energy transfer imbalance includes:

[0059] Take the natural logarithm of each probability value of the current device's energy distribution probability density function;

[0060] Calculate the difference between the natural logarithm of each probability value of the current equipment's energy distribution probability density function and the corresponding probability value of the upstream equipment's energy distribution probability density function;

[0061] The weighted difference is obtained by multiplying each probability value of the current device energy distribution probability density function by the corresponding difference value.

[0062] The summation of all weighted differences yields the KL divergence, which is used as the energy transfer imbalance.

[0063] Furthermore, when the corrected energy absorption efficiency factor continuously falls below a critical value or the energy transfer imbalance exceeds a preset transfer threshold, an alarm signal is generated, including:

[0064] Establish continuous monitoring records for the corrected energy absorption efficiency factor;

[0065] When the corrected energy absorption efficiency factor of three consecutive surge events is lower than the critical value, an equipment-level degradation alarm signal is generated.

[0066] Real-time comparison of the relationship between energy transfer imbalance and preset transfer threshold;

[0067] When the energy transfer imbalance exceeds the preset transfer threshold, a system-level imbalance alarm signal is generated.

[0068] Encapsulate device-level degradation alarm signals or system-level imbalance alarm signals using IoT communication protocols;

[0069] The encapsulated alarm signal is transmitted to the IoT monitoring center.

[0070] On the other hand, the present invention provides a remote monitoring system for power surge protectors based on the Internet of Things, comprising:

[0071] The waveform acquisition module is used to acquire the high-frequency current waveform of the power surge protector in real time through an IoT monitoring terminal.

[0072] The event triggering module is used to trigger surge event marking and extract complete current waveform segments when the phase deviation between the fundamental wave and the third harmonic in the high-frequency current waveform is less than the critical angle and the oscillation envelope coherence coefficient of the preset discrete frequency point is greater than the coherence threshold.

[0073] The efficiency calculation module is used to separate the rising edge stage and the oscillation decay stage from a complete current waveform segment, calculate the rising edge time width and the oscillation decay time constant, and calculate the ratio of the two as the energy absorption efficiency factor.

[0074] The frequency correction module is used to generate a decay mode spectrum by statistically analyzing the extreme point interval sequence of the high-frequency current waveform during the oscillation decay phase, and corrects the energy absorption efficiency factor by identifying whether the dominant oscillation frequency has shifted to the preset critical frequency band.

[0075] The balance analysis module is used to obtain the high-frequency current waveform of the upstream power surge protector of the same power distribution branch during the oscillation decay stage of the same surge event, perform empirical mode decomposition on the upstream and downstream waveforms respectively, extract the energy distribution probability density of the intrinsic mode function, and calculate the KL divergence between the two as the energy transfer imbalance.

[0076] The alarm generation module is used to generate an alarm signal when the corrected energy absorption efficiency factor is continuously lower than the critical value or the energy transfer imbalance exceeds the preset transfer threshold.

[0077] The beneficial effects of this invention are:

[0078] 1. Online evaluation of the residual voltage characteristics of surge protectors (SPDs) was achieved through high-frequency current waveform analysis, solving the problem of missed detection of latent failures. Specifically, based on the event triggering mechanism of the phase deviation between the fundamental and third harmonic waves and the coherence of the oscillation envelope, actual surge events can be accurately captured without applying external high voltage. The synchronously extracted complete current waveform segments fully record the dynamic response characteristics of the SPD. By separating the rising edge and the oscillation decay stage, the energy absorption efficiency factor is constructed using the ratio of the rising edge time width to the oscillation decay time constant, directly quantifying the energy transfer rate of the SPD under surge impact. This parameter is strongly correlated with the residual voltage characteristics, thereby enabling online diagnosis of protection performance degradation in IoT monitoring.

[0079] 2. A multi-dimensional degradation assessment system was established: By identifying the dominant oscillation frequency shift through attenuation mode spectrum, the frequency domain characteristic changes caused by the degradation of the microstructure of metal oxide rheostats can be detected; the energy transfer imbalance calculation based on the empirical mode decomposition of upstream and downstream waveforms reveals the abnormal energy distribution among multi-level SPDs; finally, through a composite alarm mechanism, proactive warnings are given when there is equipment-level degradation or system-level imbalance, breaking through the limitations of traditional monitoring that relies solely on static parameters, enabling maintenance personnel to grasp the core protection performance of SPDs in real time and eliminate overvoltage protection blind spots caused by residual voltage rise. Attached Figure Description

[0080] Figure 1 This is a flowchart of the remote monitoring method for power surge protectors based on the Internet of Things according to the present invention.

[0081] Figure 2 This is a schematic diagram of the remote monitoring system for power surge protectors based on the Internet of Things (IoT) of the present invention. Detailed Implementation

[0082] 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.

[0083] Example 1: Figure 1 The present invention provides a remote monitoring method for power surge protectors based on the Internet of Things, comprising:

[0084] S1. Real-time acquisition of high-frequency current waveforms of power surge protectors via IoT monitoring terminals;

[0085] S2. When the phase deviation between the fundamental wave and the third harmonic in the high-frequency current waveform is less than the critical angle and the coherence coefficient of the oscillation envelope at the preset discrete frequency point is greater than the coherence threshold, the surge event is triggered and the complete current waveform segment is extracted.

[0086] S3. Separate the rising edge stage and the oscillation decay stage from the complete current waveform segment, calculate the rising edge time width and the oscillation decay time constant, and calculate the ratio of the two as the energy absorption efficiency factor.

[0087] S4. The extreme point interval sequence of the high-frequency current waveform during the statistical oscillation decay stage generates the decay mode spectrum. The energy absorption efficiency factor is corrected by identifying whether the dominant oscillation frequency has shifted to the preset critical frequency band.

[0088] S5. Obtain the high-frequency current waveform of the upstream power surge protector of the same power distribution branch during the oscillation decay stage of the same surge event, perform empirical mode decomposition on the upstream and downstream waveforms respectively, extract the energy distribution probability density of the intrinsic mode function, and calculate the KL divergence between the two as the energy transfer imbalance.

[0089] S6. When the corrected energy absorption efficiency factor is continuously lower than the critical value or the energy transfer imbalance exceeds the preset transfer threshold, an alarm signal is generated.

[0090] S1. The high-frequency current waveform of the power surge protector is collected in real time through an IoT monitoring terminal. The specific implementation is as follows:

[0091] The high-frequency current signal from the grounding wire of the power surge protector is acquired via a high-frequency current sensor on an IoT monitoring terminal. The high-frequency current sensor uses an open-loop current transformer, for example, with a frequency response range of 1 kHz to 1 MHz and a measurement range covering, for example, ±2000 amp peak current. During installation, it is directly clamped to the surface of the grounding copper busbar of the power surge protector, acquiring the current signal through electromagnetic induction. The sensor output is connected to a signal conditioning circuit with overvoltage protection to prevent transient high voltage from damaging the acquisition circuit.

[0092] Anti-aliasing filtering is applied to high-frequency current signals. Multi-stage active filters are used, such as Butterworth filters with a passband cutoff frequency of 200 kHz, a stopband start frequency of 300 kHz, and an attenuation slope of at least 40 dB per decibel. The filtering process is performed locally at the monitoring terminal, eliminating radio frequency interference above the Nyquist frequency, such as suppressing radio interference signals with frequencies exceeding 300 kHz.

[0093] The filtered high-frequency current signal is converted into a digital waveform using an analog-to-digital converter (ADC). The ADC employs 16-bit resolution, a sampling rate of 1 million samples per second, and an input voltage range configured as, for example, ±5 volts. The conversion process comprises two stages: sample-and-hold and quantization. The sample-and-hold stage captures instantaneous voltage values ​​at fixed time intervals, while the quantization stage maps the voltage values ​​to integer values ​​between 0 and 65535. The converted digital waveform is stored as a two-dimensional array. The first column of the array is a timestamp with microsecond-level precision, and the second column contains the quantized current value in milliamperes (mA).

[0094] Digital waveforms are uploaded to edge computing nodes via an IoT communication module. The IoT communication module employs a low-power wide-area network (LPWAN) wireless transmission protocol, such as a 470 MHz communication band. Before transmission, the digital waveform is packetized into fixed-length packets, each with a device identifier and checksum added. The data packet transmission interval does not exceed 50 milliseconds to ensure waveform continuity. The edge computing node, a gateway device deployed in the power distribution room, receives data packets from designated monitoring terminals through a device identifier binding mechanism and performs data reassembly and integrity verification. If verification fails, a data retransmission request is triggered.

[0095] High-frequency current waveforms are generated by normalizing digital waveforms at edge computing nodes. The normalization process includes the following operations: First, a reference current segment lasting one power frequency cycle (20 milliseconds) is extracted from the digital waveform; the arithmetic mean of the current amplitude in this segment is calculated as the reference amplitude, excluding outliers exceeding ±3 standard deviations of the average amplitude; then, each quantized current value in the complete digital waveform is divided by this reference amplitude to obtain the normalized amplitude; finally, the generated high-frequency current waveform consists of a timestamp sequence and a normalized amplitude sequence, with the timestamp unit converted to milliseconds and the normalized amplitude value being dimensionless. This processing eliminates the influence of differences in sensor sensitivity and line impedance fluctuations on the amplitude.

[0096] The selection of a high-frequency current sensor must meet the requirements of high-frequency signal acquisition. For example, the measurement range must cover the maximum discharge current value of the surge protector, and the upper limit of the frequency response must be higher than the expected surge oscillation frequency. Anti-aliasing filter parameters are set according to the sampling theorem; for example, when the sampling rate is 1 million samples per second, the passband cutoff frequency should not exceed 500 kHz. The analog-to-digital conversion resolution must ensure current measurement accuracy; for example, a 16-bit resolution can provide approximately 0.06 mA resolution within a ±2000 ampere range. The normalized reference range is selected from the normal power frequency cycle without surge interference, and spectral analysis verifies the absence of high-frequency components; for example, a fast Fourier transform shows no frequency components significantly higher than 1 kHz. The output high-frequency current waveform serves as the basis for subsequent processing steps, containing both accurate time series and standardized amplitude series.

[0097] S2. When the phase deviation between the fundamental wave and the third harmonic in the high-frequency current waveform is less than the critical angle and the coherence coefficient of the oscillation envelope at the preset discrete frequency point is greater than the coherence threshold, a surge event marker is triggered and a complete current waveform segment is extracted. Specifically, the implementation is as follows:

[0098] The fundamental wave phase angle is obtained by extracting the fundamental component from a high-frequency current waveform. This extraction is achieved using a finite-length unit impulse response digital filter, such as a bandpass filter with a center frequency of 50 Hz and a bandwidth of ±5 Hz. The high-frequency current waveform is input into this filter, and the output signal is the fundamental wave component signal. The fundamental wave phase angle is calculated using a Hilbert transform: first, the fundamental wave component signal is subjected to a Hilbert transform to generate orthogonal signals; then, the phase angle sequence is calculated from the complex signal formed by the original fundamental wave component signal and the orthogonal signals; finally, the value of this phase angle sequence at a specific time point is taken as the fundamental wave phase angle, for example, the center time of the current analysis window.

[0099] The third harmonic phase angle is obtained by extracting the third harmonic component from the high-frequency current waveform. The third harmonic component extraction uses the same filtering method as the fundamental component extraction, but the filter's center frequency is adjusted to 150 Hz; for example, a bandpass filter with a passband center frequency of 150 Hz and a bandwidth of ±15 Hz is designed. The calculation method for the third harmonic phase angle is the same as that for the fundamental phase angle: after processing the third harmonic component signal using Hilbert transform, the phase angle sequence is obtained, and the values ​​at the same specific time points are taken as the third harmonic phase angle.

[0100] The difference between the fundamental phase angle and the third harmonic phase angle is calculated as the phase deviation. The phase deviation is calculated by directly subtracting the angle differences; for example, when the fundamental phase angle is 45 degrees and the third harmonic phase angle is 60 degrees, the phase deviation is -15 degrees. The calculation result is then standardized to constrain the phase deviation to a range of -180 degrees to +180 degrees. For example, when the calculated difference is 200 degrees, it is converted to -160 degrees.

[0101] The high-frequency current waveform is bandpass filtered at preset discrete frequency points to obtain the oscillation component signals at each frequency. The preset discrete frequency points are set based on typical surge spectrum characteristics, for example, 100 kHz, 300 kHz, and 500 kHz. The bandpass filter for each frequency point is designed independently; for example, the bandwidth of the 100 kHz filter is set to ±10 kHz, and the bandwidth of the 300 kHz filter is set to ±30 kHz. The filtering process employs zero-phase digital filtering technology, i.e., forward filtering followed by reverse filtering, to avoid introducing additional phase shifts. The oscillation component signals at each frequency point are stored in time series format, with the sequence length consistent with the original high-frequency current waveform.

[0102] Envelope detection is performed on each oscillating component signal to generate an envelope signal sequence. Envelope detection is implemented using the Hilbert transform method: first, the oscillating component signal is subjected to a Hilbert transform to generate an analytic signal; then, the magnitude of the analytic signal is calculated as the instantaneous envelope amplitude. All instantaneous envelope amplitudes are arranged in chronological order to form the envelope signal sequence. The envelope signal sequence at each frequency point is normalized to ensure its maximum amplitude is 1, facilitating subsequent comparisons.

[0103] The cross-correlation coefficient between the envelope signal sequences is calculated as the oscillating envelope coherence coefficient. The cross-correlation coefficient is calculated using the sliding window covariance method: a fixed-length time window is set, for example, 1 millisecond; the covariance of the two envelope signal sequences is calculated within the window; the standard deviation of each sequence is calculated; the covariance is divided by the product of the two standard deviations to obtain the correlation coefficient for that window; after the sliding window traverses the entire time range, the average of all window correlation coefficients is taken as the oscillating envelope coherence coefficient. For multiple frequency points, the pairwise coherence coefficients of all frequency point combinations are calculated and their arithmetic mean is taken.

[0104] The surge event start time is marked when the phase deviation is less than the critical angle and the oscillation envelope coherence coefficient is greater than the coherence threshold. The critical angle is set based on the phase synchronization characteristics of the surge event, for example, 30 degrees; the coherence threshold is set based on the background noise level, for example, 0.85. The judgment logic is as follows: if the phase deviation is less than the critical angle (e.g., within the range of -30 degrees to +30 degrees) and the oscillation envelope coherence coefficient is greater than the coherence threshold (e.g., greater than 0.85), then the specific time point is marked as the surge event start time. If multiple time points satisfy the conditions simultaneously, the time with the smallest phase deviation is selected as the start time.

[0105] The system extracts a segment of current waveform data for a preset time length, centered on the start time, to form a complete current waveform segment. The preset time length covers the duration of a typical surge process, such as 2 milliseconds from 0.5 milliseconds before to 1.5 milliseconds after the start time. The extraction operation preserves the complete data structure of the original high-frequency current waveform, including the timestamp sequence and the normalized amplitude sequence. An event identifier and time range label are added to the extracted data segment to form the complete current waveform segment output.

[0106] The critical angle setting needs to consider the impact of power grid frequency fluctuations; for example, maintaining a constant critical angle of 30 degrees within a frequency deviation range of ±0.5 Hz. The coherence threshold setting needs to be determined through background noise statistical analysis. Specifically, this involves collecting high-frequency current waveform data during surge-free periods, calculating the envelope coherence coefficient distribution using the same method, and taking the 95th percentile of the distribution as the threshold. Preset discrete frequency points are optimized based on the actual surge spectrum characteristics; for example, selecting energy-concentrated frequencies within the 100 kHz to 1 MHz range. The fundamental and harmonic phase angle calculations must use the same analysis time window, such as a rectangular window with a length of 10 milliseconds. Output a complete current waveform segment as input data for subsequent steps, with its time range precisely marking the core period of the surge event.

[0107] S3. Separate the rising edge stage and the oscillation decay stage from the complete current waveform segment, calculate the rising edge time width and the oscillation decay time constant, and calculate the ratio of the two as the energy absorption efficiency factor. The specific implementation is as follows:

[0108] The moment when the current amplitude first reaches its peak value within a complete current waveform segment is designated as the end of the rising edge. The positioning method employs gradient detection: the first-order difference value of each sampling point in the complete current waveform segment is calculated. When a sampling point's difference value is detected to be greater than zero and the difference value of the next sampling point is less than or equal to zero, the corresponding moment of that sampling point is marked as the end of the rising edge. Specifically, this involves: calculating the difference between adjacent sampling points to form a difference sequence from the normalized amplitude sequence of the complete current waveform segment; traversing the difference sequence to detect sign changes; and selecting the earliest candidate point if multiple candidate points exist. A difference threshold is set during the positioning process to exclude small fluctuations, for example, current amplitude changes less than 0.01 are ignored.

[0109] The waveform segment preceding the end of the rising edge is divided into rising edge phases. This division is based on a timestamp sequence: all sampling points in the complete current waveform segment with timestamps less than the rising edge end timestamp are extracted to form new waveform segments. The start time of this waveform segment is the start time of the complete current waveform segment, and the end time is the time corresponding to the rising edge end. The data structure of the rising edge phase includes a timestamp subsequence and a corresponding normalized amplitude subsequence, with the timestamp unit remaining constant in milliseconds. Data integrity must be verified during division, for example, ensuring that the rising edge phase contains at least 10 sampling points.

[0110] The waveform segment following the rising edge is divided into an oscillation decay phase. This division is also based on a timestamp sequence: all sampling points in the complete current waveform segment with timestamps greater than the rising edge's end timestamp are extracted to form another waveform segment. The data structure of the oscillation decay phase is consistent with the rising edge phase, containing an independent timestamp subsequence and a normalized amplitude subsequence. After division, a validity check is required, for example, ensuring that the duration of the oscillation decay phase is at least twice that of the rising edge phase.

[0111] The time difference between the start and end of the rising edge phase is calculated as the rising edge time width. The time difference is calculated using a direct subtraction method: subtract the timestamp value of the start of the rising edge phase from the timestamp value corresponding to the end of the rising edge phase. The time difference value is converted to microseconds; for example, if the start time is 100.50 milliseconds and the end time is 100.78 milliseconds, the time difference is 280 microseconds. The calculation result is stored as a 64-bit floating-point number, achieving microsecond-level precision. The validity of the starting point must be verified before calculation; for example, if the starting current amplitude is greater than 0.01, it is necessary to trace back to the moment it first exceeded 0.01.

[0112] An exponential decay curve was fitted to the current waveform during the oscillation decay phase. The curve fitting was achieved using a nonlinear least squares method: it was assumed that the current amplitude during the oscillation decay phase followed an exponential decay law, i.e., the amplitude decreased exponentially with time. The fitting process included three stages: parameter initialization, iterative optimization, and convergence judgment. First, the initial parameter values ​​were calculated using the first three sampling points of the oscillation decay phase; then, the parameters were adjusted to minimize the sum of squared residuals between the fitted curve and the measured data; convergence was determined when the residual change was less than 0.1% after five consecutive iterations. Data points with amplitudes less than 5% of the peak value were excluded during the fitting process to avoid noise interference. The iterative optimization used a damped least squares method, with a step size control factor set to 0.01 and a maximum number of iterations set to 100.

[0113] The decay time constant of the exponential decay curve is extracted as the oscillation decay time constant. The time constant parameter value is directly read from the fitted exponential decay model; this parameter represents the time required for the current amplitude to decay to 36.8% of its initial value, in microseconds. If fitting fails, a zero-crossing detection method is used for estimation: all local maxima are detected during the oscillation decay phase; the ratio of the amplitude of the next maximum point to the amplitude of the previous maximum point is calculated; and the time constant is obtained through logarithmic calculation. The results need to be validated for reasonableness; for example, a time constant greater than the total duration of the complete waveform segment is considered invalid.

[0114] The energy absorption efficiency factor is generated by calculating the ratio of the rise time width to the oscillation decay time constant. The ratio is calculated using division: the rise time width is divided by the oscillation decay time constant, resulting in a dimensionless value as the energy absorption efficiency factor. For example, when the rise time width is 280 microseconds and the oscillation decay time constant is 420 microseconds, the ratio is 0.6667. The calculation result is preserved to four decimal places and a timestamp is added to associate it with the corresponding surge event. Before calculation, the validity of the parameters must be verified; for example, data should be discarded if the rise time width is less than 1 microsecond or the oscillation decay time constant is less than 10 microseconds.

[0115] The division of the oscillation decay phase must ensure data integrity. For example, at least 100 sampling points should be retained after the rising edge ends; if insufficient, adjacent sampling points should be supplemented through linear interpolation. The initial parameter setting method for fitting the exponential decay curve is as follows: take the amplitude of the first maximum point in the oscillation decay phase as the initial amplitude; take the time difference between the first two adjacent maximum points multiplied by 2 as the initial time constant. When using the zero-crossing detection method to estimate the time constant, take the median of the calculation results for all intervals as the final value.

[0116] The energy absorption efficiency factor (EEPF) physically represents the ratio of energy absorption rate to energy release rate, and in practical applications, it is set to an effective range of 0.5 to 2.0. The output parameters include three sets of data: rise time width, oscillation decay time constant, and EEPF, all associated with surge event identifiers. All calculation results are marked with a quality flag (e.g., 0 for valid, 1 for questionable, 2 for invalid) for subsequent steps. Parameter validation rules include: rise time width must be less than oscillation decay time constant; the oscillation decay phase must contain at least three complete oscillation cycles; negative EEPF values ​​are immediately discarded.

[0117] S4. The extreme point interval sequence of the high-frequency current waveform during the statistical oscillation decay stage generates the decay mode spectrum. The energy absorption efficiency factor is corrected by identifying whether the dominant oscillation frequency has shifted to the preset critical frequency band. The specific implementation is as follows:

[0118] Local maxima and minima are detected in the high-frequency current waveform during the oscillation decay phase. The detection method employs the differential sign change method: the first-order difference value of the normalized amplitude sequence of the high-frequency current waveform during the oscillation decay phase is calculated; when the difference value changes from positive to negative, it is marked as a local maximum point; when the difference value changes from negative to positive, it is marked as a local minimum point. Specifically, this involves traversing all sampling points and comparing the amplitude relationship between the current sampling point and the previous and next sampling points; if the current amplitude is greater than both the previous and next sampling points, it is marked as a maximum point; if the current amplitude is less than both the previous and next sampling points, it is marked as a minimum point. An amplitude change threshold is set during the detection process to eliminate noise interference; for example, amplitude changes between adjacent sampling points less than 0.005 are ignored.

[0119] The time intervals between adjacent maxima and minima are recorded to form an extreme point interval sequence. The time interval is calculated using the timestamp difference method: detected maxima and minima are sorted chronologically; the timestamp difference between each maxima and its next adjacent minima is calculated; all time differences are arranged chronologically to form the extreme point interval sequence. The time interval unit is uniformly microseconds, and the sequence is stored as a floating-point array. The recording process includes validity verification; for example, time intervals shorter than the sampling period are considered invalid and discarded.

[0120] The extreme point interval sequence is converted into an instantaneous frequency sequence. The conversion method uses a reciprocal calculation: the reciprocal of each time interval value in the extreme point interval sequence is taken, and then multiplied by 1,000,000 to convert it into a frequency value in Hertz. Specifically, the instantaneous frequency value equals 1,000,000 divided by the time interval value. The converted sequence is an instantaneous frequency sequence of the same length as the extreme point interval sequence, with frequency values ​​retaining integer precision. The conversion process includes outlier filtering; for example, instantaneous frequency values ​​less than 10 Hertz or greater than 1000 kilohertz are considered invalid data and removed.

[0121] The attenuation mode spectrum is generated by plotting the distribution histogram of the instantaneous frequency sequence. Histogram statistics employ an equal-width binning method: the lower limit of the frequency range is set to 0 Hz, and the upper limit to 1000 kHz; the bin width is set to 10 kHz; the number of data points in each frequency bin is calculated; and the probability density value is obtained by dividing the count of each bin by the total number of valid points. The attenuation mode spectrum data structure includes a sequence of frequency bin center values ​​and the corresponding probability density value sequence. The statistical process includes data smoothing, such as performing a moving average of the probability density values ​​of three adjacent frequency bins.

[0122] The frequency with the highest probability density in the decaying mode spectrum is identified as the dominant oscillation frequency. The identification method employs a maximum value search: traversing the probability density value sequence of the decaying mode spectrum, the index corresponding to the maximum value is found; the corresponding value in the frequency bucket center value sequence is then located based on the index as the dominant oscillation frequency. When multiple identical maximum probability density values ​​exist, the arithmetic mean of the frequency bucket center values ​​is taken as the dominant oscillation frequency. The identification process includes confidence verification; for example, a maximum probability density value less than 0.3 is considered an unreliable result.

[0123] The system determines whether the dominant oscillation frequency falls within a preset critical frequency band. This preset critical frequency band is set based on the aging characteristics of the surge protector, for example, a lower limit of 300 kHz and an upper limit of 500 kHz. The judgment logic is: if the dominant oscillation frequency is greater than or equal to the lower limit and less than or equal to the upper limit, it is determined to be within the critical frequency band. The judgment process includes boundary condition handling; for example, when the dominant oscillation frequency is exactly equal to 300 kHz or 500 kHz, it is considered to be within the range. The critical frequency band range is set based on experimental statistical data of characteristic frequency shifts during the degradation of the surge protector's metal oxide rheostat.

[0124] When the dominant oscillation frequency is within a preset critical frequency band, the energy absorption efficiency factor is reduced according to a preset correction factor. The correction factor is set in stages based on the degree of frequency offset; for example, the factor is 0.9 when the dominant oscillation frequency is between 300 kHz and 400 kHz, and 0.8 when it is between 400 kHz and 500 kHz. The correction operation uses multiplication: the original energy absorption efficiency factor is multiplied by the correction factor to obtain the corrected energy absorption efficiency factor. The correction result retains four decimal places and a correction marker is added. When the dominant oscillation frequency is outside the critical frequency band, the original energy absorption efficiency factor remains unchanged.

[0125] Before extreme point detection, the high-frequency current waveform during the oscillation decay phase is smoothed, for example, by using a moving average filter with a window width of 5 sampling points. The generation of the time interval sequence must ensure correct point pair matching; when two consecutive maxima occur, a minimum point is inserted after the previous maximum for correction. Instantaneous frequency conversion is constrained to an effective range; values ​​exceeding the reasonable range are replaced with the arithmetic mean of the preceding and following effective values.

[0126] The attenuation mode spectrum statistics dynamically adjust the binning parameters based on the actual data distribution. For example, the width is reduced to 5 kHz when the data is concentrated and increased to 20 kHz when it is dispersed. Auxiliary criteria are added to the identification of the dominant oscillation frequency, such as requiring the probability density of the bin at that frequency to be at least 1.2 times that of the adjacent bins. Temperature compensation is considered when setting the critical frequency range; for example, the lower and upper frequency limits each increase by 5 kHz for every 10 degrees Celsius increase in ambient temperature.

[0127] The correction factor is set to correspond to the frequency value; for example, the correction factor is 0.85 when the dominant oscillation frequency is 350 kHz. Before correction, the validity of the energy absorption efficiency factor must be verified; no correction is performed if the original value is less than 0.1 or greater than 10. When outputting the corrected energy absorption efficiency factor, a copy of the original value is retained, and the dominant oscillation frequency and correction factor value are recorded. All intermediate calculation results are associated with surge event identifiers and corresponding timestamps.

[0128] S5. Obtain the high-frequency current waveform of the upstream power surge protector of the same power distribution branch during the oscillation decay stage of the same surge event. Perform empirical mode decomposition on the upstream and downstream waveforms respectively, extract the energy distribution probability density of the intrinsic mode function, and calculate the KL divergence between the two as the energy transfer imbalance. The specific implementation is as follows:

[0129] The high-frequency current waveform of the oscillation decay phase of the upstream power surge protector under the same surge event identifier is acquired through an IoT data bus. The IoT data bus employs a message queue architecture, such as a publish-subscribe communication protocol. The surge event identifier is generated by combining the event trigger time and device location code, for example, a string in the format "20230815103025_NodeA". The acquisition process includes: sending a subscription request containing a specific surge event identifier to the IoT data bus; verifying the event identifier matching after receiving the data packet; extracting the high-frequency current waveform data of the oscillation decay phase from the data packet; and checking data integrity, such as verifying whether the time series is continuous and uninterrupted. The upstream waveform data structure is consistent with the local waveform, containing a timestamp sequence and a normalized amplitude sequence, with the timestamps using a unified time base format.

[0130] Empirical mode decomposition (EMD) is performed on the high-frequency current waveform during the oscillation decay phase of the current power surge protector to obtain the set of inherent mode functions (IMFs) of the current device. EMD is implemented using an iterative sieving method: the remaining signal is initialized as a normalized amplitude sequence of the original high-frequency current waveform during the oscillation decay phase; local maxima and local minima of the remaining signal are detected; all local maxima are connected using an interpolation algorithm to form an upper envelope; all local minima are connected to form a lower envelope; the arithmetic mean of the upper and lower envelopes is calculated as the mean envelope; the mean envelope is subtracted from the remaining signal to obtain candidate mode signals; the above sieving process is repeated until the candidate mode signals satisfy the IMF conditions; the IMF conditions include: the number of extrema and the number of zero-crossings differ by no more than 1, and the average value of the upper and lower envelopes is zero at any given time; the candidate mode signals that meet the conditions are output as IMFs; the IMF is separated from the remaining signal; the entire process is repeated until the remaining signal becomes a monotonic or constant function. The screening stop condition is set to the standard deviation change of less than 0.1% for three consecutive screenings. The final output set of intrinsic mode functions is sorted by frequency components from high to low.

[0131] Empirical mode decomposition (EMD) is performed on the high-frequency current waveform during the oscillation decay phase of the upstream power surge protector to obtain the set of intrinsic mode functions (IMFs) of the upstream equipment. The decomposition process is exactly the same as that of the current equipment: the same screening algorithm steps; the same IMF judgment criteria; and the same screening stop conditions. To ensure the comparability of the decomposition results, the same preprocessing standards are adopted for the upstream and downstream waveforms: when the data lengths are inconsistent, the shorter waveform is symmetrically extended and padded; when the sampling rates are different, resampling is performed to unify the sampling rates. During the decomposition process, the frequency characteristics of each IMF are recorded, for example, the average frequency is calculated using the zero-crossing interval.

[0132] Calculate the energy value of each intrinsic mode function (IMF) in the current device's intrinsic mode function set. The energy value calculation uses the discrete integration method: for each IMF, calculate the square of the amplitude at each sampling point in the time series throughout the oscillation decay phase; sum all the squares; multiply the sum by the sampling time interval to obtain the final energy value. Specifically, the energy value equals the sum of the amplitude squares multiplied by the time difference (microseconds) between adjacent sampling points. The calculation result is preserved in floating-point format, with the unit being microjoules.

[0133] Calculate the energy value of each intrinsic mode function (IMF) in the upstream device's intrinsic mode function set. The calculation method is completely consistent with the current device's energy value calculation method: the same discrete integral formula; the same sampling time interval parameter; the same unit system. Before energy value calculation, mode alignment is performed: calculate the average frequency of the upstream and downstream intrinsic mode functions; pair modes with a frequency difference less than 10%; unpaired modes are considered invalid and do not participate in subsequent calculations.

[0134] The probability density function of the current device's energy distribution is generated based on the energy values ​​of all intrinsic mode functions (IMFs). The generation process involves three steps: first, the arithmetic sum of the energy values ​​of all IMFs is calculated as the total energy; then, the energy value of each IMF is divided by the total energy to obtain a normalized energy value; finally, the normalized energy value is used as the probability value of that mode function, and these values ​​are arranged in descending order of frequency to form a discrete probability distribution function. The probability density function data structure includes a sequence of IMF identifiers and a corresponding sequence of probability values, with the probability values ​​retained to four decimal places. A minimum probability threshold is set during the generation process, for example, 0.0001; probabilities less than this value are considered zero.

[0135] The upstream device's energy distribution probability density function is generated based on the energy values ​​of all intrinsic mode functions of the upstream device. The generation method is completely consistent with that of the current device: the same normalization process; the same probability value calculation rules; and the same data structure. Dimension alignment is performed before generation: when the number of upstream and downstream modes is inconsistent, a zero-probability term is added to the set with fewer modes to make the two probability distributions have the same dimension.

[0136] The KL divergence between the current device's energy distribution probability density function and the upstream device's energy distribution probability density function is calculated as the energy transfer imbalance. The calculation process consists of four steps: First, the natural logarithm of each probability value in the current device's probability density function is taken; then, the difference between this natural logarithm and the natural logarithm of the corresponding probability value of the upstream device is calculated; next, the probability value of the current device is multiplied by this difference to obtain a weighted difference value; finally, all weighted differences are summed to obtain the KL divergence value. Specific implementation includes: strictly matching probability value pairs in the same frequency order; excluding terms with probability values ​​less than 0.001 to avoid numerical instability; and retaining three decimal places for the KL divergence calculation result. A symmetry processing step is added after calculation: the upstream KL divergence to the current device is calculated simultaneously; the arithmetic mean of the KL divergences in both directions is taken as the final energy transfer imbalance.

[0137] The intrinsic mode function screening process has a maximum iteration limit, such as 100 iterations, to prevent infinite loops. Energy value calculation uses an improved trapezoidal integral method to enhance accuracy: adjacent sampling points accumulate and square their values ​​along a trapezoidal surface. The probability density function generation includes an outlier detection mechanism: when the energy value of a mode exceeds 90% of the total energy, a manual review is triggered. The KL divergence calculation has an upper limit, such as 10.0; if exceeded, it is forcibly corrected to 10.0 and an outlier is marked.

[0138] Modal alignment employs an optimized matching algorithm: a frequency similarity matrix is ​​established; the optimal pairing combination is found through an optimal allocation algorithm. A confidence assessment is added to the KL divergence calculation results: when the number of valid probability value pairs is less than 3, the result is marked as low confidence. When outputting energy transfer imbalance, complete metadata is associated: surge event identifier; upstream and downstream device location codes; timestamp information; and data quality indicators.

[0139] The empirical mode decomposition (EMD) screening process incorporates envelope optimization: when overshoot occurs in the envelope, a mirror continuation method is used for smoothing. Energy normalization includes an energy correction step: when the total energy is less than a preset threshold, all energy values ​​are proportionally amplified. KL divergence calculation incorporates smoothing: a moving average is applied to the probability value sequence before calculation. All intermediate calculation results retain their original records, including intermediate signals and energy calculation details during the decomposition process, supporting subsequent quality audits.

[0140] Data preprocessing includes time alignment correction: when there is a time offset between upstream and downstream waveforms, the optimal matching position is found through a cross-correlation algorithm. After intrinsic mode function decomposition, mode filtering is added: effective modes with an energy percentage greater than 1% are retained, while noisy modes are excluded. After probability density function generation, smoothing is added: a three-point moving average is applied to the probability value sequence. After KL divergence calculation, a physical interpretation is added: a value greater than 1 indicates a significant energy imbalance, while a value less than 0.1 indicates a balanced state. The output includes detailed diagnostic information: dominant mode frequency comparison; energy distribution difference analysis; and imbalance level assessment.

[0141] S6. When the corrected energy absorption efficiency factor is continuously lower than the critical value or the energy transfer imbalance exceeds the preset transfer threshold, an alarm signal is generated. The specific implementation is as follows:

[0142] Establish continuous monitoring records for the corrected energy absorption efficiency factor. The monitoring records are implemented using a circular buffer storage structure, with the buffer size set to, for example, the data capacity of the most recent 50 surge events. Record content includes: surge event identifier; the value of the corrected energy absorption efficiency factor; and the corresponding timestamp. Each new record overwrites the oldest record, keeping the buffer full. Record updates are triggered by completing the correction operation in step S4, for example, automatically storing the corrected energy absorption efficiency factor in the buffer when S4 outputs it. The buffer data structure includes an index number field, supporting fast retrieval by time order.

[0143] A device-level degradation alarm signal is generated when the corrected energy absorption efficiency factor (EEPF) of three consecutive surge events is below a critical value. The judgment logic is as follows: retrieve the three latest records from the monitoring record buffer; compare the corrected EEPF with the critical value for each record; if the corrected EEPF of all three records is less than the critical value, an alarm is triggered. The critical value is set based on the device aging characteristics, for example, 0.7. The alarm signal data structure includes: alarm type code; alarm trigger time; a list of identifiers for the three involved surge events; and the corresponding sequence of corrected EEPF values. The generation process includes validity verification, such as requiring the three events to occur within 72 hours.

[0144] The system performs a real-time comparison between the energy transfer imbalance and a preset transfer threshold. The comparison operation is executed immediately after each S5 step calculation: it retrieves the energy transfer imbalance value corresponding to the current surge event; reads the preset transfer threshold parameter; and performs a numerical comparison operation. The preset transfer threshold is set according to system balance requirements, for example, 1.2. The comparison result is stored as a Boolean flag and associated with the current surge event identifier. The comparison process includes data validity checks; for example, data with an energy transfer imbalance less than zero is considered invalid.

[0145] When the energy transfer imbalance exceeds a preset transfer threshold, a system-level imbalance alarm signal is generated. The trigger condition is a true comparison result, meaning the energy transfer imbalance is greater than the preset transfer threshold. The alarm signal data structure includes: alarm type code; alarm trigger time; current surge event identifier; energy transfer imbalance value; and preset transfer threshold value. The generation process includes timeliness controls, such as preventing the same device from generating the same alarm repeatedly within 24 hours. The alarm level is set in stages according to the degree of imbalance; for example, when it exceeds the threshold by 50%, it is upgraded to an emergency alarm.

[0146] Device-level degradation alarm signals or system-level imbalance alarm signals are encapsulated using an IoT communication protocol. The encapsulation process includes: adding a device location encoding field; adding a timestamp field; adding a protocol version number; and adding a data checksum. The IoT communication protocol uses a lightweight binary format, such as a custom type length value structure. The encapsulation process includes data compression, such as using differential encoding for repeated fields. The encapsulated data unit contains a fixed-length header and a variable-length body.

[0147] The encapsulated alarm signal is transmitted to the IoT monitoring center. The transmission employs a reliable mechanism: it is first sent to the edge gateway via a low-power wide-area network; the edge gateway forwards it to the cloud platform via a message queue telemetry protocol; the cloud platform parses the signal and stores it in the monitoring center's database. The transmission process includes a retransmission mechanism: if no acknowledgment is received, it is retransmitted every 30 seconds, up to a maximum of three times. The transmitted content includes complete encapsulated data and time-sensitive markers, requiring end-to-end latency to not exceed 10 seconds.

[0148] The threshold value is set based on historical equipment data analysis, for example, taking the 10th percentile value of the corrected energy absorption efficiency factor during normal operation. The preset transmission threshold is determined through system simulation, for example, testing the energy transmission imbalance distribution under typical load conditions. An alarm generation delay confirmation mechanism is added: when the conditions are met, wait 5 minutes for re-verification to avoid instantaneous interference.

[0149] Device-level degradation alarms now include a severity assessment: calculating the arithmetic mean of three consecutive corrected energy absorption efficiency factors; alarm levels are assigned based on this mean, for example, a mean of 0.6 to 0.7 is a warning level, 0.5 to 0.6 is a severe level, and below 0.5 is a critical level. System-level imbalance alarms now include upstream and downstream device comparison data: recording the codes of upstream and downstream devices. IoT encapsulation protocols define dedicated alarm fields: device-level alarms use a specific type of code; system-level alarms use another specific type of code.

[0150] Transmission failure handling includes local caching: unacknowledged alarms are stored on the monitoring terminal, and retransmission is attempted every 10 minutes. The monitoring center receiver is configured with integrity checks: verifying timestamp continuity; verifying data integrity values; and confirming the legitimacy of the device location. Alarm signals are ultimately stored in a time-series database, retaining both the original binary message and the parsed structure.

[0151] The alarm confirmation mechanism includes a manual intervention interface: maintenance personnel generate feedback instructions after confirming an alarm. Historical alarm records support combined queries: by device location; by time period; by alarm level. Critical value parameters support remote configuration: critical values ​​stored locally on the device are updated via the downlink channel. Preset transmission thresholds support dynamic adjustment: new thresholds are automatically calculated based on system operating status. All alarm events generate statistical reports: weekly summaries of device degradation trends; monthly system balance assessments.

[0152] The critical value update method is as follows: periodically collect corrected energy absorption efficiency factor data from the most recent 100 surge events; calculate the average and standard deviation of the data; and set the critical value as the average minus twice the standard deviation. The preset transmission threshold dynamic adjustment method is as follows: statistically analyze the 95th percentile value of all energy transmission imbalances within the most recent 24 hours; multiply this value by a safety factor to obtain the new threshold. The alarm delay confirmation mechanism includes waveform verification: when an alarm condition is triggered, relevant waveform data is reacquired for verification.

[0153] Equipment location coding adopts a hierarchical structure: power distribution room number plus equipment location number. Timestamps use Coordinated Universal Time (UTC) format with millisecond precision. Data checksums are generated using a cyclic redundancy check (CRC) algorithm, with a checksum length of 16 bits. Alarm signal message bodies use binary encoding: the first two bytes indicate the alarm type; the following four bytes represent the timestamp; and subsequent bytes store a list of event identifiers. The monitoring center database has an alarm deduplication mechanism: only one record of the same type of alarm from the same device is stored within a preset time window.

[0154] Example 2: Figure 2 A schematic diagram of the IoT-based remote monitoring system for power surge protectors is provided. The IoT-based remote monitoring system for power surge protectors includes:

[0155] The waveform acquisition module is used to acquire the high-frequency current waveform of the power surge protector in real time through an IoT monitoring terminal.

[0156] The event triggering module is used to trigger surge event marking and extract complete current waveform segments when the phase deviation between the fundamental wave and the third harmonic in the high-frequency current waveform is less than the critical angle and the oscillation envelope coherence coefficient of the preset discrete frequency point is greater than the coherence threshold.

[0157] The efficiency calculation module is used to separate the rising edge stage and the oscillation decay stage from a complete current waveform segment, calculate the rising edge time width and the oscillation decay time constant, and calculate the ratio of the two as the energy absorption efficiency factor.

[0158] The frequency correction module is used to generate a decay mode spectrum by statistically analyzing the extreme point interval sequence of the high-frequency current waveform during the oscillation decay phase, and corrects the energy absorption efficiency factor by identifying whether the dominant oscillation frequency has shifted to the preset critical frequency band.

[0159] The balance analysis module is used to obtain the high-frequency current waveform of the upstream power surge protector of the same power distribution branch during the oscillation decay stage of the same surge event, perform empirical mode decomposition on the upstream and downstream waveforms respectively, extract the energy distribution probability density of the intrinsic mode function, and calculate the KL divergence between the two as the energy transfer imbalance.

[0160] The alarm generation module is used to generate an alarm signal when the corrected energy absorption efficiency factor is continuously lower than the critical value or the energy transfer imbalance exceeds the preset transfer threshold.

[0161] All calculations involved in the embodiments are dimensionless numerical calculations, and the preset parameters and thresholds in the calculations are set by those skilled in the art according to the actual situation.

[0162] It should be noted that this invention can be deployed on the device itself to realize embedded applications, or it can run on a PC or other terminal with a user interface, thereby meeting various hardware environments and usage requirements.

[0163] The above embodiments can be implemented, in whole or in part, by software, hardware, firmware, or any other combination thereof. When implemented using software, the above embodiments can be implemented, in whole or in part, as a computer program product. The computer program product includes one or more computer instructions or computer programs. When the computer instructions or computer programs are loaded or executed on a computer, all or part of the processes or functions described in the embodiments of this application are generated. The computer can be a general-purpose computer, a special-purpose computer, a computer network, or other programmable device. The computer instructions can be stored in a computer-readable storage medium or transmitted from one computer-readable storage medium to another. For example, the computer instructions can be transmitted from one website, computer, server, or data center to another website, computer, server, or data center via wireless or wired transmission; wired transmission methods include optical fiber, twisted pair, coaxial cable, etc.; wireless transmission includes infrared, microwave, etc. The computer-readable storage medium can be any available medium that a computer can access or a data storage device such as a server or data center containing one or more sets of available media. The available medium can be a magnetic medium (e.g., floppy disk, hard disk, magnetic tape), an optical medium (e.g., DVD), or a semiconductor medium. A semiconductor medium can be a solid-state drive.

[0164] Those skilled in the art will understand that, for the sake of convenience and brevity, the specific working processes of the systems, devices, and modules described above can be referred to the corresponding processes in the foregoing method embodiments, and will not be repeated here.

[0165] In the several embodiments provided in this application, it should be understood that the disclosed systems, apparatuses, and methods can be implemented in other ways. For example, the apparatus embodiments described above are merely illustrative; for instance, the division of modules is only a logical functional division, and in actual implementation, there may be other division methods. For example, multiple modules or components may be combined or integrated into another system, or some features may be ignored or not executed. Furthermore, the coupling or direct coupling or communication connection shown or discussed may be through some interfaces; the indirect coupling or communication connection between apparatuses or modules may be electrical, mechanical, or other forms.

[0166] The modules described as separate components may or may not be physically separate. The components shown as modules may or may not be physical modules; they may be located in one place or distributed across multiple network modules. Some or all of the modules can be selected to achieve the purpose of this embodiment according to actual needs.

[0167] In addition, the functional modules in the various embodiments of this application can be integrated into one processing module, or each module can exist physically separately, or two or more modules can be integrated into one module.

[0168] If the aforementioned functions are implemented as software functional modules and sold or used as independent products, they can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of this application, in essence, or the part that contributes to the prior art, or a portion of the technical solution, can be embodied in the form of a software product. This computer software product is stored in a storage medium and includes several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) to execute all or part of the steps of the methods described in the various embodiments of this application. The aforementioned storage medium includes various media capable of storing program code, such as USB flash drives, portable hard drives, read-only memory (ROM), random access memory (RAM), magnetic disks, or optical disks.

[0169] The above description is merely a specific embodiment of this application, but the scope of protection of this application is not limited thereto. Any variations or substitutions that can be easily conceived by those skilled in the art within the scope of the technology disclosed in this application should be included within the scope of protection of this application. Therefore, the scope of protection of this application should be determined by the scope of the claims.

[0170] In conclusion, the above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of the present invention should be included within the protection scope of the present invention.

Claims

1. A remote monitoring method for power surge protectors based on the Internet of Things, characterized in that, include: S1. Real-time acquisition of high-frequency current waveforms of power surge protectors via IoT monitoring terminals; S2. When the phase deviation between the fundamental wave and the third harmonic in the high-frequency current waveform is less than the critical angle and the coherence coefficient of the oscillation envelope at the preset discrete frequency point is greater than the coherence threshold, the surge event is triggered and the complete current waveform segment is extracted. S3. Separate the rising edge stage and the oscillation decay stage from the complete current waveform segment, calculate the rising edge time width and the oscillation decay time constant, and calculate the ratio of the two as the energy absorption efficiency factor. S4. The extreme point interval sequence of the high-frequency current waveform during the statistical oscillation decay stage generates the decay mode spectrum. The energy absorption efficiency factor is corrected by identifying whether the dominant oscillation frequency has shifted to the preset critical frequency band. S5. Obtain the high-frequency current waveform of the upstream power surge protector of the same distribution branch during the oscillation decay stage of the same surge event. Perform empirical mode decomposition on the upstream and downstream waveforms respectively, extract the energy distribution probability density of the intrinsic mode function, and calculate the KL divergence between the two as the energy transfer imbalance, including: The high-frequency current waveform of the upstream power surge protector during the oscillation decay phase under the same surge event identifier is obtained through the Internet of Things data bus. Empirical mode decomposition is performed on the high-frequency current waveform during the oscillation decay phase of the current power surge protector to obtain the set of inherent mode functions of the current device. Empirical mode decomposition is performed on the high-frequency current waveform during the oscillation decay phase of the upstream power surge protector to obtain the set of inherent mode functions of the upstream equipment; Calculate the energy value of each intrinsic mode function in the current device intrinsic mode function set; Calculate the energy value of each intrinsic mode function in the intrinsic mode function set of the upstream equipment; Generate the energy distribution probability density function of the current device based on the energy values ​​of all intrinsic mode functions of the current device; Generate the energy distribution probability density function of the upstream equipment based on the energy values ​​of all intrinsic mode functions of the upstream equipment; The KL divergence between the current energy distribution probability density function of the equipment and the energy distribution probability density function of the upstream equipment is calculated as the energy transfer imbalance. S6. When the corrected energy absorption efficiency factor is continuously lower than the critical value or the energy transfer imbalance exceeds the preset transfer threshold, an alarm signal is generated.

2. The method for remote monitoring of power surge protectors based on the Internet of Things according to claim 1, characterized in that, The high-frequency current waveform of the power surge protector is collected in real time through an IoT monitoring terminal, including: The high-frequency current signal of the power surge protector grounding wire is collected by the high-frequency current sensor of the Internet of Things monitoring terminal. Perform anti-aliasing filtering on high-frequency current signals; The filtered high-frequency current signal is converted into a digital waveform using an analog-to-digital converter. The digital waveform is uploaded to the edge computing node via the Internet of Things (IoT) communication module; The digital waveform is normalized at the edge computing node to generate a high-frequency current waveform.

3. The method for remote monitoring of power surge protectors based on the Internet of Things according to claim 1, characterized in that, When the phase deviation between the fundamental wave and the third harmonic in a high-frequency current waveform is less than the critical angle and the coherence coefficient of the oscillation envelope at the preset discrete frequency point is greater than the coherence threshold, a surge event marker is triggered and a complete current waveform segment is extracted, including: The fundamental phase angle is obtained by extracting the fundamental component of the high-frequency current waveform; The third harmonic phase angle is obtained by extracting the third harmonic component from the high-frequency current waveform. The difference between the fundamental phase angle and the third harmonic phase angle is calculated as the phase deviation; Bandpass filtering is performed on the high-frequency current waveform at preset discrete frequency points to obtain the oscillation component signal at each frequency point; Envelope detection is performed on each oscillation component signal to generate an envelope signal sequence; The cross-correlation coefficients between the envelope signal sequences are calculated as the oscillating envelope coherence coefficients; The start time of a surge event is marked when the phase deviation is less than the critical angle and the oscillation envelope coherence coefficient is greater than the coherence threshold. The current waveform data of a preset time length is extracted from the starting moment as a complete current waveform segment.

4. The method for remote monitoring of power surge protectors based on the Internet of Things according to claim 1, characterized in that, Separate the rising edge phase and the oscillation decay phase from a complete current waveform segment, calculate the rising edge time width and the oscillation decay time constant, and calculate the ratio of the two as the energy absorption efficiency factor, including: The moment when the current amplitude first reaches its peak value in a complete current waveform segment is taken as the end point of the rising edge. The waveform segment before the end of the rising edge is divided into the rising edge phase; The waveform segment after the end of the rising edge is divided into the oscillation decay stage; The time difference between the start of the rising edge phase and the end of the rising edge phase is calculated as the rising edge time width. The current waveform during the oscillation decay phase is fitted with an exponential decay curve. Extract the decay time constant of the exponential decay curve as the oscillation decay time constant; The energy absorption efficiency factor is generated by calculating the ratio of the rise time width to the oscillation decay time constant.

5. The method for remote monitoring of power surge protectors based on the Internet of Things according to claim 1, characterized in that, The extreme point interval sequence of the high-frequency current waveform during the statistical oscillation decay phase generates the decay mode spectrum. The energy absorption efficiency factor is corrected by identifying whether the dominant oscillation frequency has shifted to a preset critical frequency band, including: Local maxima and minima are detected in the high-frequency current waveform during the oscillation decay phase. Record the time intervals between adjacent maxima and minima to form an extreme point interval sequence; Convert the extreme point interval sequence into an instantaneous frequency sequence; The attenuation mode spectrum is generated by statistically analyzing the distribution histogram of the instantaneous frequency sequence; The frequency with the highest probability density in the decay mode spectrum is identified as the dominant oscillation frequency. Determine whether the dominant oscillation frequency is within the preset critical frequency band; When the dominant oscillation frequency is within the preset critical frequency band, the energy absorption efficiency factor is reduced according to the preset correction coefficient.

6. The method for remote monitoring of power surge protectors based on the Internet of Things according to claim 1, characterized in that, The probability density function for generating the current device energy distribution includes: Calculate the energy value of each intrinsic mode function in the current device intrinsic mode function set. The energy value is the square integral of the amplitude of the corresponding intrinsic mode function during the entire oscillation decay phase. The energy values ​​of all intrinsic mode functions are normalized so that the sum of all energy values ​​equals one. The normalized energy value is used as the probability value of the corresponding intrinsic mode function to generate the current device energy distribution probability density function.

7. The method for remote monitoring of power surge protectors based on the Internet of Things according to claim 1, characterized in that, Calculating the KL divergence as a measure of energy transfer imbalance includes: Take the natural logarithm of each probability value of the current device's energy distribution probability density function; Calculate the difference between the natural logarithm of each probability value of the current equipment's energy distribution probability density function and the corresponding probability value of the upstream equipment's energy distribution probability density function; The weighted difference is obtained by multiplying each probability value of the current device energy distribution probability density function by the corresponding difference value. The summation of all weighted differences yields the KL divergence, which is used as the energy transfer imbalance.

8. The method for remote monitoring of power surge protectors based on the Internet of Things according to claim 1, characterized in that, An alarm signal is generated when the corrected energy absorption efficiency factor continuously falls below the critical value or the energy transfer imbalance exceeds the preset transfer threshold, including: Establish continuous monitoring records for the corrected energy absorption efficiency factor; When the corrected energy absorption efficiency factor of three consecutive surge events is lower than the critical value, an equipment-level degradation alarm signal is generated. Real-time comparison of the relationship between energy transfer imbalance and preset transfer threshold; When the energy transfer imbalance exceeds the preset transfer threshold, a system-level imbalance alarm signal is generated. Encapsulate device-level degradation alarm signals or system-level imbalance alarm signals using IoT communication protocols; The encapsulated alarm signal is transmitted to the IoT monitoring center.

9. A remote monitoring system for power surge protectors based on the Internet of Things (IoT), used to implement the remote monitoring method for power surge protectors based on the IoT as described in any one of claims 1-8, characterized in that, include: The waveform acquisition module is used to acquire the high-frequency current waveform of the power surge protector in real time through an IoT monitoring terminal. The event triggering module is used to trigger surge event marking and extract complete current waveform segments when the phase deviation between the fundamental wave and the third harmonic in the high-frequency current waveform is less than the critical angle and the oscillation envelope coherence coefficient of the preset discrete frequency point is greater than the coherence threshold. The efficiency calculation module is used to separate the rising edge stage and the oscillation decay stage from a complete current waveform segment, calculate the rising edge time width and the oscillation decay time constant, and calculate the ratio of the two as the energy absorption efficiency factor. The frequency correction module is used to generate a decay mode spectrum by statistically analyzing the extreme point interval sequence of the high-frequency current waveform during the oscillation decay phase, and corrects the energy absorption efficiency factor by identifying whether the dominant oscillation frequency has shifted to the preset critical frequency band. The balance analysis module is used to obtain the high-frequency current waveform of the upstream power surge protector of the same power distribution branch during the oscillation decay stage of the same surge event, perform empirical mode decomposition on the upstream and downstream waveforms respectively, extract the energy distribution probability density of the intrinsic mode function, and calculate the KL divergence between the two as the energy transfer imbalance. The alarm generation module is used to generate an alarm signal when the corrected energy absorption efficiency factor is continuously lower than the critical value or the energy transfer imbalance exceeds the preset transfer threshold.

Images