An intelligent diagnostic system for collecting abnormal electricity usage information

Abstract

This invention discloses an intelligent diagnostic system for collecting abnormal electricity consumption information, specifically relating to the field of power system monitoring technology. It addresses the waveform distortion and broadening problems caused by transmission line distribution parameters during long-distance wired transmission in existing systems. The system collects abnormal characteristic waveform data through distributed intelligent terminals, selects characteristic harmonic components from the power line frequency voltage to construct an inherent test signal, and extracts the envelope shape similarity between the response signal and the test signal as a first distortion evaluation index through envelope analysis. After statistical significance judgment, it performs instantaneous frequency stability analysis on signals with significant distortion to obtain a second distortion evaluation index. Finally, based on the dual indices, it generates corresponding waveform compensation parameters to achieve real-time compensation and correction. This effectively maintains the integrity of the abnormal characteristic waveform during transmission, significantly improving the accuracy and reliability of abnormal electricity consumption diagnosis.

Description

Technical Field

[0001] This invention relates to the field of power system monitoring technology, and more specifically, to an intelligent diagnostic system for collecting abnormal electricity consumption information. Background Technology

[0002] In the field of electricity anomaly information collection and intelligent diagnosis, existing systems typically rely on distributed intelligent terminals to collect high-frequency electrical quantities in power lines and upload the collected waveform data to a cloud analysis platform via wired communication media. Such systems can achieve real-time monitoring of signals such as voltage and current, and rely on backend computing power for anomaly identification and fault diagnosis. While existing transmission technologies can ensure data connectivity, in long-distance wired transmission environments, signals must traverse complex cable networks, and their transmission characteristics can affect waveform quality.

[0003] However, in the existing system, abnormal waveforms are distorted and broadened due to the influence of transmission line distribution parameters during waveform data transmission, resulting in waveform distortion. This makes it impossible for the signal received by the cloud to accurately reflect the original abnormal characteristics, thereby affecting the accuracy and reliability of fault diagnosis and making it difficult to meet the needs of high-precision intelligent diagnosis of power anomalies. Summary of the Invention

[0004] In order to overcome the above-mentioned defects of the prior art, the present invention provides an intelligent diagnostic system for collecting abnormal electricity consumption information to solve the problems mentioned in the background art.

[0005] To achieve the above objectives, the present invention provides the following technical solution:

[0006] An intelligent diagnostic system for collecting abnormal electricity consumption information includes:

[0007] The data acquisition module is used to collect abnormal characteristic waveform data in power lines through distributed intelligent terminals;

[0008] The harmonic selection module is used to select the pre-set characteristic harmonic components in the power line frequency voltage as the inherent test signal when abnormal characteristic waveform data is detected.

[0009] The envelope analysis module is used to obtain the response signal of the inherent test signal after passing through the transmission line, extract the envelope of the response signal, and calculate the morphological similarity between the envelope and the envelope of the inherent test signal as the first distortion evaluation index.

[0010] The significance judgment module is used to determine whether the waveform distortion is statistically significant by comparing the first distortion evaluation index with the pre-stored historical benchmark range.

[0011] The frequency stability analysis module is used to analyze the instantaneous frequency stability through Hilbert transform when it has statistical significance, and uses the difference in instantaneous frequency variance between the response signal and the inherent test signal as the second distortion evaluation index.

[0012] The compensation and correction module is used to perform real-time compensation and correction on abnormal feature waveform data by generating corresponding waveform compensation parameters through the first distortion assessment index and the second distortion assessment index based on the statistical significance judgment results; wherein, the waveform compensation parameters are envelope compensation parameters or joint compensation parameters.

[0013] Furthermore, the data acquisition module is further configured to synchronously acquire three-phase voltage waveform data in the power line, and extract waveform segments in a time window that is an integer multiple of the power frequency cycle. Waveform segments that exceed the voltage change threshold are marked as abnormal characteristic waveform data, wherein the abnormal characteristic waveform data includes a sequence of instantaneous voltage sampling values ​​and corresponding timestamps.

[0014] Furthermore, the harmonic selection module is further configured to extract the fundamental frequency from the starting point of the power frequency cycle corresponding to the timestamp of the abnormal characteristic waveform data, obtain the harmonic spectrum by decomposing the power frequency voltage of the power line through fast Fourier transform, select the third and fifth harmonic components as the preset characteristic harmonic components, and synthesize the preset characteristic harmonic components into an inherent test signal, wherein the inherent test signal includes harmonic amplitude, phase and frequency parameters.

[0015] Furthermore, the envelope analysis module is further configured to inject the inherent test signal into the transmission line and collect the output signal as the response signal, perform Hilbert transform on the response signal to extract the envelope, and simultaneously perform Hilbert transform on the inherent test signal to extract the reference envelope. The morphological similarity between the envelope of the response signal and the envelope of the inherent test signal is calculated through a dynamic time warping algorithm, and the morphological similarity is quantified into a first distortion evaluation index, wherein the first distortion evaluation index is a dimensionless value.

[0016] Furthermore, the morphological similarity calculation is further configured as follows: the dynamic time warping algorithm is used to time-align the response signal envelope sequence with the inherent test signal envelope sequence, calculate the minimum cumulative distance under the optimal warping path between the two sequences, and convert the minimum cumulative distance into a similarity value in the range of 0 to 1 as the first distortion evaluation index.

[0017] Furthermore, the significance judgment module is further configured to establish a dynamic pre-stored historical benchmark range update mechanism based on a sliding time window. The current first distortion assessment index and the dynamic benchmark range composed of historical data within the sliding time window are subjected to significance testing. When the significance test exceeds the dynamic threshold, it is marked as statistically significant; otherwise, it is marked as not statistically significant. At the same time, the current first distortion assessment index is included in the dynamic benchmark range updated by the sliding time window. The dynamic threshold is automatically adjusted according to the confidence interval of the data distribution within the sliding time window.

[0018] Furthermore, the frequency stability analysis module is further configured to select the analysis depth based on the statistical significance judgment result. When there is statistical significance, multi-scale instantaneous frequency stability analysis is initiated. First, the response signal and the inherent test signal are synchronously resampled. Then, the instantaneous frequency variance at multiple time scales is calculated. Finally, the weighted sum of the variance differences at each time scale is taken as the second distortion evaluation index. The weighting coefficient is dynamically adjusted according to the frequency stability characteristics of historical data within the sliding time window.

[0019] Furthermore, the compensation and correction module is further configured to select a compensation mode based on the statistical significance judgment result. When there is no statistical significance, envelope compensation parameters mainly used to correct the waveform envelope shape are generated based on the first distortion evaluation index through the envelope gain lookup table. When there is statistical significance, the second distortion evaluation index is used as the dominant phase compensation quantity and combined with the first distortion evaluation index as the auxiliary envelope compensation quantity. A two-parameter coupling algorithm is used to generate joint compensation parameters that are used to correct both waveform envelope shape and phase distortion. The envelope compensation parameters or joint compensation parameters are applied to the subsequently acquired abnormal feature waveform data to achieve real-time compensation and correction. The two-parameter coupling algorithm adopts a phase-envelope serial compensation structure, first performing phase compensation based on the second distortion evaluation index and then performing envelope compensation based on the first distortion evaluation index.

[0020] Furthermore, the envelope gain lookup table is established through a pre-calibration experiment. The pre-calibration experiment involves injecting a set of test signals with known attenuation levels into the transmission line, recording the first distortion evaluation index value and the ideal gain compensation value corresponding to each signal, and forming a mapping relationship database. In application, the mapping relationship database is queried based on the real-time first distortion evaluation index value, and the envelope compensation parameters are generated through interpolation calculation.

[0021] Furthermore, the dual-parameter coupling algorithm is further configured as follows: first, the time delay required for phase compensation is calculated based on the second distortion evaluation index, and time shift correction is performed on the abnormal characteristic waveform data; then, the envelope gain adjustment coefficient is calculated based on the first distortion evaluation index, and amplitude correction is performed on the time-shifted waveform data; and so on, the joint compensation parameters are generated and applied in a serial manner.

[0022] Compared with the prior art, the present invention has the following beneficial effects:

[0023] 1. By constructing an inherent test signal based on characteristic harmonic components and employing a dual distortion assessment mechanism combining envelope morphology similarity analysis and instantaneous frequency stability evaluation, the waveform distortion problem in long-distance wired transmission environments is effectively solved. The system extracts the envelope features of the response signal and the original test signal in real time, accurately quantifies the degree of waveform distortion during transmission, and intelligently selects compensation strategies based on statistical significance judgment, significantly improving the transmission fidelity of abnormal characteristic waveforms. It can effectively overcome the influence of changes in transmission line distribution parameters, ensuring that the waveform data received by the cloud truly reflects the actual abnormal characteristics of the line, and providing an accurate and reliable data foundation for subsequent intelligent diagnosis.

[0024] 2. By organically combining signal analysis and transmission compensation, closed-loop optimization of the power consumption anomaly information acquisition and transmission process is realized. It can not only monitor waveform distortion in real time, but also automatically generate targeted compensation parameters based on distortion characteristics to perform coordinated correction of envelope morphology distortion and phase distortion. This avoids the defect of waveform data and compensation control being separated in traditional systems, and greatly improves the accuracy and reliability of power consumption anomaly diagnosis. Attached Figure Description

[0025] Figure 1 This is a schematic diagram of the structure of an intelligent diagnostic system for collecting abnormal electricity consumption information according to the present invention. Detailed Implementation

[0026] 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.

[0027] Example: Figure 1 A schematic diagram of the structure of an intelligent diagnostic system for collecting abnormal electricity consumption information according to the present invention is provided. The intelligent diagnostic system for collecting abnormal electricity consumption information includes:

[0028] The data acquisition module is used to collect abnormal characteristic waveform data in power lines through distributed intelligent terminals;

[0029] The harmonic selection module is used to select the pre-set characteristic harmonic components in the power line frequency voltage as the inherent test signal when abnormal characteristic waveform data is detected.

[0030] The envelope analysis module is used to obtain the response signal of the inherent test signal after passing through the transmission line, extract the envelope of the response signal, and calculate the morphological similarity between the envelope and the envelope of the inherent test signal as the first distortion evaluation index.

[0031] The significance judgment module is used to determine whether the waveform distortion is statistically significant by comparing the first distortion evaluation index with the pre-stored historical benchmark range.

[0032] The frequency stability analysis module is used to analyze the instantaneous frequency stability through Hilbert transform when it has statistical significance, and uses the difference in instantaneous frequency variance between the response signal and the inherent test signal as the second distortion evaluation index.

[0033] The compensation and correction module is used to perform real-time compensation and correction on abnormal feature waveform data by generating corresponding waveform compensation parameters through the first distortion assessment index and the second distortion assessment index based on the statistical significance judgment results; wherein, the waveform compensation parameters are envelope compensation parameters or joint compensation parameters.

[0034] The data acquisition module synchronously collects three-phase voltage waveform data from power lines through distributed intelligent terminals. This synchronous acquisition employs GPS clock pulse-based synchronous sampling technology to ensure that each intelligent terminal acquires voltage waveforms at the exact same start time and sampling rate. The sampling rate is set to, for example, 6.4kHz to cover the harmonic analysis requirements of the power system. The acquired three-phase voltage waveform data is processed by extracting time windows that are integer multiples of the power frequency cycle. The power frequency cycle reference value is, for example, 20ms (corresponding to a 50Hz system), and the time window length is, for example, 10 power frequency cycles, or 200ms. This time window length ensures both the integrity of the waveform characteristics and meets real-time processing requirements. Within each time window, the voltage waveform data is monitored in real time. When the deviation between the instantaneous value of any phase voltage and the average value of the previous 10 power frequency cycles exceeds the voltage surge threshold, the waveform segment within that time window is marked as abnormal characteristic waveform data.

[0035] The voltage fluctuation threshold is set based on the voltage fluctuation characteristics of the power system during normal operation. Specifically, it is taken as, for example, 15% of the rated voltage of that phase. When the voltage deviation exceeds this threshold, it indicates a possible abnormal situation such as a ground fault, short circuit fault, or load fluctuation. The specific setting of the voltage fluctuation threshold is obtained through historical data analysis. The analysis method includes statistically analyzing the fluctuation range of each phase voltage over 10 minutes during normal operation, taking the average value plus three times the standard deviation as the threshold baseline, and then making appropriate adjustments according to the specific line characteristics. For example, for lines with high stability requirements, the threshold percentage can be appropriately reduced to 10%, while for lines with large fluctuations, it can be appropriately increased to 20%.

[0036] The tagged abnormal waveform data includes a complete sequence of instantaneous voltage samples and corresponding timestamps. The instantaneous voltage sample sequence is stored separately by phase, and the timestamps are accurate to the microsecond level and include GPS time information to ensure that subsequent analysis can accurately trace the time and phase of the anomaly. The abnormal waveform data is stored using a circular buffer, retaining waveform data for the five power frequency cycles before and after the anomaly, providing a complete record of the transient process for subsequent analysis. Each abnormal waveform data record also includes metadata such as the acquisition terminal identifier, line number, and phase information, forming a complete abnormal waveform data package for subsequent processing.

[0037] The time window truncation process employs a sliding window approach, with a sliding step size set to one power frequency cycle to ensure no abnormal waveform segments are missed. Distributed intelligent terminals synchronize time via a high-speed fiber optic network, achieving microsecond-level synchronization accuracy and guaranteeing time consistency across all data acquisition points. The acquired raw waveform data is digitized using a 12-bit analog-to-digital converter, covering a voltage measurement range of 0 to 1000V to meet the monitoring needs of lines with different voltage levels. The marking of abnormal waveform data utilizes a multi-criteria comprehensive judgment. In addition to the voltage mutation threshold, it also incorporates voltage change rate and waveform distortion rate criteria. For example, the voltage change rate threshold is set to 5% of the rated voltage per millisecond, and the waveform distortion rate threshold is set to 8% of the total harmonic distortion rate, ensuring accurate anomaly identification.

[0038] Each marked abnormal waveform data is accompanied by a quality identifier code, indicating whether data loss, sampling asynchrony, or signal saturation occurred during the acquisition process, providing a data quality reference for subsequent analysis. The quality identifier code uses an 8-bit binary encoding method, where the high-order bits represent data integrity, the middle bits represent synchronization status, and the low-order bits represent signal quality. For example, 00000001 indicates complete data but slight signal saturation, and 00000010 indicates complete data but synchronization deviation. A real-time verification mechanism is also implemented during data acquisition, including sampling value range checks, sampling interval consistency checks, and data continuity checks, ensuring the reliability of the acquired data.

[0039] For the storage format of abnormal waveform data, a standardized data structure is adopted, consisting of a file header and a data body. The file header contains metadata such as acquisition time, line parameters, and sampling rate, while the data body contains the sequence of voltage sampling values ​​for each phase. Data compression employs a combination of lossy and lossless compression. Lossy compression is used for normal waveform segments, with a compression ratio set to, for example, 4:1, while lossless compression is used for abnormal waveform segments to ensure that abnormal features are not lost. Data transmission adopts a frame transmission mechanism. Each frame of data includes a frame header, a data body, and a checksum. The checksum uses a cyclic redundancy check algorithm, and the check polynomial adopts a standard such as CRC-16-CCITT to ensure the integrity of data transmission.

[0040] During data acquisition, an automatic gain control mechanism is implemented to dynamically adjust the amplification factor based on the input signal amplitude, preventing signal saturation or excessively low signal-to-noise ratio. The gain adjustment step size is set to, for example, 0.5 dB, with an adjustment period of one power frequency cycle to ensure smooth gain adjustment. The acquisition system is also equipped with power failure protection, using a supercapacitor as a backup power source. This allows for at least 10 ms of data acquisition and storage operations to be maintained in the event of a main power failure, ensuring the integrity of abnormal event records. The acquisition terminal also has self-diagnostic capabilities, periodically checking the status of the sampling channel, storage unit, and communication. Upon detecting problems, it automatically issues alarms and records fault information.

[0041] The harmonic selection module first extracts the fundamental frequency from the starting point of the power frequency cycle corresponding to the timestamp of the abnormal characteristic waveform data. This starting point is determined by detecting the zero-crossing point of the voltage waveform. Specifically, a linear interpolation method is used to improve the accuracy of the zero-crossing point detection, ensuring that the starting point positioning error is less than 1 microsecond. The fundamental frequency extraction adopts an adaptive frequency tracking algorithm, which calculates the average frequency by continuously monitoring the duration of multiple power frequency cycles. For example, the average value is taken after measuring 10 consecutive cycles, achieving a frequency measurement resolution of 0.01Hz. The extracted fundamental frequency is used as a reference parameter for subsequent harmonic analysis. The frequency tracking range is set to ±2Hz of the rated frequency to accommodate power system frequency fluctuations.

[0042] When obtaining the harmonic spectrum by decomposing the power line frequency voltage using Fast Fourier Transform (FFT), the abnormal characteristic waveform data is first preprocessed. This includes removing the DC component and using a window function to reduce spectral leakage. For example, a Hanning window is chosen, with the window length consistent with an integer multiple of the power frequency cycle. The number of points in the FFT is set to an integer power of 2, such as 1024 or 2048 points. The sampling rate is maintained at 6.4 kHz, consistent with the data acquisition module, to ensure the frequency resolution meets requirements. Before spectrum analysis, the data is resampled to ensure that each power frequency cycle contains an integer number of sampling points, avoiding the picket fence effect during spectrum analysis. Polynomial interpolation is used for resampling to ensure the waveform remains undistorted.

[0043] After calculating the harmonic spectrum, the third and fifth harmonic components are selected as pre-defined characteristic harmonic components. This selection is based on experience in analyzing power system anomalies; the third harmonic typically reflects asymmetrical operation, while the fifth harmonic is often associated with distortions caused by power electronic equipment. The selection of characteristic harmonic components also considers the amplitude of each harmonic, selecting only those whose amplitude exceeds the fundamental amplitude by a certain percentage (e.g., 1%) to ensure the selected harmonics have analytical value. The amplitude and phase parameters of each harmonic are directly extracted from the harmonic spectrum. The amplitude is expressed as a percentage relative to the fundamental amplitude, and the phase is expressed as the angular difference relative to the fundamental phase. The phase measurement range covers 0° to 360°.

[0044] When synthesizing pre-defined characteristic harmonic components into a native test signal, a harmonic superposition method is used. The fundamental component and the selected characteristic harmonic components are synthesized according to their respective amplitude, phase, and frequency parameters. The synthesis formula is a linear superposition of the sinusoidal functions of each harmonic, where the frequency of each harmonic is an integer multiple of the fundamental frequency; for example, the third harmonic frequency is 3 times the fundamental frequency, and the fifth harmonic frequency is 5 times the fundamental frequency. The native test signal contains complete harmonic amplitude, phase, and frequency parameter information. These parameters are directly derived from actual measurement data rather than theoretical values, ensuring that the test signal truly reflects the system characteristics.

[0045] The duration of the inherent test signal is set to the same length as the abnormal characteristic waveform data, i.e., 10 power frequency cycles (200ms), to ensure time consistency in subsequent analysis. The phase relationship between harmonics is also considered during the synthesis process, maintaining the original phase difference to ensure that the waveform characteristics of the test signal match the actual situation. After generation, the inherent test signal undergoes amplitude normalization to ensure that its maximum amplitude does not exceed the rated voltage value, avoiding interference to the system during injection. The normalization coefficient is calculated based on the line's rated voltage and the signal's maximum amplitude.

[0046] Quality control measures are implemented during harmonic selection, including spectral signal-to-noise ratio (SNR) checks, harmonic amplitude stability checks, and phase continuity checks. An SNR threshold is set, for example, to 40 dB; harmonic components below this threshold are considered noise and not selected. For harmonic components with large amplitude fluctuations, averaging over multiple power frequency cycles is used to improve measurement accuracy. The number of averaging cycles is dynamically adjusted based on signal stability; for example, 5 cycles are used when the signal is stable, and 10 cycles are used when fluctuations are large. All selected harmonic components are accompanied by a confidence index, calculated based on the quality of the measurement data and the stability of the harmonic amplitude, providing a reliability reference for subsequent analysis.

[0047] The inherent test parameters also include harmonic distortion rate information. The total harmonic distortion rate is calculated as the ratio of the square root of the sum of the squares of the amplitudes of each characteristic harmonic to the amplitude of the fundamental wave, expressed as a percentage. The phase angle of each harmonic is uniformly converted to the absolute phase with reference to the zero-crossing point of the fundamental wave, facilitating subsequent analysis and comparison. All harmonic parameters are timestamped to ensure time synchronization with the original abnormal characteristic waveform data. After the test signal is synthesized, it undergoes distortion verification to ensure that the harmonic characteristics of the synthesized signal are consistent with those of the original signal. The maximum allowable distortion is set to, for example, 1%. If this threshold is exceeded, the harmonic selection and signal synthesis process is repeated.

[0048] Real-time calibration is also implemented during harmonic analysis. A standard signal source is used to periodically calibrate the analysis system every 24 hours to ensure measurement accuracy. The calibration process includes amplitude and phase calibration, with amplitude calibration errors controlled within 0.5% and phase calibration errors within 0.5°. All calibration data is recorded, forming a complete calibration history, ensuring the traceability of measurement results. The analysis results also include environmental temperature compensation parameters. Measurement values ​​are compensated based on the ambient temperature at the time of acquisition. The temperature compensation coefficient is experimentally determined, for example, 0.1% compensation per degree Celsius.

[0049] The envelope analysis module first injects the inherent test signal into the transmission line. This injection process is achieved through a signal generator and coupling circuit. The signal generator output amplitude is set to, for example, 5% to 10% of the line's rated voltage, covering the fundamental frequency and selected harmonic components. The injected signal is coupled to the power line via an isolation transformer. The transformer's turns ratio is selected based on the line voltage level; for example, a 10000V / 100V ratio is used for a 10kV line. The injection time is chosen near the voltage zero-crossing point to reduce transient impacts. The injection duration is consistent with the inherent test signal length, i.e., 10 power frequency cycles (200ms). The level monitoring of the injected signal employs a real-time feedback mechanism. The accuracy of the injected signal amplitude is ensured by measuring the voltage at the coupling point; if the deviation exceeds, for example, 2%, the signal generator output is automatically adjusted.

[0050] When acquiring the output signal as the response signal, the same sampling rate and synchronization method as the data acquisition module are used. The sampling rate is maintained at 6.4kHz, and the synchronization signal comes from the same GPS clock source. The response signal acquisition channel is physically isolated from the injection channel, and a differential input method is used to suppress common-mode interference. The input impedance is set to, for example, 1MΩ to reduce load effects. The acquired response signal undergoes anti-aliasing filtering. The filter cutoff frequency is set to, for example, 40% of the sampling rate, i.e., 2.56kHz. The filter type is a Butterworth low-pass filter, and the order is set to 8th order to provide sufficient stopband attenuation. Real-time quality monitoring is implemented during signal acquisition, including signal amplitude range checks, signal-to-noise ratio checks, and DC offset checks, to ensure the reliability of the acquired data.

[0051] When extracting the envelope from the response signal using Hilbert transform, the signal is first converted into an analytic signal. This is done by using a Fast Fourier Transform (FFT) to transform the signal to the frequency domain, setting the negative frequency components to zero, and then performing an inverse FFT back to the time domain. The magnitude of the analytic signal is the envelope, and the sampling rate of the envelope is consistent with the original signal. After extraction, the envelope is smoothed using a moving average filter with a window length set to, for example, 1 / 4 of the power frequency period (5 ms) to eliminate high-frequency fluctuations while preserving the main characteristics of the envelope. The smoothed envelope is then normalized to unify the amplitude range to between 0 and 1. The normalization coefficients are calculated based on the maximum and minimum values ​​of the signal.

[0052] The method for extracting the reference envelope from the intrinsic test signal using Hilbert transform is completely consistent with the response signal processing, ensuring comparability of the processing steps. The reference envelope is calculated based on the intrinsic test signal before injection and represents the envelope shape under ideal conditions without transmission line distortion. The reference envelope also undergoes the same smoothing and normalization processing, with smoothing parameters consistent with the response signal envelope processing parameters. The duration of the reference envelope is exactly the same as that of the response signal envelope, 200 ms, containing 10 complete power frequency cycles.

[0053] When calculating the morphological similarity between the response signal envelope and the intrinsic test signal envelope using the Dynamic Time Warping (VTW) algorithm, the two envelope sequences are first preprocessed, including removing DC offset and amplitude normalization. The VTW algorithm then searches for the optimal warped path between the two sequences, minimizing the cumulative distance. Euclidean distance is used as the distance metric, and the search range for the warped path is limited to a single strip region with a bandwidth set to, for example, 10% of the sequence length to control computational complexity. The algorithm is implemented using dynamic programming, recursively calculating the cumulative distance matrix and backtracking to find the optimal path.

[0054] After calculating the minimum cumulative distance under the optimal regular path between the two sequences, the minimum cumulative distance is converted into a similarity value in the range of 0 to 1 as the first distortion evaluation index.

[0055] The conversion method employs an exponential function mapping, with the formula: XS=e （-γ×S） Where XS is the similarity index, representing the calculated morphological similarity index, with a value range of (0, 1]; e is the natural constant; S is the distance, representing the minimum cumulative distance calculated by the dynamic time warping algorithm; γ is the scaling factor, determined based on the distribution of typical distance values ​​in historical data, for example, taking the reciprocal of the median distance value. The first distortion assessment index is a dimensionless value; the closer it is to 1, the more similar the envelope morphology, and the closer it is to 0, the greater the difference. γ is dynamically adjusted according to the actual data distribution to ensure that the similarity value has good discriminative power.

[0056] The entire calculation process implements multiple quality controls, including envelope signal-to-noise ratio (SNR) checks, regularization path rationality checks, and similarity value validity checks. The envelope SNR threshold is set to, for example, 20 dB; data below this threshold is flagged as having questionable quality. Regularization path rationality is judged by the monotonicity and continuity of the path; abnormal paths are rejected. Similarity value validity checks include range checks and abrupt change checks; outliers are removed and recalculated. All intermediate calculation results are timestamped and quality-identified to ensure traceability to the original data. Quality control parameters are obtained based on historical data statistical analysis; for example, the SNR threshold is taken as the 5th percentile of the SNR distribution during normal operation.

[0057] Real-time performance optimization is implemented during the computation process, employing a multi-resolution dynamic time warping strategy. First, the downsampled envelope is coarsely aligned, then finely warped at the original resolution. The downsampling factor is dynamically adjusted based on the sequence length; for example, a 4x downsampling is used for a 200ms envelope. The constraints of the warped path are adaptively adjusted based on signal characteristics, using a narrower bandwidth for stable signals and a wider bandwidth for signals with large fluctuations. All algorithm parameters are recorded in a configuration file for easy subsequent analysis and verification. Performance optimization ensures that the time consumed in a single analysis is controlled to, for example, within 10ms, meeting real-time processing requirements.

[0058] The final first distortion assessment index includes a confidence level, calculated based on multiple factors such as data quality, algorithm convergence, and result consistency. Results with a confidence level below a threshold (e.g., 0.8) are marked as pending verification and require manual review or retesting. All calculation processes meet real-time performance requirements, ensuring timely output of assessment results for subsequent processing. The confidence level is calculated using a weighted average method, with the weights of each factor derived from historical data analysis; for example, data quality has a weight of 0.4, algorithm convergence a weight of 0.3, and result consistency a weight of 0.3. The assessment result output includes the first distortion assessment index value, confidence level, timestamp, and quality management identifier, forming a complete assessment record.

[0059] The saliency assessment module first establishes a dynamic pre-stored historical benchmark range update mechanism based on a sliding time window. The length of the sliding time window is set, for example, to 100 consecutive values ​​of the first distortion assessment index, corresponding to approximately 20 seconds of data recording time. The sliding time window adopts a first-in, first-out (FIFO) update strategy; when a new first distortion assessment index arrives, the oldest data point is removed, and the new data point is added to the window. The dynamic benchmark range composed of historical data is obtained by calculating the statistical characteristics of the data within the sliding time window, including statistics such as mean, standard deviation, and percentiles. The mean reflects the central tendency of the data, the standard deviation characterizes the dispersion of the data, and the percentiles are used to describe the distribution pattern of the data.

[0060] When performing a significance test on the dynamic benchmark range formed by the current first distortion assessment index and historical data within the sliding time window, a hypothesis testing method based on the t-distribution is used. The test statistic is calculated as the difference between the current first distortion assessment index value and the sliding window mean, divided by the sliding window standard deviation, and then multiplied by the reciprocal of the square root of the window size. The significance level is set to, for example, 0.05, with the corresponding critical value obtained from the t-distribution table, and the degrees of freedom being the window size minus 1. The normality assumption of the data distribution is considered during the calculation process, and the normality of the data is verified using the Shapiro-Wilke test. When the data does not meet the normality assumption, a non-parametric test method is used.

[0061] A test result exceeding the dynamic threshold is marked as statistically significant; otherwise, it is marked as not statistically significant. The dynamic threshold is automatically adjusted based on the confidence interval of the data distribution within the sliding time window, calculated using, for example, a 95% confidence level. The dynamic threshold is calculated as the sliding window mean plus k times the sliding window standard deviation, where the value of k is determined based on the confidence level; for example, 1.96 for a 95% confidence level. The threshold adjustment employs an adaptive mechanism: when data fluctuations are large, the value of k is appropriately increased; when fluctuations are small, the value of k is decreased. The adjustment magnitude is determined based on the coefficient of variation of historical data.

[0062] Simultaneously, the current primary distortion assessment index is incorporated into the dynamic benchmark range updated within the sliding time window. The update process employs a weighted average method, with the weight of new data points dynamically adjusted based on their deviation from the current distribution. Data points with larger deviations are assigned smaller weights to avoid excessive influence of outliers on the benchmark range. The weight calculation uses an exponential decay function, with the decay coefficient determined based on historical data fluctuation characteristics. Specifically, the decay rate is adjusted by calculating the average absolute deviation of the data within the sliding window.

[0063] The dynamic threshold adjustment process also considers the shape characteristics of the data distribution. When the data exhibits a significantly skewed distribution, the median and interquartile range are used instead of the mean and standard deviation to calculate the threshold. The degree of skewness is determined by the skewness coefficient; a skewness coefficient exceeding, for example, 0.5 indicates significant skewness. Robustness adjustment is also introduced in the threshold calculation, using a pruned mean instead of a simple average, with a pruning ratio set to, for example, 5%, to reduce the impact of outliers. Distribution shape monitoring is achieved by continuously calculating the skewness and kurtosis indices of the data within a sliding window; when these indices exceed preset ranges, adaptive adjustments to the distribution shape are triggered.

[0064] The significance assessment results are accompanied by a confidence level evaluation, which is calculated based on the significance level of the test statistic and data quality. If the confidence level is below a set threshold (e.g., 0.8), the assessment result is marked as pending confirmation and requires further data validation. A detailed log is maintained for the entire assessment process, including input data, intermediate calculation results, and the final assessment conclusion, facilitating subsequent traceability and analysis. The confidence level calculation employs a multi-factor weighted method, considering factors such as data freshness, distribution consistency, and test power. The weights of each factor are obtained through historical data analysis.

[0065] The size of the sliding time window is dynamically adjusted based on the system's operating status. A larger window (e.g., 100 data points) is used during stable system operation, while a smaller window (e.g., 50 data points) is used during system transitions. Window size adjustments are based on data stability and rate of change, automatically determined by monitoring the degree of difference between consecutive data points. The data stability index is calculated as the root mean square value of consecutive differences within the sliding window; when this value exceeds a threshold, the window size is reduced. Window size adjustments are gradual, with each adjustment not exceeding 20% ​​of the original size to avoid drastic changes affecting stability assessment.

[0066] The dynamic baseline range update also considers seasonal factors and changes in operating patterns, establishing multiple baseline range models for different time periods and operating states. Model selection is based on time characteristics and system state characteristics; for example, different baseline ranges are used for weekdays and holidays, and different threshold parameters are used under different load levels. Model parameters are learned from historical data and updated regularly to adapt to system changes. Seasonal pattern recognition uses Fourier analysis to extract periodic features, and operating state classification uses cluster analysis to ensure the accuracy of model selection.

[0067] The entire significance assessment process is monitored and automatically calibrated in real time. Monitoring indicators include false positive rate and false negative rate. Parameters are automatically adjusted when performance indicators deviate from expectations. The calibration process employs a feedback control mechanism, dynamically optimizing threshold parameters and window size based on the accuracy of recent assessment results. Version information for all parameter adjustments is recorded to ensure process traceability and result reproducibility. Performance monitoring uses a sliding window statistical method, with a window length set for example, 1000 assessment results. The false positive rate is calculated and compared with the target value; when the deviation exceeds the allowable range, the parameter calibration process is triggered.

[0068] The frequency stability analysis module selects the analysis depth based on the statistical significance judgment results. When statistical significance is achieved, multi-scale instantaneous frequency stability analysis is initiated. The selection of analysis depth is based on the confidence level in the statistical significance judgment results. When the confidence level exceeds a set threshold (e.g., 0.9), full multi-scale analysis is used; when the confidence level is low, a simplified analysis mode is used. The multi-scale analysis includes three time scales, corresponding to short-time (e.g., 1 power frequency cycle), medium-time (e.g., 5 power frequency cycles), and long-time (e.g., 10 power frequency cycles) analysis windows, respectively. The length of the analysis window for each scale is dynamically adjusted according to the signal characteristics. The selection of analysis depth also considers the system operating state; a standard analysis depth is used during steady-state operation, and an enhanced analysis depth is used during transient processes.

[0069] First, the response signal and the inherent test signal are synchronously resampled. Polynomial interpolation is used to ensure time-domain continuity during resampling. The resampling rate is dynamically calculated based on the fundamental frequency of the signal, ensuring an integer number of sampling points per power frequency cycle. For example, for a 50Hz fundamental frequency, the resampling rate is set to 3200Hz, resulting in 64 sampling points per power frequency cycle. Phase alignment is implemented during resampling to ensure complete synchronization of the zero-crossing points of the two signals, with phase alignment accuracy controlled within 1 microsecond. Synchronization is achieved by using a cross-correlation algorithm to determine the time offset and linear interpolation to achieve sub-sampling precision synchronization.

[0070] Then, the instantaneous frequency variance at multiple time scales is calculated. The instantaneous frequency is obtained by extracting the phase derivative of the signal through Hilbert transform. The frequency calculation employs the central difference method to improve accuracy, with a difference step size of 2 sampling points. Variance calculation uses an unbiased estimation method, with at least 30 sampling points at each time scale to ensure statistical reliability. The frequency variance calculation also eliminates the influence of the trend term by using polynomial fitting to remove frequency drift components; the polynomial order is selected based on the signal characteristics (e.g., 3rd order). The analysis at each time scale uses an overlapping segmentation method with an overlap rate of 50% to improve statistical stability.

[0071] Finally, the weighted sum of the variance differences across all time scales is used as the second distortion assessment index. The variance difference is the absolute difference between the instantaneous frequency variance of the response signal and the instantaneous frequency variance of the intrinsic test signal. The weighting coefficients are dynamically adjusted based on the frequency stability characteristics of historical data within the sliding time window, with the principle of assigning higher weights to time scales with poorer stability. The weighting coefficients are obtained by calculating and normalizing the reciprocals of the historical variances of each scale, ensuring that the sum of the weights is 1. The correlation of variances across scales is also considered during the dynamic adjustment process; when the variances of multiple scales are highly correlated, the weights are appropriately reduced to avoid redundant calculations. The weighting coefficient update cycle is set to, for example, 10 analysis windows, and the update process employs a smooth transition to avoid abrupt changes.

[0072] Multi-scale analysis employs a hierarchical processing strategy, starting with the longest scale and gradually refining to shorter scales. Each scale's analysis results are accompanied by a quality index, calculated based on the signal-to-noise ratio and stability of the data at that scale. When the quality index for a scale falls below a threshold (e.g., 0.7), the weight of that scale is reduced accordingly, and the analysis result is marked as pending verification. Analysis at all scales is performed in parallel, using multi-threading technology to improve computational efficiency and ensure real-time requirements. Anomaly detection is implemented during the analysis process; when outlier data points are detected, robust statistical methods are used to handle them, such as replacing the mean with the median.

[0073] The dynamic adjustment of the weighting coefficients also considers environmental factors, such as the impact of temperature changes and load levels on frequency stability. Environmental factors are monitored in real time using sensors, and a correlation model between environmental factors and frequency stability is established. The model parameters are trained using historical data. When environmental factors change significantly, the weighting coefficients are adaptively adjusted, with the adjustment magnitude determined based on the severity of the environmental change. Environmental factor monitoring includes parameters such as temperature, humidity, and load current, and the sampling rate is consistent with the electrical signal sampling rate.

[0074] The entire frequency stability analysis process employs closed-loop quality control, including data validity checks, algorithm convergence verification, and result reasonableness assessment. Data validity checks ensure the signal-to-noise ratio (SNR) of the input signal is above a threshold (e.g., 30 dB). Algorithm convergence verification is achieved by monitoring the rate of change in iterative calculations. Result reasonableness assessment is based on the statistical range of historical data. If any abnormality occurs in any stage, a re-analysis mechanism is activated, or a backup analysis method is employed. Quality control parameters are adaptively optimized based on long-term operational data; for example, the SNR threshold is dynamically adjusted according to the environmental noise level.

[0075] The analysis results include the value of the second distortion assessment index, contribution analysis at each scale, quality assessment index, and timestamp information. The second distortion assessment index is dimensionless, with values ​​standardized to between 0 and 1 for easy use in subsequent processing. All intermediate calculation results and adjustment parameters are recorded to form a complete analysis log, supporting subsequent traceability and verification. The final output also includes a confidence score, calculated based on the consistency of the analysis results at each scale and data quality. The confidence score is calculated using a weighted average method, with the weights of each factor determined through historical data analysis; for example, data quality has a weight of 0.4, algorithm convergence has a weight of 0.3, and result consistency has a weight of 0.3.

[0076] The frequency stability analysis module also includes a self-calibration function, periodically verifying analysis accuracy using standard test signals. The calibration cycle is set to 24 hours, and the calibration process includes amplitude calibration, frequency calibration, and phase calibration. Calibration data is recorded in a calibration database for long-term performance monitoring and trend analysis. When calibration detects an accuracy deviation exceeding the allowable range, the algorithm parameters are automatically adjusted or a maintenance alarm is issued. Calibration results are also used to update the weighted coefficient calculation model, ensuring continuous optimization of analysis accuracy.

[0077] The compensation and correction module selects the compensation mode based on the statistical significance judgment result. When there is no statistical significance, it generates envelope compensation parameters mainly used to correct the waveform envelope shape based on the first distortion evaluation index through an envelope gain lookup table. The envelope gain lookup table is established through pre-calibration experiments. The pre-calibration experiments involve injecting a set of test signals with known attenuation levels into the transmission line. The test signal set contains multiple amplitude attenuation levels, such as an attenuation gradient from 5% to 20% in 1% steps. Each attenuation level is tested 10 times and the average value is taken to ensure data reliability. The first distortion evaluation index value and the ideal gain compensation value corresponding to each signal are recorded to form a mapping relationship database. The database is stored in a two-dimensional table structure, containing the first distortion evaluation index value and the corresponding ideal gain compensation value. The number of data points is determined according to the accuracy requirements, for example, containing 200 data points. During application, the mapping relationship database is queried based on the real-time first distortion evaluation index value, and the envelope compensation parameters are generated by linear interpolation. The interpolation calculation uses the nearest neighbor method. When the query value exceeds the range of the database, an extrapolation algorithm is used in combination with boundary conditions. The extrapolation range is limited to within 20% of the minimum and maximum values ​​in the database.

[0078] When statistical significance is achieved, the second distortion evaluation index is used as the dominant phase compensation quantity, combined with the first distortion evaluation index as the auxiliary envelope compensation quantity. A two-parameter coupling algorithm is used to generate joint compensation parameters that simultaneously correct waveform envelope morphology and phase distortion. The two-parameter coupling algorithm adopts a phase-envelope serial compensation structure. First, the time delay required for phase compensation is calculated based on the second distortion evaluation index. The time delay is calculated as the product of the second distortion evaluation index and a scaling factor. The scaling factor is determined through calibration experiments, for example, using a value of 0.5 microseconds per unit index. The calibration experiments use test signals with known phase shifts to establish the correspondence between the second distortion evaluation index and the actual phase shift. Time shift correction is performed on abnormal characteristic waveform data. Time shift correction uses sample point translation and interpolation processing. The sinc function interpolation method is selected to ensure waveform integrity, and the number of interpolation points is set to, for example, 8 points to ensure that the time shift accuracy reaches 1 / 10 of the sampling interval.

[0079] Subsequently, the envelope gain adjustment coefficient is calculated based on the first distortion evaluation index. The gain adjustment coefficient is calculated as a function of the reference gain and the first distortion evaluation index, with the function form obtained by fitting historical data, for example, using a quadratic polynomial fitting, requiring a goodness of fit of 0.95 or higher. Amplitude correction is then performed on the time-shifted waveform data, achieved through point-by-point multiplication. The gain adjustment coefficient changes smoothly over time to avoid introducing additional distortion, with the smoothing time constant set to, for example, one power frequency cycle. Joint compensation parameters are generated and applied sequentially in this manner to ensure the order and independence of phase correction and envelope correction. The compensation order is determined based on the physical mechanism of distortion generation, correcting phase distortion first and then amplitude distortion.

[0080] Envelope compensation parameters or joint compensation parameters are applied to subsequently acquired abnormal waveform data to achieve real-time compensation and correction. The application process employs a feedforward control structure, with compensation parameters dynamically updated based on real-time monitored distortion indicators. The update rate is consistent with the data acquisition rate, for example, 6400 updates per second. The compensation effect is ensured through a closed-loop verification mechanism, which compares the distortion indicators of the signal before and after compensation in real time. If the compensation effect does not meet the requirements, the compensation parameters are automatically adjusted or the compensation mode is switched. The required improvement in compensation effect is at least 30%. A smooth transition is implemented during the compensation parameter update process to avoid signal jumps caused by abrupt parameter changes; the transition time is set to, for example, 10 sampling periods.

[0081] During the establishment of the envelope gain lookup table, a standard signal generator was used to inject the test signal set, with signal amplitude accuracy controlled within 0.1% and attenuation accuracy controlled within 0.5%. Each test signal lasted for 10 power frequency cycles, and the measured value after stabilization was recorded. The measured value was then the average of the last 5 cycles. The mapping database was updated periodically, with the update cycle set according to equipment aging and environmental changes, for example, every six months. A gradual update strategy was adopted, with the weight of new data gradually increasing over time until it completely replaced the old data. The database also included data quality identifiers, indicating the measurement conditions and reliability level of each data point.

[0082] In the implementation of the two-parameter coupling algorithm, the calculation of time delay takes into account the influence of signal propagation speed, which is calculated based on line parameters; for example, 99% of the speed of light is used for overhead lines, and 66% for cable lines. Time shift correction accuracy reaches 1 / 10 of the sampling interval, achieved through oversampling and interpolation, with the oversampling factor set to 4 times. Amplitude correction accuracy is controlled within 0.5%, using a 16-bit digital-to-analog converter to ensure resolution, and the conversion rate meets real-time requirements. A serial compensation structure ensures that the two compensation processes do not interfere with each other; the compensation order is determined based on the physical mechanism of distortion generation, correcting phase distortion first and then amplitude distortion.

[0083] A real-time monitoring and protection mechanism is implemented during the application of compensation parameters. Monitoring indicators include the amplitude range, frequency range, and distortion level of the compensated signal. The amplitude range is limited to 90% to 110% of the rated value, the frequency range is limited to ±2% of the fundamental frequency, and the distortion level is assessed through total harmonic distortion (THD) and is required to be less than 5%. When an anomaly is detected, compensation is automatically stopped or switched to a safety mode to prevent improper compensation from introducing new distortions. All compensation operations are logged in detail, including compensation parameters, application time, and compensation effect information, supporting subsequent analysis and optimization. Log data is retained for 30 days.

[0084] The compensation and correction module also includes a self-learning function, continuously optimizing compensation parameters and algorithms by recording historical compensation data and effects. The learning algorithm adopts an incremental learning approach, updating model parameters after each compensation. The learning rate is dynamically adjusted based on data quality, with an initial learning rate set to 0.1, gradually decreasing as the amount of data increases. A knowledge base developed over long-term operation is used to improve the initial compensation effect, especially providing adaptive compensation when system parameters change. The knowledge base is updated every 24 hours.

[0085] The entire compensation process undergoes multiple verifications, including parameter range checks, effect verification, and safety assessments. Parameter range checks ensure that compensation parameters are within reasonable ranges; for example, the gain adjustment coefficient is limited to between 0.5 and 2.0, and the time delay is limited to ±1 power frequency cycle. Effect verification is achieved by comparing distortion indicators before and after compensation, requiring that the improvement in indicators after compensation exceed a set threshold (e.g., 30%). Safety assessments ensure that compensation will not cause system instability or equipment overload; the assessment is based on line parameters and equipment capacity, with the maximum compensation amplitude not exceeding 20% ​​of the equipment's rated capacity. All verification results influence compensation decisions; if any verification fails, corresponding measures are taken, including reducing the compensation intensity, suspending compensation, or issuing an alarm.

[0086] The compensation system also features a manual intervention function, allowing operators to adjust compensation parameters or switch compensation modes based on actual conditions. The intervention interface provides real-time data display and parameter setting functions. All intervention operations are logged, including the operator, operation time, and modifications. The system periodically generates operation reports, statistically analyzing compensation effectiveness and equipment status to provide a basis for maintenance decisions. Report content includes the number of compensations, average improvement level, equipment utilization rate, and abnormal event records. The reporting cycle can be set, such as daily, weekly, or monthly.

[0087] 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.

[0088] 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.

[0089] 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.

[0090] 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.

[0091] 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.

[0092] 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.

[0093] 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.

[0094] 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.

[0095] 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.

[0096] 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. An intelligent diagnostic system for collecting abnormal electricity usage information, characterized in that, include: The data acquisition module is used to collect abnormal characteristic waveform data in power lines through distributed intelligent terminals; The harmonic selection module is used to select the pre-set characteristic harmonic components in the power line frequency voltage as the inherent test signal when abnormal characteristic waveform data is detected. The envelope analysis module is used to obtain the response signal of the inherent test signal after passing through the transmission line, extract the envelope of the response signal, and calculate the morphological similarity between the envelope and the envelope of the inherent test signal as the first distortion evaluation index. The significance judgment module is used to determine whether the waveform distortion is statistically significant by comparing the first distortion evaluation index with the pre-stored historical benchmark range. The frequency stability analysis module is used to analyze the instantaneous frequency stability through Hilbert transform when it has statistical significance, and uses the difference in instantaneous frequency variance between the response signal and the inherent test signal as the second distortion evaluation index. The compensation and correction module is used to perform real-time compensation and correction on abnormal feature waveform data by generating corresponding waveform compensation parameters through the first distortion assessment index and the second distortion assessment index based on the statistical significance judgment results; wherein, the waveform compensation parameters are envelope compensation parameters or joint compensation parameters; The compensation and correction module is further configured to select a compensation mode based on the statistical significance judgment result. When there is no statistical significance, the envelope compensation parameter mainly used to correct the waveform envelope shape is generated based on the first distortion evaluation index through the envelope gain lookup table. When there is statistical significance, the second distortion evaluation index is used as the dominant phase compensation quantity and combined with the first distortion evaluation index as the auxiliary envelope compensation quantity. A two-parameter coupling algorithm is used to generate a joint compensation parameter that is used to correct both the waveform envelope shape and phase distortion. The envelope compensation parameter or the joint compensation parameter is applied to the subsequently acquired abnormal feature waveform data to achieve real-time compensation and correction. The two-parameter coupling algorithm adopts a phase-envelope serial compensation structure, first performing phase compensation based on the second distortion evaluation index and then performing envelope compensation based on the first distortion evaluation index.

2. The intelligent diagnostic system for collecting abnormal electricity consumption information according to claim 1, characterized in that, The data acquisition module is further configured to synchronously acquire three-phase voltage waveform data in the power line, and extract waveform segments in a time window that is an integer multiple of the power frequency cycle. Waveform segments that exceed the voltage change threshold are marked as abnormal characteristic waveform data, wherein the abnormal characteristic waveform data includes a sequence of instantaneous voltage sampling values ​​and corresponding timestamps.

3. The intelligent diagnostic system for collecting abnormal electricity consumption information according to claim 1, characterized in that, The harmonic selection module is further configured to extract the fundamental frequency from the starting point of the power frequency cycle corresponding to the timestamp of the abnormal characteristic waveform data, obtain the harmonic spectrum by decomposing the power frequency voltage of the power line through fast Fourier transform, select the third and fifth harmonic components as the preset characteristic harmonic components, and synthesize the preset characteristic harmonic components into an inherent test signal, wherein the inherent test signal includes harmonic amplitude, phase and frequency parameters.

4. The intelligent diagnostic system for collecting abnormal electricity consumption information according to claim 1, characterized in that, The envelope analysis module is further configured to inject the inherent test signal into the transmission line and collect the output signal as the response signal, perform Hilbert transform on the response signal to extract the envelope, and simultaneously perform Hilbert transform on the inherent test signal to extract the reference envelope. The morphological similarity between the envelope of the response signal and the envelope of the inherent test signal is calculated through a dynamic time warping algorithm, and the morphological similarity is quantified into a first distortion evaluation index, wherein the first distortion evaluation index is a dimensionless value.

5. The intelligent diagnostic system for collecting abnormal electricity consumption information according to claim 4, characterized in that, The calculation of morphological similarity is further configured as follows: the dynamic time warping algorithm is used to align the response signal envelope sequence with the inherent test signal envelope sequence in time, calculate the minimum cumulative distance under the optimal warping path between the two sequences, and convert the minimum cumulative distance into a similarity value in the range of 0 to 1 as the first distortion evaluation index.

6. The intelligent diagnostic system for collecting abnormal electricity consumption information according to claim 1, characterized in that, The significance judgment module is further configured to establish a dynamic pre-stored historical benchmark range update mechanism based on a sliding time window. The current first distortion assessment index and the dynamic benchmark range composed of historical data within the sliding time window are subjected to significance testing. When the significance test exceeds the dynamic threshold, it is marked as statistically significant; otherwise, it is marked as not statistically significant. At the same time, the current first distortion assessment index is included in the dynamic benchmark range updated by the sliding time window. The dynamic threshold is automatically adjusted according to the confidence interval of the data distribution within the sliding time window.

7. The intelligent diagnostic system for collecting abnormal electricity consumption information according to claim 1, characterized in that, The frequency stability analysis module is further configured to select the analysis depth based on the statistical significance judgment result. When there is statistical significance, multi-scale instantaneous frequency stability analysis is started. First, the response signal and the inherent test signal are synchronously resampled. Then, the instantaneous frequency variance at multiple time scales is calculated. Finally, the weighted sum of the variance differences at each time scale is taken as the second distortion evaluation index. The weighting coefficient is dynamically adjusted according to the frequency stability characteristics of historical data within the sliding time window.

8. The intelligent diagnostic system for collecting abnormal electricity consumption information according to claim 1, characterized in that, The envelope gain lookup table is established through a pre-calibration experiment. The pre-calibration experiment involves injecting a set of test signals with known attenuation levels into the transmission line, recording the first distortion evaluation index value and the ideal gain compensation value corresponding to each signal, and forming a mapping relationship database. In application, the mapping relationship database is queried based on the real-time first distortion evaluation index value, and the envelope compensation parameters are generated by interpolation calculation.

9. The intelligent diagnostic system for collecting abnormal electricity consumption information according to claim 1, characterized in that, The dual-parameter coupling algorithm is further configured as follows: first, the time delay required for phase compensation is calculated based on the second distortion evaluation index, and time shift correction is performed on the abnormal characteristic waveform data; then, the envelope gain adjustment coefficient is calculated based on the first distortion evaluation index, and amplitude correction is performed on the time-shifted waveform data; and so on, the joint compensation parameters are generated and applied in a serial manner.

Images